Researchers from the University of Vermont and the University of Adelaide have said it is possible to build a profile of a person from data gathered from the posts of eight or nine of their contacts.
They claim their study showed it was possible to predict what that person was likely to tweet next as accurately as if they were looking directly at the individual’s own Twitter feed.
The study also claims that even if a person leaves a social media platform such as Facebook or Twitter – or never even joined – the online posts of their friends can provide up to 95% of the “potential predictive accuracy” of a person’s future activities.
Social media and technology giants such as Google, Facebook and Twitter have come under increased scrutiny over the way they handle the personal information of users. The work has been published in the journal Nature Human Behavior.
The researchers suggest the nature of a networked society with millions of online links between people means it could be possible to build a basic profile of someone – such as their political stance, religious views and favourite consumer items – based on the information posted online by their friends.
University of Vermont mathematician James Bagrow, who led the study, said the theory worked both ways because when users sign up for a social media platform “you think you’re giving up your information, but you’re giving up your friends’ information too”.
“You alone don’t control your privacy on social media platforms,” he said.
“Your friends have a say too.”
The scientists said that while their study showed there was a mathematical upper limit on how much predictive information can be gathered from social networks, it does not matter if the person is a user of that social network or not.