# Appraisal rules What one contact does to the mind, in full. These are heuristics chosen by this project, not doctrine: the classical sources say which factors exist and what counteracts what, and say nothing about how hard a failing test should land. Every number below is arguable, and this page exists so it can be argued with without reading TypeScript. ## The five gates Every contact arrives through one gate, and the gate decides both the feeling and the factors. The same failure through `ear` (being corrected) and through `body` (a non-zero exit) is not the same event for the mind that receives it. | gate | what it is | favorable | adverse | |---|---|---|---| | 眼识 `eye` | what the agent looks at — file contents, search results, rendered output | 想 samjna 0.5 · 慧 prajna 0.35 · 作意 manaskara 0.3 | 疑 vicikitsa 0.45 · 寻 vitarka 0.4 · 想 samjna 0.2 | | 耳识 `ear` | what the agent is told — user messages, review comments, notifications | 信 sraddha 0.5 · 欲 chanda 0.35 · 念 smrti 0.3 | 惭 hri 0.5 · 念 smrti 0.35 · 无嗔 advesa 0.2 | | 鼻识 `nose` | what the agent senses without being told — code smell, stale config, drifting state | 轻安 prasrabdhi 0.4 · 定 samadhi 0.25 | 伺 vicara 0.45 · 疑 vicikitsa 0.3 · 不放逸 apramada 0.25 | | 舌识 `tongue` | what the agent tastes of its own product — test output, builds, its own diffs re-read | 轻安 prasrabdhi 0.45 · 信 sraddha 0.3 · 精进 virya 0.25 | 悔 kaukrtya 0.45 · 精进 virya 0.35 · 惭 hri 0.25 | | 身识 `body` | the world's direct resistance — non-zero exits, write failures, exceptions, timeouts | 精进 virya 0.45 · 轻安 prasrabdhi 0.3 · 定 samadhi 0.2 | 嗔 pratigha 0.45 · 忿 krodha 0.35 · 掉举 auddhatya 0.3 | Weights are scaled by the contact's intensity and combined saturatingly, so a mind can be fully irritated but never doubly irritated. ## Expectation, and how much of a contact is felt The store already holds a prediction: a seed's running valence is what this situation has felt like, and its count and surviving potency are how much that history deserves to be believed. A contact is compared against it, and the violation decides how much of the contact's magnitude actually lands. | what happened | felt at | and it stirs | |---|---|---| | never met this situation before | full magnitude | 寻 vitarka, 作意 manaskara — everything is news the first time | | contradicted a confident prediction | full magnitude | 寻 vitarka · 作意 manaskara, plus 疑 vicikitsa when adverse, 轻安 prasrabdhi when favorable | | a predicted failure failed | 0.45 of magnitude | 懈怠 kausidya · 行舍 upeksa — resignation, not anger | | a predicted success succeeded | 0.45 of magnitude | 行舍 upeksa · 定 samadhi — routine competence is quiet | A prediction is trusted in proportion to its evidence, reaching full confidence at 4 contacts and decaying with the seed. Surprise at or above 0.6 counts as news; at or below 0.25 the store called it. This is what separates the model from a meter. A meter reads the same number every time the same thing happens; the fourth failure of a test you already knew was broken should not land like the first. ## Feeling `body` and `tongue` are bodily seats, so they yield 苦 / 乐; `eye`, `ear`, and `nose` are read cognitively, so they yield 忧 / 喜. A neutral contact is 舍. Mental feeling is the stronger of the two (±0.8 against ±0.6) before intensity scaling — being told you are wrong lands harder than a command exiting 1. ## Universal factors Present in every contact regardless of gate or outcome: 触 sparsa 0.6, 作意 manaskara 0.4, 受 vedana 0.5, 思 cetana 0.3. ## Thresholds | constant | value | what it decides | |---|---|---| | `STREAK_CONCEIT` | 3 | a favorable run this long starts stirring 憍 mada and 放逸 pramada, and raises 我慢 | | `STREAK_CHURN` | 2 | an adverse run this long starts stirring 掉举 auddhatya and 恨 upanaha | | `STREAK_DULL` | 4 | an adverse run this long starts stirring 昏沉 styana, 懈怠 kausidya, and 疑 vicikitsa | | `SEED_ENTRENCHED` | 3 | a seed re-perfumed adversely this often reads as a position being defended: 恶见 drsti, and 我见 rises | | `DEFENSIVE_GRIP` | 0.5 | above this grip a correction is met with 覆 mraksa + 嗔 pratigha instead of 惭 hri + 愧 apatrapya | The last row is the one worth staring at. It is where the model stops being a mood tracker: the object is not met bare, it is met through the self that receives it, so the same correction becomes two different mental events depending on how tightly the agent is currently holding its own record. ## Epistemic status of a reading When a contact carries a `nature`, it moves self-delusion directly — this is the one place the agent grades itself. | nature | factors stirred | 我痴 | |---|---|---| | 遍计所执 `parikalpita` (assumed) | 痴 moha 0.3 | +0.10 | | 依他起 `paratantra` (inferred) | 伺 vicara 0.25 | unchanged | | 圆成实 `parinispanna` (verified) | 无痴 amoha 0.4 · 慧 prajna 0.3 | −0.12 | A favorable contact through `tongue` or `body` lowers 我痴 by a further 0.05, because running it and touching it are the gates that can be wrong about the world. Reading and being told cannot, however convincing they feel. ## Self-grasping | movement | when | |---|---| | 我慢 −0.12, 我见 −0.06 | a correction received while the grip is loose | | 我慢 +0.05, 我见 +0.05 | the same correction while the grip is tight | | 我慢 +0.05 per excess win | a favorable run at or past `STREAK_CONCEIT`, capped at +0.2 | | 我见 +0.10 | an adverse contact on a seed at or past `SEED_ENTRENCHED` | | 我爱 +0.08 | the object is the agent's own output (`self:` situation) | | 我爱 −0.03 | any other object | ## Decay | what | half-life | |---|---| | a mental factor, and the feeling's arousal | 5 minutes | | self-grasping | ten times a factor's — a standing disposition, not a mood | | a seed's potency | 14 days | Anything below 0.02 is dropped rather than carried as noise. ## Transformation Which wisdom answers an affliction follows from what kind of affliction it is: not-seeing turns through 妙观察智, distorting the record through 大圆镜智, slackness through 成所作智, and everything else — grabbing, recoiling, and all four self-graspings — through 平等性智. The antidote itself is never invented here: it is the classical pairing carried on each of the 51 factors.