Agency, Security & Policy:
The Application of Multi-Agent Simulations to National Security Issues


    The Department of Defense is faced by a number of strategic, operational and tactical challenges that would benefit from novel insights and technical innovation. A number of these challenges are characterized by the existence of multiple agents, extensive structural hierarchy, and systemic outcomes that extend across time. The ability to model these phenomenon, or human complex systems, allows for a more critical assessment of the nature of these systems, and the implications of prescribed actions aimed at mitigating or accentuating the outcomes associated with these systems.
    One type of technical innovation that is being actively developed and used for these problems is that of multi-agent simulation. The ability to consider the actions and interactions of heterogeneous actors and organizations, as well as compare theoretical assumptions, provides an effective means to explore a number of the issues facing national security organizations. Counter balancing the potential of this modeling approach are limitations associated with appropriate referent representation and model use.  This paper contrasts the potentials and limitations of multi-agent simulations, as well as a scheme for considering the appropriate development of innovative technologies in this area.  The author will consider in detail issues of ontology and epistemology as they relate to national security issues. Specific examples from the national security realm include the performance of highly-networked organizations or systems, leadership behavior under varying regime structures, novel hypothesis generation in support of intelligence analysis, and operational planning.  


Desmond Saunders-Newton
Defense Advanced Research Project Agency ® Information Awareness Office
www.darpa.mil/iao or www.saunders-newton.com
dez@saunders-newton.com