Abstract

This talk will describe some recent joint work with Peter Key (Microsoft Research) and Neil Walton (University of Amsterdam).

Ad-auctions are real-time algorithms that determine the adverts shown, for example, on a Google search. In current Ad-auctions there is an information asymmetry between the platform and advertisers: the platform typically knows more than an advertiser about the search being conducted, such as information about the searcher. Hence the platform can potentially choose prices and an allocation that depends on the platform’s additional information. In contrast, the advertiser has to rely on more coarse-grained information, perhaps just the search terms of a query together with a crude categorization of the searcher.

The talk will describe how the information asymmetry can be used to develop a simple mechanism for advert assignment and pricing that incentivizes truthful bidding and encourages convergence to a unique Nash equilibrium that is socially optimal.