Agent-based computer models to peer through the mists of complexity

More than 20 years ago, sociologist and organizational theorist Charles Perrow developed his “normal accident theory,” which was summarized thusly: “as the complexity and interconnectedness of systems increase, major accidents become inevitable and thus in a sense, ‘normal’.” MIT Sloan Mgt. Rev., Vol. 48, Summer 2007 at 64, explains one technique to help us anticipate normal accidents that may arise from complex situations.

“Agent-based modeling is a computer technique that simulates complex systems from the bottom up to capture their emergent properties.” The software has artificial intelligence rules enabling it to make decisions and act based on those roles. Additionally, the software agents may change or evolve, allowing new behaviors to emerge. Someday we may see an agent-based simulations for hostile tender offers, complex litigation, or merger negotiations.