Methodology

Our editorial process is public because our credibility depends on it. This page explains how we select, verify, produce, and correct our work.

How Stories Are Selected

Every story candidate is evaluated against six criteria before it enters the editorial pipeline:

  • Relevance — Does this affect public understanding of a material issue?
  • Timeliness — Is this happening now, or is there a time-sensitive element?
  • Evidence availability — Can the claims be checked against primary sources?
  • Contrast potential — Is there a meaningful gap between what was said and what can be shown?
  • Legal risk — Can this be published responsibly and defensibly?
  • Platform suitability — Does this work as short-form video, long-form article, or both?

We do not chase outrage. We follow evidence.

How Sources Are Verified

  • Claims are separated from facts before any editorial work begins.
  • Multiple-source corroboration is required for every factual assertion.
  • Unknowns and contradictions are explicitly flagged — never papered over.
  • Primary sources are preferred. Secondary sources are cited only when primary sources are unavailable, and the gap is noted.

We do not label something a lie unless the contradiction is explicit and well-supported.

Where AI Is Used

AI is a tool in our editorial process. It handles volume and structure so editorial focus can go where it matters. Specifically:

  • Story clustering and candidate scoring
  • Verification packet assembly
  • Script and article drafting
  • Narration generation
  • Video assembly from templates

AI assists the editorial process. It does not replace editorial judgment.

Where Humans Review

Human review is mandatory before publication for any content involving:

  • Named-person accusations
  • Criminal allegations
  • Breaking news with weak corroboration
  • Claims of lying or deception
  • High-risk political content
  • Uncertain licensing or rights status

These categories are non-negotiable. If a piece touches any of these areas, it does not publish without explicit human sign-off.

How Corrections Work

Mistakes happen. What matters is how they are handled. Our correction policy:

  • Visible corrections are placed on affected pages with a clear description of what changed.
  • Retractions are issued when a published piece is materially wrong.
  • Update notes are added when context changes after publication.
  • Original audit history is preserved — we do not silently edit published work.

All corrections are tracked on our Corrections & Retractions page.