CMU Theory Lunch: Kanat Tangwongsan, Regret Minimization and Parallel Algorithms for Approximately Solving Positive LPs/title>  <link rel="stylesheet" href="lunch-main.css"/> </head> <body> <div id="page"> <div id="page_contents">  <div id="header"> <div id="page_title">Theory Lunch</div> <div id="scs_logo"><img src="cmu-scs.gif" alt="School of Computer Science, Carnegie Mellon University" width="290px" border="0" /></div> </div> <div id="main_container">  <div id="sched_header"> Asymmetric Spite in Auctions</div> <div id="abs_talk_speaker"> Ankit Sharma </a> </div> <div id="abs_date">Feb 24, 2010</div>  <div id="abs">  <p>ABSTRACT:</p> <p> This talk deals with "spiteful bidding" in auctions. In general, research literature on auctions assumes that a bidder, on losing the auction, has zero utility. However, is it the case that the losing bidder's utility is zero? To take the case of an online advertising auction, the losing bidder has not only lost the auction but also the customers who now frequent the winner's website and buy his product. Hence, the loser's utility is in fact negative. A losing bidder would therefore try to minimize his negative utility and this would reflect in his bidding strategy. This is known as spiteful bidding. </p><p> While prior work on spiteful bidding has only considered the case where bidders are symmetric in their spite i.e. all have the same utility function on losing, in our work we do not assume symmetry since it is an unrealistic assumption and in practice different bidders have different utility functions. We show interesting results on the way bidders place bids in first-price and second-price auctions. Spite makes agents bid higher than what the zero-utility-model predicts which in turn implies higher revenue generation for the auctioneer. We present cases where a spiteful-bidder with significantly lower valuation than rest of the bidders wins the auction which is never the case in the no-spite model. The results also indicate that first-price and second-price auctions no longer generate equal expected revenue as is the case in the no-spite model. </p><p> This is joint work with Prof. Tuomas Sandholm. </p> </div>  </body>