Public figures who make explicit probabilistic forecasts should pre‑register their predictions with stated credences and then publish a standardized postmortem showing hits, misses, calibration statistics and causal lessons. That routine would convert messy punditry into traceable epistemic practice and create public learning about what forecasting methods work.
— Normalizing pre‑registration and public postmortems for high‑visibility predictions would raise civic epistemic standards, reduce overconfidence-driven misinformation, and create auditable incentives for humility among media and policy influencers.
Wolfgang Munchau
2026.04.19
75% relevant
The article critiques single‑point forecasts and emphasizes assessing ranges and extreme scenarios (citing Rumsfeld, Knight, and the IMF's scenario work); that links directly to the existing idea that forecasting should be evaluated, calibrated and held to postmortem accountability rather than treated as authoritative ex ante.
Eli McKown-Dawson
2026.01.14
80% relevant
The article’s mini report‑card and discussion of polling misses embodies the need for transparent postmortems (who erred, by how much, and why). It shows pollsters and forecasters should publish robustness maps and calibration data after salient misses.
Nate Silver
2026.01.14
85% relevant
Silver’s Predictive Plus‑Minus is an operationalization of ex‑ante credibility for pollsters and functions like the recommended preregistration/postmortem practice: it quantifies past performance and provides a provenance‑aware prediction of future error that forecasters and reporters can use to calibrate trust.
Matthew Yglesias
2026.01.05
100% relevant
Matthew Yglesias explicitly lists his 2025 predictions, reports he assigned ~80% credence to them, and then publicly tallies outcomes and lessons — exactly the behavior a standardized postmortem regime would capture.