Beyond Final Accuracy: Auditing Communication in LLM Multi-Agent Systems

arXiv 2610.01042 ↗First posted 1 Oct 2026cs.AI cs.CLcited by 0

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks also fits: Social simulation and multi-agent debate

Abstract (arXiv)

Multi-agent communication aims to help agents benefit from one another's information. Yet improvements in system performance leave a fundamental ambiguity: do they reflect effective communication, a favorable agent architecture, or simply additional reasoning? Because communication methods are commonly evaluated within the systems they were designed for, these factors are difficult to disentangle. Final accuracy further merges corrected errors and corrupted answers into a single outcome, obscuring how communication changes decisions. We introduce Independent--Communicate--Revise (ICR), a controlled framework that evaluates communication as answer revision following independent reasoning. ICR fixes initial reasoning trajectories, measures correction and preservation conditional on both agents' initial correctness, and uses a no-message revision control to quantify gains beyond additional reasoning. Across four reasoning benchmarks, our audit of textual and latent communication reveals that similar aggregate accuracy can conceal substantially different revision behaviors. Compared with transmitting answers alone, full reasoning increases correction while reducing preservation on all four benchmarks, so richer messages amplify beneficial and harmful influence alike. Receiver-policy comparisons on MedQA and GPQA-D further show that a structured verification policy shifts every channel toward greater preservation and lower correction, while its effect on selectivity varies across channels and tasks. These findings challenge treating communication quality as an intrinsic property of a channel. ICR therefore recenters evaluation on selective revision, providing a unified framework for examining how message content and receiver policies jointly produce benefits and harms.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 42 of its 44 references that are in the map. First posted on arXiv 1 Oct 2026.

Builds on

42 of 44 references.

Closest ancestors · 9

Same direction, newest first

Other references · 33

2021Training Verifiers to Solve Math Word Problemscited by 9,743
2025Qwen3 Technical Reportcited by 8,195
2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,710
2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,748
2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,906
2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
2023Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generationcited by 1,332
2020What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Examscited by 1,324
2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 995

21 more

2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
2023LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compressioncited by 443
2024LLMLingua-2: Data Distillation for Efficient and Faithful Task-Agnostic Prompt Compressioncited by 343
2023LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Modelscited by 279
2023Examining Inter-Consistency of Large Language Models Collaboration: An In-depth Analysis via Debatecited by 94
2025Peacemaker or Troublemaker: How Sycophancy Shapes Multi-Agent Debatecited by 29
2025Free-MAD: Consensus-Free Multi-Agent Debatecited by 27
2025When Identity Skews Debate: Anonymization for Bias-Reduced Multi-Agent Reasoningcited by 20
2025Learning to Deliberate: Meta-policy Collaboration for Agentic LLMs with Multi-agent Reinforcement Learningcited by 18
2026Auditing Multi-Agent LLM Reasoning Trees Outperforms Majority Vote and LLM-as-Judgecited by 16
2025VeriMoA: A Mixture-of-Agents Framework for Spec-to-HDL Generationcited by 15
2026Self-Compression of Chain-of-Thought via Multi-Agent Reinforcement Learningcited by 13
2026POET: Power-Oriented Evolutionary Tuning for LLM-Based RTL PPA Optimizationcited by 11
2026FORTIS: Benchmarking Over-Privilege in Agent Skillscited by 11
2026Adaptive Collaboration with Humans: Metacognitive Policy Optimization for Multi-Agent LLMs with Continual Learningcited by 9
2026OASES: Outcome-Aligned Search-Evaluation Co-Training for Agentic Searchcited by 7
2026Tournament-GRPO: Group-Wise Tournament Rewards for Reinforcement Learning in Open-Ended Long-Form Generationcited by 7
2026UnityMAS-O: A General RL Optimization Framework for LLM-Based Multi-Agent Systemscited by 6
2026RaMem: Contextual Reinstatement for Long-term Agentic Memorycited by 5
2026Minority Sentinel: When to Overturn Majority Voting in Multi-Agent LLM Debatescited by 4
2026ReM-MoA: Reasoning Memory Sustains Mixture-of-Agents Scalingcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way