Superforecasters vs. explanation: debate over whether predictions need models
Sam_kuyp · x · 2026-09-11
A debate on the methodology of superforecasting:
- Samkuyp's position: forecasters may be predictive, but predictions must rest on models and explanations. Why did Evan predict 10% rather than 1% or 50%? Past successes don't answer that—only a model can.
- davidmanheim's rebuttal: science succeeded before Popper explained why; ignoring demonstrably effective forecasting because you don't understand the mechanism is predictably worse. He compares the objection to rejecting a cake recipe as delicious-adjacent merely because its tastiness can't be justified.
- Sam draws an analogy to induction: science's success doesn't prove induction is how science works, and the same applies to superforecasters.
The core dispute: must well-calibrated forecasts be grounded in explainable models, or is empirical effectiveness justification enough?
Related event: Debate Erupts Over Whether Superforecasters Need Explicit Models(2 posts)→
More from AGI Musings
- Blogger predicts 2027: every major AI lab will match OpenAI's superintelligence — Dr_Singularity · 2026-09-11
- Four-Color Theorem Gets a Rare New Proof, Revisiting Its Controversial 1970s Computer-Assisted Solution — soumitrashukla9 · 2026-09-11
- Debating the '10% extinction chance' framing: attention now, credibility loss later — paulnovosad · 2026-09-11
- Karl Friston interviews Mark Solms on psychoanalysis, active inference and consciousness — MacrinePhD · 2026-09-11
- Why So Many AI Researchers Think the Machines Could Kill Everyone — connoraxiotes · 2026-09-11
- OpenRouter agents now out-consume humans as AI usage arrives in three waves — AccBalanced · 2026-09-11