Synthetic-layer assembly — audit v1.9 Phase 8 (2026-06-23)
Run: v1.9 Plato-cohort comprehensive audit · branch claude/audit-2026-06-23
Mode: Claude Code — governance halts are filesystem/maintainer decisions.
Method: mechanical harvest (harvest-claim-candidates.py) + the ~34 in-register Plato candidates; 6 claim-promotion-reviewer subagents (opus 1M, read-only, advisory) rigorously applied the 3-test live gate + the ≥2-anchored-bullets/≥2-distinct-dialogue auto-threshold; all spine writes + the promotion decisions on the main thread.
Register delta
| before | after | Δ | |
|---|---|---|---|
| supported | 22 | 23 | +1 |
| live | 121 | 142 | +21 |
| candidate | 135 | 113 | −22 |
| contested | 1 | 1 | — |
| retired | 1 | 1 | — |
| total | 280 | 280 | 0 |
22 candidates promoted (21→live, 1→supported); total unchanged (promotions redistribute status). Cap-exceed (22 vs ~10–15) maintainer-authorized for the Plato cohort (precedent: v1.8 run-13 promoted 19 under pre-authorization). The Plato register layer is now substantially promoted from its as-ingested all-candidate state.
Promotions
→ supported (1; maintainer-authorized — durable HALT cleared):
plato-virtue-teachability-staged-twice— the cohort's first Platosupportedclaim. Strongest-anchored: Protagoras 319a–d/361a + Meno 81a–86b/98c + Euthydemus 282c, three distinct dialogues all extraction-verified (the Protagoras extraction note pre-flagged this exact claim). 5-test: payoff (teachability staged twice as parasitic on the unanswered "what is virtue?") survives the complementary-stages counter; confidencemedium.
→ live (21; candidate→live, 3-test + ≥2/≥2, reviewer-verified):
plato-apology-philosophy-as-highest-piety, plato-charmides-anticipates-intellectualism (held at confidence low — gate passes but the claim self-describes as "not yet an argument"; maintainer-authorized), plato-critique-of-writing-spine, plato-disclaimed-authorship-device, plato-dubia-systematize-the-aporetic, plato-elenchus-fails-to-define-the-virtues, plato-hippias-minor-craft-analogy-reductio, plato-laches-protagoras-courage-mirror, plato-letters-philosopher-king-biographical-test, plato-lysis-seeds-symposium-erotics, plato-measure-saves-divergent-verdicts, plato-minos-supplies-the-laws-missing-definition-of-law, plato-moral-psychology-intellectualism-to-partition, plato-not-beings-spine, plato-poet-lacks-knowledge-two-routes, plato-political-art-staged-aporia-to-law, plato-republic-timaeus-critias-enacted-paradigm, plato-rhetoric-critique-genre-complete, plato-three-counterfeits-of-logos, plato-two-epistemologies, plato-two-routes-to-the-kalon.
Stay-candidate (5; not promoted)
- Test-2 FAIL (gestural 2nd source — only 1 dialogue genuinely anchored):
plato-self-is-soul-minimal-model(onlyplato-alcibiades-1129d/130c anchored; the Republic + Laws bullets are gestural) ·plato-statesman-ideal-ruler-vs-philosopher-king(onlyplato-statesman292c/293c/294a anchored; the Republic bullet is gestural). Single-source live would HALT; held at candidate. A future pass should either anchor the 2nd dialogue or keep at candidate. - Single-source (1 distinct source):
parmenides-self-critique-retains-forms,plato-spuria-academy-reception-stratum,symposium-dramatizes-its-counter-thesis. Held at candidate.
§Cousin disposition — the 7 cross-tradition cousins (HELD + classified; NOT promoted)
Per the v1.9 Phase-8 disposition rule + bridge-card-signoff §12e (a Latent-Adjacent cross-tradition cousin relation is below claim-grade by design), all 7 are held at candidate; each received a Status-History disposition note. The false-friend audit PASSED on all 3 high-risk pairs in Phase 1 (anamnesis≠Stiftung; khōra≠flesh; Form-of-the-Good≠Being/aletheia) — disanalogies are recorded in each Counterpressure.
