Make bold predictions — then check them

State a specific, testable prediction about the future and record it before you see the outcome.

Why it works

Popper favored bold, highly specific predictions because they carry high falsification risk — they are easy to prove wrong, so surviving them is genuinely informative. Vague predictions survive almost any outcome; bold ones do not. Making predictions in advance and tracking them also builds the feedback loop that calibration requires: without records, outcomes are absorbed into narrative and the real hit rate is invisible.

How to do it

1. Before acting on a belief, write a specific prediction: "If X is true, then Y will happen by [date]."
2. Set a reminder to review the outcome at the specified time.
3. Score the prediction as accurate, partially accurate, or wrong.
4. Adjust the confidence in the underlying belief based on the score.

Evidence

Prediction scoring (Brier scores, calibration metrics) is the empirical method behind the Good Judgment Project and prediction markets. Forecasters who track and score predictions show measurably better calibration over time. (observational)

Benefits depend on honest scoring; selective memory of successes and forgetting failures undermines the calibration loop entirely.

Sources

• Tetlock & Gardner (2015), Superforecasting — prediction tracking and calibration improvement

Common mistake

Making a bold prediction and then, when it fails, redefining what you predicted — "I didn’t mean literally by that date" — which converts a falsifiable prediction into an unfalsifiable one.