OPINION
Technical arguments, their assumptions and supporting evidence.
Authors argue that more computation outside text could weaken chain-of-thought oversight.The revised conclusion distinguishes supported claims from broader preferences.
Original publication ↗Their central position is that agent evaluations should control costs, distinguish model research from downstream procurement, and test generalization with meaningful held-out tasks.
Original publication ↗Aggregate scores conceal weak items and construct mismatches. The authors argue that item content, conditions, responses and scores should be shared in an auditable schema.
Original publication ↗Their proposed workflow starts with the claim and asks which kinds of validity evidence are missing before extending a benchmark result to a practical decision.
Original publication ↗
NUWA