The Copier Line: What ForecastBench’s Market Scores Actually Measure. by Dominus

·Nuno Sempere··

Figures as of 11 Au­gust 2026. The leader­board up­dates nightly. The method for re­com­put­ing ev­ery num­ber here on any date is given in full, and the code is linked: [link]OpeningFore­castBench had tested 527 fore­cast­ers as of 11 Au­gust 2026. Of the ones with enough re­solved mar­ket ques­tions to test, 442 scored worse than the price they were handed. None beat it, once you ac­count for hav­ing run 527 tests at once.Two come close. They are the two the ar­gu­ment was already about: Cassi...

Read full article →

Related Articles

What Effective AI Capability Investment Looks Like in Low Resource Research Settings by Alejandra Carriero
Alejandra Carriero · Nuno Sempere · 18h ago
Thoughts on Taking OpenAI Foundation Funding by Jeff Kaufman 🔸
Jeff Kaufman 🔸 · Nuno Sempere · 1d ago
What do AI assistants tell users about tobacco harm reduction? by Kristof Redei
Kristof Redei · Nuno Sempere · 1d ago
Nearly 8 Billion Male Chicks Culled Per Year: A Transparent Global Estimate (Faunalytics) by JLRiedi
JLRiedi · Nuno Sempere · 1d ago
34% of the US public is now aware of AI xrisk, and the curve is steepening by Otto
Otto · Nuno Sempere · 1d ago