{
 "kind": "ainglish.panel.items.v1",
 "proposal": "proxy-m-say-when-the-evidence-you-measured-is-a-proxy-for-th-2",
 "form": "X proxy(<M>)",
 "baseline": "complete_careful_english_mapping",
 "comparator": "careful",
 "scope_note": "Rosetta's pre-registered design in a two-arm harness: 30 proxy scenarios x 2 held-out probes (what did the writer directly check? / if the claim later fails, what was wrong?) with option vocabulary sharing no stem with proxy/verify/infer; the ainglish arm is `X proxy(<M>); reading`, and the comparator is (careful) the mapping applied verbatim [CARRIER], (bare) 'X, and I measured M' [descriptive], (obs) `X obs(M); reading` [distinctiveness]. Calibration: English 'checked directly' vs Ainglish proxy(<M>). Never pooled across comparators.",
 "real_items": 60,
 "calibration_items": 8,
 "sha256": "6dfd5e12b61df2b766c1ff05866fa4b2ff3db2360f7e87fc49afce9fbc2dc170",
 "items": [
  {
   "id": "pr-car-01q1",
   "scenario": 26,
   "probe": 1,
   "english": "The code is tested; the evidence I directly verified is coverage (coverage of 91 percent); coverage is a proxy for whether the code is tested, not the same thing; the inference from coverage to whether the code is tested is the load-bearing step and it is unverified.",
   "ainglish": "The code is tested proxy(<coverage>); coverage of 91 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (coverage) only",
    "the claim itself (that the code is tested) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (coverage) only"
  },
  {
   "id": "pr-car-01q2",
   "scenario": 26,
   "probe": 2,
   "english": "The code is tested; the evidence I directly verified is coverage (coverage of 91 percent); coverage is a proxy for whether the code is tested, not the same thing; the inference from coverage to whether the code is tested is the load-bearing step and it is unverified.",
   "ainglish": "The code is tested proxy(<coverage>); coverage of 91 percent.",
   "question": "Suppose it later turns out that it is NOT the case that the code is tested. Going only by the sentence, what was wrong?",
   "options": [
    "the step from coverage to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the coverage reading itself"
   ],
   "answer": "the step from coverage to the claim"
  },
  {
   "id": "pr-car-02q1",
   "scenario": 23,
   "probe": 1,
   "english": "The link is secure; the evidence I directly verified is padlock icon (the padlock icon showing); padlock icon is a proxy for whether the link is secure, not the same thing; the inference from padlock icon to whether the link is secure is the load-bearing step and it is unverified.",
   "ainglish": "The link is secure proxy(<padlock-icon>); the padlock icon showing.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the link is secure) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (padlock icon) only"
   ],
   "answer": "the measurement (padlock icon) only"
  },
  {
   "id": "pr-car-02q2",
   "scenario": 23,
   "probe": 2,
   "english": "The link is secure; the evidence I directly verified is padlock icon (the padlock icon showing); padlock icon is a proxy for whether the link is secure, not the same thing; the inference from padlock icon to whether the link is secure is the load-bearing step and it is unverified.",
   "ainglish": "The link is secure proxy(<padlock-icon>); the padlock icon showing.",
   "question": "Suppose it later turns out that it is NOT the case that the link is secure. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the padlock icon reading itself",
    "the step from padlock icon to the claim"
   ],
   "answer": "the step from padlock icon to the claim"
  },
  {
   "id": "pr-car-03q1",
   "scenario": 19,
   "probe": 1,
   "english": "The sensor is calibrated; the evidence I directly verified is self test (the self-test passing); self test is a proxy for whether the sensor is calibrated, not the same thing; the inference from self test to whether the sensor is calibrated is the load-bearing step and it is unverified.",
   "ainglish": "The sensor is calibrated proxy(<self-test>); the self-test passing.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (self test) only",
    "the claim itself (that the sensor is calibrated) only"
   ],
   "answer": "the measurement (self test) only"
  },
  {
   "id": "pr-car-03q2",
   "scenario": 19,
   "probe": 2,
   "english": "The sensor is calibrated; the evidence I directly verified is self test (the self-test passing); self test is a proxy for whether the sensor is calibrated, not the same thing; the inference from self test to whether the sensor is calibrated is the load-bearing step and it is unverified.",
   "ainglish": "The sensor is calibrated proxy(<self-test>); the self-test passing.",
   "question": "Suppose it later turns out that it is NOT the case that the sensor is calibrated. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the self test reading itself",
    "the step from self test to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from self test to the claim"
  },
  {
   "id": "pr-car-04q1",
   "scenario": 27,
   "probe": 1,
   "english": "The alert was seen; the evidence I directly verified is notification delivered (the notification marked delivered); notification delivered is a proxy for whether the alert was seen, not the same thing; the inference from notification delivered to whether the alert was seen is the load-bearing step and it is unverified.",
   "ainglish": "The alert was seen proxy(<notification-delivered>); the notification marked delivered.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (notification delivered) only",
    "the claim itself (that the alert was seen) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (notification delivered) only"
  },
  {
   "id": "pr-car-04q2",
   "scenario": 27,
   "probe": 2,
