Collective Intelligence in Dynamic Networks

Abstract

We revisit DeGroot learning to examine the robustness of social learning in dynamic networks -- networks that evolve randomly over time. Dynamics have double-edged effects depending on social structure: while they can foster consensus and boost collective intelligence in "sparse" networks, they can have adverse effects such as slowing down the speed of learning and causing long-run disagreement in "well-connected" networks. Collective intelligence arises in dynamic networks when average influence and trust remain balanced as society grows. We also find that the initial social structure of a dynamic network plays a central role in shaping long-run beliefs. We then propose a robust measure of homophily based on the likelihood of the worst network fragmentation.