About
No human writes these pages. A scheduled job runs once a day, fetches public data, writes one JSON file, and commits it to a version-controlled archive. That commit is the publication.
Five independent layers, captured concurrently. Any one can fail without losing the day.
| Layer | Question it answers | Source | License |
|---|---|---|---|
| attention | What did the most people look up? | Wikipedia pageviews | CC BY-SA 4.0 |
| frontier | What did builders argue about? | Hacker News | Public API, credited + linked |
| research | What was published? | arXiv cs.AI | Open metadata |
| history | What happened on this calendar day? | Wikipedia On this day | CC BY-SA 4.0 |
| census | How does the web treat machines? | Direct measurement of 2,110 origins | Our own observation |
Records are immutable.
Once a day is captured with all four sources, its file is never rewritten. Corrections would defeat the purpose: this is a record of what was true to observers on that day, not a tidied summary. Partial days stay eligible for repair until complete, and are labeled as partial until then.
Zero dependencies.
The whole system is plain JavaScript on Node's standard library. Nothing to install, no lockfile, no supply chain, no dependency that can rot. A machine meant to run for decades should have nothing in it that expires.
Gaps are stated, never hidden.
If a source is down, the record says which sources it captured and marks itself partial. An archive that quietly papers over its holes is worse than no archive.
Everything links home.
No item appears without a link to where it came from. This is an index of public sources, not a replacement for them.
The census is the part that cannot be faked.
Four of the five layers aggregate what others published, and could in principle be rebuilt by anyone from the same public APIs. The census could not. It is a direct observation of what 2,110 origins were serving on a given morning, and nobody — including us — can go back and measure a day we did not measure. That is the whole reason it runs every single day.
The archive inherits CC BY-SA 4.0 from Wikipedia. Use it freely, credit it, and share derivatives alike. The code is MIT. If you build something on this, a link back is all that is asked.
curl https://28081999.me/api/index.json curl https://28081999.me/api/day/2026-08-25.json curl https://28081999.me/api/since/2026-05-01.json