paper
Contagion Networks: Evaluator Bias Propagation in Multi-Agent LLM Systems
paperactiveprovisional
contagion-networks-evaluator-bias-propagation-in-multi-agent-llm-systems-148dfdcf·1 events·first seen 47h agoAliases: Contagion Networks: Evaluator Bias Propagation in Multi-Agent LLM Systems
Co-occurring entities
More like this (12)
multi-level agent evaluationMulti-Component LLM AgentMulti-Turn Evaluation of Deep Research Agents Under Process-Level FeedbackWill the Agent Recuse Itself? Measuring LLM-Agent Compliance with In-Band Access-Deny Signalsmulti-agent cooperative frameworkRole-Agent: Bootstrapping LLM Agents via Dual-Role EvolutionAgentopia: Long-Term Life Simulation and Learning in Agent SocietiesLeadership as Coordination Control: Behavioral Signatures and the Recovery-Advantage Boundary in Multi-Agent LLM TeamsDistributed Agent AttackReward Modeling for Multi-Agent OrchestrationStylisticBias: A Few Human Visual Cues Drive Most Social Biases in MLLMsSkill-RM: Unifying Heterogeneous Evaluation Criteria via Agent Skill
Recent events (1)
Contagion Networks: formal framework for measuring evaluator bias propagation in multi-agent LLM systems
A new arXiv preprint introduces Contagion Networks, a formal framework for quantifying how systematic evaluation biases spread across interacting LLM agents in multi-agent systems. Using a controlled 3-agent experiment with DeepSeek-chat, the authors measure a Cross-Agent Contagion Matrix and find that homogeneous-model agents produce contagion coefficients 3-5x weaker than cross-model settings. A key practical finding is that increasing evaluator committee size from k=1 to k=3 reduces effective contagion by 72.4%, offering a concrete mitigation strategy. The authors release an open-source experimental framework alongside the paper.