Project Aims
This project aims to develop and apply novel market-based methods to the design, management, evolution and control of complex, distributed computational systems. To do so, it brings together a multi-disciplinary team with expertise covering multi-agent systems, economic theory, evolutionary computation, and adaptive systems.
Market-based methods view computational systems as virtual environments, akin to marketplaces, in which participants are economic agents interacting through some form of distributed decision mechanism. Although decision-making by these agents is only local, economics and evolutionary theory provide the means to generate and predict desirable system-level properties. Thus, market-based methods have potential application to many distributed computational systems, including scientific Grid networks, peer-to-peer content delivery networks, and commercial and military systems for real-time co-ordination of autonomous entities.
We believe that the techniques developed in this project will alter fundamentally the way in which distributed systems are designed, used and managed. This expectation is shared by the leading industrial participants supporting this proposal, HP, BT, IBM and BAE Systems, who have all recognized this promise. The potential of market mechanisms for controlling complex computing systems is a relatively recent phenomenon, and designers of distributed systems using such methods currently require the assistance of an expert game theorist or economist. This creates a resource bottleneck if these methods are to be widely deployed. In addition, current theories cannot adequately cover the dynamics of these systems as they evolve (a critical shortcoming, since most complex systems change over time).
Against this background, our long-term goal is no less than Automated Mechanism Design (AMD): the automation of interaction mechanism design and the automation of strategy-selection for participants in distributed computational systems. We believe this bold objective is ultimately achievable, and the current project represents a sound, feasible but nonetheless ambitious step towards this goal. As a result of our research, designers or controllers of complex systems should be enabled to design, deploy and manage market-based mechanisms on an on-going basis for complex, distributed computer-based systems, without requiring human expertise in economic mechanisms, strategy design or evolutionary theory.