Reasoning About Computational Economies

The aim of this project is to develop, apply, and evaluate RACE: a robust and practical model checking tool for use in the formal specification, verification, and analysis of computational economies – computer systems in which system components are assumed to have their own goals/preferences about the overall behaviour of the system, and where these system components are assumed to behave selfishly and strategically in the furtherance of their goals/preferences. The key deliverables of the project will of course include the RACE system itself, and in addition: new formalisms for representing and reasoning about computational economies, suitable for use in the RACE system and elsewhere; theoretical results (e.g., complexity analyses, axiom systems, . . . ) relating to the use of these formalisms; new algorithms and data structures for the verification and analysis of computational economies; and a library of case studies, demonstrating the use of RACE in a variety of settings. The project is both timely and essential. It is timely because computer networks populated by multiple self-interested computational entities are increasingly the reality of computing in the 21st century, and as a consequence, research in this area has witnessed a huge explosion of interest recently. It is essential because current formalisms, tools, and techniques for the specification, analysis, and verification of systems were not intended for, and are not appropriate for, this emerging reality. The project will build on two decades of enormously influential research by the PI, who is among the most highly cited researchers in computer science and artificial intelligence today.