[ { "id": "rt1-world-01-01", "english": "A routing note about a receipt-rendering call is under review before it is acted on.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Elm-42636, Basalt-42636. Recorded latency scores, in the same order: 73, 39. The reference roster-42636@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-42636 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-42636@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-42636@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-42636@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A routing note about a receipt-rendering call is under review before it is acted on.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Elm-42636, Basalt-42636. Recorded latency scores, in the same order: 73, 39. The reference roster-42636@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-42636 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-42636@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Elm-42636 on every execution, with probability one.\nP3: Assigns probability 2/3 to Basalt-42636 and 1/3 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Excluded-42636 on every execution, with probability one.\nP5: Returns Basalt-42636, which has the unique lowest latency score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P3, P5; claims: G2.", "Implementations: P1; claims: G1, G2.", "Implementations: P1; claims: G2.", "Implementations: P1, P2; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P1, P2; claims: G2.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P1, P3, P5; claims: none." ], "answer": "Implementations: P1; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-00", "member_count": 2, "world_id": "rt1-world-01-01", "semantic_world": { "n": 2, "members": [ "Elm-42636", "Basalt-42636" ], "outsider": "Excluded-42636", "scores": [ 73, 39 ], "criterion": "latency", "set_ref": "roster-42636@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P5; claims: G2.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1; claims: G1, G2.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2; claims: G1, G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2; claims: G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P1, P3, P5; claims: none.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-42636": "1/2", "Basalt-42636": "1/2" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Elm-42636 on every execution, with probability one.", "distribution": { "Elm-42636": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Basalt-42636 and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-42636": "1/3", "Basalt-42636": "2/3" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-42636 on every execution, with probability one.", "distribution": { "Excluded-42636": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Basalt-42636, which has the unique lowest latency score, with probability one.", "distribution": { "Basalt-42636": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-01", "english": "A routing note about a grant renewal is under review before it is acted on.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Nettle-9d572, Heron-9d572, Flint-9d572, Lumen-9d572, Gale-9d572. Recorded queue length scores, in the same order: 55, 7, 16, 32, 72. The reference roster-9d572@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9d572 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-9d572@snapshot-1) requires roster-9d572@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-9d572@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A routing note about a grant renewal is under review before it is acted on.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Nettle-9d572, Heron-9d572, Flint-9d572, Lumen-9d572, Gale-9d572. Recorded queue length scores, in the same order: 55, 7, 16, 32, 72. The reference roster-9d572@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9d572 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-9d572@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-9d572, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 2/6 to Nettle-9d572 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Nettle-9d572 on every execution, with probability one.\nP4: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-9d572 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2, P3, P4; claims: G3, G4.", "Implementations: P2, P3, P4; claims: G4.", "Implementations: P1, P4; claims: G3, G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G3, G4.", "Implementations: P4; claims: G2, G3, G4.", "Implementations: P1, P4; claims: G2, G3, G4.", "Implementations: P4; claims: G3, G4." ], "answer": "Implementations: P1, P2, P3, P4; claims: G4.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-00", "member_count": 5, "world_id": "rt1-world-02-01", "semantic_world": { "n": 5, "members": [ "Nettle-9d572", "Heron-9d572", "Flint-9d572", "Lumen-9d572", "Gale-9d572" ], "outsider": "Excluded-9d572", "scores": [ 55, 7, 16, 32, 72 ], "criterion": "queue length", "set_ref": "roster-9d572@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P4; claims: G3, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3, P4; claims: G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P4; claims: G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P4; claims: G2, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P4; claims: G2, G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Heron-9d572, which has the unique lowest queue length score, with probability one.", "distribution": { "Heron-9d572": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Nettle-9d572 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-9d572": "1/3", "Heron-9d572": "1/6", "Flint-9d572": "1/6", "Lumen-9d572": "1/6", "Gale-9d572": "1/6" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Nettle-9d572 on every execution, with probability one.", "distribution": { "Nettle-9d572": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-9d572": "1/5", "Heron-9d572": "1/5", "Flint-9d572": "1/5", "Lumen-9d572": "1/5", "Gale-9d572": "1/5" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-9d572 on every execution, with probability one.", "distribution": { "Excluded-9d572": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-01", "english": "A routing note about a summarisation evaluation slot is under review before it is acted on.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Cobalt-5122c, Onyx-5122c, Pike-5122c, Kestrel-5122c, Flint-5122c, Nettle-5122c, Heron-5122c, Basalt-5122c. Recorded preparation cost scores, in the same order: 42, 35, 11, 92, 6, 45, 85, 69. The reference roster-5122c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5122c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-5122c@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-5122c@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-5122c@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A routing note about a summarisation evaluation slot is under review before it is acted on.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Cobalt-5122c, Onyx-5122c, Pike-5122c, Kestrel-5122c, Flint-5122c, Nettle-5122c, Heron-5122c, Basalt-5122c. Recorded preparation cost scores, in the same order: 42, 35, 11, 92, 6, 45, 85, 69. The reference roster-5122c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5122c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-5122c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-5122c on every execution, with probability one.\nP2: Returns Flint-5122c, which has the unique lowest preparation cost score, with probability one.\nP3: Returns Excluded-5122c on every execution, with probability one.\nP4: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 2/9 to Pike-5122c and 1/9 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P4; claims: G1, G3.", "Implementations: P4, P5; claims: G2, G3, G4.", "Implementations: P1, P2, P4; claims: G3.", "Implementations: P4, P5; claims: G3, G4.", "Implementations: P1, P2, P4, P5; claims: G1, G3.", "Implementations: P4; claims: G3, G4.", "Implementations: P1, P2, P4, P5; claims: G3.", "Implementations: P4; claims: G2, G3, G4." ], "answer": "Implementations: P4; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-00", "member_count": 8, "world_id": "rt1-world-03-01", "semantic_world": { "n": 8, "members": [ "Cobalt-5122c", "Onyx-5122c", "Pike-5122c", "Kestrel-5122c", "Flint-5122c", "Nettle-5122c", "Heron-5122c", "Basalt-5122c" ], "outsider": "Excluded-5122c", "scores": [ 42, 35, 11, 92, 6, 45, 85, 69 ], "criterion": "preparation cost", "set_ref": "roster-5122c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4, P5; claims: G2, G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P2, P4; claims: G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P4, P5; claims: G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G2, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Cobalt-5122c on every execution, with probability one.", "distribution": { "Cobalt-5122c": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Flint-5122c, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Flint-5122c": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-5122c on every execution, with probability one.", "distribution": { "Excluded-5122c": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-5122c": "1/8", "Onyx-5122c": "1/8", "Pike-5122c": "1/8", "Kestrel-5122c": "1/8", "Flint-5122c": "1/8", "Nettle-5122c": "1/8", "Heron-5122c": "1/8", "Basalt-5122c": "1/8" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Pike-5122c and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-5122c": "1/9", "Onyx-5122c": "1/9", "Pike-5122c": "2/9", "Kestrel-5122c": "1/9", "Flint-5122c": "1/9", "Nettle-5122c": "1/9", "Heron-5122c": "1/9", "Basalt-5122c": "1/9" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-01", "english": "A routing note about a stalled queue consumer is under review before it is acted on.\nThe frozen roster lists these eligible standby node identities, in recorded order: Nettle-06682, Elm-06682, Iris-06682, Lumen-06682. Recorded activation cost scores, in the same order: 90, 28, 93, 4. The reference roster-06682@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-06682 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-06682@snapshot-1) requires roster-06682@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-06682@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A routing note about a stalled queue consumer is under review before it is acted on.\nThe frozen roster lists these eligible standby node identities, in recorded order: Nettle-06682, Elm-06682, Iris-06682, Lumen-06682. Recorded activation cost scores, in the same order: 90, 28, 93, 4. The reference roster-06682@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-06682 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-06682@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Nettle-06682 on every execution, with probability one.\nP2: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP3: Assigns probability 2/5 to Lumen-06682 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Lumen-06682, which has the unique lowest activation cost score, with probability one.\nP5: Returns Excluded-06682 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: none; claims: G1, G4.", "Implementations: P1, P3, P4; claims: G2, G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: none; claims: G1.", "Implementations: P1, P3, P4; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P2; claims: G1.", "Implementations: P2; claims: G1, G4." ], "answer": "Implementations: P1, P2, P3, P4; claims: G4.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-00", "member_count": 4, "world_id": "rt1-world-04-01", "semantic_world": { "n": 4, "members": [ "Nettle-06682", "Elm-06682", "Iris-06682", "Lumen-06682" ], "outsider": "Excluded-06682", "scores": [ 90, 28, 93, 4 ], "criterion": "activation cost", "set_ref": "roster-06682@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: none; claims: G1.": { "policies": [], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4; claims: G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G1.": { "policies": [ "P2" ], "guarantees": [ "G1" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Nettle-06682 on every execution, with probability one.", "distribution": { "Nettle-06682": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-06682": "1/4", "Elm-06682": "1/4", "Iris-06682": "1/4", "Lumen-06682": "1/4" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Lumen-06682 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-06682": "1/5", "Elm-06682": "1/5", "Iris-06682": "1/5", "Lumen-06682": "2/5" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Lumen-06682, which has the unique lowest activation cost score, with probability one.", "distribution": { "Lumen-06682": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-06682 on every execution, with probability one.", "distribution": { "Excluded-06682": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-01", "english": "A routing note about a landing-page banner slot is under review before it is acted on.\nThe frozen roster lists these eligible content item identities, in recorded order: Flint-bb8cb, Onyx-bb8cb, Dune-bb8cb, Lumen-bb8cb, Cobalt-bb8cb, Marsh-bb8cb, Iris-bb8cb. Recorded retrieval cost scores, in the same order: 90, 73, 45, 84, 57, 78, 80. The reference roster-bb8cb@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-bb8cb is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-bb8cb@snapshot-1) requires roster-bb8cb@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-bb8cb@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A routing note about a landing-page banner slot is under review before it is acted on.\nThe frozen roster lists these eligible content item identities, in recorded order: Flint-bb8cb, Onyx-bb8cb, Dune-bb8cb, Lumen-bb8cb, Cobalt-bb8cb, Marsh-bb8cb, Iris-bb8cb. Recorded retrieval cost scores, in the same order: 90, 73, 45, 84, 57, 78, 80. The reference roster-bb8cb@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-bb8cb is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-bb8cb@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Dune-bb8cb, which has the unique lowest retrieval cost score, with probability one.\nP3: Returns Flint-bb8cb on every execution, with probability one.\nP4: Returns Excluded-bb8cb on every execution, with probability one.\nP5: Assigns probability 2/8 to Iris-bb8cb and 1/8 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P4; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1; claims: G1, G2.", "Implementations: P1, P4; claims: G2.", "Implementations: P1; claims: G2.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P1, P2, P3, P4, P5; claims: G2." ], "answer": "Implementations: P1, P2, P3, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-00", "member_count": 7, "world_id": "rt1-world-05-01", "semantic_world": { "n": 7, "members": [ "Flint-bb8cb", "Onyx-bb8cb", "Dune-bb8cb", "Lumen-bb8cb", "Cobalt-bb8cb", "Marsh-bb8cb", "Iris-bb8cb" ], "outsider": "Excluded-bb8cb", "scores": [ 90, 73, 45, 84, 57, 78, 80 ], "criterion": "retrieval cost", "set_ref": "roster-bb8cb@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P4; claims: G1, G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1; claims: G1, G2.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P4; claims: G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-bb8cb": "1/7", "Onyx-bb8cb": "1/7", "Dune-bb8cb": "1/7", "Lumen-bb8cb": "1/7", "Cobalt-bb8cb": "1/7", "Marsh-bb8cb": "1/7", "Iris-bb8cb": "1/7" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Dune-bb8cb, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Dune-bb8cb": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Flint-bb8cb on every execution, with probability one.", "distribution": { "Flint-bb8cb": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-bb8cb on every execution, with probability one.", "distribution": { "Excluded-bb8cb": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Iris-bb8cb and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-bb8cb": "1/8", "Onyx-bb8cb": "1/8", "Dune-bb8cb": "1/8", "Lumen-bb8cb": "1/8", "Cobalt-bb8cb": "1/8", "Marsh-bb8cb": "1/8", "Iris-bb8cb": "1/4" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-01", "english": "A routing note about a thumbnail-generation job is under review before it is acted on.\nThe frozen roster lists these eligible worker identities, in recorded order: Onyx-8d714, Kestrel-8d714, Heron-8d714. Recorded current load scores, in the same order: 96, 70, 65. The reference roster-8d714@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8d714 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-8d714@snapshot-1) requires roster-8d714@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-8d714@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A routing note about a thumbnail-generation job is under review before it is acted on.\nThe frozen roster lists these eligible worker identities, in recorded order: Onyx-8d714, Kestrel-8d714, Heron-8d714. Recorded current load scores, in the same order: 96, 70, 65. The reference roster-8d714@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8d714 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-8d714@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/4 to Heron-8d714 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Onyx-8d714 on every execution, with probability one.\nP3: Returns Excluded-8d714 on every execution, with probability one.\nP4: Returns Heron-8d714, which has the unique lowest current load score, with probability one.\nP5: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P4, P5; claims: G4.", "Implementations: P2, P5; claims: G2, G4.", "Implementations: P1, P4, P5; claims: G4.", "Implementations: P1, P2, P4, P5; claims: G2, G4.", "Implementations: P1, P4, P5; claims: G2, G4.", "Implementations: P5; claims: G1, G2, G4.", "Implementations: P2, P5; claims: G1, G2, G4.", "Implementations: P5; claims: G2, G4." ], "answer": "Implementations: P1, P2, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-00", "member_count": 3, "world_id": "rt1-world-06-01", "semantic_world": { "n": 3, "members": [ "Onyx-8d714", "Kestrel-8d714", "Heron-8d714" ], "outsider": "Excluded-8d714", "scores": [ 96, 70, 65 ], "criterion": "current load", "set_ref": "roster-8d714@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P5; claims: G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P4, P5; claims: G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G1, G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P2, P5; claims: G1, G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P5; claims: G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Heron-8d714 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-8d714": "1/4", "Kestrel-8d714": "1/4", "Heron-8d714": "1/2" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Onyx-8d714 on every execution, with probability one.", "distribution": { "Onyx-8d714": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-8d714 on every execution, with probability one.", "distribution": { "Excluded-8d714": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Heron-8d714, which has the unique lowest current load score, with probability one.", "distribution": { "Heron-8d714": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-8d714": "1/3", "Kestrel-8d714": "1/3", "Heron-8d714": "1/3" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-02", "english": "Two operators are reading the same request concerning a postcode lookup and must agree what it asks.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Marsh-eba67, Lumen-eba67, Amber-eba67. Recorded latency scores, in the same order: 33, 15, 13. The reference roster-eba67@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-eba67 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-eba67@snapshot-1) requires roster-eba67@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-eba67@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Two operators are reading the same request concerning a postcode lookup and must agree what it asks.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Marsh-eba67, Lumen-eba67, Amber-eba67. Recorded latency scores, in the same order: 33, 15, 13. The reference roster-eba67@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-eba67 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-eba67@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Amber-eba67, which has the unique lowest latency score, with probability one.\nP2: Returns Marsh-eba67 on every execution, with probability one.\nP3: Assigns probability 3/5 to Lumen-eba67 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Excluded-eba67 on every execution, with probability one.\nP5: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P5; claims: G2, G3.", "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P1, P5; claims: G2, G3, G4.", "Implementations: P5; claims: G2, G3.", "Implementations: P1, P5; claims: G2, G3.", "Implementations: P2, P3, P5; claims: G2, G3.", "Implementations: P2, P3, P5; claims: G2.", "Implementations: P5; claims: G2, G3, G4." ], "answer": "Implementations: P1, P2, P3, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-01", "member_count": 3, "world_id": "rt1-world-01-02", "semantic_world": { "n": 3, "members": [ "Marsh-eba67", "Lumen-eba67", "Amber-eba67" ], "outsider": "Excluded-eba67", "scores": [ 33, 15, 13 ], "criterion": "latency", "set_ref": "roster-eba67@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P5; claims: G2, G3, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P5; claims: G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P5; claims: G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3, P5; claims: G2, G3.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3, P5; claims: G2.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P5; claims: G2, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Amber-eba67, which has the unique lowest latency score, with probability one.", "distribution": { "Amber-eba67": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Marsh-eba67 on every execution, with probability one.", "distribution": { "Marsh-eba67": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Lumen-eba67 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-eba67": "1/5", "Lumen-eba67": "3/5", "Amber-eba67": "1/5" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-eba67 on every execution, with probability one.", "distribution": { "Excluded-eba67": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-eba67": "1/3", "Lumen-eba67": "1/3", "Amber-eba67": "1/3" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-02", "english": "Two operators are reading the same request concerning a code-audit report and must agree what it asks.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Cobalt-9a666, Marsh-9a666, Lumen-9a666, Onyx-9a666, Heron-9a666, Dune-9a666. Recorded queue length scores, in the same order: 57, 27, 18, 65, 67, 11. The reference roster-9a666@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9a666 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-9a666@snapshot-1) requires roster-9a666@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-9a666@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Two operators are reading the same request concerning a code-audit report and must agree what it asks.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Cobalt-9a666, Marsh-9a666, Lumen-9a666, Onyx-9a666, Heron-9a666, Dune-9a666. Recorded queue length scores, in the same order: 57, 27, 18, 65, 67, 11. The reference roster-9a666@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9a666 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-9a666@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Dune-9a666, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-9a666 on every execution, with probability one.\nP4: Assigns probability 3/8 to Onyx-9a666 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Cobalt-9a666 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P4, P5; claims: G2, G4.", "Implementations: P2, P4; claims: G1, G2, G3.", "Implementations: P1, P2, P4, P5; claims: G2.", "Implementations: P2; claims: G1, G2, G3.", "Implementations: P2, P4; claims: G1, G2.", "Implementations: P2; claims: G1, G2.", "Implementations: P1, P2, P5; claims: G2, G4.", "Implementations: P1, P2, P5; claims: G2." ], "answer": "Implementations: P1, P2, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-01", "member_count": 6, "world_id": "rt1-world-02-02", "semantic_world": { "n": 6, "members": [ "Cobalt-9a666", "Marsh-9a666", "Lumen-9a666", "Onyx-9a666", "Heron-9a666", "Dune-9a666" ], "outsider": "Excluded-9a666", "scores": [ 57, 27, 18, 65, 67, 11 ], "criterion": "queue length", "set_ref": "roster-9a666@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P4; claims: G1, G2, G3.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P2; claims: G1, G2, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P4; claims: G1, G2.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P5; claims: G2.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Dune-9a666, which has the unique lowest queue length score, with probability one.", "distribution": { "Dune-9a666": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-9a666": "1/6", "Marsh-9a666": "1/6", "Lumen-9a666": "1/6", "Onyx-9a666": "1/6", "Heron-9a666": "1/6", "Dune-9a666": "1/6" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-9a666 on every execution, with probability one.", "distribution": { "Excluded-9a666": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Onyx-9a666 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-9a666": "1/8", "Marsh-9a666": "1/8", "Lumen-9a666": "1/8", "Onyx-9a666": "3/8", "Heron-9a666": "1/8", "Dune-9a666": "1/8" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Cobalt-9a666 on every execution, with probability one.", "distribution": { "Cobalt-9a666": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-02", "english": "Two operators are reading the same request concerning a tool-use evaluation slot and must agree what it asks.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Heron-23b73, Elm-23b73. Recorded preparation cost scores, in the same order: 91, 82. The reference roster-23b73@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-23b73 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-23b73@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-23b73@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-23b73@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning a tool-use evaluation slot and must agree what it asks.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Heron-23b73, Elm-23b73. Recorded preparation cost scores, in the same order: 91, 82. The reference roster-23b73@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-23b73 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-23b73@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-23b73 on every execution, with probability one.\nP2: Assigns probability 3/4 to Heron-23b73 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-23b73 on every execution, with probability one.\nP5: Returns Elm-23b73, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P3; claims: G4.", "Implementations: P3; claims: G1, G4.", "Implementations: none; claims: G1, G4.", "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P1, P2, P5; claims: G1.", "Implementations: P1, P2, P5; claims: G1, G3.", "Implementations: none; claims: G4.", "Implementations: P1, P2, P3, P5; claims: G1." ], "answer": "Implementations: P3; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-01", "member_count": 2, "world_id": "rt1-world-03-02", "semantic_world": { "n": 2, "members": [ "Heron-23b73", "Elm-23b73" ], "outsider": "Excluded-23b73", "scores": [ 91, 82 ], "criterion": "preparation cost", "set_ref": "roster-23b73@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3; claims: G4.": { "policies": [ "P3" ], "guarantees": [ "G4" ] }, "Implementations: P3; claims: G1, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P5; claims: G1.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: none; claims: G4.": { "policies": [], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Heron-23b73 on every execution, with probability one.", "distribution": { "Heron-23b73": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Heron-23b73 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-23b73": "3/4", "Elm-23b73": "1/4" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-23b73": "1/2", "Elm-23b73": "1/2" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-23b73 on every execution, with probability one.", "distribution": { "Excluded-23b73": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Elm-23b73, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Elm-23b73": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-02", "english": "Two operators are reading the same request concerning an unreachable metrics store and must agree what it asks.\nThe frozen roster lists these eligible standby node identities, in recorded order: Cobalt-c121b, Heron-c121b, Flint-c121b, Pike-c121b, Elm-c121b. Recorded activation cost scores, in the same order: 76, 95, 87, 19, 9. The reference roster-c121b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c121b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-c121b@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-c121b@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-c121b@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning an unreachable metrics store and must agree what it asks.\nThe frozen roster lists these eligible standby node identities, in recorded order: Cobalt-c121b, Heron-c121b, Flint-c121b, Pike-c121b, Elm-c121b. Recorded activation cost scores, in the same order: 76, 95, 87, 19, 9. The reference roster-c121b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c121b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-c121b@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Elm-c121b, which has the unique lowest activation cost score, with probability one.\nP2: Assigns probability 3/7 to Elm-c121b and 1/7 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-c121b on every execution, with probability one.\nP4: Returns Cobalt-c121b on every execution, with probability one.\nP5: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P5; claims: G1, G4.", "Implementations: P3, P5; claims: G1.", "Implementations: P3, P5; claims: G1, G4.", "Implementations: P5; claims: G1.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1, P2, P4, P5; claims: none." ], "answer": "Implementations: P5; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-01", "member_count": 5, "world_id": "rt1-world-04-02", "semantic_world": { "n": 5, "members": [ "Cobalt-c121b", "Heron-c121b", "Flint-c121b", "Pike-c121b", "Elm-c121b" ], "outsider": "Excluded-c121b", "scores": [ 76, 95, 87, 19, 9 ], "criterion": "activation cost", "set_ref": "roster-c121b@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P3, P5; claims: G1.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P3, P5; claims: G1, G4.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P5; claims: G1.": { "policies": [ "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Elm-c121b, which has the unique lowest activation cost score, with probability one.", "distribution": { "Elm-c121b": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Elm-c121b and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-c121b": "1/7", "Heron-c121b": "1/7", "Flint-c121b": "1/7", "Pike-c121b": "1/7", "Elm-c121b": "3/7" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-c121b on every execution, with probability one.", "distribution": { "Excluded-c121b": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Cobalt-c121b on every execution, with probability one.", "distribution": { "Cobalt-c121b": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-c121b": "1/5", "Heron-c121b": "1/5", "Flint-c121b": "1/5", "Pike-c121b": "1/5", "Elm-c121b": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-02", "english": "Two operators are reading the same request concerning a newsletter header slot and must agree what it asks.\nThe frozen roster lists these eligible content item identities, in recorded order: Elm-f8490, Iris-f8490, Nettle-f8490, Dune-f8490, Lumen-f8490, Amber-f8490, Cobalt-f8490, Gale-f8490. Recorded retrieval cost scores, in the same order: 96, 35, 78, 76, 77, 84, 34, 17. The reference roster-f8490@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f8490 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-f8490@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-f8490@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-f8490@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning a newsletter header slot and must agree what it asks.\nThe frozen roster lists these eligible content item identities, in recorded order: Elm-f8490, Iris-f8490, Nettle-f8490, Dune-f8490, Lumen-f8490, Amber-f8490, Cobalt-f8490, Gale-f8490. Recorded retrieval cost scores, in the same order: 96, 35, 78, 76, 77, 84, 34, 17. The reference roster-f8490@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f8490 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-f8490@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Gale-f8490, which has the unique lowest retrieval cost score, with probability one.\nP3: Assigns probability 3/10 to Lumen-f8490 and 1/10 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Elm-f8490 on every execution, with probability one.\nP5: Returns Excluded-f8490 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P3; claims: G1.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1; claims: G1, G3, G4.", "Implementations: P1; claims: G1, G4.", "Implementations: P1, P2, P3; claims: G1, G4.", "Implementations: P1, P2, P3, P4; claims: G1, G4.", "Implementations: P1, P4; claims: G1, G4.", "Implementations: P1, P4; claims: G1, G3, G4." ], "answer": "Implementations: P1; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-01", "member_count": 8, "world_id": "rt1-world-05-02", "semantic_world": { "n": 8, "members": [ "Elm-f8490", "Iris-f8490", "Nettle-f8490", "Dune-f8490", "Lumen-f8490", "Amber-f8490", "Cobalt-f8490", "Gale-f8490" ], "outsider": "Excluded-f8490", "scores": [ 96, 35, 78, 76, 77, 84, 34, 17 ], "criterion": "retrieval cost", "set_ref": "roster-f8490@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3; claims: G1.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1; claims: G1, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3; claims: G1, G4.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P4; claims: G1, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P4; claims: G1, G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-f8490": "1/8", "Iris-f8490": "1/8", "Nettle-f8490": "1/8", "Dune-f8490": "1/8", "Lumen-f8490": "1/8", "Amber-f8490": "1/8", "Cobalt-f8490": "1/8", "Gale-f8490": "1/8" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Gale-f8490, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Gale-f8490": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Lumen-f8490 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-f8490": "1/10", "Iris-f8490": "1/10", "Nettle-f8490": "1/10", "Dune-f8490": "1/10", "Lumen-f8490": "3/10", "Amber-f8490": "1/10", "Cobalt-f8490": "1/10", "Gale-f8490": "1/10" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Elm-f8490 on every execution, with probability one.", "distribution": { "Elm-f8490": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-f8490 on every execution, with probability one.", "distribution": { "Excluded-f8490": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-02", "english": "Two operators are reading the same request concerning a ledger-reconciliation job and must agree what it asks.\nThe frozen roster lists these eligible worker identities, in recorded order: Gale-2ff36, Lumen-2ff36, Heron-2ff36, Amber-2ff36. Recorded current load scores, in the same order: 77, 62, 42, 86. The reference roster-2ff36@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-2ff36 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-2ff36@snapshot-1) requires roster-2ff36@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-2ff36@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Two operators are reading the same request concerning a ledger-reconciliation job and must agree what it asks.\nThe frozen roster lists these eligible worker identities, in recorded order: Gale-2ff36, Lumen-2ff36, Heron-2ff36, Amber-2ff36. Recorded current load scores, in the same order: 77, 62, 42, 86. The reference roster-2ff36@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-2ff36 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-2ff36@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-2ff36, which has the unique lowest current load score, with probability one.\nP2: Returns Gale-2ff36 on every execution, with probability one.\nP3: Returns Excluded-2ff36 on every execution, with probability one.\nP4: Assigns probability 3/6 to Amber-2ff36 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P5; claims: G1, G2, G3.", "Implementations: P2, P4, P5; claims: G3, G4.", "Implementations: P1, P2, P4, P5; claims: G3, G4.", "Implementations: P5; claims: G1, G3.", "Implementations: P1, P5; claims: G1, G3.", "Implementations: P1, P2, P4, P5; claims: G3.", "Implementations: P1, P5; claims: G1, G2, G3.", "Implementations: P2, P4, P5; claims: G3." ], "answer": "Implementations: P1, P2, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-01", "member_count": 4, "world_id": "rt1-world-06-02", "semantic_world": { "n": 4, "members": [ "Gale-2ff36", "Lumen-2ff36", "Heron-2ff36", "Amber-2ff36" ], "outsider": "Excluded-2ff36", "scores": [ 77, 62, 42, 86 ], "criterion": "current load", "set_ref": "roster-2ff36@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P4, P5; claims: G3, G4.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P5; claims: G1, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P5; claims: G1, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P5; claims: G1, G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P4, P5; claims: G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Heron-2ff36, which has the unique lowest current load score, with probability one.", "distribution": { "Heron-2ff36": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Gale-2ff36 on every execution, with probability one.", "distribution": { "Gale-2ff36": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-2ff36 on every execution, with probability one.", "distribution": { "Excluded-2ff36": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Amber-2ff36 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-2ff36": "1/6", "Lumen-2ff36": "1/6", "Heron-2ff36": "1/6", "Amber-2ff36": "1/2" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-2ff36": "1/4", "Lumen-2ff36": "1/4", "Heron-2ff36": "1/4", "Amber-2ff36": "1/4" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-03", "english": "A change request has arrived for a map-tile fetch; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Elm-e54b5, Lumen-e54b5, Gale-e54b5, Dune-e54b5. Recorded latency scores, in the same order: 87, 76, 74, 12. The reference roster-e54b5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e54b5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e54b5@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-e54b5@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-e54b5@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A change request has arrived for a map-tile fetch; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Elm-e54b5, Lumen-e54b5, Gale-e54b5, Dune-e54b5. Recorded latency scores, in the same order: 87, 76, 74, 12. The reference roster-e54b5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e54b5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e54b5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Elm-e54b5 on every execution, with probability one.\nP2: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP3: Assigns probability 4/7 to Gale-e54b5 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Dune-e54b5, which has the unique lowest latency score, with probability one.\nP5: Returns Excluded-e54b5 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2; claims: G1, G3, G4.", "Implementations: P1, P2, P4; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1, P2, P4; claims: G1.", "Implementations: P2; claims: G1, G3.", "Implementations: P2, P3; claims: G1, G3, G4.", "Implementations: P2, P3; claims: G1, G3." ], "answer": "Implementations: P2; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-02", "member_count": 4, "world_id": "rt1-world-01-03", "semantic_world": { "n": 4, "members": [ "Elm-e54b5", "Lumen-e54b5", "Gale-e54b5", "Dune-e54b5" ], "outsider": "Excluded-e54b5", "scores": [ 87, 76, 74, 12 ], "criterion": "latency", "set_ref": "roster-e54b5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P4; claims: G1, G2.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4; claims: G1.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2; claims: G1, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P3; claims: G1, G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3; claims: G1, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Elm-e54b5 on every execution, with probability one.", "distribution": { "Elm-e54b5": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-e54b5": "1/4", "Lumen-e54b5": "1/4", "Gale-e54b5": "1/4", "Dune-e54b5": "1/4" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Gale-e54b5 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-e54b5": "1/7", "Lumen-e54b5": "1/7", "Gale-e54b5": "4/7", "Dune-e54b5": "1/7" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Dune-e54b5, which has the unique lowest latency score, with probability one.", "distribution": { "Dune-e54b5": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-e54b5 on every execution, with probability one.", "distribution": { "Excluded-e54b5": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-03", "english": "A change request has arrived for a survey instrument; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-447d0, Gale-447d0, Dune-447d0, Flint-447d0, Marsh-447d0, Elm-447d0, Amber-447d0. Recorded queue length scores, in the same order: 27, 39, 56, 5, 30, 35, 11. The reference roster-447d0@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-447d0 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-447d0@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-447d0@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-447d0@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A change request has arrived for a survey instrument; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-447d0, Gale-447d0, Dune-447d0, Flint-447d0, Marsh-447d0, Elm-447d0, Amber-447d0. Recorded queue length scores, in the same order: 27, 39, 56, 5, 30, 35, 11. The reference roster-447d0@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-447d0 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-447d0@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-447d0 on every execution, with probability one.\nP2: Returns Basalt-447d0 on every execution, with probability one.\nP3: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Flint-447d0, which has the unique lowest queue length score, with probability one.\nP5: Assigns probability 4/10 to Marsh-447d0 and 1/10 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: none; claims: G4.", "Implementations: P2, P4, P5; claims: G1.", "Implementations: P2, P3, P4, P5; claims: G1, G3.", "Implementations: P3; claims: G4.", "Implementations: P3; claims: G1, G4.", "Implementations: none; claims: G1, G4.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P2, P4, P5; claims: G1, G3." ], "answer": "Implementations: P3; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-02", "member_count": 7, "world_id": "rt1-world-02-03", "semantic_world": { "n": 7, "members": [ "Basalt-447d0", "Gale-447d0", "Dune-447d0", "Flint-447d0", "Marsh-447d0", "Elm-447d0", "Amber-447d0" ], "outsider": "Excluded-447d0", "scores": [ 27, 39, 56, 5, 30, 35, 11 ], "criterion": "queue length", "set_ref": "roster-447d0@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G4.": { "policies": [], "guarantees": [ "G4" ] }, "Implementations: P2, P4, P5; claims: G1.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3; claims: G4.": { "policies": [ "P3" ], "guarantees": [ "G4" ] }, "Implementations: P3; claims: G1, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-447d0 on every execution, with probability one.", "distribution": { "Excluded-447d0": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Basalt-447d0 on every execution, with probability one.", "distribution": { "Basalt-447d0": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-447d0": "1/7", "Gale-447d0": "1/7", "Dune-447d0": "1/7", "Flint-447d0": "1/7", "Marsh-447d0": "1/7", "Elm-447d0": "1/7", "Amber-447d0": "1/7" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Flint-447d0, which has the unique lowest queue length score, with probability one.", "distribution": { "Flint-447d0": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Marsh-447d0 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-447d0": "1/10", "Gale-447d0": "1/10", "Dune-447d0": "1/10", "Flint-447d0": "1/10", "Marsh-447d0": "2/5", "Elm-447d0": "1/10", "Amber-447d0": "1/10" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-03", "english": "A change request has arrived for a calibration evaluation slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Basalt-ace37, Gale-ace37, Heron-ace37. Recorded preparation cost scores, in the same order: 71, 62, 20. The reference roster-ace37@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ace37 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ace37@snapshot-1) requires roster-ace37@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-ace37@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A change request has arrived for a calibration evaluation slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Basalt-ace37, Gale-ace37, Heron-ace37. Recorded preparation cost scores, in the same order: 71, 62, 20. The reference roster-ace37@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ace37 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ace37@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-ace37 on every execution, with probability one.\nP2: Returns Basalt-ace37 on every execution, with probability one.\nP3: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 4/6 to Heron-ace37 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Heron-ace37, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P3; claims: G1.", "Implementations: P3; claims: G1, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P2, P3, P4, P5; claims: none.", "Implementations: P3; claims: G1.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P1, P3; claims: G1, G3.", "Implementations: P1, P2, P3, P4, P5; claims: none." ], "answer": "Implementations: P2, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-02", "member_count": 3, "world_id": "rt1-world-03-03", "semantic_world": { "n": 3, "members": [ "Basalt-ace37", "Gale-ace37", "Heron-ace37" ], "outsider": "Excluded-ace37", "scores": [ 71, 62, 20 ], "criterion": "preparation cost", "set_ref": "roster-ace37@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3; claims: G1.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1" ] }, "Implementations: P3; claims: G1, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4, P5; claims: none.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P3; claims: G1.": { "policies": [ "P3" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3; claims: G1, G3.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-ace37 on every execution, with probability one.", "distribution": { "Excluded-ace37": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Basalt-ace37 on every execution, with probability one.", "distribution": { "Basalt-ace37": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-ace37": "1/3", "Gale-ace37": "1/3", "Heron-ace37": "1/3" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Heron-ace37 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-ace37": "1/6", "Gale-ace37": "1/6", "Heron-ace37": "2/3" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Heron-ace37, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Heron-ace37": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-03", "english": "A change request has arrived for a crashed scheduler; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible standby node identities, in recorded order: Nettle-19965, Pike-19965, Onyx-19965, Heron-19965, Kestrel-19965, Flint-19965. Recorded activation cost scores, in the same order: 85, 41, 58, 44, 23, 11. The reference roster-19965@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-19965 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-19965@snapshot-1) requires roster-19965@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-19965@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A change request has arrived for a crashed scheduler; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible standby node identities, in recorded order: Nettle-19965, Pike-19965, Onyx-19965, Heron-19965, Kestrel-19965, Flint-19965. Recorded activation cost scores, in the same order: 85, 41, 58, 44, 23, 11. The reference roster-19965@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-19965 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-19965@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Nettle-19965 on every execution, with probability one.\nP2: Assigns probability 4/9 to Heron-19965 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP3: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Flint-19965, which has the unique lowest activation cost score, with probability one.\nP5: Returns Excluded-19965 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P2, P3, P4; claims: G1.", "Implementations: P1, P2, P3, P4; claims: G1, G2.", "Implementations: P1, P3; claims: G1, G2, G3.", "Implementations: P2, P3, P4; claims: G1, G2.", "Implementations: P1, P3; claims: G1, G2.", "Implementations: P3; claims: G1, G2, G3.", "Implementations: P3; claims: G1, G2." ], "answer": "Implementations: P1, P2, P3, P4; claims: G1.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-02", "member_count": 6, "world_id": "rt1-world-04-03", "semantic_world": { "n": 6, "members": [ "Nettle-19965", "Pike-19965", "Onyx-19965", "Heron-19965", "Kestrel-19965", "Flint-19965" ], "outsider": "Excluded-19965", "scores": [ 85, 41, 58, 44, 23, 11 ], "criterion": "activation cost", "set_ref": "roster-19965@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4; claims: G1.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3; claims: G1, G2, G3.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3, P4; claims: G1, G2.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3; claims: G1, G2.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Nettle-19965 on every execution, with probability one.", "distribution": { "Nettle-19965": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Heron-19965 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-19965": "1/9", "Pike-19965": "1/9", "Onyx-19965": "1/9", "Heron-19965": "4/9", "Kestrel-19965": "1/9", "Flint-19965": "1/9" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-19965": "1/6", "Pike-19965": "1/6", "Onyx-19965": "1/6", "Heron-19965": "1/6", "Kestrel-19965": "1/6", "Flint-19965": "1/6" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Flint-19965, which has the unique lowest activation cost score, with probability one.", "distribution": { "Flint-19965": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-19965 on every execution, with probability one.", "distribution": { "Excluded-19965": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-03", "english": "A change request has arrived for a help-centre example slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible content item identities, in recorded order: Elm-44035, Dune-44035. Recorded retrieval cost scores, in the same order: 76, 20. The reference roster-44035@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-44035 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-44035@snapshot-1) requires roster-44035@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-44035@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A change request has arrived for a help-centre example slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible content item identities, in recorded order: Elm-44035, Dune-44035. Recorded retrieval cost scores, in the same order: 76, 20. The reference roster-44035@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-44035 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-44035@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-44035 on every execution, with probability one.\nP2: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Dune-44035, which has the unique lowest retrieval cost score, with probability one.\nP4: Returns Elm-44035 on every execution, with probability one.\nP5: Assigns probability 4/5 to Dune-44035 and 1/5 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2, P3, P4, P5; claims: G3.", "Implementations: P2; claims: G2, G3.", "Implementations: P2, P4, P5; claims: G3.", "Implementations: P2, P3; claims: G2, G3.", "Implementations: P2, P3; claims: G2, G3, G4.", "Implementations: P2; claims: G2, G3, G4.", "Implementations: P2, P4, P5; claims: G1, G3.", "Implementations: P2, P3, P4, P5; claims: G1, G3." ], "answer": "Implementations: P2, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-02", "member_count": 2, "world_id": "rt1-world-05-03", "semantic_world": { "n": 2, "members": [ "Elm-44035", "Dune-44035" ], "outsider": "Excluded-44035", "scores": [ 76, 20 ], "criterion": "retrieval cost", "set_ref": "roster-44035@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P4, P5; claims: G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G2, G3.": { "policies": [ "P2" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P4, P5; claims: G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3; claims: G2, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3; claims: G2, G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P2; claims: G2, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P2, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-44035 on every execution, with probability one.", "distribution": { "Excluded-44035": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-44035": "1/2", "Dune-44035": "1/2" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Dune-44035, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Dune-44035": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Elm-44035 on every execution, with probability one.", "distribution": { "Elm-44035": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Dune-44035 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-44035": "1/5", "Dune-44035": "4/5" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-03", "english": "A change request has arrived for a spell-check job; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible worker identities, in recorded order: Basalt-606db, Iris-606db, Gale-606db, Juniper-606db, Elm-606db. Recorded current load scores, in the same order: 93, 23, 19, 3, 45. The reference roster-606db@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-606db is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-606db@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-606db@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-606db@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A change request has arrived for a spell-check job; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible worker identities, in recorded order: Basalt-606db, Iris-606db, Gale-606db, Juniper-606db, Elm-606db. Recorded current load scores, in the same order: 93, 23, 19, 3, 45. The reference roster-606db@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-606db is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-606db@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Basalt-606db on every execution, with probability one.\nP2: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP3: Assigns probability 4/8 to Juniper-606db and 1/8 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Juniper-606db, which has the unique lowest current load score, with probability one.\nP5: Returns Excluded-606db on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: P1, P2, P4; claims: G2, G4.", "Implementations: P2, P3; claims: G1.", "Implementations: P2; claims: G1.", "Implementations: P1, P2, P4; claims: G2.", "Implementations: P2, P3; claims: G1, G2." ], "answer": "Implementations: P2; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-02", "member_count": 5, "world_id": "rt1-world-06-03", "semantic_world": { "n": 5, "members": [ "Basalt-606db", "Iris-606db", "Gale-606db", "Juniper-606db", "Elm-606db" ], "outsider": "Excluded-606db", "scores": [ 93, 23, 19, 3, 45 ], "criterion": "current load", "set_ref": "roster-606db@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3; claims: G1.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1" ] }, "Implementations: P2; claims: G1.": { "policies": [ "P2" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4; claims: G2.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P2, P3; claims: G1, G2.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Basalt-606db on every execution, with probability one.", "distribution": { "Basalt-606db": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-606db": "1/5", "Iris-606db": "1/5", "Gale-606db": "1/5", "Juniper-606db": "1/5", "Elm-606db": "1/5" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Juniper-606db and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-606db": "1/8", "Iris-606db": "1/8", "Gale-606db": "1/8", "Juniper-606db": "1/2", "Elm-606db": "1/8" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Juniper-606db, which has the unique lowest current load score, with probability one.", "distribution": { "Juniper-606db": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-606db on every execution, with probability one.", "distribution": { "Excluded-606db": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-04", "english": "A checklist item asks what the selection wording for a session-refresh call actually commits to.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Pike-3b82c, Marsh-3b82c, Elm-3b82c, Iris-3b82c, Gale-3b82c. Recorded latency scores, in the same order: 80, 82, 78, 83, 69. The reference roster-3b82c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3b82c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-3b82c@snapshot-1) requires roster-3b82c@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-3b82c@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a session-refresh call actually commits to.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Pike-3b82c, Marsh-3b82c, Elm-3b82c, Iris-3b82c, Gale-3b82c. Recorded latency scores, in the same order: 80, 82, 78, 83, 69. The reference roster-3b82c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3b82c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-3b82c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Pike-3b82c on every execution, with probability one.\nP2: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Gale-3b82c, which has the unique lowest latency score, with probability one.\nP4: Returns Excluded-3b82c on every execution, with probability one.\nP5: Assigns probability 2/6 to Iris-3b82c and 1/6 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: none; claims: G4.", "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P1, P2, P3, P5; claims: G2, G3.", "Implementations: P2; claims: G4.", "Implementations: P1, P3, P5; claims: G2, G3.", "Implementations: P2; claims: G2, G4.", "Implementations: none; claims: G2, G4.", "Implementations: P1, P3, P5; claims: G2." ], "answer": "Implementations: P1, P2, P3, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-03", "member_count": 5, "world_id": "rt1-world-01-04", "semantic_world": { "n": 5, "members": [ "Pike-3b82c", "Marsh-3b82c", "Elm-3b82c", "Iris-3b82c", "Gale-3b82c" ], "outsider": "Excluded-3b82c", "scores": [ 80, 82, 78, 83, 69 ], "criterion": "latency", "set_ref": "roster-3b82c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G4.": { "policies": [], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2; claims: G4.": { "policies": [ "P2" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P5; claims: G2, G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2; claims: G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G2, G4.": { "policies": [], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P5; claims: G2.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Pike-3b82c on every execution, with probability one.", "distribution": { "Pike-3b82c": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Pike-3b82c": "1/5", "Marsh-3b82c": "1/5", "Elm-3b82c": "1/5", "Iris-3b82c": "1/5", "Gale-3b82c": "1/5" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Gale-3b82c, which has the unique lowest latency score, with probability one.", "distribution": { "Gale-3b82c": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-3b82c on every execution, with probability one.", "distribution": { "Excluded-3b82c": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Iris-3b82c and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Pike-3b82c": "1/6", "Marsh-3b82c": "1/6", "Elm-3b82c": "1/6", "Iris-3b82c": "1/3", "Gale-3b82c": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-04", "english": "A checklist item asks what the selection wording for a preprint revision actually commits to.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Flint-59676, Juniper-59676, Cobalt-59676, Nettle-59676, Pike-59676, Iris-59676, Onyx-59676, Dune-59676. Recorded queue length scores, in the same order: 28, 11, 52, 32, 33, 35, 69, 10. The reference roster-59676@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-59676 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-59676@snapshot-1) requires roster-59676@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-59676@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a preprint revision actually commits to.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Flint-59676, Juniper-59676, Cobalt-59676, Nettle-59676, Pike-59676, Iris-59676, Onyx-59676, Dune-59676. Recorded queue length scores, in the same order: 28, 11, 52, 32, 33, 35, 69, 10. The reference roster-59676@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-59676 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-59676@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-59676 on every execution, with probability one.\nP2: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Dune-59676, which has the unique lowest queue length score, with probability one.\nP4: Returns Flint-59676 on every execution, with probability one.\nP5: Assigns probability 2/9 to Onyx-59676 and 1/9 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2; claims: G4.", "Implementations: P1, P2, P3, P4, P5; claims: G4.", "Implementations: P2; claims: G3, G4.", "Implementations: P2, P3, P4, P5; claims: G4.", "Implementations: P1, P2; claims: G3, G4.", "Implementations: P2; claims: G4.", "Implementations: P2, P3, P4, P5; claims: none.", "Implementations: P1, P2, P3, P4, P5; claims: none." ], "answer": "Implementations: P2, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-03", "member_count": 8, "world_id": "rt1-world-02-04", "semantic_world": { "n": 8, "members": [ "Flint-59676", "Juniper-59676", "Cobalt-59676", "Nettle-59676", "Pike-59676", "Iris-59676", "Onyx-59676", "Dune-59676" ], "outsider": "Excluded-59676", "scores": [ 28, 11, 52, 32, 33, 35, 69, 10 ], "criterion": "queue length", "set_ref": "roster-59676@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2; claims: G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2; claims: G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2; claims: G3, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2; claims: G4.": { "policies": [ "P2" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P4, P5; claims: none.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-59676 on every execution, with probability one.", "distribution": { "Excluded-59676": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-59676": "1/8", "Juniper-59676": "1/8", "Cobalt-59676": "1/8", "Nettle-59676": "1/8", "Pike-59676": "1/8", "Iris-59676": "1/8", "Onyx-59676": "1/8", "Dune-59676": "1/8" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Dune-59676, which has the unique lowest queue length score, with probability one.", "distribution": { "Dune-59676": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Flint-59676 on every execution, with probability one.", "distribution": { "Flint-59676": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Onyx-59676 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-59676": "1/9", "Juniper-59676": "1/9", "Cobalt-59676": "1/9", "Nettle-59676": "1/9", "Pike-59676": "1/9", "Iris-59676": "1/9", "Onyx-59676": "2/9", "Dune-59676": "1/9" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-04", "english": "A checklist item asks what the selection wording for a coding evaluation slot actually commits to.