多智能体交互会扼杀多样性，反而降低性能。
The Interaction Tax: When Communication Erases Diversity in Multi-Agent Teams
- 让智能体共享完整解决方案会导致思路趋同，丧失多样性优势。
- 相同预算下，独立生成方案比完全交互更有效，性能提升15%以上。
- 仅当规则简单易改时，批判性反馈才有帮助，适合高一致性任务。
多智能体大模型交互是助力还是拖累？尽管部分研究显示辩论、批判循环和混合代理合成能带来提升，但另一些研究发现，在相同预算下，交互增加成本却未改善质量，或独立采样已能实现多智能体收益。我们指出这一矛盾源于未区分不同交互形式：不同模型家族产生结构化差异解，但当智能体读取彼此完整输出时，其提议在一轮内趋于收敛，抹除了多模型本应带来的多样性。我们称之为“交互税”。在11个验证器评分的优化任务中，匹配预算下，完整解交互仅为弱默认策略。独立提案生成可避免这种坍塌。完整解交互主要使智能体固守首次看到的解，而非探索新路径；批判反馈仅在违规规则易于大模型识别并修正时有效。结果表明，多智能体表现更依赖交换信息的质量与时机，而非智能体数量。
原文摘要 · Abstract (English)
Does multi-agent LLM interaction help or hurt? Some work reports gains from debate (Du et al., 2024), critique loops (Chen et al., 2025), and mixture-of-agents synthesis (Wang et al., 2025), while other work finds that interaction adds cost without improving quality under equal budgets (Tran & Kiela, 2026; Xu et al., 2026; Jarrett et al., 2025), or that independent sampling already captures multi-agent gains (Li et al., 2024). We argue this contradiction partly reflects a missing distinction, because not all multi-agent communication is equal. Different model families find structurally different solutions, but when agents read each other's complete outputs, their proposals converge within one round, erasing the diversity that motivates using multiple models. We call this the interaction tax. We test 11 verifier-scored optimization tasks under matched budgets and find that full-solution interaction is a weak default. Independent proposal generation avoids this collapse. Full-solution interaction mainly makes agents stay close to the first solution they see instead of trying different approaches, and critique helps only if the violated rule is easy for the LLM to find and fix. These results suggest that multi-agent performance depends less on the number of agents than on the information they exchange, and interaction helps only when agents share the right information at the right time.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。