Multiagent systems, and the search for appropriate foundations

Over 25 years ago, faced with a request to provide a short description of my research, I ventured that my work focused on the use of economic theory, voting theory, and game theory to establish appropriate foundations for Multiagent Systems (though the original wording was slightly different). That has remained an accurate description of my research, and it is a description that I still use. In this talk, I will discuss how this meta-description found instantiation in a wide range of my research group's work, the ways in which my perspective has changed over the years, and my understanding of our short-term and long-term challenges in Multiagent Systems. Along the way, I'll elaborate on some specific observations, about research in the field and the field itself, including: - Some of the disruptions that have occurred in MAS, including the move to include self-motivated agents as an important subject for research, and the acceptance of mechanism design as a legitimate topic to be included in artificial intelligence; - Some of the wider trends in AI that have had a more minor impact on MAS; - Ways in which intuitions about interaction have driven my formal research; - The heterogeneous origins of the field, and the lasting impact that has had on its development.