Privacy Interfaces for Location Sharing: when is too much transparency a bad thing?

Most smartphones can store and share the owners' location either to provide location-based services or to automatically augment social network activity. Ordinary privacy interfaces are notoriously complex because they require users to try to specify exactly what kids of information they wish to share with whom. Interfaces for controlling location privacy are even more complex because of the added dimensions of time and space (or context). We addressed this problem by conducting a number of studies on location privacy interface requirements. We found that context-sensitive feedback was important and that Erickson and Kellogg's social translucence principles addressed a number of user privacy concerns: that is, users get feedback on who is viewing information about them.

We developed a simple continuous location sharing social networking application and tested on both close-knit groups (families) and looser groups of friends and colleagues. We first implemented simple translucency through text feedback and explored the effects on participant behaviour. Next we implemented a context-sensitive privacy feedback interface which incorporated social translucency using a machine learning approach. This provided feedback to each user in a way that was appropriate for them for their current context. In this talk we present the results of these studies, including the unexpected result that certain groups of users would prefer to remain ignorant about who is tracking them.