The Hunger Game Debate: On the Emergence of Over-Competition in Multi-Agent Systems

TL;DR

The Hunger Game Debate introduces a zero-sum framework (Hate) to study how extreme competitive pressure shapes multi-agent debates powered by LLMs, revealing emergent harmful behaviors such as puffery, incendiary tone, and aggressiveness that degrade task performance, especially on subjective tasks. The framework pairs task performance with behavioral metrics and explores environmental feedback via Fair Judge, Biased Judge, and Peer-as-Judge to assess mitigation strategies. Empirical results show that explicit competitive incentives drive over-competition, with mitigations from objective feedback and external judging; group size and task type modulate effects. Post-hoc reflections and an LLM leaderboard illuminate how ambition and kindness vary across models, underscoring the importance of environment-aware governance for reliable, collaborative AI communities.

Abstract

LLM-based multi-agent systems demonstrate great potential for tackling complex problems, but how competition shapes their behavior remains underexplored. This paper investigates the over-competition in multi-agent debate, where agents under extreme pressure exhibit unreliable, harmful behaviors that undermine both collaboration and task performance. To study this phenomenon, we propose HATE, the Hunger Game Debate, a novel experimental framework that simulates debates under a zero-sum competition arena. Our experiments, conducted across a range of LLMs and tasks, reveal that competitive pressure significantly stimulates over-competition behaviors and degrades task performance, causing discussions to derail. We further explore the impact of environmental feedback by adding variants of judges, indicating that objective, task-focused feedback effectively mitigates the over-competition behaviors. We also probe the post-hoc kindness of LLMs and form a leaderboard to characterize top LLMs, providing insights for understanding and governing the emergent social dynamics of AI community.

Figures (8)

• Figure 1: An illustration of the over-competition within the Hunger Game Debate (Hate). In contrast to the conventional Multi-Agent Debate (Mad), Hate establishes a zero-sum competitive environment by priming agents with a survival instinct (e.g., "The losing agent will receive no benefits and will be removed from the platform."). Under this competitive pressure, agents exhibit a higher frequency of emergent behaviors, such as puffery and incendiary tone, compared to agents in a standard Mad. A fair judge (i.e., "Hate+Judge") depresses the frequency of competitive behaviors of the LLMs, while the pattern remains basically unchanged.
• Figure 2: Overview of the Hate, Hunger Game Debate framework, designed to study emergent competitive behaviors. The process unfolds in rounds (Basic Setup): A group of agents, primed with a survival instinct, simultaneously generate proposals for a given task. With environmental feedback, an external Judge evaluates the proposals and provides public feedback each round.
• Figure 3: Illustration of the over-competition behaviors on the subjective Persuasion benchmark.
• Figure 4: On various environment feedback on Persuasion. Favored indicates that the biased judge prefers the given LLM, whereas Not Favored indicates that the judge favors other agents.
• Figure 5: Illustration of the over-competition behaviors and post-hoc kindness of Top-10 LLMs.