Playing games in many possible worlds
EC '06 Proceedings of the 7th ACM conference on Electronic commerce
| Hi-index | 0.00 |
The frameworks of game theory and mechanism design have exerted significant influence on formal models of multiagent systems by providing tools for designing and analyzing systems in order to guarantee certain desirable outcomes. However, many game theoretic models assume idealized rational decision makers interacting in prescribed ways. In particular, the models often ignore the fact that in many multiagent systems, the agents are not fully rational. Instead they are computational agents who have time and cost constraints that hinder them from optimally determining their utilities from the game and which strategies are best to follow. Because of this, the game theoretic equilibrium for rational agents does not generally remain the same for agents with bounds on their computational capabilities. This creates a potentially hazardous gap in game theory and automated negotiation since computationally bounded agents are not motivated to behave in the desired way. My thesis statement is that it is possible to bridge this gap. By incorporating computational actions into the strategies of agents, I provide a theory of interaction for self-interested computationally-bounded agents. This allows one to formally study the impact that bounded rationality has one agents' strategic behavior. It also provides a foundation for game-theory and mechanism design for computationally-limited agents. First, this thesis introduces a model of bounded rationality where agents must compute in order to determine their preferences. The computing resources of the agents is restricted so that the agents must carefully decide how to best use their computation. I present a fully normative mode of deliberation control, the performance profile tree. Not only does this structure provide full normativity in theory, but I also show that in real-world applications it improves deliberation control over other methods. This thesis proposes explicitly incorporating the deliberation actions of agents into a game-theoretic framework. I introduce a new game-theoretic solution concept, the deliberation equilibrium. This provides me with an approach for understanding and analyzing the strategic use of computation. Using this approach I analyze different negotiation protocols for computationally-limited agents. I highlight a new form of strategic behavior—strategic deliberation—which arises when agents are computationally limited, as well as provide algorithms that agents can use to determine their optimal strategies, when possible. Finally, this thesis studies the problem of designing mechanisms specifically for computationally-limited agents, in order to guarantee both good economic properties as well as good deliberative properties.