Principlev1
Record predictions with specified confidence levels before
Record predictions with specified confidence levels before outcomes are known, then track calibration across predictions to identify domains of systematic over- or under-confidence.
Why This Is a Principle
Derives from Hindsight Bias and Calibration Necessity (hindsight bias), Systematic Overconfidence Taxonomy (systematic overconfidence), and Memory is reconstructive, not reproductive—humans rebuild (reconstructive memory). This principle operationalizes calibration tracking as a metacognitive practice, prescribing how to build reliable self-assessment through structured prediction logging.