网络拓扑无关紧要：LLM群体中经典共识的失效
原标题：Network Topology Does Not Matter: The Breakdown of Classical Consensus in LLM Populations
AI 摘要
该论文研究多智能体语言模型网络的集体行为，发现经典分布式共识理论的三项结构假设（更新为凸组合、群体均值守恒、权重与内容无关）在LLM群体中全部失效。作者通过token级概率确定性地提取连续置信度，在8个指令微调模型、73,080次受控交互上测量LLM更新算子，发现LLM会更新到观测意见凸包之外并表现出负权重（主动排斥）。将拟合算子扩展到10,000个智能体的模拟后，代数连通性几乎无法解释最终结果方差，群体动态由模型内部更新偏差和消息措辞主导，其中4个模型（含7B模型）的群体无论网络拓扑如何都会自信地认同虚假主张。
正文节选
Network Topology Does Not Matter: The Breakdown of Classical Consensus in LLM Populations Abstract As multi-agent language model networks become widely deployed, predicting their collective behavior is a critical challenge. Classical distributed consensus theory relies on network topology to predict convergence, structurally assuming that agent updates are convex combinations, population means are conserved, and weights are content-independent. In this work, we demonstrate that these foundationa