Asymmetric Wargaming: Toward a Game Theoretic Perspective

摘要

As we enter the 21st century the art and practice of warfare is radically changing. The US has emerged as the dominant conventional military power only to find its adversaries working their way out of the box. The Defense Advanced Research Projects Agency Information Systems Office (DARPA/ISO) which is seeking new approaches to asymmetric threat modeling, analysis and prediction sponsored this work as well as several related research efforts during FY 2000. This paper enumerates some of the main features of the asymmetric environment and summarizes shortfalls in our current wargame technology. It is argued that contemporary developments in game theory provide a flexible and promising framework in which to efficiently model adversarial motivation and to generate representative asymmetric strategies for improved automation of behaviors in simulations and to support Information Operations analysis and planning. Genetic programming and reinforcement learning are suggested approaches for extraction and refinement of multi-player models from historical data.