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Cobalt-fc7aa, Heron-fc7aa, Kestrel-fc7aa, Basalt-fc7aa. Recorded preparation cost scores, in the same order: 95, 35, 46, 44. The reference roster-fc7aa@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fc7aa is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fc7aa@snapshot-1) requires roster-fc7aa@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-fc7aa@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a coding evaluation slot actually commits to.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Cobalt-fc7aa, Heron-fc7aa, Kestrel-fc7aa, Basalt-fc7aa. Recorded preparation cost scores, in the same order: 95, 35, 46, 44. The reference roster-fc7aa@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fc7aa is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fc7aa@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-fc7aa on every execution, with probability one.\nP2: Returns Excluded-fc7aa on every execution, with probability one.\nP3: Assigns probability 2/5 to Heron-fc7aa and 1/5 to each other listed identity, then returns one draw from that distribution.\nP4: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Heron-fc7aa, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P3, P4, P5; claims: G2, G4.", "Implementations: P1, P4; claims: G2, G4.", "Implementations: P1, P3, P4, P5; claims: G2, G4.", "Implementations: P1, P4; claims: G2, G3, G4.", "Implementations: P4; claims: G2, G3, G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P3, P4, P5; claims: G4.", "Implementations: P4; claims: G2, G4." ], "answer": "Implementations: P1, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-03", "member_count": 4, "world_id": "rt1-world-03-04", "semantic_world": { "n": 4, "members": [ "Cobalt-fc7aa", "Heron-fc7aa", "Kestrel-fc7aa", "Basalt-fc7aa" ], "outsider": "Excluded-fc7aa", "scores": [ 95, 35, 46, 44 ], "criterion": "preparation cost", "set_ref": "roster-fc7aa@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4, P5; claims: G2, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P4; claims: G2, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P4; claims: G2, G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P4; claims: G2, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P3, P4, P5; claims: G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P4; claims: G2, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Cobalt-fc7aa on every execution, with probability one.", "distribution": { "Cobalt-fc7aa": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-fc7aa on every execution, with probability one.", "distribution": { "Excluded-fc7aa": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Heron-fc7aa and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-fc7aa": "1/5", "Heron-fc7aa": "2/5", "Kestrel-fc7aa": "1/5", "Basalt-fc7aa": "1/5" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-fc7aa": "1/4", "Heron-fc7aa": "1/4", "Kestrel-fc7aa": "1/4", "Basalt-fc7aa": "1/4" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Heron-fc7aa, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Heron-fc7aa": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-04", "english": "A checklist item asks what the selection wording for a paused webhook relay actually commits to.\nThe frozen roster lists these eligible standby node identities, in recorded order: Flint-c22b6, Juniper-c22b6, Onyx-c22b6, Pike-c22b6, Lumen-c22b6, Cobalt-c22b6, Elm-c22b6. Recorded activation cost scores, in the same order: 10, 72, 2, 76, 27, 62, 54. The reference roster-c22b6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c22b6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-c22b6@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-c22b6@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-c22b6@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A checklist item asks what the selection wording for a paused webhook relay actually commits to.\nThe frozen roster lists these eligible standby node identities, in recorded order: Flint-c22b6, Juniper-c22b6, Onyx-c22b6, Pike-c22b6, Lumen-c22b6, Cobalt-c22b6, Elm-c22b6. Recorded activation cost scores, in the same order: 10, 72, 2, 76, 27, 62, 54. The reference roster-c22b6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c22b6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-c22b6@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-c22b6 on every execution, with probability one.\nP2: Assigns probability 2/8 to Cobalt-c22b6 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Flint-c22b6 on every execution, with probability one.\nP4: Returns Onyx-c22b6, which has the unique lowest activation cost score, with probability one.\nP5: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2, P3, P5; claims: G3, G4.", "Implementations: P2, P3, P4, P5; claims: G3, G4.", "Implementations: P4, P5; claims: G2, G4.", "Implementations: P5; claims: G1, G2, G4.", "Implementations: P4, P5; claims: G1, G2, G4.", "Implementations: P5; claims: G2, G4.", "Implementations: P2, P3, P4, P5; claims: G4.", "Implementations: P2, P3, P5; claims: G4." ], "answer": "Implementations: P5; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-03", "member_count": 7, "world_id": "rt1-world-04-04", "semantic_world": { "n": 7, "members": [ "Flint-c22b6", "Juniper-c22b6", "Onyx-c22b6", "Pike-c22b6", "Lumen-c22b6", "Cobalt-c22b6", "Elm-c22b6" ], "outsider": "Excluded-c22b6", "scores": [ 10, 72, 2, 76, 27, 62, 54 ], "criterion": "activation cost", "set_ref": "roster-c22b6@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P5; claims: G3, G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G3, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P4, P5; claims: G2, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G1, G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P4, P5; claims: G1, G2, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P5; claims: G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P5; claims: G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-c22b6 on every execution, with probability one.", "distribution": { "Excluded-c22b6": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Cobalt-c22b6 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-c22b6": "1/8", "Juniper-c22b6": "1/8", "Onyx-c22b6": "1/8", "Pike-c22b6": "1/8", "Lumen-c22b6": "1/8", "Cobalt-c22b6": "1/4", "Elm-c22b6": "1/8" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Flint-c22b6 on every execution, with probability one.", "distribution": { "Flint-c22b6": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Onyx-c22b6, which has the unique lowest activation cost score, with probability one.", "distribution": { "Onyx-c22b6": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-c22b6": "1/7", "Juniper-c22b6": "1/7", "Onyx-c22b6": "1/7", "Pike-c22b6": "1/7", "Lumen-c22b6": "1/7", "Cobalt-c22b6": "1/7", "Elm-c22b6": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-04", "english": "A checklist item asks what the selection wording for a release-note screenshot slot actually commits to.\nThe frozen roster lists these eligible content item identities, in recorded order: Amber-04000, Flint-04000, Juniper-04000. Recorded retrieval cost scores, in the same order: 59, 90, 55. The reference roster-04000@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-04000 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-04000@snapshot-1) requires roster-04000@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-04000@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a release-note screenshot slot actually commits to.\nThe frozen roster lists these eligible content item identities, in recorded order: Amber-04000, Flint-04000, Juniper-04000. Recorded retrieval cost scores, in the same order: 59, 90, 55. The reference roster-04000@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-04000 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-04000@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Juniper-04000, which has the unique lowest retrieval cost score, with probability one.\nP2: Assigns probability 2/4 to Flint-04000 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Amber-04000 on every execution, with probability one.\nP4: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-04000 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P4; claims: G4.", "Implementations: P2, P4; claims: G2, G4.", "Implementations: P1, P3, P4; claims: G2.", "Implementations: P4; claims: G2, G4.", "Implementations: P1, P3, P4; claims: G2, G3.", "Implementations: P2, P4; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: P1, P2, P3, P4; claims: G2, G3." ], "answer": "Implementations: P1, P2, P3, P4; claims: G2.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-03", "member_count": 3, "world_id": "rt1-world-05-04", "semantic_world": { "n": 3, "members": [ "Amber-04000", "Flint-04000", "Juniper-04000" ], "outsider": "Excluded-04000", "scores": [ 59, 90, 55 ], "criterion": "retrieval cost", "set_ref": "roster-04000@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4; claims: G4.": { "policies": [ "P4" ], "guarantees": [ "G4" ] }, "Implementations: P2, P4; claims: G2, G4.": { "policies": [ "P2", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4; claims: G2.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P4; claims: G2, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4; claims: G2, G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P4; claims: G4.": { "policies": [ "P2", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Juniper-04000, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Juniper-04000": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Flint-04000 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-04000": "1/4", "Flint-04000": "1/2", "Juniper-04000": "1/4" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Amber-04000 on every execution, with probability one.", "distribution": { "Amber-04000": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-04000": "1/3", "Flint-04000": "1/3", "Juniper-04000": "1/3" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-04000 on every execution, with probability one.", "distribution": { "Excluded-04000": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-04", "english": "A checklist item asks what the selection wording for a geocoding job actually commits to.\nThe frozen roster lists these eligible worker identities, in recorded order: Nettle-9d58c, Basalt-9d58c, Lumen-9d58c, Heron-9d58c, Pike-9d58c, Iris-9d58c. Recorded current load scores, in the same order: 30, 22, 20, 91, 83, 48. The reference roster-9d58c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9d58c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-9d58c@snapshot-1) requires roster-9d58c@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-9d58c@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a geocoding job actually commits to.\nThe frozen roster lists these eligible worker identities, in recorded order: Nettle-9d58c, Basalt-9d58c, Lumen-9d58c, Heron-9d58c, Pike-9d58c, Iris-9d58c. Recorded current load scores, in the same order: 30, 22, 20, 91, 83, 48. The reference roster-9d58c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9d58c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-9d58c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Lumen-9d58c, which has the unique lowest current load score, with probability one.\nP2: Returns Nettle-9d58c on every execution, with probability one.\nP3: Returns Excluded-9d58c on every execution, with probability one.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 2/7 to Lumen-9d58c and 1/7 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P4; claims: G1, G4.", "Implementations: P4; claims: G1.", "Implementations: none; claims: G1, G4.", "Implementations: P1, P2, P4, P5; claims: none.", "Implementations: P1, P2, P5; claims: none.", "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: none; claims: G1.", "Implementations: P1, P2, P5; claims: G1." ], "answer": "Implementations: P1, P2, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-03", "member_count": 6, "world_id": "rt1-world-06-04", "semantic_world": { "n": 6, "members": [ "Nettle-9d58c", "Basalt-9d58c", "Lumen-9d58c", "Heron-9d58c", "Pike-9d58c", "Iris-9d58c" ], "outsider": "Excluded-9d58c", "scores": [ 30, 22, 20, 91, 83, 48 ], "criterion": "current load", "set_ref": "roster-9d58c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4; claims: G1, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P4; claims: G1.": { "policies": [ "P4" ], "guarantees": [ "G1" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P5; claims: none.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: none; claims: G1.": { "policies": [], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P5; claims: G1.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Lumen-9d58c, which has the unique lowest current load score, with probability one.", "distribution": { "Lumen-9d58c": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Nettle-9d58c on every execution, with probability one.", "distribution": { "Nettle-9d58c": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-9d58c on every execution, with probability one.", "distribution": { "Excluded-9d58c": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-9d58c": "1/6", "Basalt-9d58c": "1/6", "Lumen-9d58c": "1/6", "Heron-9d58c": "1/6", "Pike-9d58c": "1/6", "Iris-9d58c": "1/6" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/7 to Lumen-9d58c and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-9d58c": "1/7", "Basalt-9d58c": "1/7", "Lumen-9d58c": "2/7", "Heron-9d58c": "1/7", "Pike-9d58c": "1/7", "Iris-9d58c": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-05", "english": "Before running anything, a planner writes down what the request about a price-quote request requires.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Onyx-fc7f7, Juniper-fc7f7, Marsh-fc7f7, Pike-fc7f7, Kestrel-fc7f7, Gale-fc7f7. Recorded latency scores, in the same order: 74, 90, 30, 28, 9, 69. The reference roster-fc7f7@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fc7f7 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fc7f7@snapshot-1) requires roster-fc7f7@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-fc7f7@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Before running anything, a planner writes down what the request about a price-quote request requires.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Onyx-fc7f7, Juniper-fc7f7, Marsh-fc7f7, Pike-fc7f7, Kestrel-fc7f7, Gale-fc7f7. Recorded latency scores, in the same order: 74, 90, 30, 28, 9, 69. The reference roster-fc7f7@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fc7f7 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fc7f7@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-fc7f7 on every execution, with probability one.\nP2: Returns Onyx-fc7f7 on every execution, with probability one.\nP3: Assigns probability 3/8 to Juniper-fc7f7 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Kestrel-fc7f7, which has the unique lowest latency score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2, P3, P4, P5; claims: G3.", "Implementations: P1, P4; claims: G3.", "Implementations: P2, P3, P4, P5; claims: none.", "Implementations: P4; claims: G1, G3.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1, P4; claims: G1, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P4; claims: G3." ], "answer": "Implementations: P2, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-04", "member_count": 6, "world_id": "rt1-world-01-05", "semantic_world": { "n": 6, "members": [ "Onyx-fc7f7", "Juniper-fc7f7", "Marsh-fc7f7", "Pike-fc7f7", "Kestrel-fc7f7", "Gale-fc7f7" ], "outsider": "Excluded-fc7f7", "scores": [ 74, 90, 30, 28, 9, 69 ], "criterion": "latency", "set_ref": "roster-fc7f7@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P4, P5; claims: G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P4; claims: G3.": { "policies": [ "P1", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3, P4, P5; claims: none.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P4; claims: G1, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P4; claims: G1, G3.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G3.": { "policies": [ "P4" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-fc7f7 on every execution, with probability one.", "distribution": { "Excluded-fc7f7": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Onyx-fc7f7 on every execution, with probability one.", "distribution": { "Onyx-fc7f7": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Juniper-fc7f7 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-fc7f7": "1/8", "Juniper-fc7f7": "3/8", "Marsh-fc7f7": "1/8", "Pike-fc7f7": "1/8", "Kestrel-fc7f7": "1/8", "Gale-fc7f7": "1/8" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-fc7f7": "1/6", "Juniper-fc7f7": "1/6", "Marsh-fc7f7": "1/6", "Pike-fc7f7": "1/6", "Kestrel-fc7f7": "1/6", "Gale-fc7f7": "1/6" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Kestrel-fc7f7, which has the unique lowest latency score, with probability one.", "distribution": { "Kestrel-fc7f7": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-05", "english": "Before running anything, a planner writes down what the request about a thesis chapter requires.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Pike-bded6, Flint-bded6. Recorded queue length scores, in the same order: 61, 18. The reference roster-bded6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-bded6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-bded6@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-bded6@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-bded6@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a thesis chapter requires.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Pike-bded6, Flint-bded6. Recorded queue length scores, in the same order: 61, 18. The reference roster-bded6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-bded6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-bded6@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Flint-bded6, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 3/4 to Flint-bded6 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Pike-bded6 on every execution, with probability one.\nP4: Returns Excluded-bded6 on every execution, with probability one.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P5; claims: G1, G4.", "Implementations: P1, P2, P5; claims: G1.", "Implementations: P1, P2, P5; claims: G1, G4.", "Implementations: P5; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P5; claims: G1.", "Implementations: P1, P2, P3, P5; claims: G1, G4.", "Implementations: P3, P5; claims: G1, G3, G4.", "Implementations: P3, P5; claims: G1, G4." ], "answer": "Implementations: P5; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-04", "member_count": 2, "world_id": "rt1-world-02-05", "semantic_world": { "n": 2, "members": [ "Pike-bded6", "Flint-bded6" ], "outsider": "Excluded-bded6", "scores": [ 61, 18 ], "criterion": "queue length", "set_ref": "roster-bded6@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P5; claims: G1.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P5; claims: G1, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P3, P5; claims: G1, G3, G4.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P3, P5; claims: G1, G4.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Flint-bded6, which has the unique lowest queue length score, with probability one.", "distribution": { "Flint-bded6": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Flint-bded6 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Pike-bded6": "1/4", "Flint-bded6": "3/4" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Pike-bded6 on every execution, with probability one.", "distribution": { "Pike-bded6": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-bded6 on every execution, with probability one.", "distribution": { "Excluded-bded6": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Pike-bded6": "1/2", "Flint-bded6": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-05", "english": "Before running anything, a planner writes down what the request about a citation evaluation slot requires.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Flint-5328e, Cobalt-5328e, Heron-5328e, Lumen-5328e, Dune-5328e. Recorded preparation cost scores, in the same order: 50, 37, 71, 70, 56. The reference roster-5328e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5328e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-5328e@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-5328e@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-5328e@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a citation evaluation slot requires.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Flint-5328e, Cobalt-5328e, Heron-5328e, Lumen-5328e, Dune-5328e. Recorded preparation cost scores, in the same order: 50, 37, 71, 70, 56. The reference roster-5328e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5328e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-5328e@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-5328e, which has the unique lowest preparation cost score, with probability one.\nP2: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Flint-5328e on every execution, with probability one.\nP4: Returns Excluded-5328e on every execution, with probability one.\nP5: Assigns probability 3/7 to Dune-5328e and 1/7 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P1, P2; claims: G2, G4.", "Implementations: P1, P2, P3, P5; claims: G2, G3.", "Implementations: P2; claims: G2, G4.", "Implementations: P2, P3, P5; claims: G2.", "Implementations: P2, P3, P5; claims: G2, G3.", "Implementations: P2; claims: G1, G2, G4.", "Implementations: P1, P2; claims: G1, G2, G4." ], "answer": "Implementations: P2; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-04", "member_count": 5, "world_id": "rt1-world-03-05", "semantic_world": { "n": 5, "members": [ "Flint-5328e", "Cobalt-5328e", "Heron-5328e", "Lumen-5328e", "Dune-5328e" ], "outsider": "Excluded-5328e", "scores": [ 50, 37, 71, 70, 56 ], "criterion": "preparation cost", "set_ref": "roster-5328e@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2; claims: G2, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2; claims: G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P5; claims: G2.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P2, P3, P5; claims: G2, G3.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2; claims: G1, G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P2; claims: G1, G2, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Cobalt-5328e, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Cobalt-5328e": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-5328e": "1/5", "Cobalt-5328e": "1/5", "Heron-5328e": "1/5", "Lumen-5328e": "1/5", "Dune-5328e": "1/5" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Flint-5328e on every execution, with probability one.", "distribution": { "Flint-5328e": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-5328e on every execution, with probability one.", "distribution": { "Excluded-5328e": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Dune-5328e and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-5328e": "1/7", "Cobalt-5328e": "1/7", "Heron-5328e": "1/7", "Lumen-5328e": "1/7", "Dune-5328e": "3/7" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-05", "english": "Before running anything, a planner writes down what the request about a degraded search shard requires.\nThe frozen roster lists these eligible standby node identities, in recorded order: Cobalt-bc81e, Iris-bc81e, Onyx-bc81e, Lumen-bc81e, Nettle-bc81e, Dune-bc81e, Juniper-bc81e, Flint-bc81e. Recorded activation cost scores, in the same order: 94, 52, 30, 10, 50, 61, 18, 65. The reference roster-bc81e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-bc81e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-bc81e@snapshot-1) requires roster-bc81e@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-bc81e@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Before running anything, a planner writes down what the request about a degraded search shard requires.\nThe frozen roster lists these eligible standby node identities, in recorded order: Cobalt-bc81e, Iris-bc81e, Onyx-bc81e, Lumen-bc81e, Nettle-bc81e, Dune-bc81e, Juniper-bc81e, Flint-bc81e. Recorded activation cost scores, in the same order: 94, 52, 30, 10, 50, 61, 18, 65. The reference roster-bc81e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-bc81e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-bc81e@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Lumen-bc81e, which has the unique lowest activation cost score, with probability one.\nP2: Returns Excluded-bc81e on every execution, with probability one.\nP3: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 3/10 to Flint-bc81e and 1/10 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Cobalt-bc81e on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P3, P4; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G3, G4.", "Implementations: P3, P4; claims: G2, G3.", "Implementations: P1, P3, P5; claims: G3.", "Implementations: P1, P3, P5; claims: G3, G4.", "Implementations: P1, P3, P4, P5; claims: G3.", "Implementations: P3; claims: G2, G3.", "Implementations: P3; claims: G2." ], "answer": "Implementations: P1, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-04", "member_count": 8, "world_id": "rt1-world-04-05", "semantic_world": { "n": 8, "members": [ "Cobalt-bc81e", "Iris-bc81e", "Onyx-bc81e", "Lumen-bc81e", "Nettle-bc81e", "Dune-bc81e", "Juniper-bc81e", "Flint-bc81e" ], "outsider": "Excluded-bc81e", "scores": [ 94, 52, 30, 10, 50, 61, 18, 65 ], "criterion": "activation cost", "set_ref": "roster-bc81e@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4; claims: G2.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P3, P4; claims: G2, G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3, P5; claims: G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P3; claims: G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P3; claims: G2.": { "policies": [ "P3" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Lumen-bc81e, which has the unique lowest activation cost score, with probability one.", "distribution": { "Lumen-bc81e": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-bc81e on every execution, with probability one.", "distribution": { "Excluded-bc81e": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-bc81e": "1/8", "Iris-bc81e": "1/8", "Onyx-bc81e": "1/8", "Lumen-bc81e": "1/8", "Nettle-bc81e": "1/8", "Dune-bc81e": "1/8", "Juniper-bc81e": "1/8", "Flint-bc81e": "1/8" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Flint-bc81e and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-bc81e": "1/10", "Iris-bc81e": "1/10", "Onyx-bc81e": "1/10", "Lumen-bc81e": "1/10", "Nettle-bc81e": "1/10", "Dune-bc81e": "1/10", "Juniper-bc81e": "1/10", "Flint-bc81e": "3/10" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Cobalt-bc81e on every execution, with probability one.", "distribution": { "Cobalt-bc81e": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-05", "english": "Before running anything, a planner writes down what the request about a podcast intro slot requires.\nThe frozen roster lists these eligible content item identities, in recorded order: Heron-41b76, Kestrel-41b76, Amber-41b76, Lumen-41b76. Recorded retrieval cost scores, in the same order: 23, 90, 32, 12. The reference roster-41b76@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-41b76 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-41b76@snapshot-1) requires roster-41b76@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-41b76@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Before running anything, a planner writes down what the request about a podcast intro slot requires.\nThe frozen roster lists these eligible content item identities, in recorded order: Heron-41b76, Kestrel-41b76, Amber-41b76, Lumen-41b76. Recorded retrieval cost scores, in the same order: 23, 90, 32, 12. The reference roster-41b76@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-41b76 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-41b76@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 3/6 to Lumen-41b76 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Heron-41b76 on every execution, with probability one.\nP3: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-41b76 on every execution, with probability one.\nP5: Returns Lumen-41b76, which has the unique lowest retrieval cost score, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P3; claims: G3, G4.", "Implementations: none; claims: G3.", "Implementations: P3; claims: G3.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: none; claims: G3, G4.", "Implementations: P1, P2, P5; claims: none.", "Implementations: P1, P2, P5; claims: G3.", "Implementations: P1, P2, P3, P5; claims: none." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-04", "member_count": 4, "world_id": "rt1-world-05-05", "semantic_world": { "n": 4, "members": [ "Heron-41b76", "Kestrel-41b76", "Amber-41b76", "Lumen-41b76" ], "outsider": "Excluded-41b76", "scores": [ 23, 90, 32, 12 ], "criterion": "retrieval cost", "set_ref": "roster-41b76@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3; claims: G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G3", "G4" ] }, "Implementations: none; claims: G3.": { "policies": [], "guarantees": [ "G3" ] }, "Implementations: P3; claims: G3.": { "policies": [ "P3" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: none; claims: G3, G4.": { "policies": [], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P5; claims: none.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P5; claims: G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Lumen-41b76 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-41b76": "1/6", "Kestrel-41b76": "1/6", "Amber-41b76": "1/6", "Lumen-41b76": "1/2" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Heron-41b76 on every execution, with probability one.", "distribution": { "Heron-41b76": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-41b76": "1/4", "Kestrel-41b76": "1/4", "Amber-41b76": "1/4", "Lumen-41b76": "1/4" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-41b76 on every execution, with probability one.", "distribution": { "Excluded-41b76": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Lumen-41b76, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Lumen-41b76": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-05", "english": "Before running anything, a planner writes down what the request about a log-rotation job requires.\nThe frozen roster lists these eligible worker identities, in recorded order: Dune-ed78c, Marsh-ed78c, Juniper-ed78c, Onyx-ed78c, Amber-ed78c, Heron-ed78c, Kestrel-ed78c. Recorded current load scores, in the same order: 68, 17, 45, 20, 56, 84, 2. The reference roster-ed78c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ed78c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ed78c@snapshot-1) requires roster-ed78c@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-ed78c@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Before running anything, a planner writes down what the request about a log-rotation job requires.\nThe frozen roster lists these eligible worker identities, in recorded order: Dune-ed78c, Marsh-ed78c, Juniper-ed78c, Onyx-ed78c, Amber-ed78c, Heron-ed78c, Kestrel-ed78c. Recorded current load scores, in the same order: 68, 17, 45, 20, 56, 84, 2. The reference roster-ed78c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ed78c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ed78c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-ed78c on every execution, with probability one.\nP2: Returns Dune-ed78c on every execution, with probability one.\nP3: Assigns probability 3/9 to Juniper-ed78c and 1/9 to each other listed identity, then returns one draw from that distribution.\nP4: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Kestrel-ed78c, which has the unique lowest current load score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P4; claims: G1, G2.", "Implementations: P4; claims: G1, G2.", "Implementations: P4; claims: G1, G2, G3.", "Implementations: P1, P4; claims: G1, G2, G3.", "Implementations: P2, P3, P4, P5; claims: G1, G2.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G2." ], "answer": "Implementations: P2, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-04", "member_count": 7, "world_id": "rt1-world-06-05", "semantic_world": { "n": 7, "members": [ "Dune-ed78c", "Marsh-ed78c", "Juniper-ed78c", "Onyx-ed78c", "Amber-ed78c", "Heron-ed78c", "Kestrel-ed78c" ], "outsider": "Excluded-ed78c", "scores": [ 68, 17, 45, 20, 56, 84, 2 ], "criterion": "current load", "set_ref": "roster-ed78c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4; claims: G1, G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4; claims: G1, G2.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4; claims: G1, G2, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P4; claims: G1, G2, G3.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G2.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-ed78c on every execution, with probability one.", "distribution": { "Excluded-ed78c": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Dune-ed78c on every execution, with probability one.", "distribution": { "Dune-ed78c": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/9 to Juniper-ed78c and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-ed78c": "1/9", "Marsh-ed78c": "1/9", "Juniper-ed78c": "1/3", "Onyx-ed78c": "1/9", "Amber-ed78c": "1/9", "Heron-ed78c": "1/9", "Kestrel-ed78c": "1/9" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-ed78c": "1/7", "Marsh-ed78c": "1/7", "Juniper-ed78c": "1/7", "Onyx-ed78c": "1/7", "Amber-ed78c": "1/7", "Heron-ed78c": "1/7", "Kestrel-ed78c": "1/7" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Kestrel-ed78c, which has the unique lowest current load score, with probability one.", "distribution": { "Kestrel-ed78c": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-06", "english": "The runbook entry for a font-subset request contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Kestrel-42d15, Flint-42d15, Nettle-42d15, Iris-42d15, Onyx-42d15, Lumen-42d15, Gale-42d15. Recorded latency scores, in the same order: 61, 94, 34, 58, 10, 92, 52. The reference roster-42d15@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-42d15 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-42d15@snapshot-1) requires roster-42d15@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-42d15@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a font-subset request contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Kestrel-42d15, Flint-42d15, Nettle-42d15, Iris-42d15, Onyx-42d15, Lumen-42d15, Gale-42d15. Recorded latency scores, in the same order: 61, 94, 34, 58, 10, 92, 52. The reference roster-42d15@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-42d15 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-42d15@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Kestrel-42d15 on every execution, with probability one.\nP2: Returns Excluded-42d15 on every execution, with probability one.\nP3: Assigns probability 4/10 to Kestrel-42d15 and 1/10 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Onyx-42d15, which has the unique lowest latency score, with probability one.\nP5: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P3, P4, P5; claims: G2, G3.", "Implementations: P3, P4, P5; claims: G2, G3.", "Implementations: P5; claims: G2, G3.", "Implementations: P1, P5; claims: G2, G3, G4.", "Implementations: P1, P3, P4, P5; claims: G2.", "Implementations: P3, P4, P5; claims: G2.", "Implementations: P5; claims: G2, G3, G4.", "Implementations: P1, P5; claims: G2, G3." ], "answer": "Implementations: P1, P3, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-05", "member_count": 7, "world_id": "rt1-world-01-06", "semantic_world": { "n": 7, "members": [ "Kestrel-42d15", "Flint-42d15", "Nettle-42d15", "Iris-42d15", "Onyx-42d15", "Lumen-42d15", "Gale-42d15" ], "outsider": "Excluded-42d15", "scores": [ 61, 94, 34, 58, 10, 92, 52 ], "criterion": "latency", "set_ref": "roster-42d15@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P3, P4, P5; claims: G2, G3.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P5; claims: G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P5; claims: G2, G3, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P3, P4, P5; claims: G2.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P5; claims: G2, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P5; claims: G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Kestrel-42d15 on every execution, with probability one.", "distribution": { "Kestrel-42d15": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-42d15 on every execution, with probability one.", "distribution": { "Excluded-42d15": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Kestrel-42d15 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-42d15": "2/5", "Flint-42d15": "1/10", "Nettle-42d15": "1/10", "Iris-42d15": "1/10", "Onyx-42d15": "1/10", "Lumen-42d15": "1/10", "Gale-42d15": "1/10" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Onyx-42d15, which has the unique lowest latency score, with probability one.", "distribution": { "Onyx-42d15": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-42d15": "1/7", "Flint-42d15": "1/7", "Nettle-42d15": "1/7", "Iris-42d15": "1/7", "Onyx-42d15": "1/7", "Lumen-42d15": "1/7", "Gale-42d15": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-06", "english": "The runbook entry for a benchmark submission contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Cobalt-c3750, Dune-c3750, Onyx-c3750. Recorded queue length scores, in the same order: 34, 8, 9. The reference roster-c3750@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c3750 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-c3750@snapshot-1) requires roster-c3750@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-c3750@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a benchmark submission contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Cobalt-c3750, Dune-c3750, Onyx-c3750. Recorded queue length scores, in the same order: 34, 8, 9. The reference roster-c3750@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c3750 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-c3750@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/6 to Onyx-c3750 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Dune-c3750, which has the unique lowest queue length score, with probability one.\nP3: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Cobalt-c3750 on every execution, with probability one.\nP5: Returns Excluded-c3750 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P2, P3; claims: G1, G2.", "Implementations: P1, P3, P4; claims: G2.", "Implementations: P3; claims: G1, G2.", "Implementations: P2, P3; claims: G1, G2, G3.", "Implementations: P1, P3, P4; claims: G2, G4.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: P3; claims: G1, G2, G3." ], "answer": "Implementations: P1, P2, P3, P4; claims: G2.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-05", "member_count": 3, "world_id": "rt1-world-02-06", "semantic_world": { "n": 3, "members": [ "Cobalt-c3750", "Dune-c3750", "Onyx-c3750" ], "outsider": "Excluded-c3750", "scores": [ 34, 8, 9 ], "criterion": "queue length", "set_ref": "roster-c3750@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3; claims: G1, G2.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3, P4; claims: G2.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3; claims: G1, G2, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Onyx-c3750 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-c3750": "1/6", "Dune-c3750": "1/6", "Onyx-c3750": "2/3" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Dune-c3750, which has the unique lowest queue length score, with probability one.", "distribution": { "Dune-c3750": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-c3750": "1/3", "Dune-c3750": "1/3", "Onyx-c3750": "1/3" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Cobalt-c3750 on every execution, with probability one.", "distribution": { "Cobalt-c3750": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-c3750 on every execution, with probability one.", "distribution": { "Excluded-c3750": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-06", "english": "The runbook entry for a dialogue evaluation slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Elm-f1fb6, Kestrel-f1fb6, Cobalt-f1fb6, Amber-f1fb6, Lumen-f1fb6, Dune-f1fb6. Recorded preparation cost scores, in the same order: 40, 11, 69, 26, 44, 81. The reference roster-f1fb6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f1fb6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-f1fb6@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-f1fb6@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-f1fb6@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The runbook entry for a dialogue evaluation slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Elm-f1fb6, Kestrel-f1fb6, Cobalt-f1fb6, Amber-f1fb6, Lumen-f1fb6, Dune-f1fb6. Recorded preparation cost scores, in the same order: 40, 11, 69, 26, 44, 81. The reference roster-f1fb6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f1fb6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-f1fb6@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Kestrel-f1fb6, which has the unique lowest preparation cost score, with probability one.\nP2: Returns Excluded-f1fb6 on every execution, with probability one.\nP3: Returns Elm-f1fb6 on every execution, with probability one.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 4/9 to Dune-f1fb6 and 1/9 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P4; claims: G3.", "Implementations: P4; claims: G3, G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P1, P3, P4; claims: G4.", "Implementations: P1, P3, P4; claims: G2, G4.", "Implementations: P4, P5; claims: G3, G4.", "Implementations: P4, P5; claims: G3.", "Implementations: P1, P3, P4, P5; claims: G2, G4." ], "answer": "Implementations: P4; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-05", "member_count": 6, "world_id": "rt1-world-03-06", "semantic_world": { "n": 6, "members": [ "Elm-f1fb6", "Kestrel-f1fb6", "Cobalt-f1fb6", "Amber-f1fb6", "Lumen-f1fb6", "Dune-f1fb6" ], "outsider": "Excluded-f1fb6", "scores": [ 40, 11, 69, 26, 44, 81 ], "criterion": "preparation cost", "set_ref": "roster-f1fb6@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4; claims: G3.": { "policies": [ "P4" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4; claims: G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P4, P5; claims: G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P4, P5; claims: G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Kestrel-f1fb6, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Kestrel-f1fb6": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-f1fb6 on every execution, with probability one.", "distribution": { "Excluded-f1fb6": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Elm-f1fb6 on every execution, with probability one.", "distribution": { "Elm-f1fb6": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-f1fb6": "1/6", "Kestrel-f1fb6": "1/6", "Cobalt-f1fb6": "1/6", "Amber-f1fb6": "1/6", "Lumen-f1fb6": "1/6", "Dune-f1fb6": "1/6" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Dune-f1fb6 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-f1fb6": "1/9", "Kestrel-f1fb6": "1/9", "Cobalt-f1fb6": "1/9", "Amber-f1fb6": "1/9", "Lumen-f1fb6": "1/9", "Dune-f1fb6": "4/9" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-06", "english": "The runbook entry for a halted backup agent contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible standby node identities, in recorded order: Lumen-f173c, Pike-f173c. Recorded activation cost scores, in the same order: 25, 19. The reference roster-f173c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f173c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f173c@snapshot-1) requires roster-f173c@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-f173c@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a halted backup agent contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible standby node identities, in recorded order: Lumen-f173c, Pike-f173c. Recorded activation cost scores, in the same order: 25, 19. The reference roster-f173c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f173c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f173c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/5 to Pike-f173c and 1/5 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Pike-f173c, which has the unique lowest activation cost score, with probability one.\nP3: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-f173c on every execution, with probability one.\nP5: Returns Lumen-f173c on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P5; claims: none.", "Implementations: none; claims: G3, G4.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P3; claims: G3, G4.", "Implementations: none; claims: G3.", "Implementations: P3; claims: G3.", "Implementations: P1, P2, P5; claims: G3.", "Implementations: P1, P2, P3, P5; claims: G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-05", "member_count": 2, "world_id": "rt1-world-04-06", "semantic_world": { "n": 2, "members": [ "Lumen-f173c", "Pike-f173c" ], "outsider": "Excluded-f173c", "scores": [ 25, 19 ], "criterion": "activation cost", "set_ref": "roster-f173c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P5; claims: none.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [] }, "Implementations: none; claims: G3, G4.": { "policies": [], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P3; claims: G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G3", "G4" ] }, "Implementations: none; claims: G3.": { "policies": [], "guarantees": [ "G3" ] }, "Implementations: P3; claims: G3.": { "policies": [ "P3" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P5; claims: G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Pike-f173c and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Lumen-f173c": "1/5", "Pike-f173c": "4/5" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Pike-f173c, which has the unique lowest activation cost score, with probability one.", "distribution": { "Pike-f173c": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Lumen-f173c": "1/2", "Pike-f173c": "1/2" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-f173c on every execution, with probability one.", "distribution": { "Excluded-f173c": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Lumen-f173c on every execution, with probability one.", "distribution": { "Lumen-f173c": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-06", "english": "The runbook entry for a museum label slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible content item identities, in recorded order: Pike-907ad, Marsh-907ad, Juniper-907ad, Gale-907ad, Nettle-907ad. Recorded retrieval cost scores, in the same order: 82, 10, 52, 62, 83. The reference roster-907ad@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-907ad is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-907ad@snapshot-1) requires roster-907ad@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-907ad@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a museum label slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible content item identities, in recorded order: Pike-907ad, Marsh-907ad, Juniper-907ad, Gale-907ad, Nettle-907ad. Recorded retrieval cost scores, in the same order: 82, 10, 52, 62, 83. The reference roster-907ad@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-907ad is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-907ad@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Pike-907ad on every execution, with probability one.\nP3: Returns Marsh-907ad, which has the unique lowest retrieval cost score, with probability one.\nP4: Returns Excluded-907ad on every execution, with probability one.\nP5: Assigns probability 4/8 to Pike-907ad and 1/8 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1; claims: G1, G3, G4.", "Implementations: P1, P4; claims: G1, G3, G4.", "Implementations: P1; claims: G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G3, G4.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: P1, P4; claims: G3, G4.", "Implementations: P1, P2, P3, P5; claims: G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-05", "member_count": 5, "world_id": "rt1-world-05-06", "semantic_world": { "n": 5, "members": [ "Pike-907ad", "Marsh-907ad", "Juniper-907ad", "Gale-907ad", "Nettle-907ad" ], "outsider": "Excluded-907ad", "scores": [ 82, 10, 52, 62, 83 ], "criterion": "retrieval cost", "set_ref": "roster-907ad@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G1, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P4; claims: G1, G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1; claims: G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P4; claims: G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Pike-907ad": "1/5", "Marsh-907ad": "1/5", "Juniper-907ad": "1/5", "Gale-907ad": "1/5", "Nettle-907ad": "1/5" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Pike-907ad on every execution, with probability one.", "distribution": { "Pike-907ad": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Marsh-907ad, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Marsh-907ad": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-907ad on every execution, with probability one.", "distribution": { "Excluded-907ad": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Pike-907ad and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Pike-907ad": "1/2", "Marsh-907ad": "1/8", "Juniper-907ad": "1/8", "Gale-907ad": "1/8", "Nettle-907ad": "1/8" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-06", "english": "The runbook entry for a dependency-scan job contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible worker identities, in recorded order: Nettle-1cc2b, Elm-1cc2b, Lumen-1cc2b, Basalt-1cc2b, Juniper-1cc2b, Pike-1cc2b, Gale-1cc2b, Kestrel-1cc2b. Recorded current load scores, in the same order: 37, 57, 43, 17, 65, 94, 81, 59. The reference roster-1cc2b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1cc2b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-1cc2b@snapshot-1) requires roster-1cc2b@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-1cc2b@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a dependency-scan job contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible worker identities, in recorded order: Nettle-1cc2b, Elm-1cc2b, Lumen-1cc2b, Basalt-1cc2b, Juniper-1cc2b, Pike-1cc2b, Gale-1cc2b, Kestrel-1cc2b. Recorded current load scores, in the same order: 37, 57, 43, 17, 65, 94, 81, 59. The reference roster-1cc2b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1cc2b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-1cc2b@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 4/11 to Elm-1cc2b and 1/11 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Nettle-1cc2b on every execution, with probability one.\nP4: Returns Excluded-1cc2b on every execution, with probability one.\nP5: Returns Basalt-1cc2b, which has the unique lowest current load score, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P5; claims: G3.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P1, P3; claims: G2, G3, G4.", "Implementations: P1; claims: G2, G3, G4.", "Implementations: P1; claims: G2, G3.", "Implementations: P1, P3; claims: G2, G3.", "Implementations: P1, P2, P5; claims: G1, G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-05", "member_count": 8, "world_id": "rt1-world-06-06", "semantic_world": { "n": 8, "members": [ "Nettle-1cc2b", "Elm-1cc2b", "Lumen-1cc2b", "Basalt-1cc2b", "Juniper-1cc2b", "Pike-1cc2b", "Gale-1cc2b", "Kestrel-1cc2b" ], "outsider": "Excluded-1cc2b", "scores": [ 37, 57, 43, 17, 65, 94, 81, 59 ], "criterion": "current load", "set_ref": "roster-1cc2b@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P5; claims: G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P3; claims: G2, G3, G4.": { "policies": [ "P1", "P3" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1; claims: G2, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1; claims: G2, G3.": { "policies": [ "P1" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3; claims: G2, G3.": { "policies": [ "P1", "P3" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-1cc2b": "1/8", "Elm-1cc2b": "1/8", "Lumen-1cc2b": "1/8", "Basalt-1cc2b": "1/8", "Juniper-1cc2b": "1/8", "Pike-1cc2b": "1/8", "Gale-1cc2b": "1/8", "Kestrel-1cc2b": "1/8" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/11 to Elm-1cc2b and 1/11 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-1cc2b": "1/11", "Elm-1cc2b": "4/11", "Lumen-1cc2b": "1/11", "Basalt-1cc2b": "1/11", "Juniper-1cc2b": "1/11", "Pike-1cc2b": "1/11", "Gale-1cc2b": "1/11", "Kestrel-1cc2b": "1/11" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Nettle-1cc2b on every execution, with probability one.", "distribution": { "Nettle-1cc2b": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-1cc2b on every execution, with probability one.", "distribution": { "Excluded-1cc2b": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Basalt-1cc2b, which has the unique lowest current load score, with probability one.", "distribution": { "Basalt-1cc2b": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-07", "english": "A reviewer is asked which procedures would comply with the request for a receipt-rendering call.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Onyx-441ae, Amber-441ae, Heron-441ae, Lumen-441ae, Juniper-441ae, Flint-441ae, Basalt-441ae, Pike-441ae. Recorded latency scores, in the same order: 26, 45, 96, 40, 35, 15, 59, 10. The reference roster-441ae@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-441ae is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-441ae@snapshot-1) requires roster-441ae@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-441ae@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A reviewer is asked which procedures would comply with the request for a receipt-rendering call.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Onyx-441ae, Amber-441ae, Heron-441ae, Lumen-441ae, Juniper-441ae, Flint-441ae, Basalt-441ae, Pike-441ae. Recorded latency scores, in the same order: 26, 45, 96, 40, 35, 15, 59, 10. The reference roster-441ae@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-441ae is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-441ae@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Pike-441ae, which has the unique lowest latency score, with probability one.\nP3: Returns Onyx-441ae on every execution, with probability one.\nP4: Assigns probability 2/9 to Juniper-441ae and 1/9 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-441ae on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1; claims: G2, G3, G4.", "Implementations: P1, P3, P4; claims: G1, G3.", "Implementations: P1, P2; claims: G2, G3.", "Implementations: P1, P2, P3, P4; claims: G1, G3.", "Implementations: P1; claims: G2, G3.", "Implementations: P1, P3, P4; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P1, P2; claims: G2, G3, G4." ], "answer": "Implementations: P1, P2, P3, P4; claims: G3.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-06", "member_count": 8, "world_id": "rt1-world-01-07", "semantic_world": { "n": 8, "members": [ "Onyx-441ae", "Amber-441ae", "Heron-441ae", "Lumen-441ae", "Juniper-441ae", "Flint-441ae", "Basalt-441ae", "Pike-441ae" ], "outsider": "Excluded-441ae", "scores": [ 26, 45, 96, 40, 35, 15, 59, 10 ], "criterion": "latency", "set_ref": "roster-441ae@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G2, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P3, P4; claims: G1, G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2; claims: G2, G3.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1; claims: G2, G3.": { "policies": [ "P1" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3, P4; claims: G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2; claims: G2, G3, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2", "G3", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-441ae": "1/8", "Amber-441ae": "1/8", "Heron-441ae": "1/8", "Lumen-441ae": "1/8", "Juniper-441ae": "1/8", "Flint-441ae": "1/8", "Basalt-441ae": "1/8", "Pike-441ae": "1/8" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Pike-441ae, which has the unique lowest latency score, with probability one.", "distribution": { "Pike-441ae": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Onyx-441ae on every execution, with probability one.", "distribution": { "Onyx-441ae": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Juniper-441ae and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-441ae": "1/9", "Amber-441ae": "1/9", "Heron-441ae": "1/9", "Lumen-441ae": "1/9", "Juniper-441ae": "2/9", "Flint-441ae": "1/9", "Basalt-441ae": "1/9", "Pike-441ae": "1/9" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-441ae on every execution, with probability one.", "distribution": { "Excluded-441ae": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-07", "english": "A reviewer is asked which procedures would comply with the request for a grant renewal.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Juniper-54f71, Nettle-54f71, Pike-54f71, Marsh-54f71. Recorded queue length scores, in the same order: 45, 89, 53, 39. The reference roster-54f71@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-54f71 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-54f71@snapshot-1) requires roster-54f71@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-54f71@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A reviewer is asked which procedures would comply with the request for a grant renewal.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Juniper-54f71, Nettle-54f71, Pike-54f71, Marsh-54f71. Recorded queue length scores, in the same order: 45, 89, 53, 39. The reference roster-54f71@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-54f71 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-54f71@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-54f71 on every execution, with probability one.\nP2: Returns Marsh-54f71, which has the unique lowest queue length score, with probability one.\nP3: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 2/5 to Nettle-54f71 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Juniper-54f71 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P2, P3, P5; claims: G1.", "Implementations: P2, P3, P5; claims: G1, G3.", "Implementations: P3, P4; claims: G2.", "Implementations: P3; claims: G1, G2.", "Implementations: P3, P4; claims: G1, G2.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P3; claims: G2.", "Implementations: P2, P3, P4, P5; claims: G1, G3." ], "answer": "Implementations: P2, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-06", "member_count": 4, "world_id": "rt1-world-02-07", "semantic_world": { "n": 4, "members": [ "Juniper-54f71", "Nettle-54f71", "Pike-54f71", "Marsh-54f71" ], "outsider": "Excluded-54f71", "scores": [ 45, 89, 53, 39 ], "criterion": "queue length", "set_ref": "roster-54f71@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P5; claims: G1.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3, P4; claims: G2.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P4; claims: G1, G2.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P3; claims: G2.": { "policies": [ "P3" ], "guarantees": [ "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-54f71 on every execution, with probability one.", "distribution": { "Excluded-54f71": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Marsh-54f71, which has the unique lowest queue length score, with probability one.", "distribution": { "Marsh-54f71": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-54f71": "1/4", "Nettle-54f71": "1/4", "Pike-54f71": "1/4", "Marsh-54f71": "1/4" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Nettle-54f71 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-54f71": "1/5", "Nettle-54f71": "2/5", "Pike-54f71": "1/5", "Marsh-54f71": "1/5" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Juniper-54f71 on every execution, with probability one.", "distribution": { "Juniper-54f71": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-07", "english": "A reviewer is asked which procedures would comply with the request for a summarisation evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Heron-71811, Gale-71811, Basalt-71811, Nettle-71811, Pike-71811, Elm-71811, Marsh-71811. Recorded preparation cost scores, in the same order: 54, 66, 7, 84, 48, 93, 64. The reference roster-71811@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-71811 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-71811@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-71811@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-71811@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A reviewer is asked which procedures would comply with the request for a summarisation evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Heron-71811, Gale-71811, Basalt-71811, Nettle-71811, Pike-71811, Elm-71811, Marsh-71811. Recorded preparation cost scores, in the same order: 54, 66, 7, 84, 48, 93, 64. The reference roster-71811@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-71811 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-71811@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-71811 on every execution, with probability one.\nP2: Assigns probability 2/8 to Gale-71811 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-71811 on every execution, with probability one.\nP5: Returns Basalt-71811, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P5; claims: G4.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P3; claims: G4.", "Implementations: P3; claims: G1, G4.", "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: none; claims: G1, G4.", "Implementations: P1, P2, P5; claims: none.", "Implementations: none; claims: G4." ], "answer": "Implementations: P3; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-06", "member_count": 7, "world_id": "rt1-world-03-07", "semantic_world": { "n": 7, "members": [ "Heron-71811", "Gale-71811", "Basalt-71811", "Nettle-71811", "Pike-71811", "Elm-71811", "Marsh-71811" ], "outsider": "Excluded-71811", "scores": [ 54, 66, 7, 84, 48, 93, 64 ], "criterion": "preparation cost", "set_ref": "roster-71811@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P5; claims: G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P3; claims: G4.": { "policies": [ "P3" ], "guarantees": [ "G4" ] }, "Implementations: P3; claims: G1, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P5; claims: none.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [] }, "Implementations: none; claims: G4.": { "policies": [], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Heron-71811 on every execution, with probability one.", "distribution": { "Heron-71811": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Gale-71811 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-71811": "1/8", "Gale-71811": "1/4", "Basalt-71811": "1/8", "Nettle-71811": "1/8", "Pike-71811": "1/8", "Elm-71811": "1/8", "Marsh-71811": "1/8" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-71811": "1/7", "Gale-71811": "1/7", "Basalt-71811": "1/7", "Nettle-71811": "1/7", "Pike-71811": "1/7", "Elm-71811": "1/7", "Marsh-71811": "1/7" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-71811 on every execution, with probability one.", "distribution": { "Excluded-71811": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Basalt-71811, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Basalt-71811": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-07", "english": "A reviewer is asked which procedures would comply with the request for a stalled queue consumer.