Scott Alexander, curated
← Back to curation

2016 Predictions: Calibration Results

Quality
55
Solid
Claude Shift
28
Slight
RWI
2
of 10

Summary

Year-end scoring of Scott's 100 calibrated predictions for 2016 (world events + personal/community), with strikethrough marking misses, bucketed hit-rates per confidence level, a calibration curve, and a brief reflection that he is mildly underconfident at 70% (consistent across 2014-15). A clean instance of the public-calibration ritual that models good epistemic practice, but as a standalone post it's mostly a results dump with light commentary.

Why this score

Quality 55 · Solid. Solid (55). Competent and mildly interesting -- the public calibration discipline is exemplary as practice and the underconfidence observation has some reflective value -- but it carries little new insight or craft as an essay; it's a housekeeping/ritual artifact. Sits in the Solid band, a notch above a bare list because of the calibration analysis and graph.

Claude’s paradigm shift 28 · Slight. Slight (28). Public scoring of calibrated predictions was an established rationalist practice by 2016 (Scott had done 2014-15; PredictionBook/LW precede him). A competent instance of a known ritual, essentially no novelty in its moment.

Real-world impact 2 · Minor. 2 -- minor/within-blog. The annual prediction-and-scoring ritual modestly helped normalize public calibrated forecasting in the rationalist sphere, but a single year's results post has only small-community influence.