Breaking the Martingale Curse: Multi-Agent Debate via Asymmetric Cognitive Potential Energy

2026-03-06

AI Summary

Standard multi-agent debate often converges to erroneous consensus due to relevance errors and is fundamentally limited by the correctness ceiling of majority voting—a phenomenon known as the “martingale curse.” This work proposes AceMAD, a novel framework that reveals, for the first time, how truth-holding agents can anticipate collective misjudgments and establish an asymmetric cognitive potential gap. This gap transforms the debate dynamics from a random walk into a submartingale process with positive drift, thereby guiding convergence toward truth. AceMAD integrates peer prediction, strictly proper scoring rules, and a nonlinear aggregation mechanism to effectively quantify and harness cognitive potential to steer deliberation. Evaluated on challenging subsets of six benchmark tasks, AceMAD significantly outperforms existing baselines and successfully recovers sparse correct signals even when the initial majority opinion is incorrect.

Technology Category

• Multiagent Systems: Adversarial Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Multi-Agent Debate (MAD) has emerged as a promising paradigm for enhancing large language model reasoning. However, recent work reveals a limitation:standard MAD cannot improve belief correctness beyond majority voting; we refer to this as the Martingale Curse. This curse arises because correlated errors cause agents to converge toward erroneous consensus, where debate merely reinforces collective mistakes rather than filtering noise. We propose AceMAD, a framework that breaks the Martingale Curse by harnessing asymmetric cognitive potential energy to transform MAD from a random walk into a directed convergence process with positive drift. Through a peer-prediction mechanism, agents predict their peers'belief distributions, revealing asymmetric cognitive potential: truth-holders not only know the correct answer but also anticipate the crowd's misconceptions, while the hallucinating majority remains blind to their collective error. This asymmetry creates a potential energy gap that we quantify via strictly proper scoring rules. We prove this cognitive potential manifests as information-theoretic superiority and, under nonlinear aggregation, converts into submartingale drift toward truth, directly breaking the Martingale Curse. Experiments on challenging subsets across six benchmarks show AceMAD recovers sparse truth signals even when initial majorities are incorrect, substantially outperforming baseline methods.

Problem

• Multi-Agent Debate
• Martingale Curse
• correlated errors
• erroneous consensus
• belief correctness

Innovation

• Multi-Agent Debate
• Martingale Curse
• Asymmetric Cognitive Potential Energy
• Peer Prediction
• Submartingale Drift