Scientists from the Psychometrics Centre at the University of Cambridge have found, in a study of 86,000 Facebook profiles, that an individual’s character can be predicted based on their “likes” on the social network
A video of a labrador puppy having a bath. A photograph of a Monster Munch sandwich. Waitrose; Neighbours; a friend’s spare room for rent; a colleague’s travel updates from Australia.
This motley list makes up the things I have “liked” on Facebook in the past few days. Clicking the blue thumbs-up beside a post on the social network is something many of us do without a second thought. Sometimes it’s because we genuinely like it. Sometimes it’s a shortcut to saying hello to an old acquaintance. Other times it’s simply something to do. It’s an absent-minded act, as transient and inconsequential as the changing stories in our news feed.
Or is it? For it has been revealed that as we while away our lives browsing Facebook on our mobile phones and tablets (Britons spend a cumulative 62 million hours a day on social media, according to a national survey), Facebook is now watching us back.
There have recent rumblings that the social network is more than the benign data dump it appears – last month, its “Year in Review” feature, which used an algorithm to collate users’ most popular photographs of 2014 into a cheery collage, offended some by stirring up unhappy memories. Earlier last year, too, it caused outrage by allowing researchers to tweak news feeds for the purposes of a sociological experiment.
But the latest development goes further still. Scientists from the Psychometrics Centre at the University of Cambridge have found, in a study of 86,000 Facebook profiles, that an individual’s character can be predicted based on their “likes” on the social network.
Personalities are measured according to the five main traits used in psychological assessments – openness, conscientiousness, extraversion, agreeableness and neuroticism – and the results go as far as forecasting users’ age, gender, sexual orientation, political and religious views, life satisfaction and intelligence.
“Given that so much of our lives is lived through digital devices, all of them collecting information about you, it’s a great opportunity for researchers to look at large numbers of people in more detail than ever before,” explains Dr David Stillwell, co-author of the prediction software, curiously titled ‘Apply Magic Sauce’. “Some things about it are interesting and some are a little scary, in terms of what computers are capable of.”
Anyone can take the free test – which has been likened to artificial intelligence featured in the film Her, starring Scarlet Johansson, in which a man develops a relationship with a female operating system – but reactions have been mixed. “It was fun to do – I would say it only has a 50 per cent chance of being right, though,” says one friend who let the software access her profile. “It says I’m nearly seven years younger than I am, and that I’m single but 7 per cent lesbian. I’m in a three-year relationship with a guy, so that’s not quite right either.”
In research, the computer was able to predict personality traits based on just 10 “likes”, ranging from books to films, sports and celebrities. A preference for Doctor Who and the game Minecraft, for example, indicates a shy and reserved personality; liking gothic rock and Buffy the Vampire Slayer suggests emotional instability; while a fondness for tanning and meeting new people implies extroversion.
Though some of these may seem obvious, other outcomes are more insightful – and accuracy improves as the number of “likes” increases (the average Facebook profile has 227). Seventy means the computer can identify sides to you that a friend couldn’t; 150 puts it in a better place than a parent or sibling, and analysing 500 “likes” enables predictions on a par with a spouse.
Sadly, my profile doesn’t have enough “likes” listed to produce a character study – something of a relief as my recent activity probably suggests I’m childish, perennially hungry and desperate for a holiday. “You’ll need to track down a trendy 20-year-old,” suggests Dr Stillwell – so my colleague, Alice, acts as a guinea pig (see panel).
Encouragingly, he has taken the test himself. “It says I’m very introverted, which is true. And it says I’m less open minded than I would like to think I am – but colleagues say it’s about right. It seems to know more about me than I do.”
And there’s the unsettling bit. The fact that social networking behaviour can create such accurate personality portraits has raised concern among experts.
“It is very Big Brothery,” says Nathalie Nahai, a web psychologist and author of Webs of Influence. “There is much more intimate surveillance online than people know about, and it’s crucial we understand that we have a choice – and what we’re giving up by opting in. As long as we’re seduced by the convenience of these networking platforms, I worry we’ll sleepwalk into giving up more information than we intend.”
Any research in this area is, inevitably, limited by the disparity between our online and offline personalities – Facebook profiles tend to reflect an idealised image (“liking” that cool band, for example, or using a flattering profile picture) rather than a realistic one.
But Dr Stillwell insists this doesn’t undermine his study. “Of course there are issues about self-presentation, but it’s a bit like applying for a job. Even if everyone presents themselves as being more conscientious or agreeable than they are, the most and least conscientious or agreeable will still rank in the same place in their social group.”
So, should you take the Facebook test? It’s a bit of harmless fun, and should be understood as such. Real-life friends are, and always will be, a better gauge of personality than a machine – but you don’t need a computer to tell you that.
“Be careful,” warns Nahai. “Something as throwaway as a ‘like’ can give away rich personal information that you can’t get back once it’s in the public domain.” Something to bear in mind next time your cursor hovers over that little, innocuous thumbs-up.