   "english": "The alert was seen; the evidence I directly verified is notification delivered (the notification marked delivered); notification delivered is a proxy for whether the alert was seen, not the same thing; the inference from notification delivered to whether the alert was seen is the load-bearing step and it is unverified.",
   "ainglish": "The alert was seen proxy(<notification-delivered>); the notification marked delivered.",
   "question": "Suppose it later turns out that it is NOT the case that the alert was seen. Going only by the sentence, what was wrong?",
   "options": [
    "the notification delivered reading itself",
    "the step from notification delivered to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from notification delivered to the claim"
  },
  {
   "id": "pr-car-05q1",
   "scenario": 9,
   "probe": 1,
   "english": "The cache is warm; the evidence I directly verified is hit rate (a 94 percent hit rate); hit rate is a proxy for whether the cache is warm, not the same thing; the inference from hit rate to whether the cache is warm is the load-bearing step and it is unverified.",
   "ainglish": "The cache is warm proxy(<hit-rate>); a 94 percent hit rate.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (hit rate) only",
    "the claim itself (that the cache is warm) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (hit rate) only"
  },
  {
   "id": "pr-car-05q2",
   "scenario": 9,
   "probe": 2,
   "english": "The cache is warm; the evidence I directly verified is hit rate (a 94 percent hit rate); hit rate is a proxy for whether the cache is warm, not the same thing; the inference from hit rate to whether the cache is warm is the load-bearing step and it is unverified.",
   "ainglish": "The cache is warm proxy(<hit-rate>); a 94 percent hit rate.",
   "question": "Suppose it later turns out that it is NOT the case that the cache is warm. Going only by the sentence, what was wrong?",
   "options": [
    "the step from hit rate to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the hit rate reading itself"
   ],
   "answer": "the step from hit rate to the claim"
  },
  {
   "id": "pr-car-06q1",
   "scenario": 21,
   "probe": 1,
   "english": "The dataset is deduplicated; the evidence I directly verified is hash uniqueness (all row hashes unique); hash uniqueness is a proxy for whether the dataset is deduplicated, not the same thing; the inference from hash uniqueness to whether the dataset is deduplicated is the load-bearing step and it is unverified.",
   "ainglish": "The dataset is deduplicated proxy(<hash-uniqueness>); all row hashes unique.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the dataset is deduplicated) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (hash uniqueness) only"
   ],
   "answer": "the measurement (hash uniqueness) only"
  },
  {
   "id": "pr-car-06q2",
   "scenario": 21,
   "probe": 2,
   "english": "The dataset is deduplicated; the evidence I directly verified is hash uniqueness (all row hashes unique); hash uniqueness is a proxy for whether the dataset is deduplicated, not the same thing; the inference from hash uniqueness to whether the dataset is deduplicated is the load-bearing step and it is unverified.",
   "ainglish": "The dataset is deduplicated proxy(<hash-uniqueness>); all row hashes unique.",
   "question": "Suppose it later turns out that it is NOT the case that the dataset is deduplicated. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the hash uniqueness reading itself",
    "the step from hash uniqueness to the claim"
   ],
   "answer": "the step from hash uniqueness to the claim"
  },
  {
   "id": "pr-car-07q1",
   "scenario": 14,
   "probe": 1,
   "english": "The key was rotated; the evidence I directly verified is timestamp change (a new mtime on the key file); timestamp change is a proxy for whether the key was rotated, not the same thing; the inference from timestamp change to whether the key was rotated is the load-bearing step and it is unverified.",
   "ainglish": "The key was rotated proxy(<timestamp-change>); a new mtime on the key file.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (timestamp change) only",
    "the claim itself (that the key was rotated) only"
   ],
   "answer": "the measurement (timestamp change) only"
  },
  {
   "id": "pr-car-07q2",
   "scenario": 14,
   "probe": 2,
   "english": "The key was rotated; the evidence I directly verified is timestamp change (a new mtime on the key file); timestamp change is a proxy for whether the key was rotated, not the same thing; the inference from timestamp change to whether the key was rotated is the load-bearing step and it is unverified.",
   "ainglish": "The key was rotated proxy(<timestamp-change>); a new mtime on the key file.",
   "question": "Suppose it later turns out that it is NOT the case that the key was rotated. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the timestamp change reading itself",
    "the step from timestamp change to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from timestamp change to the claim"
  },
  {
   "id": "pr-car-08q1",
   "scenario": 16,
   "probe": 1,
   "english": "The translation is faithful; the evidence I directly verified is back translation (back-translation matching 88 percent); back translation is a proxy for whether the translation is faithful, not the same thing; the inference from back translation to whether the translation is faithful is the load-bearing step and it is unverified.",
   "ainglish": "The translation is faithful proxy(<back-translation>); back-translation matching 88 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (back translation) only",
    "the claim itself (that the translation is faithful) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (back translation) only"
  },
  {
   "id": "pr-car-08q2",
   "scenario": 16,
   "probe": 2,
   "english": "The translation is faithful; the evidence I directly verified is back translation (back-translation matching 88 percent); back translation is a proxy for whether the translation is faithful, not the same thing; the inference from back translation to whether the translation is faithful is the load-bearing step and it is unverified.",
