Distributed Problem Solving: Adaptive Networks with a Computer Intermediary Resource

Abstract : Distributed problem solving networks are omnipresent decision-making structures with often untapped potential. The power of these networks cannot fully be recognized until the basic processes within the network are understood. A major source of disruptions in the network is the degradation of processes caused by uncertainty. Many uncertainty sources have technological solutions, such as increased sensor capability, increased communication bandwidth, or increased processing power. Inter-agent uncertainty is a more fundamental group dynamics problem without an obvious technological solution. Establishing and understanding inter-agent uncertainty reduction processes is necessary for a more effective utilization of the distributed problem-solving network. Forcing disagreeing agents to resolve their differences under control of a predefined set of rules is a proposed method of attacking this problem. The first report proposes a testbed design to study the process using input/output analysis. The difference resolving rules are inputs; the measures of group performance, the outputs. The second focus is the assessment programs/instructions.