200 Chess Games Test Suggests In-Context Learning Falls Short

A 200-game chess experiment found in-context learning largely ineffective, comparing Astra and Opus head-to-head. Discussion notes chess's simple transition functions favor exhaustive search like Alpha-Beta pruning, while math's semantic transitions may explain why CoT outperforms AlphaProof.

2026-10-16 ~ 2026-10-16 · 2 related posts

• 200 chess games suggest in-context learning doesn't work yet, experimenter says — akbirthko · 2026-10-16
• Chess has simple transitions, math doesn't: why CoT beats AlphaProof-style search — akbirthko · 2026-10-16