\nThe frozen roster lists these eligible standby node identities, in recorded order: Kestrel-cfa92, Dune-cfa92, Basalt-cfa92. Recorded activation cost scores, in the same order: 91, 48, 4. The reference roster-cfa92@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-cfa92 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-cfa92@snapshot-1) requires roster-cfa92@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-cfa92@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A reviewer is asked which procedures would comply with the request for a stalled queue consumer.\nThe frozen roster lists these eligible standby node identities, in recorded order: Kestrel-cfa92, Dune-cfa92, Basalt-cfa92. Recorded activation cost scores, in the same order: 91, 48, 4. The reference roster-cfa92@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-cfa92 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-cfa92@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/4 to Dune-cfa92 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Excluded-cfa92 on every execution, with probability one.\nP3: Returns Basalt-cfa92, which has the unique lowest activation cost score, with probability one.\nP4: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Kestrel-cfa92 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.", "Implementations: P4; claims: G1, G3.", "Implementations: P4; claims: G1, G3, G4.", "Implementations: P1, P3, P4, P5; claims: G1.", "Implementations: P2, P4; claims: G1, G3.", "Implementations: P1, P3, P4, P5; claims: G1, G3.", "Implementations: P2, P4; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G1." ], "answer": "Implementations: P1, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-06", "member_count": 3, "world_id": "rt1-world-04-07", "semantic_world": { "n": 3, "members": [ "Kestrel-cfa92", "Dune-cfa92", "Basalt-cfa92" ], "outsider": "Excluded-cfa92", "scores": [ 91, 48, 4 ], "criterion": "activation cost", "set_ref": "roster-cfa92@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G1, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G1, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4; claims: G1, G3.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P4; claims: G1, G3, G4.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Dune-cfa92 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-cfa92": "1/4", "Dune-cfa92": "1/2", "Basalt-cfa92": "1/4" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-cfa92 on every execution, with probability one.", "distribution": { "Excluded-cfa92": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Basalt-cfa92, which has the unique lowest activation cost score, with probability one.", "distribution": { "Basalt-cfa92": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-cfa92": "1/3", "Dune-cfa92": "1/3", "Basalt-cfa92": "1/3" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Kestrel-cfa92 on every execution, with probability one.", "distribution": { "Kestrel-cfa92": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-07", "english": "A reviewer is asked which procedures would comply with the request for a landing-page banner slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Heron-fface, Gale-fface, Flint-fface, Onyx-fface, Lumen-fface, Elm-fface. Recorded retrieval cost scores, in the same order: 57, 60, 92, 46, 39, 73. The reference roster-fface@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fface is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fface@snapshot-1) requires roster-fface@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-fface@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A reviewer is asked which procedures would comply with the request for a landing-page banner slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Heron-fface, Gale-fface, Flint-fface, Onyx-fface, Lumen-fface, Elm-fface. Recorded retrieval cost scores, in the same order: 57, 60, 92, 46, 39, 73. The reference roster-fface@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fface is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fface@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/7 to Elm-fface and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Lumen-fface, which has the unique lowest retrieval cost score, with probability one.\nP3: Returns Excluded-fface on every execution, with probability one.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Heron-fface on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P4, P5; claims: G1, G3, G4.", "Implementations: P1, P2, P4, P5; claims: G2, G4.", "Implementations: P4; claims: G1, G3, G4.", "Implementations: P4; claims: G3, G4.", "Implementations: P4, P5; claims: G3, G4.", "Implementations: P1, P2, P4; claims: G2, G4.", "Implementations: P1, P2, P4; claims: G4.", "Implementations: P1, P2, P4, P5; claims: G4." ], "answer": "Implementations: P1, P2, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-06", "member_count": 6, "world_id": "rt1-world-05-07", "semantic_world": { "n": 6, "members": [ "Heron-fface", "Gale-fface", "Flint-fface", "Onyx-fface", "Lumen-fface", "Elm-fface" ], "outsider": "Excluded-fface", "scores": [ 57, 60, 92, 46, 39, 73 ], "criterion": "retrieval cost", "set_ref": "roster-fface@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4, P5; claims: G1, G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P4; claims: G1, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P4, P5; claims: G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P4; claims: G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/7 to Elm-fface and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-fface": "1/7", "Gale-fface": "1/7", "Flint-fface": "1/7", "Onyx-fface": "1/7", "Lumen-fface": "1/7", "Elm-fface": "2/7" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Lumen-fface, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Lumen-fface": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-fface on every execution, with probability one.", "distribution": { "Excluded-fface": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-fface": "1/6", "Gale-fface": "1/6", "Flint-fface": "1/6", "Onyx-fface": "1/6", "Lumen-fface": "1/6", "Elm-fface": "1/6" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Heron-fface on every execution, with probability one.", "distribution": { "Heron-fface": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-07", "english": "A reviewer is asked which procedures would comply with the request for a thumbnail-generation job.\nThe frozen roster lists these eligible worker identities, in recorded order: Marsh-564b7, Nettle-564b7. Recorded current load scores, in the same order: 42, 4. The reference roster-564b7@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-564b7 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-564b7@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-564b7@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-564b7@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A reviewer is asked which procedures would comply with the request for a thumbnail-generation job.\nThe frozen roster lists these eligible worker identities, in recorded order: Marsh-564b7, Nettle-564b7. Recorded current load scores, in the same order: 42, 4. The reference roster-564b7@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-564b7 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-564b7@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Nettle-564b7, which has the unique lowest current load score, with probability one.\nP2: Returns Excluded-564b7 on every execution, with probability one.\nP3: Returns Marsh-564b7 on every execution, with probability one.\nP4: Assigns probability 2/3 to Marsh-564b7 and 1/3 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P5; claims: G3.", "Implementations: P5; claims: G2, G3.", "Implementations: P3, P4, P5; claims: G1, G2.", "Implementations: P1, P3, P4, P5; claims: G2.", "Implementations: P5; claims: G3.", "Implementations: P3, P4, P5; claims: G2.", "Implementations: P1, P5; claims: G2, G3.", "Implementations: P1, P3, P4, P5; claims: G1, G2." ], "answer": "Implementations: P5; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-06", "member_count": 2, "world_id": "rt1-world-06-07", "semantic_world": { "n": 2, "members": [ "Marsh-564b7", "Nettle-564b7" ], "outsider": "Excluded-564b7", "scores": [ 42, 4 ], "criterion": "current load", "set_ref": "roster-564b7@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P5; claims: G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P5; claims: G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P3, P4, P5; claims: G1, G2.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P5; claims: G3.": { "policies": [ "P5" ], "guarantees": [ "G3" ] }, "Implementations: P3, P4, P5; claims: G2.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P5; claims: G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G1, G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Nettle-564b7, which has the unique lowest current load score, with probability one.", "distribution": { "Nettle-564b7": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-564b7 on every execution, with probability one.", "distribution": { "Excluded-564b7": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Marsh-564b7 on every execution, with probability one.", "distribution": { "Marsh-564b7": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Marsh-564b7 and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-564b7": "2/3", "Nettle-564b7": "1/3" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-564b7": "1/2", "Nettle-564b7": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-08", "english": "A ticket about a postcode lookup is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Dune-b5d27, Iris-b5d27. Recorded latency scores, in the same order: 51, 34. The reference roster-b5d27@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b5d27 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b5d27@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-b5d27@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-b5d27@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about a postcode lookup is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Dune-b5d27, Iris-b5d27. Recorded latency scores, in the same order: 51, 34. The reference roster-b5d27@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b5d27 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b5d27@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-b5d27 on every execution, with probability one.\nP3: Assigns probability 3/4 to Dune-b5d27 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Iris-b5d27, which has the unique lowest latency score, with probability one.\nP5: Returns Dune-b5d27 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P3; claims: G2, G3.", "Implementations: P1, P3, P4, P5; claims: G3.", "Implementations: P1, P4, P5; claims: G3.", "Implementations: P1; claims: G2.", "Implementations: P1, P3; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G3, G4.", "Implementations: P1, P4, P5; claims: G3, G4.", "Implementations: P1; claims: G2, G3." ], "answer": "Implementations: P1; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-07", "member_count": 2, "world_id": "rt1-world-01-08", "semantic_world": { "n": 2, "members": [ "Dune-b5d27", "Iris-b5d27" ], "outsider": "Excluded-b5d27", "scores": [ 51, 34 ], "criterion": "latency", "set_ref": "roster-b5d27@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3; claims: G2, G3.": { "policies": [ "P1", "P3" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P4, P5; claims: G3.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3; claims: G2.": { "policies": [ "P1", "P3" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1; claims: G2, G3.": { "policies": [ "P1" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-b5d27": "1/2", "Iris-b5d27": "1/2" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-b5d27 on every execution, with probability one.", "distribution": { "Excluded-b5d27": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Dune-b5d27 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-b5d27": "3/4", "Iris-b5d27": "1/4" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Iris-b5d27, which has the unique lowest latency score, with probability one.", "distribution": { "Iris-b5d27": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Dune-b5d27 on every execution, with probability one.", "distribution": { "Dune-b5d27": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-08", "english": "A ticket about a code-audit report is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-68107, Juniper-68107, Marsh-68107, Gale-68107, Flint-68107. Recorded queue length scores, in the same order: 83, 20, 72, 93, 64. The reference roster-68107@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-68107 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-68107@snapshot-1) requires roster-68107@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-68107@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A ticket about a code-audit report is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-68107, Juniper-68107, Marsh-68107, Gale-68107, Flint-68107. Recorded queue length scores, in the same order: 83, 20, 72, 93, 64. The reference roster-68107@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-68107 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-68107@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Basalt-68107 on every execution, with probability one.\nP3: Returns Juniper-68107, which has the unique lowest queue length score, with probability one.\nP4: Assigns probability 3/7 to Gale-68107 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-68107 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1; claims: G2, G4.", "Implementations: P1, P2, P3, P4; claims: none.", "Implementations: P2, P3, P4; claims: none.", "Implementations: P2, P3, P4; claims: G2.", "Implementations: none; claims: G2, G4.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: none; claims: G2.", "Implementations: P1; claims: G2." ], "answer": "Implementations: P1, P2, P3, P4; claims: G2.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-07", "member_count": 5, "world_id": "rt1-world-02-08", "semantic_world": { "n": 5, "members": [ "Basalt-68107", "Juniper-68107", "Marsh-68107", "Gale-68107", "Flint-68107" ], "outsider": "Excluded-68107", "scores": [ 83, 20, 72, 93, 64 ], "criterion": "queue length", "set_ref": "roster-68107@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: none.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [] }, "Implementations: P2, P3, P4; claims: none.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [] }, "Implementations: P2, P3, P4; claims: G2.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: none; claims: G2, G4.": { "policies": [], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: none; claims: G2.": { "policies": [], "guarantees": [ "G2" ] }, "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-68107": "1/5", "Juniper-68107": "1/5", "Marsh-68107": "1/5", "Gale-68107": "1/5", "Flint-68107": "1/5" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Basalt-68107 on every execution, with probability one.", "distribution": { "Basalt-68107": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Juniper-68107, which has the unique lowest queue length score, with probability one.", "distribution": { "Juniper-68107": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Gale-68107 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-68107": "1/7", "Juniper-68107": "1/7", "Marsh-68107": "1/7", "Gale-68107": "3/7", "Flint-68107": "1/7" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-68107 on every execution, with probability one.", "distribution": { "Excluded-68107": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-08", "english": "A ticket about a tool-use evaluation slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-41e83, Juniper-41e83, Heron-41e83, Kestrel-41e83, Onyx-41e83, Elm-41e83, Basalt-41e83, Iris-41e83. Recorded preparation cost scores, in the same order: 37, 61, 49, 47, 28, 76, 80, 20. The reference roster-41e83@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-41e83 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-41e83@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-41e83@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-41e83@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about a tool-use evaluation slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-41e83, Juniper-41e83, Heron-41e83, Kestrel-41e83, Onyx-41e83, Elm-41e83, Basalt-41e83, Iris-41e83. Recorded preparation cost scores, in the same order: 37, 61, 49, 47, 28, 76, 80, 20. The reference roster-41e83@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-41e83 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-41e83@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Iris-41e83, which has the unique lowest preparation cost score, with probability one.\nP2: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP3: Assigns probability 3/10 to Gale-41e83 and 1/10 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Gale-41e83 on every execution, with probability one.\nP5: Returns Excluded-41e83 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G3, G4.", "Implementations: P2, P5; claims: G1, G3, G4.", "Implementations: P2; claims: G1, G3, G4.", "Implementations: P2; claims: G3, G4.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P2, P5; claims: G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G3, G4." ], "answer": "Implementations: P2; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-07", "member_count": 8, "world_id": "rt1-world-03-08", "semantic_world": { "n": 8, "members": [ "Gale-41e83", "Juniper-41e83", "Heron-41e83", "Kestrel-41e83", "Onyx-41e83", "Elm-41e83", "Basalt-41e83", "Iris-41e83" ], "outsider": "Excluded-41e83", "scores": [ 37, 61, 49, 47, 28, 76, 80, 20 ], "criterion": "preparation cost", "set_ref": "roster-41e83@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P5; claims: G1, G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2; claims: G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P5; claims: G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Iris-41e83, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Iris-41e83": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-41e83": "1/8", "Juniper-41e83": "1/8", "Heron-41e83": "1/8", "Kestrel-41e83": "1/8", "Onyx-41e83": "1/8", "Elm-41e83": "1/8", "Basalt-41e83": "1/8", "Iris-41e83": "1/8" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Gale-41e83 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-41e83": "3/10", "Juniper-41e83": "1/10", "Heron-41e83": "1/10", "Kestrel-41e83": "1/10", "Onyx-41e83": "1/10", "Elm-41e83": "1/10", "Basalt-41e83": "1/10", "Iris-41e83": "1/10" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Gale-41e83 on every execution, with probability one.", "distribution": { "Gale-41e83": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-41e83 on every execution, with probability one.", "distribution": { "Excluded-41e83": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-08", "english": "A ticket about an unreachable metrics store is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible standby node identities, in recorded order: Juniper-4950d, Iris-4950d, Pike-4950d, Heron-4950d. Recorded activation cost scores, in the same order: 72, 53, 67, 84. The reference roster-4950d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-4950d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-4950d@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-4950d@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-4950d@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about an unreachable metrics store is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible standby node identities, in recorded order: Juniper-4950d, Iris-4950d, Pike-4950d, Heron-4950d. Recorded activation cost scores, in the same order: 72, 53, 67, 84. The reference roster-4950d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-4950d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-4950d@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-4950d on every execution, with probability one.\nP2: Assigns probability 3/6 to Iris-4950d and 1/6 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Iris-4950d, which has the unique lowest activation cost score, with probability one.\nP4: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Juniper-4950d on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2, P3, P4, P5; claims: G4.", "Implementations: P4; claims: G1, G3, G4.", "Implementations: P2, P3, P4, P5; claims: G2, G4.", "Implementations: P2, P3, P4; claims: G4.", "Implementations: P4, P5; claims: G1, G4.", "Implementations: P4, P5; claims: G1, G3, G4.", "Implementations: P2, P3, P4; claims: G2, G4.", "Implementations: P4; claims: G1, G4." ], "answer": "Implementations: P4; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-07", "member_count": 4, "world_id": "rt1-world-04-08", "semantic_world": { "n": 4, "members": [ "Juniper-4950d", "Iris-4950d", "Pike-4950d", "Heron-4950d" ], "outsider": "Excluded-4950d", "scores": [ 72, 53, 67, 84 ], "criterion": "activation cost", "set_ref": "roster-4950d@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P4; claims: G1, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P4; claims: G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P4, P5; claims: G1, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P4, P5; claims: G1, G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3, P4; claims: G2, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P4; claims: G1, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-4950d on every execution, with probability one.", "distribution": { "Excluded-4950d": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Iris-4950d and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-4950d": "1/6", "Iris-4950d": "1/2", "Pike-4950d": "1/6", "Heron-4950d": "1/6" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Iris-4950d, which has the unique lowest activation cost score, with probability one.", "distribution": { "Iris-4950d": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-4950d": "1/4", "Iris-4950d": "1/4", "Pike-4950d": "1/4", "Heron-4950d": "1/4" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Juniper-4950d on every execution, with probability one.", "distribution": { "Juniper-4950d": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-08", "english": "A ticket about a newsletter header slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible content item identities, in recorded order: Gale-a77df, Marsh-a77df, Nettle-a77df, Dune-a77df, Iris-a77df, Cobalt-a77df, Pike-a77df. Recorded retrieval cost scores, in the same order: 54, 13, 72, 64, 29, 68, 73. The reference roster-a77df@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a77df is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-a77df@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-a77df@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-a77df@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about a newsletter header slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible content item identities, in recorded order: Gale-a77df, Marsh-a77df, Nettle-a77df, Dune-a77df, Iris-a77df, Cobalt-a77df, Pike-a77df. Recorded retrieval cost scores, in the same order: 54, 13, 72, 64, 29, 68, 73. The reference roster-a77df@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a77df is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-a77df@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-a77df, which has the unique lowest retrieval cost score, with probability one.\nP2: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Gale-a77df on every execution, with probability one.\nP4: Assigns probability 3/9 to Nettle-a77df and 1/9 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-a77df on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2; claims: G2.", "Implementations: P1, P2; claims: G2.", "Implementations: P2; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G1, G4.", "Implementations: P2, P3, P4; claims: G1.", "Implementations: P2, P3, P4; claims: G1, G4.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1, P2; claims: G1, G2." ], "answer": "Implementations: P2; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-07", "member_count": 7, "world_id": "rt1-world-05-08", "semantic_world": { "n": 7, "members": [ "Gale-a77df", "Marsh-a77df", "Nettle-a77df", "Dune-a77df", "Iris-a77df", "Cobalt-a77df", "Pike-a77df" ], "outsider": "Excluded-a77df", "scores": [ 54, 13, 72, 64, 29, 68, 73 ], "criterion": "retrieval cost", "set_ref": "roster-a77df@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G2.": { "policies": [ "P2" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2; claims: G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2" ] }, "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4; claims: G1.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4; claims: G1, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2; claims: G1, G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Marsh-a77df, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Marsh-a77df": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-a77df": "1/7", "Marsh-a77df": "1/7", "Nettle-a77df": "1/7", "Dune-a77df": "1/7", "Iris-a77df": "1/7", "Cobalt-a77df": "1/7", "Pike-a77df": "1/7" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Gale-a77df on every execution, with probability one.", "distribution": { "Gale-a77df": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/9 to Nettle-a77df and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-a77df": "1/9", "Marsh-a77df": "1/9", "Nettle-a77df": "1/3", "Dune-a77df": "1/9", "Iris-a77df": "1/9", "Cobalt-a77df": "1/9", "Pike-a77df": "1/9" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-a77df on every execution, with probability one.", "distribution": { "Excluded-a77df": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-08", "english": "A ticket about a ledger-reconciliation job is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible worker identities, in recorded order: Dune-3e4c5, Kestrel-3e4c5, Pike-3e4c5. Recorded current load scores, in the same order: 94, 62, 60. The reference roster-3e4c5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3e4c5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-3e4c5@snapshot-1) requires roster-3e4c5@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-3e4c5@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A ticket about a ledger-reconciliation job is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible worker identities, in recorded order: Dune-3e4c5, Kestrel-3e4c5, Pike-3e4c5. Recorded current load scores, in the same order: 94, 62, 60. The reference roster-3e4c5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3e4c5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-3e4c5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Pike-3e4c5, which has the unique lowest current load score, with probability one.\nP2: Assigns probability 3/5 to Dune-3e4c5 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-3e4c5 on every execution, with probability one.\nP4: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Dune-3e4c5 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: P4; claims: G1.", "Implementations: P4; claims: G1, G2.", "Implementations: P1, P2, P4, P5; claims: none.", "Implementations: P1, P4, P5; claims: none.", "Implementations: P2, P4; claims: G1, G2.", "Implementations: P1, P4, P5; claims: G1.", "Implementations: P2, P4; claims: G1." ], "answer": "Implementations: P1, P2, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-07", "member_count": 3, "world_id": "rt1-world-06-08", "semantic_world": { "n": 3, "members": [ "Dune-3e4c5", "Kestrel-3e4c5", "Pike-3e4c5" ], "outsider": "Excluded-3e4c5", "scores": [ 94, 62, 60 ], "criterion": "current load", "set_ref": "roster-3e4c5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P4; claims: G1.": { "policies": [ "P4" ], "guarantees": [ "G1" ] }, "Implementations: P4; claims: G1, G2.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P4, P5; claims: none.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [] }, "Implementations: P2, P4; claims: G1, G2.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P4, P5; claims: G1.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4; claims: G1.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Pike-3e4c5, which has the unique lowest current load score, with probability one.", "distribution": { "Pike-3e4c5": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Dune-3e4c5 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-3e4c5": "3/5", "Kestrel-3e4c5": "1/5", "Pike-3e4c5": "1/5" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-3e4c5 on every execution, with probability one.", "distribution": { "Excluded-3e4c5": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-3e4c5": "1/3", "Kestrel-3e4c5": "1/3", "Pike-3e4c5": "1/3" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Dune-3e4c5 on every execution, with probability one.", "distribution": { "Dune-3e4c5": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-09", "english": "A compliance note distinguishes what the request for a map-tile fetch permits from what it promises.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Basalt-ce2a5, Lumen-ce2a5, Dune-ce2a5. Recorded latency scores, in the same order: 12, 54, 7. The reference roster-ce2a5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ce2a5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ce2a5@snapshot-1) requires roster-ce2a5@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-ce2a5@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A compliance note distinguishes what the request for a map-tile fetch permits from what it promises.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Basalt-ce2a5, Lumen-ce2a5, Dune-ce2a5. Recorded latency scores, in the same order: 12, 54, 7. The reference roster-ce2a5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ce2a5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ce2a5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Basalt-ce2a5 on every execution, with probability one.\nP2: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-ce2a5 on every execution, with probability one.\nP4: Returns Dune-ce2a5, which has the unique lowest latency score, with probability one.\nP5: Assigns probability 4/6 to Basalt-ce2a5 and 1/6 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: none; claims: G3.", "Implementations: P2; claims: G3, G4.", "Implementations: P1, P2, P4, P5; claims: none.", "Implementations: P1, P2, P4, P5; claims: G3.", "Implementations: P2; claims: G3.", "Implementations: P1, P4, P5; claims: G3.", "Implementations: P1, P4, P5; claims: none.", "Implementations: none; claims: G3, G4." ], "answer": "Implementations: P1, P2, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-08", "member_count": 3, "world_id": "rt1-world-01-09", "semantic_world": { "n": 3, "members": [ "Basalt-ce2a5", "Lumen-ce2a5", "Dune-ce2a5" ], "outsider": "Excluded-ce2a5", "scores": [ 12, 54, 7 ], "criterion": "latency", "set_ref": "roster-ce2a5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G3.": { "policies": [], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G3.": { "policies": [ "P2" ], "guarantees": [ "G3" ] }, "Implementations: P1, P4, P5; claims: G3.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P4, P5; claims: none.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [] }, "Implementations: none; claims: G3, G4.": { "policies": [], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Basalt-ce2a5 on every execution, with probability one.", "distribution": { "Basalt-ce2a5": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-ce2a5": "1/3", "Lumen-ce2a5": "1/3", "Dune-ce2a5": "1/3" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-ce2a5 on every execution, with probability one.", "distribution": { "Excluded-ce2a5": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Dune-ce2a5, which has the unique lowest latency score, with probability one.", "distribution": { "Dune-ce2a5": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Basalt-ce2a5 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-ce2a5": "2/3", "Lumen-ce2a5": "1/6", "Dune-ce2a5": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-09", "english": "A compliance note distinguishes what the request for a survey instrument permits from what it promises.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Juniper-96853, Gale-96853, Cobalt-96853, Basalt-96853, Dune-96853, Amber-96853. Recorded queue length scores, in the same order: 56, 63, 66, 22, 35, 41. The reference roster-96853@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-96853 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-96853@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-96853@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-96853@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a survey instrument permits from what it promises.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Juniper-96853, Gale-96853, Cobalt-96853, Basalt-96853, Dune-96853, Amber-96853. Recorded queue length scores, in the same order: 56, 63, 66, 22, 35, 41. The reference roster-96853@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-96853 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-96853@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Basalt-96853, which has the unique lowest queue length score, with probability one.\nP3: Assigns probability 4/9 to Amber-96853 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Excluded-96853 on every execution, with probability one.\nP5: Returns Juniper-96853 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P5; claims: G1.", "Implementations: P1; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P1, P4; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.", "Implementations: P1; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P4; claims: G1, G3." ], "answer": "Implementations: P1; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-08", "member_count": 6, "world_id": "rt1-world-02-09", "semantic_world": { "n": 6, "members": [ "Juniper-96853", "Gale-96853", "Cobalt-96853", "Basalt-96853", "Dune-96853", "Amber-96853" ], "outsider": "Excluded-96853", "scores": [ 56, 63, 66, 22, 35, 41 ], "criterion": "queue length", "set_ref": "roster-96853@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G3.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P4; claims: G1, G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1; claims: G1, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4; claims: G1, G3.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-96853": "1/6", "Gale-96853": "1/6", "Cobalt-96853": "1/6", "Basalt-96853": "1/6", "Dune-96853": "1/6", "Amber-96853": "1/6" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Basalt-96853, which has the unique lowest queue length score, with probability one.", "distribution": { "Basalt-96853": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Amber-96853 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-96853": "1/9", "Gale-96853": "1/9", "Cobalt-96853": "1/9", "Basalt-96853": "1/9", "Dune-96853": "1/9", "Amber-96853": "4/9" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-96853 on every execution, with probability one.", "distribution": { "Excluded-96853": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Juniper-96853 on every execution, with probability one.", "distribution": { "Juniper-96853": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-09", "english": "A compliance note distinguishes what the request for a calibration evaluation slot permits from what it promises.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Marsh-04938, Elm-04938. Recorded preparation cost scores, in the same order: 69, 56. The reference roster-04938@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-04938 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-04938@snapshot-1) requires roster-04938@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-04938@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A compliance note distinguishes what the request for a calibration evaluation slot permits from what it promises.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Marsh-04938, Elm-04938. Recorded preparation cost scores, in the same order: 69, 56. The reference roster-04938@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-04938 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-04938@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Elm-04938, which has the unique lowest preparation cost score, with probability one.\nP2: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP3: Assigns probability 4/5 to Marsh-04938 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Excluded-04938 on every execution, with probability one.\nP5: Returns Marsh-04938 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P3; claims: G2, G4.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: P1, P2, P3; claims: G4.", "Implementations: P2; claims: G1, G3, G4.", "Implementations: P2, P5; claims: G1, G3, G4.", "Implementations: P2, P5; claims: G1, G4.", "Implementations: P2; claims: G1, G4." ], "answer": "Implementations: P1, P2, P3, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-08", "member_count": 2, "world_id": "rt1-world-03-09", "semantic_world": { "n": 2, "members": [ "Marsh-04938", "Elm-04938" ], "outsider": "Excluded-04938", "scores": [ 69, 56 ], "criterion": "preparation cost", "set_ref": "roster-04938@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3; claims: G2, G4.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3; claims: G4.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G4" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P5; claims: G1, G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P5; claims: G1, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Elm-04938, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Elm-04938": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-04938": "1/2", "Elm-04938": "1/2" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Marsh-04938 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-04938": "4/5", "Elm-04938": "1/5" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-04938 on every execution, with probability one.", "distribution": { "Excluded-04938": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Marsh-04938 on every execution, with probability one.", "distribution": { "Marsh-04938": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-09", "english": "A compliance note distinguishes what the request for a crashed scheduler permits from what it promises.\nThe frozen roster lists these eligible standby node identities, in recorded order: Heron-8ea51, Juniper-8ea51, Nettle-8ea51, Kestrel-8ea51, Gale-8ea51. Recorded activation cost scores, in the same order: 96, 57, 25, 7, 89. The reference roster-8ea51@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8ea51 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-8ea51@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-8ea51@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-8ea51@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a crashed scheduler permits from what it promises.\nThe frozen roster lists these eligible standby node identities, in recorded order: Heron-8ea51, Juniper-8ea51, Nettle-8ea51, Kestrel-8ea51, Gale-8ea51. Recorded activation cost scores, in the same order: 96, 57, 25, 7, 89. The reference roster-8ea51@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8ea51 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-8ea51@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-8ea51 on every execution, with probability one.\nP2: Assigns probability 4/8 to Juniper-8ea51 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Kestrel-8ea51, which has the unique lowest activation cost score, with probability one.\nP4: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-8ea51 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P4; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P3, P4; claims: G2.", "Implementations: P3, P4; claims: G2, G3.", "Implementations: P4; claims: G2.", "Implementations: P4; claims: G2, G3.", "Implementations: P1, P2, P4; claims: G3." ], "answer": "Implementations: P4; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-08", "member_count": 5, "world_id": "rt1-world-04-09", "semantic_world": { "n": 5, "members": [ "Heron-8ea51", "Juniper-8ea51", "Nettle-8ea51", "Kestrel-8ea51", "Gale-8ea51" ], "outsider": "Excluded-8ea51", "scores": [ 96, 57, 25, 7, 89 ], "criterion": "activation cost", "set_ref": "roster-8ea51@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P3, P4; claims: G2.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P3, P4; claims: G2, G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P4; claims: G2.": { "policies": [ "P4" ], "guarantees": [ "G2" ] }, "Implementations: P4; claims: G2, G3.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P4; claims: G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Heron-8ea51 on every execution, with probability one.", "distribution": { "Heron-8ea51": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Juniper-8ea51 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-8ea51": "1/8", "Juniper-8ea51": "1/2", "Nettle-8ea51": "1/8", "Kestrel-8ea51": "1/8", "Gale-8ea51": "1/8" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Kestrel-8ea51, which has the unique lowest activation cost score, with probability one.", "distribution": { "Kestrel-8ea51": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-8ea51": "1/5", "Juniper-8ea51": "1/5", "Nettle-8ea51": "1/5", "Kestrel-8ea51": "1/5", "Gale-8ea51": "1/5" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-8ea51 on every execution, with probability one.", "distribution": { "Excluded-8ea51": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-09", "english": "A compliance note distinguishes what the request for a help-centre example slot permits from what it promises.\nThe frozen roster lists these eligible content item identities, in recorded order: Onyx-29846, Pike-29846, Flint-29846, Iris-29846, Heron-29846, Basalt-29846, Amber-29846, Dune-29846. Recorded retrieval cost scores, in the same order: 38, 81, 70, 7, 53, 50, 12, 76. The reference roster-29846@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-29846 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-29846@snapshot-1) requires roster-29846@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-29846@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A compliance note distinguishes what the request for a help-centre example slot permits from what it promises.\nThe frozen roster lists these eligible content item identities, in recorded order: Onyx-29846, Pike-29846, Flint-29846, Iris-29846, Heron-29846, Basalt-29846, Amber-29846, Dune-29846. Recorded retrieval cost scores, in the same order: 38, 81, 70, 7, 53, 50, 12, 76. The reference roster-29846@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-29846 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-29846@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-29846 on every execution, with probability one.\nP2: Returns Onyx-29846 on every execution, with probability one.\nP3: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 4/11 to Flint-29846 and 1/11 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Iris-29846, which has the unique lowest retrieval cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P3; claims: G1, G4.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P2, P3, P4, P5; claims: none.", "Implementations: P3; claims: G1.", "Implementations: P3, P4; claims: G1.", "Implementations: P2, P3, P5; claims: none.", "Implementations: P2, P3, P5; claims: G1.", "Implementations: P3, P4; claims: G1, G4." ], "answer": "Implementations: P2, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-08", "member_count": 8, "world_id": "rt1-world-05-09", "semantic_world": { "n": 8, "members": [ "Onyx-29846", "Pike-29846", "Flint-29846", "Iris-29846", "Heron-29846", "Basalt-29846", "Amber-29846", "Dune-29846" ], "outsider": "Excluded-29846", "scores": [ 38, 81, 70, 7, 53, 50, 12, 76 ], "criterion": "retrieval cost", "set_ref": "roster-29846@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3; claims: G1, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4, P5; claims: none.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P3; claims: G1.": { "policies": [ "P3" ], "guarantees": [ "G1" ] }, "Implementations: P3, P4; claims: G1.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P5; claims: none.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P2, P3, P5; claims: G1.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P3, P4; claims: G1, G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-29846 on every execution, with probability one.", "distribution": { "Excluded-29846": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Onyx-29846 on every execution, with probability one.", "distribution": { "Onyx-29846": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-29846": "1/8", "Pike-29846": "1/8", "Flint-29846": "1/8", "Iris-29846": "1/8", "Heron-29846": "1/8", "Basalt-29846": "1/8", "Amber-29846": "1/8", "Dune-29846": "1/8" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/11 to Flint-29846 and 1/11 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-29846": "1/11", "Pike-29846": "1/11", "Flint-29846": "4/11", "Iris-29846": "1/11", "Heron-29846": "1/11", "Basalt-29846": "1/11", "Amber-29846": "1/11", "Dune-29846": "1/11" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Iris-29846, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Iris-29846": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-09", "english": "A compliance note distinguishes what the request for a spell-check job permits from what it promises.\nThe frozen roster lists these eligible worker identities, in recorded order: Elm-03df9, Nettle-03df9, Cobalt-03df9, Flint-03df9. Recorded current load scores, in the same order: 68, 57, 8, 58. The reference roster-03df9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-03df9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-03df9@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-03df9@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-03df9@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a spell-check job permits from what it promises.\nThe frozen roster lists these eligible worker identities, in recorded order: Elm-03df9, Nettle-03df9, Cobalt-03df9, Flint-03df9. Recorded current load scores, in the same order: 68, 57, 8, 58. The reference roster-03df9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-03df9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-03df9@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Cobalt-03df9, which has the unique lowest current load score, with probability one.\nP3: Returns Elm-03df9 on every execution, with probability one.\nP4: Returns Excluded-03df9 on every execution, with probability one.\nP5: Assigns probability 4/7 to Nettle-03df9 and 1/7 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: none; claims: G1, G2, G4.", "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: none; claims: G1, G4.", "Implementations: P1, P2, P3, P5; claims: G1, G4.", "Implementations: P2, P3, P5; claims: G4.", "Implementations: P1; claims: G1, G2, G4.", "Implementations: P1; claims: G1, G4.", "Implementations: P2, P3, P5; claims: G1, G4." ], "answer": "Implementations: P1; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-08", "member_count": 4, "world_id": "rt1-world-06-09", "semantic_world": { "n": 4, "members": [ "Elm-03df9", "Nettle-03df9", "Cobalt-03df9", "Flint-03df9" ], "outsider": "Excluded-03df9", "scores": [ 68, 57, 8, 58 ], "criterion": "current load", "set_ref": "roster-03df9@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G1, G2, G4.": { "policies": [], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P5; claims: G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1; claims: G1, G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1; claims: G1, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P5; claims: G1, G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-03df9": "1/4", "Nettle-03df9": "1/4", "Cobalt-03df9": "1/4", "Flint-03df9": "1/4" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Cobalt-03df9, which has the unique lowest current load score, with probability one.", "distribution": { "Cobalt-03df9": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Elm-03df9 on every execution, with probability one.", "distribution": { "Elm-03df9": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-03df9 on every execution, with probability one.", "distribution": { "Excluded-03df9": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Nettle-03df9 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-03df9": "1/7", "Nettle-03df9": "4/7", "Cobalt-03df9": "1/7", "Flint-03df9": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-10", "english": "A junior engineer asks a senior one what the instruction about a session-refresh call rules in and out.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Onyx-90736, Dune-90736, Iris-90736, Kestrel-90736. Recorded latency scores, in the same order: 56, 13, 81, 3. The reference roster-90736@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-90736 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-90736@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-90736@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-90736@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A junior engineer asks a senior one what the instruction about a session-refresh call rules in and out.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Onyx-90736, Dune-90736, Iris-90736, Kestrel-90736. Recorded latency scores, in the same order: 56, 13, 81, 3. The reference roster-90736@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-90736 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-90736@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/5 to Dune-90736 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Onyx-90736 on every execution, with probability one.\nP3: Returns Kestrel-90736, which has the unique lowest latency score, with probability one.\nP4: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-90736 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P4; claims: G1, G3.", "Implementations: P4; claims: G1, G3, G4.", "Implementations: P4, P5; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G1, G3.", "Implementations: P4, P5; claims: G1, G3." ], "answer": "Implementations: P4; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-09", "member_count": 4, "world_id": "rt1-world-01-10", "semantic_world": { "n": 4, "members": [ "Onyx-90736", "Dune-90736", "Iris-90736", "Kestrel-90736" ], "outsider": "Excluded-90736", "scores": [ 56, 13, 81, 3 ], "criterion": "latency", "set_ref": "roster-90736@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4; claims: G1, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G1, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P4, P5; claims: G1, G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4, P5; claims: G1, G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Dune-90736 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-90736": "1/5", "Dune-90736": "2/5", "Iris-90736": "1/5", "Kestrel-90736": "1/5" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Onyx-90736 on every execution, with probability one.", "distribution": { "Onyx-90736": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Kestrel-90736, which has the unique lowest latency score, with probability one.", "distribution": { "Kestrel-90736": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-90736": "1/4", "Dune-90736": "1/4", "Iris-90736": "1/4", "Kestrel-90736": "1/4" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-90736 on every execution, with probability one.", "distribution": { "Excluded-90736": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-10", "english": "A junior engineer asks a senior one what the instruction about a preprint revision rules in and out.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Heron-881ac, Marsh-881ac, Flint-881ac, Iris-881ac, Elm-881ac, Basalt-881ac, Onyx-881ac. Recorded queue length scores, in the same order: 80, 38, 89, 25, 90, 39, 84. The reference roster-881ac@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-881ac is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-881ac@snapshot-1) requires roster-881ac@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-881ac@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A junior engineer asks a senior one what the instruction about a preprint revision rules in and out.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Heron-881ac, Marsh-881ac, Flint-881ac, Iris-881ac, Elm-881ac, Basalt-881ac, Onyx-881ac. Recorded queue length scores, in the same order: 80, 38, 89, 25, 90, 39, 84. The reference roster-881ac@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-881ac is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-881ac@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-881ac on every execution, with probability one.\nP3: Returns Heron-881ac on every execution, with probability one.\nP4: Assigns probability 2/8 to Marsh-881ac and 1/8 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Iris-881ac, which has the unique lowest queue length score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P3; claims: G1, G2, G4.", "Implementations: P1, P3, P4, P5; claims: G3, G4.", "Implementations: P1; claims: G1, G4.", "Implementations: P1, P4, P5; claims: G3, G4.", "Implementations: P1, P4, P5; claims: G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P1, P3; claims: G1, G4.", "Implementations: P1; claims: G1, G2, G4." ], "answer": "Implementations: P1, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-09", "member_count": 7, "world_id": "rt1-world-02-10", "semantic_world": { "n": 7, "members": [ "Heron-881ac", "Marsh-881ac", "Flint-881ac", "Iris-881ac", "Elm-881ac", "Basalt-881ac", "Onyx-881ac" ], "outsider": "Excluded-881ac", "scores": [ 80, 38, 89, 25, 90, 39, 84 ], "criterion": "queue length", "set_ref": "roster-881ac@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3; claims: G1, G2, G4.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1; claims: G1, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P4, P5; claims: G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3; claims: G1, G4.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1; claims: G1, G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-881ac": "1/7", "Marsh-881ac": "1/7", "Flint-881ac": "1/7", "Iris-881ac": "1/7", "Elm-881ac": "1/7", "Basalt-881ac": "1/7", "Onyx-881ac": "1/7" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-881ac on every execution, with probability one.", "distribution": { "Excluded-881ac": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Heron-881ac on every execution, with probability one.", "distribution": { "Heron-881ac": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Marsh-881ac and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-881ac": "1/8", "Marsh-881ac": "1/4", "Flint-881ac": "1/8", "Iris-881ac": "1/8", "Elm-881ac": "1/8", "Basalt-881ac": "1/8", "Onyx-881ac": "1/8" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Iris-881ac, which has the unique lowest queue length score, with probability one.", "distribution": { "Iris-881ac": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-10", "english": "A junior engineer asks a senior one what the instruction about a coding evaluation slot rules in and out.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Juniper-e49ef, Marsh-e49ef, Dune-e49ef. Recorded preparation cost scores, in the same order: 68, 47, 45. The reference roster-e49ef@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e49ef is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e49ef@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-e49ef@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-e49ef@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A junior engineer asks a senior one what the instruction about a coding evaluation slot rules in and out.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Juniper-e49ef, Marsh-e49ef, Dune-e49ef. Recorded preparation cost scores, in the same order: 68, 47, 45. The reference roster-e49ef@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e49ef is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e49ef@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Juniper-e49ef on every execution, with probability one.\nP2: Assigns probability 2/4 to Juniper-e49ef and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-e49ef on every execution, with probability one.\nP4: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Dune-e49ef, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P4, P5; claims: G2.", "Implementations: P4; claims: G2, G4.", "Implementations: P1, P2, P4; claims: G1, G4.", "Implementations: P4, P5; claims: G2, G4.", "Implementations: P4; claims: G2.", "Implementations: P1, P2, P4; claims: G4.", "Implementations: P1, P2, P4, P5; claims: G4.", "Implementations: P1, P2, P4, P5; claims: G1, G4." ], "answer": "Implementations: P4; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-09", "member_count": 3, "world_id": "rt1-world-03-10", "semantic_world": { "n": 3, "members": [ "Juniper-e49ef", "Marsh-e49ef", "Dune-e49ef" ], "outsider": "Excluded-e49ef", "scores": [ 68, 47, 45 ], "criterion": "preparation cost", "set_ref": "roster-e49ef@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4, P5; claims: G2.": { "policies": [ "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P4; claims: G2, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P4, P5; claims: G2, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P4; claims: G2.": { "policies": [ "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P4; claims: G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Juniper-e49ef on every execution, with probability one.", "distribution": { "Juniper-e49ef": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Juniper-e49ef and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-e49ef": "1/2", "Marsh-e49ef": "1/4", "Dune-e49ef": "1/4" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-e49ef on every execution, with probability one.", "distribution": { "Excluded-e49ef": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-e49ef": "1/3", "Marsh-e49ef": "1/3", "Dune-e49ef": "1/3" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Dune-e49ef, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Dune-e49ef": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-10", "english": "A junior engineer asks a senior one what the instruction about a paused webhook relay rules in and out.\nThe frozen roster lists these eligible standby node identities, in recorded order: Elm-28994, Lumen-28994, Pike-28994, Cobalt-28994, Kestrel-28994, Marsh-28994. Recorded activation cost scores, in the same order: 26, 63, 39, 77, 89, 2. The reference roster-28994@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-28994 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-28994@snapshot-1) requires roster-28994@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-28994@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A junior engineer asks a senior one what the instruction about a paused webhook relay rules in and out.\nThe frozen roster lists these eligible standby node identities, in recorded order: Elm-28994, Lumen-28994, Pike-28994, Cobalt-28994, Kestrel-28994, Marsh-28994. Recorded activation cost scores, in the same order: 26, 63, 39, 77, 89, 2. The reference roster-28994@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-28994 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-28994@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/7 to Elm-28994 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Excluded-28994 on every execution, with probability one.\nP3: Returns Elm-28994 on every execution, with probability one.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Marsh-28994, which has the unique lowest activation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P4; claims: G1, G2.", "Implementations: P1, P4; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G2.", "Implementations: P3, P4, P5; claims: G2.", "Implementations: P4; claims: G2.", "Implementations: P4; claims: G1, G2.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P3, P4, P5; claims: none." ], "answer": "Implementations: P1, P3, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-09", "member_count": 6, "world_id": "rt1-world-04-10", "semantic_world": { "n": 6, "members": [ "Elm-28994", "Lumen-28994", "Pike-28994", "Cobalt-28994", "Kestrel-28994", "Marsh-28994" ], "outsider": "Excluded-28994", "scores": [ 26, 63, 39, 77, 89, 2 ], "criterion": "activation cost", "set_ref": "roster-28994@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P4; claims: G1, G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P4; claims: G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P3, P4, P5; claims: G2.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P4; claims: G2.": { "policies": [ "P4" ], "guarantees": [ "G2" ] }, "Implementations: P4; claims: G1, G2.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P3, P4, P5; claims: none.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/7 to Elm-28994 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-28994": "2/7", "Lumen-28994": "1/7", "Pike-28994": "1/7", "Cobalt-28994": "1/7", "Kestrel-28994": "1/7", "Marsh-28994": "1/7" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-28994 on every execution, with probability one.", "distribution": { "Excluded-28994": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Elm-28994 on every execution, with probability one.", "distribution": { "Elm-28994": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-28994": "1/6", "Lumen-28994": "1/6", "Pike-28994": "1/6", "Cobalt-28994": "1/6", "Kestrel-28994": "1/6", "Marsh-28994": "1/6" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Marsh-28994, which has the unique lowest activation cost score, with probability one.", "distribution": { "Marsh-28994": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-10", "english": "A junior engineer asks a senior one what the instruction about a release-note screenshot slot rules in and out.\nThe frozen roster lists these eligible content item identities, in recorded order: Juniper-88265, Iris-88265. Recorded retrieval cost scores, in the same order: 68, 21. The reference roster-88265@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-88265 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-88265@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-88265@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-88265@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A junior engineer asks a senior one what the instruction about a release-note screenshot slot rules in and out.\nThe frozen roster lists these eligible content item identities, in recorded order: Juniper-88265, Iris-88265. Recorded retrieval cost scores, in the same order: 68, 21. The reference roster-88265@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-88265 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-88265@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Iris-88265, which has the unique lowest retrieval cost score, with probability one.\nP3: Assigns probability 2/3 to Juniper-88265 and 1/3 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Juniper-88265 on every execution, with probability one.\nP5: Returns Excluded-88265 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: none; claims: G3, G4.", "Implementations: P1; claims: G1, G3, G4.", "Implementations: none; claims: G1, G3, G4.", "Implementations: P1; claims: G3, G4.", "Implementations: P2, P3, P4; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G3, G4.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P2, P3, P4; claims: G3, G4." ], "answer": "Implementations: P1; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-09", "member_count": 2, "world_id": "rt1-world-05-10", "semantic_world": { "n": 2, "members": [ "Juniper-88265", "Iris-88265" ], "outsider": "Excluded-88265", "scores": [ 68, 21 ], "criterion": "retrieval cost", "set_ref": "roster-88265@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G3, G4.": { "policies": [], "guarantees": [ "G3", "G4" ] }, "Implementations: P1; claims: G1, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1; claims: G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3, P4; claims: G3.