A Joosr Guide to… Superforecasting by Philip Tetlock and Dan Gardner

Most of the tasks we complete in life are learned through a similar technique: try, fail, analyze, adjust, try again.

This approach works well in any scenario: simply consider the average, then adjust it according to any additional or unique information that you have.

Whenever you make a prediction, always remember that no event or situation is entirely unique.

Good predictions involve much more than just a yes, no, or maybe guess; they contain comprehensive detail that can be measured, and adjusted over time.

To make a good prediction, a forecaster must avoid ambiguous language, instead using specific numbers for probabilities—and the more detail, the better.

trong forecasts should use specific numbers for probabilities, rather than vague phrases, or what is known as “elastic language” (when we are told there is a “fair chance” or a “serious possibility” of something happening). Numbers make a forecast exact, which decreases the risk of confusion. In addition, it shows the observer that the prediction is an estimate—a conclusion reached by the forecaster that has been developed due to evidence-based research and comparison, rather than a subjective judgment.