Private Data Manipulation in Optimal Sponsored Search Auction

In this paper, We revisit the sponsored search auction as a repeated auction. We view it as a learning and exploiting task of the seller against the private data distribution of the buyers. We model such a game between the seller and buyers by a Private Data Manipulation (PDM) game: the auction seller first announces an auction for which allocation and payment rules are based on the value distributions submitted by buyers. The seller’s expected revenue depends on the design of the protocol and the game played among the buyers in their choice on the submitted (fake) value distributions. Under the PDM game, we re-evaluate the theory, methodology, and techniques in the sponsored search auctions that have been the most intensively studied in Internet economics.