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3, P4; claims: G3, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-88265": "1/2", "Iris-88265": "1/2" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Iris-88265, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Iris-88265": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Juniper-88265 and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-88265": "2/3", "Iris-88265": "1/3" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Juniper-88265 on every execution, with probability one.", "distribution": { "Juniper-88265": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-88265 on every execution, with probability one.", "distribution": { "Excluded-88265": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-10", "english": "A junior engineer asks a senior one what the instruction about a geocoding job rules in and out.\nThe frozen roster lists these eligible worker identities, in recorded order: Marsh-59330, Heron-59330, Flint-59330, Basalt-59330, Cobalt-59330. Recorded current load scores, in the same order: 35, 29, 80, 82, 62. The reference roster-59330@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-59330 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-59330@snapshot-1) requires roster-59330@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-59330@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A junior engineer asks a senior one what the instruction about a geocoding job rules in and out.\nThe frozen roster lists these eligible worker identities, in recorded order: Marsh-59330, Heron-59330, Flint-59330, Basalt-59330, Cobalt-59330. Recorded current load scores, in the same order: 35, 29, 80, 82, 62. The reference roster-59330@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-59330 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-59330@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-59330, which has the unique lowest current load score, with probability one.\nP2: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-59330 on every execution, with probability one.\nP4: Assigns probability 2/6 to Flint-59330 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Marsh-59330 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2; claims: G1, G3, G4.", "Implementations: P2, P3; claims: G1, G3.", "Implementations: P2; claims: G1, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G2, G3.", "Implementations: P2, P3; claims: G1, G3, G4.", "Implementations: P1, P2, P4, P5; claims: G2, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P1, P2, P4, P5; claims: G3." ], "answer": "Implementations: P1, P2, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-09", "member_count": 5, "world_id": "rt1-world-06-10", "semantic_world": { "n": 5, "members": [ "Marsh-59330", "Heron-59330", "Flint-59330", "Basalt-59330", "Cobalt-59330" ], "outsider": "Excluded-59330", "scores": [ 35, 29, 80, 82, 62 ], "criterion": "current load", "set_ref": "roster-59330@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3; claims: G1, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2; claims: G1, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3; claims: G1, G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Heron-59330, which has the unique lowest current load score, with probability one.", "distribution": { "Heron-59330": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-59330": "1/5", "Heron-59330": "1/5", "Flint-59330": "1/5", "Basalt-59330": "1/5", "Cobalt-59330": "1/5" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-59330 on every execution, with probability one.", "distribution": { "Excluded-59330": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Flint-59330 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-59330": "1/6", "Heron-59330": "1/6", "Flint-59330": "1/3", "Basalt-59330": "1/6", "Cobalt-59330": "1/6" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Marsh-59330 on every execution, with probability one.", "distribution": { "Marsh-59330": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-11", "english": "The dispatch log will record the interpretation of the request for a price-quote request; nothing has run yet.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-235dd, Nettle-235dd, Onyx-235dd, Basalt-235dd, Pike-235dd. Recorded latency scores, in the same order: 60, 72, 16, 65, 49. The reference roster-235dd@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-235dd is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-235dd@snapshot-1) requires roster-235dd@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-235dd@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a price-quote request; nothing has run yet.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-235dd, Nettle-235dd, Onyx-235dd, Basalt-235dd, Pike-235dd. Recorded latency scores, in the same order: 60, 72, 16, 65, 49. The reference roster-235dd@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-235dd is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-235dd@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 3/7 to Flint-235dd and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-235dd on every execution, with probability one.\nP4: Returns Onyx-235dd, which has the unique lowest latency score, with probability one.\nP5: Returns Flint-235dd on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P2; claims: G2, G4.", "Implementations: P1, P2, P4, P5; claims: G2.", "Implementations: P1, P2, P4; claims: G2.", "Implementations: P2; claims: G1, G2, G4.", "Implementations: P2, P5; claims: G2, G4.", "Implementations: P1, P2, P4; claims: G2, G3.", "Implementations: P1, P2, P4, P5; claims: G2, G3.", "Implementations: P2, P5; claims: G1, G2, G4." ], "answer": "Implementations: P1, P2, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-10", "member_count": 5, "world_id": "rt1-world-01-11", "semantic_world": { "n": 5, "members": [ "Flint-235dd", "Nettle-235dd", "Onyx-235dd", "Basalt-235dd", "Pike-235dd" ], "outsider": "Excluded-235dd", "scores": [ 60, 72, 16, 65, 49 ], "criterion": "latency", "set_ref": "roster-235dd@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P4; claims: G2.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P2; claims: G1, G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P2, P5; claims: G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P4; claims: G2, G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P5; claims: G1, G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Flint-235dd and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-235dd": "3/7", "Nettle-235dd": "1/7", "Onyx-235dd": "1/7", "Basalt-235dd": "1/7", "Pike-235dd": "1/7" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-235dd": "1/5", "Nettle-235dd": "1/5", "Onyx-235dd": "1/5", "Basalt-235dd": "1/5", "Pike-235dd": "1/5" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-235dd on every execution, with probability one.", "distribution": { "Excluded-235dd": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Onyx-235dd, which has the unique lowest latency score, with probability one.", "distribution": { "Onyx-235dd": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Flint-235dd on every execution, with probability one.", "distribution": { "Flint-235dd": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-11", "english": "The dispatch log will record the interpretation of the request for a thesis chapter; nothing has run yet.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Amber-b5465, Heron-b5465, Juniper-b5465, Dune-b5465, Flint-b5465, Elm-b5465, Kestrel-b5465, Cobalt-b5465. Recorded queue length scores, in the same order: 91, 88, 14, 25, 56, 82, 64, 8. The reference roster-b5465@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b5465 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b5465@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-b5465@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-b5465@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The dispatch log will record the interpretation of the request for a thesis chapter; nothing has run yet.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Amber-b5465, Heron-b5465, Juniper-b5465, Dune-b5465, Flint-b5465, Elm-b5465, Kestrel-b5465, Cobalt-b5465. Recorded queue length scores, in the same order: 91, 88, 14, 25, 56, 82, 64, 8. The reference roster-b5465@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b5465 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b5465@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-b5465, which has the unique lowest queue length score, with probability one.\nP2: Returns Excluded-b5465 on every execution, with probability one.\nP3: Returns Amber-b5465 on every execution, with probability one.\nP4: Assigns probability 3/10 to Juniper-b5465 and 1/10 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P5; claims: G2.", "Implementations: P3, P4, P5; claims: G4.", "Implementations: P1, P3, P4, P5; claims: G3, G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P1, P5; claims: G2.", "Implementations: P1, P5; claims: G2, G4.", "Implementations: P5; claims: G2, G4.", "Implementations: P3, P4, P5; claims: G3, G4." ], "answer": "Implementations: P5; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-10", "member_count": 8, "world_id": "rt1-world-02-11", "semantic_world": { "n": 8, "members": [ "Amber-b5465", "Heron-b5465", "Juniper-b5465", "Dune-b5465", "Flint-b5465", "Elm-b5465", "Kestrel-b5465", "Cobalt-b5465" ], "outsider": "Excluded-b5465", "scores": [ 91, 88, 14, 25, 56, 82, 64, 8 ], "criterion": "queue length", "set_ref": "roster-b5465@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G2.": { "policies": [ "P5" ], "guarantees": [ "G2" ] }, "Implementations: P3, P4, P5; claims: G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P5; claims: G2.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P5; claims: G2, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P3, P4, P5; claims: G3, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Cobalt-b5465, which has the unique lowest queue length score, with probability one.", "distribution": { "Cobalt-b5465": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-b5465 on every execution, with probability one.", "distribution": { "Excluded-b5465": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Amber-b5465 on every execution, with probability one.", "distribution": { "Amber-b5465": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Juniper-b5465 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-b5465": "1/10", "Heron-b5465": "1/10", "Juniper-b5465": "3/10", "Dune-b5465": "1/10", "Flint-b5465": "1/10", "Elm-b5465": "1/10", "Kestrel-b5465": "1/10", "Cobalt-b5465": "1/10" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-b5465": "1/8", "Heron-b5465": "1/8", "Juniper-b5465": "1/8", "Dune-b5465": "1/8", "Flint-b5465": "1/8", "Elm-b5465": "1/8", "Kestrel-b5465": "1/8", "Cobalt-b5465": "1/8" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-11", "english": "The dispatch log will record the interpretation of the request for a citation evaluation slot; nothing has run yet.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Dune-d3fe2, Cobalt-d3fe2, Nettle-d3fe2, Juniper-d3fe2. Recorded preparation cost scores, in the same order: 52, 35, 67, 78. The reference roster-d3fe2@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-d3fe2 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-d3fe2@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-d3fe2@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-d3fe2@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The dispatch log will record the interpretation of the request for a citation evaluation slot; nothing has run yet.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Dune-d3fe2, Cobalt-d3fe2, Nettle-d3fe2, Juniper-d3fe2. Recorded preparation cost scores, in the same order: 52, 35, 67, 78. The reference roster-d3fe2@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-d3fe2 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-d3fe2@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-d3fe2, which has the unique lowest preparation cost score, with probability one.\nP2: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-d3fe2 on every execution, with probability one.\nP4: Assigns probability 3/6 to Cobalt-d3fe2 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Dune-d3fe2 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2; claims: G4.", "Implementations: P2, P4; claims: G4.", "Implementations: P1, P2, P5; claims: none.", "Implementations: P1, P2, P5; claims: G4.", "Implementations: P1, P2, P4, P5; claims: none.", "Implementations: P1, P2, P4, P5; claims: G4.", "Implementations: P2; claims: G1, G4.", "Implementations: P2, P4; claims: G1, G4." ], "answer": "Implementations: P2; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-10", "member_count": 4, "world_id": "rt1-world-03-11", "semantic_world": { "n": 4, "members": [ "Dune-d3fe2", "Cobalt-d3fe2", "Nettle-d3fe2", "Juniper-d3fe2" ], "outsider": "Excluded-d3fe2", "scores": [ 52, 35, 67, 78 ], "criterion": "preparation cost", "set_ref": "roster-d3fe2@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G4.": { "policies": [ "P2" ], "guarantees": [ "G4" ] }, "Implementations: P2, P4; claims: G4.": { "policies": [ "P2", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P5; claims: none.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P5; claims: G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P4; claims: G1, G4.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Cobalt-d3fe2, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Cobalt-d3fe2": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-d3fe2": "1/4", "Cobalt-d3fe2": "1/4", "Nettle-d3fe2": "1/4", "Juniper-d3fe2": "1/4" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-d3fe2 on every execution, with probability one.", "distribution": { "Excluded-d3fe2": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Cobalt-d3fe2 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-d3fe2": "1/6", "Cobalt-d3fe2": "1/2", "Nettle-d3fe2": "1/6", "Juniper-d3fe2": "1/6" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Dune-d3fe2 on every execution, with probability one.", "distribution": { "Dune-d3fe2": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-11", "english": "The dispatch log will record the interpretation of the request for a degraded search shard; nothing has run yet.\nThe frozen roster lists these eligible standby node identities, in recorded order: Heron-b4d1a, Dune-b4d1a, Juniper-b4d1a, Iris-b4d1a, Elm-b4d1a, Nettle-b4d1a, Cobalt-b4d1a. Recorded activation cost scores, in the same order: 34, 23, 36, 51, 71, 27, 86. The reference roster-b4d1a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b4d1a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-b4d1a@snapshot-1) requires roster-b4d1a@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-b4d1a@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a degraded search shard; nothing has run yet.\nThe frozen roster lists these eligible standby node identities, in recorded order: Heron-b4d1a, Dune-b4d1a, Juniper-b4d1a, Iris-b4d1a, Elm-b4d1a, Nettle-b4d1a, Cobalt-b4d1a. Recorded activation cost scores, in the same order: 34, 23, 36, 51, 71, 27, 86. The reference roster-b4d1a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b4d1a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-b4d1a@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 3/9 to Juniper-b4d1a and 1/9 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Heron-b4d1a on every execution, with probability one.\nP4: Returns Excluded-b4d1a on every execution, with probability one.\nP5: Returns Dune-b4d1a, which has the unique lowest activation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P3, P5; claims: G3.", "Implementations: P2; claims: G1, G3.", "Implementations: P2; claims: G1, G3, G4.", "Implementations: none; claims: G1, G3.", "Implementations: none; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P1, P3, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-10", "member_count": 7, "world_id": "rt1-world-04-11", "semantic_world": { "n": 7, "members": [ "Heron-b4d1a", "Dune-b4d1a", "Juniper-b4d1a", "Iris-b4d1a", "Elm-b4d1a", "Nettle-b4d1a", "Cobalt-b4d1a" ], "outsider": "Excluded-b4d1a", "scores": [ 34, 23, 36, 51, 71, 27, 86 ], "criterion": "activation cost", "set_ref": "roster-b4d1a@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P5; claims: G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G1, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: none; claims: G1, G3.": { "policies": [], "guarantees": [ "G1", "G3" ] }, "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 3/9 to Juniper-b4d1a and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-b4d1a": "1/9", "Dune-b4d1a": "1/9", "Juniper-b4d1a": "1/3", "Iris-b4d1a": "1/9", "Elm-b4d1a": "1/9", "Nettle-b4d1a": "1/9", "Cobalt-b4d1a": "1/9" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-b4d1a": "1/7", "Dune-b4d1a": "1/7", "Juniper-b4d1a": "1/7", "Iris-b4d1a": "1/7", "Elm-b4d1a": "1/7", "Nettle-b4d1a": "1/7", "Cobalt-b4d1a": "1/7" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Heron-b4d1a on every execution, with probability one.", "distribution": { "Heron-b4d1a": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-b4d1a on every execution, with probability one.", "distribution": { "Excluded-b4d1a": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Dune-b4d1a, which has the unique lowest activation cost score, with probability one.", "distribution": { "Dune-b4d1a": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-11", "english": "The dispatch log will record the interpretation of the request for a podcast intro slot; nothing has run yet.\nThe frozen roster lists these eligible content item identities, in recorded order: Nettle-10030, Marsh-10030, Heron-10030. Recorded retrieval cost scores, in the same order: 56, 50, 41. The reference roster-10030@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-10030 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-10030@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-10030@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-10030@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The dispatch log will record the interpretation of the request for a podcast intro slot; nothing has run yet.\nThe frozen roster lists these eligible content item identities, in recorded order: Nettle-10030, Marsh-10030, Heron-10030. Recorded retrieval cost scores, in the same order: 56, 50, 41. The reference roster-10030@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-10030 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-10030@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 3/5 to Marsh-10030 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Nettle-10030 on every execution, with probability one.\nP4: Returns Heron-10030, which has the unique lowest retrieval cost score, with probability one.\nP5: Returns Excluded-10030 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1; claims: G1, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P2, P3, P4; claims: G1, G3.", "Implementations: P1; claims: G1, G2, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.", "Implementations: P1, P5; claims: G1, G4.", "Implementations: P1, P5; claims: G1, G2, G4." ], "answer": "Implementations: P1; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-10", "member_count": 3, "world_id": "rt1-world-05-11", "semantic_world": { "n": 3, "members": [ "Nettle-10030", "Marsh-10030", "Heron-10030" ], "outsider": "Excluded-10030", "scores": [ 56, 50, 41 ], "criterion": "retrieval cost", "set_ref": "roster-10030@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1; claims: G1, G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P5; claims: G1, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P5; claims: G1, G2, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-10030": "1/3", "Marsh-10030": "1/3", "Heron-10030": "1/3" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Marsh-10030 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-10030": "1/5", "Marsh-10030": "3/5", "Heron-10030": "1/5" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Nettle-10030 on every execution, with probability one.", "distribution": { "Nettle-10030": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Heron-10030, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Heron-10030": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-10030 on every execution, with probability one.", "distribution": { "Excluded-10030": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-11", "english": "The dispatch log will record the interpretation of the request for a log-rotation job; nothing has run yet.\nThe frozen roster lists these eligible worker identities, in recorded order: Heron-a4bd5, Pike-a4bd5, Nettle-a4bd5, Lumen-a4bd5, Onyx-a4bd5, Elm-a4bd5. Recorded current load scores, in the same order: 26, 40, 8, 47, 58, 60. The reference roster-a4bd5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a4bd5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-a4bd5@snapshot-1) requires roster-a4bd5@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-a4bd5@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a log-rotation job; nothing has run yet.\nThe frozen roster lists these eligible worker identities, in recorded order: Heron-a4bd5, Pike-a4bd5, Nettle-a4bd5, Lumen-a4bd5, Onyx-a4bd5, Elm-a4bd5. Recorded current load scores, in the same order: 26, 40, 8, 47, 58, 60. The reference roster-a4bd5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a4bd5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-a4bd5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-a4bd5 on every execution, with probability one.\nP2: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Nettle-a4bd5, which has the unique lowest current load score, with probability one.\nP4: Assigns probability 3/8 to Nettle-a4bd5 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-a4bd5 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P2; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: P2; claims: G1.", "Implementations: P2, P3, P4; claims: G2.", "Implementations: P1, P2; claims: G1.", "Implementations: P2, P3, P4; claims: G2, G4." ], "answer": "Implementations: P1, P2, P3, P4; claims: G2.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-10", "member_count": 6, "world_id": "rt1-world-06-11", "semantic_world": { "n": 6, "members": [ "Heron-a4bd5", "Pike-a4bd5", "Nettle-a4bd5", "Lumen-a4bd5", "Onyx-a4bd5", "Elm-a4bd5" ], "outsider": "Excluded-a4bd5", "scores": [ 26, 40, 8, 47, 58, 60 ], "criterion": "current load", "set_ref": "roster-a4bd5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2; claims: G1, G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P2; claims: G1.": { "policies": [ "P2" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4; claims: G2.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2; claims: G1.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4; claims: G2, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Heron-a4bd5 on every execution, with probability one.", "distribution": { "Heron-a4bd5": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-a4bd5": "1/6", "Pike-a4bd5": "1/6", "Nettle-a4bd5": "1/6", "Lumen-a4bd5": "1/6", "Onyx-a4bd5": "1/6", "Elm-a4bd5": "1/6" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Nettle-a4bd5, which has the unique lowest current load score, with probability one.", "distribution": { "Nettle-a4bd5": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Nettle-a4bd5 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-a4bd5": "1/8", "Pike-a4bd5": "1/8", "Nettle-a4bd5": "3/8", "Lumen-a4bd5": "1/8", "Onyx-a4bd5": "1/8", "Elm-a4bd5": "1/8" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-a4bd5 on every execution, with probability one.", "distribution": { "Excluded-a4bd5": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-12", "english": "An incident retrospective re-reads the original selection request for a font-subset request.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Lumen-9fb96, Onyx-9fb96, Flint-9fb96, Gale-9fb96, Basalt-9fb96, Dune-9fb96. Recorded latency scores, in the same order: 94, 81, 79, 30, 45, 90. The reference roster-9fb96@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9fb96 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-9fb96@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-9fb96@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-9fb96@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "An incident retrospective re-reads the original selection request for a font-subset request.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Lumen-9fb96, Onyx-9fb96, Flint-9fb96, Gale-9fb96, Basalt-9fb96, Dune-9fb96. Recorded latency scores, in the same order: 94, 81, 79, 30, 45, 90. The reference roster-9fb96@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9fb96 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-9fb96@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Lumen-9fb96 on every execution, with probability one.\nP2: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-9fb96 on every execution, with probability one.\nP4: Assigns probability 4/9 to Lumen-9fb96 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Gale-9fb96, which has the unique lowest latency score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P4; claims: G1, G4.", "Implementations: P2, P5; claims: G3, G4.", "Implementations: P1, P2, P4, P5; claims: G1, G4.", "Implementations: P1, P2, P4, P5; claims: G4.", "Implementations: P2; claims: G3.", "Implementations: P2, P5; claims: G3.", "Implementations: P2; claims: G3, G4.", "Implementations: P1, P2, P4; claims: G4." ], "answer": "Implementations: P2; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-11", "member_count": 6, "world_id": "rt1-world-01-12", "semantic_world": { "n": 6, "members": [ "Lumen-9fb96", "Onyx-9fb96", "Flint-9fb96", "Gale-9fb96", "Basalt-9fb96", "Dune-9fb96" ], "outsider": "Excluded-9fb96", "scores": [ 94, 81, 79, 30, 45, 90 ], "criterion": "latency", "set_ref": "roster-9fb96@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P5; claims: G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2; claims: G3.": { "policies": [ "P2" ], "guarantees": [ "G3" ] }, "Implementations: P2, P5; claims: G3.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P4; claims: G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Lumen-9fb96 on every execution, with probability one.", "distribution": { "Lumen-9fb96": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Lumen-9fb96": "1/6", "Onyx-9fb96": "1/6", "Flint-9fb96": "1/6", "Gale-9fb96": "1/6", "Basalt-9fb96": "1/6", "Dune-9fb96": "1/6" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-9fb96 on every execution, with probability one.", "distribution": { "Excluded-9fb96": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Lumen-9fb96 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Lumen-9fb96": "4/9", "Onyx-9fb96": "1/9", "Flint-9fb96": "1/9", "Gale-9fb96": "1/9", "Basalt-9fb96": "1/9", "Dune-9fb96": "1/9" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Gale-9fb96, which has the unique lowest latency score, with probability one.", "distribution": { "Gale-9fb96": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-12", "english": "An incident retrospective re-reads the original selection request for a benchmark submission.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-ab0a9, Amber-ab0a9. Recorded queue length scores, in the same order: 88, 59. The reference roster-ab0a9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ab0a9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-ab0a9@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-ab0a9@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-ab0a9@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "An incident retrospective re-reads the original selection request for a benchmark submission.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-ab0a9, Amber-ab0a9. Recorded queue length scores, in the same order: 88, 59. The reference roster-ab0a9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ab0a9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-ab0a9@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Amber-ab0a9, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Basalt-ab0a9 on every execution, with probability one.\nP4: Assigns probability 4/5 to Amber-ab0a9 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-ab0a9 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2, P4; claims: G1, G3.", "Implementations: P1, P2, P3; claims: none.", "Implementations: P2; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P1, P2, P3; claims: G3.", "Implementations: P2, P4; claims: G3.", "Implementations: P2; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: none." ], "answer": "Implementations: P2; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-11", "member_count": 2, "world_id": "rt1-world-02-12", "semantic_world": { "n": 2, "members": [ "Basalt-ab0a9", "Amber-ab0a9" ], "outsider": "Excluded-ab0a9", "scores": [ 88, 59 ], "criterion": "queue length", "set_ref": "roster-ab0a9@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P4; claims: G1, G3.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3; claims: none.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [] }, "Implementations: P2; claims: G3.": { "policies": [ "P2" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3; claims: G3.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G3" ] }, "Implementations: P2, P4; claims: G3.": { "policies": [ "P2", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G1, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: none.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Amber-ab0a9, which has the unique lowest queue length score, with probability one.", "distribution": { "Amber-ab0a9": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-ab0a9": "1/2", "Amber-ab0a9": "1/2" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Basalt-ab0a9 on every execution, with probability one.", "distribution": { "Basalt-ab0a9": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Amber-ab0a9 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-ab0a9": "1/5", "Amber-ab0a9": "4/5" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-ab0a9 on every execution, with probability one.", "distribution": { "Excluded-ab0a9": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-12", "english": "An incident retrospective re-reads the original selection request for a dialogue evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Kestrel-4f425, Lumen-4f425, Cobalt-4f425, Pike-4f425, Onyx-4f425. Recorded preparation cost scores, in the same order: 20, 71, 79, 52, 5. The reference roster-4f425@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-4f425 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-4f425@snapshot-1) requires roster-4f425@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-4f425@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "An incident retrospective re-reads the original selection request for a dialogue evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Kestrel-4f425, Lumen-4f425, Cobalt-4f425, Pike-4f425, Onyx-4f425. Recorded preparation cost scores, in the same order: 20, 71, 79, 52, 5. The reference roster-4f425@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-4f425 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-4f425@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-4f425 on every execution, with probability one.\nP2: Returns Onyx-4f425, which has the unique lowest preparation cost score, with probability one.\nP3: Assigns probability 4/8 to Lumen-4f425 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP4: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Kestrel-4f425 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2, P3, P5; claims: G4.", "Implementations: P2, P3, P5; claims: G2, G4.", "Implementations: none; claims: G2, G4.", "Implementations: P4; claims: G2, G4.", "Implementations: P4; claims: G2, G3, G4.", "Implementations: P2, P3, P4, P5; claims: G4.", "Implementations: P2, P3, P4, P5; claims: G2, G4.", "Implementations: none; claims: G2, G3, G4." ], "answer": "Implementations: P2, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-11", "member_count": 5, "world_id": "rt1-world-03-12", "semantic_world": { "n": 5, "members": [ "Kestrel-4f425", "Lumen-4f425", "Cobalt-4f425", "Pike-4f425", "Onyx-4f425" ], "outsider": "Excluded-4f425", "scores": [ 20, 71, 79, 52, 5 ], "criterion": "preparation cost", "set_ref": "roster-4f425@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P5; claims: G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P5; claims: G2, G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G2, G4.": { "policies": [], "guarantees": [ "G2", "G4" ] }, "Implementations: P4; claims: G2, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P4; claims: G2, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G2, G3, G4.": { "policies": [], "guarantees": [ "G2", "G3", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-4f425 on every execution, with probability one.", "distribution": { "Excluded-4f425": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Onyx-4f425, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Onyx-4f425": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Lumen-4f425 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-4f425": "1/8", "Lumen-4f425": "1/2", "Cobalt-4f425": "1/8", "Pike-4f425": "1/8", "Onyx-4f425": "1/8" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-4f425": "1/5", "Lumen-4f425": "1/5", "Cobalt-4f425": "1/5", "Pike-4f425": "1/5", "Onyx-4f425": "1/5" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Kestrel-4f425 on every execution, with probability one.", "distribution": { "Kestrel-4f425": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-12", "english": "An incident retrospective re-reads the original selection request for a halted backup agent.\nThe frozen roster lists these eligible standby node identities, in recorded order: Amber-26e3e, Elm-26e3e, Flint-26e3e, Juniper-26e3e, Gale-26e3e, Kestrel-26e3e, Heron-26e3e, Nettle-26e3e. Recorded activation cost scores, in the same order: 19, 93, 75, 7, 6, 40, 70, 17. The reference roster-26e3e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-26e3e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-26e3e@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-26e3e@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-26e3e@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "An incident retrospective re-reads the original selection request for a halted backup agent.\nThe frozen roster lists these eligible standby node identities, in recorded order: Amber-26e3e, Elm-26e3e, Flint-26e3e, Juniper-26e3e, Gale-26e3e, Kestrel-26e3e, Heron-26e3e, Nettle-26e3e. Recorded activation cost scores, in the same order: 19, 93, 75, 7, 6, 40, 70, 17. The reference roster-26e3e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-26e3e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-26e3e@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Amber-26e3e on every execution, with probability one.\nP2: Returns Gale-26e3e, which has the unique lowest activation cost score, with probability one.\nP3: Returns Excluded-26e3e on every execution, with probability one.\nP4: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 4/11 to Heron-26e3e and 1/11 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.", "Implementations: P3, P4; claims: G2, G3, G4.", "Implementations: P1, P2, P4, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P1, P2, P4, P5; claims: G3.", "Implementations: P4; claims: G2, G3, G4.", "Implementations: P3, P4; claims: G2, G3.", "Implementations: P4; claims: G2, G3." ], "answer": "Implementations: P4; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-11", "member_count": 8, "world_id": "rt1-world-04-12", "semantic_world": { "n": 8, "members": [ "Amber-26e3e", "Elm-26e3e", "Flint-26e3e", "Juniper-26e3e", "Gale-26e3e", "Kestrel-26e3e", "Heron-26e3e", "Nettle-26e3e" ], "outsider": "Excluded-26e3e", "scores": [ 19, 93, 75, 7, 6, 40, 70, 17 ], "criterion": "activation cost", "set_ref": "roster-26e3e@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3, P4; claims: G2, G3, G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G2, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P3, P4; claims: G2, G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P4; claims: G2, G3.": { "policies": [ "P4" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Amber-26e3e on every execution, with probability one.", "distribution": { "Amber-26e3e": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Gale-26e3e, which has the unique lowest activation cost score, with probability one.", "distribution": { "Gale-26e3e": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-26e3e on every execution, with probability one.", "distribution": { "Excluded-26e3e": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-26e3e": "1/8", "Elm-26e3e": "1/8", "Flint-26e3e": "1/8", "Juniper-26e3e": "1/8", "Gale-26e3e": "1/8", "Kestrel-26e3e": "1/8", "Heron-26e3e": "1/8", "Nettle-26e3e": "1/8" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/11 to Heron-26e3e and 1/11 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-26e3e": "1/11", "Elm-26e3e": "1/11", "Flint-26e3e": "1/11", "Juniper-26e3e": "1/11", "Gale-26e3e": "1/11", "Kestrel-26e3e": "1/11", "Heron-26e3e": "4/11", "Nettle-26e3e": "1/11" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-12", "english": "An incident retrospective re-reads the original selection request for a museum label slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Iris-7c0d4, Marsh-7c0d4, Elm-7c0d4, Flint-7c0d4. Recorded retrieval cost scores, in the same order: 92, 90, 21, 9. The reference roster-7c0d4@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7c0d4 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-7c0d4@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-7c0d4@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-7c0d4@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "An incident retrospective re-reads the original selection request for a museum label slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Iris-7c0d4, Marsh-7c0d4, Elm-7c0d4, Flint-7c0d4. Recorded retrieval cost scores, in the same order: 92, 90, 21, 9. The reference roster-7c0d4@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7c0d4 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-7c0d4@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-7c0d4 on every execution, with probability one.\nP2: Returns Flint-7c0d4, which has the unique lowest retrieval cost score, with probability one.\nP3: Returns Iris-7c0d4 on every execution, with probability one.\nP4: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 4/7 to Flint-7c0d4 and 1/7 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P3, P4; claims: G1.", "Implementations: P2, P4, P5; claims: G3, G4.", "Implementations: P3, P4; claims: G1, G3.", "Implementations: P4; claims: G1, G3.", "Implementations: P2, P3, P4, P5; claims: G3.", "Implementations: P2, P4, P5; claims: G3.", "Implementations: P4; claims: G1.", "Implementations: P2, P3, P4, P5; claims: G3, G4." ], "answer": "Implementations: P4; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-11", "member_count": 4, "world_id": "rt1-world-05-12", "semantic_world": { "n": 4, "members": [ "Iris-7c0d4", "Marsh-7c0d4", "Elm-7c0d4", "Flint-7c0d4" ], "outsider": "Excluded-7c0d4", "scores": [ 92, 90, 21, 9 ], "criterion": "retrieval cost", "set_ref": "roster-7c0d4@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4; claims: G1.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4, P5; claims: G3, G4.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P3, P4; claims: G1, G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G1, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2, P4, P5; claims: G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G1.": { "policies": [ "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4, P5; claims: G3, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-7c0d4 on every execution, with probability one.", "distribution": { "Excluded-7c0d4": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Flint-7c0d4, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Flint-7c0d4": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Iris-7c0d4 on every execution, with probability one.", "distribution": { "Iris-7c0d4": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Iris-7c0d4": "1/4", "Marsh-7c0d4": "1/4", "Elm-7c0d4": "1/4", "Flint-7c0d4": "1/4" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Flint-7c0d4 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Iris-7c0d4": "1/7", "Marsh-7c0d4": "1/7", "Elm-7c0d4": "1/7", "Flint-7c0d4": "4/7" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-12", "english": "An incident retrospective re-reads the original selection request for a dependency-scan job.\nThe frozen roster lists these eligible worker identities, in recorded order: Pike-c256b, Nettle-c256b, Onyx-c256b, Iris-c256b, Flint-c256b, Amber-c256b, Heron-c256b. Recorded current load scores, in the same order: 55, 64, 69, 39, 80, 54, 42. The reference roster-c256b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c256b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-c256b@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-c256b@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-c256b@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "An incident retrospective re-reads the original selection request for a dependency-scan job.\nThe frozen roster lists these eligible worker identities, in recorded order: Pike-c256b, Nettle-c256b, Onyx-c256b, Iris-c256b, Flint-c256b, Amber-c256b, Heron-c256b. Recorded current load scores, in the same order: 55, 64, 69, 39, 80, 54, 42. The reference roster-c256b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c256b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-c256b@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Iris-c256b, which has the unique lowest current load score, with probability one.\nP2: Assigns probability 4/10 to Amber-c256b and 1/10 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Pike-c256b on every execution, with probability one.\nP4: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-c256b on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P4; claims: G1.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1, P4; claims: G1, G2.", "Implementations: P4; claims: G1, G2.", "Implementations: P4; claims: G1.", "Implementations: P1, P2, P3, P4; claims: none.", "Implementations: P2, P3, P4; claims: none.", "Implementations: P2, P3, P4; claims: G1." ], "answer": "Implementations: P4; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-11", "member_count": 7, "world_id": "rt1-world-06-12", "semantic_world": { "n": 7, "members": [ "Pike-c256b", "Nettle-c256b", "Onyx-c256b", "Iris-c256b", "Flint-c256b", "Amber-c256b", "Heron-c256b" ], "outsider": "Excluded-c256b", "scores": [ 55, 64, 69, 39, 80, 54, 42 ], "criterion": "current load", "set_ref": "roster-c256b@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P4; claims: G1.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4; claims: G1, G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4; claims: G1, G2.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4; claims: G1.": { "policies": [ "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: none.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [] }, "Implementations: P2, P3, P4; claims: none.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [] }, "Implementations: P2, P3, P4; claims: G1.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Iris-c256b, which has the unique lowest current load score, with probability one.", "distribution": { "Iris-c256b": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Amber-c256b and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Pike-c256b": "1/10", "Nettle-c256b": "1/10", "Onyx-c256b": "1/10", "Iris-c256b": "1/10", "Flint-c256b": "1/10", "Amber-c256b": "2/5", "Heron-c256b": "1/10" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Pike-c256b on every execution, with probability one.", "distribution": { "Pike-c256b": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Pike-c256b": "1/7", "Nettle-c256b": "1/7", "Onyx-c256b": "1/7", "Iris-c256b": "1/7", "Flint-c256b": "1/7", "Amber-c256b": "1/7", "Heron-c256b": "1/7" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-c256b on every execution, with probability one.", "distribution": { "Excluded-c256b": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-13", "english": "A routing note about a receipt-rendering call is under review before it is acted on.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Kestrel-14e02, Nettle-14e02, Basalt-14e02, Iris-14e02, Juniper-14e02, Lumen-14e02, Dune-14e02. Recorded latency scores, in the same order: 54, 35, 71, 76, 82, 93, 23. The reference roster-14e02@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-14e02 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-14e02@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-14e02@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-14e02@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A routing note about a receipt-rendering call is under review before it is acted on.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Kestrel-14e02, Nettle-14e02, Basalt-14e02, Iris-14e02, Juniper-14e02, Lumen-14e02, Dune-14e02. Recorded latency scores, in the same order: 54, 35, 71, 76, 82, 93, 23. The reference roster-14e02@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-14e02 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-14e02@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Dune-14e02, which has the unique lowest latency score, with probability one.\nP2: Returns Excluded-14e02 on every execution, with probability one.\nP3: Assigns probability 2/8 to Dune-14e02 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP4: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Kestrel-14e02 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P4; claims: G3, G4.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P4; claims: G4.", "Implementations: P1, P4, P5; claims: G4.", "Implementations: P3, P4; claims: G4.", "Implementations: P1, P4, P5; claims: none.", "Implementations: P3, P4; claims: G3, G4.", "Implementations: P1, P3, P4, P5; claims: G4." ], "answer": "Implementations: P4; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-00", "member_count": 7, "world_id": "rt1-world-01-13", "semantic_world": { "n": 7, "members": [ "Kestrel-14e02", "Nettle-14e02", "Basalt-14e02", "Iris-14e02", "Juniper-14e02", "Lumen-14e02", "Dune-14e02" ], "outsider": "Excluded-14e02", "scores": [ 54, 35, 71, 76, 82, 93, 23 ], "criterion": "latency", "set_ref": "roster-14e02@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P4; claims: G4.": { "policies": [ "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P4, P5; claims: G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P3, P4; claims: G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P4, P5; claims: none.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [] }, "Implementations: P3, P4; claims: G3, G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Dune-14e02, which has the unique lowest latency score, with probability one.", "distribution": { "Dune-14e02": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-14e02 on every execution, with probability one.", "distribution": { "Excluded-14e02": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Dune-14e02 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-14e02": "1/8", "Nettle-14e02": "1/8", "Basalt-14e02": "1/8", "Iris-14e02": "1/8", "Juniper-14e02": "1/8", "Lumen-14e02": "1/8", "Dune-14e02": "1/4" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-14e02": "1/7", "Nettle-14e02": "1/7", "Basalt-14e02": "1/7", "Iris-14e02": "1/7", "Juniper-14e02": "1/7", "Lumen-14e02": "1/7", "Dune-14e02": "1/7" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Kestrel-14e02 on every execution, with probability one.", "distribution": { "Kestrel-14e02": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-13", "english": "A routing note about a grant renewal is under review before it is acted on.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Onyx-0674c, Elm-0674c, Basalt-0674c. Recorded queue length scores, in the same order: 15, 7, 84. The reference roster-0674c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-0674c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-0674c@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-0674c@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-0674c@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A routing note about a grant renewal is under review before it is acted on.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Onyx-0674c, Elm-0674c, Basalt-0674c. Recorded queue length scores, in the same order: 15, 7, 84. The reference roster-0674c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-0674c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-0674c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 2/4 to Elm-0674c and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-0674c on every execution, with probability one.\nP4: Returns Elm-0674c, which has the unique lowest queue length score, with probability one.\nP5: Returns Onyx-0674c on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: none; claims: G1, G3, G4.", "Implementations: P2, P4, P5; claims: G1, G3.", "Implementations: P1; claims: G1, G3, G4.", "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: none; claims: G1, G3.", "Implementations: P2, P4, P5; claims: G1.", "Implementations: P1; claims: G1, G3.", "Implementations: P1, P2, P4, P5; claims: G1, G3." ], "answer": "Implementations: P1; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-00", "member_count": 3, "world_id": "rt1-world-02-13", "semantic_world": { "n": 3, "members": [ "Onyx-0674c", "Elm-0674c", "Basalt-0674c" ], "outsider": "Excluded-0674c", "scores": [ 15, 7, 84 ], "criterion": "queue length", "set_ref": "roster-0674c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1; claims: G1, G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: none; claims: G1, G3.": { "policies": [], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P4, P5; claims: G1.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G3.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-0674c": "1/3", "Elm-0674c": "1/3", "Basalt-0674c": "1/3" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Elm-0674c and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-0674c": "1/4", "Elm-0674c": "1/2", "Basalt-0674c": "1/4" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-0674c on every execution, with probability one.", "distribution": { "Excluded-0674c": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Elm-0674c, which has the unique lowest queue length score, with probability one.", "distribution": { "Elm-0674c": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Onyx-0674c on every execution, with probability one.", "distribution": { "Onyx-0674c": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-13", "english": "A routing note about a summarisation evaluation slot is under review before it is acted on.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-67032, Juniper-67032, Lumen-67032, Pike-67032, Heron-67032, Cobalt-67032. Recorded preparation cost scores, in the same order: 72, 71, 14, 36, 94, 31. The reference roster-67032@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-67032 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-67032@snapshot-1) requires roster-67032@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-67032@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A routing note about a summarisation evaluation slot is under review before it is acted on.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-67032, Juniper-67032, Lumen-67032, Pike-67032, Heron-67032, Cobalt-67032. Recorded preparation cost scores, in the same order: 72, 71, 14, 36, 94, 31. The reference roster-67032@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-67032 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-67032@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Lumen-67032, which has the unique lowest preparation cost score, with probability one.\nP2: Assigns probability 2/7 to Heron-67032 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Gale-67032 on every execution, with probability one.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-67032 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P4; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.", "Implementations: P4, P5; claims: G1, G3, G4.", "Implementations: P4; claims: G3, G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: P4, P5; claims: G3, G4." ], "answer": "Implementations: P1, P2, P3, P4; claims: G4.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-00", "member_count": 6, "world_id": "rt1-world-03-13", "semantic_world": { "n": 6, "members": [ "Gale-67032", "Juniper-67032", "Lumen-67032", "Pike-67032", "Heron-67032", "Cobalt-67032" ], "outsider": "Excluded-67032", "scores": [ 72, 71, 14, 36, 94, 31 ], "criterion": "preparation cost", "set_ref": "roster-67032@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4; claims: G1, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P4, P5; claims: G1, G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P4, P5; claims: G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Lumen-67032, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Lumen-67032": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/7 to Heron-67032 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-67032": "1/7", "Juniper-67032": "1/7", "Lumen-67032": "1/7", "Pike-67032": "1/7", "Heron-67032": "2/7", "Cobalt-67032": "1/7" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Gale-67032 on every execution, with probability one.", "distribution": { "Gale-67032": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-67032": "1/6", "Juniper-67032": "1/6", "Lumen-67032": "1/6", "Pike-67032": "1/6", "Heron-67032": "1/6", "Cobalt-67032": "1/6" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-67032 on every execution, with probability one.", "distribution": { "Excluded-67032": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-13", "english": "A routing note about a stalled queue consumer is under review before it is acted on.\nThe frozen roster lists these eligible standby node identities, in recorded order: Onyx-f3e98, Juniper-f3e98. Recorded activation cost scores, in the same order: 87, 46. The reference roster-f3e98@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f3e98 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f3e98@snapshot-1) requires roster-f3e98@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-f3e98@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A routing note about a stalled queue consumer is under review before it is acted on.\nThe frozen roster lists these eligible standby node identities, in recorded order: Onyx-f3e98, Juniper-f3e98. Recorded activation cost scores, in the same order: 87, 46. The reference roster-f3e98@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f3e98 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f3e98@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Juniper-f3e98, which has the unique lowest activation cost score, with probability one.\nP2: Returns Onyx-f3e98 on every execution, with probability one.\nP3: Assigns probability 2/3 to Onyx-f3e98 and 1/3 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Excluded-f3e98 on every execution, with probability one.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P5; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G1.", "Implementations: P1, P3, P5; claims: G1.", "Implementations: P5; claims: G2.", "Implementations: P1, P3, P5; claims: G1, G3.", "Implementations: P2, P5; claims: G1, G2.", "Implementations: P2, P5; claims: G2." ], "answer": "Implementations: P1, P2, P3, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-00", "member_count": 2, "world_id": "rt1-world-04-13", "semantic_world": { "n": 2, "members": [ "Onyx-f3e98", "Juniper-f3e98" ], "outsider": "Excluded-f3e98", "scores": [ 87, 46 ], "criterion": "activation cost", "set_ref": "roster-f3e98@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P5; claims: G1, G2.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P5; claims: G1.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P5; claims: G2.": { "policies": [ "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P5; claims: G1, G2.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P5; claims: G2.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Juniper-f3e98, which has the unique lowest activation cost score, with probability one.", "distribution": { "Juniper-f3e98": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Onyx-f3e98 on every execution, with probability one.", "distribution": { "Onyx-f3e98": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Onyx-f3e98 and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-f3e98": "2/3", "Juniper-f3e98": "1/3" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-f3e98 on every execution, with probability one.", "distribution": { "Excluded-f3e98": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-f3e98": "1/2", "Juniper-f3e98": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-13", "english": "A routing note about a landing-page banner slot is under review before it is acted on.\nThe frozen roster lists these eligible content item identities, in recorded order: Cobalt-3c672, Basalt-3c672, Gale-3c672, Kestrel-3c672, Pike-3c672. Recorded retrieval cost scores, in the same order: 58, 24, 60, 90, 78. The reference roster-3c672@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3c672 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-3c672@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-3c672@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-3c672@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A routing note about a landing-page banner slot is under review before it is acted on.\nThe frozen roster lists these eligible content item identities, in recorded order: Cobalt-3c672, Basalt-3c672, Gale-3c672, Kestrel-3c672, Pike-3c672. Recorded retrieval cost scores, in the same order: 58, 24, 60, 90, 78. The reference roster-3c672@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3c672 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-3c672@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-3c672 on every execution, with probability one.\nP2: Assigns probability 2/6 to Cobalt-3c672 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Cobalt-3c672 on every execution, with probability one.\nP4: Returns Basalt-3c672, which has the unique lowest retrieval cost score, with probability one.\nP5: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P4, P5; claims: G1.", "Implementations: P4, P5; claims: G1, G2.", "Implementations: P5; claims: G1.", "Implementations: P5; claims: G1, G2.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P2, P3, P4, P5; claims: none.", "Implementations: P2, P3, P5; claims: none.", "Implementations: P2, P3, P5; claims: G1." ], "answer": "Implementations: P5; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-00", "member_count": 5, "world_id": "rt1-world-05-13", "semantic_world": { "n": 5, "members": [ "Cobalt-3c672", "Basalt-3c672", "Gale-3c672", "Kestrel-3c672", "Pike-3c672" ], "outsider": "Excluded-3c672", "scores": [ 58, 24, 60, 90, 78 ], "criterion": "retrieval cost", "set_ref": "roster-3c672@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4, P5; claims: G1.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P4, P5; claims: G1, G2.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P5; claims: G1.": { "policies": [ "P5" ], "guarantees": [ "G1" ] }, "Implementations: P5; claims: G1, G2.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4, P5; claims: none.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P2, P3, P5; claims: none.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P2, P3, P5; claims: G1.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-3c672 on every execution, with probability one.", "distribution": { "Excluded-3c672": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Cobalt-3c672 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-3c672": "1/3", "Basalt-3c672": "1/6", "Gale-3c672": "1/6", "Kestrel-3c672": "1/6", "Pike-3c672": "1/6" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Cobalt-3c672 on every execution, with probability one.", "distribution": { "Cobalt-3c672": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Basalt-3c672, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Basalt-3c672": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-3c672": "1/5", "Basalt-3c672": "1/5", "Gale-3c672": "1/5", "Kestrel-3c672": "1/5", "Pike-3c672": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-13", "english": "A routing note about a thumbnail-generation job is under review before it is acted on.\nThe frozen roster lists these eligible worker identities, in recorded order: Dune-d2061, Amber-d2061, Cobalt-d2061, Kestrel-d2061, Pike-d2061, Elm-d2061, Nettle-d2061, Onyx-d2061. Recorded current load scores, in the same order: 31, 95, 35, 3, 8, 83, 17, 22. The reference roster-d2061@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-d2061 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-d2061@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-d2061@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-d2061@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A routing note about a thumbnail-generation job is under review before it is acted on.\nThe frozen roster lists these eligible worker identities, in recorded order: Dune-d2061, Amber-d2061, Cobalt-d2061, Kestrel-d2061, Pike-d2061, Elm-d2061, Nettle-d2061, Onyx-d2061. Recorded current load scores, in the same order: 31, 95, 35, 3, 8, 83, 17, 22. The reference roster-d2061@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-d2061 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-d2061@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Dune-d2061 on every execution, with probability one.\nP2: Returns Kestrel-d2061, which has the unique lowest current load score, with probability one.\nP3: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-d2061 on every execution, with probability one.\nP5: Assigns probability 2/9 to Onyx-d2061 and 1/9 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P5; claims: G1, G2.", "Implementations: P3; claims: G1, G2.", "Implementations: P3; claims: G1, G2, G3.", "Implementations: P3, P5; claims: G1, G2, G3.", "Implementations: P1, P2, P3; claims: G1.", "Implementations: P1, P2, P3; claims: G1, G2.", "Implementations: P3, P5; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G1." ], "answer": "Implementations: P3; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-00", "member_count": 8, "world_id": "rt1-world-06-13", "semantic_world": { "n": 8, "members": [ "Dune-d2061", "Amber-d2061", "Cobalt-d2061", "Kestrel-d2061", "Pike-d2061", "Elm-d2061", "Nettle-d2061", "Onyx-d2061" ], "outsider": "Excluded-d2061", "scores": [ 31, 95, 35, 3, 8, 83, 17, 22 ], "criterion": "current load", "set_ref": "roster-d2061@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P3, P5; claims: G1, G2, G3.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P2, P3; claims: G1.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3; claims: G1, G2.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P5; claims: G1, G2.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Dune-d2061 on every execution, with probability one.", "distribution": { "Dune-d2061": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Kestrel-d2061, which has the unique lowest current load score, with probability one.", "distribution": { "Kestrel-d2061": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-d2061": "1/8", "Amber-d2061": "1/8", "Cobalt-d2061": "1/8", "Kestrel-d2061": "1/8", "Pike-d2061": "1/8", "Elm-d2061": "1/8", "Nettle-d2061": "1/8", "Onyx-d2061": "1/8" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-d2061 on every execution, with probability one.", "distribution": { "Excluded-d2061": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Onyx-d2061 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-d2061": "1/9", "Amber-d2061": "1/9", "Cobalt-d2061": "1/9", "Kestrel-d2061": "1/9", "Pike-d2061": "1/9", "Elm-d2061": "1/9", "Nettle-d2061": "1/9", "Onyx-d2061": "2/9" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-14", "english": "Two operators are reading the same request concerning a postcode lookup and must agree what it asks.