Prediction Without Precision

Philip Tetlock spent 20 years running the Good Judgment Project. His central finding: human forecasters can be meaningfully calibrated. The better ones don't just get direction right — they know how right they are.

Prediction markets ignored this. For 60 years, they've been built on a binary: yes or no, up or down. A 51% correct prediction and a 99% correct prediction pay out the same way. The market treats them identically.

Nobody questioned it. Until Trepa.