judgement
Fine probabilities matter
The rule
Small steps in your odds really do matter. The gap between 63% and 65% carries real meaning that lazy rounding throws away.
Where it flips
Past a point, extra decimals are just theatre. Nobody knows a stock is 63.4% likely to rise. The fix is to use as much fineness as your evidence supports, and no more.
It sounds like fake precision, but the proof is clear. The best guessers tell apart fine steps of confidence. When they round their 63% to a lazy "about two-thirds," their calls get worse. So treat your odds like a dial with many settings, not three rough buckets of low, medium, and high. That extra fineness is real signal, not decoration.
A worked example
A guesser who files calls as 60/65/70% beats one who lumps everything into "likely." Across many calls, those fine gaps track real differences in the world. [illustrative]
How to spot it
- ·confidence lumped into few buckets
- ·fine gradations dismissed as false precision
- ·information lost to rounding
Philip Tetlock & Dan Gardner · Superforecasting