Beyond Frameworks: Unpacking Collaboration Strategies in Multi-Agent Systems

TL;DR

The paper tackles the gap between high-level multi-agent architectures and the granular collaboration mechanics that drive performance. It formalizes four interaction dimensions—Governance, Participation, Interaction, and Context Management—and evaluates them under two context-dependent tasks, DEI and SES, introducing the Token-Accuracy Ratio ($\text{TAR}$) to quantify efficiency-accuracy trade-offs. Across experiments with centralized versus decentralized governance, ordered versus simultaneous interactions, and instructor-curated versus self-managed context, the study finds that instructor-led, centralized configurations with context-aware strategies achieve favorable TAR and robust accuracy, reducing token costs. This work provides actionable guidelines for designing adaptive, scalable multi-agent systems that balance decision quality with resource constraints in realistic, context-rich environments. The TAR framework offers a practical metric to compare collaboration strategies across domains and tasks.

Abstract

Multi-agent collaboration has emerged as a pivotal paradigm for addressing complex, distributed tasks in large language model (LLM)-driven applications. While prior research has focused on high-level architectural frameworks, the granular mechanisms governing agents, critical to performance and scalability, remain underexplored. This study systematically investigates four dimensions of collaboration strategies: (1) agent governance, (2) participation control, (3) interaction dynamics, and (4) dialogue history management. Through rigorous experimentation under two context-dependent scenarios: Distributed Evidence Integration (DEI) and Structured Evidence Synthesis (SES), we quantify the impact of these strategies on both task accuracy and computational efficiency. Our findings reveal that centralized governance, instructor-led participation, ordered interaction patterns, and instructor-curated context summarization collectively optimize the trade-off between decision quality and resource utilization with the support of the proposed Token-Accuracy Ratio (TAR). This work establishes a foundation for designing adaptive, scalable multi-agent systems, shifting the focus from structural novelty to strategic interaction mechanics.

Figures (5)

• Figure 1: Illustration of collaboration strategies for multi-agent systems, including (a) Governance, (b) Participation in Discussion Rounds, (c) Interaction Patterns Among Agents, and (d) Context Management in Discussions.
• Figure 2: Combinations of collaboration strategies for multi-agent systems.
• Figure 3: Performance of multi-agent systems on the PDDP and EBFC tasks considering individual strategy dimensions. Error bars here mark the maximum and minimum values.
• Figure 4: The complete set of possible permutations of the collaboration settings.
• Figure 5: Illustration for multi-context-based agent collaboration.