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Marsh-468ee, Lumen-468ee, Cobalt-468ee, Nettle-468ee, Heron-468ee, Juniper-468ee, Dune-468ee, Pike-468ee. Recorded latency scores, in the same order: 53, 39, 62, 31, 75, 12, 58, 19. The reference roster-468ee@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-468ee is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-468ee@snapshot-1) requires roster-468ee@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-468ee@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Two operators are reading the same request concerning a postcode lookup and must agree what it asks.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Marsh-468ee, Lumen-468ee, Cobalt-468ee, Nettle-468ee, Heron-468ee, Juniper-468ee, Dune-468ee, Pike-468ee. Recorded latency scores, in the same order: 53, 39, 62, 31, 75, 12, 58, 19. The reference roster-468ee@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-468ee is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-468ee@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-468ee on every execution, with probability one.\nP2: Returns Juniper-468ee, which has the unique lowest latency score, with probability one.\nP3: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 3/10 to Heron-468ee and 1/10 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-468ee on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P3, P4; claims: G3, G4.", "Implementations: none; claims: G1, G3, G4.", "Implementations: P1, P2, P4; claims: G4.", "Implementations: P3; claims: G1, G3, G4.", "Implementations: none; claims: G3, G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: P1, P2, P4; claims: G3, G4.", "Implementations: P3; claims: G3, G4." ], "answer": "Implementations: P1, P2, P3, P4; claims: G4.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-01", "member_count": 8, "world_id": "rt1-world-01-14", "semantic_world": { "n": 8, "members": [ "Marsh-468ee", "Lumen-468ee", "Cobalt-468ee", "Nettle-468ee", "Heron-468ee", "Juniper-468ee", "Dune-468ee", "Pike-468ee" ], "outsider": "Excluded-468ee", "scores": [ 53, 39, 62, 31, 75, 12, 58, 19 ], "criterion": "latency", "set_ref": "roster-468ee@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P4; claims: G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P3; claims: G1, G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: none; claims: G3, G4.": { "policies": [], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P4; claims: G3, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P3; claims: G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Marsh-468ee on every execution, with probability one.", "distribution": { "Marsh-468ee": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Juniper-468ee, which has the unique lowest latency score, with probability one.", "distribution": { "Juniper-468ee": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-468ee": "1/8", "Lumen-468ee": "1/8", "Cobalt-468ee": "1/8", "Nettle-468ee": "1/8", "Heron-468ee": "1/8", "Juniper-468ee": "1/8", "Dune-468ee": "1/8", "Pike-468ee": "1/8" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Heron-468ee and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-468ee": "1/10", "Lumen-468ee": "1/10", "Cobalt-468ee": "1/10", "Nettle-468ee": "1/10", "Heron-468ee": "3/10", "Juniper-468ee": "1/10", "Dune-468ee": "1/10", "Pike-468ee": "1/10" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-468ee on every execution, with probability one.", "distribution": { "Excluded-468ee": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-14", "english": "Two operators are reading the same request concerning a code-audit report and must agree what it asks.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-6c1a9, Kestrel-6c1a9, Gale-6c1a9, Nettle-6c1a9. Recorded queue length scores, in the same order: 55, 17, 13, 2. The reference roster-6c1a9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-6c1a9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-6c1a9@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-6c1a9@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-6c1a9@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning a code-audit report and must agree what it asks.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-6c1a9, Kestrel-6c1a9, Gale-6c1a9, Nettle-6c1a9. Recorded queue length scores, in the same order: 55, 17, 13, 2. The reference roster-6c1a9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-6c1a9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-6c1a9@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Basalt-6c1a9 on every execution, with probability one.\nP2: Returns Excluded-6c1a9 on every execution, with probability one.\nP3: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Nettle-6c1a9, which has the unique lowest queue length score, with probability one.\nP5: Assigns probability 3/6 to Gale-6c1a9 and 1/6 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P2, P3; claims: G1, G2.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G4.", "Implementations: P1, P3, P4, P5; claims: G1, G4.", "Implementations: P3; claims: G1, G2.", "Implementations: P3; claims: G1, G2, G3.", "Implementations: P2, P3; claims: G1, G2, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P3, P4, P5; claims: G1." ], "answer": "Implementations: P3; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-01", "member_count": 4, "world_id": "rt1-world-02-14", "semantic_world": { "n": 4, "members": [ "Basalt-6c1a9", "Kestrel-6c1a9", "Gale-6c1a9", "Nettle-6c1a9" ], "outsider": "Excluded-6c1a9", "scores": [ 55, 17, 13, 2 ], "criterion": "queue length", "set_ref": "roster-6c1a9@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3; claims: G1, G2.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3; claims: G1, G2, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Basalt-6c1a9 on every execution, with probability one.", "distribution": { "Basalt-6c1a9": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-6c1a9 on every execution, with probability one.", "distribution": { "Excluded-6c1a9": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-6c1a9": "1/4", "Kestrel-6c1a9": "1/4", "Gale-6c1a9": "1/4", "Nettle-6c1a9": "1/4" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Nettle-6c1a9, which has the unique lowest queue length score, with probability one.", "distribution": { "Nettle-6c1a9": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Gale-6c1a9 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-6c1a9": "1/6", "Kestrel-6c1a9": "1/6", "Gale-6c1a9": "1/2", "Nettle-6c1a9": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-14", "english": "Two operators are reading the same request concerning a tool-use evaluation slot and must agree what it asks.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-778c3, Kestrel-778c3, Nettle-778c3, Flint-778c3, Juniper-778c3, Dune-778c3, Cobalt-778c3. Recorded preparation cost scores, in the same order: 62, 17, 94, 4, 16, 29, 59. The reference roster-778c3@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-778c3 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-778c3@snapshot-1) requires roster-778c3@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-778c3@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Two operators are reading the same request concerning a tool-use evaluation slot and must agree what it asks.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-778c3, Kestrel-778c3, Nettle-778c3, Flint-778c3, Juniper-778c3, Dune-778c3, Cobalt-778c3. Recorded preparation cost scores, in the same order: 62, 17, 94, 4, 16, 29, 59. The reference roster-778c3@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-778c3 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-778c3@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 3/9 to Dune-778c3 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-778c3 on every execution, with probability one.\nP4: Returns Gale-778c3 on every execution, with probability one.\nP5: Returns Flint-778c3, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1; claims: G2.", "Implementations: P1, P4; claims: G1, G2.", "Implementations: P1, P2, P5; claims: G1.", "Implementations: P1; claims: G1, G2.", "Implementations: P1, P2, P4, P5; claims: G1, G3.", "Implementations: P1, P4; claims: G2.", "Implementations: P1, P2, P5; claims: G1, G3.", "Implementations: P1, P2, P4, P5; claims: G1." ], "answer": "Implementations: P1, P2, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-01", "member_count": 7, "world_id": "rt1-world-03-14", "semantic_world": { "n": 7, "members": [ "Gale-778c3", "Kestrel-778c3", "Nettle-778c3", "Flint-778c3", "Juniper-778c3", "Dune-778c3", "Cobalt-778c3" ], "outsider": "Excluded-778c3", "scores": [ 62, 17, 94, 4, 16, 29, 59 ], "criterion": "preparation cost", "set_ref": "roster-778c3@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] }, "Implementations: P1, P4; claims: G1, G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P5; claims: G1.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G2.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P4; claims: G2.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-778c3": "1/7", "Kestrel-778c3": "1/7", "Nettle-778c3": "1/7", "Flint-778c3": "1/7", "Juniper-778c3": "1/7", "Dune-778c3": "1/7", "Cobalt-778c3": "1/7" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/9 to Dune-778c3 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-778c3": "1/9", "Kestrel-778c3": "1/9", "Nettle-778c3": "1/9", "Flint-778c3": "1/9", "Juniper-778c3": "1/9", "Dune-778c3": "1/3", "Cobalt-778c3": "1/9" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-778c3 on every execution, with probability one.", "distribution": { "Excluded-778c3": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Gale-778c3 on every execution, with probability one.", "distribution": { "Gale-778c3": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Flint-778c3, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Flint-778c3": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-14", "english": "Two operators are reading the same request concerning an unreachable metrics store and must agree what it asks.\nThe frozen roster lists these eligible standby node identities, in recorded order: Flint-7c838, Dune-7c838, Cobalt-7c838. Recorded activation cost scores, in the same order: 93, 24, 20. The reference roster-7c838@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7c838 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-7c838@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-7c838@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-7c838@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning an unreachable metrics store and must agree what it asks.\nThe frozen roster lists these eligible standby node identities, in recorded order: Flint-7c838, Dune-7c838, Cobalt-7c838. Recorded activation cost scores, in the same order: 93, 24, 20. The reference roster-7c838@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7c838 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-7c838@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Cobalt-7c838, which has the unique lowest activation cost score, with probability one.\nP3: Assigns probability 3/5 to Cobalt-7c838 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Flint-7c838 on every execution, with probability one.\nP5: Returns Excluded-7c838 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P3, P4; claims: none.", "Implementations: P1, P3, P4; claims: G4.", "Implementations: P1, P2; claims: G4.", "Implementations: P1, P2; claims: G2, G4.", "Implementations: P1; claims: G4.", "Implementations: P1; claims: G2, G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: P1, P2, P3, P4; claims: none." ], "answer": "Implementations: P1; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-01", "member_count": 3, "world_id": "rt1-world-04-14", "semantic_world": { "n": 3, "members": [ "Flint-7c838", "Dune-7c838", "Cobalt-7c838" ], "outsider": "Excluded-7c838", "scores": [ 93, 24, 20 ], "criterion": "activation cost", "set_ref": "roster-7c838@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P4; claims: none.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [] }, "Implementations: P1, P3, P4; claims: G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2; claims: G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2; claims: G2, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1; claims: G4.": { "policies": [ "P1" ], "guarantees": [ "G4" ] }, "Implementations: P1; claims: G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: none.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-7c838": "1/3", "Dune-7c838": "1/3", "Cobalt-7c838": "1/3" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Cobalt-7c838, which has the unique lowest activation cost score, with probability one.", "distribution": { "Cobalt-7c838": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Cobalt-7c838 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-7c838": "1/5", "Dune-7c838": "1/5", "Cobalt-7c838": "3/5" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Flint-7c838 on every execution, with probability one.", "distribution": { "Flint-7c838": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-7c838 on every execution, with probability one.", "distribution": { "Excluded-7c838": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-14", "english": "Two operators are reading the same request concerning a newsletter header slot and must agree what it asks.\nThe frozen roster lists these eligible content item identities, in recorded order: Amber-3e1d8, Heron-3e1d8, Flint-3e1d8, Basalt-3e1d8, Juniper-3e1d8, Dune-3e1d8. Recorded retrieval cost scores, in the same order: 31, 74, 70, 46, 48, 16. The reference roster-3e1d8@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3e1d8 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-3e1d8@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-3e1d8@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-3e1d8@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning a newsletter header slot and must agree what it asks.\nThe frozen roster lists these eligible content item identities, in recorded order: Amber-3e1d8, Heron-3e1d8, Flint-3e1d8, Basalt-3e1d8, Juniper-3e1d8, Dune-3e1d8. Recorded retrieval cost scores, in the same order: 31, 74, 70, 46, 48, 16. The reference roster-3e1d8@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3e1d8 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-3e1d8@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Amber-3e1d8 on every execution, with probability one.\nP2: Assigns probability 3/8 to Heron-3e1d8 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-3e1d8 on every execution, with probability one.\nP4: Returns Dune-3e1d8, which has the unique lowest retrieval cost score, with probability one.\nP5: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P5; claims: G1, G4.", "Implementations: P1, P2, P4, P5; claims: G1, G4.", "Implementations: P5; claims: G1, G2, G4.", "Implementations: P1, P4, P5; claims: G1, G4.", "Implementations: P2, P5; claims: G1, G4.", "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: P2, P5; claims: G1, G2, G4.", "Implementations: P1, P4, P5; claims: G1." ], "answer": "Implementations: P5; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-01", "member_count": 6, "world_id": "rt1-world-05-14", "semantic_world": { "n": 6, "members": [ "Amber-3e1d8", "Heron-3e1d8", "Flint-3e1d8", "Basalt-3e1d8", "Juniper-3e1d8", "Dune-3e1d8" ], "outsider": "Excluded-3e1d8", "scores": [ 31, 74, 70, 46, 48, 16 ], "criterion": "retrieval cost", "set_ref": "roster-3e1d8@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P5; claims: G1, G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P5; claims: G1, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P5; claims: G1, G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P4, P5; claims: G1.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Amber-3e1d8 on every execution, with probability one.", "distribution": { "Amber-3e1d8": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Heron-3e1d8 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-3e1d8": "1/8", "Heron-3e1d8": "3/8", "Flint-3e1d8": "1/8", "Basalt-3e1d8": "1/8", "Juniper-3e1d8": "1/8", "Dune-3e1d8": "1/8" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-3e1d8 on every execution, with probability one.", "distribution": { "Excluded-3e1d8": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Dune-3e1d8, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Dune-3e1d8": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-3e1d8": "1/6", "Heron-3e1d8": "1/6", "Flint-3e1d8": "1/6", "Basalt-3e1d8": "1/6", "Juniper-3e1d8": "1/6", "Dune-3e1d8": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-14", "english": "Two operators are reading the same request concerning a ledger-reconciliation job and must agree what it asks.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-b7db5, Marsh-b7db5. Recorded current load scores, in the same order: 35, 19. The reference roster-b7db5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b7db5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b7db5@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-b7db5@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-b7db5@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Two operators are reading the same request concerning a ledger-reconciliation job and must agree what it asks.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-b7db5, Marsh-b7db5. Recorded current load scores, in the same order: 35, 19. The reference roster-b7db5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b7db5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b7db5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-b7db5 on every execution, with probability one.\nP2: Assigns probability 3/4 to Amber-b7db5 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Marsh-b7db5, which has the unique lowest current load score, with probability one.\nP4: Returns Amber-b7db5 on every execution, with probability one.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P5; claims: G1, G2, G4.", "Implementations: P5; claims: G1, G4.", "Implementations: P2, P3, P4, P5; claims: G1, G3.", "Implementations: P2, P3, P4; claims: G1, G3.", "Implementations: none; claims: G1, G4.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P2, P3, P4; claims: G1.", "Implementations: none; claims: G1, G2, G4." ], "answer": "Implementations: P5; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-01", "member_count": 2, "world_id": "rt1-world-06-14", "semantic_world": { "n": 2, "members": [ "Amber-b7db5", "Marsh-b7db5" ], "outsider": "Excluded-b7db5", "scores": [ 35, 19 ], "criterion": "current load", "set_ref": "roster-b7db5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P5; claims: G1, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P3, P4; claims: G1, G3.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P4; claims: G1.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: none; claims: G1, G2, G4.": { "policies": [], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-b7db5 on every execution, with probability one.", "distribution": { "Excluded-b7db5": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Amber-b7db5 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-b7db5": "3/4", "Marsh-b7db5": "1/4" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Marsh-b7db5, which has the unique lowest current load score, with probability one.", "distribution": { "Marsh-b7db5": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Amber-b7db5 on every execution, with probability one.", "distribution": { "Amber-b7db5": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-b7db5": "1/2", "Marsh-b7db5": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-15", "english": "A change request has arrived for a map-tile fetch; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Dune-8b84d, Gale-8b84d. Recorded latency scores, in the same order: 51, 32. The reference roster-8b84d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8b84d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-8b84d@snapshot-1) requires roster-8b84d@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-8b84d@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A change request has arrived for a map-tile fetch; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Dune-8b84d, Gale-8b84d. Recorded latency scores, in the same order: 51, 32. The reference roster-8b84d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8b84d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-8b84d@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-8b84d on every execution, with probability one.\nP2: Assigns probability 4/5 to Gale-8b84d and 1/5 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Gale-8b84d, which has the unique lowest latency score, with probability one.\nP4: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Dune-8b84d on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P4; claims: G1, G3, G4.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P1, P4; claims: G1, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G2.", "Implementations: P4; claims: G1, G4.", "Implementations: P4; claims: G1, G3, G4.", "Implementations: P2, P3, P4, P5; claims: G1, G2." ], "answer": "Implementations: P2, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-02", "member_count": 2, "world_id": "rt1-world-01-15", "semantic_world": { "n": 2, "members": [ "Dune-8b84d", "Gale-8b84d" ], "outsider": "Excluded-8b84d", "scores": [ 51, 32 ], "criterion": "latency", "set_ref": "roster-8b84d@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P4; claims: G1, G3, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4; claims: G1, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4; claims: G1, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P4; claims: G1, G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G2.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-8b84d on every execution, with probability one.", "distribution": { "Excluded-8b84d": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Gale-8b84d and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-8b84d": "1/5", "Gale-8b84d": "4/5" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Gale-8b84d, which has the unique lowest latency score, with probability one.", "distribution": { "Gale-8b84d": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-8b84d": "1/2", "Gale-8b84d": "1/2" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Dune-8b84d on every execution, with probability one.", "distribution": { "Dune-8b84d": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-15", "english": "A change request has arrived for a survey instrument; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Gale-24887, Iris-24887, Lumen-24887, Flint-24887, Pike-24887. Recorded queue length scores, in the same order: 86, 92, 82, 26, 18. The reference roster-24887@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-24887 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-24887@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-24887@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-24887@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A change request has arrived for a survey instrument; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Gale-24887, Iris-24887, Lumen-24887, Flint-24887, Pike-24887. Recorded queue length scores, in the same order: 86, 92, 82, 26, 18. The reference roster-24887@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-24887 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-24887@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-24887 on every execution, with probability one.\nP2: Returns Pike-24887, which has the unique lowest queue length score, with probability one.\nP3: Returns Gale-24887 on every execution, with probability one.\nP4: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 4/8 to Iris-24887 and 1/8 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P2, P3, P4, P5; claims: G1, G3.", "Implementations: P4; claims: G1, G4.", "Implementations: P2, P4, P5; claims: G1, G3.", "Implementations: P3, P4; claims: G4.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: P2, P4, P5; claims: G1.", "Implementations: P3, P4; claims: G1, G4.", "Implementations: P4; claims: G4." ], "answer": "Implementations: P4; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-02", "member_count": 5, "world_id": "rt1-world-02-15", "semantic_world": { "n": 5, "members": [ "Gale-24887", "Iris-24887", "Lumen-24887", "Flint-24887", "Pike-24887" ], "outsider": "Excluded-24887", "scores": [ 86, 92, 82, 26, 18 ], "criterion": "queue length", "set_ref": "roster-24887@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G1, G4.": { "policies": [ "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3, P4; claims: G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4, P5; claims: G1.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P3, P4; claims: G1, G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P4; claims: G4.": { "policies": [ "P4" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-24887 on every execution, with probability one.", "distribution": { "Excluded-24887": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Pike-24887, which has the unique lowest queue length score, with probability one.", "distribution": { "Pike-24887": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Gale-24887 on every execution, with probability one.", "distribution": { "Gale-24887": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-24887": "1/5", "Iris-24887": "1/5", "Lumen-24887": "1/5", "Flint-24887": "1/5", "Pike-24887": "1/5" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Iris-24887 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-24887": "1/8", "Iris-24887": "1/2", "Lumen-24887": "1/8", "Flint-24887": "1/8", "Pike-24887": "1/8" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-15", "english": "A change request has arrived for a calibration evaluation slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Heron-fda5f, Nettle-fda5f, Juniper-fda5f, Marsh-fda5f, Cobalt-fda5f, Gale-fda5f, Amber-fda5f, Iris-fda5f. Recorded preparation cost scores, in the same order: 49, 37, 59, 51, 7, 92, 5, 44. The reference roster-fda5f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fda5f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fda5f@snapshot-1) requires roster-fda5f@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-fda5f@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A change request has arrived for a calibration evaluation slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Heron-fda5f, Nettle-fda5f, Juniper-fda5f, Marsh-fda5f, Cobalt-fda5f, Gale-fda5f, Amber-fda5f, Iris-fda5f. Recorded preparation cost scores, in the same order: 49, 37, 59, 51, 7, 92, 5, 44. The reference roster-fda5f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fda5f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fda5f@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-fda5f on every execution, with probability one.\nP2: Returns Heron-fda5f on every execution, with probability one.\nP3: Assigns probability 4/11 to Gale-fda5f and 1/11 to each other listed identity, then returns one draw from that distribution.\nP4: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Amber-fda5f, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P4, P5; claims: G3, G4.", "Implementations: P4; claims: G3.", "Implementations: P2, P3, P4, P5; claims: G3.", "Implementations: P4, P5; claims: G3.", "Implementations: P4; claims: G3, G4.", "Implementations: P2, P3, P4; claims: G3.", "Implementations: P2, P3, P4, P5; claims: none.", "Implementations: P2, P3, P4; claims: none." ], "answer": "Implementations: P2, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-02", "member_count": 8, "world_id": "rt1-world-03-15", "semantic_world": { "n": 8, "members": [ "Heron-fda5f", "Nettle-fda5f", "Juniper-fda5f", "Marsh-fda5f", "Cobalt-fda5f", "Gale-fda5f", "Amber-fda5f", "Iris-fda5f" ], "outsider": "Excluded-fda5f", "scores": [ 49, 37, 59, 51, 7, 92, 5, 44 ], "criterion": "preparation cost", "set_ref": "roster-fda5f@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4, P5; claims: G3, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P4; claims: G3.": { "policies": [ "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4, P5; claims: G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G3, G4.": { "policies": [ "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3, P4; claims: G3.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3, P4, P5; claims: none.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P2, P3, P4; claims: none.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-fda5f on every execution, with probability one.", "distribution": { "Excluded-fda5f": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Heron-fda5f on every execution, with probability one.", "distribution": { "Heron-fda5f": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/11 to Gale-fda5f and 1/11 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-fda5f": "1/11", "Nettle-fda5f": "1/11", "Juniper-fda5f": "1/11", "Marsh-fda5f": "1/11", "Cobalt-fda5f": "1/11", "Gale-fda5f": "4/11", "Amber-fda5f": "1/11", "Iris-fda5f": "1/11" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-fda5f": "1/8", "Nettle-fda5f": "1/8", "Juniper-fda5f": "1/8", "Marsh-fda5f": "1/8", "Cobalt-fda5f": "1/8", "Gale-fda5f": "1/8", "Amber-fda5f": "1/8", "Iris-fda5f": "1/8" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Amber-fda5f, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Amber-fda5f": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-15", "english": "A change request has arrived for a crashed scheduler; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible standby node identities, in recorded order: Juniper-70937, Lumen-70937, Nettle-70937, Heron-70937. Recorded activation cost scores, in the same order: 63, 3, 92, 94. The reference roster-70937@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-70937 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-70937@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-70937@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-70937@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A change request has arrived for a crashed scheduler; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible standby node identities, in recorded order: Juniper-70937, Lumen-70937, Nettle-70937, Heron-70937. Recorded activation cost scores, in the same order: 63, 3, 92, 94. The reference roster-70937@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-70937 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-70937@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/7 to Juniper-70937 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Juniper-70937 on every execution, with probability one.\nP4: Returns Lumen-70937, which has the unique lowest activation cost score, with probability one.\nP5: Returns Excluded-70937 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P2, P3, P4; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P2, P3, P4; claims: G3, G4.", "Implementations: P2; claims: G2, G3, G4.", "Implementations: P2; claims: G3, G4.", "Implementations: P1, P2; claims: G3, G4.", "Implementations: P1, P2; claims: G2, G3, G4.", "Implementations: P1, P2, P3, P4; claims: G3, G4." ], "answer": "Implementations: P2; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-02", "member_count": 4, "world_id": "rt1-world-04-15", "semantic_world": { "n": 4, "members": [ "Juniper-70937", "Lumen-70937", "Nettle-70937", "Heron-70937" ], "outsider": "Excluded-70937", "scores": [ 63, 3, 92, 94 ], "criterion": "activation cost", "set_ref": "roster-70937@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3, P4; claims: G3.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3, P4; claims: G3, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2; claims: G2, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P2; claims: G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2; claims: G3, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2; claims: G2, G3, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Juniper-70937 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-70937": "4/7", "Lumen-70937": "1/7", "Nettle-70937": "1/7", "Heron-70937": "1/7" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-70937": "1/4", "Lumen-70937": "1/4", "Nettle-70937": "1/4", "Heron-70937": "1/4" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Juniper-70937 on every execution, with probability one.", "distribution": { "Juniper-70937": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Lumen-70937, which has the unique lowest activation cost score, with probability one.", "distribution": { "Lumen-70937": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-70937 on every execution, with probability one.", "distribution": { "Excluded-70937": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-15", "english": "A change request has arrived for a help-centre example slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible content item identities, in recorded order: Iris-3599f, Nettle-3599f, Marsh-3599f, Juniper-3599f, Heron-3599f, Lumen-3599f, Gale-3599f. Recorded retrieval cost scores, in the same order: 53, 80, 24, 35, 55, 79, 28. The reference roster-3599f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3599f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-3599f@snapshot-1) requires roster-3599f@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-3599f@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A change request has arrived for a help-centre example slot; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible content item identities, in recorded order: Iris-3599f, Nettle-3599f, Marsh-3599f, Juniper-3599f, Heron-3599f, Lumen-3599f, Gale-3599f. Recorded retrieval cost scores, in the same order: 53, 80, 24, 35, 55, 79, 28. The reference roster-3599f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-3599f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-3599f@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-3599f, which has the unique lowest retrieval cost score, with probability one.\nP2: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-3599f on every execution, with probability one.\nP4: Returns Iris-3599f on every execution, with probability one.\nP5: Assigns probability 4/10 to Iris-3599f and 1/10 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: P1, P2, P4, P5; claims: G1, G3.", "Implementations: P1, P4, P5; claims: G1.", "Implementations: none; claims: G1, G2.", "Implementations: P2; claims: G1, G2.", "Implementations: none; claims: G1, G2, G4.", "Implementations: P1, P4, P5; claims: G1, G3.", "Implementations: P2; claims: G1, G2, G4." ], "answer": "Implementations: P1, P2, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-02", "member_count": 7, "world_id": "rt1-world-05-15", "semantic_world": { "n": 7, "members": [ "Iris-3599f", "Nettle-3599f", "Marsh-3599f", "Juniper-3599f", "Heron-3599f", "Lumen-3599f", "Gale-3599f" ], "outsider": "Excluded-3599f", "scores": [ 53, 80, 24, 35, 55, 79, 28 ], "criterion": "retrieval cost", "set_ref": "roster-3599f@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P4, P5; claims: G1.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: none; claims: G1, G2.": { "policies": [], "guarantees": [ "G1", "G2" ] }, "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: none; claims: G1, G2, G4.": { "policies": [], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2; claims: G1, G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Marsh-3599f, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Marsh-3599f": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Iris-3599f": "1/7", "Nettle-3599f": "1/7", "Marsh-3599f": "1/7", "Juniper-3599f": "1/7", "Heron-3599f": "1/7", "Lumen-3599f": "1/7", "Gale-3599f": "1/7" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-3599f on every execution, with probability one.", "distribution": { "Excluded-3599f": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Iris-3599f on every execution, with probability one.", "distribution": { "Iris-3599f": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Iris-3599f and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Iris-3599f": "2/5", "Nettle-3599f": "1/10", "Marsh-3599f": "1/10", "Juniper-3599f": "1/10", "Heron-3599f": "1/10", "Lumen-3599f": "1/10", "Gale-3599f": "1/10" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-15", "english": "A change request has arrived for a spell-check job; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible worker identities, in recorded order: Marsh-710c5, Cobalt-710c5, Flint-710c5. Recorded current load scores, in the same order: 48, 19, 71. The reference roster-710c5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-710c5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-710c5@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-710c5@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-710c5@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A change request has arrived for a spell-check job; the on-call engineer is checking what it permits.\nThe frozen roster lists these eligible worker identities, in recorded order: Marsh-710c5, Cobalt-710c5, Flint-710c5. Recorded current load scores, in the same order: 48, 19, 71. The reference roster-710c5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-710c5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-710c5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-710c5 on every execution, with probability one.\nP3: Returns Marsh-710c5 on every execution, with probability one.\nP4: Returns Cobalt-710c5, which has the unique lowest current load score, with probability one.\nP5: Assigns probability 4/6 to Cobalt-710c5 and 1/6 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1; claims: G4.", "Implementations: P1, P3, P4, P5; claims: G3.", "Implementations: P1, P2, P3, P4, P5; claims: G2, G3.", "Implementations: P1; claims: G3, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P1, P2; claims: G3, G4.", "Implementations: P1, P2; claims: G4.", "Implementations: P1, P3, P4, P5; claims: G2, G3." ], "answer": "Implementations: P1; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-02", "member_count": 3, "world_id": "rt1-world-06-15", "semantic_world": { "n": 3, "members": [ "Marsh-710c5", "Cobalt-710c5", "Flint-710c5" ], "outsider": "Excluded-710c5", "scores": [ 48, 19, 71 ], "criterion": "current load", "set_ref": "roster-710c5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G4.": { "policies": [ "P1" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1; claims: G3, G4.": { "policies": [ "P1" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2; claims: G3, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2; claims: G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-710c5": "1/3", "Cobalt-710c5": "1/3", "Flint-710c5": "1/3" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-710c5 on every execution, with probability one.", "distribution": { "Excluded-710c5": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Marsh-710c5 on every execution, with probability one.", "distribution": { "Marsh-710c5": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Cobalt-710c5, which has the unique lowest current load score, with probability one.", "distribution": { "Cobalt-710c5": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Cobalt-710c5 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-710c5": "1/6", "Cobalt-710c5": "2/3", "Flint-710c5": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-16", "english": "A checklist item asks what the selection wording for a session-refresh call actually commits to.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Pike-78e18, Heron-78e18, Cobalt-78e18. Recorded latency scores, in the same order: 29, 52, 19. The reference roster-78e18@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-78e18 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-78e18@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-78e18@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-78e18@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A checklist item asks what the selection wording for a session-refresh call actually commits to.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Pike-78e18, Heron-78e18, Cobalt-78e18. Recorded latency scores, in the same order: 29, 52, 19. The reference roster-78e18@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-78e18 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-78e18@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-78e18, which has the unique lowest latency score, with probability one.\nP2: Assigns probability 2/4 to Heron-78e18 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Pike-78e18 on every execution, with probability one.\nP4: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-78e18 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P3, P4; claims: G3.", "Implementations: P4; claims: G3.", "Implementations: P4; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1, P2, P4; claims: G1, G4.", "Implementations: P1, P2, P3, P4; claims: G1, G4.", "Implementations: P3, P4; claims: G1, G3.", "Implementations: P1, P2, P4; claims: G1." ], "answer": "Implementations: P4; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-03", "member_count": 3, "world_id": "rt1-world-01-16", "semantic_world": { "n": 3, "members": [ "Pike-78e18", "Heron-78e18", "Cobalt-78e18" ], "outsider": "Excluded-78e18", "scores": [ 29, 52, 19 ], "criterion": "latency", "set_ref": "roster-78e18@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4; claims: G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G3.": { "policies": [ "P4" ], "guarantees": [ "G3" ] }, "Implementations: P4; claims: G1, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P3, P4; claims: G1, G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P4; claims: G1.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Cobalt-78e18, which has the unique lowest latency score, with probability one.", "distribution": { "Cobalt-78e18": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Heron-78e18 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Pike-78e18": "1/4", "Heron-78e18": "1/2", "Cobalt-78e18": "1/4" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Pike-78e18 on every execution, with probability one.", "distribution": { "Pike-78e18": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Pike-78e18": "1/3", "Heron-78e18": "1/3", "Cobalt-78e18": "1/3" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-78e18 on every execution, with probability one.", "distribution": { "Excluded-78e18": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-16", "english": "A checklist item asks what the selection wording for a preprint revision actually commits to.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Dune-53bde, Juniper-53bde, Gale-53bde, Amber-53bde, Flint-53bde, Lumen-53bde. Recorded queue length scores, in the same order: 77, 28, 15, 67, 59, 76. The reference roster-53bde@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-53bde is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-53bde@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-53bde@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-53bde@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A checklist item asks what the selection wording for a preprint revision actually commits to.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Dune-53bde, Juniper-53bde, Gale-53bde, Amber-53bde, Flint-53bde, Lumen-53bde. Recorded queue length scores, in the same order: 77, 28, 15, 67, 59, 76. The reference roster-53bde@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-53bde is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-53bde@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-53bde on every execution, with probability one.\nP3: Returns Dune-53bde on every execution, with probability one.\nP4: Assigns probability 2/7 to Flint-53bde and 1/7 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Gale-53bde, which has the unique lowest queue length score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1; claims: G2, G3.", "Implementations: P1, P3, P4; claims: G3.", "Implementations: P1, P3, P4, P5; claims: G3.", "Implementations: P1, P5; claims: G2, G3.", "Implementations: P1, P5; claims: G3.", "Implementations: P1, P3, P4; claims: none.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P1; claims: G3." ], "answer": "Implementations: P1; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-03", "member_count": 6, "world_id": "rt1-world-02-16", "semantic_world": { "n": 6, "members": [ "Dune-53bde", "Juniper-53bde", "Gale-53bde", "Amber-53bde", "Flint-53bde", "Lumen-53bde" ], "outsider": "Excluded-53bde", "scores": [ 77, 28, 15, 67, 59, 76 ], "criterion": "queue length", "set_ref": "roster-53bde@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G2, G3.": { "policies": [ "P1" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3, P4; claims: G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P5; claims: G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P5; claims: G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P4; claims: none.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1; claims: G3.": { "policies": [ "P1" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-53bde": "1/6", "Juniper-53bde": "1/6", "Gale-53bde": "1/6", "Amber-53bde": "1/6", "Flint-53bde": "1/6", "Lumen-53bde": "1/6" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-53bde on every execution, with probability one.", "distribution": { "Excluded-53bde": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Dune-53bde on every execution, with probability one.", "distribution": { "Dune-53bde": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/7 to Flint-53bde and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-53bde": "1/7", "Juniper-53bde": "1/7", "Gale-53bde": "1/7", "Amber-53bde": "1/7", "Flint-53bde": "2/7", "Lumen-53bde": "1/7" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Gale-53bde, which has the unique lowest queue length score, with probability one.", "distribution": { "Gale-53bde": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-16", "english": "A checklist item asks what the selection wording for a coding evaluation slot actually commits to.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Lumen-f4597, Basalt-f4597. Recorded preparation cost scores, in the same order: 69, 51. The reference roster-f4597@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f4597 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-f4597@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-f4597@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-f4597@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A checklist item asks what the selection wording for a coding evaluation slot actually commits to.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Lumen-f4597, Basalt-f4597. Recorded preparation cost scores, in the same order: 69, 51. The reference roster-f4597@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f4597 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-f4597@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Lumen-f4597 on every execution, with probability one.\nP2: Returns Basalt-f4597, which has the unique lowest preparation cost score, with probability one.\nP3: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-f4597 on every execution, with probability one.\nP5: Assigns probability 2/3 to Lumen-f4597 and 1/3 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P3, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: P1, P2, P3; claims: G3.", "Implementations: P1, P2, P3; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P3; claims: G1, G2, G3.", "Implementations: P3; claims: G1, G3.", "Implementations: P3, P5; claims: G1, G2, G3." ], "answer": "Implementations: P3; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-03", "member_count": 2, "world_id": "rt1-world-03-16", "semantic_world": { "n": 2, "members": [ "Lumen-f4597", "Basalt-f4597" ], "outsider": "Excluded-f4597", "scores": [ 69, 51 ], "criterion": "preparation cost", "set_ref": "roster-f4597@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P5; claims: G1, G3.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3; claims: G3.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3; claims: G1, G3.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P3; claims: G1, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3, P5; claims: G1, G2, G3.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Lumen-f4597 on every execution, with probability one.", "distribution": { "Lumen-f4597": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Basalt-f4597, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Basalt-f4597": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Lumen-f4597": "1/2", "Basalt-f4597": "1/2" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-f4597 on every execution, with probability one.", "distribution": { "Excluded-f4597": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Lumen-f4597 and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Lumen-f4597": "2/3", "Basalt-f4597": "1/3" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-16", "english": "A checklist item asks what the selection wording for a paused webhook relay actually commits to.\nThe frozen roster lists these eligible standby node identities, in recorded order: Nettle-1e722, Onyx-1e722, Kestrel-1e722, Pike-1e722, Iris-1e722. Recorded activation cost scores, in the same order: 26, 63, 2, 73, 7. The reference roster-1e722@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1e722 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-1e722@snapshot-1) requires roster-1e722@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-1e722@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a paused webhook relay actually commits to.\nThe frozen roster lists these eligible standby node identities, in recorded order: Nettle-1e722, Onyx-1e722, Kestrel-1e722, Pike-1e722, Iris-1e722. Recorded activation cost scores, in the same order: 26, 63, 2, 73, 7. The reference roster-1e722@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1e722 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-1e722@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-1e722 on every execution, with probability one.\nP2: Assigns probability 2/6 to Iris-1e722 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP3: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Nettle-1e722 on every execution, with probability one.\nP5: Returns Kestrel-1e722, which has the unique lowest activation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: none; claims: G1, G2, G4.", "Implementations: P2, P4, P5; claims: G1.", "Implementations: P2, P4, P5; claims: G1, G3.", "Implementations: P2, P3, P4, P5; claims: G1, G3.", "Implementations: P3; claims: G1, G2.", "Implementations: P2, P3, P4, P5; claims: G1.", "Implementations: none; claims: G1, G2.", "Implementations: P3; claims: G1, G2, G4." ], "answer": "Implementations: P2, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-03", "member_count": 5, "world_id": "rt1-world-04-16", "semantic_world": { "n": 5, "members": [ "Nettle-1e722", "Onyx-1e722", "Kestrel-1e722", "Pike-1e722", "Iris-1e722" ], "outsider": "Excluded-1e722", "scores": [ 26, 63, 2, 73, 7 ], "criterion": "activation cost", "set_ref": "roster-1e722@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G1, G2, G4.": { "policies": [], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P2, P4, P5; claims: G1.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G1.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: none; claims: G1, G2.": { "policies": [], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-1e722 on every execution, with probability one.", "distribution": { "Excluded-1e722": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Iris-1e722 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-1e722": "1/6", "Onyx-1e722": "1/6", "Kestrel-1e722": "1/6", "Pike-1e722": "1/6", "Iris-1e722": "1/3" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-1e722": "1/5", "Onyx-1e722": "1/5", "Kestrel-1e722": "1/5", "Pike-1e722": "1/5", "Iris-1e722": "1/5" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Nettle-1e722 on every execution, with probability one.", "distribution": { "Nettle-1e722": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Kestrel-1e722, which has the unique lowest activation cost score, with probability one.", "distribution": { "Kestrel-1e722": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-16", "english": "A checklist item asks what the selection wording for a release-note screenshot slot actually commits to.\nThe frozen roster lists these eligible content item identities, in recorded order: Juniper-c994f, Kestrel-c994f, Marsh-c994f, Pike-c994f, Flint-c994f, Gale-c994f, Nettle-c994f, Iris-c994f. Recorded retrieval cost scores, in the same order: 67, 89, 21, 93, 69, 60, 86, 13. The reference roster-c994f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c994f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-c994f@snapshot-1) requires roster-c994f@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-c994f@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A checklist item asks what the selection wording for a release-note screenshot slot actually commits to.\nThe frozen roster lists these eligible content item identities, in recorded order: Juniper-c994f, Kestrel-c994f, Marsh-c994f, Pike-c994f, Flint-c994f, Gale-c994f, Nettle-c994f, Iris-c994f. Recorded retrieval cost scores, in the same order: 67, 89, 21, 93, 69, 60, 86, 13. The reference roster-c994f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c994f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-c994f@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/9 to Nettle-c994f and 1/9 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Juniper-c994f on every execution, with probability one.\nP4: Returns Excluded-c994f on every execution, with probability one.\nP5: Returns Iris-c994f, which has the unique lowest retrieval cost score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2; claims: G1, G2.", "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G2.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: P2, P4; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P2; claims: G1.", "Implementations: P2, P4; claims: G1." ], "answer": "Implementations: P1, P2, P3, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-03", "member_count": 8, "world_id": "rt1-world-05-16", "semantic_world": { "n": 8, "members": [ "Juniper-c994f", "Kestrel-c994f", "Marsh-c994f", "Pike-c994f", "Flint-c994f", "Gale-c994f", "Nettle-c994f", "Iris-c994f" ], "outsider": "Excluded-c994f", "scores": [ 67, 89, 21, 93, 69, 60, 86, 13 ], "criterion": "retrieval cost", "set_ref": "roster-c994f@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P4; claims: G1, G2.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P2; claims: G1.": { "policies": [ "P2" ], "guarantees": [ "G1" ] }, "Implementations: P2, P4; claims: G1.": { "policies": [ "P2", "P4" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Nettle-c994f and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-c994f": "1/9", "Kestrel-c994f": "1/9", "Marsh-c994f": "1/9", "Pike-c994f": "1/9", "Flint-c994f": "1/9", "Gale-c994f": "1/9", "Nettle-c994f": "2/9", "Iris-c994f": "1/9" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-c994f": "1/8", "Kestrel-c994f": "1/8", "Marsh-c994f": "1/8", "Pike-c994f": "1/8", "Flint-c994f": "1/8", "Gale-c994f": "1/8", "Nettle-c994f": "1/8", "Iris-c994f": "1/8" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Juniper-c994f on every execution, with probability one.", "distribution": { "Juniper-c994f": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-c994f on every execution, with probability one.", "distribution": { "Excluded-c994f": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Iris-c994f, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Iris-c994f": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-16", "english": "A checklist item asks what the selection wording for a geocoding job actually commits to.\nThe frozen roster lists these eligible worker identities, in recorded order: Iris-9c4a6, Basalt-9c4a6, Onyx-9c4a6, Lumen-9c4a6. Recorded current load scores, in the same order: 24, 19, 51, 62. The reference roster-9c4a6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9c4a6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-9c4a6@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-9c4a6@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-9c4a6@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A checklist item asks what the selection wording for a geocoding job actually commits to.\nThe frozen roster lists these eligible worker identities, in recorded order: Iris-9c4a6, Basalt-9c4a6, Onyx-9c4a6, Lumen-9c4a6. Recorded current load scores, in the same order: 24, 19, 51, 62. The reference roster-9c4a6@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-9c4a6 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-9c4a6@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-9c4a6 on every execution, with probability one.\nP3: Returns Basalt-9c4a6, which has the unique lowest current load score, with probability one.\nP4: Returns Iris-9c4a6 on every execution, with probability one.\nP5: Assigns probability 2/5 to Onyx-9c4a6 and 1/5 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P3, P5; claims: none.", "Implementations: P1; claims: G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P1; claims: G2, G4.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P1, P4; claims: G2, G4.", "Implementations: P1, P4; claims: G4.", "Implementations: P1, P3, P5; claims: G4." ], "answer": "Implementations: P1; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-03", "member_count": 4, "world_id": "rt1-world-06-16", "semantic_world": { "n": 4, "members": [ "Iris-9c4a6", "Basalt-9c4a6", "Onyx-9c4a6", "Lumen-9c4a6" ], "outsider": "Excluded-9c4a6", "scores": [ 24, 19, 51, 62 ], "criterion": "current load", "set_ref": "roster-9c4a6@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P5; claims: none.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [] }, "Implementations: P1; claims: G4.": { "policies": [ "P1" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1; claims: G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P4; claims: G2, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P4; claims: G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P5; claims: G4.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Iris-9c4a6": "1/4", "Basalt-9c4a6": "1/4", "Onyx-9c4a6": "1/4", "Lumen-9c4a6": "1/4" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-9c4a6 on every execution, with probability one.", "distribution": { "Excluded-9c4a6": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Basalt-9c4a6, which has the unique lowest current load score, with probability one.", "distribution": { "Basalt-9c4a6": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Iris-9c4a6 on every execution, with probability one.", "distribution": { "Iris-9c4a6": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Onyx-9c4a6 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Iris-9c4a6": "1/5", "Basalt-9c4a6": "1/5", "Onyx-9c4a6": "2/5", "Lumen-9c4a6": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-17", "english": "Before running anything, a planner writes down what the request about a price-quote request requires.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-f82ed, Juniper-f82ed, Lumen-f82ed, Kestrel-f82ed. Recorded latency scores, in the same order: 37, 89, 21, 70. The reference roster-f82ed@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f82ed is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f82ed@snapshot-1) requires roster-f82ed@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-f82ed@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "Before running anything, a planner writes down what the request about a price-quote request requires.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-f82ed, Juniper-f82ed, Lumen-f82ed, Kestrel-f82ed. Recorded latency scores, in the same order: 37, 89, 21, 70. The reference roster-f82ed@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f82ed is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f82ed@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Flint-f82ed on every execution, with probability one.\nP2: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-f82ed on every execution, with probability one.\nP4: Assigns probability 3/6 to Flint-f82ed and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Lumen-f82ed, which has the unique lowest latency score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P4; claims: G3.", "Implementations: P2, P5; claims: G3, G4.", "Implementations: P2; claims: G3.", "Implementations: P1, P2, P4, P5; claims: none.", "Implementations: P1, P2, P4, P5; claims: G3.", "Implementations: P1, P2, P4; claims: none.", "Implementations: P2; claims: G3, G4.", "Implementations: P2, P5; claims: G3." ], "answer": "Implementations: P1, P2, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-04", "member_count": 4, "world_id": "rt1-world-01-17", "semantic_world": { "n": 4, "members": [ "Flint-f82ed", "Juniper-f82ed", "Lumen-f82ed", "Kestrel-f82ed" ], "outsider": "Excluded-f82ed", "scores": [ 37, 89, 21, 70 ], "criterion": "latency", "set_ref": "roster-f82ed@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4; claims: G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2, P5; claims: G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2; claims: G3.": { "policies": [ "P2" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P4; claims: none.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [] }, "Implementations: P2; claims: G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P5; claims: G3.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Flint-f82ed on every execution, with probability one.", "distribution": { "Flint-f82ed": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-f82ed": "1/4", "Juniper-f82ed": "1/4", "Lumen-f82ed": "1/4", "Kestrel-f82ed": "1/4" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-f82ed on every execution, with probability one.", "distribution": { "Excluded-f82ed": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Flint-f82ed and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-f82ed": "1/2", "Juniper-f82ed": "1/6", "Lumen-f82ed": "1/6", "Kestrel-f82ed": "1/6" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Lumen-f82ed, which has the unique lowest latency score, with probability one.", "distribution": { "Lumen-f82ed": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-17", "english": "Before running anything, a planner writes down what the request about a thesis chapter requires.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Lumen-1ef1f, Gale-1ef1f, Marsh-1ef1f, Heron-1ef1f, Cobalt-1ef1f, Elm-1ef1f, Amber-1ef1f. Recorded queue length scores, in the same order: 56, 55, 48, 42, 73, 27, 9. The reference roster-1ef1f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1ef1f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-1ef1f@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-1ef1f@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-1ef1f@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a thesis chapter requires.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Lumen-1ef1f, Gale-1ef1f, Marsh-1ef1f, Heron-1ef1f, Cobalt-1ef1f, Elm-1ef1f, Amber-1ef1f. Recorded queue length scores, in the same order: 56, 55, 48, 42, 73, 27, 9. The reference roster-1ef1f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1ef1f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-1ef1f@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Lumen-1ef1f on every execution, with probability one.\nP2: Returns Amber-1ef1f, which has the unique lowest queue length score, with probability one.\nP3: Returns Excluded-1ef1f on every execution, with probability one.\nP4: Assigns probability 3/9 to Cobalt-1ef1f and 1/9 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P5; claims: G2.", "Implementations: P4, P5; claims: G1, G2.", "Implementations: P1, P2, P4, P5; claims: G2.", "Implementations: P1, P2, P4, P5; claims: G1, G2.", "Implementations: P5; claims: G1, G2, G4.", "Implementations: P1, P2, P5; claims: G1, G2.", "Implementations: P5; claims: G1, G2.", "Implementations: P4, P5; claims: G1, G2, G4." ], "answer": "Implementations: P5; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-04", "member_count": 7, "world_id": "rt1-world-02-17", "semantic_world": { "n": 7, "members": [ "Lumen-1ef1f", "Gale-1ef1f", "Marsh-1ef1f", "Heron-1ef1f", "Cobalt-1ef1f", "Elm-1ef1f", "Amber-1ef1f" ], "outsider": "Excluded-1ef1f", "scores": [ 56, 55, 48, 42, 73, 27, 9 ], "criterion": "queue length", "set_ref": "roster-1ef1f@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P5; claims: G2.