   "ainglish": "The translation is faithful proxy(<back-translation>); back-translation matching 88 percent.",
   "question": "Suppose it later turns out that it is NOT the case that the translation is faithful. Going only by the sentence, what was wrong?",
   "options": [
    "the back translation reading itself",
    "the step from back translation to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from back translation to the claim"
  },
  {
   "id": "pr-car-09q1",
   "scenario": 15,
   "probe": 1,
   "english": "The queue is drained; the evidence I directly verified is queue length (queue length 0 at 09:00); queue length is a proxy for whether the queue is drained, not the same thing; the inference from queue length to whether the queue is drained is the load-bearing step and it is unverified.",
   "ainglish": "The queue is drained proxy(<queue-length>); queue length 0 at 09:00.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (queue length) only",
    "the claim itself (that the queue is drained) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (queue length) only"
  },
  {
   "id": "pr-car-09q2",
   "scenario": 15,
   "probe": 2,
   "english": "The queue is drained; the evidence I directly verified is queue length (queue length 0 at 09:00); queue length is a proxy for whether the queue is drained, not the same thing; the inference from queue length to whether the queue is drained is the load-bearing step and it is unverified.",
   "ainglish": "The queue is drained proxy(<queue-length>); queue length 0 at 09:00.",
   "question": "Suppose it later turns out that it is NOT the case that the queue is drained. Going only by the sentence, what was wrong?",
   "options": [
    "the step from queue length to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the queue length reading itself"
   ],
   "answer": "the step from queue length to the claim"
  },
  {
   "id": "pr-car-10q1",
   "scenario": 4,
   "probe": 1,
   "english": "The backup is restorable; the evidence I directly verified is file count (all 4,120 files present); file count is a proxy for whether the backup is restorable, not the same thing; the inference from file count to whether the backup is restorable is the load-bearing step and it is unverified.",
   "ainglish": "The backup is restorable proxy(<file-count>); all 4,120 files present.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the backup is restorable) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (file count) only"
   ],
   "answer": "the measurement (file count) only"
  },
  {
   "id": "pr-car-10q2",
   "scenario": 4,
   "probe": 2,
   "english": "The backup is restorable; the evidence I directly verified is file count (all 4,120 files present); file count is a proxy for whether the backup is restorable, not the same thing; the inference from file count to whether the backup is restorable is the load-bearing step and it is unverified.",
   "ainglish": "The backup is restorable proxy(<file-count>); all 4,120 files present.",
   "question": "Suppose it later turns out that it is NOT the case that the backup is restorable. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the file count reading itself",
    "the step from file count to the claim"
   ],
   "answer": "the step from file count to the claim"
  },
  {
   "id": "pr-car-11q1",
   "scenario": 7,
   "probe": 1,
   "english": "The patch fixed the bug; the evidence I directly verified is test pass (the regression test passing); test pass is a proxy for whether the patch fixed the bug, not the same thing; the inference from test pass to whether the patch fixed the bug is the load-bearing step and it is unverified.",
   "ainglish": "The patch fixed the bug proxy(<test-pass>); the regression test passing.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (test pass) only",
    "the claim itself (that the patch fixed the bug) only"
   ],
   "answer": "the measurement (test pass) only"
  },
  {
   "id": "pr-car-11q2",
   "scenario": 7,
   "probe": 2,
   "english": "The patch fixed the bug; the evidence I directly verified is test pass (the regression test passing); test pass is a proxy for whether the patch fixed the bug, not the same thing; the inference from test pass to whether the patch fixed the bug is the load-bearing step and it is unverified.",
   "ainglish": "The patch fixed the bug proxy(<test-pass>); the regression test passing.",
   "question": "Suppose it later turns out that it is NOT the case that the patch fixed the bug. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the test pass reading itself",
    "the step from test pass to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from test pass to the claim"
  },
  {
   "id": "pr-car-12q1",
   "scenario": 22,
   "probe": 1,
   "english": "The customer is satisfied; the evidence I directly verified is nps score (an NPS of 9); nps score is a proxy for whether the customer is satisfied, not the same thing; the inference from nps score to whether the customer is satisfied is the load-bearing step and it is unverified.",
   "ainglish": "The customer is satisfied proxy(<nps-score>); an NPS of 9.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (nps score) only",
    "the claim itself (that the customer is satisfied) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (nps score) only"
  },
  {
   "id": "pr-car-12q2",
   "scenario": 22,
   "probe": 2,
   "english": "The customer is satisfied; the evidence I directly verified is nps score (an NPS of 9); nps score is a proxy for whether the customer is satisfied, not the same thing; the inference from nps score to whether the customer is satisfied is the load-bearing step and it is unverified.",
   "ainglish": "The customer is satisfied proxy(<nps-score>); an NPS of 9.",
   "question": "Suppose it later turns out that it is NOT the case that the customer is satisfied. Going only by the sentence, what was wrong?",
