Measuring privacy
In general, privacy is a hard thing to measure, especially since it’s hard even for users themselves to quantify. For example, photos alone are likely to have wildly varying privacy requirements, depending on who is in the photo, where it was taken, etc. In our survey, we simply treated all privacy violations as being equal, even though this is certainly not true in practice. In future work, we will explore mechanisms for measuring the “importance” of the various privacy violations, potentially by asking the users or using machine learning approaches on content metadata