Privacy Preferences, Trade-Offs, and Risk: New Methods and Directions

People across the world are facing challenges to privacy with the rise of large-scale sensing, learning, and inference from personal data. I will describe efforts to elucidate and manage tradeoffs with the benefits, risks, and uncertainties with sharing personal data. First I will review studies on principles of community sensing aimed at limiting sensing to the most important data for providing a service of wide benefit.  Then, I will describe research on balancing and controlling privacy-personalization tradeoffs. The work considers studies with balancing the benefits and costs of sharing information, including efforts to understand peoples’ different sensitivities about sharing different kinds of demographic and behavioral data with online services. Then, I will introduce new work on stochastic privacy, where guarantees are provided to users’ of online services about the small likelihood that data will be accessed. The methods protect users by bounding risk while providing online services with means for enhancing services.