We all have friends who smugly boast of their lack of involvement in social media—and their heightened privacy as a result. But new research reveals they may have more to worry about than they thought.
A team of researchers from Switzerland's ETH Zurich University believes that social networks are learning about everyone, whether they're signed up to the site or not. Though they couldn't probe information held by the likes of Facebook or Twitter, the team studied publicly available data archived from that old gem of a site Friendster. They found that the data locked away in there held some amazing nuggets of information about non-users—including sexual orientation, age, relationship status, occupation, and political affiliation.
In a paper, David Garcia, a postdoctoral researcher working on the project, showed that algorithms could mine the data to predict sexual orientation with an accuracy of 60 per cent—compared to just 5 per cent if the software merely guessed with nothing to go on. He reckons that similar analysis could provide similar results for the other categories, too. The insights are made possible by what Garcia refers to as "shadow profiles." Based on using statistical analysis to study the tastes, opinions and relationships of people that do use the service, it's possible to apply insight to people who don't use it. And in turn, infer plenty about them.
While the Friendster data is over a decade old, similar information is available in all social networks—it's just a question of whether or not it's used. In other words, Facebook has this kind of ability at its fingertips, and the kind of engineers and statisticians who could whip up the code to mine it in next to no time. While Facebook has stated in the past that it "doesn't have shadow accounts or profiles – hidden or otherwise – for people who haven't signed up for our service," that doesn't mean it doesn't analyse the data at its fingertips to glean more insight. And honestly? We might never really know if it does or not. [arXiv via Wired]