The EJ-1 standard

Empirical Journalism Standard (EJ-1) · v1.0

EJ-1 governs how a reader assesses a claim published here without having to learn forensic notation. The Trustless protocol underneath it carries the granular evidence layer: hashes, archive links, chain of custody, and a methodology with its required sections. EJ-1 sits above that layer and states, in public and in advance, what the publication binds itself to.

A publication declaring EJ-1 commits to all six of the following. A publication that cannot meet one of them may declare partial compliance and document the gap. It may not claim full compliance and hope nobody checks.

1. Graded claims

Every claim carrying the argument is published with a grade, the receipt behind it, and its disproof condition, under Receipts v1.0. The five grades — Established, Supported, Reported, Thin, Disputed — replace the E1–E5 public tier used in earlier versions of this standard, and replace the MHEES six-axis code and the Admiralty grade outright.

The grade is set by the three questions in the Receipts spec, answered in writing before publication. An analyst’s confidence never promotes a claim. Corroboration, provenance, and a short testable leap promote it.

2. The evidence profile and the honesty flag

Each piece publishes the spread of its grades and names, out loud, the weakest claim its conclusion depends on. A summary that hides which weak claim carries the roof is a violation of this standard, not a stylistic choice.

3. Binding right of reply

Every named subject facing a serious allegation receives the substance of it before publication and a 72-hour hold in which to respond. The response is carried in the piece. This is binding: the hold elapses before publication, and a subject’s silence is reported as silence rather than treated as consent.

4. Append-only corrections, no silent edits

Corrections are appended with a date and the substance of what changed. Nothing is quietly revised. Superseded figures are disavowed in the record rather than replaced, which is why an earlier draft’s numbers remain visible in the published packages alongside the reason they did not survive. See the corrections ledger.

5. Documented chain of custody

Every cited source is archived where it was read and fingerprinted with a SHA-256 hash, and the chain from source to sentence is recorded. Fetch failures are logged as failures. An archive with gaps reports itself as incomplete rather than passing; a gate that waves through a gap is not a gate.

6. A killed thesis stays in the record

A hypothesis disconfirmed by its own disproof condition is recorded as refuted, with the date and the reason, and kept. A deleted failure is a lie of omission.

Partial compliance

Where a piece meets some commitments and not others, it declares partial compliance and names the gap on the page. The published water-deliberation package is a live example: its archive is missing web-archive snapshots for two of eleven sources, its gate does not currently pass, and its page says so in the reader contract instead of implying a completeness the archive has not earned.

Adoption

EJ-1 is written to be adoptable by someone who is not Brian Nuckols. Any journalist or outlet may declare it, and any contributor publishing here publishes under it. The standard is versioned and public so the work can be held to the same rules it claims.