Simulating Complex Social-Behavioral Systems

摘要

Social-behavioral systems are complex. Consequently, they are often assessed using simulation. Particularly valuable is the artificial society approach, in which the social-behavioral system is examined using an agent-based dynamic-network simulation. Historically such models were far removed from real-world systems, costly to build, difficult to maintain, and costly to validate. While these problems still exist, there are practices that, if followed, reduce these problems, support model re-uses, and increase model scalability and veridicality. The practices are predicated on the assumption that context matters. Consequently the purpose of the simulation drives how the model is built, analyzed and validated. These practices are rooted in the unalterable fact that human systems are distinct from physical and mechanical systems and so require fundamentally different types of, and guidelines for, modeling and validation.