On the Poisson Follower Model

• Published in: arXiv.org 2024-09
• Main Authors: Dragovic, Natasa, Baccelli, Francois
• Format: Article
• Language: English
• Subjects: Discrete event systems; Dynamics; Euclidean geometry; Euclidean space; Leadership; Social networks
• ISSN: 2331-8422

We introduce a stochastic geometry dynamics inspired by opinion dynamics, which captures the essence of modern asymmetric social networks, with leaders and followers. Opinions are represented by points in the Euclidean space and the leader of an agent is the agent with the closest opinion. In this dynamics, each follower updates its opinion by halving the distance to its leader. We show that this simple dynamics and its iterates exhibit several interesting purely geometric phenomena pertaining to the evolution of leadership and of opinion clusters, which are reminiscent of those observed in social network. We also show that when the initial opinions are randomly distributed as a stationary Poisson point process, the probability of occurrence of each of these phenomena can be represented by an integral geometry formula involving semi-agebraic domains. Furthermore, we prove this property for step 0 and step 1 of the dynamics using percolation techniques. We f inally analyze numerically the limiting behavior of this follower dynamics. In the Poisson case, the agents fall into two categories, that of ultimate followers, which never stop updating their opinions, and that of ultimate leaders, which adopt a fixed opinion in a finite time. All our findings are complemented by spatial discrete event simulations.