Abstract
Recent efforts have dedicated significant attention to agent-based modeling in a spectrum of fields, including modern warfare. Far fewer computational social science studies have directly modeled nonviolent campaigns, which is needed to provide a practical alternative to destructive armed conflicts. This paper describes a sequence of experiments that sought to identify the model parameters that would be most beneficial to the success of a nonviolent campaign. The discussion offers both quantitative assessment and qualitative insights into opportunities for additional modeling that can serve as the basis for a quantitative science of nonviolence.