Principlev1PR-253
Aggregate predictions by confidence level and compare stated
Aggregate predictions by confidence level and compare stated probability to actual frequency across dozens of cases rather than evaluating single predictions, as systematic patterns reveal calibration while individual outcomes contain too much noise.
Why This Is a Principle
This principle follows from Systematic Overconfidence Taxonomy (overconfidence as systematic) and Hindsight Bias and Calibration Necessity (need for calibration systems). It prescribes aggregation as the method for detecting systematic error that individual cases cannot reveal—a statistical principle applied to personal epistemology.