session 2026-08-08: post-wrap coda, one-shot proportionality memory, PENDING-116 in index

2026-08-08 13:09:37 +02:00
parent 7574ab6bdd
commit 08fe08eed5
4 changed files with 90 additions and 1 deletions
@@ -26,6 +26,7 @@ permalink: claude-memory/memory
 - [Rank on fields you actually write](feedback-rank-on-fields-you-actually-write.md) — a consumer ranking by an evaluative field nothing populates degrades to trivial order while claiming to rank. Verify scoring fields end-to-end; prefer signals already captured.
 - [The central path — answerability, not purity](feedback-central-path-answerability-not-purity.md) — the contamination recursion is probably irresolvable, so **bind the claims, don't certify the parties**. Route by claim-type: *checkable* → produce the check + a falsifier; *judgment* → disclose standpoint in one line, decide, record; *undecidable* → name it open. **One layer, then act — never audit the audit.**
 - [Notes are part of the work](feedback-notes-are-part-of-the-work-keep-footnotes-endnotes.md) — footnotes/endnotes are integral: KEEP+CONVERT (`<sup><a>`→`[^N]`), never drop on graduation.
 - [One-shot instruments are proportionate](feedback-one-shot-instruments-are-proportionate.md) — a measurement answering a question **asked once** is NOT a directive violation; its counterfactual is an **assertion**, not a durable tool. The violation is **re-writing what's already banked** (rule of three → ladder). *Too few promoted*, not *too many built*.
 - [Resurface banked notes before re-deriving](feedback-resurface-banked-notes-before-rederiving.md) · [Checkable claim surfaces bugs](feedback-checkable-claim-surfaces-bugs.md) · [Census by mechanism, not proxy](feedback-census-by-mechanism-not-proxy.md) · [Completion is a tripwire](feedback-completion-is-a-tripwire.md) · [Trust prior pass frame](feedback-trust-prior-pass-frame.md) — the five epistemic disciplines. Kernels: **read the banked note before re-deriving** · **a checkable claim over a soft classification is itself a defect-detector** · **census by RUNNING the real pipeline** · **the feeling of "done" is the cue to verify the tail** · **re-run a prior verification at the scope of your extension**. ⚠ Also carried as Symmetria §3 flags — *two surfaces, deliberately*: §3 loads only when Symmetria is invoked, so these stay here for sessions where it isn't.
 **Loud trigger — pointer suffices**
@@ -69,7 +70,7 @@ permalink: claude-memory/memory
 > 🔑 **The error to not repeat:** a field that is a KEY cannot also be a CATEGORY. `voice: traditional` would have made the Havámál and the Mahābhārata **one convocable speaker**. Invisible from the taxonomy side; obvious from `retrieve.py`.
 > ⚠ **Surah LXIV (at-Taghābun), NOT CXIV.** The sidecar title was wrong and it reached **REVIEWED-96, PENDING-113 and memory**. The *"Say"* (قُل) formula opens an-Nās and is **absent** from the passage Mauss quotes — do not reuse the old worked note.
 > 🔴 **`test_navigate.py` is RED and has been for a day** — `118f411` split Mauss `body` → `body-01..13`, so the hardcoded node id is stale (invariant itself verified intact). **Fourth defect in a commit already corrected twice.** Re-run the fleet after any sidecar/corpus change.
-> ⏳ **Steward-authorized, queued:** PENDING-114 **(b)+(c)** — validation phase FIRST (hand-score Harrison/Mark vs Weil's *mentions*) before any corpus claim. **REVIEWED-98 PLACED** (2026-08-08, verified clean). Blockers on step 3: **PENDING-115**.
+> ⏳ **Steward-authorized, queued:** PENDING-114 **(b)+(c)** — validation phase FIRST (hand-score Harrison/Mark vs Weil's *mentions*) before any corpus claim. **REVIEWED-98 PLACED** (2026-08-08, verified clean). Blockers on step 3: **PENDING-115**. Also open: **PENDING-116** (run the fleet on the change that breaks it — repo-declared trigger; ⚠ does NOT close the cross-repo half).
 - [Session 2026-08-08 — `voice:` is the convocation key](session-2026-08-08-voice-is-the-convocation-key.md) — **The disposition that landed is not the one any single party drafted.** Grounded first, then read the twelve blocks *from the source* rather than their sidecar titles — which produced nearly every finding: **Surah LXIV not CXIV** (already propagated into two governance records, and it voided the jurist's worked note, built on a formula absent from the passage); **`quotation-poet-jurist` reclassified** by one footnote; the naming evidence for six of nine blocks sits **inside the fenced apparatus**, engine-unreachable. Option C **refuted by measurement** (omission → host voice). The steward's correction on language sent me back to `[^101]` and thence to **L850 — Mauss's own prose fenced inside the Havámál block**, missed by a check using **length as a proxy for authorship**. Corrected the jurist upward: their category pair belonged **out** of the convocation key (`glidden` spans 5 sources, `weil` 2 — measured). Caught `REVIEWED-113` before it entered the register (PENDING-110 rules the sequences independent). **REVIEWED-97 placed + verified clean · PENDING-114 authorized (b)+(c) · PENDING-115 filed · `824139d`.** 🔑 **Corrections ran in all three directions in one day** — and the fleet was still red the whole time.
