Sociophysics of social conflict

We welcome Frank Schweitzer’s article, “Sociophysics” (Physics Today, February 2018, page 40 ), which looks back at antecedents of current physics applications to social phenomena. Schweitzer calls for overcoming challenges that derive from method-driven (physics) perspectives. To his list of examples we propose to add another: problem-driven work. We—an urban planner and two physicists—started out with social conflict, looked at its real problems, and reached for statistical mechanics tools to explore its dynamics over time.

Social conflicts are embedded in complex systems and are themselves complex. They defy description. Moreover, rarely do we know enough about the links between cause and effect to predict outcomes or even to narrow their range meaningfully. Therefore, it is difficult for conflicting parties using negotiation, decision, and planning theories to strategize effectively.

For example, as Democrats and Republicans confront each other over climate change, security threats, and immigration policies, they harbor a range of attitudes, from total opposition toward the other group to openness about bridging differences. Similar tugs-of-war emerge at local levels around transportation and policing initiatives; at regional levels around environmental protection and conservation; and at state levels around regulatory policies. The international level also has no dearth of disputes between two ethnic groups or cross-border conflicts between two neighboring countries. While seemingly different in location, scale, stakes, decision processes, and levels of tractability, two-group conflicts are alike in dynamics.

Observed conflict outcomes can be rationalized, but predicting them is still nigh impossible. Consequently, groups in dispute, interveners, and others concerned with the effects of conflict need to evaluate several courses of action and their possible outcomes. Planners and decision makers are increasingly switching from prediction to scenario-based anticipation and preparing responses for a range of outcomes. Physics tools can be used to generate such anticipatory scenarios ¹ and to pose what-if questions (in toy-model fashion) to help parties contend with an uncertain and complex environment.

For example, we used a multiplex network model ^{2 , 3} to study conflict dynamics among two groups. We explored possible outcomes of two 2016 events rife with conflict—the Brexit referendum and the US presidential elections. We examined the multiplex model with mean-field theory, where the range of interactions between individuals is infinite, and with Monte Carlo simulations of a more realistic version in which interactions decline with distance—geographic or social. The mean attitudes of the two groups exhibit time oscillations and chaotic behavior.

Do people behave like electrons? At large scales it seems as if they do because spatial averages retain only common macroscopic aspects of an ensemble, whereas individual (atypical) behaviors wash out in the averaging. Therefore, statistical-physics models can be used to study some social phenomena involving large numbers of participants and to derive insights not obtainable by other means.