Erratic Extremism causes Dynamic Consensus (a new model for one-dimensional opinion dynamics) (1809.06049v3)

Abstract: A society of agents, with ideological positions, or "opinions" measured by real values ranging from $-\infty$ (the "far left") to $+\infty$ (the "far right"), is considered. At fixed (unit) time intervals agents repeatedly reconsider and change their opinions if and only if they find themselves at the extremes of the range of ideological positions held by members of the society. Extremist agents are erratic: they become either more radical, and move away from the positions of other agents, with probability $\varepsilon$, or more moderate, and move towards the positions held by peers, with probability $(1 - \varepsilon)$. The change in the opinion of the extremists is one unit on the real line. We prove that the agent positions cluster in time, with all non-extremist agents located within a unit interval. However, the consensus opinion is dynamic. Due to the extremists' erratic behavior the clustered opinion set performs a "sluggish" random walk on the entire range of possible ideological positions (the real line). The inertia of the group, the reluctance of the society's agents to change their consensus opinion, increases with the size of the group. The extremists perform biased random walk excursions to the right and left and, in time, their actions succeed to move the society of agents in random directions. The "far left" agent effectively pushes the group consensus toward the right, while the "far right" agent counter-balances the push and causes the consensus to move toward the left. We believe that this model, and some of its variations, has the potential to explain the real world swings in societal ideologies that we see around us.