Reviewed by
Bruce Edmonds
Centre for Policy Modelling, Manchester Metropolitan University
On the face of it, one would expect that a book that focuses on automating negotiation would be interesting from the point of view of suggesting approaches for simulating human negotiation. However, almost all the chapters in this book take an extremely narrow view of negotiation, namely that of haggling over continuous dimensions such as price. In these chapters the only communication between parties is the offer of sets of price and options to each other and their acceptance (or not). Observations of human negotiation (e.g. van Boven and Thompson 2003) note that such haggling takes up only a small proportion of the conversations, with a large part of it probing positions, giving information on ones situation, talking about underlying goals, setting frameworks, and discussing what is possible. Only one chapter in this book even touches upon this. This is odd, since a few chapters quote fairly general definitions of negotiation (e.g. where "two parties aim at reaching a joint agreement") and some motivate their work by referring to a richer world of negotiation. The "Complex" in the title refers to the fact that they are negotiating over multiple and sometimes interdependent issues rather than single dimensional haggling or where the issues are monotone and independent of each other.Thus in this book most chapters restrict themselves to a series of offers and counter-offers of single continuous dimensions. One chapter describes a tool for producing a test bed for testing negotiation strategies (chapter 5). Another chapter looks at how a standard price-haggling architecture could be best parameterised using Hofstede's indicators of national characteristics (chapter 3).
That leaves only two chapters of any interest. The first looks at the organisation of convoys through a difficult region (chapter 7, "Acting while negotiating in the Convoy Formation Problem"). This, at least, situates the haggling within a slightly richer problem - that of deciding upon mutual convoys (determined by a 2D place and time of joining up and those of splitting apart again) whilst the agents are already taking actions that will affect the agreed upon outcome. This leads to a variety of different haggling strategies: optimistic vs. pessimistic (depending on how long an agent is willing to hang around for a good deal), ones that do or do not make minimal models of opponents' beliefs, and the sort of concession strategy. However the negotiation still only involves adapting offer values and filtering what offers to consider. No other kind of information is communicated and all agents are aware of all offers.
The most interesting chapter by far is chapter 9: "Automated Negotiation through a Cooperative-Competitive Model". Unlike the other chapters, this takes the process of negotiation a bit more seriously. It distinguishes between two kinds of situation: "Position Based Negotiation" which is where parties are committed to their bargaining goals and "Interest Based Negotiation" which focuses on satisfying the underlying reasons (Fisher and Ury 1983). It then goes on to discuss the kinds of approaches: game-theoretic, heuristic, or argumentation-based. This chapter seriously considers how one might do interest based negotiation using argumentation. Their technique will work for multi-party situations where there is a mixture of cooperation and competition. Communication includes passing/suggesting sub-goals and exchanging information in the search for a mutually acceptable solution, as well as the usual offers and counter offers.
Interestingly they list five capabilities that a good negotiation strategy should exhibit (p. 165):
All in all, this is a very disappointing book for a social simulator who wants some ideas as to how to model observed human negotiation. It is like a book claiming to tell you how to play chess but that only discusses the end game - it will be of very limited use.
ReferencesFISHER, R and Ury, W (1983) Getting to Yes: Negotiating Agreement without giving in. Penguin Books, NY
Return to Contents
of this issue