Safe Exchange Planner

• Venue: ICMAS '00 Proceedings of the Fourth International Conference on MultiAgent Systems (ICMAS-2000)
• Year: 2000

Nondelivery is a major problem in exchanges, especially in electronic commerce: the supplier might not deliver the goods or the demander might not pay. Enforcement is often difficult if the exchange parties are software agents, anonymous, bound by different laws, or if litigation and escrow services are expensive. Recently, a game-theoretic self-enforcing method for carrying out exchanges was presented [Sandholm PhD-96, Sandholm & Lesser IJCAI-95]. The exchange is divided into chunks, and each exchange party delivers its next chunk only after the other party has completed the previous chunk. The chunking and chunk sequencing should be done so that at every point of the exchange, both parties gain more by completing the exchange than by vanishing. This paper operationalizes the theory by presenting a design of a safe exchange planner, which is offered on the web. Exchanges differ based on whether there is one or multiple distinguishable items to exchange, whether there is one or multiple indistinguishable units of each item, whether the items are independent or not in terms of the supplier's cost and the demander's valuation, whether the units are dependent or independent in this sense, and whether the goods are countable or uncountable. For these different settings, we present chunking and chunk-sequencing algorithms that provably find safe exchanges that optimize different simplicity criteria. We also present interface designs that minimize the amount of information that the user has to input in each setting.