Bounded confidence dynamics generates opinion cascades on growing scale-free networks

Abstract

We study the pairwise bounded confidence model on scale-free networks where new agents regularly arrive over time. The probability that arriving agents form links to preexisting ones depends on both agent degree and opinion proximity. In parameter value ranges where both factors impact the link choice, a new phenomenon is observed. Minor clusters continuously form on the periphery of the opinion space and remain stable for a time, before suddenly merging with a major cluster in the network. We label these processes as "opinion cascades", and analyse their origin and behavior. They are triggered by the arrival of agents acting as "bridges" between the previously disconnected minor and major clusters. Lastly, we propose theoretical approximations to describe the varying shapes and merging behavior of opinion cascades under different conditions.