Flexibility through Multiagent Systems: Solution or Illusion?

Multiagent software systems are known to exhibit a system-level behavior that rarely can be predicted from the description of individual agents but must be observed in simulation or real-life. On the other hand there are indications that agent technology is superior in situations that are non-deterministic or so ill-structured as to appear non-deterministic. This paper examines whether one can give a more precise characterization of those situations where multiagent systems hold great promise, and to test the corresponding hypothesis by simulating a real-world production scenario. After refining the hypothesis the paper examines whether one can guarantee the reliability of the agents in the presence of disturbances, because otherwise the flexibility of the multiagent system could become uncontrollable. After examining various transactional approaches that all pose major challenges the paper concludes that to go beyond an illusion still requires intensive research.