Evidence
Evidence and corrections
Selected corrections, finite findings and unresolved verification gaps.
Source-text parity
The earlier Study 14 check could accept matching A/B identifiers and order despite changed source text. Its successor rejects the text mismatch. This establishes one tested checker property; the full empirical design remains a separate validation question.
Projection-oracle convention
The earlier unrestricted projection specification treated a strict-improvement witness as a strict hull vertex. The correction separates valid boundary witnesses from counted strict vertices. It concerns the parametric-closure engine, not the bounded semiorder theorem.
Finite counterexample
The #413 first-four-offset conjecture failed a recorded finite check. The exact counterexample and associated records remain under release review. Neither infinitude nor global held-out secrecy is established.
Prototype coverage
Earlier tests missed intended propagation and authority behavior. Their passing status did not establish complete semantic coverage; repair of every defect remains unverified.
Evidence scope
Proofs support stated claims under their assumptions; finite checkers cover exercised domains; hashes identify bytes; public releases establish availability.
Public mathematical releases contain arguments and replay materials. Private episode records, correspondence and internal ledgers remain outside this site.