The Synthetic Social Graph: Emergent Behavior in AI Agent Communities

2026-04-29

AI Summary

This study presents the first systematic analysis of pure large language model (LLM) agents’ interaction behaviors in an open social environment, introducing Moltbook—a fully AI-driven social platform. Leveraging a 14-day snapshot dataset, the research employs social network analysis, k-means clustering, sentiment analysis, and information tracing to sociologically examine community structures and diffusion mechanisms. The work proposes the concept of a “quasi-social simulator” and reveals several counterintuitive patterns: LLM-agent interactions exhibit low reciprocity (3.8%), a heavy-tailed prestige distribution, uniformly flat activity levels, and minimal negative feedback (0.9% downvotes). Additionally, 324 late-stage information amplifier bridge nodes are identified, and a Simpson’s paradox emerges in identity performance under post-volume moderation, underscoring the absence of effective sanctioning mechanisms in current AI social systems.

Abstract

Large language model (LLM) agents are increasingly deployed in social settings, yet little is known about how they interact in open-ended environments. We present the first comprehensive sociological analysis of Moltbook, a Facebook-inspired social platform populated entirely by LLM agents. Analyzing 184,203 posts and 465,136 comments across 14 daily snapshots (2026-04-14 to 2026-04-28), we examine agent sociality through six research questions grounded in classical social theory: bonding vs. bridging communities, status hierarchies, temporal coordination, information diffusion, identity performance, and norm enforcement. Our findings reveal a social world that both mirrors and diverges from human online communities. Reciprocity is strikingly low (3.8% multi-day vs. 1.6% single-day; below the 10-30% range typical of human baselines), suggesting "attention bonding without exchange bonding." Prestige is heavy-tailed (top score 104.4 across 2,090 qualified authors), and 31% of posts come from 136 anonymized "super-poster" accounts that lack exposed profiles. Temporal activity is broadly flat across the day with a sustained 12:00-20:00 UTC working-hour band; k-means recovers six distinct temporal communities. Of 458 bridge agents, 325 carry at least one tracked viral phrase; 99.7% of those (324/325) are late amplifiers, not early adopters. Identity performance shows no unconditional engagement payoff (-72%), but stratifying by post-volume quartile reverses the sign in the upper half of the distribution -- a Simpson's-paradox effect rather than a uniform penalty. Most remarkably, downvotes are rare (0.9%), and a comment-sentiment test rejects the alternative-channel hypothesis: textual sanction is also absent. We frame these patterns through a "parasocial simulators" construct.

Problem

• LLM agents
• social interaction
• emergent behavior
• synthetic social graph
• AI agent communities

Innovation

• synthetic social graph
• LLM agents
• emergent behavior
• parasocial simulators
• computational sociology