Recommendation Bias and Sponsored Search in Information Gatekeepers

We interact frequently with many forms of information gatekeepers -- recommender systems, Internet search engines, Shopbots, ratings agencies etc. These gatekeepers invest significant resources to develop technology -- expertise, data, algorithms -- that enables them to efficiently make recommendations in response to user queries. Sometimes, however, these gatekeepers also incorporate monetary payments from merchants who wish to be recommended, in determining the recommendation list. Two unique characteristics of sponsored recommendations on the Internet (e.g., in search engines such as Google) are (a) performance-based pricing ... the sponsorship fee is determined ex-post based on ow much performance it generates (e.g., in terms of a user click), and (b) sponsored search slots are awarded through repeated execution of online auctions to choose between merchants who bid for specific search words or phrases. Our research investigates several topics related to information gatekeepers' use of sponsorship in making recommendations, including

• 
  
• How should search engines allocate sponsored search slots to competing merchants when the payment mechanism is performance based pricing (pay per click)? How should it weigh wilingness to pay for advertising vs. the likelihood of click-throughs, especially when the click-through rate is not known?
  
  J. Feng, H. K. Bhargava, and D. M. Pennock. Implementing sponsored search in web search engines:
  Computational evaluation of alternative mechanisms. INFORMS Journal on Computing, 19(1):137–148,
  2007.
  
• How should information gatekeepers determine the optimal mix of sponsored and objective recommendations? Will the share of sponsored results, in equilibrium, be too few or too many? How does the technical quality of the gatekeeper, or competition with other gatekeepers, affect the mix? What are the conditions under which sponsored recommendations enhance and when do they harm user benefit from using the recommender?
  
  H. K. Bhargava and J. Feng, “Pay for Play: Sponsored Recommendations in Information
  Gatekeepers", working paper UC Davis.
  
  
• As a form of advertising, sponsored results can serve as a signaling mechanism when consumers are uncertain about merchant quality levels. How does recommender intelligence impact the merchants' allocation of signaling effort between price and advertising? How does it impact product prices, consumer welfare, and the recommmender's revenues from sponsorship? How does the adoption of sponsored results by the recommender affect consumer surplus?
  
  H. K. Bhargava and J. Feng, “Sponsored Results in Intelligent Recommenders: Impact on Quality
  Signaling and Consumer Welfare", under review.