Method

Start with the claim. End with the record.

The system is designed so a reader can retrace the audit instead of being asked to trust the person who performed it.

1. State the exact claim.

Before research begins, the factual proposition is written as precisely as possible. If the circulating statement contains multiple propositions, those propositions are separated before classification.

2. Identify where it came from.

The review looks for the original post, document, transcript, report, dataset, article, filing, or closest traceable source. A screenshot or clipped quote is treated as a starting point when the underlying material can still be found.

3. Trace the citation chain.

If Source A cites Source B, the audit checks Source B. If ten accounts all repeat Source A, that does not create ten independent sources.

Repetition can measure reach. It cannot manufacture provenance.

4. Separate evidence from interpretation.

The review distinguishes what a source directly establishes from what a writer, creator, lawyer, commentator, or reader inferred from it. Questions, argument, speculation, and opinion are not silently promoted to evidence.

5. Compare each component with the record.

Each material subclaim is tested independently. One supported fact cannot rescue an unsupported accusation attached to it.

6. Publish the classification and its limits.

A finding says what is supported, contradicted, unsupported, unresolved, or not fact-testable. It also states what the cited evidence cannot establish.

7. Correct the record when better evidence arrives.

Material changes are documented. The current finding is updated, but the correction history remains visible so the audit itself can be audited.

Future submissions

Reader submissions will eventually be allowed, but they will enter as leads—not votes and not accepted proof. The submission must identify its source well enough for the source itself to be checked.

Next: how source provenance is evaluated →