The Record

Judged by Reality

Every call on this page was filed in writing before the outcome, with a confidence number and a stated condition that would prove it wrong. Reality does the grading. Nothing is curated, nothing is deleted — the misses are the point.
Stamps — RIGHT / WRONG / OPEN. Reality assigns them, not us. A wrong stamp never disappears.
Confidence — how sure we were, in writing, before the outcome. 60% means "we expect to be wrong 4 times in 10." Being wrong at 60% is normal; being wrong at 95% is the scandal.
Accuracy score — how much better our confidence numbers are than coin-flipping. For professionals: it's a Brier-based skill score; for everyone else: higher is better, zero means guessing.
The Ledger
Open calls take your prediction too — you'll be scored against ours when reality rules.
The Post-Mortems
When we're wrong, we say so in first person — and name the reason.
Us vs the Crowd
Our probability against the prediction-market price on the same question — shown only after reality has ruled.
The Stories That Evaporated
Last week's loud headlines that no one ever confirmed. We keep the list nobody else keeps.
What Went Quiet
Topics that stopped being reported without being resolved. Silence, timestamped.
Sources Report Card
We keep score on the outlets we read: does what they print get confirmed — or does it vanish?
SourceGradeRecord