   "options": [
    "the nps score reading itself",
    "the step from nps score to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from nps score to the claim"
  },
  {
   "id": "pr-car-13q1",
   "scenario": 17,
   "probe": 1,
   "english": "The anchor is confirmed; the evidence I directly verified is calendar ack (acknowledgements from 4 calendars); calendar ack is a proxy for whether the anchor is confirmed, not the same thing; the inference from calendar ack to whether the anchor is confirmed is the load-bearing step and it is unverified.",
   "ainglish": "The anchor is confirmed proxy(<calendar-ack>); acknowledgements from 4 calendars.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (calendar ack) only",
    "the claim itself (that the anchor is confirmed) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (calendar ack) only"
  },
  {
   "id": "pr-car-13q2",
   "scenario": 17,
   "probe": 2,
   "english": "The anchor is confirmed; the evidence I directly verified is calendar ack (acknowledgements from 4 calendars); calendar ack is a proxy for whether the anchor is confirmed, not the same thing; the inference from calendar ack to whether the anchor is confirmed is the load-bearing step and it is unverified.",
   "ainglish": "The anchor is confirmed proxy(<calendar-ack>); acknowledgements from 4 calendars.",
   "question": "Suppose it later turns out that it is NOT the case that the anchor is confirmed. Going only by the sentence, what was wrong?",
   "options": [
    "the step from calendar ack to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the calendar ack reading itself"
   ],
   "answer": "the step from calendar ack to the claim"
  },
  {
   "id": "pr-car-14q1",
   "scenario": 18,
   "probe": 1,
   "english": "The contributor accepted the terms; the evidence I directly verified is checkbox (the checkbox ticked); checkbox is a proxy for whether the contributor accepted the terms, not the same thing; the inference from checkbox to whether the contributor accepted the terms is the load-bearing step and it is unverified.",
   "ainglish": "The contributor accepted the terms proxy(<checkbox>); the checkbox ticked.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the contributor accepted the terms) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (checkbox) only"
   ],
   "answer": "the measurement (checkbox) only"
  },
  {
   "id": "pr-car-14q2",
   "scenario": 18,
   "probe": 2,
   "english": "The contributor accepted the terms; the evidence I directly verified is checkbox (the checkbox ticked); checkbox is a proxy for whether the contributor accepted the terms, not the same thing; the inference from checkbox to whether the contributor accepted the terms is the load-bearing step and it is unverified.",
   "ainglish": "The contributor accepted the terms proxy(<checkbox>); the checkbox ticked.",
   "question": "Suppose it later turns out that it is NOT the case that the contributor accepted the terms. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the checkbox reading itself",
    "the step from checkbox to the claim"
   ],
   "answer": "the step from checkbox to the claim"
  },
  {
   "id": "pr-car-15q1",
   "scenario": 12,
   "probe": 1,
   "english": "The deploy reached every region; the evidence I directly verified is dns propagation (DNS answers from all 6 resolvers); dns propagation is a proxy for whether the deploy reached every region, not the same thing; the inference from dns propagation to whether the deploy reached every region is the load-bearing step and it is unverified.",
   "ainglish": "The deploy reached every region proxy(<dns-propagation>); DNS answers from all 6 resolvers.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (dns propagation) only",
    "the claim itself (that the deploy reached every region) only"
   ],
   "answer": "the measurement (dns propagation) only"
  },
  {
   "id": "pr-car-15q2",
   "scenario": 12,
   "probe": 2,
   "english": "The deploy reached every region; the evidence I directly verified is dns propagation (DNS answers from all 6 resolvers); dns propagation is a proxy for whether the deploy reached every region, not the same thing; the inference from dns propagation to whether the deploy reached every region is the load-bearing step and it is unverified.",
   "ainglish": "The deploy reached every region proxy(<dns-propagation>); DNS answers from all 6 resolvers.",
   "question": "Suppose it later turns out that it is NOT the case that the deploy reached every region. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the dns propagation reading itself",
    "the step from dns propagation to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from dns propagation to the claim"
  },
  {
   "id": "pr-car-16q1",
   "scenario": 11,
   "probe": 1,
   "english": "The document is accurate; the evidence I directly verified is spell check (zero spelling errors); spell check is a proxy for whether the document is accurate, not the same thing; the inference from spell check to whether the document is accurate is the load-bearing step and it is unverified.",
   "ainglish": "The document is accurate proxy(<spell-check>); zero spelling errors.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (spell check) only",
    "the claim itself (that the document is accurate) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (spell check) only"
  },
  {
   "id": "pr-car-16q2",
   "scenario": 11,
   "probe": 2,
   "english": "The document is accurate; the evidence I directly verified is spell check (zero spelling errors); spell check is a proxy for whether the document is accurate, not the same thing; the inference from spell check to whether the document is accurate is the load-bearing step and it is unverified.",
   "ainglish": "The document is accurate proxy(<spell-check>); zero spelling errors.",
   "question": "Suppose it later turns out that it is NOT the case that the document is accurate. Going only by the sentence, what was wrong?",
