2020 Predictions: Calibration Results

The annual calibration-scoring post — Scott grades his 100 predictions for 2020, bins them, and finds that for the first time he was consistently overconfident in every bin, driven by two correlated misses (expecting a much shorter COVID lockdown and missing the NYT affair). The genuinely interesting bit is the meta-argument: 2020 was his weirdest year of black swans, and if some years are normal and some weird but you can't tell which in advance, correct long-run calibration means looking slightly underconfident in normal years and significantly overconfident in weird ones — so a single bad year isn't necessarily miscalibration (he resolves to check whether it averages out). Minor lessons (overestimates friends' relationships lasting; bad at predicting his own purchases). A worthwhile recurring accountability exercise, consistent with the SSC calibration-results band (49-58).

Quality 60 · Strong. Solid (60, top of the calibration-post band). The black-swan-averaging insight lifts it slightly above the typical annual calibration post (SSC 49-58), but it remains a recurring self-scoring exercise, largely a list plus binning.

Claude’s paradigm shift 40 · Moderate. Slight-Moderate (40). Public calibration-scoring is good epistemic hygiene; the black-swan-averaging point is a nice sharpening, not a novel frame.

Real-world impact 1 · Negligible. 1 — a personal prediction-scoring post; negligible real-world effect.