RoundTable: Investigating Group Decision-Making Mechanism in Multi-Agent Collaboration

2026-04-11

AI Summary

This study investigates the impact of decentralized voting mechanisms on collaboration quality and efficiency in multi-agent systems (MAS). Through controlled experiments and multi-round decision modeling, it systematically evaluates diverse voting rules, revealing that majority voting often induces inefficient collaboration, while unanimity voting severely degrades initial performance. The work first identifies a critical pattern of message redundancy explosion across rounds—message length increases by 84% and inter-round similarity reaches 90%. Leveraging this insight, it proposes a language-driven dynamic early-stopping mechanism: using large language models to assess real-time consensus states, enabling halving of collaboration rounds (−50%) while achieving near-Oracle performance (+13%). Key contributions include: (1) quantifying the strong negative correlation between communication redundancy and decision efficiency; (2) identifying the optimal voting mechanism for MAS collaboration; and (3) introducing the first language-model-based early-stopping paradigm tailored for text-based MAS coordination.

Technology Category

Multiagent Systems: Mechanism DesignGame Theory and Economic Paradigms: Social Choice / VotingIntelligent Robots: Multi-Robot Systems

Application Category

Economics, Online Markets and Human Computation: LLM based quality controls for crowd workResponsible Web: Machine-in-the-loop, human agency and autonomyWeb Mining and Content Analysis: Content-based information diffusion

Abstract

Effective group decision-making is critical in Multi-Agent Systems (MAS). Yet, how different mechanisms for reaching consensus impact collaboration quality and efficiency remains understudied. We conduct a systematic study on group decision-making mechanisms in a decentralized setting. Through controlled experiments, we analyze how different voting rules affect decision quality and efficiency in a multi-round collaboration. Results reveal that majority voting often cause inefficient collaboration due to its strict acceptance criteria. At the extreme, unanimous voting gives 87% lower initial performance than the best-performing method. Our qualitative analysis of cross-agent communication shows that messages become longer and more repetitive over time: while message length increases by 84%, similarity to the previous round increases to 90%. Based on these insights, language-based early stopping methods make the performance 13% closer to oracle while reducing rounds by 50%. Our findings highlight the crucial role of group decision-making in optimizing MAS collaboration.

Problem

• Study impact of consensus mechanisms on multi-agent collaboration quality
• Analyze voting rules' effect on decision quality and efficiency
• Evaluate language-based early stopping to optimize collaboration performance

Innovation

• Analyzing voting rules' impact on decision quality
• Language-based early stopping improves efficiency
• Majority voting causes inefficient collaboration