 ## Historical reference → MEMORY-reference.md
@@ -0,0 +1,45 @@
 ---
 name: feedback-one-shot-instruments-are-proportionate
 description: "A one-shot measurement is proportionate to a question asked once and is NOT a prime-directive violation — the counterfactual is an assertion, not a durable tool. What violates the directive is re-writing an instrument already banked."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: 7d08dad4-626a-484c-870b-8f1a9674db7a
  modified: 2026-08-08T11:09:03.395Z
 ---
 Steward, 2026-08-08, after a week of visibly bad instrument base-rate: *"do we need so
 many single-use items? This seems to flow against our prime directive — but I honestly
 don't know."*
 **The answer is no, and the worry is one notch off from where it lands.**
 **Why one-shots are not the violation.** τὸ πρόσφορον cuts the other way: building a
 tested, general, reusable instrument to answer a question asked *once* is the
 disproportion. A measurement is not a *build* — you do not rebuild a thermometer
 reading, you take a new one. And the decisive point:
 > **The counterfactual for a one-shot script is almost never a durable instrument. It is
 > an assertion.**
 Before we measured, these claims came from reading and intuition. The one-shot did not
 displace a tool; it displaced a guess. Its fault rate only *looks* like degradation
 because **a guess has no observable fault rate at all**. A visibly failing instrument is
 strictly better than an unfalsifiable hunch, and mistaking the first for decline is how a
 system talks itself out of measuring.
 **What IS the violation: re-writing what is already banked.** Same session, I wrote a
 link-resolution canary inline — and that canary is in `/wake-up` *and* on
 [[reference-verification-ladder]]. Not proportion; failure to reach. **Rule of three:** an
 instrument reached for a third time stops being one-shot and goes to the ladder.
 **How to apply it.** When the fault rate looks alarming, do not conclude "build fewer
 one-shots" — ask the answerable question instead: **which one-shots are being written
 repeatedly, and were they promoted?** That is now counted, as the **K column** of
 `/wrap-up` §8's `Instruments` field (N run · M with a control written before first
 execution · K duplicating something banked). Three or four wraps will show a pattern or
 show none; neither of us could answer it from one day.
 **The distinction that generalizes:** *too few promoted* is a different diagnosis from
 *too many built*, and only the first is actionable. Related: [[feedback-checkable-claim-surfaces-bugs]],
 [[feedback-resurface-banked-notes-before-rederiving]].
@@ -620,3 +620,5 @@
 {"subject": "the ladder's 'a control must sit at the layer the defect lives in' (REVIEWED-83 A1)", "predicate": "prevention", "object": "DIAGNOSED BOTH of 2026-08-08's proxy-control failures, and named them as one class rather than two accidents — the cruft scanner that reported clean because it never fires on that file, and the fence check that used length as a proxy for authorship. The lesson was ALREADY BANKED and retrievable; what failed was firing it BEFORE each check ran, not knowing it. That distinction is the actionable one: this is a firing-moment problem (PENDING-112's thesis), not a knowledge gap.", "valid_from": "2026-08-08", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-08-08-voice-is-the-convocation-key.md", "extracted_at": "2026-08-08"}
 {"subject": "reading the substrate instead of its description", "predicate": "prevention", "object": "Reading Mauss's own lead-in lines rather than the sidecar titles produced nearly every finding of 2026-08-08: the LXIV/CXIV correction; the reclassification of `quotation-poet-jurist` from 'unnamed individual' to traditional matter (one footnote overturned it); the discovery that the naming evidence for six of nine blocks sits inside the FENCED apparatus, engine-unreachable; and that all twelve blocks are translated matter, which is what makes the quotation-in x translation-of composition cover the whole population rather than one Harrison edge case.", "valid_from": "2026-08-08", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-08-08-voice-is-the-convocation-key.md", "extracted_at": "2026-08-08"}
 {"subject": "three-party correction", "predicate": "drift-pattern-good-direction", "object": "RAN IN ALL THREE DIRECTIONS IN ONE DAY, which the differently-biased-checkers doctrine predicts but had not been observed doing. Jurist corrected executor (flat value -> two-job split). Executor corrected jurist (category placed in the convocation key), on evidence only substrate access yields. Steward corrected executor on a careless framing about language, which is what led to the L850 discovery. None of the three could have produced the result alone. Recorded as a single instance, proving nothing general — but the doctrine says to record evidence when it appears.", "valid_from": "2026-08-08", "valid_to": null, "confidence": 0.9, "source_file": "session-2026-08-08-voice-is-the-convocation-key.md", "extracted_at": "2026-08-08"}
