Edmund H. Durfee and Jeffrey S. Rosenschein
In this paper, our principal goal is to revisit these terms, work to clarify what they might mean, and encourage the community to consider useful decompositions of the broader research objectives of DAI. For that reason, the reader is forewarned that, in the bulk of the remaining paper, our use of the term "multiagent system" takes on the more narrow meaning as was first intended, and as derived from the history of the DAI field (Section 2). We then consider several views of how multiagent system (MAS) research differs from distributed problem solving (DPS) research (Section 3). Each of the views provides some insight important questions in the field, and into different ways of solving problems and designing systems. We conclude by urging the community to not lose track of useful distinctions within the field, and to universally adopt terms to describe distinctive subfields (Section 4).