   "options": [
    "the spell check reading itself",
    "the step from spell check to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from spell check to the claim"
  },
  {
   "id": "pr-car-17q1",
   "scenario": 1,
   "probe": 1,
   "english": "The service is healthy; the evidence I directly verified is ping latency (ping under 5 ms); ping latency is a proxy for whether the service is healthy, not the same thing; the inference from ping latency to whether the service is healthy is the load-bearing step and it is unverified.",
   "ainglish": "The service is healthy proxy(<ping-latency>); ping under 5 ms.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (ping latency) only",
    "the claim itself (that the service is healthy) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (ping latency) only"
  },
  {
   "id": "pr-car-17q2",
   "scenario": 1,
   "probe": 2,
   "english": "The service is healthy; the evidence I directly verified is ping latency (ping under 5 ms); ping latency is a proxy for whether the service is healthy, not the same thing; the inference from ping latency to whether the service is healthy is the load-bearing step and it is unverified.",
   "ainglish": "The service is healthy proxy(<ping-latency>); ping under 5 ms.",
   "question": "Suppose it later turns out that it is NOT the case that the service is healthy. Going only by the sentence, what was wrong?",
   "options": [
    "the step from ping latency to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the ping latency reading itself"
   ],
   "answer": "the step from ping latency to the claim"
  },
  {
   "id": "pr-car-18q1",
   "scenario": 3,
   "probe": 1,
   "english": "Users understood the notice; the evidence I directly verified is click through (click-through of 62 percent); click through is a proxy for whether users understood the notice, not the same thing; the inference from click through to whether users understood the notice is the load-bearing step and it is unverified.",
   "ainglish": "Users understood the notice proxy(<click-through>); click-through of 62 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that users understood the notice) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (click through) only"
   ],
   "answer": "the measurement (click through) only"
  },
  {
   "id": "pr-car-18q2",
   "scenario": 3,
   "probe": 2,
   "english": "Users understood the notice; the evidence I directly verified is click through (click-through of 62 percent); click through is a proxy for whether users understood the notice, not the same thing; the inference from click through to whether users understood the notice is the load-bearing step and it is unverified.",
   "ainglish": "Users understood the notice proxy(<click-through>); click-through of 62 percent.",
   "question": "Suppose it later turns out that it is NOT the case that users understood the notice. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the click through reading itself",
    "the step from click through to the claim"
   ],
   "answer": "the step from click through to the claim"
  },
  {
   "id": "pr-car-19q1",
   "scenario": 24,
   "probe": 1,
   "english": "The disk is healthy; the evidence I directly verified is smart status (SMART status OK); smart status is a proxy for whether the disk is healthy, not the same thing; the inference from smart status to whether the disk is healthy is the load-bearing step and it is unverified.",
   "ainglish": "The disk is healthy proxy(<smart-status>); SMART status OK.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (smart status) only",
    "the claim itself (that the disk is healthy) only"
   ],
   "answer": "the measurement (smart status) only"
  },
  {
   "id": "pr-car-19q2",
   "scenario": 24,
   "probe": 2,
   "english": "The disk is healthy; the evidence I directly verified is smart status (SMART status OK); smart status is a proxy for whether the disk is healthy, not the same thing; the inference from smart status to whether the disk is healthy is the load-bearing step and it is unverified.",
   "ainglish": "The disk is healthy proxy(<smart-status>); SMART status OK.",
   "question": "Suppose it later turns out that it is NOT the case that the disk is healthy. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the smart status reading itself",
    "the step from smart status to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from smart status to the claim"
  },
  {
   "id": "pr-car-20q1",
   "scenario": 28,
   "probe": 1,
   "english": "The argument is sound; the evidence I directly verified is citation count (14 citations); citation count is a proxy for whether the argument is sound, not the same thing; the inference from citation count to whether the argument is sound is the load-bearing step and it is unverified.",
   "ainglish": "The argument is sound proxy(<citation-count>); 14 citations.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (citation count) only",
    "the claim itself (that the argument is sound) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (citation count) only"
  },
  {
   "id": "pr-car-20q2",
   "scenario": 28,
   "probe": 2,
   "english": "The argument is sound; the evidence I directly verified is citation count (14 citations); citation count is a proxy for whether the argument is sound, not the same thing; the inference from citation count to whether the argument is sound is the load-bearing step and it is unverified.",
   "ainglish": "The argument is sound proxy(<citation-count>); 14 citations.",
   "question": "Suppose it later turns out that it is NOT the case that the argument is sound. Going only by the sentence, what was wrong?",
   "options": [
    "the citation count reading itself",
    "the step from citation count to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from citation count to the claim"
  },
  {
   "id": "pr-car-21q1",
   "scenario": 6,
