You may have heard of Facebook. They’ve been all over the news lately. It’s a $500 billion company that specializes in social networking services. Last week, its founder Mark Zuckerberg appeared in Congress to address mounting concerns regarding his company’s alleged carelessness with its users’ data.
Earlier this year, media outlets exposed to the public a massive breach of Facebook data allegedly orchestrated by political firm Cambridge Analytica. The data, gathered from more than 80 million Facebook users, was used to benefit political campaigns around the world, including Donald Trump’s, according to The Guardian. Journalists have been sounding the alarm about political misuse of Facebook data since December 2015, when it was found that Ted Cruz utilized Facebook data during his presidential run, but it was only this year that the issue gained significant bipartisan traction, even prompting #deletefacebook.
Scientists have for years been demonstrating how well Facebook knows its users. For example, a casual article by the New York Times, published barely a year before the Cruz expose, reports on a study by researchers from Cambridge University and Stanford University, who developed a computer algorithm that can determine a person’s personality more accurately than even their closest friends and relatives — just by analyzing their Facebook likes.
To test how accurate this algorithm is against the intimate knowledge of those closest to the person, the researchers conducted surveys on 17,000 Facebook users, rating their personalities based on the OCEAN personality traits, openness, conscientiousness, extraversion, agreeableness, and neuroticism. Friends, relatives, and even work colleagues were also tasked to describe the participants through the personality surveys.
ALSO ON PSYCHATORY: #DeleteFacebook Could Be Good For Your Mental Health, Too, But There’s A Catch
The results were surprising — or not, depending on who you ask. It only took the algorithm 10 Facebook likes to predict a user’s personality traits more accurately than a person from work. To defeat a roommate, the algorithm needed 70 likes. To edge out a parent or sibling, it needed 150 likes.
More disturbingly, 300 likes were all it required to predict a person’s personality traits better than their spouse. That’s not an unreasonably high number of likes for someone who has been actively using the social media site for years.
And it’s not only personality that these Facebook likes can potentially predict. Intelligence and even gender orientation have been successfully predicted by algorithms using this seemingly innocuous data. According to Wired in 2013, researchers from Microsoft and Cambridge University analyzed 58,000 Facebook users from the United States using an app called myPersonality, which “allowed users to take real psychometric tests.” The scientists were then able to infer a user’s level of intelligence, among other traits, from the resulting statistical models.
Some of the results were a no-brainer. A user who likes Sarah Palin and Rush Limbaugh on Facebook, for example, can easily be assumed a conservative even without complex data analyses, and it’s exactly what the models showed. But the study, published on Proceedings on the National Academy of Science, was able to find some alarmingly accurate predictions from likes that did not appear to have any link with the trait in question. In one instance, the algorithm was able to determine liking curly fries as a reliable predictor for high intelligence. Liking Lord of the Rings or Mozart was also found to be a predictor of smartness. On the other hand, those with lower intelligence levels were more likely to like Sephora or Tyler Perry.
This applied to determining a user’s sexual orientation too. Gay men liked Adam Levine and Kathy Griffin; straight men liked Bruce Lee and WWE.
Changes in algorithm as well as user behavior could mean these no longer apply in 2018. Either these likes are no longer as predictive or they could be more predictive than ever. With data gathered from unwitting users, third parties like Cambridge Analytica can now find out for themselves and use the results for personal and political gain. Some are not too bothered by this data scandal, but it’s likely because they don’t realize information gathered from them goes way beyond their names and hometowns. It includes the capability to figure things out about your person from something as mundane as liking curly fries.
[Featured Image by Thought Catalog via Unsplash]