无需说服的一致：单轮辩论抹去了验证所需的异议

中文摘要 AI 辅助

辩论小组可以在不变得更正确的情况下达成一致。这对下游安全机制是危险的：被替换的验证选票只能改变微弱优势的投票，而更丰富的仲裁者会失去异议这一自然定位信号。我们表明，一轮辩论就能抹去该资源，且无需说服。通过追踪一个由7名法官组成的异质小组，在600个代码正确性候选上经历一轮盲评和三轮辩论，固定队列上的一致率从第1轮的39.5%跃升至95.2%（占总崩溃的93.1%），而准确率变动不到1个百分点，且96.3%的判决翻转跟随显示的同行多数。基于执行的验证选票在辩论前的2,037个候选替换实例中纠正了8个，但在后续每一轮中纠正为零；到第3轮，每个错误决定都是一致的，抹去了曾标记小组三分之二错误的异议。相同队列的对照组解释了原因：无同行重新考虑重现了79.6%的崩溃，无推理的真实标签重现了91.5%，而随机标签将翻转引向其显示的任何内容；完整辩论条件在仅标签基础上增加了4.3个百分点（聚类95%置信区间0.7--8.1）。单轮崩溃在另外两次真实运行和两个假标签种子中重现，在小组规模3--7下保持，并出现在MATH-500中。解析失败集中在有争议的候选上（p<0.001），使得流失不可忽略。设计含义是操作性的：在任何第二轮评估或同行暴露之前进行验证，且绝不将辩论后的一致视为可靠性的独立证据。

英文摘要

A debate panel can become unanimous without becoming more correct. This is dangerous for downstream safeguards: a substituted verification ballot can change only narrow-margin votes, while richer arbiters lose disagreement as a natural targeting signal. We show that one debate round can erase that resource without requiring persuasion. Tracking a heterogeneous 7-judge panel through a blind round and three debate rounds on 600 code-correctness candidates, unanimity on a fixed cohort jumps from 39.5\% to 95.2\% in round 1 (93.1\% of the total collapse), while accuracy moves by less than one point and 96.3\% of verdict flips follow the displayed peer majority. An execution-based verification ballot corrects 8 of 2,037 pre-debate candidate-substitution instances but changes zero in every later round; by round 3 every wrong decision is unanimous, erasing dissent that had flagged two-thirds of the panel's errors. Identical-cohort controls explain why: no-peer reconsideration reproduces 79.6\% of the collapse, real labels without reasoning reproduce 91.5\%, and random labels steer flips toward whatever they display; the full-debate condition adds 4.3 percentage points over labels only (clustered 95\% CI 0.7--8.1). The one-round collapse reproduces in two additional real runs and two fake-label seeds, remains under panel sizes 3--7, and appears in MATH-500. Parse failures concentrate on contested candidates ($p<0.001$), making attrition non-ignorable. The design implication is operational: verify before any second-pass evaluation or peer exposure, and never treat post-debate unanimity as independent evidence of reliability.

发表机构

• Zhejiang University（浙江大学）

机构由 AI 辅助整理，请以论文原文为准。