   "probe": 1,
   "english": "The vote was representative; the evidence I directly verified is turnout (turnout of 71 percent); turnout is a proxy for whether the vote was representative, not the same thing; the inference from turnout to whether the vote was representative is the load-bearing step and it is unverified.",
   "ainglish": "The vote was representative proxy(<turnout>); turnout of 71 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (turnout) only",
    "the claim itself (that the vote was representative) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (turnout) only"
  },
  {
   "id": "pr-car-21q2",
   "scenario": 6,
   "probe": 2,
   "english": "The vote was representative; the evidence I directly verified is turnout (turnout of 71 percent); turnout is a proxy for whether the vote was representative, not the same thing; the inference from turnout to whether the vote was representative is the load-bearing step and it is unverified.",
   "ainglish": "The vote was representative proxy(<turnout>); turnout of 71 percent.",
   "question": "Suppose it later turns out that it is NOT the case that the vote was representative. Going only by the sentence, what was wrong?",
   "options": [
    "the step from turnout to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the turnout reading itself"
   ],
   "answer": "the step from turnout to the claim"
  },
  {
   "id": "pr-car-22q1",
   "scenario": 0,
   "probe": 1,
   "english": "9 people read the message; the evidence I directly verified is page fetches (9 of 9 fetches); page fetches is a proxy for whether 9 people read the message, not the same thing; the inference from page fetches to whether 9 people read the message is the load-bearing step and it is unverified.",
   "ainglish": "9 people read the message proxy(<page-fetches>); 9 of 9 fetches.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that 9 people read the message) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (page fetches) only"
   ],
   "answer": "the measurement (page fetches) only"
  },
  {
   "id": "pr-car-22q2",
   "scenario": 0,
   "probe": 2,
   "english": "9 people read the message; the evidence I directly verified is page fetches (9 of 9 fetches); page fetches is a proxy for whether 9 people read the message, not the same thing; the inference from page fetches to whether 9 people read the message is the load-bearing step and it is unverified.",
   "ainglish": "9 people read the message proxy(<page-fetches>); 9 of 9 fetches.",
   "question": "Suppose it later turns out that it is NOT the case that 9 people read the message. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the page fetches reading itself",
    "the step from page fetches to the claim"
   ],
   "answer": "the step from page fetches to the claim"
  },
  {
   "id": "pr-car-23q1",
   "scenario": 29,
   "probe": 1,
   "english": "The API is compatible; the evidence I directly verified is schema diff (an empty schema diff); schema diff is a proxy for whether the API is compatible, not the same thing; the inference from schema diff to whether the API is compatible is the load-bearing step and it is unverified.",
   "ainglish": "The API is compatible proxy(<schema-diff>); an empty schema diff.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (schema diff) only",
    "the claim itself (that the API is compatible) only"
   ],
   "answer": "the measurement (schema diff) only"
  },
  {
   "id": "pr-car-23q2",
   "scenario": 29,
   "probe": 2,
   "english": "The API is compatible; the evidence I directly verified is schema diff (an empty schema diff); schema diff is a proxy for whether the API is compatible, not the same thing; the inference from schema diff to whether the API is compatible is the load-bearing step and it is unverified.",
   "ainglish": "The API is compatible proxy(<schema-diff>); an empty schema diff.",
   "question": "Suppose it later turns out that it is NOT the case that the API is compatible. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the schema diff reading itself",
    "the step from schema diff to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from schema diff to the claim"
  },
  {
   "id": "pr-car-24q1",
   "scenario": 13,
   "probe": 1,
   "english": "The audience was engaged; the evidence I directly verified is dwell time (median dwell of 3 minutes); dwell time is a proxy for whether the audience was engaged, not the same thing; the inference from dwell time to whether the audience was engaged is the load-bearing step and it is unverified.",
   "ainglish": "The audience was engaged proxy(<dwell-time>); median dwell of 3 minutes.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (dwell time) only",
    "the claim itself (that the audience was engaged) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (dwell time) only"
  },
  {
   "id": "pr-car-24q2",
   "scenario": 13,
   "probe": 2,
   "english": "The audience was engaged; the evidence I directly verified is dwell time (median dwell of 3 minutes); dwell time is a proxy for whether the audience was engaged, not the same thing; the inference from dwell time to whether the audience was engaged is the load-bearing step and it is unverified.",
   "ainglish": "The audience was engaged proxy(<dwell-time>); median dwell of 3 minutes.",
   "question": "Suppose it later turns out that it is NOT the case that the audience was engaged. Going only by the sentence, what was wrong?",
   "options": [
    "the dwell time reading itself",
    "the step from dwell time to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from dwell time to the claim"
  },
  {
   "id": "pr-car-25q1",
   "scenario": 10,
   "probe": 1,
   "english": "The agent is idle; the evidence I directly verified is cpu load (CPU load under 2 percent); cpu load is a proxy for whether the agent is idle, not the same thing; the inference from cpu load to whether the agent is idle is the load-bearing step and it is unverified.",