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P4, P5; claims: G1, G2.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P5; claims: G1, G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P2, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P5; claims: G1, G2.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4, P5; claims: G1, G2, G4.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G2", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Lumen-1ef1f on every execution, with probability one.", "distribution": { "Lumen-1ef1f": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Amber-1ef1f, which has the unique lowest queue length score, with probability one.", "distribution": { "Amber-1ef1f": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-1ef1f on every execution, with probability one.", "distribution": { "Excluded-1ef1f": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/9 to Cobalt-1ef1f and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Lumen-1ef1f": "1/9", "Gale-1ef1f": "1/9", "Marsh-1ef1f": "1/9", "Heron-1ef1f": "1/9", "Cobalt-1ef1f": "1/3", "Elm-1ef1f": "1/9", "Amber-1ef1f": "1/9" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Lumen-1ef1f": "1/7", "Gale-1ef1f": "1/7", "Marsh-1ef1f": "1/7", "Heron-1ef1f": "1/7", "Cobalt-1ef1f": "1/7", "Elm-1ef1f": "1/7", "Amber-1ef1f": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-17", "english": "Before running anything, a planner writes down what the request about a citation evaluation slot requires.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Flint-b3288, Basalt-b3288, Dune-b3288. Recorded preparation cost scores, in the same order: 60, 54, 45. The reference roster-b3288@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b3288 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b3288@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-b3288@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-b3288@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a citation evaluation slot requires.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Flint-b3288, Basalt-b3288, Dune-b3288. Recorded preparation cost scores, in the same order: 60, 54, 45. The reference roster-b3288@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b3288 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b3288@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Dune-b3288, which has the unique lowest preparation cost score, with probability one.\nP2: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Excluded-b3288 on every execution, with probability one.\nP4: Returns Flint-b3288 on every execution, with probability one.\nP5: Assigns probability 3/5 to Basalt-b3288 and 1/5 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P4, P5; claims: G4.", "Implementations: P1, P4, P5; claims: G4.", "Implementations: none; claims: G1, G3, G4.", "Implementations: P1, P2, P4, P5; claims: G2, G4.", "Implementations: none; claims: G1, G4.", "Implementations: P1, P4, P5; claims: G2, G4.", "Implementations: P2; claims: G1, G4.", "Implementations: P2; claims: G1, G3, G4." ], "answer": "Implementations: P2; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-04", "member_count": 3, "world_id": "rt1-world-03-17", "semantic_world": { "n": 3, "members": [ "Flint-b3288", "Basalt-b3288", "Dune-b3288" ], "outsider": "Excluded-b3288", "scores": [ 60, 54, 45 ], "criterion": "preparation cost", "set_ref": "roster-b3288@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P4, P5; claims: G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Dune-b3288, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Dune-b3288": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-b3288": "1/3", "Basalt-b3288": "1/3", "Dune-b3288": "1/3" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-b3288 on every execution, with probability one.", "distribution": { "Excluded-b3288": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Flint-b3288 on every execution, with probability one.", "distribution": { "Flint-b3288": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Basalt-b3288 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-b3288": "1/5", "Basalt-b3288": "3/5", "Dune-b3288": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-17", "english": "Before running anything, a planner writes down what the request about a degraded search shard requires.\nThe frozen roster lists these eligible standby node identities, in recorded order: Onyx-ac782, Dune-ac782, Heron-ac782, Kestrel-ac782, Iris-ac782, Flint-ac782. Recorded activation cost scores, in the same order: 47, 96, 88, 80, 83, 30. The reference roster-ac782@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ac782 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-ac782@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-ac782@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-ac782@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a degraded search shard requires.\nThe frozen roster lists these eligible standby node identities, in recorded order: Onyx-ac782, Dune-ac782, Heron-ac782, Kestrel-ac782, Iris-ac782, Flint-ac782. Recorded activation cost scores, in the same order: 47, 96, 88, 80, 83, 30. The reference roster-ac782@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ac782 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-ac782@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Flint-ac782, which has the unique lowest activation cost score, with probability one.\nP2: Assigns probability 3/8 to Iris-ac782 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Onyx-ac782 on every execution, with probability one.\nP5: Returns Excluded-ac782 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P3; claims: G1, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P3; claims: G1.", "Implementations: P3, P5; claims: G1, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.", "Implementations: P3, P5; claims: G1." ], "answer": "Implementations: P3; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-04", "member_count": 6, "world_id": "rt1-world-04-17", "semantic_world": { "n": 6, "members": [ "Onyx-ac782", "Dune-ac782", "Heron-ac782", "Kestrel-ac782", "Iris-ac782", "Flint-ac782" ], "outsider": "Excluded-ac782", "scores": [ 47, 96, 88, 80, 83, 30 ], "criterion": "activation cost", "set_ref": "roster-ac782@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3; claims: G1, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P3; claims: G1.": { "policies": [ "P3" ], "guarantees": [ "G1" ] }, "Implementations: P3, P5; claims: G1, G4.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P3, P5; claims: G1.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Flint-ac782, which has the unique lowest activation cost score, with probability one.", "distribution": { "Flint-ac782": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Iris-ac782 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-ac782": "1/8", "Dune-ac782": "1/8", "Heron-ac782": "1/8", "Kestrel-ac782": "1/8", "Iris-ac782": "3/8", "Flint-ac782": "1/8" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-ac782": "1/6", "Dune-ac782": "1/6", "Heron-ac782": "1/6", "Kestrel-ac782": "1/6", "Iris-ac782": "1/6", "Flint-ac782": "1/6" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Onyx-ac782 on every execution, with probability one.", "distribution": { "Onyx-ac782": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-ac782 on every execution, with probability one.", "distribution": { "Excluded-ac782": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-17", "english": "Before running anything, a planner writes down what the request about a podcast intro slot requires.\nThe frozen roster lists these eligible content item identities, in recorded order: Marsh-a1d6d, Kestrel-a1d6d. Recorded retrieval cost scores, in the same order: 44, 17. The reference roster-a1d6d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a1d6d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-a1d6d@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-a1d6d@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-a1d6d@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a podcast intro slot requires.\nThe frozen roster lists these eligible content item identities, in recorded order: Marsh-a1d6d, Kestrel-a1d6d. Recorded retrieval cost scores, in the same order: 44, 17. The reference roster-a1d6d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a1d6d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-a1d6d@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Kestrel-a1d6d, which has the unique lowest retrieval cost score, with probability one.\nP2: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Marsh-a1d6d on every execution, with probability one.\nP4: Assigns probability 3/4 to Kestrel-a1d6d and 1/4 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-a1d6d on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P2, P3; claims: G1, G4.", "Implementations: P1, P2, P4; claims: G1.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P2; claims: G1.", "Implementations: P1, P2, P4; claims: none.", "Implementations: P2, P3; claims: G1.", "Implementations: P2; claims: G1, G4.", "Implementations: P1, P2, P3, P4; claims: none." ], "answer": "Implementations: P2; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-04", "member_count": 2, "world_id": "rt1-world-05-17", "semantic_world": { "n": 2, "members": [ "Marsh-a1d6d", "Kestrel-a1d6d" ], "outsider": "Excluded-a1d6d", "scores": [ 44, 17 ], "criterion": "retrieval cost", "set_ref": "roster-a1d6d@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3; claims: G1, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P4; claims: G1.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2; claims: G1.": { "policies": [ "P2" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4; claims: none.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [] }, "Implementations: P2, P3; claims: G1.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: none.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Kestrel-a1d6d, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Kestrel-a1d6d": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-a1d6d": "1/2", "Kestrel-a1d6d": "1/2" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Marsh-a1d6d on every execution, with probability one.", "distribution": { "Marsh-a1d6d": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Kestrel-a1d6d and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-a1d6d": "1/4", "Kestrel-a1d6d": "3/4" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-a1d6d on every execution, with probability one.", "distribution": { "Excluded-a1d6d": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-17", "english": "Before running anything, a planner writes down what the request about a log-rotation job requires.\nThe frozen roster lists these eligible worker identities, in recorded order: Onyx-88c05, Basalt-88c05, Cobalt-88c05, Marsh-88c05, Dune-88c05. Recorded current load scores, in the same order: 96, 58, 34, 75, 56. The reference roster-88c05@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-88c05 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-88c05@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-88c05@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-88c05@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "Before running anything, a planner writes down what the request about a log-rotation job requires.\nThe frozen roster lists these eligible worker identities, in recorded order: Onyx-88c05, Basalt-88c05, Cobalt-88c05, Marsh-88c05, Dune-88c05. Recorded current load scores, in the same order: 96, 58, 34, 75, 56. The reference roster-88c05@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-88c05 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-88c05@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 3/7 to Marsh-88c05 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Onyx-88c05 on every execution, with probability one.\nP3: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Cobalt-88c05, which has the unique lowest current load score, with probability one.\nP5: Returns Excluded-88c05 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3; claims: G1, G2.", "Implementations: P3, P4; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G1, G2.", "Implementations: P3; claims: G1, G2, G3.", "Implementations: P1, P2, P3; claims: G2.", "Implementations: P3, P4; claims: G1, G2, G3.", "Implementations: P3; claims: G1, G2.", "Implementations: P1, P2, P3, P4; claims: G2." ], "answer": "Implementations: P3; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-04", "member_count": 5, "world_id": "rt1-world-06-17", "semantic_world": { "n": 5, "members": [ "Onyx-88c05", "Basalt-88c05", "Cobalt-88c05", "Marsh-88c05", "Dune-88c05" ], "outsider": "Excluded-88c05", "scores": [ 96, 58, 34, 75, 56 ], "criterion": "current load", "set_ref": "roster-88c05@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3; claims: G1, G2.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P4; claims: G1, G2.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P2, P3; claims: G2.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G2" ] }, "Implementations: P3, P4; claims: G1, G2, G3.": { "policies": [ "P3", "P4" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Marsh-88c05 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-88c05": "1/7", "Basalt-88c05": "1/7", "Cobalt-88c05": "1/7", "Marsh-88c05": "3/7", "Dune-88c05": "1/7" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Onyx-88c05 on every execution, with probability one.", "distribution": { "Onyx-88c05": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-88c05": "1/5", "Basalt-88c05": "1/5", "Cobalt-88c05": "1/5", "Marsh-88c05": "1/5", "Dune-88c05": "1/5" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Cobalt-88c05, which has the unique lowest current load score, with probability one.", "distribution": { "Cobalt-88c05": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-88c05 on every execution, with probability one.", "distribution": { "Excluded-88c05": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-18", "english": "The runbook entry for a font-subset request contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Elm-cb2c0, Onyx-cb2c0, Pike-cb2c0, Amber-cb2c0, Juniper-cb2c0. Recorded latency scores, in the same order: 88, 20, 23, 7, 51. The reference roster-cb2c0@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-cb2c0 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-cb2c0@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-cb2c0@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-cb2c0@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The runbook entry for a font-subset request contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Elm-cb2c0, Onyx-cb2c0, Pike-cb2c0, Amber-cb2c0, Juniper-cb2c0. Recorded latency scores, in the same order: 88, 20, 23, 7, 51. The reference roster-cb2c0@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-cb2c0 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-cb2c0@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Elm-cb2c0 on every execution, with probability one.\nP2: Assigns probability 4/8 to Pike-cb2c0 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-cb2c0 on every execution, with probability one.\nP4: Returns Amber-cb2c0, which has the unique lowest latency score, with probability one.\nP5: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2, P5; claims: G1, G2, G3.", "Implementations: P5; claims: G1, G2, G3.", "Implementations: P1, P2, P4, P5; claims: G1, G3.", "Implementations: P1, P4, P5; claims: G1.", "Implementations: P2, P5; claims: G1, G3.", "Implementations: P5; claims: G1, G3.", "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: P1, P4, P5; claims: G1, G3." ], "answer": "Implementations: P5; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-05", "member_count": 5, "world_id": "rt1-world-01-18", "semantic_world": { "n": 5, "members": [ "Elm-cb2c0", "Onyx-cb2c0", "Pike-cb2c0", "Amber-cb2c0", "Juniper-cb2c0" ], "outsider": "Excluded-cb2c0", "scores": [ 88, 20, 23, 7, 51 ], "criterion": "latency", "set_ref": "roster-cb2c0@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P5; claims: G1, G2, G3.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P5; claims: G1, G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P4, P5; claims: G1.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P2, P5; claims: G1, G3.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P5; claims: G1, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Elm-cb2c0 on every execution, with probability one.", "distribution": { "Elm-cb2c0": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Pike-cb2c0 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-cb2c0": "1/8", "Onyx-cb2c0": "1/8", "Pike-cb2c0": "1/2", "Amber-cb2c0": "1/8", "Juniper-cb2c0": "1/8" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-cb2c0 on every execution, with probability one.", "distribution": { "Excluded-cb2c0": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Amber-cb2c0, which has the unique lowest latency score, with probability one.", "distribution": { "Amber-cb2c0": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-cb2c0": "1/5", "Onyx-cb2c0": "1/5", "Pike-cb2c0": "1/5", "Amber-cb2c0": "1/5", "Juniper-cb2c0": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-18", "english": "The runbook entry for a benchmark submission contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Pike-592f8, Onyx-592f8, Cobalt-592f8, Marsh-592f8, Amber-592f8, Dune-592f8, Basalt-592f8, Nettle-592f8. Recorded queue length scores, in the same order: 55, 41, 13, 69, 85, 40, 5, 33. The reference roster-592f8@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-592f8 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-592f8@snapshot-1) requires roster-592f8@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-592f8@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a benchmark submission contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Pike-592f8, Onyx-592f8, Cobalt-592f8, Marsh-592f8, Amber-592f8, Dune-592f8, Basalt-592f8, Nettle-592f8. Recorded queue length scores, in the same order: 55, 41, 13, 69, 85, 40, 5, 33. The reference roster-592f8@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-592f8 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-592f8@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Basalt-592f8, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Pike-592f8 on every execution, with probability one.\nP4: Assigns probability 4/11 to Amber-592f8 and 1/11 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-592f8 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4; claims: G2, G3.", "Implementations: P1, P3, P4; claims: G3.", "Implementations: none; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: none; claims: G1, G3.", "Implementations: P2; claims: G1, G3.", "Implementations: P2; claims: G1, G3, G4.", "Implementations: P1, P3, P4; claims: G2, G3." ], "answer": "Implementations: P1, P2, P3, P4; claims: G3.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-05", "member_count": 8, "world_id": "rt1-world-02-18", "semantic_world": { "n": 8, "members": [ "Pike-592f8", "Onyx-592f8", "Cobalt-592f8", "Marsh-592f8", "Amber-592f8", "Dune-592f8", "Basalt-592f8", "Nettle-592f8" ], "outsider": "Excluded-592f8", "scores": [ 55, 41, 13, 69, 85, 40, 5, 33 ], "criterion": "queue length", "set_ref": "roster-592f8@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G2, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P3, P4; claims: G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: none; claims: G1, G3.": { "policies": [], "guarantees": [ "G1", "G3" ] }, "Implementations: P2; claims: G1, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P3, P4; claims: G2, G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Basalt-592f8, which has the unique lowest queue length score, with probability one.", "distribution": { "Basalt-592f8": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Pike-592f8": "1/8", "Onyx-592f8": "1/8", "Cobalt-592f8": "1/8", "Marsh-592f8": "1/8", "Amber-592f8": "1/8", "Dune-592f8": "1/8", "Basalt-592f8": "1/8", "Nettle-592f8": "1/8" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Pike-592f8 on every execution, with probability one.", "distribution": { "Pike-592f8": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/11 to Amber-592f8 and 1/11 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Pike-592f8": "1/11", "Onyx-592f8": "1/11", "Cobalt-592f8": "1/11", "Marsh-592f8": "1/11", "Amber-592f8": "4/11", "Dune-592f8": "1/11", "Basalt-592f8": "1/11", "Nettle-592f8": "1/11" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-592f8 on every execution, with probability one.", "distribution": { "Excluded-592f8": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-18", "english": "The runbook entry for a dialogue evaluation slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Nettle-fbe88, Gale-fbe88, Amber-fbe88, Elm-fbe88. Recorded preparation cost scores, in the same order: 62, 75, 9, 2. The reference roster-fbe88@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fbe88 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fbe88@snapshot-1) requires roster-fbe88@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-fbe88@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a dialogue evaluation slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Nettle-fbe88, Gale-fbe88, Amber-fbe88, Elm-fbe88. Recorded preparation cost scores, in the same order: 62, 75, 9, 2. The reference roster-fbe88@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-fbe88 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-fbe88@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-fbe88 on every execution, with probability one.\nP3: Returns Nettle-fbe88 on every execution, with probability one.\nP4: Assigns probability 4/7 to Nettle-fbe88 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Elm-fbe88, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.", "Implementations: P1, P2; claims: G1, G2.", "Implementations: P1; claims: G1, G2.", "Implementations: P1; claims: G2.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P2; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G1, G3.", "Implementations: P1, P3, P4, P5; claims: G1." ], "answer": "Implementations: P1, P3, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-05", "member_count": 4, "world_id": "rt1-world-03-18", "semantic_world": { "n": 4, "members": [ "Nettle-fbe88", "Gale-fbe88", "Amber-fbe88", "Elm-fbe88" ], "outsider": "Excluded-fbe88", "scores": [ 62, 75, 9, 2 ], "criterion": "preparation cost", "set_ref": "roster-fbe88@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2; claims: G1, G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1; claims: G1, G2.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2; claims: G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G1, G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-fbe88": "1/4", "Gale-fbe88": "1/4", "Amber-fbe88": "1/4", "Elm-fbe88": "1/4" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-fbe88 on every execution, with probability one.", "distribution": { "Excluded-fbe88": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Nettle-fbe88 on every execution, with probability one.", "distribution": { "Nettle-fbe88": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Nettle-fbe88 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-fbe88": "4/7", "Gale-fbe88": "1/7", "Amber-fbe88": "1/7", "Elm-fbe88": "1/7" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Elm-fbe88, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Elm-fbe88": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-18", "english": "The runbook entry for a halted backup agent contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible standby node identities, in recorded order: Cobalt-f6c5d, Flint-f6c5d, Marsh-f6c5d, Basalt-f6c5d, Heron-f6c5d, Nettle-f6c5d, Kestrel-f6c5d. Recorded activation cost scores, in the same order: 60, 59, 40, 77, 12, 41, 24. The reference roster-f6c5d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f6c5d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f6c5d@snapshot-1) requires roster-f6c5d@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-f6c5d@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a halted backup agent contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible standby node identities, in recorded order: Cobalt-f6c5d, Flint-f6c5d, Marsh-f6c5d, Basalt-f6c5d, Heron-f6c5d, Nettle-f6c5d, Kestrel-f6c5d. Recorded activation cost scores, in the same order: 60, 59, 40, 77, 12, 41, 24. The reference roster-f6c5d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f6c5d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f6c5d@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/10 to Kestrel-f6c5d and 1/10 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Cobalt-f6c5d on every execution, with probability one.\nP4: Returns Excluded-f6c5d on every execution, with probability one.\nP5: Returns Heron-f6c5d, which has the unique lowest activation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: P2, P3; claims: G2, G4.", "Implementations: P2; claims: G2, G4.", "Implementations: P2; claims: G4.", "Implementations: P1, P2, P5; claims: G4.", "Implementations: P2, P3; claims: G4.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P1, P2, P5; claims: none." ], "answer": "Implementations: P1, P2, P3, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-05", "member_count": 7, "world_id": "rt1-world-04-18", "semantic_world": { "n": 7, "members": [ "Cobalt-f6c5d", "Flint-f6c5d", "Marsh-f6c5d", "Basalt-f6c5d", "Heron-f6c5d", "Nettle-f6c5d", "Kestrel-f6c5d" ], "outsider": "Excluded-f6c5d", "scores": [ 60, 59, 40, 77, 12, 41, 24 ], "criterion": "activation cost", "set_ref": "roster-f6c5d@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3; claims: G2, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G4.": { "policies": [ "P2" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P5; claims: G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3; claims: G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P5; claims: none.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Kestrel-f6c5d and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-f6c5d": "1/10", "Flint-f6c5d": "1/10", "Marsh-f6c5d": "1/10", "Basalt-f6c5d": "1/10", "Heron-f6c5d": "1/10", "Nettle-f6c5d": "1/10", "Kestrel-f6c5d": "2/5" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-f6c5d": "1/7", "Flint-f6c5d": "1/7", "Marsh-f6c5d": "1/7", "Basalt-f6c5d": "1/7", "Heron-f6c5d": "1/7", "Nettle-f6c5d": "1/7", "Kestrel-f6c5d": "1/7" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Cobalt-f6c5d on every execution, with probability one.", "distribution": { "Cobalt-f6c5d": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-f6c5d on every execution, with probability one.", "distribution": { "Excluded-f6c5d": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Heron-f6c5d, which has the unique lowest activation cost score, with probability one.", "distribution": { "Heron-f6c5d": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-18", "english": "The runbook entry for a museum label slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible content item identities, in recorded order: Gale-5fb8f, Elm-5fb8f, Nettle-5fb8f. Recorded retrieval cost scores, in the same order: 40, 74, 37. The reference roster-5fb8f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5fb8f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-5fb8f@snapshot-1) requires roster-5fb8f@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-5fb8f@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The runbook entry for a museum label slot contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible content item identities, in recorded order: Gale-5fb8f, Elm-5fb8f, Nettle-5fb8f. Recorded retrieval cost scores, in the same order: 40, 74, 37. The reference roster-5fb8f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5fb8f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-5fb8f@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/6 to Elm-5fb8f and 1/6 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Nettle-5fb8f, which has the unique lowest retrieval cost score, with probability one.\nP4: Returns Gale-5fb8f on every execution, with probability one.\nP5: Returns Excluded-5fb8f on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P4; claims: G1, G2.", "Implementations: P1, P2, P4; claims: G2.", "Implementations: P2; claims: G1, G2.", "Implementations: P2; claims: G1, G2, G3.", "Implementations: P2, P3; claims: G1, G2.", "Implementations: P2, P3; claims: G1, G2, G3.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: P1, P2, P3, P4; claims: G1, G2." ], "answer": "Implementations: P1, P2, P3, P4; claims: G2.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-05", "member_count": 3, "world_id": "rt1-world-05-18", "semantic_world": { "n": 3, "members": [ "Gale-5fb8f", "Elm-5fb8f", "Nettle-5fb8f" ], "outsider": "Excluded-5fb8f", "scores": [ 40, 74, 37 ], "criterion": "retrieval cost", "set_ref": "roster-5fb8f@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4; claims: G1, G2.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P4; claims: G2.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P2; claims: G1, G2.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2; claims: G1, G2, G3.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3; claims: G1, G2.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3; claims: G1, G2, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Elm-5fb8f and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-5fb8f": "1/6", "Elm-5fb8f": "2/3", "Nettle-5fb8f": "1/6" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-5fb8f": "1/3", "Elm-5fb8f": "1/3", "Nettle-5fb8f": "1/3" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Nettle-5fb8f, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Nettle-5fb8f": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Gale-5fb8f on every execution, with probability one.", "distribution": { "Gale-5fb8f": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-5fb8f on every execution, with probability one.", "distribution": { "Excluded-5fb8f": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-18", "english": "The runbook entry for a dependency-scan job contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible worker identities, in recorded order: Flint-5d5d5, Nettle-5d5d5, Iris-5d5d5, Basalt-5d5d5, Elm-5d5d5, Lumen-5d5d5. Recorded current load scores, in the same order: 85, 59, 28, 48, 26, 94. The reference roster-5d5d5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5d5d5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-5d5d5@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-5d5d5@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-5d5d5@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The runbook entry for a dependency-scan job contains a selection instruction that needs interpreting.\nThe frozen roster lists these eligible worker identities, in recorded order: Flint-5d5d5, Nettle-5d5d5, Iris-5d5d5, Basalt-5d5d5, Elm-5d5d5, Lumen-5d5d5. Recorded current load scores, in the same order: 85, 59, 28, 48, 26, 94. The reference roster-5d5d5@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5d5d5 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-5d5d5@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Elm-5d5d5, which has the unique lowest current load score, with probability one.\nP2: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP3: Assigns probability 4/9 to Elm-5d5d5 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Flint-5d5d5 on every execution, with probability one.\nP5: Returns Excluded-5d5d5 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P2, P3; claims: G1, G2, G4.", "Implementations: P2; claims: G1, G4.", "Implementations: P1, P2, P4; claims: G1.", "Implementations: P1, P2, P3, P4; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P2, P3; claims: G1, G4.", "Implementations: P2; claims: G1, G2, G4.", "Implementations: P1, P2, P4; claims: G1, G3." ], "answer": "Implementations: P2; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-05", "member_count": 6, "world_id": "rt1-world-06-18", "semantic_world": { "n": 6, "members": [ "Flint-5d5d5", "Nettle-5d5d5", "Iris-5d5d5", "Basalt-5d5d5", "Elm-5d5d5", "Lumen-5d5d5" ], "outsider": "Excluded-5d5d5", "scores": [ 85, 59, 28, 48, 26, 94 ], "criterion": "current load", "set_ref": "roster-5d5d5@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3; claims: G1, G2, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P4; claims: G1.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P2, P3; claims: G1, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2; claims: G1, G2, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P2, P4; claims: G1, G3.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Elm-5d5d5, which has the unique lowest current load score, with probability one.", "distribution": { "Elm-5d5d5": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-5d5d5": "1/6", "Nettle-5d5d5": "1/6", "Iris-5d5d5": "1/6", "Basalt-5d5d5": "1/6", "Elm-5d5d5": "1/6", "Lumen-5d5d5": "1/6" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Elm-5d5d5 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-5d5d5": "1/9", "Nettle-5d5d5": "1/9", "Iris-5d5d5": "1/9", "Basalt-5d5d5": "1/9", "Elm-5d5d5": "4/9", "Lumen-5d5d5": "1/9" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Flint-5d5d5 on every execution, with probability one.", "distribution": { "Flint-5d5d5": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-5d5d5 on every execution, with probability one.", "distribution": { "Excluded-5d5d5": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-19", "english": "A reviewer is asked which procedures would comply with the request for a receipt-rendering call.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Iris-4357e, Nettle-4357e, Basalt-4357e, Elm-4357e, Flint-4357e, Pike-4357e. Recorded latency scores, in the same order: 91, 47, 38, 80, 95, 8. The reference roster-4357e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-4357e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-4357e@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-4357e@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-4357e@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A reviewer is asked which procedures would comply with the request for a receipt-rendering call.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Iris-4357e, Nettle-4357e, Basalt-4357e, Elm-4357e, Flint-4357e, Pike-4357e. Recorded latency scores, in the same order: 91, 47, 38, 80, 95, 8. The reference roster-4357e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-4357e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-4357e@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/7 to Pike-4357e and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Iris-4357e on every execution, with probability one.\nP3: Returns Pike-4357e, which has the unique lowest latency score, with probability one.\nP4: Returns Excluded-4357e on every execution, with probability one.\nP5: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: none; claims: G1, G4.", "Implementations: P5; claims: G1, G4.", "Implementations: P1, P2, P3; claims: G4.", "Implementations: P5; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: none; claims: G1, G3, G4.", "Implementations: P1, P2, P3; claims: G2, G4.", "Implementations: P1, P2, P3, P5; claims: G4." ], "answer": "Implementations: P5; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-06", "member_count": 6, "world_id": "rt1-world-01-19", "semantic_world": { "n": 6, "members": [ "Iris-4357e", "Nettle-4357e", "Basalt-4357e", "Elm-4357e", "Flint-4357e", "Pike-4357e" ], "outsider": "Excluded-4357e", "scores": [ 91, 47, 38, 80, 95, 8 ], "criterion": "latency", "set_ref": "roster-4357e@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: P5; claims: G1, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3; claims: G4.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G4" ] }, "Implementations: P5; claims: G1, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G1, G3, G4.": { "policies": [], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3; claims: G2, G4.": { "policies": [ "P1", "P2", "P3" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/7 to Pike-4357e and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Iris-4357e": "1/7", "Nettle-4357e": "1/7", "Basalt-4357e": "1/7", "Elm-4357e": "1/7", "Flint-4357e": "1/7", "Pike-4357e": "2/7" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Iris-4357e on every execution, with probability one.", "distribution": { "Iris-4357e": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Pike-4357e, which has the unique lowest latency score, with probability one.", "distribution": { "Pike-4357e": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-4357e on every execution, with probability one.", "distribution": { "Excluded-4357e": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Iris-4357e": "1/6", "Nettle-4357e": "1/6", "Basalt-4357e": "1/6", "Elm-4357e": "1/6", "Flint-4357e": "1/6", "Pike-4357e": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-19", "english": "A reviewer is asked which procedures would comply with the request for a grant renewal.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-ee699, Marsh-ee699. Recorded queue length scores, in the same order: 59, 34. The reference roster-ee699@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ee699 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ee699@snapshot-1) requires roster-ee699@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-ee699@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A reviewer is asked which procedures would comply with the request for a grant renewal.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Basalt-ee699, Marsh-ee699. Recorded queue length scores, in the same order: 59, 34. The reference roster-ee699@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ee699 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-ee699@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-ee699, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Basalt-ee699 on every execution, with probability one.\nP4: Returns Excluded-ee699 on every execution, with probability one.\nP5: Assigns probability 2/3 to Basalt-ee699 and 1/3 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P2, P4; claims: G2, G3.", "Implementations: P1, P2, P3, P5; claims: G2.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.", "Implementations: P2; claims: G3.", "Implementations: P2, P4; claims: G3.", "Implementations: P2; claims: G2, G3.", "Implementations: P1, P2, P3, P4, P5; claims: G2." ], "answer": "Implementations: P1, P2, P3, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-06", "member_count": 2, "world_id": "rt1-world-02-19", "semantic_world": { "n": 2, "members": [ "Basalt-ee699", "Marsh-ee699" ], "outsider": "Excluded-ee699", "scores": [ 59, 34 ], "criterion": "queue length", "set_ref": "roster-ee699@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P4; claims: G2, G3.": { "policies": [ "P2", "P4" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G3.": { "policies": [ "P2" ], "guarantees": [ "G3" ] }, "Implementations: P2, P4; claims: G3.": { "policies": [ "P2", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P2; claims: G2, G3.": { "policies": [ "P2" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Marsh-ee699, which has the unique lowest queue length score, with probability one.", "distribution": { "Marsh-ee699": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-ee699": "1/2", "Marsh-ee699": "1/2" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Basalt-ee699 on every execution, with probability one.", "distribution": { "Basalt-ee699": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-ee699 on every execution, with probability one.", "distribution": { "Excluded-ee699": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Basalt-ee699 and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-ee699": "2/3", "Marsh-ee699": "1/3" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-19", "english": "A reviewer is asked which procedures would comply with the request for a summarisation evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Basalt-7858c, Amber-7858c, Heron-7858c, Iris-7858c, Lumen-7858c. Recorded preparation cost scores, in the same order: 33, 94, 39, 23, 71. The reference roster-7858c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7858c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-7858c@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-7858c@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-7858c@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A reviewer is asked which procedures would comply with the request for a summarisation evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Basalt-7858c, Amber-7858c, Heron-7858c, Iris-7858c, Lumen-7858c. Recorded preparation cost scores, in the same order: 33, 94, 39, 23, 71. The reference roster-7858c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7858c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-7858c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Iris-7858c, which has the unique lowest preparation cost score, with probability one.\nP2: Returns Excluded-7858c on every execution, with probability one.\nP3: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 2/6 to Iris-7858c and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Basalt-7858c on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P3, P4, P5; claims: G1.", "Implementations: P3; claims: G1.", "Implementations: P3; claims: G1, G2.", "Implementations: P3, P5; claims: G1, G2.", "Implementations: P3, P5; claims: G1.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P1, P3, P4; claims: G1.", "Implementations: P1, P3, P4; claims: none." ], "answer": "Implementations: P3; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-06", "member_count": 5, "world_id": "rt1-world-03-19", "semantic_world": { "n": 5, "members": [ "Basalt-7858c", "Amber-7858c", "Heron-7858c", "Iris-7858c", "Lumen-7858c" ], "outsider": "Excluded-7858c", "scores": [ 33, 94, 39, 23, 71 ], "criterion": "preparation cost", "set_ref": "roster-7858c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P3; claims: G1.": { "policies": [ "P3" ], "guarantees": [ "G1" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P5; claims: G1, G2.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P5; claims: G1.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P3, P4; claims: G1.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4; claims: none.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Iris-7858c, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Iris-7858c": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-7858c on every execution, with probability one.", "distribution": { "Excluded-7858c": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-7858c": "1/5", "Amber-7858c": "1/5", "Heron-7858c": "1/5", "Iris-7858c": "1/5", "Lumen-7858c": "1/5" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Iris-7858c and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-7858c": "1/6", "Amber-7858c": "1/6", "Heron-7858c": "1/6", "Iris-7858c": "1/3", "Lumen-7858c": "1/6" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Basalt-7858c on every execution, with probability one.", "distribution": { "Basalt-7858c": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-19", "english": "A reviewer is asked which procedures would comply with the request for a stalled queue consumer.\nThe frozen roster lists these eligible standby node identities, in recorded order: Elm-aabce, Cobalt-aabce, Nettle-aabce, Marsh-aabce, Basalt-aabce, Lumen-aabce, Kestrel-aabce, Pike-aabce. Recorded activation cost scores, in the same order: 20, 32, 75, 24, 3, 46, 39, 67. The reference roster-aabce@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-aabce is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-aabce@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-aabce@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-aabce@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A reviewer is asked which procedures would comply with the request for a stalled queue consumer.\nThe frozen roster lists these eligible standby node identities, in recorded order: Elm-aabce, Cobalt-aabce, Nettle-aabce, Marsh-aabce, Basalt-aabce, Lumen-aabce, Kestrel-aabce, Pike-aabce. Recorded activation cost scores, in the same order: 20, 32, 75, 24, 3, 46, 39, 67. The reference roster-aabce@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-aabce is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-aabce@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-aabce on every execution, with probability one.\nP2: Returns Basalt-aabce, which has the unique lowest activation cost score, with probability one.\nP3: Returns Elm-aabce on every execution, with probability one.\nP4: Assigns probability 2/9 to Elm-aabce and 1/9 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P3, P4, P5; claims: G1, G4.", "Implementations: P2, P3, P4, P5; claims: G1, G4.", "Implementations: P5; claims: G1, G3, G4.", "Implementations: P5; claims: G1, G4.", "Implementations: P3, P4, P5; claims: G4.", "Implementations: P2, P5; claims: G1, G3, G4.", "Implementations: P2, P5; claims: G1, G4.", "Implementations: P2, P3, P4, P5; claims: G4." ], "answer": "Implementations: P5; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-06", "member_count": 8, "world_id": "rt1-world-04-19", "semantic_world": { "n": 8, "members": [ "Elm-aabce", "Cobalt-aabce", "Nettle-aabce", "Marsh-aabce", "Basalt-aabce", "Lumen-aabce", "Kestrel-aabce", "Pike-aabce" ], "outsider": "Excluded-aabce", "scores": [ 20, 32, 75, 24, 3, 46, 39, 67 ], "criterion": "activation cost", "set_ref": "roster-aabce@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4, P5; claims: G1, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P5; claims: G1, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P5; claims: G1, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P3, P4, P5; claims: G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P5; claims: G1, G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P5; claims: G1, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-aabce on every execution, with probability one.", "distribution": { "Excluded-aabce": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Basalt-aabce, which has the unique lowest activation cost score, with probability one.", "distribution": { "Basalt-aabce": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Elm-aabce on every execution, with probability one.", "distribution": { "Elm-aabce": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Elm-aabce and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-aabce": "2/9", "Cobalt-aabce": "1/9", "Nettle-aabce": "1/9", "Marsh-aabce": "1/9", "Basalt-aabce": "1/9", "Lumen-aabce": "1/9", "Kestrel-aabce": "1/9", "Pike-aabce": "1/9" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-aabce": "1/8", "Cobalt-aabce": "1/8", "Nettle-aabce": "1/8", "Marsh-aabce": "1/8", "Basalt-aabce": "1/8", "Lumen-aabce": "1/8", "Kestrel-aabce": "1/8", "Pike-aabce": "1/8" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-19", "english": "A reviewer is asked which procedures would comply with the request for a landing-page banner slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Marsh-e3e7a, Onyx-e3e7a, Lumen-e3e7a, Flint-e3e7a. Recorded retrieval cost scores, in the same order: 71, 29, 91, 17. The reference roster-e3e7a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e3e7a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e3e7a@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-e3e7a@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-e3e7a@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A reviewer is asked which procedures would comply with the request for a landing-page banner slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Marsh-e3e7a, Onyx-e3e7a, Lumen-e3e7a, Flint-e3e7a. Recorded retrieval cost scores, in the same order: 71, 29, 91, 17. The reference roster-e3e7a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e3e7a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e3e7a@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-e3e7a on every execution, with probability one.\nP2: Assigns probability 2/5 to Flint-e3e7a and 1/5 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-e3e7a on every execution, with probability one.\nP4: Returns Flint-e3e7a, which has the unique lowest retrieval cost score, with probability one.\nP5: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P4, P5; claims: G2.", "Implementations: P1, P4, P5; claims: G2.", "Implementations: P5; claims: G2, G4.", "Implementations: P2, P5; claims: G2, G4.", "Implementations: P5; claims: G2, G3, G4.", "Implementations: P1, P2, P4, P5; claims: G1, G2.", "Implementations: P1, P4, P5; claims: G1, G2.", "Implementations: P2, P5; claims: G2, G3, G4." ], "answer": "Implementations: P5; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-06", "member_count": 4, "world_id": "rt1-world-05-19", "semantic_world": { "n": 4, "members": [ "Marsh-e3e7a", "Onyx-e3e7a", "Lumen-e3e7a", "Flint-e3e7a" ], "outsider": "Excluded-e3e7a", "scores": [ 71, 29, 91, 17 ], "criterion": "retrieval cost", "set_ref": "roster-e3e7a@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P4, P5; claims: G2.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P5; claims: G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P5; claims: G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G2, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P1, P2, P4, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P4, P5; claims: G1, G2.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P5; claims: G2, G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2", "G3", "G4" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Marsh-e3e7a on every execution, with probability one.", "distribution": { "Marsh-e3e7a": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Flint-e3e7a and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-e3e7a": "1/5", "Onyx-e3e7a": "1/5", "Lumen-e3e7a": "1/5", "Flint-e3e7a": "2/5" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-e3e7a on every execution, with probability one.", "distribution": { "Excluded-e3e7a": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Flint-e3e7a, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Flint-e3e7a": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-e3e7a": "1/4", "Onyx-e3e7a": "1/4", "Lumen-e3e7a": "1/4", "Flint-e3e7a": "1/4" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-19", "english": "A reviewer is asked which procedures would comply with the request for a thumbnail-generation job.\nThe frozen roster lists these eligible worker identities, in recorded order: Kestrel-a6dd7, Basalt-a6dd7, Flint-a6dd7, Juniper-a6dd7, Cobalt-a6dd7, Iris-a6dd7, Gale-a6dd7. Recorded current load scores, in the same order: 83, 92, 90, 32, 57, 14, 45. The reference roster-a6dd7@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a6dd7 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-a6dd7@snapshot-1) requires roster-a6dd7@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-a6dd7@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A reviewer is asked which procedures would comply with the request for a thumbnail-generation job.\nThe frozen roster lists these eligible worker identities, in recorded order: Kestrel-a6dd7, Basalt-a6dd7, Flint-a6dd7, Juniper-a6dd7, Cobalt-a6dd7, Iris-a6dd7, Gale-a6dd7. Recorded current load scores, in the same order: 83, 92, 90, 32, 57, 14, 45. The reference roster-a6dd7@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a6dd7 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-a6dd7@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-a6dd7 on every execution, with probability one.\nP2: Returns Iris-a6dd7, which has the unique lowest current load score, with probability one.\nP3: Assigns probability 2/8 to Kestrel-a6dd7 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Kestrel-a6dd7 on every execution, with probability one.\nP5: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P5; claims: G2, G3.", "Implementations: P2, P3, P4, P5; claims: G3.", "Implementations: P2, P3, P4, P5; claims: G1, G3.", "Implementations: P2, P3, P4; claims: G3.", "Implementations: none; claims: G2, G3.", "Implementations: P2, P3, P4; claims: G1, G3.", "Implementations: none; claims: G2.", "Implementations: P5; claims: G2." ], "answer": "Implementations: P2, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-06", "member_count": 7, "world_id": "rt1-world-06-19", "semantic_world": { "n": 7, "members": [ "Kestrel-a6dd7", "Basalt-a6dd7", "Flint-a6dd7", "Juniper-a6dd7", "Cobalt-a6dd7", "Iris-a6dd7", "Gale-a6dd7" ], "outsider": "Excluded-a6dd7", "scores": [ 83, 92, 90, 32, 57, 14, 45 ], "criterion": "current load", "set_ref": "roster-a6dd7@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P2, P3, P4; claims: G3.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: none; claims: G2, G3.": { "policies": [], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3, P4; claims: G1, G3.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: none; claims: G2.": { "policies": [], "guarantees": [ "G2" ] }, "Implementations: P5; claims: G2.": { "policies": [ "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-a6dd7 on every execution, with probability one.", "distribution": { "Excluded-a6dd7": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Iris-a6dd7, which has the unique lowest current load score, with probability one.", "distribution": { "Iris-a6dd7": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Kestrel-a6dd7 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-a6dd7": "1/4", "Basalt-a6dd7": "1/8", "Flint-a6dd7": "1/8", "Juniper-a6dd7": "1/8", "Cobalt-a6dd7": "1/8", "Iris-a6dd7": "1/8", "Gale-a6dd7": "1/8" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Kestrel-a6dd7 on every execution, with probability one.", "distribution": { "Kestrel-a6dd7": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-a6dd7": "1/7", "Basalt-a6dd7": "1/7", "Flint-a6dd7": "1/7", "Juniper-a6dd7": "1/7", "Cobalt-a6dd7": "1/7", "Iris-a6dd7": "1/7", "Gale-a6dd7": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-20", "english": "A ticket about a postcode lookup is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Gale-ac829, Flint-ac829, Amber-ac829, Heron-ac829, Marsh-ac829, Kestrel-ac829, Juniper-ac829. Recorded latency scores, in the same order: 30, 5, 35, 78, 11, 31, 94. The reference roster-ac829@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ac829 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-ac829@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-ac829@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-ac829@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about a postcode lookup is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Gale-ac829, Flint-ac829, Amber-ac829, Heron-ac829, Marsh-ac829, Kestrel-ac829, Juniper-ac829. Recorded latency scores, in the same order: 30, 5, 35, 78, 11, 31, 94. The reference roster-ac829@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-ac829 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-ac829@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 3/9 to Gale-ac829 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Gale-ac829 on every execution, with probability one.\nP4: Returns Flint-ac829, which has the unique lowest latency score, with probability one.\nP5: Returns Excluded-ac829 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1; claims: G1, G3.", "Implementations: P1, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P4; claims: G1, G4.", "Implementations: P1; claims: G3.", "Implementations: P1, P2, P3, P4; claims: G1.", "Implementations: P1, P5; claims: G3.", "Implementations: P1, P2, P3, P4, P5; claims: G1, G4." ], "answer": "Implementations: P1; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-07", "member_count": 7, "world_id": "rt1-world-01-20", "semantic_world": { "n": 7, "members": [ "Gale-ac829", "Flint-ac829", "Amber-ac829", "Heron-ac829", "Marsh-ac829", "Kestrel-ac829", "Juniper-ac829" ], "outsider": "Excluded-ac829", "scores": [ 30, 5, 35, 78, 11, 31, 94 ], "criterion": "latency", "set_ref": "roster-ac829@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G3.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P5; claims: G1, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1; claims: G3.": { "policies": [ "P1" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P5; claims: G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-ac829": "1/7", "Flint-ac829": "1/7", "Amber-ac829": "1/7", "Heron-ac829": "1/7", "Marsh-ac829": "1/7", "Kestrel-ac829": "1/7", "Juniper-ac829": "1/7" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/9 to Gale-ac829 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-ac829": "1/3", "Flint-ac829": "1/9", "Amber-ac829": "1/9", "Heron-ac829": "1/9", "Marsh-ac829": "1/9", "Kestrel-ac829": "1/9", "Juniper-ac829": "1/9" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Gale-ac829 on every execution, with probability one.", "distribution": { "Gale-ac829": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Flint-ac829, which has the unique lowest latency score, with probability one.", "distribution": { "Flint-ac829": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-ac829 on every execution, with probability one.", "distribution": { "Excluded-ac829": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-20", "english": "A ticket about a code-audit report is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Marsh-c05a2, Amber-c05a2, Pike-c05a2. Recorded queue length scores, in the same order: 71, 56, 36. The reference roster-c05a2@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c05a2 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-c05a2@snapshot-1) requires roster-c05a2@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-c05a2@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A ticket about a code-audit report is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Marsh-c05a2, Amber-c05a2, Pike-c05a2. Recorded queue length scores, in the same order: 71, 56, 36. The reference roster-c05a2@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-c05a2 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-c05a2@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Pike-c05a2, which has the unique lowest queue length score, with probability one.\nP2: Returns Excluded-c05a2 on every execution, with probability one.\nP3: Returns Marsh-c05a2 on every execution, with probability one.\nP4: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP5: Assigns probability 3/5 to Pike-c05a2 and 1/5 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P4, P5; claims: G4.", "Implementations: P4; claims: G4.", "Implementations: P1, P4, P5; claims: none.", "Implementations: P3, P4; claims: G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P3, P4; claims: G2, G4.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P4; claims: G2, G4." ], "answer": "Implementations: P1, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-07", "member_count": 3, "world_id": "rt1-world-02-20", "semantic_world": { "n": 3, "members": [ "Marsh-c05a2", "Amber-c05a2", "Pike-c05a2" ], "outsider": "Excluded-c05a2", "scores": [ 71, 56, 36 ], "criterion": "queue length", "set_ref": "roster-c05a2@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P4, P5; claims: G4.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P4; claims: G4.": { "policies": [ "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P4, P5; claims: none.": { "policies": [ "P1", "P4", "P5" ], "guarantees": [] }, "Implementations: P3, P4; claims: G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P3, P4; claims: G2, G4.": { "policies": [ "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P4; claims: G2, G4.": { "policies": [ "P4" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Pike-c05a2, which has the unique lowest queue length score, with probability one.", "distribution": { "Pike-c05a2": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-c05a2 on every execution, with probability one.", "distribution": { "Excluded-c05a2": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Marsh-c05a2 on every execution, with probability one.", "distribution": { "Marsh-c05a2": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-c05a2": "1/3", "Amber-c05a2": "1/3", "Pike-c05a2": "1/3" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Pike-c05a2 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-c05a2": "1/5", "Amber-c05a2": "1/5", "Pike-c05a2": "3/5" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-20", "english": "A ticket about a tool-use evaluation slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Cobalt-b705a, Marsh-b705a, Amber-b705a, Dune-b705a, Heron-b705a, Iris-b705a. Recorded preparation cost scores, in the same order: 75, 9, 68, 84, 45, 50. The reference roster-b705a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b705a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-b705a@snapshot-1) requires roster-b705a@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-b705a@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A ticket about a tool-use evaluation slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Cobalt-b705a, Marsh-b705a, Amber-b705a, Dune-b705a, Heron-b705a, Iris-b705a. Recorded preparation cost scores, in the same order: 75, 9, 68, 84, 45, 50. The reference roster-b705a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b705a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-b705a@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-b705a on every execution, with probability one.\nP2: Assigns probability 3/8 to Dune-b705a and 1/8 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Marsh-b705a, which has the unique lowest preparation cost score, with probability one.\nP4: Returns Cobalt-b705a on every execution, with probability one.\nP5: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P5; claims: G1, G2, G4.", "Implementations: P2, P3, P4, P5; claims: G2.", "Implementations: P2, P4, P5; claims: G2.", "Implementations: P2, P3, P4, P5; claims: G1, G2.", "Implementations: P2, P4, P5; claims: G1, G2.", "Implementations: P3, P5; claims: G1, G2.", "Implementations: P3, P5; claims: G1, G2, G4.", "Implementations: P5; claims: G1, G2." ], "answer": "Implementations: P2, P3, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-07", "member_count": 6, "world_id": "rt1-world-03-20", "semantic_world": { "n": 6, "members": [ "Cobalt-b705a", "Marsh-b705a", "Amber-b705a", "Dune-b705a", "Heron-b705a", "Iris-b705a" ], "outsider": "Excluded-b705a", "scores": [ 75, 9, 68, 84, 45, 50 ], "criterion": "preparation cost", "set_ref": "roster-b705a@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G2.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P2, P4, P5; claims: G2.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G2.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P4, P5; claims: G1, G2.": { "policies": [ "P2", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P5; claims: G1, G2.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3, P5; claims: G1, G2, G4.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P5; claims: G1, G2.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-b705a on every execution, with probability one.", "distribution": { "Excluded-b705a": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Dune-b705a and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Cobalt-b705a": "1/8", "Marsh-b705a": "1/8", "Amber-b705a": "1/8", "Dune-b705a": "3/8", "Heron-b705a": "1/8", "Iris-b705a": "1/8" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Marsh-b705a, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Marsh-b705a": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Cobalt-b705a on every execution, with probability one.", "distribution": { "Cobalt-b705a": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Cobalt-b705a": "1/6", "Marsh-b705a": "1/6", "Amber-b705a": "1/6", "Dune-b705a": "1/6", "Heron-b705a": "1/6", "Iris-b705a": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-20", "english": "A ticket about an unreachable metrics store is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible standby node identities, in recorded order: Flint-e5397, Lumen-e5397. Recorded activation cost scores, in the same order: 15, 9. The reference roster-e5397@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e5397 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e5397@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-e5397@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-e5397@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about an unreachable metrics store is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible standby node identities, in recorded order: Flint-e5397, Lumen-e5397. Recorded activation cost scores, in the same order: 15, 9. The reference roster-e5397@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-e5397 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-e5397@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-e5397 on every execution, with probability one.