Autonomous Cyber Agents: Rules for Collaboration and Concurrency

Any program, machine or person engaged in computerenabled work is a cyber agent. Can such agents be made collaborate effectively? Both empirical evidence and theory suggest that there are simple rules for designing certain problem-solving organizations, called asynchronous teams, in which collaboration is automatic and scaleeffective (adding cyber agents tends to improve solutionquality, and adding computers tends to improve solutionspeed). This paper develops some of these rules.