Modeling Complex Adaptive Systems as if They Were Voting Processes
Papers from the 2011 AAAI Spring Symposium
Arnold B. Urken, Program Chair
The goal of this symposium is to explore voting models as explanatory tools for understanding and controlling complex adaptive systems. Voting systems are communications structures for assembling and fusing information about simple or complex decision tasks to form collective outcomes. In voting systems, simple or multidimensional information at a low level is represented by data (votes), which are fused across client-server or peer-to-peer networks to form collective outcomes in an emergent and predictable pattern. These voting outcomes yield inferences about facts or preferences associated with a group’s behavior. Recently, error-resilient data fusion (ERDF) patterns of voting behavior have been discovered in which the collective outcome can be inferred on the basis of incomplete and imperfect information. ERDF models relate the time-to-decision in collective decision processes to the systemic attributes of voting systems designed to represent and fuse data. These models can yield alternative theoretical explanations of data fusion techniques in swarm and quorum sensing behavior, collective decision mechanisms in robotics, human-agent collaboration, and the design of resilient and sustainable critical infrastructure.