Iterative auctions have many computational advantages over sealed-bid auctions, but can present new possibilities for strategic manipulation. We propose a two-stage technique to make iterative auctions that compute optimal allocations with myopic best-response bidding strategies more robust to manipulation. First, introduce proxy bidding agents to constrain bidding strategies to (possibly untruthful) myopic bestresponse. Second, after the auction terminates adjust the prices towards those given in the Vickrey auction, a sealedbid auction in which truth-revelation is optimal. We present an application of this methodology to iBundle, an iterative combinatorial auction which gives optimal allocations for myopic best-response agents.
David C. Parkes, Lyle H. Ungar