Corrections and release criteria
Report a suspected error to [email protected]. Include the page URL, publication version, affected claim or figure when known, and a short description. Do not attach private research inputs or sensitive local files.
The address is the correction channel for this public review.
What happens to a report
The report is acknowledged and triaged as metadata, presentation, evidence, method, thesis, or conclusion affecting.
Affected claims return to pending review. Source anchors or signed evidence traces are regenerated and independently checked.
A correction creates a new immutable version and names the affected claims, results, figures, and conclusions. Released pages are not silently rewritten.
Superseded versions remain reachable and point prominently to the correcting version. Serious errors are preserved and labeled.
Stable-release criteria
Version 1.0.0 remains blocked until all of these conditions hold:
public review has run for at least 30 calendar days;
a second claim-ledger review is complete;
every metric discrepancy is resolved or explicitly documented;
no unresolved correction would materially change the thesis;
browser, accessibility, static-fallback, and canonical-build gates pass; and
citation, rights, and archival packaging pass their release review.
The candidate date for version 0.1.0 is 2026-07-22, but the minimum review window begins only with approved public deployment. Deployment on that date would make 2026-08-21 the earliest completion; a later deployment requires this record to be regenerated before approval.
A material change to the thesis, evidence, analysis method, or conclusions creates a new 0.x release and restarts the minimum review window. Minor conclusion-preserving presentation or metadata corrections do not.