Trust Receipt

Content parity: HTML pages and Markdown mirrors

The public Markdown mirrors for essays, field notes, and numbered book chapters matched their canonical HTML pages under the textual comparator used in the build gate.

Verdict
passed
Textual parity, not a semantic proof.
Freshness
2026-07-19
Git-derived input date.
Checks
57
57 passed, 0 failed.
Thresholds
2.5% delta max
0.98 overlap min

Boundary

This receipt is not a safety certification or a claim that every site surface has a Markdown twin. It proves point-in-time textual parity for the mirrored writing documents listed in the report.

Passed

Mirror parity gate
passed
Compared 57 mirrored documents. Worst word-count delta was 1.38%. Weakest token overlap was 0.988.
python3 scripts/check-content-parity.py
Published report
passed
The JSON report is served at /parity-report.json and the summary appears on /ai.
parity-report.json

Evidence

MeasureValue
Documents checked57
Total HTML words88339
Total Markdown words88505
Report/parity-report.json

Intentionally Excluded

Navigation, footers, scripts, styles, and buttons
The comparator strips page chrome so it checks document text, not the surrounding website shell.
Pages without Markdown mirrors
Standalone tools, dashboards, data assets, and indexes are outside this receipt unless they gain mirrors later.
Semantic equivalence
The check catches textual divergence. It does not prove two texts mean the same thing in every context.

Limitations

Token overlap is robust enough for dropped text, but it is not a plagiarism detector or a semantic model.
Checksums are included for drift triage only; exact equality is not required because HTML and Markdown extractors intentionally differ.
Future pages need mirrors and gate coverage before they are covered by this receipt.

Reproduce Or Inspect

Run the comparator.
python3 scripts/check-content-parity.py
Refresh the public report and receipt.
python3 scripts/check-content-parity.py --write