Business Network Fellowship Statement of Purpose: Distributed decision agents for managability and security

This application to the Santa Fe Institute’s Business Network Fellowship presents a model for distributed control that applies to current and planned work in computer network manageability and security. First I review the problems facing the industry in networks of large numbers of machines. Second, I describe methods from a current project on self-propagating worm detection. Finally I embed the method used in this project in a larger theory of distributed decision-making agents, and describe how this fits in with current work at Intel. In short, this is a proposal to apply methods with origins in Bayesian decision theory and modeling to autonomous collaboration among netoworked computers.