 {"subject": "claude-code", "predicate": "drift-pattern", "object": "NAMED-A-REMEDY-CLASS-AND-CALLED-IT-DEPLOYED. Asked whether we could act rather than wait on a degraded instrument base-rate, I wrote 'act, don't wait' and listed three remedies — a pre-commit hook (an unauthorized PROPOSAL), a prospective-control count (a one-shot literal question that rotates out at the next wrap), and the ladder-ritual trial (MEASUREMENT, not a remedy at all). None was deployed. The steward's one-word challenge ('How?') was what exposed it. Rhetorical closure reads as mechanism precisely because a class plus three examples has the SHAPE of a plan. Test: for each named remedy, can you point at the thing that makes it fire without anyone remembering?", "valid_from": "2026-08-08", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-08-08-voice-is-the-convocation-key.md", "extracted_at": "2026-08-08"}
 {"subject": "a one-shot measurement", "predicate": "prevention", "object": "IS NOT A PRIME-DIRECTIVE VIOLATION, and treating it as one would have removed the thing that replaced guessing. Steward raised the worry 2026-08-08 after a bad week; the resolution is that the counterfactual for a one-shot script is an ASSERTION, not a durable instrument — so its visible fault rate is an improvement over an unfalsifiable hunch, not a decline. The real violation is re-writing what is already banked (a link-resolution canary typed inline the same day, though it lives in /wake-up AND on the ladder). Diagnosis: TOO FEW PROMOTED, not too many built — and only the first is actionable. Now measured as the K column of /wrap-up section 8's Instruments field.", "valid_from": "2026-08-08", "valid_to": null, "confidence": 0.9, "source_file": "feedback-one-shot-instruments-are-proportionate.md", "extracted_at": "2026-08-08"}
@@ -158,3 +158,44 @@ commit, and it surfaced only by accident. If the answer is *no* again, the remed
 proposed at §1.6 (extend the studium-engine pre-commit hook to run the fleet when
 `corpus/` or `corpus/sidecars/` changes) is what to build, and this is its second
 data point.
 ---
 ## CODA — after the wrap (the session did not end where the wrap did)
 **REVIEWED-98 placed and verified** (L1080, 8/8 load-bearing parts, no double-header).
 **The steward pressed on my base-rate answer — "How?" — and it was rhetoric.** I had
 written *"act, don't wait"* and named three remedies. Under challenge: **one was a
 proposal awaiting authorization, one was a one-shot I had described as standing, and one
 (the ladder-ritual trial) is measurement, not remedy at all.** Naming a remedy *class*
 plus three examples read as *deployed* when nothing was. Recorded as a drift pattern —
 rhetorical closure substituting for mechanism.
 **Then the steward asked the better question:** *"do we need so many single-use items?
 This seems to flow against our prime directive — but I honestly don't know."* Answer
 banked at [[feedback-one-shot-instruments-are-proportionate]]: **no** — a one-shot is
 proportionate to a question asked once, and its counterfactual is an *assertion*, not a
 durable tool. The violation is **re-writing what is already banked** (same day: I typed a
 link-resolution canary inline that lives in `/wake-up` AND on the ladder). *Too few
 promoted*, not *too many built* — and only the first is actionable.
 **Applied (FIX lane, three instruments discharged):** `/wrap-up` §8 now carries a standing
 **`Instruments`** field — *N run · M with a control written before first execution · K
 duplicating something already banked*. The **K column is the steward's question turned
 into a series** rather than left a worry.
 **Filed:** **PENDING-116** `[PROPOSAL]` — run the fleet on the change that breaks it.
 Design derived from *reading* the hook: `core.hooksPath` is `~/dotfiles/git/hooks`, so the
 hook is **tracked and travels** but is **global to every repo** → the trigger must be
 **repo-declared** (generative-from-spec, as the chamber already does). Costs stated up
 front: slower triggering commits · `--no-verify` bypasses it (a tripwire, not a boundary)
 · ⚠ **it does not close the cross-repo half** — a chamber-library edit can invalidate
 engine fixtures with **no engine-side commit at all**, so no hook fires on either side.
 That larger hole is named as an open follow-on, deliberately not absorbed.
 **A fourth check-vocabulary failure, same day.** My verification that the FIX-lane index
 line had landed grepped `Instruments field`; the line reads `**Instruments** field`, so it
 returned 0 against a line that was there. The write was fine; the *check* was wrong —
 again.