None of the previous chapter's analyses calls into question a theoretical expectation that disagreement will be short-lived among people who communicate on a regular basis. The agent-based models of dyadic, one-on-one interaction clearly lead to a conclusion that time will erase diversity among interacting agents. The present chapter modifies the design of these agent-based models to incorporate the idea that people react to new information only after they compare it against the input they receive from the other agents in their communication networks. Several startling results emerge from this simple premise of autoregressive influence. Self-organized networks develop in which agents experience diversity and respond to it in an understandable way. These networks reflect the micro-level impact of the experience that agents accumulate through a continuing series of dyadic interactions. The agents have no conscious drive to expose themselves to diversity. Rather, they simply adjust to new information in a way that conforms to their experience within networks of acquaintance – to the majority sentiment among other agents with whom they are most familiar. Perhaps surprisingly, diversity is preserved in a system that is often, but not always, stable in its response to exogenous disturbances.

The previous chapter provides scant support to those friends of democratic politics who might hope for the persistent and enduring experience of political disagreement among citizens. In every model, political homogeneity appears to be the long-run stable consequence of interaction among agents.