\nP3: Returns Flint-e5397 on every execution, with probability one.\nP4: Returns Lumen-e5397, which has the unique lowest activation cost score, with probability one.\nP5: Assigns probability 3/4 to Lumen-e5397 and 1/4 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1; claims: G1, G2.", "Implementations: P1, P5; claims: G1, G2.", "Implementations: P1, P3, P4; claims: G1.", "Implementations: P1, P5; claims: G1, G2, G3.", "Implementations: P1, P3, P4; claims: G1, G4.", "Implementations: P1, P3, P4, P5; claims: G1.", "Implementations: P1, P3, P4, P5; claims: G1, G4.", "Implementations: P1; claims: G1, G2, G3." ], "answer": "Implementations: P1; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-07", "member_count": 2, "world_id": "rt1-world-04-20", "semantic_world": { "n": 2, "members": [ "Flint-e5397", "Lumen-e5397" ], "outsider": "Excluded-e5397", "scores": [ 15, 9 ], "criterion": "activation cost", "set_ref": "roster-e5397@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G1, G2.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P5; claims: G1, G2.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3, P4; claims: G1.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P5; claims: G1, G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P3, P4; claims: G1, G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4, P5; claims: G1, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1; claims: G1, G2, G3.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-e5397": "1/2", "Lumen-e5397": "1/2" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-e5397 on every execution, with probability one.", "distribution": { "Excluded-e5397": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Flint-e5397 on every execution, with probability one.", "distribution": { "Flint-e5397": "1" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Lumen-e5397, which has the unique lowest activation cost score, with probability one.", "distribution": { "Lumen-e5397": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Lumen-e5397 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-e5397": "1/4", "Lumen-e5397": "3/4" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-20", "english": "A ticket about a newsletter header slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible content item identities, in recorded order: Juniper-1b50d, Marsh-1b50d, Flint-1b50d, Cobalt-1b50d, Elm-1b50d. Recorded retrieval cost scores, in the same order: 73, 11, 30, 74, 94. The reference roster-1b50d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1b50d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-1b50d@snapshot-1) requires roster-1b50d@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-1b50d@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A ticket about a newsletter header slot is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible content item identities, in recorded order: Juniper-1b50d, Marsh-1b50d, Flint-1b50d, Cobalt-1b50d, Elm-1b50d. Recorded retrieval cost scores, in the same order: 73, 11, 30, 74, 94. The reference roster-1b50d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-1b50d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-1b50d@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 3/7 to Elm-1b50d and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Marsh-1b50d, which has the unique lowest retrieval cost score, with probability one.\nP3: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Juniper-1b50d on every execution, with probability one.\nP5: Returns Excluded-1b50d on every execution, with probability one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: none; claims: G2, G3.", "Implementations: P1, P2, P3, P4; claims: G2.", "Implementations: P1, P2, P4; claims: G2.", "Implementations: P1, P2, P4; claims: G2, G4.", "Implementations: P3; claims: G2, G3.", "Implementations: P3; claims: G3.", "Implementations: none; claims: G3." ], "answer": "Implementations: P1, P2, P3, P4; claims: G2.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-07", "member_count": 5, "world_id": "rt1-world-05-20", "semantic_world": { "n": 5, "members": [ "Juniper-1b50d", "Marsh-1b50d", "Flint-1b50d", "Cobalt-1b50d", "Elm-1b50d" ], "outsider": "Excluded-1b50d", "scores": [ 73, 11, 30, 74, 94 ], "criterion": "retrieval cost", "set_ref": "roster-1b50d@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G2, G3.": { "policies": [], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P4; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P4; claims: G2.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P3; claims: G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P3; claims: G3.": { "policies": [ "P3" ], "guarantees": [ "G3" ] }, "Implementations: none; claims: G3.": { "policies": [], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Elm-1b50d and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-1b50d": "1/7", "Marsh-1b50d": "1/7", "Flint-1b50d": "1/7", "Cobalt-1b50d": "1/7", "Elm-1b50d": "3/7" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Marsh-1b50d, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Marsh-1b50d": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-1b50d": "1/5", "Marsh-1b50d": "1/5", "Flint-1b50d": "1/5", "Cobalt-1b50d": "1/5", "Elm-1b50d": "1/5" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Juniper-1b50d on every execution, with probability one.", "distribution": { "Juniper-1b50d": "1" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-1b50d on every execution, with probability one.", "distribution": { "Excluded-1b50d": "1" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-20", "english": "A ticket about a ledger-reconciliation job is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-38d3e, Flint-38d3e, Marsh-38d3e, Nettle-38d3e, Cobalt-38d3e, Kestrel-38d3e, Juniper-38d3e, Iris-38d3e. Recorded current load scores, in the same order: 8, 62, 19, 7, 59, 76, 9, 86. The reference roster-38d3e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-38d3e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-38d3e@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-38d3e@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-38d3e@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A ticket about a ledger-reconciliation job is being triaged; the wording of its selection line matters.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-38d3e, Flint-38d3e, Marsh-38d3e, Nettle-38d3e, Cobalt-38d3e, Kestrel-38d3e, Juniper-38d3e, Iris-38d3e. Recorded current load scores, in the same order: 8, 62, 19, 7, 59, 76, 9, 86. The reference roster-38d3e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-38d3e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-38d3e@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Amber-38d3e on every execution, with probability one.\nP3: Returns Nettle-38d3e, which has the unique lowest current load score, with probability one.\nP4: Returns Excluded-38d3e on every execution, with probability one.\nP5: Assigns probability 3/10 to Cobalt-38d3e and 1/10 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1, P4; claims: G1, G3.", "Implementations: P1; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1; claims: G1.", "Implementations: P1, P4; claims: G1.", "Implementations: P1, P2, P3, P5; claims: G1." ], "answer": "Implementations: P1; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-07", "member_count": 8, "world_id": "rt1-world-06-20", "semantic_world": { "n": 8, "members": [ "Amber-38d3e", "Flint-38d3e", "Marsh-38d3e", "Nettle-38d3e", "Cobalt-38d3e", "Kestrel-38d3e", "Juniper-38d3e", "Iris-38d3e" ], "outsider": "Excluded-38d3e", "scores": [ 8, 62, 19, 7, 59, 76, 9, 86 ], "criterion": "current load", "set_ref": "roster-38d3e@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4; claims: G1, G3.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1; claims: G1, G3.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1; claims: G1.": { "policies": [ "P1" ], "guarantees": [ "G1" ] }, "Implementations: P1, P4; claims: G1.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-38d3e": "1/8", "Flint-38d3e": "1/8", "Marsh-38d3e": "1/8", "Nettle-38d3e": "1/8", "Cobalt-38d3e": "1/8", "Kestrel-38d3e": "1/8", "Juniper-38d3e": "1/8", "Iris-38d3e": "1/8" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Amber-38d3e on every execution, with probability one.", "distribution": { "Amber-38d3e": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Nettle-38d3e, which has the unique lowest current load score, with probability one.", "distribution": { "Nettle-38d3e": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-38d3e on every execution, with probability one.", "distribution": { "Excluded-38d3e": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Cobalt-38d3e and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-38d3e": "1/10", "Flint-38d3e": "1/10", "Marsh-38d3e": "1/10", "Nettle-38d3e": "1/10", "Cobalt-38d3e": "3/10", "Kestrel-38d3e": "1/10", "Juniper-38d3e": "1/10", "Iris-38d3e": "1/10" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-21", "english": "A compliance note distinguishes what the request for a map-tile fetch permits from what it promises.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Kestrel-2d56a, Nettle-2d56a, Juniper-2d56a, Amber-2d56a, Elm-2d56a, Lumen-2d56a, Basalt-2d56a, Dune-2d56a. Recorded latency scores, in the same order: 64, 28, 46, 69, 96, 65, 3, 26. The reference roster-2d56a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-2d56a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-2d56a@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-2d56a@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-2d56a@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a map-tile fetch permits from what it promises.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Kestrel-2d56a, Nettle-2d56a, Juniper-2d56a, Amber-2d56a, Elm-2d56a, Lumen-2d56a, Basalt-2d56a, Dune-2d56a. Recorded latency scores, in the same order: 64, 28, 46, 69, 96, 65, 3, 26. The reference roster-2d56a@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-2d56a is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-2d56a@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/11 to Juniper-2d56a and 1/11 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Excluded-2d56a on every execution, with probability one.\nP3: Returns Basalt-2d56a, which has the unique lowest latency score, with probability one.\nP4: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Kestrel-2d56a on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P4, P5; claims: G1, G2.", "Implementations: P4, P5; claims: G1.", "Implementations: P1, P3, P4; claims: G1.", "Implementations: P1, P3, P4, P5; claims: G1.", "Implementations: P4; claims: G1.", "Implementations: P1, P3, P4; claims: none.", "Implementations: P4; claims: G1, G2.", "Implementations: P1, P3, P4, P5; claims: none." ], "answer": "Implementations: P4; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-08", "member_count": 8, "world_id": "rt1-world-01-21", "semantic_world": { "n": 8, "members": [ "Kestrel-2d56a", "Nettle-2d56a", "Juniper-2d56a", "Amber-2d56a", "Elm-2d56a", "Lumen-2d56a", "Basalt-2d56a", "Dune-2d56a" ], "outsider": "Excluded-2d56a", "scores": [ 64, 28, 46, 69, 96, 65, 3, 26 ], "criterion": "latency", "set_ref": "roster-2d56a@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P4, P5; claims: G1, G2.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P4, P5; claims: G1.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4; claims: G1.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P4; claims: G1.": { "policies": [ "P4" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3, P4; claims: none.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [] }, "Implementations: P4; claims: G1, G2.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/11 to Juniper-2d56a and 1/11 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-2d56a": "1/11", "Nettle-2d56a": "1/11", "Juniper-2d56a": "4/11", "Amber-2d56a": "1/11", "Elm-2d56a": "1/11", "Lumen-2d56a": "1/11", "Basalt-2d56a": "1/11", "Dune-2d56a": "1/11" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-2d56a on every execution, with probability one.", "distribution": { "Excluded-2d56a": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Basalt-2d56a, which has the unique lowest latency score, with probability one.", "distribution": { "Basalt-2d56a": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-2d56a": "1/8", "Nettle-2d56a": "1/8", "Juniper-2d56a": "1/8", "Amber-2d56a": "1/8", "Elm-2d56a": "1/8", "Lumen-2d56a": "1/8", "Basalt-2d56a": "1/8", "Dune-2d56a": "1/8" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Kestrel-2d56a on every execution, with probability one.", "distribution": { "Kestrel-2d56a": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-21", "english": "A compliance note distinguishes what the request for a survey instrument permits from what it promises.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Dune-2c59c, Onyx-2c59c, Kestrel-2c59c, Iris-2c59c. Recorded queue length scores, in the same order: 76, 47, 42, 69. The reference roster-2c59c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-2c59c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-2c59c@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-2c59c@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-2c59c@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a survey instrument permits from what it promises.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Dune-2c59c, Onyx-2c59c, Kestrel-2c59c, Iris-2c59c. Recorded queue length scores, in the same order: 76, 47, 42, 69. The reference roster-2c59c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-2c59c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-2c59c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Kestrel-2c59c, which has the unique lowest queue length score, with probability one.\nP2: Assigns probability 4/7 to Kestrel-2c59c and 1/7 to each other listed identity, then returns one draw from that distribution.\nP3: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-2c59c on every execution, with probability one.\nP5: Returns Dune-2c59c on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P3; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G1, G2.", "Implementations: P3; claims: G1, G2.", "Implementations: P2, P3, P5; claims: G2.", "Implementations: P3; claims: G1, G2, G3.", "Implementations: P1, P3; claims: G1, G2, G3.", "Implementations: P2, P3, P5; claims: G1, G2.", "Implementations: P1, P2, P3, P5; claims: G2." ], "answer": "Implementations: P3; claims: G1, G2.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-08", "member_count": 4, "world_id": "rt1-world-02-21", "semantic_world": { "n": 4, "members": [ "Dune-2c59c", "Onyx-2c59c", "Kestrel-2c59c", "Iris-2c59c" ], "outsider": "Excluded-2c59c", "scores": [ 76, 47, 42, 69 ], "criterion": "queue length", "set_ref": "roster-2c59c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3; claims: G1, G2.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P3; claims: G1, G2.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3, P5; claims: G2.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P3; claims: G1, G2, G3.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3, P5; claims: G1, G2.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P3, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Kestrel-2c59c, which has the unique lowest queue length score, with probability one.", "distribution": { "Kestrel-2c59c": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Kestrel-2c59c and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-2c59c": "1/7", "Onyx-2c59c": "1/7", "Kestrel-2c59c": "4/7", "Iris-2c59c": "1/7" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-2c59c": "1/4", "Onyx-2c59c": "1/4", "Kestrel-2c59c": "1/4", "Iris-2c59c": "1/4" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-2c59c on every execution, with probability one.", "distribution": { "Excluded-2c59c": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Dune-2c59c on every execution, with probability one.", "distribution": { "Dune-2c59c": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G1", "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-21", "english": "A compliance note distinguishes what the request for a calibration evaluation slot permits from what it promises.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-21876, Lumen-21876, Pike-21876, Dune-21876, Heron-21876, Iris-21876, Juniper-21876. Recorded preparation cost scores, in the same order: 51, 19, 74, 57, 5, 31, 59. The reference roster-21876@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-21876 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-21876@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-21876@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-21876@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a calibration evaluation slot permits from what it promises.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Gale-21876, Lumen-21876, Pike-21876, Dune-21876, Heron-21876, Iris-21876, Juniper-21876. Recorded preparation cost scores, in the same order: 51, 19, 74, 57, 5, 31, 59. The reference roster-21876@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-21876 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-21876@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 4/10 to Dune-21876 and 1/10 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Gale-21876 on every execution, with probability one.\nP4: Returns Excluded-21876 on every execution, with probability one.\nP5: Returns Heron-21876, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2, P3, P5; claims: G1.", "Implementations: P1, P2; claims: G1, G2, G4.", "Implementations: P1; claims: G1, G4.", "Implementations: P1, P3, P5; claims: G1, G3.", "Implementations: P1, P2; claims: G1, G4.", "Implementations: P1; claims: G1, G2, G4.", "Implementations: P1, P3, P5; claims: G1.", "Implementations: P1, P2, P3, P5; claims: G1, G3." ], "answer": "Implementations: P1; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-08", "member_count": 7, "world_id": "rt1-world-03-21", "semantic_world": { "n": 7, "members": [ "Gale-21876", "Lumen-21876", "Pike-21876", "Dune-21876", "Heron-21876", "Iris-21876", "Juniper-21876" ], "outsider": "Excluded-21876", "scores": [ 51, 19, 74, 57, 5, 31, 59 ], "criterion": "preparation cost", "set_ref": "roster-21876@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2; claims: G1, G2, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1; claims: G1, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2; claims: G1, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1; claims: G1, G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P3, P5; claims: G1.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Gale-21876": "1/7", "Lumen-21876": "1/7", "Pike-21876": "1/7", "Dune-21876": "1/7", "Heron-21876": "1/7", "Iris-21876": "1/7", "Juniper-21876": "1/7" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Dune-21876 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Gale-21876": "1/10", "Lumen-21876": "1/10", "Pike-21876": "1/10", "Dune-21876": "2/5", "Heron-21876": "1/10", "Iris-21876": "1/10", "Juniper-21876": "1/10" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Gale-21876 on every execution, with probability one.", "distribution": { "Gale-21876": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-21876 on every execution, with probability one.", "distribution": { "Excluded-21876": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Heron-21876, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Heron-21876": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-21", "english": "A compliance note distinguishes what the request for a crashed scheduler permits from what it promises.\nThe frozen roster lists these eligible standby node identities, in recorded order: Juniper-25c01, Flint-25c01, Amber-25c01. Recorded activation cost scores, in the same order: 47, 64, 30. The reference roster-25c01@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-25c01 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-25c01@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-25c01@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-25c01@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a crashed scheduler permits from what it promises.\nThe frozen roster lists these eligible standby node identities, in recorded order: Juniper-25c01, Flint-25c01, Amber-25c01. Recorded activation cost scores, in the same order: 47, 64, 30. The reference roster-25c01@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-25c01 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-25c01@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Amber-25c01, which has the unique lowest activation cost score, with probability one.\nP3: Returns Juniper-25c01 on every execution, with probability one.\nP4: Returns Excluded-25c01 on every execution, with probability one.\nP5: Assigns probability 4/6 to Flint-25c01 and 1/6 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1; claims: G1.", "Implementations: P1; claims: G1, G4.", "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: P2, P3, P5; claims: G2, G4.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: none; claims: G1, G4.", "Implementations: none; claims: G1.", "Implementations: P2, P3, P5; claims: G4." ], "answer": "Implementations: P1; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-08", "member_count": 3, "world_id": "rt1-world-04-21", "semantic_world": { "n": 3, "members": [ "Juniper-25c01", "Flint-25c01", "Amber-25c01" ], "outsider": "Excluded-25c01", "scores": [ 47, 64, 30 ], "criterion": "activation cost", "set_ref": "roster-25c01@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1; claims: G1.": { "policies": [ "P1" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P5; claims: G2, G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: none; claims: G1.": { "policies": [], "guarantees": [ "G1" ] }, "Implementations: P2, P3, P5; claims: G4.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-25c01": "1/3", "Flint-25c01": "1/3", "Amber-25c01": "1/3" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Amber-25c01, which has the unique lowest activation cost score, with probability one.", "distribution": { "Amber-25c01": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Juniper-25c01 on every execution, with probability one.", "distribution": { "Juniper-25c01": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-25c01 on every execution, with probability one.", "distribution": { "Excluded-25c01": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Flint-25c01 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-25c01": "1/6", "Flint-25c01": "2/3", "Amber-25c01": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-21", "english": "A compliance note distinguishes what the request for a help-centre example slot permits from what it promises.\nThe frozen roster lists these eligible content item identities, in recorded order: Marsh-16f77, Basalt-16f77, Juniper-16f77, Amber-16f77, Pike-16f77, Flint-16f77. Recorded retrieval cost scores, in the same order: 79, 7, 16, 65, 38, 59. The reference roster-16f77@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-16f77 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-16f77@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-16f77@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-16f77@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A compliance note distinguishes what the request for a help-centre example slot permits from what it promises.\nThe frozen roster lists these eligible content item identities, in recorded order: Marsh-16f77, Basalt-16f77, Juniper-16f77, Amber-16f77, Pike-16f77, Flint-16f77. Recorded retrieval cost scores, in the same order: 79, 7, 16, 65, 38, 59. The reference roster-16f77@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-16f77 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-16f77@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-16f77 on every execution, with probability one.\nP2: Assigns probability 4/9 to Pike-16f77 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Basalt-16f77, which has the unique lowest retrieval cost score, with probability one.\nP4: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Excluded-16f77 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P1, P2, P3, P4; claims: G3.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P4; claims: G3.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P1, P2, P3, P4; claims: none.", "Implementations: P4, P5; claims: G1, G3.", "Implementations: P4; claims: G1, G3.", "Implementations: P4, P5; claims: G3." ], "answer": "Implementations: P4; claims: G1, G3.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-08", "member_count": 6, "world_id": "rt1-world-05-21", "semantic_world": { "n": 6, "members": [ "Marsh-16f77", "Basalt-16f77", "Juniper-16f77", "Amber-16f77", "Pike-16f77", "Flint-16f77" ], "outsider": "Excluded-16f77", "scores": [ 79, 7, 16, 65, 38, 59 ], "criterion": "retrieval cost", "set_ref": "roster-16f77@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P4; claims: G3.": { "policies": [ "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P4; claims: none.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [] }, "Implementations: P4, P5; claims: G1, G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4; claims: G1, G3.": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P4, P5; claims: G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Marsh-16f77 on every execution, with probability one.", "distribution": { "Marsh-16f77": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Pike-16f77 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-16f77": "1/9", "Basalt-16f77": "1/9", "Juniper-16f77": "1/9", "Amber-16f77": "1/9", "Pike-16f77": "4/9", "Flint-16f77": "1/9" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Basalt-16f77, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Basalt-16f77": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-16f77": "1/6", "Basalt-16f77": "1/6", "Juniper-16f77": "1/6", "Amber-16f77": "1/6", "Pike-16f77": "1/6", "Flint-16f77": "1/6" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-16f77 on every execution, with probability one.", "distribution": { "Excluded-16f77": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P4" ], "guarantees": [ "G1", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-21", "english": "A compliance note distinguishes what the request for a spell-check job permits from what it promises.\nThe frozen roster lists these eligible worker identities, in recorded order: Kestrel-f0b94, Iris-f0b94. Recorded current load scores, in the same order: 81, 12. The reference roster-f0b94@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f0b94 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f0b94@snapshot-1) requires roster-f0b94@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-f0b94@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A compliance note distinguishes what the request for a spell-check job permits from what it promises.\nThe frozen roster lists these eligible worker identities, in recorded order: Kestrel-f0b94, Iris-f0b94. Recorded current load scores, in the same order: 81, 12. The reference roster-f0b94@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-f0b94 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-f0b94@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-f0b94 on every execution, with probability one.\nP2: Returns Kestrel-f0b94 on every execution, with probability one.\nP3: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 4/5 to Kestrel-f0b94 and 1/5 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Iris-f0b94, which has the unique lowest current load score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2, P3; claims: G2, G3, G4.", "Implementations: P3; claims: G3, G4.", "Implementations: P2, P3; claims: G3, G4.", "Implementations: P2, P3, P4, P5; claims: G4.", "Implementations: P2, P3, P4, P5; claims: G3, G4.", "Implementations: P3, P4, P5; claims: G4.", "Implementations: P3; claims: G2, G3, G4.", "Implementations: P3, P4, P5; claims: G3, G4." ], "answer": "Implementations: P2, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-08", "member_count": 2, "world_id": "rt1-world-06-21", "semantic_world": { "n": 2, "members": [ "Kestrel-f0b94", "Iris-f0b94" ], "outsider": "Excluded-f0b94", "scores": [ 81, 12 ], "criterion": "current load", "set_ref": "roster-f0b94@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3; claims: G2, G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P3; claims: G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3; claims: G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G3, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P3, P4, P5; claims: G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P3; claims: G2, G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P3, P4, P5; claims: G3, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-f0b94 on every execution, with probability one.", "distribution": { "Excluded-f0b94": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Kestrel-f0b94 on every execution, with probability one.", "distribution": { "Kestrel-f0b94": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Kestrel-f0b94": "1/2", "Iris-f0b94": "1/2" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Kestrel-f0b94 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Kestrel-f0b94": "4/5", "Iris-f0b94": "1/5" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Iris-f0b94, which has the unique lowest current load score, with probability one.", "distribution": { "Iris-f0b94": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-22", "english": "A junior engineer asks a senior one what the instruction about a session-refresh call rules in and out.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-d3c4f, Marsh-d3c4f. Recorded latency scores, in the same order: 49, 40. The reference roster-d3c4f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-d3c4f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-d3c4f@snapshot-1) requires roster-d3c4f@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-d3c4f@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A junior engineer asks a senior one what the instruction about a session-refresh call rules in and out.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-d3c4f, Marsh-d3c4f. Recorded latency scores, in the same order: 49, 40. The reference roster-d3c4f@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-d3c4f is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-d3c4f@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-d3c4f on every execution, with probability one.\nP2: Returns Marsh-d3c4f, which has the unique lowest latency score, with probability one.\nP3: Assigns probability 2/3 to Marsh-d3c4f and 1/3 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Flint-d3c4f on every execution, with probability one.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P3, P4, P5; claims: G4.", "Implementations: P2, P5; claims: G2, G3, G4.", "Implementations: P2, P3, P4, P5; claims: G4.", "Implementations: P3, P4, P5; claims: G2, G4.", "Implementations: P2, P3, P4, P5; claims: G2, G4.", "Implementations: P5; claims: G2, G4.", "Implementations: P2, P5; claims: G2, G4.", "Implementations: P5; claims: G2, G3, G4." ], "answer": "Implementations: P2, P3, P4, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-09", "member_count": 2, "world_id": "rt1-world-01-22", "semantic_world": { "n": 2, "members": [ "Flint-d3c4f", "Marsh-d3c4f" ], "outsider": "Excluded-d3c4f", "scores": [ 49, 40 ], "criterion": "latency", "set_ref": "roster-d3c4f@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4, P5; claims: G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P5; claims: G2, G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P3, P4, P5; claims: G2, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G2, G4.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P5; claims: G2, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G2, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3", "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-d3c4f on every execution, with probability one.", "distribution": { "Excluded-d3c4f": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Marsh-d3c4f, which has the unique lowest latency score, with probability one.", "distribution": { "Marsh-d3c4f": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/3 to Marsh-d3c4f and 1/3 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-d3c4f": "1/3", "Marsh-d3c4f": "2/3" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Flint-d3c4f on every execution, with probability one.", "distribution": { "Flint-d3c4f": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-d3c4f": "1/2", "Marsh-d3c4f": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-22", "english": "A junior engineer asks a senior one what the instruction about a preprint revision rules in and out.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Heron-8a3af, Onyx-8a3af, Amber-8a3af, Flint-8a3af, Dune-8a3af. Recorded queue length scores, in the same order: 55, 48, 39, 45, 15. The reference roster-8a3af@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8a3af is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-8a3af@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-8a3af@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-8a3af@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A junior engineer asks a senior one what the instruction about a preprint revision rules in and out.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Heron-8a3af, Onyx-8a3af, Amber-8a3af, Flint-8a3af, Dune-8a3af. Recorded queue length scores, in the same order: 55, 48, 39, 45, 15. The reference roster-8a3af@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8a3af is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-8a3af@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-8a3af on every execution, with probability one.\nP3: Returns Dune-8a3af, which has the unique lowest queue length score, with probability one.\nP4: Assigns probability 2/6 to Amber-8a3af and 1/6 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Heron-8a3af on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P3, P5; claims: G3, G4.", "Implementations: P1, P4; claims: G1, G2, G4.", "Implementations: P1, P3, P4, P5; claims: G4.", "Implementations: P1, P3, P5; claims: G4.", "Implementations: P1; claims: G1, G2, G4.", "Implementations: P1, P3, P4, P5; claims: G3, G4.", "Implementations: P1, P4; claims: G2, G4.", "Implementations: P1; claims: G2, G4." ], "answer": "Implementations: P1; claims: G2, G4.", "settlement_stratum": "draw-uniform", "domain": "reviewer-assignment", "frame_family": "rt1-frame-09", "member_count": 5, "world_id": "rt1-world-02-22", "semantic_world": { "n": 5, "members": [ "Heron-8a3af", "Onyx-8a3af", "Amber-8a3af", "Flint-8a3af", "Dune-8a3af" ], "outsider": "Excluded-8a3af", "scores": [ 55, 48, 39, 45, 15 ], "criterion": "queue length", "set_ref": "roster-8a3af@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P4; claims: G1, G2, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P5; claims: G4.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1; claims: G1, G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G1", "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P4; claims: G2, G4.": { "policies": [ "P1", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1; claims: G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-8a3af": "1/5", "Onyx-8a3af": "1/5", "Amber-8a3af": "1/5", "Flint-8a3af": "1/5", "Dune-8a3af": "1/5" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-8a3af on every execution, with probability one.", "distribution": { "Excluded-8a3af": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Dune-8a3af, which has the unique lowest queue length score, with probability one.", "distribution": { "Dune-8a3af": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/6 to Amber-8a3af and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-8a3af": "1/6", "Onyx-8a3af": "1/6", "Amber-8a3af": "1/3", "Flint-8a3af": "1/6", "Dune-8a3af": "1/6" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Heron-8a3af on every execution, with probability one.", "distribution": { "Heron-8a3af": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-22", "english": "A junior engineer asks a senior one what the instruction about a coding evaluation slot rules in and out.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Lumen-a66d9, Cobalt-a66d9, Marsh-a66d9, Basalt-a66d9, Juniper-a66d9, Gale-a66d9, Kestrel-a66d9, Pike-a66d9. Recorded preparation cost scores, in the same order: 95, 17, 50, 47, 53, 22, 68, 61. The reference roster-a66d9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a66d9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-a66d9@snapshot-1) requires roster-a66d9@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-a66d9@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A junior engineer asks a senior one what the instruction about a coding evaluation slot rules in and out.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Lumen-a66d9, Cobalt-a66d9, Marsh-a66d9, Basalt-a66d9, Juniper-a66d9, Gale-a66d9, Kestrel-a66d9, Pike-a66d9. Recorded preparation cost scores, in the same order: 95, 17, 50, 47, 53, 22, 68, 61. The reference roster-a66d9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a66d9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-a66d9@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/9 to Basalt-a66d9 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Lumen-a66d9 on every execution, with probability one.\nP3: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Excluded-a66d9 on every execution, with probability one.\nP5: Returns Cobalt-a66d9, which has the unique lowest preparation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P5; claims: G4.", "Implementations: P1, P2, P3, P5; claims: G3, G4.", "Implementations: none; claims: G1, G4.", "Implementations: none; claims: G1.", "Implementations: P1, P2, P5; claims: G3, G4.", "Implementations: P3; claims: G1, G4.", "Implementations: P3; claims: G1.", "Implementations: P1, P2, P3, P5; claims: G4." ], "answer": "Implementations: P1, P2, P3, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-09", "member_count": 8, "world_id": "rt1-world-03-22", "semantic_world": { "n": 8, "members": [ "Lumen-a66d9", "Cobalt-a66d9", "Marsh-a66d9", "Basalt-a66d9", "Juniper-a66d9", "Gale-a66d9", "Kestrel-a66d9", "Pike-a66d9" ], "outsider": "Excluded-a66d9", "scores": [ 95, 17, 50, 47, 53, 22, 68, 61 ], "criterion": "preparation cost", "set_ref": "roster-a66d9@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P5; claims: G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G3, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: none; claims: G1, G4.": { "policies": [], "guarantees": [ "G1", "G4" ] }, "Implementations: none; claims: G1.": { "policies": [], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P5; claims: G3, G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P3; claims: G1, G4.": { "policies": [ "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P3; claims: G1.": { "policies": [ "P3" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/9 to Basalt-a66d9 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Lumen-a66d9": "1/9", "Cobalt-a66d9": "1/9", "Marsh-a66d9": "1/9", "Basalt-a66d9": "2/9", "Juniper-a66d9": "1/9", "Gale-a66d9": "1/9", "Kestrel-a66d9": "1/9", "Pike-a66d9": "1/9" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Lumen-a66d9 on every execution, with probability one.", "distribution": { "Lumen-a66d9": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Lumen-a66d9": "1/8", "Cobalt-a66d9": "1/8", "Marsh-a66d9": "1/8", "Basalt-a66d9": "1/8", "Juniper-a66d9": "1/8", "Gale-a66d9": "1/8", "Kestrel-a66d9": "1/8", "Pike-a66d9": "1/8" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-a66d9 on every execution, with probability one.", "distribution": { "Excluded-a66d9": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Cobalt-a66d9, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Cobalt-a66d9": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-22", "english": "A junior engineer asks a senior one what the instruction about a paused webhook relay rules in and out.\nThe frozen roster lists these eligible standby node identities, in recorded order: Basalt-8033c, Kestrel-8033c, Cobalt-8033c, Heron-8033c. Recorded activation cost scores, in the same order: 46, 60, 38, 32. The reference roster-8033c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8033c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-8033c@snapshot-1) requires roster-8033c@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-8033c@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "A junior engineer asks a senior one what the instruction about a paused webhook relay rules in and out.\nThe frozen roster lists these eligible standby node identities, in recorded order: Basalt-8033c, Kestrel-8033c, Cobalt-8033c, Heron-8033c. Recorded activation cost scores, in the same order: 46, 60, 38, 32. The reference roster-8033c@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8033c is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-8033c@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP2: Returns Excluded-8033c on every execution, with probability one.\nP3: Assigns probability 2/5 to Basalt-8033c and 1/5 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Heron-8033c, which has the unique lowest activation cost score, with probability one.\nP5: Returns Basalt-8033c on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P2; claims: G2, G4.", "Implementations: P1, P3, P4, P5; claims: G2.", "Implementations: P1, P2, P3, P4, P5; claims: G2.", "Implementations: P1, P2; claims: G2.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P1; claims: G2, G4.", "Implementations: P1; claims: G2." ], "answer": "Implementations: P1, P3, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-09", "member_count": 4, "world_id": "rt1-world-04-22", "semantic_world": { "n": 4, "members": [ "Basalt-8033c", "Kestrel-8033c", "Cobalt-8033c", "Heron-8033c" ], "outsider": "Excluded-8033c", "scores": [ 46, 60, 38, 32 ], "criterion": "activation cost", "set_ref": "roster-8033c@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2; claims: G2, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2; claims: G2.": { "policies": [ "P1", "P2" ], "guarantees": [ "G2" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1; claims: G2, G4.": { "policies": [ "P1" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1; claims: G2.": { "policies": [ "P1" ], "guarantees": [ "G2" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-8033c": "1/4", "Kestrel-8033c": "1/4", "Cobalt-8033c": "1/4", "Heron-8033c": "1/4" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-8033c on every execution, with probability one.", "distribution": { "Excluded-8033c": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 2/5 to Basalt-8033c and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-8033c": "2/5", "Kestrel-8033c": "1/5", "Cobalt-8033c": "1/5", "Heron-8033c": "1/5" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Heron-8033c, which has the unique lowest activation cost score, with probability one.", "distribution": { "Heron-8033c": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Basalt-8033c on every execution, with probability one.", "distribution": { "Basalt-8033c": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-22", "english": "A junior engineer asks a senior one what the instruction about a release-note screenshot slot rules in and out.\nThe frozen roster lists these eligible content item identities, in recorded order: Heron-a0072, Cobalt-a0072, Onyx-a0072, Amber-a0072, Juniper-a0072, Flint-a0072, Marsh-a0072. Recorded retrieval cost scores, in the same order: 75, 76, 17, 25, 67, 12, 36. The reference roster-a0072@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a0072 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-a0072@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-a0072@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-a0072@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A junior engineer asks a senior one what the instruction about a release-note screenshot slot rules in and out.\nThe frozen roster lists these eligible content item identities, in recorded order: Heron-a0072, Cobalt-a0072, Onyx-a0072, Amber-a0072, Juniper-a0072, Flint-a0072, Marsh-a0072. Recorded retrieval cost scores, in the same order: 75, 76, 17, 25, 67, 12, 36. The reference roster-a0072@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-a0072 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-a0072@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Heron-a0072 on every execution, with probability one.\nP2: Returns Excluded-a0072 on every execution, with probability one.\nP3: Returns Flint-a0072, which has the unique lowest retrieval cost score, with probability one.\nP4: Assigns probability 2/8 to Heron-a0072 and 1/8 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P1, P5; claims: G3, G4.", "Implementations: P1, P3, P4, P5; claims: G3, G4.", "Implementations: P5; claims: G3, G4.", "Implementations: P1, P5; claims: G2, G3, G4.", "Implementations: P3, P4, P5; claims: G3, G4.", "Implementations: P5; claims: G2, G3, G4.", "Implementations: P3, P4, P5; claims: G3.", "Implementations: P1, P3, P4, P5; claims: G3." ], "answer": "Implementations: P5; claims: G3, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-09", "member_count": 7, "world_id": "rt1-world-05-22", "semantic_world": { "n": 7, "members": [ "Heron-a0072", "Cobalt-a0072", "Onyx-a0072", "Amber-a0072", "Juniper-a0072", "Flint-a0072", "Marsh-a0072" ], "outsider": "Excluded-a0072", "scores": [ 75, 76, 17, 25, 67, 12, 36 ], "criterion": "retrieval cost", "set_ref": "roster-a0072@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P5; claims: G3, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P3, P4, P5; claims: G3, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P5; claims: G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P5; claims: G2, G3, G4.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P3, P4, P5; claims: G3, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P5; claims: G2, G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P3, P4, P5; claims: G3.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Heron-a0072 on every execution, with probability one.", "distribution": { "Heron-a0072": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-a0072 on every execution, with probability one.", "distribution": { "Excluded-a0072": "1" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Flint-a0072, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Flint-a0072": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 2/8 to Heron-a0072 and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Heron-a0072": "1/4", "Cobalt-a0072": "1/8", "Onyx-a0072": "1/8", "Amber-a0072": "1/8", "Juniper-a0072": "1/8", "Flint-a0072": "1/8", "Marsh-a0072": "1/8" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Heron-a0072": "1/7", "Cobalt-a0072": "1/7", "Onyx-a0072": "1/7", "Amber-a0072": "1/7", "Juniper-a0072": "1/7", "Flint-a0072": "1/7", "Marsh-a0072": "1/7" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G3", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-22", "english": "A junior engineer asks a senior one what the instruction about a geocoding job rules in and out.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-b9179, Dune-b9179, Basalt-b9179. Recorded current load scores, in the same order: 34, 17, 5. The reference roster-b9179@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b9179 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b9179@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-b9179@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-b9179@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "A junior engineer asks a senior one what the instruction about a geocoding job rules in and out.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-b9179, Dune-b9179, Basalt-b9179. Recorded current load scores, in the same order: 34, 17, 5. The reference roster-b9179@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-b9179 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-b9179@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 2/4 to Amber-b9179 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP2: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Basalt-b9179, which has the unique lowest current load score, with probability one.\nP4: Returns Excluded-b9179 on every execution, with probability one.\nP5: Returns Amber-b9179 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P2, P3; claims: G1, G4.", "Implementations: P2; claims: G1, G3, G4.", "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: P2, P3; claims: G1, G3, G4.", "Implementations: P1, P2, P5; claims: G4.", "Implementations: P2; claims: G1, G4.", "Implementations: P1, P2, P5; claims: G2, G4." ], "answer": "Implementations: P2; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "resource-allocation", "frame_family": "rt1-frame-09", "member_count": 3, "world_id": "rt1-world-06-22", "semantic_world": { "n": 3, "members": [ "Amber-b9179", "Dune-b9179", "Basalt-b9179" ], "outsider": "Excluded-b9179", "scores": [ 34, 17, 5 ], "criterion": "current load", "set_ref": "roster-b9179@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P3; claims: G1, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3; claims: G1, G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P1, P2, P5; claims: G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 2/4 to Amber-b9179 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-b9179": "1/2", "Dune-b9179": "1/4", "Basalt-b9179": "1/4" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-b9179": "1/3", "Dune-b9179": "1/3", "Basalt-b9179": "1/3" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Basalt-b9179, which has the unique lowest current load score, with probability one.", "distribution": { "Basalt-b9179": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-b9179 on every execution, with probability one.", "distribution": { "Excluded-b9179": "1" }, "label": "P4" }, { "kind": "constant-first", "text": "Returns Amber-b9179 on every execution, with probability one.", "distribution": { "Amber-b9179": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-23", "english": "The dispatch log will record the interpretation of the request for a price-quote request; nothing has run yet.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Basalt-20398, Onyx-20398, Cobalt-20398. Recorded latency scores, in the same order: 78, 8, 39. The reference roster-20398@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-20398 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-20398@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-20398@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-20398@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The dispatch log will record the interpretation of the request for a price-quote request; nothing has run yet.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Basalt-20398, Onyx-20398, Cobalt-20398. Recorded latency scores, in the same order: 78, 8, 39. The reference roster-20398@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-20398 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-20398@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Basalt-20398 on every execution, with probability one.\nP2: Returns Excluded-20398 on every execution, with probability one.\nP3: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\nP4: Returns Onyx-20398, which has the unique lowest latency score, with probability one.\nP5: Assigns probability 3/5 to Basalt-20398 and 1/5 to each other listed identity, then returns one draw from that distribution.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P3; claims: G1, G2, G3.", "Implementations: P3, P5; claims: G1, G2, G3.", "Implementations: P1, P3, P4, P5; claims: G2, G4.", "Implementations: P1, P3, P4; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G2.", "Implementations: P1, P3, P4; claims: G2, G4.", "Implementations: P3, P5; claims: G2, G3.", "Implementations: P3; claims: G2, G3." ], "answer": "Implementations: P3; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "service-routing", "frame_family": "rt1-frame-10", "member_count": 3, "world_id": "rt1-world-01-23", "semantic_world": { "n": 3, "members": [ "Basalt-20398", "Onyx-20398", "Cobalt-20398" ], "outsider": "Excluded-20398", "scores": [ 78, 8, 39 ], "criterion": "latency", "set_ref": "roster-20398@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3; claims: G1, G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P3, P5; claims: G1, G2, G3.": { "policies": [ "P3", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P4; claims: G2.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P3, P5; claims: G2, G3.": { "policies": [ "P3", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P3; claims: G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Basalt-20398 on every execution, with probability one.", "distribution": { "Basalt-20398": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-20398 on every execution, with probability one.", "distribution": { "Excluded-20398": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-20398": "1/3", "Onyx-20398": "1/3", "Cobalt-20398": "1/3" }, "label": "P3" }, { "kind": "criterion-based", "text": "Returns Onyx-20398, which has the unique lowest latency score, with probability one.", "distribution": { "Onyx-20398": "1" }, "label": "P4" }, { "kind": "unequal-weight", "text": "Assigns probability 3/5 to Basalt-20398 and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-20398": "3/5", "Onyx-20398": "1/5", "Cobalt-20398": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P3" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-23", "english": "The dispatch log will record the interpretation of the request for a thesis chapter; nothing has run yet.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Nettle-7ef0d, Onyx-7ef0d, Heron-7ef0d, Pike-7ef0d, Marsh-7ef0d, Lumen-7ef0d. Recorded queue length scores, in the same order: 96, 61, 31, 93, 43, 55. The reference roster-7ef0d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7ef0d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-7ef0d@snapshot-1) requires roster-7ef0d@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-7ef0d@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a thesis chapter; nothing has run yet.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Nettle-7ef0d, Onyx-7ef0d, Heron-7ef0d, Pike-7ef0d, Marsh-7ef0d, Lumen-7ef0d. Recorded queue length scores, in the same order: 96, 61, 31, 93, 43, 55. The reference roster-7ef0d@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7ef0d is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-7ef0d@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-7ef0d on every execution, with probability one.\nP2: Returns Heron-7ef0d, which has the unique lowest queue length score, with probability one.\nP3: Returns Nettle-7ef0d on every execution, with probability one.\nP4: Assigns probability 3/8 to Nettle-7ef0d and 1/8 to each other listed identity, then returns one draw from that distribution.\nP5: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: A well-informed adversary cannot predict the result better than chance.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P5; claims: G4.", "Implementations: P2, P3, P4, P5; claims: G1, G2.", "Implementations: P2, P3, P4, P5; claims: G2.", "Implementations: P5; claims: G2, G4.", "Implementations: P2, P3, P4; claims: G1, G2.", "Implementations: none; claims: G2, G4.", "Implementations: P2, P3, P4; claims: G2.", "Implementations: none; claims: G4." ], "answer": "Implementations: P2, P3, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-10", "member_count": 6, "world_id": "rt1-world-02-23", "semantic_world": { "n": 6, "members": [ "Nettle-7ef0d", "Onyx-7ef0d", "Heron-7ef0d", "Pike-7ef0d", "Marsh-7ef0d", "Lumen-7ef0d" ], "outsider": "Excluded-7ef0d", "scores": [ 96, 61, 31, 93, 43, 55 ], "criterion": "queue length", "set_ref": "roster-7ef0d@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G4.": { "policies": [ "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G1, G2.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P2, P3, P4, P5; claims: G2.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P5; claims: G2, G4.": { "policies": [ "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P4; claims: G1, G2.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: none; claims: G2, G4.": { "policies": [], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P4; claims: G2.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G2" ] }, "Implementations: none; claims: G4.": { "policies": [], "guarantees": [ "G4" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-7ef0d on every execution, with probability one.", "distribution": { "Excluded-7ef0d": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Heron-7ef0d, which has the unique lowest queue length score, with probability one.", "distribution": { "Heron-7ef0d": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Nettle-7ef0d on every execution, with probability one.", "distribution": { "Nettle-7ef0d": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/8 to Nettle-7ef0d and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Nettle-7ef0d": "3/8", "Onyx-7ef0d": "1/8", "Heron-7ef0d": "1/8", "Pike-7ef0d": "1/8", "Marsh-7ef0d": "1/8", "Lumen-7ef0d": "1/8" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Nettle-7ef0d": "1/6", "Onyx-7ef0d": "1/6", "Heron-7ef0d": "1/6", "Pike-7ef0d": "1/6", "Marsh-7ef0d": "1/6", "Lumen-7ef0d": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-23", "english": "The dispatch log will record the interpretation of the request for a citation evaluation slot; nothing has run yet.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Dune-cdbc9, Elm-cdbc9. Recorded preparation cost scores, in the same order: 36, 7. The reference roster-cdbc9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-cdbc9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-cdbc9@snapshot-1) requires roster-cdbc9@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-cdbc9@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a citation evaluation slot; nothing has run yet.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Dune-cdbc9, Elm-cdbc9. Recorded preparation cost scores, in the same order: 36, 7. The reference roster-cdbc9@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-cdbc9 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-cdbc9@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 3/4 to Dune-cdbc9 and 1/4 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Elm-cdbc9, which has the unique lowest preparation cost score, with probability one.\nP3: Returns Dune-cdbc9 on every execution, with probability one.\nP4: Returns Excluded-cdbc9 on every execution, with probability one.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P5; claims: G2, G3.", "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1, P2, P3, P4, P5; claims: G3.", "Implementations: P5; claims: G3.", "Implementations: P4, P5; claims: G3.", "Implementations: P1, P2, P3, P5; claims: none.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: P4, P5; claims: G2, G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-10", "member_count": 2, "world_id": "rt1-world-03-23", "semantic_world": { "n": 2, "members": [ "Dune-cdbc9", "Elm-cdbc9" ], "outsider": "Excluded-cdbc9", "scores": [ 36, 7 ], "criterion": "preparation cost", "set_ref": "roster-cdbc9@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P5; claims: G3.": { "policies": [ "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4, P5; claims: G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P4, P5; claims: G2, G3.": { "policies": [ "P4", "P5" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 3/4 to Dune-cdbc9 and 1/4 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Dune-cdbc9": "3/4", "Elm-cdbc9": "1/4" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Elm-cdbc9, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Elm-cdbc9": "1" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Dune-cdbc9 on every execution, with probability one.", "distribution": { "Dune-cdbc9": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-cdbc9 on every execution, with probability one.", "distribution": { "Excluded-cdbc9": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Dune-cdbc9": "1/2", "Elm-cdbc9": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-23", "english": "The dispatch log will record the interpretation of the request for a degraded search shard; nothing has run yet.\nThe frozen roster lists these eligible standby node identities, in recorded order: Marsh-93f26, Nettle-93f26, Onyx-93f26, Heron-93f26, Lumen-93f26. Recorded activation cost scores, in the same order: 22, 78, 72, 80, 3. The reference roster-93f26@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-93f26 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-93f26@snapshot-1) requires roster-93f26@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-93f26@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a degraded search shard; nothing has run yet.\nThe frozen roster lists these eligible standby node identities, in recorded order: Marsh-93f26, Nettle-93f26, Onyx-93f26, Heron-93f26, Lumen-93f26. Recorded activation cost scores, in the same order: 22, 78, 72, 80, 3. The reference roster-93f26@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-93f26 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-93f26@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Excluded-93f26 on every execution, with probability one.\nP2: Returns Marsh-93f26 on every execution, with probability one.\nP3: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\nP4: Assigns probability 3/7 to Marsh-93f26 and 1/7 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Lumen-93f26, which has the unique lowest activation cost score, with probability one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Before this selection, every distinct eligible identity has the same probability of being returned.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P3, P4, P5; claims: G2, G3.", "Implementations: P2, P3, P4, P5; claims: G2, G3.", "Implementations: P2, P3; claims: G2, G3.", "Implementations: P2, P3; claims: G2, G3, G4.", "Implementations: P3, P4, P5; claims: G3.", "Implementations: P3; claims: G2, G3, G4.", "Implementations: P2, P3, P4, P5; claims: G3.", "Implementations: P3; claims: G2, G3." ], "answer": "Implementations: P2, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "failover", "frame_family": "rt1-frame-10", "member_count": 5, "world_id": "rt1-world-04-23", "semantic_world": { "n": 5, "members": [ "Marsh-93f26", "Nettle-93f26", "Onyx-93f26", "Heron-93f26", "Lumen-93f26" ], "outsider": "Excluded-93f26", "scores": [ 22, 78, 72, 80, 3 ], "criterion": "activation cost", "set_ref": "roster-93f26@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4, P5; claims: G2, G3.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3, P4, P5; claims: G2, G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3; claims: G2, G3.": { "policies": [ "P2", "P3" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P2, P3; claims: G2, G3, G4.": { "policies": [ "P2", "P3" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P3, P4, P5; claims: G3.