Improving the Military Content of Strategy Analysis Using Automated War Games
A Technical Approach and an Agenda for Research
The RAND Strategy Assessment Center seeks to improve U.S. strategy analysis by combining the best features of political-military war gaming and analytical modeling. Earlier RAND work demonstrated the feasibility of automating the war game by replacing some or all teams by computer models based on artificial intelligence techniques. This Note describes new concepts for building a high degree of military content into the automated war games. The concepts involve rule-based logic structures that are called "analytic war plans," and higher-resolution versions referred to as "branched scripts." These logic structures define the range of strategies and grand tactics from which the computer models ("Red Agent" and "Blue Agent") can choose. The Note also describes the intended approach to combat modeling and explains how realism will be achieved and reflect effects of numerous special phenomena usually ignored by modelers (e.g., attacks on C3I systems and disruptions of rear areas).