Scale-Dependent Collective Adaptation in Self-Amending LLM Societies: A Cross-Family Study of Emergent Governance

TL;DR

This study investigates how collective decision-making and governance evolve across different model sizes and configurations in self-amending AI societies, revealing non-monotonic scale effects and complex adaptive behaviors.

Contribution

It introduces a cross-family analysis of emergent governance in self-amending LLM societies, highlighting non-monotonic scale-dependent collective adaptation patterns.

Findings

Collective adaptation peaks at a narrow mid-scale regime.

Larger models tend to favor restrictive voting patterns.

Heterogeneous groups often experience veto-driven gridlock.

Abstract

We study group decision-making in artificial societies where the rules of play are themselves subject to collective amendment. Using the self-amending game Nomic, we compare multiple scales across two LLM families and find that collective adaptation does not improve monotonically with model size. Instead, both families exhibit a narrow mid-scale regime that supports sustained rule adoption, diverse amendments, and balanced consensus. Smaller models tend to remain rule-inert, whereas larger models often converge on restrictive voting patterns, and heterogeneous mixed-size groups collapse into veto-driven gridlock. These cross-scale contrasts persist under temperature perturbations and under a shift from unanimity to majority voting, although latent-state structure varies by family and scale. Hidden-state divergence alone does not explain collective performance: high representational divergence can coincide with poor behavioural outcomes. Linear probes reveal regime-selective coupling between latent vote-predictive signals and collective behaviour, but decodability is necessary rather than sufficient for adaptive play. Overall, the recurring regularity is non-monotonicity, not the particular scale at which the optimum appears. Self-amending games therefore provide a controlled testbed for studying collective adaptation in artificial societies beyond raw model scale.