Guidance Under Uncertainty: Employing a Mediator Framework in Bilateral Incomplete-Information Negotiations

Bilateral incomplete-information negotiations of multiple issues present a difficult yet common negotiation problem that is complicated to solve from a mechanism design perspective. Unlike multilateral situations, where the individual aspirations of multiple agents can potentially be used against one another to achieve socially desirable outcomes, bilateral negotiations only involve two agents; this makes the negotiations appear to be a zero-sum game pitting agent against agent. While this is essentially true, the gain of one agent is the loss of the other, with multiple issues, it is not unusual that issues are valued asymmetrically such that agents can gain on issues important to them but suffer losses on issues of less importance. Being able to make trade-offs amongst the issues to take advantage of this asymmetry allows both agents to experience overall benefit. The major complication is negotiating under the uncertainty of incomplete information, where agents do not know each other's preferences and neither agent wants to be taken advantage of by revealing its private information to the other agent, or by being too generous in its negotiating. This leaves agents stumbling in the dark trying to find appropriate trade-offs amongst issues. In this work, we introduce the Bilateral Automated Mediation (BAM) framework. The BAM framework is aimed at helping agents alleviate the difficulties of negotiating under uncertainty by formulating a negotiation environment that is suitable for creating agreements that benefit both agents jointly. Our mediator is a composition of many different negotiation ideas and methods put together in a novel third-party framework that guides agents through the agreement space of the negotiation, but instead of arbitrating a final agreement, it allows the agents themselves to ratify the final agreement