FROM ARGUMENTATION TO NEGOTIATIONS: THE GAME OF DETERRENCE PATH
Michel Rudnianski · 2012
∗ Abstract. Negotiations are a process of social interaction between parties defending their own interests. To that end argumentation is an essential component for the success of the process, since it may help one party to convince the other parties that an agreement proposed by the former is appropriate.. A vast literature on the subject has developed in the last two decades (1). The present paper uses a specific category of qualitative games, called games of deterrence to bridge argumentation and negotiations. Initially developed for analyzing nuclear doctrines (15), these games consider .two kinds of outcomes, acceptable and unacceptable ones. This binary feature enables games of deterrence to model argumentation through associating a strategy with each argument presented by a party. The playability of this strategy is assessed, which in turn determines the truthfulness of the argument (18, 19). Likewise, when applied to bargaining, games of deterrence enable to find feasible agreements as the result of a sequential process taking into account the players' postures, i.e. the preferences of each player on the set of possible outcome pairs (16, 17). The bridge between argumentation and negotiation is then built by associating with each set of arguments presented by the parties and recognized as true, a set of agreements consistent with these arguments, and in turn associating with each agreement an outcome for each party. It is assumed in particular that any agreement proposal associated with a set of inconsistent agreement is unacceptable for all parties. The model is tested on the case of a bilateral negotiation, in which the parties are two companies negotiating the conditions of a joint venture. Two versions of the case are developed. In the first one each pair of true arguments is associated in a bilateral relation with a single agreement, while in the second one the arguments pairs are replaced by pairs of arguments sets, such that different pairs of arguments' sets can be associated with the same agreement.