Personality mix shapes social network polarization and collective smarts
Diverse Minds, Divided Networks? Personality Composition, Polarization, and Collective Intelligence in LLM-Based Social Simulations
Computation and Language
Summary
This paper looks at how the personalities of people in online groups affect how much they disagree and how well they work together. The authors used computer simulations with language model agents that mimic human traits like openness and agreeableness. They found that groups with more diverse personalities have opinions spread out but less likely to split into hostile sides, while similar groups tend to form echo chambers. Importantly, having less polarization did not mean the group made worse decisions, challenging the idea that reducing conflict harms group intelligence.
What this means in practice
- •For online platform designers: Design social spaces with diverse personality compositions to reduce group segregation and prevent echo chambers without harming collective decision-making.
- •For team managers: Structure teams with personality diversity to foster broader viewpoints while maintaining effective collective problem solving.
Authors
Raad Bin Tareaf
Abstract
Simulated societies of large language model agents are used to study online polarization, and separately to study collective intelligence, but the two are rarely measured in the same system. It is therefore difficult to say whether a society's personality composition shapes both, or whether reducing polarization costs collective competence. We present TraitMix, an experimental design in which the Big Five composition of a simulated social network, both trait levels and trait heterogeneity, is a controlled experimental variable, and in which polarization and collective performance are measured in the same runs. Across 991 simulations of hundred-agent societies, spanning six contested topics and six language models, trait heterogeneity has the largest measured effects, acting in opposite directions on two faces of polarization: varied societies hold more dispersed opinions while being less segregated into camps, so homogeneous societies are not moderate but consensual echo chambers. Trait effects are not additive, as Agreeableness determines the sign of Openness, an interaction that replicates across models although the primary model's estimate is influence-driven. Contrary to the trade-off the study was designed to measure, no polarization measure predicts poorer collective performance, and cross-cutting interaction is the only one of four whose association with collective accuracy survives partialling on the aggregation identity. We report ablations removing two potential measurement circularities, an induction gate applied to every model, and the measures that failed them.