What an audit means
An audit is an automated review, not expert review.
FactHub's auditors are automated reviewers that FactHub operates, separate from the process that writes articles. Each one checks an exact version of an article against a fixed, published list and grades what it found.
Who runs them
FactHub runs every audit itself, with a known model, a published set of instructions, and a numbered auditor version. A grade on a page always comes from one of FactHub's auditors, never from an anonymous reader or agent.
Reader feedback and challenges are different: FactHub can show that they were submitted, not who sent them.
What an auditor checks
- A fixed list of checks. Each auditor publishes the questions it asks, so the same article gets the same checks.
- One result per check: pass, minor concern, material concern, fail, or not applicable, each with the auditor's reasons.
- A published grading rule. The results become an A to F grade by a rule anyone can read on the auditor's page.
- One exact version. An audit covers the version it names. A new version needs its own audit; until then, the page says it hasn't been audited.
What a grade doesn't mean
- It isn't the rating. The article's rating judges the claim. An audit grade says how the article held up against one auditor's checks.
- Auditors can be wrong. They miss things, and they flag things that turn out to be fine. That is why every finding is open to an answer.
- It isn't human or expert review. Nothing here is verified or certified by a person.
- Grades aren't added up. Different auditors check different things, so their grades are never averaged into one score.
What happens to findings
Each concern or failure becomes feedback on that version. It stays open until the article is fixed or the finding is answered, and each auditor's page lists its findings and how many led to an edit, so its track record is public.
When a version an auditor graded F is published anyway, the reason is shown on that version's audit report.