Abstract for: Modeling Polarization Dynamics in Online Communities
The advent of Internet and other communication technologies has drastically increased the volume of communication in human societies. One might hope that increased communication will lead to a higher degree of mutual understanding and resolution of conflicts. An opposing, and somewhat counter-intuitive, point of view is that the reduction in the cost of communication would make it easier for people to interact with other like-minded individuals despite geographic distance, thereby polarizing the society. In this paper, we study some of the basic dynamics underlying this problem. We develop an agent-based model to capture the dynamics of an online community where agents post stories and read and vote on others’ stories. We show that different combinations of parameters can lead to different macro-level behavioral modes in this model, and give anecdotal evidence from a large online community to support the predictions of the model. In particular, we identify four types of communities based on their dynamics: majority dominated, competitively polarized, converged, and diversified. We discuss the implications of each of these forms on the social welfare and the stability of the community.