You might have the privacy settings locked down on all your social networks, but it's the weak links in the system that make the difference. A team at the University of Rochester in New York, for instance, say they can predict where you are to within 100m, just by analysing the location of your friends on Twitter.
New Scientist reports that Adam Sadilek has developed a system that can track the loaction of a Twitter user by looking at where their friends are. To do that, the software looks at the time-varying location of the friends, and how it varies depending on the day of the week and the time. Rolling that together using some fancy Bayesian statistics lets Sadilek pinpoint the user. So far, it's been used on data mined from four million tweets in Los Angeles and New York City. New Scientist describes the accuracy of the technique:
They found a couple of weeks of location data on an individual, combined with location data from their two most sharing friends, is enough to place that person within a 100-metre radius with 77 per cent accuracy. That rises to nearly 85 per cent when you combine information from nine friends. Even someone who has never shared their location can be pinpointed with 47 per cent accuracy from information available from two friends, rising to 57 per cent with nine.
Depending on your perspective, that's insanely impressive or absolutely terrifying. Especially when you consider that the scheme could also be used iteratively, moving from user to user, to make the results more accurate. The technique's demonstrated using Twitter, but there's no reason it couldn't be used with Facebook or any other network that takes location into account.
Still, let's focus on the positives: it won't necessarily be used for stalking, right? It could be used for tracking disease spread, or umm... look, I'm sure there are plenty of applications, it's just that nobody has thought of them yet. [New Scientist]