EvoCorps: An Evolutionary Multi-Agent Framework for Depolarizing Online Discourse

Abstract: Polarization in online discourse erodes social trust and accelerates misinformation, yet technical responses remain largely diagnostic and post-hoc. Current governance approaches suffer from inherent latency and static policies, struggling to counter coordinated adversarial amplification that evolves in real-time. We present EvoCorps, an evolutionary multi-agent framework for proactive depolarization. EvoCorps frames discourse governance as a dynamic social game and coordinates roles for monitoring, planning, grounded generation, and multi-identity diffusion. A retrieval-augmented collective cognition core provides factual grounding and action--outcome memory, while closed-loop evolutionary learning adapts strategies as the environment and attackers change. We implement EvoCorps on the MOSAIC social-AI simulation platform for controlled evaluation in a multi-source news stream with adversarial injection and amplification. Across emotional polarization, viewpoint extremity, and argumentative rationality, EvoCorps improves discourse outcomes over an adversarial baseline, pointing to a practical path from detection and post-hoc mitigation to in-process, closed-loop intervention. The code is available at https://github.com/ln2146/EvoCorps.

• The paper employs a multi-agent framework that leverages evolutionary strategies and a Stackelberg–Mean-Field Control paradigm to proactively mitigate online polarization.
• The experimental results show reduced sentiment extremity and improved argumentative rationality compared to traditional post-hoc interventions.
• The framework’s real-time, evidence-driven interventions mark a shift from reactive to anticipatory governance in managing digital discourse.

EvoCorps: An Evolutionary Multi-Agent Framework for Depolarizing Online Discourse

Introduction

"EvoCorps: An Evolutionary Multi-Agent Framework for Depolarizing Online Discourse" presents an innovative architecture aimed at proactively managing polarization in online environments. With the rise of coordinated adversarial activities jeopardizing social harmony, traditional governance strategies have proven inadequate, being reactive and lacking adaptability. EvoCorps is designed to transition from the diagnostic approaches to active intervention, framing discourse governance as a dynamic social game with strategic agents controlling the narrative flow.

Figure 1: Online interaction evolves from normal user communication to malicious agents spreading harmful information, leading to amplified negative emotions and polarization.

Framework and Methodology

EvoCorps employs the Stackelberg--Mean-Field Control paradigm to model discourse governance as a dynamic game. This framework, implemented on the MOSAIC simulation platform, consists of several agent roles—Analyst, Strategist, Leader, and Amplifier—each tailored for specific intervention tasks.

Agent Dynamics: The Analyst agent evaluates the discourse environment, identifying sentiment and extremity metrics critical for decision-making. Upon trigger events suggesting heightened polarization, the Strategist formulates intervention strategies integrating historical data, aimed at balancing narratives.

Execution Cycle: The Leader agent, equipped with a retrieval-augmented cognition core, synthesizes evidence-based responses supported by persuasive argumentation. Amplifiers then disseminate these outputs across the network, ensuring adoption through engagement amplification strategies.

Figure 2: Workflow of the closed-loop evolutionary system.

Experimental Results

The EvoCorps framework was evaluated under conditions simulating adversarial amplification and demonstrated improvements in measures of emotional polarization and argumentative rationality when compared to traditional post-hoc interventions.

Figure 3: Sentiment trajectories over time under Case~1/2/3/4. The dashed line marks clarification at $t=5$. Case~2 continues to decline, Case~3 partially mitigates, and Case~4 declines more slowly and stabilizes relative to Case~2/3.

Metric Analysis: The experiments showed that EvoCorps moderated viewpoint extremity and enhanced sentiment stabilization, diverging from antagonist baselines shortly after corrective content introduction. Such proactive governance effectively prevents the entrenchment of adversarial narratives and hostility.

Figure 4: Reward trajectories across intervention rounds.

Implications and Future Directions

The deployment of EvoCorps marks a shift towards anticipatory governance in digital discourse, potentially influencing future AI-driven interventions. The integration of evolutionary learning within multi-agent systems showcases adaptability by retaining knowledge patterns offering strategic intervention while discarding ineffectiveness.

Practical implications include the transition from post-event analysis to real-time strategic adaptations that could redefine regulatory practices on platforms facing adversarial challenges. EvoCorps thus lays foundational groundwork for sophisticated AI systems fostering online harmony amidst complex socio-political landscapes.

Conclusion

EvoCorps, by leveraging evolutionary multi-agent strategies, enhances discourse quality through proactive intervention. Its design not only addresses latency inherent in conventional mechanisms but also demonstrates the potential of coordinated AI-driven strategies in achieving sustained discourse stability. Future research may explore its scalability and effectiveness within increasingly multimodal and complex online environments.