A register of machine-citation findings
Court findings about fabricated and mischaracterized citations, matched to the issuing court file, with every correction recorded.
What this is
A published record of instances in which a court has made a finding about a citation or assertion that a machine produced and a person filed. Each entry is matched against the court's own file. The methodology is published, the corrections are published, and the data is downloadable with its manifest and timestamp proofs.
What it does not claim
- It does not rank vendors. 283 of 366 findings — 77.3% — do not identify the system involved, because the court had no reason to name it. There is no denominator for a league table and none is published.
- It publishes no distribution of sanction amounts. Only 9 of 366 findings — 2.5% — establish a remedy at all, and the dominant form is fee-shifting with the amount fixed later by invoice. Any amount statistic from this corpus would describe amounts stated in a first order, which is a biased subsample that excludes the usual remedy.
- It is not a list of people. Outcomes attach to citations and claims. Where a court anonymized a party, the register uses the court's caption and nothing else.
- It is not a search product. This is a publication. There is no account to create and nothing to subscribe to in order to read it.
Sections
The register itself, revision 4, with the court file matched to each entry.
How an entry is made, what counts as a match, and the nine guards that stop a build.
The policy, and how to report something wrong.
CSV, JSON, manifest, timestamp proofs, contribution schema, system registry.
Every revision, what changed in it, and its hash.
Why a register rather than a tool
Tools that check citations already exist and are in wide use. None of them leaves an artifact a third party can inspect: the check happens privately and produces a result its user is asked to trust. A register is the other half of that — a published record, with a stated method and a correction history, that someone who was not present can examine. That is the whole of the claim being made here.