| # | Cousin claim | Class | Next action |
|---|---|---|---|
| 1 | anamnesis-stiftung-cross-tradition-cousins |
RELATION-ONLY | follow-up weave: retire-and-relocate → *has cross-tradition cousin* (already on anamnesis) + motifs.md BRIDGE entry (added this run). Retirement HALTS for user confirmation. |
| 2 | mp-writing-and-living-inverts-phaedrus |
RELATION-ONLY | follow-up weave: retire-and-relocate → *inverts* (on writing-and-living) + BRIDGE entry. Keep candidate only if MP's literature course is shown to engage Plato directly. |
| 3 | peras-apeiron-khora-cousins |
RELATION-ONLY | follow-up weave: retire-and-relocate → typed connection (intra-Plato; the existing *shares mechanism with* is registrally over-strong per coherence §B) + BRIDGE-routed note. |
| 4 | republic-cave-aletheia-orthotes |
RELATION-ONLY (attribution) | follow-up weave: retire-and-relocate → existing typed connections + BRIDGE entry, retraction-caveat attached. Do NOT promote — the wiki's own aletheia page treats the warrant as contested (Heidegger's 1964 self-retraction). |
| 5 | sophist-not-being-as-difference-vs-determinate-negation |
CONTESTABLE-THESIS | stays candidate. The corrective-to-Hegel's-Parmenides-charge thesis is the contestable core; promotion blocked pending a Hegel-Sophist-reception cross-check. |
| 6 | khora-third-kind-mp-third-domain |
CONTESTABLE-THESIS | stays candidate (speculative). The genealogical primary-anchor-of-the-matrix-HUB thesis is the core; promotion blocked until Derrida's Khōra/Dissemination is ingested. |
| 7 | peras-apeiron-hegel-masz |
CONTESTABLE-THESIS | stays candidate; → Phase-8.6 merge-check vs collection-and-division↔bestimmte-negation (possible duplicate parallel re-described). |
Net: 4 relation-only (→ the separate cross-tradition weave run executes the retire-and-relocate, user-confirmed) + 3 contestable-thesis (stay candidate). The 6 cross-tradition BRIDGE motifs.md entries added in Phase 3 are the relocation targets for the relation-only set.
Carried-forward v1.8 candidates — 2nd-anchor re-test
The preflight asked whether any Plato dialogue now supplies a 2nd anchor for the carried-forward single-source v1.8 candidates (science-secretes-ontology-non-foundational-reciprocity, solus-ipse-dissolved-by-removing-its-origin, + the 12 D7 + 4 D4). Result: no. These are MP/Husserl-specific theses; the Plato cohort does not supply a second cross-author anchor for any of them (a Plato dialogue is not evidence for an MP-internal claim). They remain single-source, deferred to a future non-cohort Phase-8 pass (out of this Plato-scoped run, like v1.8 deferred its non-cohort harvest).
New candidates surfaced (flagged, not added this run)
The corpus-wide harvest surfaced ~82 candidates, overwhelmingly non-Plato carried-forward backlog (Heidegger/Husserl/Kee/Klossowski) — out of scope for this Plato-scoped run. Two genuinely-new Plato candidates were flagged for a follow-up pass (not added, to bound the claims.md edit):
ion-enthousiasmos-phaedrus-mania(Ion's enthousiasmos ↔ Phaedrus' poetic mania as one divine-madness figure) — structural-parallel across 2 dialogues; a plausible candidate.- A to prepon counterpressure touch for
plato-measure-saves-divergent-verdicts(the same Greek word across Statesman/Hippias-Major).
Cite-back (Step 8.9) — DEFERRED
The 22 promoted claims are cited on ~20+ concept/source pages, several with now-stale (candidate) framing that should become (live claim …) / normal citation for the supported one. Cite-back across that many pages exceeds the >20-file batch gate; deferred to a scoped follow-up sweep (precedent: v1.7/v1.8 deferred cite-backs). Not lint-breaking (the prose framing is not a structured field). Backlog item recorded in log.md.
Outputs
wiki/claims.md(22 promotions + 7 cousin disposition notes) + regenerated Summary table.wiki/index.mdClaims section + count (vianormalize_claims_table.py).- This report. Phase 7 addendum: appended below after the post-promotion confidence re-pass. Phase 8.6 queue:
claim-retirement-queue-2026-06-23.md.
Phase 7 addendum (post-Phase-8 confidence re-pass)
Per Step 8.8, re-passed confidence on pages whose evidential standing changed via the promotions. Result: 0 frontmatter changes. Promoting a claim from candidate→live/supported records the wiki's growing confidence in the synthetic thesis, but does not change how accurately a concept/source page represents its sources (the meaning of the confidence: field). The Plato pages were already correctly calibrated (Phase 7 main pass); no page's source-representation confidence shifted because a claim it homes was promoted. charmides was deliberately kept at claim-confidence low even at live. No page edits.