We consider algorithmic problems in a distributed setting where the participants cannot be assumed to follow the algorithm but rather their own self-interest. As such participants, termed agents, are capable of manipulating the algorithm, the algorithm designer should ensure in advance that the agents ’ interests are best served by behaving correctly. Following notions from the field of mechanism design, we suggest a framework for studying such algorithms. Our main technical contribution concerns the study of a representative task scheduling problem for which the standard mechanism design tools do not suffice. Journal of Economic Literature

Complex, real-world domains require a rethinking of traditional approaches to AI planning. Planning and executing the resulting plans in a dynamic environment requires a continual approachinwhich planning and execution are interleaved, there may be uncertaintyin the current and projected world state, and replanning may be required when the situation changes or planned actions fail. Furthermore, complex planning and execution problems may require multiple computational agents and human planners to collaborate on a solution. In this article, we describe a new paradigm for planning in complex, dynamic environments, whichweterm distributed,continual planning (DCP). We argue that developing DCP systems will be necessary in order for planning applications to be successful in these environments. We give a historical overview of research leading up to the current state of the art in DCP, and describe research in distributed and continual planning. The increasing emphasis on r...

Abstract This paper considers algorithmic problems in a distributed setting where the participants cannot be assumed to follow the algorithm but rather their own self-interest. Such scenarios arise, in particular, when computers or users aim to cooperate or trade over the Internet. As such participants, termed agents, are capable of manipulating the algorithm, the algorithm designer should ensure in advance that the agents ' interests are best served by behaving correctly. This exposition presents a model to formally study such algorithms. This model, based on the field of mechanism design, is taken from the author's joint work with Amir Ronen, and is similar to approaches taken in the distributed AI community in recent years. Using this model, we demonstrate how some of the techniques of mechanism design can be applied towards distributed computation problems. We then exhibit some issues that arise in distributed computation which require going beyond the existing theory of mechanism design. 1 Introduction A large part of research in computer science is concerned with protocols and algorithms for inter-connected collections of computers. The designer of such an algorithm or protocol always makes an implicit assumption that the participating computers will act as instructed- except, perhaps, for the faulty or malicious ones.

We discuss the design of a trading-agent competition, to be held in conjunction with ICMAS-00. This design will be revised based on deliberations of a committee comprising active researchers in the fieldJ This AmEC-99 presentation constitutes an official announcement of the competition. We solicit feedback from the community regarding further operational and design details, and are hopeful that AmEC attendees will consider participating in the competition.

In an automated contracting environment, where a \customer" agent must negotiate with other self-interested \supplier " agents in order to execute its plans, there is a tradeo between giving the suppliers su cient exibility to incorporate the requirements of the customer's call-for-bids into their own resource schedules, and ensuring the customer that any bids received can be composed into a feasible plan. In this paper, we introduce a bid evaluation process that incorporates cost, task coverage, temporal feasibility, and risk estimation. Using this evaluation process, we provide an empirical study of the tradeo s between exibility, plan feasibility, and cost in the context of our MAGNET multi-agent contracting market infrastructure. Our experimental results demonstrate that the advantage of increasing supplier exibility is dependent on the number of available suppliers. In other words, if the number of suppliers is small, the risk of plan infeasibility outweighs the advantage of added exibility. On the other hand, if the number of suppliers is large, the more exible plan speci cations result in lower-risk plans. 1 1

Distributed operating systems provide users with transparent access to networkwide resources. As changes occur to network resources or as user locations and preferences evolve, the system must be able to adapt by reallocating, replicating, or moving its resources. We explore a market-based approach to resource allocation in suchenvironments, and describe initial experiments in designing and building computational markets for distributed operating system resources. We focus on the problem of redistributing network file allocations for a mobile user population in order to improvelocality and accessibility.

Abstract. We present an approach to the bid-evaluation problem in a system for multi-agent contract negotiation, called MAGNET. The MAGNET market infrastructure provides support for a variety oftypes of transactions, from simple buying and selling of goods and services to complex multi-agent contract negotiations. In the latter case, MAGNET is designed to negotiate contracts based on temporal and precedence constraints, and includes facilities for dealing with time-based contingencies. One responsibility of a customer agent in the MAGNET system is to select an optimal bid combination. We present an e cient anytime algorithm for a customer agent to select bids submitted by supplier agents in response to a call for bids. Bids might include combinations of subtasks and might include discounts for combinations. In an experimental study we explore the behavior of the algorithm based on the interactions of factors such as bid prices, number of bids, and number of subtasks. The results of experiments we present show that the algorithm is extremely e cient even for large number of bids. 1

We discuss the design of a trading-agent competition, to be held in conjunction with ICMAS-00. This design will be revised based on deliberations of a committee comprising activeresearchers in the field.