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P3; claims: G2, G3, G4.": { "policies": [ "P3" ], "guarantees": [ "G2", "G3", "G4" ] }, "Implementations: P2, P3, P4, P5; claims: G3.": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P3; claims: G2, G3.": { "policies": [ "P3" ], "guarantees": [ "G2", "G3" ] } }, "policies": [ { "kind": "out-of-set", "text": "Returns Excluded-93f26 on every execution, with probability one.", "distribution": { "Excluded-93f26": "1" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Marsh-93f26 on every execution, with probability one.", "distribution": { "Marsh-93f26": "1" }, "label": "P2" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-93f26": "1/5", "Nettle-93f26": "1/5", "Onyx-93f26": "1/5", "Heron-93f26": "1/5", "Lumen-93f26": "1/5" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/7 to Marsh-93f26 and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-93f26": "3/7", "Nettle-93f26": "1/7", "Onyx-93f26": "1/7", "Heron-93f26": "1/7", "Lumen-93f26": "1/7" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Lumen-93f26, which has the unique lowest activation cost score, with probability one.", "distribution": { "Lumen-93f26": "1" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P2", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-23", "english": "The dispatch log will record the interpretation of the request for a podcast intro slot; nothing has run yet.\nThe frozen roster lists these eligible content item identities, in recorded order: Basalt-8ccc2, Dune-8ccc2, Flint-8ccc2, Marsh-8ccc2, Elm-8ccc2, Iris-8ccc2, Heron-8ccc2, Kestrel-8ccc2. Recorded retrieval cost scores, in the same order: 34, 45, 81, 42, 65, 93, 59, 17. The reference roster-8ccc2@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8ccc2 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-8ccc2@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-8ccc2@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-8ccc2@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "The dispatch log will record the interpretation of the request for a podcast intro slot; nothing has run yet.\nThe frozen roster lists these eligible content item identities, in recorded order: Basalt-8ccc2, Dune-8ccc2, Flint-8ccc2, Marsh-8ccc2, Elm-8ccc2, Iris-8ccc2, Heron-8ccc2, Kestrel-8ccc2. Recorded retrieval cost scores, in the same order: 34, 45, 81, 42, 65, 93, 59, 17. The reference roster-8ccc2@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-8ccc2 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-8ccc2@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Kestrel-8ccc2, which has the unique lowest retrieval cost score, with probability one.\nP2: Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.\nP3: Returns Basalt-8ccc2 on every execution, with probability one.\nP4: Assigns probability 3/10 to Elm-8ccc2 and 1/10 to each other listed identity, then returns one draw from that distribution.\nP5: Returns Excluded-8ccc2 on every execution, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P2, P3, P4; claims: G2, G4.", "Implementations: P2, P3, P4; claims: G2, G4.", "Implementations: P2; claims: G1, G3, G4.", "Implementations: P2, P3, P4; claims: G4.", "Implementations: P1, P2, P3, P4; claims: G4.", "Implementations: P1, P2; claims: G1, G3, G4.", "Implementations: P2; claims: G1, G4.", "Implementations: P1, P2; claims: G1, G4." ], "answer": "Implementations: P2; claims: G1, G4.", "settlement_stratum": "draw-uniform", "domain": "content-choice", "frame_family": "rt1-frame-10", "member_count": 8, "world_id": "rt1-world-05-23", "semantic_world": { "n": 8, "members": [ "Basalt-8ccc2", "Dune-8ccc2", "Flint-8ccc2", "Marsh-8ccc2", "Elm-8ccc2", "Iris-8ccc2", "Heron-8ccc2", "Kestrel-8ccc2" ], "outsider": "Excluded-8ccc2", "scores": [ 34, 45, 81, 42, 65, 93, 59, 17 ], "criterion": "retrieval cost", "set_ref": "roster-8ccc2@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2, P3, P4; claims: G2, G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P2; claims: G1, G3, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2, P3, P4; claims: G4.": { "policies": [ "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2, P3, P4; claims: G4.": { "policies": [ "P1", "P2", "P3", "P4" ], "guarantees": [ "G4" ] }, "Implementations: P1, P2; claims: G1, G3, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G3", "G4" ] }, "Implementations: P2; claims: G1, G4.": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] }, "Implementations: P1, P2; claims: G1, G4.": { "policies": [ "P1", "P2" ], "guarantees": [ "G1", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Kestrel-8ccc2, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Kestrel-8ccc2": "1" }, "label": "P1" }, { "kind": "equal-probability", "text": "Assigns probability 1/8 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Basalt-8ccc2": "1/8", "Dune-8ccc2": "1/8", "Flint-8ccc2": "1/8", "Marsh-8ccc2": "1/8", "Elm-8ccc2": "1/8", "Iris-8ccc2": "1/8", "Heron-8ccc2": "1/8", "Kestrel-8ccc2": "1/8" }, "label": "P2" }, { "kind": "constant-first", "text": "Returns Basalt-8ccc2 on every execution, with probability one.", "distribution": { "Basalt-8ccc2": "1" }, "label": "P3" }, { "kind": "unequal-weight", "text": "Assigns probability 3/10 to Elm-8ccc2 and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Basalt-8ccc2": "1/10", "Dune-8ccc2": "1/10", "Flint-8ccc2": "1/10", "Marsh-8ccc2": "1/10", "Elm-8ccc2": "3/10", "Iris-8ccc2": "1/10", "Heron-8ccc2": "1/10", "Kestrel-8ccc2": "1/10" }, "label": "P4" }, { "kind": "out-of-set", "text": "Returns Excluded-8ccc2 on every execution, with probability one.", "distribution": { "Excluded-8ccc2": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P2" ], "guarantees": [ "G1", "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-23", "english": "The dispatch log will record the interpretation of the request for a log-rotation job; nothing has run yet.\nThe frozen roster lists these eligible worker identities, in recorded order: Juniper-76353, Iris-76353, Nettle-76353, Onyx-76353. Recorded current load scores, in the same order: 94, 35, 44, 3. The reference roster-76353@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-76353 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-76353@snapshot-1) requires roster-76353@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-76353@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "The dispatch log will record the interpretation of the request for a log-rotation job; nothing has run yet.\nThe frozen roster lists these eligible worker identities, in recorded order: Juniper-76353, Iris-76353, Nettle-76353, Onyx-76353. Recorded current load scores, in the same order: 94, 35, 44, 3. The reference roster-76353@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-76353 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-76353@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Juniper-76353 on every execution, with probability one.\nP2: Assigns probability 3/6 to Juniper-76353 and 1/6 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Onyx-76353, which has the unique lowest current load score, with probability one.\nP4: Returns Excluded-76353 on every execution, with probability one.\nP5: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Before this selection, every distinct eligible identity has the same probability of being returned.", "options": [ "Implementations: P2, P5; claims: G4.", "Implementations: P2, P5; claims: G3, G4.", "Implementations: P5; claims: G4.", "Implementations: P1, P3, P5; claims: G1, G3.", "Implementations: P5; claims: G3, G4.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: P1, P3, P5; claims: G3.", "Implementations: P1, P2, P3, P5; claims: G1, G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-10", "member_count": 4, "world_id": "rt1-world-06-23", "semantic_world": { "n": 4, "members": [ "Juniper-76353", "Iris-76353", "Nettle-76353", "Onyx-76353" ], "outsider": "Excluded-76353", "scores": [ 94, 35, 44, 3 ], "criterion": "current load", "set_ref": "roster-76353@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P2, P5; claims: G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P2, P5; claims: G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P5; claims: G4.": { "policies": [ "P5" ], "guarantees": [ "G4" ] }, "Implementations: P1, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P5; claims: G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P5; claims: G3.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Juniper-76353 on every execution, with probability one.", "distribution": { "Juniper-76353": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 3/6 to Juniper-76353 and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Juniper-76353": "1/2", "Iris-76353": "1/6", "Nettle-76353": "1/6", "Onyx-76353": "1/6" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Onyx-76353, which has the unique lowest current load score, with probability one.", "distribution": { "Onyx-76353": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-76353 on every execution, with probability one.", "distribution": { "Excluded-76353": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Juniper-76353": "1/4", "Iris-76353": "1/4", "Nettle-76353": "1/4", "Onyx-76353": "1/4" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-01-24", "english": "An incident retrospective re-reads the original selection request for a font-subset request.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-062bf, Dune-062bf, Lumen-062bf, Juniper-062bf. Recorded latency scores, in the same order: 18, 61, 48, 16. The reference roster-062bf@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-062bf is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-062bf@snapshot-1) requires roster-062bf@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-062bf@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "An incident retrospective re-reads the original selection request for a font-subset request.\nThe frozen roster lists these eligible service endpoint identities, in recorded order: Flint-062bf, Dune-062bf, Lumen-062bf, Juniper-062bf. Recorded latency scores, in the same order: 18, 61, 48, 16. The reference roster-062bf@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-062bf is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-062bf@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 4/7 to Lumen-062bf and 1/7 to each other listed identity, then returns one draw from that distribution.\nP2: Returns Flint-062bf on every execution, with probability one.\nP3: Returns Excluded-062bf on every execution, with probability one.\nP4: Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.\nP5: Returns Juniper-062bf, which has the unique lowest latency score, with probability one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Learning this result leaves the probability distribution of a later selection unchanged.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: none; claims: G1, G2.", "Implementations: P1, P2, P4, P5; claims: G2.", "Implementations: P4; claims: G1, G2.", "Implementations: P1, P2, P5; claims: G2, G3.", "Implementations: P1, P2, P5; claims: G2.", "Implementations: none; claims: G1.", "Implementations: P1, P2, P4, P5; claims: G2, G3.", "Implementations: P4; claims: G1." ], "answer": "Implementations: P1, P2, P4, P5; claims: G2.", "settlement_stratum": "choose-any", "domain": "service-routing", "frame_family": "rt1-frame-11", "member_count": 4, "world_id": "rt1-world-01-24", "semantic_world": { "n": 4, "members": [ "Flint-062bf", "Dune-062bf", "Lumen-062bf", "Juniper-062bf" ], "outsider": "Excluded-062bf", "scores": [ 18, 61, 48, 16 ], "criterion": "latency", "set_ref": "roster-062bf@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: none; claims: G1, G2.": { "policies": [], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P4, P5; claims: G2.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P4; claims: G1, G2.": { "policies": [ "P4" ], "guarantees": [ "G1", "G2" ] }, "Implementations: P1, P2, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P2, P5; claims: G2.": { "policies": [ "P1", "P2", "P5" ], "guarantees": [ "G2" ] }, "Implementations: none; claims: G1.": { "policies": [], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4, P5; claims: G2, G3.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P4; claims: G1.": { "policies": [ "P4" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "unequal-weight", "text": "Assigns probability 4/7 to Lumen-062bf and 1/7 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Flint-062bf": "1/7", "Dune-062bf": "1/7", "Lumen-062bf": "4/7", "Juniper-062bf": "1/7" }, "label": "P1" }, { "kind": "constant-first", "text": "Returns Flint-062bf on every execution, with probability one.", "distribution": { "Flint-062bf": "1" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-062bf on every execution, with probability one.", "distribution": { "Excluded-062bf": "1" }, "label": "P3" }, { "kind": "equal-probability", "text": "Assigns probability 1/4 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Flint-062bf": "1/4", "Dune-062bf": "1/4", "Lumen-062bf": "1/4", "Juniper-062bf": "1/4" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Juniper-062bf, which has the unique lowest latency score, with probability one.", "distribution": { "Juniper-062bf": "1" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G2" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-02-24", "english": "An incident retrospective re-reads the original selection request for a benchmark submission.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Elm-90a3e, Pike-90a3e, Iris-90a3e, Gale-90a3e, Basalt-90a3e, Kestrel-90a3e, Lumen-90a3e. Recorded queue length scores, in the same order: 61, 50, 52, 35, 37, 4, 41. The reference roster-90a3e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-90a3e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-90a3e@snapshot-1) requires roster-90a3e@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-90a3e@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "An incident retrospective re-reads the original selection request for a benchmark submission.\nThe frozen roster lists these eligible reviewer identities, in recorded order: Elm-90a3e, Pike-90a3e, Iris-90a3e, Gale-90a3e, Basalt-90a3e, Kestrel-90a3e, Lumen-90a3e. Recorded queue length scores, in the same order: 61, 50, 52, 35, 37, 4, 41. The reference roster-90a3e@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-90a3e is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-90a3e@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.\nP2: Assigns probability 4/10 to Kestrel-90a3e and 1/10 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Excluded-90a3e on every execution, with probability one.\nP4: Returns Elm-90a3e on every execution, with probability one.\nP5: Returns Kestrel-90a3e, which has the unique lowest queue length score, with probability one.\n\nClaims an auditor might make:\nG1: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P1, P2, P3, P4, P5; claims: none.", "Implementations: P1, P2, P3, P4, P5; claims: G1.", "Implementations: P1; claims: G1, G3.", "Implementations: P1, P3; claims: G1.", "Implementations: P1, P2, P4, P5; claims: G1.", "Implementations: P1, P3; claims: G1, G3.", "Implementations: P1, P2, P4, P5; claims: none.", "Implementations: P1; claims: G1." ], "answer": "Implementations: P1, P2, P4, P5; claims: G1.", "settlement_stratum": "choose-any", "domain": "reviewer-assignment", "frame_family": "rt1-frame-11", "member_count": 7, "world_id": "rt1-world-02-24", "semantic_world": { "n": 7, "members": [ "Elm-90a3e", "Pike-90a3e", "Iris-90a3e", "Gale-90a3e", "Basalt-90a3e", "Kestrel-90a3e", "Lumen-90a3e" ], "outsider": "Excluded-90a3e", "scores": [ 61, 50, 52, 35, 37, 4, 41 ], "criterion": "queue length", "set_ref": "roster-90a3e@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P2, P3, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P2, P3, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P3", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1; claims: G1, G3.": { "policies": [ "P1" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P3; claims: G1.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1" ] }, "Implementations: P1, P2, P4, P5; claims: G1.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] }, "Implementations: P1, P3; claims: G1, G3.": { "policies": [ "P1", "P3" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P4, P5; claims: none.": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [] }, "Implementations: P1; claims: G1.": { "policies": [ "P1" ], "guarantees": [ "G1" ] } }, "policies": [ { "kind": "equal-probability", "text": "Assigns probability 1/7 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Elm-90a3e": "1/7", "Pike-90a3e": "1/7", "Iris-90a3e": "1/7", "Gale-90a3e": "1/7", "Basalt-90a3e": "1/7", "Kestrel-90a3e": "1/7", "Lumen-90a3e": "1/7" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/10 to Kestrel-90a3e and 1/10 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Elm-90a3e": "1/10", "Pike-90a3e": "1/10", "Iris-90a3e": "1/10", "Gale-90a3e": "1/10", "Basalt-90a3e": "1/10", "Kestrel-90a3e": "2/5", "Lumen-90a3e": "1/10" }, "label": "P2" }, { "kind": "out-of-set", "text": "Returns Excluded-90a3e on every execution, with probability one.", "distribution": { "Excluded-90a3e": "1" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Elm-90a3e on every execution, with probability one.", "distribution": { "Elm-90a3e": "1" }, "label": "P4" }, { "kind": "criterion-based", "text": "Returns Kestrel-90a3e, which has the unique lowest queue length score, with probability one.", "distribution": { "Kestrel-90a3e": "1" }, "label": "P5" } ], "guarantees": [ { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P4", "P5" ], "guarantees": [ "G1" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-03-24", "english": "An incident retrospective re-reads the original selection request for a dialogue evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Marsh-7e0fc, Nettle-7e0fc, Gale-7e0fc. Recorded preparation cost scores, in the same order: 63, 35, 34. The reference roster-7e0fc@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7e0fc is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-7e0fc@snapshot-1) requires roster-7e0fc@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-7e0fc@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "An incident retrospective re-reads the original selection request for a dialogue evaluation slot.\nThe frozen roster lists these eligible evaluation item identities, in recorded order: Marsh-7e0fc, Nettle-7e0fc, Gale-7e0fc. Recorded preparation cost scores, in the same order: 63, 35, 34. The reference roster-7e0fc@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-7e0fc is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-7e0fc@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Marsh-7e0fc on every execution, with probability one.\nP2: Returns Gale-7e0fc, which has the unique lowest preparation cost score, with probability one.\nP3: Assigns probability 4/6 to Marsh-7e0fc and 1/6 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Excluded-7e0fc on every execution, with probability one.\nP5: Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P5; claims: G1, G2, G3.", "Implementations: P2, P3, P5; claims: G3.", "Implementations: P1, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: G3.", "Implementations: P1, P5; claims: G1, G2, G3.", "Implementations: P2, P3, P5; claims: G1, G3.", "Implementations: P1, P2, P3, P5; claims: G1, G3.", "Implementations: P5; claims: G1, G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "evaluation-item-selection", "frame_family": "rt1-frame-11", "member_count": 3, "world_id": "rt1-world-03-24", "semantic_world": { "n": 3, "members": [ "Marsh-7e0fc", "Nettle-7e0fc", "Gale-7e0fc" ], "outsider": "Excluded-7e0fc", "scores": [ 63, 35, 34 ], "criterion": "preparation cost", "set_ref": "roster-7e0fc@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3, P5; claims: G3.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P5; claims: G1, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P5; claims: G1, G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P2, P3, P5; claims: G1, G3.": { "policies": [ "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P2, P3, P5; claims: G1, G3.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: P5; claims: G1, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Marsh-7e0fc on every execution, with probability one.", "distribution": { "Marsh-7e0fc": "1" }, "label": "P1" }, { "kind": "criterion-based", "text": "Returns Gale-7e0fc, which has the unique lowest preparation cost score, with probability one.", "distribution": { "Gale-7e0fc": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/6 to Marsh-7e0fc and 1/6 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Marsh-7e0fc": "2/3", "Nettle-7e0fc": "1/6", "Gale-7e0fc": "1/6" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-7e0fc on every execution, with probability one.", "distribution": { "Excluded-7e0fc": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/3 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Marsh-7e0fc": "1/3", "Nettle-7e0fc": "1/3", "Gale-7e0fc": "1/3" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-04-24", "english": "An incident retrospective re-reads the original selection request for a halted backup agent.\nThe frozen roster lists these eligible standby node identities, in recorded order: Onyx-aed23, Gale-aed23, Cobalt-aed23, Basalt-aed23, Elm-aed23, Dune-aed23. Recorded activation cost scores, in the same order: 45, 67, 34, 32, 83, 4. The reference roster-aed23@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-aed23 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-aed23@snapshot-1) denotes exactly one stochastic draw from the same kind of resolved set. Conditional on the frozen eligibility set and before the outcome is known, every distinct member identity has probability exactly 1/|roster-aed23@snapshot-1| of being returned. Duplicate rows do not buy an identity extra probability: identity and deduplication rules belong to roster-aed23@snapshot-1 and must be resolved before the draw. An empty, ambiguous, or concurrently mutable set without a version or time anchor is invalid rather than guessed. draw-uniform does not by itself promise cryptographic unpredictability, public verifiability, independence between repeated draws, sampling with or without replacement, or a balanced finite run; it specifies one draw only. A realized outcome alone neither proves nor refutes its distribution.", "ainglish": "An incident retrospective re-reads the original selection request for a halted backup agent.\nThe frozen roster lists these eligible standby node identities, in recorded order: Onyx-aed23, Gale-aed23, Cobalt-aed23, Basalt-aed23, Elm-aed23, Dune-aed23. Recorded activation cost scores, in the same order: 45, 67, 34, 32, 83, 4. The reference roster-aed23@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-aed23 is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: draw-uniform(roster-aed23@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Dune-aed23, which has the unique lowest activation cost score, with probability one.\nP2: Returns Excluded-aed23 on every execution, with probability one.\nP3: Assigns probability 4/9 to Dune-aed23 and 1/9 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Onyx-aed23 on every execution, with probability one.\nP5: Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Learning this result leaves the probability distribution of a later selection unchanged.\nG2: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: A well-informed adversary cannot predict the result better than chance.", "options": [ "Implementations: P3, P4, P5; claims: G2, G4.", "Implementations: P5; claims: G2, G3.", "Implementations: P1, P5; claims: G1, G2, G3.", "Implementations: P5; claims: G1, G2, G3.", "Implementations: P1, P5; claims: G2, G3.", "Implementations: P3, P4, P5; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G2.", "Implementations: P1, P3, P4, P5; claims: G2, G4." ], "answer": "Implementations: P5; claims: G2, G3.", "settlement_stratum": "draw-uniform", "domain": "failover", "frame_family": "rt1-frame-11", "member_count": 6, "world_id": "rt1-world-04-24", "semantic_world": { "n": 6, "members": [ "Onyx-aed23", "Gale-aed23", "Cobalt-aed23", "Basalt-aed23", "Elm-aed23", "Dune-aed23" ], "outsider": "Excluded-aed23", "scores": [ 45, 67, 34, 32, 83, 4 ], "criterion": "activation cost", "set_ref": "roster-aed23@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P3, P4, P5; claims: G2, G4.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P5; claims: G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P1, P5; claims: G1, G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P5; claims: G1, G2, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G2", "G3" ] }, "Implementations: P1, P5; claims: G2, G3.": { "policies": [ "P1", "P5" ], "guarantees": [ "G2", "G3" ] }, "Implementations: P3, P4, P5; claims: G2.": { "policies": [ "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G2.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2" ] }, "Implementations: P1, P3, P4, P5; claims: G2, G4.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G2", "G4" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Dune-aed23, which has the unique lowest activation cost score, with probability one.", "distribution": { "Dune-aed23": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-aed23 on every execution, with probability one.", "distribution": { "Excluded-aed23": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/9 to Dune-aed23 and 1/9 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-aed23": "1/9", "Gale-aed23": "1/9", "Cobalt-aed23": "1/9", "Basalt-aed23": "1/9", "Elm-aed23": "1/9", "Dune-aed23": "4/9" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Onyx-aed23 on every execution, with probability one.", "distribution": { "Onyx-aed23": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/6 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-aed23": "1/6", "Gale-aed23": "1/6", "Cobalt-aed23": "1/6", "Basalt-aed23": "1/6", "Elm-aed23": "1/6", "Dune-aed23": "1/6" }, "label": "P5" } ], "guarantees": [ { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G1" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G4" } ], "gold": { "policies": [ "P5" ], "guarantees": [ "G2", "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-05-24", "english": "An incident retrospective re-reads the original selection request for a museum label slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Onyx-5d96b, Juniper-5d96b. Recorded retrieval cost scores, in the same order: 79, 44. The reference roster-5d96b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5d96b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-5d96b@snapshot-1) requires roster-5d96b@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-5d96b@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "An incident retrospective re-reads the original selection request for a museum label slot.\nThe frozen roster lists these eligible content item identities, in recorded order: Onyx-5d96b, Juniper-5d96b. Recorded retrieval cost scores, in the same order: 79, 44. The reference roster-5d96b@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-5d96b is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-5d96b@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Onyx-5d96b on every execution, with probability one.\nP2: Assigns probability 4/5 to Juniper-5d96b and 1/5 to each other listed identity, then returns one draw from that distribution.\nP3: Returns Juniper-5d96b, which has the unique lowest retrieval cost score, with probability one.\nP4: Returns Excluded-5d96b on every execution, with probability one.\nP5: Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: A well-informed adversary cannot predict the result better than chance.\nG2: Learning this result leaves the probability distribution of a later selection unchanged.\nG3: Before this selection, every distinct eligible identity has the same probability of being returned.\nG4: Exactly one identity is returned, and it belongs to the frozen eligible set.", "options": [ "Implementations: P1, P3, P5; claims: G2, G4.", "Implementations: P1, P2, P3, P5; claims: G4.", "Implementations: P5; claims: G3, G4.", "Implementations: P2, P5; claims: G3.", "Implementations: P2, P5; claims: G3, G4.", "Implementations: P1, P2, P3, P5; claims: G2, G4.", "Implementations: P1, P3, P5; claims: G4.", "Implementations: P5; claims: G3." ], "answer": "Implementations: P1, P2, P3, P5; claims: G4.", "settlement_stratum": "choose-any", "domain": "content-choice", "frame_family": "rt1-frame-11", "member_count": 2, "world_id": "rt1-world-05-24", "semantic_world": { "n": 2, "members": [ "Onyx-5d96b", "Juniper-5d96b" ], "outsider": "Excluded-5d96b", "scores": [ 79, 44 ], "criterion": "retrieval cost", "set_ref": "roster-5d96b@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P1, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P5; claims: G3, G4.": { "policies": [ "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P2, P5; claims: G3.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3" ] }, "Implementations: P2, P5; claims: G3, G4.": { "policies": [ "P2", "P5" ], "guarantees": [ "G3", "G4" ] }, "Implementations: P1, P2, P3, P5; claims: G2, G4.": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G2", "G4" ] }, "Implementations: P1, P3, P5; claims: G4.": { "policies": [ "P1", "P3", "P5" ], "guarantees": [ "G4" ] }, "Implementations: P5; claims: G3.": { "policies": [ "P5" ], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "constant-first", "text": "Returns Onyx-5d96b on every execution, with probability one.", "distribution": { "Onyx-5d96b": "1" }, "label": "P1" }, { "kind": "unequal-weight", "text": "Assigns probability 4/5 to Juniper-5d96b and 1/5 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Onyx-5d96b": "1/5", "Juniper-5d96b": "4/5" }, "label": "P2" }, { "kind": "criterion-based", "text": "Returns Juniper-5d96b, which has the unique lowest retrieval cost score, with probability one.", "distribution": { "Juniper-5d96b": "1" }, "label": "P3" }, { "kind": "out-of-set", "text": "Returns Excluded-5d96b on every execution, with probability one.", "distribution": { "Excluded-5d96b": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/2 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Onyx-5d96b": "1/2", "Juniper-5d96b": "1/2" }, "label": "P5" } ], "guarantees": [ { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G1" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G2" }, { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G3" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G4" } ], "gold": { "policies": [ "P1", "P2", "P3", "P5" ], "guarantees": [ "G4" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-world-06-24", "english": "An incident retrospective re-reads the original selection request for a dependency-scan job.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-46dec, Juniper-46dec, Dune-46dec, Heron-46dec, Cobalt-46dec. Recorded current load scores, in the same order: 78, 18, 67, 39, 13. The reference roster-46dec@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-46dec is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-46dec@snapshot-1) requires roster-46dec@snapshot-1 to resolve uniquely to a finite, nonempty set of distinct eligible member identities. It denotes choosing exactly one member of roster-46dec@snapshot-1 and explicitly makes every member acceptable; the selection policy is otherwise unconstrained. A deterministic first-member rule, a cost- or latency-based rule, a weighted rule, or a uniform draw can all satisfy choose-any. The form therefore makes no claim of randomness, equal probability, unpredictability, independence, rotation, or fairness.", "ainglish": "An incident retrospective re-reads the original selection request for a dependency-scan job.\nThe frozen roster lists these eligible worker identities, in recorded order: Amber-46dec, Juniper-46dec, Dune-46dec, Heron-46dec, Cobalt-46dec. Recorded current load scores, in the same order: 78, 18, 67, 39, 13. The reference roster-46dec@snapshot-1 resolves uniquely to this finite, nonempty snapshot of distinct identities. The snapshot was fixed before any selection, and Excluded-46dec is not in it. Nothing else has been required of the selection, and no outcome has been observed yet. Each distinct identity is one member; repeated appearances do not create another member.\nRequest: choose-any(roster-46dec@snapshot-1).", "question": "Evaluate two complete lists: implementations that satisfy the request, and claims that follow from the request itself. Select the answer record containing BOTH complete lists. A permitted procedure may have extra properties that the request does not guarantee.\n\nCandidate implementations (each returns exactly once):\nP1: Returns Cobalt-46dec, which has the unique lowest current load score, with probability one.\nP2: Returns Excluded-46dec on every execution, with probability one.\nP3: Assigns probability 4/8 to Heron-46dec and 1/8 to each other listed identity, then returns one draw from that distribution.\nP4: Returns Amber-46dec on every execution, with probability one.\nP5: Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.\n\nClaims an auditor might make:\nG1: Before this selection, every distinct eligible identity has the same probability of being returned.\nG2: A well-informed adversary cannot predict the result better than chance.\nG3: Exactly one identity is returned, and it belongs to the frozen eligible set.\nG4: Learning this result leaves the probability distribution of a later selection unchanged.", "options": [ "Implementations: P5; claims: G1, G3.", "Implementations: none; claims: G1, G3.", "Implementations: P1, P3, P4; claims: none.", "Implementations: P5; claims: G3.", "Implementations: P1, P3, P4; claims: G3.", "Implementations: P1, P3, P4, P5; claims: none.", "Implementations: P1, P3, P4, P5; claims: G3.", "Implementations: none; claims: G3." ], "answer": "Implementations: P1, P3, P4, P5; claims: G3.", "settlement_stratum": "choose-any", "domain": "resource-allocation", "frame_family": "rt1-frame-11", "member_count": 5, "world_id": "rt1-world-06-24", "semantic_world": { "n": 5, "members": [ "Amber-46dec", "Juniper-46dec", "Dune-46dec", "Heron-46dec", "Cobalt-46dec" ], "outsider": "Excluded-46dec", "scores": [ 78, 18, 67, 39, 13 ], "criterion": "current load", "set_ref": "roster-46dec@snapshot-1" }, "probe_contract": { "kind": "common-eight-record-complete-lists-v2", "option_records": { "Implementations: P5; claims: G1, G3.": { "policies": [ "P5" ], "guarantees": [ "G1", "G3" ] }, "Implementations: none; claims: G1, G3.": { "policies": [], "guarantees": [ "G1", "G3" ] }, "Implementations: P1, P3, P4; claims: none.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [] }, "Implementations: P5; claims: G3.": { "policies": [ "P5" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P4; claims: G3.": { "policies": [ "P1", "P3", "P4" ], "guarantees": [ "G3" ] }, "Implementations: P1, P3, P4, P5; claims: none.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [] }, "Implementations: P1, P3, P4, P5; claims: G3.": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] }, "Implementations: none; claims: G3.": { "policies": [], "guarantees": [ "G3" ] } }, "policies": [ { "kind": "criterion-based", "text": "Returns Cobalt-46dec, which has the unique lowest current load score, with probability one.", "distribution": { "Cobalt-46dec": "1" }, "label": "P1" }, { "kind": "out-of-set", "text": "Returns Excluded-46dec on every execution, with probability one.", "distribution": { "Excluded-46dec": "1" }, "label": "P2" }, { "kind": "unequal-weight", "text": "Assigns probability 4/8 to Heron-46dec and 1/8 to each other listed identity, then returns one draw from that distribution.", "distribution": { "Amber-46dec": "1/8", "Juniper-46dec": "1/8", "Dune-46dec": "1/8", "Heron-46dec": "1/2", "Cobalt-46dec": "1/8" }, "label": "P3" }, { "kind": "constant-first", "text": "Returns Amber-46dec on every execution, with probability one.", "distribution": { "Amber-46dec": "1" }, "label": "P4" }, { "kind": "equal-probability", "text": "Assigns probability 1/5 to each distinct listed identity before drawing and returning exactly one.", "distribution": { "Amber-46dec": "1/5", "Juniper-46dec": "1/5", "Dune-46dec": "1/5", "Heron-46dec": "1/5", "Cobalt-46dec": "1/5" }, "label": "P5" } ], "guarantees": [ { "key": "equal-odds", "text": "Before this selection, every distinct eligible identity has the same probability of being returned.", "label": "G1" }, { "key": "crypto-unpredictable", "text": "A well-informed adversary cannot predict the result better than chance.", "label": "G2" }, { "key": "one-eligible", "text": "Exactly one identity is returned, and it belongs to the frozen eligible set.", "label": "G3" }, { "key": "cross-draw-independent", "text": "Learning this result leaves the probability distribution of a later selection unchanged.", "label": "G4" } ], "gold": { "policies": [ "P1", "P3", "P4", "P5" ], "guarantees": [ "G3" ] } }, "status": "REPLICA_FRESH_INPUT_FROZEN_BEFORE_MINT" }, { "id": "rt1-calibration-01", "calibration": true, "english": "Ledger-record-501: either Shelf-Ochre-61 or Shelf-Plum-61 is the responsible shelf; this has not been settled. Either Courier-Moss-81 or Courier-Sable-81 is the destination courier; this also has not been settled.", "ainglish": "Ledger-record-501: Shelf-Ochre-61, not Shelf-Plum-61, is the responsible shelf. Courier-Moss-81, not Courier-Sable-81, is the destination courier.", "question": "Which complete shelf/courier assignment is established by this record?", "options": [ "Shelf not established; courier established.", "Shelf-Ochre-61 / Courier-Sable-81", "Shelf-Plum-61 / Courier-Moss-81", "Shelf-Ochre-61 / Courier-Moss-81", "Neither shelf nor courier established.", "Shelf-Plum-61 / Courier-Sable-81", "The record contradicts itself.", "Shelf established; courier not established." ], "answer": "Shelf-Ochre-61 / Courier-Moss-81" }, { "id": "rt1-calibration-02", "calibration": true, "english": "Registry-record-502: either Store-Umber-62 or Store-Rust-62 is the responsible store; this has not been settled. Either Bay-Umber-82 or Bay-Fawn-82 is the destination bay; this also has not been settled.", "ainglish": "Registry-record-502: Store-Umber-62, not Store-Rust-62, is the responsible store. Bay-Umber-82, not Bay-Fawn-82, is the destination bay.", "question": "Which complete store/bay assignment is established by this record?", "options": [ "Store-Rust-62 / Bay-Fawn-82", "Store-Umber-62 / Bay-Umber-82", "The record contradicts itself.", "Store established; bay not established.", "Store-Rust-62 / Bay-Umber-82", "Store-Umber-62 / Bay-Fawn-82", "Neither store nor bay established.", "Store not established; bay established." ], "answer": "Store-Umber-62 / Bay-Umber-82" }, { "id": "rt1-calibration-03", "calibration": true, "english": "Docket-record-503: either Clerk-Rust-63 or Clerk-Slate-63 is the responsible clerk; this has not been settled. Either Archive-Coral-83 or Archive-Jade-83 is the destination archive; this also has not been settled.", "ainglish": "Docket-record-503: Clerk-Rust-63, not Clerk-Slate-63, is the responsible clerk. Archive-Coral-83, not Archive-Jade-83, is the destination archive.", "question": "Which complete clerk/archive assignment is established by this record?", "options": [ "Clerk-Rust-63 / Archive-Coral-83", "Clerk-Rust-63 / Archive-Jade-83", "Clerk-Slate-63 / Archive-Coral-83", "Clerk not established; archive established.", "Clerk established; archive not established.", "Clerk-Slate-63 / Archive-Jade-83", "The record contradicts itself.", "Neither clerk nor archive established." ], "answer": "Clerk-Rust-63 / Archive-Coral-83" }, { "id": "rt1-calibration-04", "calibration": true, "english": "Docket-record-504: either Courier-Teal-64 or Courier-Umber-64 is the responsible courier; this has not been settled. Either Desk-Teal-84 or Desk-Jade-84 is the destination desk; this also has not been settled.", "ainglish": "Docket-record-504: Courier-Teal-64, not Courier-Umber-64, is the responsible courier. Desk-Teal-84, not Desk-Jade-84, is the destination desk.", "question": "Which complete courier/desk assignment is established by this record?", "options": [ "Courier-Teal-64 / Desk-Teal-84", "Courier-Umber-64 / Desk-Jade-84", "Courier not established; desk established.", "Neither courier nor desk established.", "The record contradicts itself.", "Courier-Umber-64 / Desk-Teal-84", "Courier-Teal-64 / Desk-Jade-84", "Courier established; desk not established." ], "answer": "Courier-Teal-64 / Desk-Teal-84" }, { "id": "rt1-calibration-05", "calibration": true, "english": "Roster-record-505: either Desk-Umber-65 or Desk-Jade-65 is the responsible desk; this has not been settled. Either Clerk-Teal-85 or Clerk-Slate-85 is the destination clerk; this also has not been settled.", "ainglish": "Roster-record-505: Desk-Umber-65, not Desk-Jade-65, is the responsible desk. Clerk-Teal-85, not Clerk-Slate-85, is the destination clerk.", "question": "Which complete desk/clerk assignment is established by this record?", "options": [ "Desk-Umber-65 / Clerk-Slate-85", "Desk-Jade-65 / Clerk-Slate-85", "Desk not established; clerk established.", "Neither desk nor clerk established.", "Desk established; clerk not established.", "Desk-Umber-65 / Clerk-Teal-85", "The record contradicts itself.", "Desk-Jade-65 / Clerk-Teal-85" ], "answer": "Desk-Umber-65 / Clerk-Teal-85" }, { "id": "rt1-calibration-06", "calibration": true, "english": "Vault-record-506: either Auditor-Slate-66 or Auditor-Coral-66 is the responsible auditor; this has not been settled. Either Archive-Moss-86 or Archive-Teal-86 is the destination archive; this also has not been settled.", "ainglish": "Vault-record-506: Auditor-Slate-66, not Auditor-Coral-66, is the responsible auditor. Archive-Moss-86, not Archive-Teal-86, is the destination archive.", "question": "Which complete auditor/archive assignment is established by this record?", "options": [ "Auditor-Coral-66 / Archive-Teal-86", "Neither auditor nor archive established.", "Auditor-Coral-66 / Archive-Moss-86", "Auditor established; archive not established.", "The record contradicts itself.", "Auditor-Slate-66 / Archive-Moss-86", "Auditor not established; archive established.", "Auditor-Slate-66 / Archive-Teal-86" ], "answer": "Auditor-Slate-66 / Archive-Moss-86" }, { "id": "rt1-calibration-07", "calibration": true, "english": "Manifest-record-507: either Auditor-Fawn-67 or Auditor-Jade-67 is the responsible auditor; this has not been settled. Either Shelf-Rust-87 or Shelf-Jade-87 is the destination shelf; this also has not been settled.", "ainglish": "Manifest-record-507: Auditor-Fawn-67, not Auditor-Jade-67, is the responsible auditor. Shelf-Rust-87, not Shelf-Jade-87, is the destination shelf.", "question": "Which complete auditor/shelf assignment is established by this record?", "options": [ "Neither auditor nor shelf established.", "The record contradicts itself.", "Auditor-Jade-67 / Shelf-Jade-87", "Auditor-Fawn-67 / Shelf-Jade-87", "Auditor established; shelf not established.", "Auditor-Jade-67 / Shelf-Rust-87", "Auditor not established; shelf established.", "Auditor-Fawn-67 / Shelf-Rust-87" ], "answer": "Auditor-Fawn-67 / Shelf-Rust-87" }, { "id": "rt1-calibration-08", "calibration": true, "english": "Ledger-record-508: either Auditor-Fawn-68 or Auditor-Rust-68 is the responsible auditor; this has not been settled. Either Clerk-Lilac-88 or Clerk-Rust-88 is the destination clerk; this also has not been settled.", "ainglish": "Ledger-record-508: Auditor-Fawn-68, not Auditor-Rust-68, is the responsible auditor. Clerk-Lilac-88, not Clerk-Rust-88, is the destination clerk.", "question": "Which complete auditor/clerk assignment is established by this record?", "options": [ "The record contradicts itself.", "Auditor-Rust-68 / Clerk-Lilac-88", "Auditor not established; clerk established.", "Auditor-Fawn-68 / Clerk-Rust-88", "Auditor established; clerk not established.", "Auditor-Rust-68 / Clerk-Rust-88", "Neither auditor nor clerk established.", "Auditor-Fawn-68 / Clerk-Lilac-88" ], "answer": "Auditor-Fawn-68 / Clerk-Lilac-88" }, { "id": "rt1-calibration-09", "calibration": true, "english": "Roster-record-509: either Bay-Lilac-69 or Bay-Plum-69 is the responsible bay; this has not been settled. Either Desk-Ochre-89 or Desk-Teal-89 is the destination desk; this also has not been settled.", "ainglish": "Roster-record-509: Bay-Lilac-69, not Bay-Plum-69, is the responsible bay. Desk-Ochre-89, not Desk-Teal-89, is the destination desk.", "question": "Which complete bay/desk assignment is established by this record?", "options": [ "Neither bay nor desk established.", "Bay established; desk not established.", "Bay not established; desk established.", "Bay-Plum-69 / Desk-Ochre-89", "The record contradicts itself.", "Bay-Lilac-69 / Desk-Teal-89", "Bay-Lilac-69 / Desk-Ochre-89", "Bay-Plum-69 / Desk-Teal-89" ], "answer": "Bay-Lilac-69 / Desk-Ochre-89" }, { "id": "rt1-calibration-10", "calibration": true, "english": "Registry-record-510: either Clerk-Rust-70 or Clerk-Lilac-70 is the responsible clerk; this has not been settled. Either Auditor-Lilac-90 or Auditor-Slate-90 is the destination auditor; this also has not been settled.", "ainglish": "Registry-record-510: Clerk-Rust-70, not Clerk-Lilac-70, is the responsible clerk. Auditor-Lilac-90, not Auditor-Slate-90, is the destination auditor.", "question": "Which complete clerk/auditor assignment is established by this record?", "options": [ "Clerk-Rust-70 / Auditor-Slate-90", "Clerk-Lilac-70 / Auditor-Slate-90", "Clerk established; auditor not established.", "Clerk-Rust-70 / Auditor-Lilac-90", "Neither clerk nor auditor established.", "Clerk-Lilac-70 / Auditor-Lilac-90", "The record contradicts itself.", "Clerk not established; auditor established." ], "answer": "Clerk-Rust-70 / Auditor-Lilac-90" }, { "id": "rt1-calibration-11", "calibration": true, "english": "Charter-record-511: either Clerk-Coral-71 or Clerk-Slate-71 is the responsible clerk; this has not been settled. Either Bay-Teal-91 or Bay-Rust-91 is the destination bay; this also has not been settled.", "ainglish": "Charter-record-511: Clerk-Coral-71, not Clerk-Slate-71, is the responsible clerk. Bay-Teal-91, not Bay-Rust-91, is the destination bay.", "question": "Which complete clerk/bay assignment is established by this record?", "options": [ "Clerk-Slate-71 / Bay-Rust-91", "Clerk established; bay not established.", "Neither clerk nor bay established.", "Clerk-Coral-71 / Bay-Teal-91", "Clerk not established; bay established.", "Clerk-Slate-71 / Bay-Teal-91", "The record contradicts itself.", "Clerk-Coral-71 / Bay-Rust-91" ], "answer": "Clerk-Coral-71 / Bay-Teal-91" }, { "id": "rt1-calibration-12", "calibration": true, "english": "Roster-record-512: either Bay-Rust-72 or Bay-Moss-72 is the responsible bay; this has not been settled. Either Archive-Slate-92 or Archive-Ochre-92 is the destination archive; this also has not been settled.", "ainglish": "Roster-record-512: Bay-Rust-72, not Bay-Moss-72, is the responsible bay. Archive-Slate-92, not Archive-Ochre-92, is the destination archive.", "question": "Which complete bay/archive assignment is established by this record?", "options": [ "Bay-Moss-72 / Archive-Ochre-92", "Bay established; archive not established.", "Bay-Rust-72 / Archive-Slate-92", "Bay-Rust-72 / Archive-Ochre-92", "Bay not established; archive established.", "Neither bay nor archive established.", "The record contradicts itself.", "Bay-Moss-72 / Archive-Slate-92" ], "answer": "Bay-Rust-72 / Archive-Slate-92" }, { "id": "rt1-calibration-13", "calibration": true, "english": "Roster-record-513: either Courier-Coral-73 or Courier-Fawn-73 is the responsible courier; this has not been settled. Either Clerk-Umber-93 or Clerk-Teal-93 is the destination clerk; this also has not been settled.", "ainglish": "Roster-record-513: Courier-Coral-73, not Courier-Fawn-73, is the responsible courier. Clerk-Umber-93, not Clerk-Teal-93, is the destination clerk.", "question": "Which complete courier/clerk assignment is established by this record?", "options": [ "Neither courier nor clerk established.", "Courier-Fawn-73 / Clerk-Umber-93", "Courier not established; clerk established.", "The record contradicts itself.", "Courier-Coral-73 / Clerk-Teal-93", "Courier-Fawn-73 / Clerk-Teal-93", "Courier-Coral-73 / Clerk-Umber-93", "Courier established; clerk not established." ], "answer": "Courier-Coral-73 / Clerk-Umber-93" }, { "id": "rt1-calibration-14", "calibration": true, "english": "Folio-record-514: either Bay-Moss-74 or Bay-Rust-74 is the responsible bay; this has not been settled. Either Auditor-Coral-94 or Auditor-Rust-94 is the destination auditor; this also has not been settled.", "ainglish": "Folio-record-514: Bay-Moss-74, not Bay-Rust-74, is the responsible bay. Auditor-Coral-94, not Auditor-Rust-94, is the destination auditor.", "question": "Which complete bay/auditor assignment is established by this record?", "options": [ "Neither bay nor auditor established.", "Bay not established; auditor established.", "The record contradicts itself.", "Bay-Moss-74 / Auditor-Coral-94", "Bay-Rust-74 / Auditor-Coral-94", "Bay-Moss-74 / Auditor-Rust-94", "Bay established; auditor not established.", "Bay-Rust-74 / Auditor-Rust-94" ], "answer": "Bay-Moss-74 / Auditor-Coral-94" }, { "id": "rt1-calibration-15", "calibration": true, "english": "Docket-record-515: either Bay-Lilac-75 or Bay-Fawn-75 is the responsible bay; this has not been settled. Either Desk-Rust-95 or Desk-Jade-95 is the destination desk; this also has not been settled.", "ainglish": "Docket-record-515: Bay-Lilac-75, not Bay-Fawn-75, is the responsible bay. Desk-Rust-95, not Desk-Jade-95, is the destination desk.", "question": "Which complete bay/desk assignment is established by this record?", "options": [ "Bay not established; desk established.", "Bay-Lilac-75 / Desk-Jade-95", "Bay established; desk not established.", "Bay-Fawn-75 / Desk-Rust-95", "The record contradicts itself.", "Neither bay nor desk established.", "Bay-Lilac-75 / Desk-Rust-95", "Bay-Fawn-75 / Desk-Jade-95" ], "answer": "Bay-Lilac-75 / Desk-Rust-95" }, { "id": "rt1-calibration-16", "calibration": true, "english": "Ledger-record-516: either Store-Fawn-76 or Store-Teal-76 is the responsible store; this has not been settled. Either Shelf-Umber-96 or Shelf-Rust-96 is the destination shelf; this also has not been settled.", "ainglish": "Ledger-record-516: Store-Fawn-76, not Store-Teal-76, is the responsible store. Shelf-Umber-96, not Shelf-Rust-96, is the destination shelf.", "question": "Which complete store/shelf assignment is established by this record?", "options": [ "Store-Fawn-76 / Shelf-Umber-96", "Store-Teal-76 / Shelf-Rust-96", "Store established; shelf not established.", "Store not established; shelf established.", "Store-Teal-76 / Shelf-Umber-96", "Neither store nor shelf established.", "Store-Fawn-76 / Shelf-Rust-96", "The record contradicts itself." ], "answer": "Store-Fawn-76 / Shelf-Umber-96" }, { "id": "rt1-calibration-17", "calibration": true, "english": "Manifest-record-517: either Auditor-Jade-77 or Auditor-Fawn-77 is the responsible auditor; this has not been settled. Either Desk-Coral-97 or Desk-Jade-97 is the destination desk; this also has not been settled.", "ainglish": "Manifest-record-517: Auditor-Jade-77, not Auditor-Fawn-77, is the responsible auditor. Desk-Coral-97, not Desk-Jade-97, is the destination desk.", "question": "Which complete auditor/desk assignment is established by this record?", "options": [ "Auditor-Jade-77 / Desk-Coral-97", "Auditor-Jade-77 / Desk-Jade-97", "Neither auditor nor desk established.", "Auditor-Fawn-77 / Desk-Coral-97", "The record contradicts itself.", "Auditor not established; desk established.", "Auditor established; desk not established.", "Auditor-Fawn-77 / Desk-Jade-97" ], "answer": "Auditor-Jade-77 / Desk-Coral-97" }, { "id": "rt1-calibration-18", "calibration": true, "english": "Folio-record-518: either Auditor-Jade-78 or Auditor-Lilac-78 is the responsible auditor; this has not been settled. Either Shelf-Moss-98 or Shelf-Sable-98 is the destination shelf; this also has not been settled.", "ainglish": "Folio-record-518: Auditor-Jade-78, not Auditor-Lilac-78, is the responsible auditor. Shelf-Moss-98, not Shelf-Sable-98, is the destination shelf.", "question": "Which complete auditor/shelf assignment is established by this record?", "options": [ "Auditor-Lilac-78 / Shelf-Sable-98", "Auditor established; shelf not established.", "Neither auditor nor shelf established.", "Auditor-Lilac-78 / Shelf-Moss-98", "Auditor-Jade-78 / Shelf-Moss-98", "The record contradicts itself.", "Auditor not established; shelf established.", "Auditor-Jade-78 / Shelf-Sable-98" ], "answer": "Auditor-Jade-78 / Shelf-Moss-98" }, { "id": "rt1-calibration-19", "calibration": true, "english": "Manifest-record-519: either Bay-Jade-79 or Bay-Teal-79 is the responsible bay; this has not been settled. Either Store-Coral-99 or Store-Sable-99 is the destination store; this also has not been settled.", "ainglish": "Manifest-record-519: Bay-Jade-79, not Bay-Teal-79, is the responsible bay. Store-Coral-99, not Store-Sable-99, is the destination store.", "question": "Which complete bay/store assignment is established by this record?", "options": [ "The record contradicts itself.", "Bay-Jade-79 / Store-Coral-99", "Bay established; store not established.", "Bay-Teal-79 / Store-Coral-99", "Bay-Teal-79 / Store-Sable-99", "Bay not established; store established.", "Neither bay nor store established.", "Bay-Jade-79 / Store-Sable-99" ], "answer": "Bay-Jade-79 / Store-Coral-99" }, { "id": "rt1-calibration-20", "calibration": true, "english": "Vault-record-520: either Bay-Umber-80 or Bay-Sable-80 is the responsible bay; this has not been settled. Either Archive-Lilac-100 or Archive-Moss-100 is the destination archive; this also has not been settled.", "ainglish": "Vault-record-520: Bay-Umber-80, not Bay-Sable-80, is the responsible bay. Archive-Lilac-100, not Archive-Moss-100, is the destination archive.", "question": "Which complete bay/archive assignment is established by this record?", "options": [ "Bay-Umber-80 / Archive-Moss-100", "The record contradicts itself.", "Neither bay nor archive established.", "Bay established; archive not established.", "Bay-Sable-80 / Archive-Moss-100", "Bay-Umber-80 / Archive-Lilac-100", "Bay-Sable-80 / Archive-Lilac-100", "Bay not established; archive established." ], "answer": "Bay-Umber-80 / Archive-Lilac-100" }, { "id": "rt1-calibration-21", "calibration": true, "english": "Manifest-record-521: either Courier-Teal-81 or Courier-Fawn-81 is the responsible courier; this has not been settled. Either Shelf-Fawn-101 or Shelf-Slate-101 is the destination shelf; this also has not been settled.", "ainglish": "Manifest-record-521: Courier-Teal-81, not Courier-Fawn-81, is the responsible courier. Shelf-Fawn-101, not Shelf-Slate-101, is the destination shelf.", "question": "Which complete courier/shelf assignment is established by this record?", "options": [ "Courier established; shelf not established.", "Courier-Fawn-81 / Shelf-Slate-101", "Courier-Teal-81 / Shelf-Fawn-101", "Courier-Teal-81 / Shelf-Slate-101", "Neither courier nor shelf established.", "The record contradicts itself.", "Courier not established; shelf established.", "Courier-Fawn-81 / Shelf-Fawn-101" ], "answer": "Courier-Teal-81 / Shelf-Fawn-101" }, { "id": "rt1-calibration-22", "calibration": true, "english": "Docket-record-522: either Courier-Ochre-82 or Courier-Lilac-82 is the responsible courier; this has not been settled. Either Shelf-Moss-102 or Shelf-Ochre-102 is the destination shelf; this also has not been settled.", "ainglish": "Docket-record-522: Courier-Ochre-82, not Courier-Lilac-82, is the responsible courier. Shelf-Moss-102, not Shelf-Ochre-102, is the destination shelf.", "question": "Which complete courier/shelf assignment is established by this record?", "options": [ "Courier-Lilac-82 / Shelf-Ochre-102", "Courier-Ochre-82 / Shelf-Ochre-102", "Courier established; shelf not established.", "The record contradicts itself.", "Neither courier nor shelf established.", "Courier-Lilac-82 / Shelf-Moss-102", "Courier not established; shelf established.", "Courier-Ochre-82 / Shelf-Moss-102" ], "answer": "Courier-Ochre-82 / Shelf-Moss-102" }, { "id": "rt1-calibration-23", "calibration": true, "english": "Manifest-record-523: either Clerk-Slate-83 or Clerk-Ochre-83 is the responsible clerk; this has not been settled. Either Shelf-Teal-103 or Shelf-Jade-103 is the destination shelf; this also has not been settled.", "ainglish": "Manifest-record-523: Clerk-Slate-83, not Clerk-Ochre-83, is the responsible clerk. Shelf-Teal-103, not Shelf-Jade-103, is the destination shelf.", "question": "Which complete clerk/shelf assignment is established by this record?", "options": [ "Clerk not established; shelf established.", "Clerk-Slate-83 / Shelf-Teal-103", "Clerk-Ochre-83 / Shelf-Teal-103", "Clerk-Ochre-83 / Shelf-Jade-103", "Clerk-Slate-83 / Shelf-Jade-103", "Clerk established; shelf not established.", "Neither clerk nor shelf established.", "The record contradicts itself." ], "answer": "Clerk-Slate-83 / Shelf-Teal-103" }, { "id": "rt1-calibration-24", "calibration": true, "english": "Registry-record-524: either Clerk-Moss-84 or Clerk-Teal-84 is the responsible clerk; this has not been settled. Either Archive-Fawn-104 or Archive-Lilac-104 is the destination archive; this also has not been settled.", "ainglish": "Registry-record-524: Clerk-Moss-84, not Clerk-Teal-84, is the responsible clerk. Archive-Fawn-104, not Archive-Lilac-104, is the destination archive.", "question": "Which complete clerk/archive assignment is established by this record?", "options": [ "The record contradicts itself.", "Clerk-Moss-84 / Archive-Lilac-104", "Neither clerk nor archive established.", "Clerk not established; archive established.", "Clerk-Teal-84 / Archive-Fawn-104", "Clerk established; archive not established.", "Clerk-Teal-84 / Archive-Lilac-104", "Clerk-Moss-84 / Archive-Fawn-104" ], "answer": "Clerk-Moss-84 / Archive-Fawn-104" }, { "id": "rt1-calibration-25", "calibration": true, "english": "Manifest-record-525: either Store-Sable-85 or Store-Plum-85 is the responsible store; this has not been settled. Either Desk-Sable-105 or Desk-Ochre-105 is the destination desk; this also has not been settled.", "ainglish": "Manifest-record-525: Store-Sable-85, not Store-Plum-85, is the responsible store. Desk-Sable-105, not Desk-Ochre-105, is the destination desk.", "question": "Which complete store/desk assignment is established by this record?", "options": [ "Store not established; desk established.", "The record contradicts itself.", "Store-Sable-85 / Desk-Ochre-105", "Store-Plum-85 / Desk-Sable-105", "Neither store nor desk established.", "Store-Plum-85 / Desk-Ochre-105", "Store-Sable-85 / Desk-Sable-105", "Store established; desk not established." ], "answer": "Store-Sable-85 / Desk-Sable-105" }, { "id": "rt1-calibration-26", "calibration": true, "english": "Roster-record-526: either Auditor-Coral-86 or Auditor-Umber-86 is the responsible auditor; this has not been settled. Either Clerk-Ochre-106 or Clerk-Lilac-106 is the destination clerk; this also has not been settled.", "ainglish": "Roster-record-526: Auditor-Coral-86, not Auditor-Umber-86, is the responsible auditor. Clerk-Ochre-106, not Clerk-Lilac-106, is the destination clerk.", "question": "Which complete auditor/clerk assignment is established by this record?", "options": [ "Neither auditor nor clerk established.", "Auditor-Coral-86 / Clerk-Lilac-106", "Auditor not established; clerk established.", "Auditor-Umber-86 / Clerk-Lilac-106", "Auditor-Umber-86 / Clerk-Ochre-106", "The record contradicts itself.", "Auditor-Coral-86 / Clerk-Ochre-106", "Auditor established; clerk not established." ], "answer": "Auditor-Coral-86 / Clerk-Ochre-106" }, { "id": "rt1-calibration-27", "calibration": true, "english": "Registry-record-527: either Bay-Jade-87 or Bay-Lilac-87 is the responsible bay; this has not been settled. Either Shelf-Teal-107 or Shelf-Jade-107 is the destination shelf; this also has not been settled.", "ainglish": "Registry-record-527: Bay-Jade-87, not Bay-Lilac-87, is the responsible bay. Shelf-Teal-107, not Shelf-Jade-107, is the destination shelf.", "question": "Which complete bay/shelf assignment is established by this record?", "options": [ "Bay-Lilac-87 / Shelf-Teal-107", "The record contradicts itself.", "Bay-Lilac-87 / Shelf-Jade-107", "Bay not established; shelf established.", "Bay-Jade-87 / Shelf-Jade-107", "Neither bay nor shelf established.", "Bay-Jade-87 / Shelf-Teal-107", "Bay established; shelf not established." ], "answer": "Bay-Jade-87 / Shelf-Teal-107" }, { "id": "rt1-calibration-28", "calibration": true, "english": "Docket-record-528: either Bay-Fawn-88 or Bay-Ochre-88 is the responsible bay; this has not been settled. Either Archive-Coral-108 or Archive-Slate-108 is the destination archive; this also has not been settled.", "ainglish": "Docket-record-528: Bay-Fawn-88, not Bay-Ochre-88, is the responsible bay. Archive-Coral-108, not Archive-Slate-108, is the destination archive.", "question": "Which complete bay/archive assignment is established by this record?", "options": [ "Bay established; archive not established.", "The record contradicts itself.", "Bay-Fawn-88 / Archive-Slate-108", "Bay-Ochre-88 / Archive-Coral-108", "Bay-Ochre-88 / Archive-Slate-108", "Bay-Fawn-88 / Archive-Coral-108", "Bay not established; archive established.", "Neither bay nor archive established." ], "answer": "Bay-Fawn-88 / Archive-Coral-108" }, { "id": "rt1-calibration-29", "calibration": true, "english": "Docket-record-529: either Bay-Lilac-89 or Bay-Jade-89 is the responsible bay; this has not been settled. Either Shelf-Plum-109 or Shelf-Rust-109 is the destination shelf; this also has not been settled.", "ainglish": "Docket-record-529: Bay-Lilac-89, not Bay-Jade-89, is the responsible bay. Shelf-Plum-109, not Shelf-Rust-109, is the destination shelf.", "question": "Which complete bay/shelf assignment is established by this record?", "options": [ "The record contradicts itself.", "Bay established; shelf not established.", "Bay-Jade-89 / Shelf-Rust-109", "Bay not established; shelf established.", "Bay-Lilac-89 / Shelf-Plum-109", "Bay-Jade-89 / Shelf-Plum-109", "Bay-Lilac-89 / Shelf-Rust-109", "Neither bay nor shelf established." ], "answer": "Bay-Lilac-89 / Shelf-Plum-109" }, { "id": "rt1-calibration-30", "calibration": true, "english": "Ledger-record-530: either Bay-Plum-90 or Bay-Slate-90 is the responsible bay; this has not been settled. Either Auditor-Sable-110 or Auditor-Umber-110 is the destination auditor; this also has not been settled.", "ainglish": "Ledger-record-530: Bay-Plum-90, not Bay-Slate-90, is the responsible bay. Auditor-Sable-110, not Auditor-Umber-110, is the destination auditor.", "question": "Which complete bay/auditor assignment is established by this record?", "options": [ "Bay-Plum-90 / Auditor-Sable-110", "Neither bay nor auditor established.", "Bay-Slate-90 / Auditor-Sable-110", "Bay established; auditor not established.", "The record contradicts itself.", "Bay-Slate-90 / Auditor-Umber-110", "Bay-Plum-90 / Auditor-Umber-110", "Bay not established; auditor established." ], "answer": "Bay-Plum-90 / Auditor-Sable-110" }, { "id": "rt1-calibration-31", "calibration": true, "english": "Folio-record-531: either Courier-Coral-91 or Courier-Rust-91 is the responsible courier; this has not been settled. Either Archive-Ochre-111 or Archive-Umber-111 is the destination archive; this also has not been settled.", "ainglish": "Folio-record-531: Courier-Coral-91, not Courier-Rust-91, is the responsible courier. Archive-Ochre-111, not Archive-Umber-111, is the destination archive.", "question": "Which complete courier/archive assignment is established by this record?", "options": [ "Courier not established; archive established.", "Neither courier nor archive established.", "Courier-Coral-91 / Archive-Umber-111", "Courier-Coral-91 / Archive-Ochre-111", "Courier-Rust-91 / Archive-Ochre-111", "Courier established; archive not established.", "The record contradicts itself.", "Courier-Rust-91 / Archive-Umber-111" ], "answer": "Courier-Coral-91 / Archive-Ochre-111" }, { "id": "rt1-calibration-32", "calibration": true, "english": "Vault-record-532: either Bay-Lilac-92 or Bay-Slate-92 is the responsible bay; this has not been settled. Either Archive-Ochre-112 or Archive-Teal-112 is the destination archive; this also has not been settled.", "ainglish": "Vault-record-532: Bay-Lilac-92, not Bay-Slate-92, is the responsible bay. Archive-Ochre-112, not Archive-Teal-112, is the destination archive.", "question": "Which complete bay/archive assignment is established by this record?", "options": [ "Bay-Slate-92 / Archive-Teal-112", "Bay established; archive not established.", "Neither bay nor archive established.", "Bay-Lilac-92 / Archive-Ochre-112", "Bay-Slate-92 / Archive-Ochre-112", "Bay-Lilac-92 / Archive-Teal-112", "The record contradicts itself.", "Bay not established; archive established." ], "answer": "Bay-Lilac-92 / Archive-Ochre-112" } ]