About
Last updated: September 2, 2026
What this is
HistoryRewritten detects unusual Wikipedia revision activity around an event date you choose and shows source-grounded before/after evidence: a deterministic, model-free activity score measured against the article’s own history, and an exact comparison of two pinned revisions with links to the official diff. It was built by one person who suspected that certain articles had been reframed after October 7, 2023, and wanted a tool that could tell him he was wrong — one where control articles have to come back typical for the method to mean anything. The why page tells that story with sources; the methodology and FAQ say the rest.
Independence
HistoryRewritten is independent and unfunded: no sponsors, no grants, no advertising, no paid tiers, no affiliation with any advocacy organization. Hosting and model costs are paid by the maker. It is not affiliated with or endorsed by the Wikimedia Foundation or Wikipedia. It reads the same public Wikimedia APIs anyone can, following the published API etiquette, and reuses article text under CC BY-SA 4.0 with attribution to the exact revisions.
Open source
The website, API, and Chrome extension are one repository under the MIT License: github.com/cheesejaguar/historyrewritten.org. Every constant that shapes a score lives in one configuration file, the algorithm is documented in the repository, and the unit tests run against recorded fixtures without touching the network. If you want to check a claim on this site against the code, that is where to look.
What it will not do
- Call activity proof of anything. Activity is not proof of bias or misconduct. Every surface says so, and no surface ranks articles as “worst” or names offenders.
- Show who edited. The product returns aggregate editor statistics only — never usernames, never inferences about identity, ideology, or intent. This is a hard boundary, not a setting.
- Score thin data. Articles without enough pre-event history say so instead of pretending to a verdict, and missing metrics surface as flags, never as a low number.
- Collect more than it needs. No accounts, no tracking cookies; see privacy.
How to reach the project
- Questions, corrections, wrong findings: GitHub issues. Every report has a Report an issue link that prefills the report URL and identifiers.
- Security vulnerabilities: please use GitHub’s private vulnerability reporting rather than a public issue. The machine-readable contact is at /.well-known/security.txt.
- Accessibility barriers: the site aims at WCAG 2.1 AA; report anything that gets in your way as an issue and it will be fixed.
See also the terms of use and the changelog.