Objective vs. subjective coordination in agent-based systems: A case study

This paper aims at showing the benefits of objective coordination in the design and development of agent-based distributed applications. We compare the subjective and objective coordination approaches in the engineering of a simple case study - a distributed MP3 encoding application - pointing out the benefits of the objective ones. In particular, we discuss the design and development of the sample application using three different solutions according to such approaches: a subjective solution, based on conversation and middle-agents, as often found in Distributed Artificial Intelligence and in Multi-Agent Systems; JavaSpaces, as a notable example of loosely-objective approach, not expressive enough to gain all the advantages of objective coordination; and TuCSoN as a fully-objective approach, providing a hybrid coordination model able to exploit the full potential of objective coordination.