The Method

What our tags mean

A breakdown may carry one descriptive tag, chosen by a human editor. Tags are categories, not a scale: one isn't "worse" than another, and many breakdowns have no tag at all. Source articles we link to are never tagged.

Competing narratives

Credible sources tell incompatible versions of the same event, and the evidence does not yet settle it.

Emotional framing

The facts are accurate, but word choice and emphasis are built to make you feel something.

Selective evidence

Real facts are picked or left out so the picture tilts one way.

Unverified claim

A central claim rests on a single or anonymous source and hasn't been confirmed.

False claim

A central claim is contradicted by strong evidence. This is about the claim, not the motive.

Documented deception

The record shows someone set out to mislead. This needs evidence of intent, not just a false statement.

A false statement isn't the same as deception. People repeat false things they believe. We only use Documented deception when there's evidence of intent, such as planning records, forged material, or an admission. A False claim needs strong contrary evidence, and says nothing about motive.

Framing isn't automatically false. Every story has to be framed somehow. Emotional or selective framing can sit on top of entirely accurate facts, and we say so.

Repeated wording isn't automatically coordination. Outlets share wording for ordinary reasons: wire copy (AP, Reuters, AFP) reprinted by many papers, quotes from the same speech, publishers posting the same story on their site and on X, press releases, and common phrases. We count a publisher's own copies as one origin, ignore quoted remarks, and skip generic phrases. Even then, shared wording is a reason to look closer, not proof.

"Earliest observed" isn't necessarily the origin. We only see the sources we track. When we say a source ran something first, we mean first among those sources. Some outlets only give a date, not a time, so the order within that day is uncertain and we mark it.

Automated preparation, human review

Source selection. Software collects coverage from a fixed list of outlets and X accounts across wire services, legacy media, independent and alternative outlets, and state media, plus news aggregators. That list will always be incomplete.

Automated preparation. The software groups stories that push the same message into narratives, finds the competing frames, flags repeated phrasing, and prepares a draft breakdown.

Human review. Nothing is published until an editor reviews, edits and approves it. Approved breakdowns show the time they were published and who reviewed them. Assessment blocks marked Draft, pending editorial review haven't been signed off yet.

Corrections. Spot an error? Email corrections@psyophq.com. When we fix a factual error, the article shows an updated time.

Limits. We work from published reporting, not our own reporting on the ground. Assessments can change as evidence comes in. Our numeric scores are editorial judgments, not measurements.