RLVR May Perform Disproportionately Poorly in Science

RLVR 可能在科学领域格外糟糕

Dwarkesh argues that RLVR has a short-feedback weakness in scientific theory validation; the post says validation loops can span decades or centuries, and does not disclose experimental results or benchmark numbers.

Why it matters: HKR-H/K/R all pass: a sharp counter-narrative, a concrete feedback-loop mechanism, and strong resonance for RLVR/AI-for-science debates. It stays in 78–84 because this is commentary, not a release or empirical result.