Software Can Suggest Privacy Settings
Social media privacy settings are supposed to be easily understood and managed, but with the privacy policies for big social networks proving to be something of a moving target, a lot of users have largely given up trying to control their social content. That’s unfortunate, of course, since there can be all kinds of unpleasant personal and professional ramifications from inadvertent sharing.
But help is on the way: managing your social privacy settings may become easier thanks to software which can suggest privacy settings for content you share with different groups -- essentially predicting what degree of privacy you would choose by analyzing the structure of your social network.
The software, developed by researchers at Penn State and the Missouri University of Science and Technology, employs data-mining to identify specific groups and connections in your personal social network, then predicts a privacy setting for each new individual you add as well as each new piece of content you share. For example, it might automatically limit a photo of a vacation or dinner party to personal friends on Facebook, but exclude casual acquaintances.
According to the researchers the software is 77% accurate in predicting what privacy levels people would choose to assign to any given piece of social media content. They noted that it is more effective for some social sharing situations than others, achieving an error rate as low as 3% in some cases, but with less accuracy for situations where the user is re-uploading content found from other users.
The researchers added that it isn’t necessarily easier to predict what people would share with close friends versus less intimate relations: “Interestingly, when inferring policies targeting general friends, people with common interest and people with similar background, accuracy is up to 90 percent. For policies targeting family, colleagues, close friends, the policy prediction accuracy is lower, about 78 percent.”
While this might seem counterintuitive at first, it actually makes sense: while your personal policies for what you share with, say, fellow hobbyists or other casual acquaintances are probably pretty uniform -- e.g., just snowboarding pics or party photos -- the volume and variety of stuff you share with close friends and family is probably much greater, and thus harder to analyze accurately. Also, the stakes are higher in more intimate relationships, so you are more likely to make subtle distinctions between individuals: it’s easy to imagine sharing content with your siblings that you wouldn’t want to share with your parents, for example.