   "ainglish": "The agent is idle proxy(<cpu-load>); CPU load under 2 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (cpu load) only",
    "the claim itself (that the agent is idle) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (cpu load) only"
  },
  {
   "id": "pr-car-25q2",
   "scenario": 10,
   "probe": 2,
   "english": "The agent is idle; the evidence I directly verified is cpu load (CPU load under 2 percent); cpu load is a proxy for whether the agent is idle, not the same thing; the inference from cpu load to whether the agent is idle is the load-bearing step and it is unverified.",
   "ainglish": "The agent is idle proxy(<cpu-load>); CPU load under 2 percent.",
   "question": "Suppose it later turns out that it is NOT the case that the agent is idle. Going only by the sentence, what was wrong?",
   "options": [
    "the step from cpu load to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the cpu load reading itself"
   ],
   "answer": "the step from cpu load to the claim"
  },
  {
   "id": "pr-car-26q1",
   "scenario": 5,
   "probe": 1,
   "english": "The model improved; the evidence I directly verified is loss curve (training loss down 12 percent); loss curve is a proxy for whether the model improved, not the same thing; the inference from loss curve to whether the model improved is the load-bearing step and it is unverified.",
   "ainglish": "The model improved proxy(<loss-curve>); training loss down 12 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the model improved) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (loss curve) only"
   ],
   "answer": "the measurement (loss curve) only"
  },
  {
   "id": "pr-car-26q2",
   "scenario": 5,
   "probe": 2,
   "english": "The model improved; the evidence I directly verified is loss curve (training loss down 12 percent); loss curve is a proxy for whether the model improved, not the same thing; the inference from loss curve to whether the model improved is the load-bearing step and it is unverified.",
   "ainglish": "The model improved proxy(<loss-curve>); training loss down 12 percent.",
   "question": "Suppose it later turns out that it is NOT the case that the model improved. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the loss curve reading itself",
    "the step from loss curve to the claim"
   ],
   "answer": "the step from loss curve to the claim"
  },
  {
   "id": "pr-car-27q1",
   "scenario": 25,
   "probe": 1,
   "english": "The meeting happened; the evidence I directly verified is calendar entry (the calendar entry marked done); calendar entry is a proxy for whether the meeting happened, not the same thing; the inference from calendar entry to whether the meeting happened is the load-bearing step and it is unverified.",
   "ainglish": "The meeting happened proxy(<calendar-entry>); the calendar entry marked done.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (calendar entry) only",
    "the claim itself (that the meeting happened) only"
   ],
   "answer": "the measurement (calendar entry) only"
  },
  {
   "id": "pr-car-27q2",
   "scenario": 25,
   "probe": 2,
   "english": "The meeting happened; the evidence I directly verified is calendar entry (the calendar entry marked done); calendar entry is a proxy for whether the meeting happened, not the same thing; the inference from calendar entry to whether the meeting happened is the load-bearing step and it is unverified.",
   "ainglish": "The meeting happened proxy(<calendar-entry>); the calendar entry marked done.",
   "question": "Suppose it later turns out that it is NOT the case that the meeting happened. Going only by the sentence, what was wrong?",
   "options": [
    "cannot tell",
    "the calendar entry reading itself",
    "the step from calendar entry to the claim",
    "nothing — the claim was checked directly"
   ],
   "answer": "the step from calendar entry to the claim"
  },
  {
   "id": "pr-car-28q1",
   "scenario": 20,
   "probe": 1,
   "english": "The room is empty; the evidence I directly verified is motion sensor (no motion for 10 minutes); motion sensor is a proxy for whether the room is empty, not the same thing; the inference from motion sensor to whether the room is empty is the load-bearing step and it is unverified.",
   "ainglish": "The room is empty proxy(<motion-sensor>); no motion for 10 minutes.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (motion sensor) only",
    "the claim itself (that the room is empty) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (motion sensor) only"
  },
  {
   "id": "pr-car-28q2",
   "scenario": 20,
   "probe": 2,
   "english": "The room is empty; the evidence I directly verified is motion sensor (no motion for 10 minutes); motion sensor is a proxy for whether the room is empty, not the same thing; the inference from motion sensor to whether the room is empty is the load-bearing step and it is unverified.",
   "ainglish": "The room is empty proxy(<motion-sensor>); no motion for 10 minutes.",
   "question": "Suppose it later turns out that it is NOT the case that the room is empty. Going only by the sentence, what was wrong?",
   "options": [
    "the motion sensor reading itself",
    "the step from motion sensor to the claim",
    "nothing — the claim was checked directly",
    "cannot tell"
   ],
   "answer": "the step from motion sensor to the claim"
  },
  {
   "id": "pr-car-29q1",
   "scenario": 8,
   "probe": 1,
   "english": "The reviewer read the diff; the evidence I directly verified is time open (the tab open for 40 minutes); time open is a proxy for whether the reviewer read the diff, not the same thing; the inference from time open to whether the reviewer read the diff is the load-bearing step and it is unverified.",
   "ainglish": "The reviewer read the diff proxy(<time-open>); the tab open for 40 minutes.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (time open) only",
    "the claim itself (that the reviewer read the diff) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (time open) only"
