PROOF.md said 'there is no false-positive rate on this page, because we have
not measured one' while corpus/README.md, same version and same date, published
the full FPR/TPR table with confidence intervals. Both cannot be true, and the
page asserting the absence is the one people read first.
The honest version is narrower than either: the rate is measured, and it is
withheld as a headline because it fails this repo's own publication gate on the
register test. That is a stronger statement than 'we have not measured one' --
it names the number, names the gate, and names which clause the number fails.
npm run self-scan:check: PASS (PROOF.md 13 exempt against a budget of 20).
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds a preservation validator, a self-scan with PROOF.md, over-correction guardrails, a false-positive intake, and a two-class control corpus with the project's first FPR, TPR, and AUC.
Splits Tier 1 into 1A frequency markers and 1B clarity edits, and fixes two detector defects the measurement found (#67, #69).
Findings and open decisions: #70, #71, #72, #73.
Conor Bronsdon