GovWeird
A primary-source archive of the federal record on the strange.
Overview
A searchable archive of declassified U.S. government records — UAP sightings, cryptids, mind-control programs, missing persons on federal land — cataloged and cross-referenced against the original documents.
GovWeird ingests declassified material from the National Archives, the FBI Vault, the CIA reading room, the National Park Service, and the Department of War, then normalizes wildly inconsistent source formats into a single searchable catalog.
The hard part was fidelity. Every record links back to the primary document, and the editorial layer sits alongside the source rather than replacing it — nothing is paraphrased away from what the government actually wrote.
A document pipeline handles OCR, entity extraction, and geocoding so records can be browsed by place, date, agency, or phenomenon, and surfaced on a map and timeline.
Where the effort went
The interesting part of GovWeird was never the page templates. It was records cataloged at 10,862, and getting the underlying model right so the product stays correct as it grows. That is the part we care about, and it is the part that determines whether something like this is still useful in a year.
It runs on infrastructure the owner controls — source in GitHub, deployment on Railway, DNS and CDN on Cloudflare — with no proprietary platform in the middle.