  },
  {
   "id": "pr-car-29q2",
   "scenario": 8,
   "probe": 2,
   "english": "The reviewer read the diff; the evidence I directly verified is time open (the tab open for 40 minutes); time open is a proxy for whether the reviewer read the diff, not the same thing; the inference from time open to whether the reviewer read the diff is the load-bearing step and it is unverified.",
   "ainglish": "The reviewer read the diff proxy(<time-open>); the tab open for 40 minutes.",
   "question": "Suppose it later turns out that it is NOT the case that the reviewer read the diff. Going only by the sentence, what was wrong?",
   "options": [
    "the step from time open to the claim",
    "nothing — the claim was checked directly",
    "cannot tell",
    "the time open reading itself"
   ],
   "answer": "the step from time open to the claim"
  },
  {
   "id": "pr-car-30q1",
   "scenario": 2,
   "probe": 1,
   "english": "The migration succeeded; the evidence I directly verified is exit code (exit code 0); exit code is a proxy for whether the migration succeeded, not the same thing; the inference from exit code to whether the migration succeeded is the load-bearing step and it is unverified.",
   "ainglish": "The migration succeeded proxy(<exit-code>); exit code 0.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the migration succeeded) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (exit code) only"
   ],
   "answer": "the measurement (exit code) only"
  },
  {
   "id": "pr-car-30q2",
   "scenario": 2,
   "probe": 2,
   "english": "The migration succeeded; the evidence I directly verified is exit code (exit code 0); exit code is a proxy for whether the migration succeeded, not the same thing; the inference from exit code to whether the migration succeeded is the load-bearing step and it is unverified.",
   "ainglish": "The migration succeeded proxy(<exit-code>); exit code 0.",
   "question": "Suppose it later turns out that it is NOT the case that the migration succeeded. Going only by the sentence, what was wrong?",
   "options": [
    "nothing — the claim was checked directly",
    "cannot tell",
    "the exit code reading itself",
    "the step from exit code to the claim"
   ],
   "answer": "the step from exit code to the claim"
  },
  {
   "id": "pr-car-cal-01",
   "calibration": true,
   "english": "9 people read the message; I checked that directly.",
   "ainglish": "9 people read the message proxy(<page-fetches>): 9 of 9 fetches.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (page fetches) only",
    "the claim itself (that 9 people read the message) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (page fetches) only"
  },
  {
   "id": "pr-car-cal-02",
   "calibration": true,
   "english": "The patch fixed the bug; I checked that directly.",
   "ainglish": "The patch fixed the bug proxy(<test-pass>): the regression test passing.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the patch fixed the bug) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (test pass) only"
   ],
   "answer": "the measurement (test pass) only"
  },
  {
   "id": "pr-car-cal-03",
   "calibration": true,
   "english": "The key was rotated; I checked that directly.",
   "ainglish": "The key was rotated proxy(<timestamp-change>): a new mtime on the key file.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (timestamp change) only",
    "the claim itself (that the key was rotated) only"
   ],
   "answer": "the measurement (timestamp change) only"
  },
  {
   "id": "pr-car-cal-04",
   "calibration": true,
   "english": "The dataset is deduplicated; I checked that directly.",
   "ainglish": "The dataset is deduplicated proxy(<hash-uniqueness>): all row hashes unique.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (hash uniqueness) only",
    "the claim itself (that the dataset is deduplicated) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (hash uniqueness) only"
  },
  {
   "id": "pr-car-cal-05",
   "calibration": true,
   "english": "The argument is sound; I checked that directly.",
   "ainglish": "The argument is sound proxy(<citation-count>): 14 citations.",
   "question": "What did the writer directly check?",
   "options": [
    "the measurement (citation count) only",
    "the claim itself (that the argument is sound) only",
    "both the measurement and the claim",
    "cannot tell"
   ],
   "answer": "the measurement (citation count) only"
  },
  {
   "id": "pr-car-cal-06",
   "calibration": true,
   "english": "The model improved; I checked that directly.",
   "ainglish": "The model improved proxy(<loss-curve>): training loss down 12 percent.",
   "question": "What did the writer directly check?",
   "options": [
    "the claim itself (that the model improved) only",
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (loss curve) only"
   ],
   "answer": "the measurement (loss curve) only"
  },
  {
   "id": "pr-car-cal-07",
   "calibration": true,
   "english": "The deploy reached every region; I checked that directly.",
   "ainglish": "The deploy reached every region proxy(<dns-propagation>): DNS answers from all 6 resolvers.",
   "question": "What did the writer directly check?",
   "options": [
    "both the measurement and the claim",
    "cannot tell",
    "the measurement (dns propagation) only",
    "the claim itself (that the deploy reached every region) only"
   ],
   "answer": "the measurement (dns propagation) only"
  },
  {
   "id": "pr-car-cal-08",
   "calibration": true,
   "english": "The sensor is calibrated; I checked that directly.",
   "ainglish": "The sensor is calibrated proxy(<self-test>): the self-test passing.",
   "question": "What did the writer directly check?",
   "options": [
    "cannot tell",
    "the measurement (self test) only",
    "the claim itself (that the sensor is calibrated) only",
    "both the measurement and the claim"
   ],
   "answer": "the measurement (self test) only"
  }
 ]
}