A new algorithm released by MIT students can determine the sexual orientation of a person just by analyzing the sexual preferences of their friend networks.
It didn’t begin as an insidious way to publicly out Facebook friends. The pair of MIT students who designed the algorithm did so for a class that was focused on ethics and law on the internet. The assignment was simple enough: Exactly how much information do we unwittingly reveal about ourselves when we friend people, “Like” certain topics, or become a fan of certain things?
The answer is: A lot more than you realized.
According to an analysis of friends networks on Facebook, students found that gay men had more gay friends than straight friends. Using that logic, a pair of MIT students created an algorithm that essentially “outs” people just by analyzing the breakup of their friends networks. Early tests of the equation successfully “outed” several gay people, just by analyzing the sexual preferences of their personal networks of friends. However, this same algorithm only worked to identify gay men; lesbians and bisexual men or women were not as easily outed.
“When they first did it, it was absolutely striking – we said, ‘Oh my God – you can actually put some computation behind that,'” states Hal Aberson, a computer science professor at MIT. “That pulls the rug out from a whole policy and technology perspective that the point is to give you control over your information – because you don’t have control over your information.”
The project focused on the “homophily principle,” basically the fact that like attracts like. If you’re into dogs, you’re more likely to friend people who also like dogs. If you’re gay, you’re more likely to hang out with people who are also gay. Using this basic principle, researchers are able to pick out a person’s political affiliation, gender and even what breed of dog they prefer.
This really puts a new spin on internet privacy. It’s not as easy as removing or hiding personal information on Facebook or Twitter. We can reveal truly personal information just by the internet crowd we pal around with.