Agent-Based Modeling

Agent-based modeling (ABM) is a computational method for simulating Complex adaptive systems by implementing the local rules of individual agents and observing the emergent system-level behavior. Unlike equation-based models that describe aggregate dynamics, ABM explicitly represents heterogeneous agents, their interaction topology, and their adaptive strategies.

The canonical ABM is Thomas Schelling's segregation model (1971): agents prefer neighbors similar to themselves, but do not require homogeneous neighborhoods. Each agent applies a simple rule — "move if fewer than 30% of neighbors share my type." The emergent result is near-total segregation, despite no agent preferring it. The model demonstrates that macro-level patterns (segregation) can arise from micro-level preferences (mild homophily) without requiring macro-level intent.

ABM is the natural tool for systems where centralized equations fail: disease spread, financial markets, traffic flow, ecosystems. The cost is that ABM produces scenario landscapes rather than general laws — you can see what happens under specific parameter settings, but parameter sweeps do not yield closed-form predictions. This is not a limitation of the method; it is a reflection of the irreducibility of the systems it models.