The sourcing rule

No claim about a living person enters the record without a source.

Enforced at the door

The requirement is checked when a record is written, not when it is approved.

This ordering is deliberate and it was a decision, not an accident. Enforcing it only at approval time leaves uncited claims about real people sitting in an inbox — read by reviewers, present in the system, one mistake away from being rendered. The door is the right place for it.

The same rule applies with extra force to the highest-risk submissions: reporting a problem law and flagging a worrisome challenger both demand sources at submission, and neither is ever published on the strength of one submission.

What a source has to be

Something a stranger who does not trust us can go and check. A roll call. A committee calendar. A filing. A video of the person saying it. A published document.

A source is not: another page on this site, an assertion by somebody we like, or a summary of a source that nobody linked.

Provenance

For anything the movement downloaded, there is a provenance record: what was fetched, from where, and when. Voting records, politician photos, district maps and transcriptions all carry them.

So a disagreement about our data can be resolved by looking at what we actually pulled, rather than becoming an argument about whose memory of it is right.

State what somebody did, not what they are

The sourcing rule has a companion that governs wording. Records say what a person did or said, with a citation — never what they are, what they wanted, or who they were working for. Motive is an argument. Acts and quotes are a record.

Gaps are published

Where the record is thin, it says so. A named gap is a real finding, it is more useful than a confident guess, and it tells a citizen exactly what contribution would close it.