Thoughts on “Empirical Models of Privacy in Location Sharing”

I am really interested in ubiquitous computing and location-based technologies so I was looking forward to this paper. In describing their methodology and specifically the concept of “location entropy”, I would have liked a more operational definition of “diversity” of people visiting that space- whether they took into consideration economic, social, ethnic, gender differences and how they qualified those variables. There is an interesting link to spatio-temporal GIS in the observation that more complex privacy preferences are usually linked to a specific time window at a given premises (ie. 9-5 on weekdays on company premises) (pg 130.)

I thought it was a novel approach to focus on the attributes of the locations at which people were sharing their locations rather than the personal characteristics of the individuals which might influence their decision to share their location at one point or another. This inverse format lends itself to generalization across subjects and the formation of universal principles about which kinds of places most inspire location-sharing.

There is an emphasis in the paper on “requests” and the explicit invitation to share one’s location in a social network, but the majority of users supply their location unwittingly or without a formal request. Although this is an important difference, it stands to reason that the authors’ observations about the nature of the request (ie. what app is using the info.) or the context (who the information is broadcast to, whether a network of acquaintances or anonymous gamers), influences an individual’s decision to share their location even in the absence of a formal request.

The Locaccino interface (brilliant branding there) looks very much like Find Friends, an app that I know some of my friends use regularly. It’s great in some ways that we are able to empirically test hypotheses about the kinds of environments and behavioural conditions which promote or discourage location sharing using these real-world datasets.




Comments are closed.