2014 Predictions: Calibration Results

Scott's annual calibration scorecard: he grades his 59 confidence-tagged 2014 predictions (world events plus personal life, including micronation Shireroth in-jokes) and tabulates hit rates by confidence bucket (50%->50%, 60%->77%, 70%->75%, 80%->80%, 90%/95%/99% all 100%). Concludes 'I declare myself to be impressively well-calibrated. You should all trust me about everything.' Valuable mainly as a clean instance of the calibration practice — modeling good epistemic behavior — rather than as an essay.

Quality 56 · Solid. Solid (56). A well-executed instance of a genuinely valuable rationalist practice (public calibration scoring), but as a post it is largely a prediction list plus a results table with minimal prose.

Claude’s paradigm shift 42 · Moderate. Moderate, low (42). The annual-prediction-scoring format and calibration concept were already established in the LW/forecasting world; this is a competent instance, not a novel one.

Real-world impact 2 · Minor. Minor (2). The public-calibration-post practice spread within the forecasting/EA community; a single year's scorecard is a modest contribution to it.