FINITE NOISY AUDITS WITH LEARNED BASELINES
Technical addendum v1.0 - 9 November2026

Start with manuscript.html, manuscript.pdf, or manuscript.txt.
REPRODUCE.txt gives offline mathematical verification commands.
source-proof-ledger.json maps each proposition to assumptions, proofs, tests,
sources and limits. bibliography.txt records source reading depth.
ai-review-summary.txt and responses-to-reviewers.txt describe internal AI review.
publication-manifest.json records release metadata and hashes.

Suggested destination: a technical addendum attached to the Reflective Self-Models
line at https://www.b-theory.com/papers/ . No new public route is claimed created.
This archive is a local release candidate; creating it did not upload anything.
Named author credit and site metadata remain the publication coordinator's
editorial decision. The reading copy currently uses B-Theory technical-addendum
attribution without inventing an individual author line.

The public package contains only this manuscript, bibliography/claim ledger,
public review material, mathematical verifier source, and finite data/ROMs.
No private review-session identifiers, local machine paths, scratch notes, secret
material, or unrelated research are intended for this archive.
