Opinion Dynamics Optimization Through Noncooperative Differential Games

TL;DR

This paper models opinion formation in social networks as a noncooperative differential game, deriving equilibrium strategies and analyzing opinion trajectories, showing opinions tend toward the network average without necessarily reaching consensus.

Contribution

It introduces a differential game framework for opinion dynamics on social networks, deriving Nash equilibria and applying them to real-world network data.

Findings

Opinions tend to move toward the network average.

Consensus is not guaranteed in non-stubborn networks.

Equilibrium strategies can be computed using Pontryagin's principle.

Abstract

In this paper, I study optimizing the opinion formation of a social network of a population of individuals on a graph whose opinion evolves according to the Hegselmann-Krause model for opinion dynamics. I propose an optimization problem based on a differential game for a population of individuals who are not stubborn. The objective of each individual is to seek an optimal control policy for her own opinion evolution by optimizing a personal performance index. The Nash equilibrium actions and the associated opinion trajectory with the equilibrium actions are derived for the opinion optimization model using Pontryagin's principle. The game strategies were executed on the well-known Zachary's Karate Club social network. The resulting opinion trajectories associated with the game strategies showed that in non-stubborn Zachary's network, the opinions moved toward the average opinion of the network, but a consensus of final opinions did not necessarily emerge.