What makes a prediction testable
Four things have to be present. Most public predictions have two or three, which is why so few of them ever get marked right or wrong.
August 29, 2026 · figures counted October 4, 2026
Four things have to be present before a prediction can ever be marked right or wrong: a measurable outcome, a threshold, a deadline, and a source that publishes the answer. Miss one and there is nothing to check.
In this archive 65 of 126 records carry written criteria and 63 of 126 carry a deadline.
The four
A measurable outcome, not a mood. “AI will change everything” cannot be settled; “more than half of X by date Y” can.
A threshold, so that “grows a lot” becomes a number. A deadline, without which a claim is never wrong, only not yet right. And a source — someone has to publish the answer, and both sides have to agree in advance that it counts.
Where they are written down
On a record here the four live in the resolution criteria, fixed before the outcome is known and shown alongside the claim. The point is not rigour for its own sake. It is that a reader can see the test and judge it themselves rather than taking a verdict on trust.
Every claim here is checkable.
The exact words, the source and its date, the test fixed before the outcome, and what happened when it landed.