How we verify

"Verified" is an overused word. Here is exactly what it means on this site today, and what it does not mean yet.

What we do measure

An item's independent-source count is a real tally of distinct outlets carrying the same claim. Our Global Confidence synthesis is deterministic (no language model), built from five measured factors -- it is a confidence estimate about the evidence we hold, not a truth verdict.

What we do not yet do

Scenario risk scores, trend indicators and forecast confidence are declared assessments -- model- and analyst-derived -- not yet validated against resolved outcomes. Read them as signposts, not facts.

Calibration, honestly

We publish a public calibration scorecard that will compare our forecasts to what actually happened, using the Brier score. As of this writing it shows zero resolved forecasts -- an honest empty state, not a fabricated track record -- because no forecast horizon has come due yet.

Corrections

Every illustration we remove or correct, and any article-level correction, is recorded in our public corrections log.