Think in probabilities, not certainties

Replace "I think this will happen" with "I think there is a 70% chance this will happen."

Why it works

Binary thinking — will/won’t, right/wrong — conceals the uncertainty that is actually present in almost every forecast. Expressing beliefs as probabilities forces intellectual honesty about confidence and creates a framework for updating: if you say 70% and the event happens 40% of the time, your calibration is off and you can improve it. Binary predictions are unfalsifiable until something fails completely.

How to do it

1. When you form a belief about what will happen, assign a rough percentage confidence: 90%, 70%, 50%.
2. Use the percentage to communicate uncertainty to yourself and others rather than false certainty.
3. Track predictions over time and compare outcomes to stated probabilities.
4. Update probabilities when new information arrives rather than holding fixed positions.

Evidence

Calibration training — learning to assign accurate probabilities — demonstrably improves forecasting accuracy. Superforecasters who express views probabilistically and track them significantly outperform analysts who do not. (observational)

Calibration improvement requires feedback over many predictions; for one-off novel decisions, probability estimates are harder to calibrate and more uncertain.

Sources

• Tetlock & Gardner (2015), Superforecasting — probabilistic forecasters vs expert political judgment

Common mistake

Using round numbers (50/50, 99%) as defaults rather than genuinely differentiating between 60% and 75% — probability thinking requires real discrimination.