THE DATA
Six registers, published open, downloadable as CSV.
Everything on this site is built from these files. They are published so you can check the site against them rather than take its word.
No login, no key, no terms. Open them in a spreadsheet. If a number on a page disagrees with the register behind it, the register is right and the page is a bug, and we want to hear about it.
DOWNLOAD
The six registers
Row counts are as published. Every file is UTF-8 CSV with a header row.
| Register | What is in it | Rows |
|---|---|---|
| sources.csv | Every source behind every claim, with its date, its publisher, and an archive link where one exists. 124 of the 249 carry an archived copy. | 249 |
| claims.csv | Every claim the site makes, each tied to the sources that support it and labelled with how confirmed it is. | 76 |
| rebuttals.csv | The rebuttal index: a public statement, what the record says, and the document that settles it. | 34 |
| timeline.csv | The dated chronology. One row per event, each with its source. | 30 |
| news.csv | Coverage and public statements as they land. | 21 |
| glossary.csv | Every term of art on this site, defined once. | 17 |
SCHEMA
The files declare their own structure
datapackage.json is a Frictionless Data Package covering all six.
It names every field, its type, and which fields are required. A validator runs against it before anything ships, and a file that does not match its own schema stops the release. That is the reason these counts can be trusted rather than asserted.
How the numbers are produced, and what each one does and does not include, is written up at data and methods. The record they support is at the record.