Facebook Knows You Better than You Know Yourself

Image for post
Image for post
Photo by Hugh Han on Unsplash

Our fears around online data privacy are only just beginning. Social networks are a centerpiece in our daily lives. We use them to absorb news, connect with friends, and develop careers. When a service becomes that fundamental, you expect people to share personal details while using it. And I’m not just talking about sharing names and addresses at sign-up. I’m talking about sharing our moods, personalities, likes, dislikes, and beliefs, as we use digital platforms every single day.

The deeper sides of our personalities have always been somewhat hidden, only revealed to those who are closest to us. Our personalities do affect our outward behavior, so they are not completely private to us as we interact with people. But we can be a little more friendly, thoughtful, or outgoing than normal when we choose to. With new people, we tend to show our best sides because we are better able to keep our undesirable traits under control.

Once we have a romantic partner who sees us constantly, controlling our flaws is a much bigger challenge. We all have moments of weakness, and those moments are often the most revealing about us. We are less conscious about projecting our best possible selves to the people who love us unconditionally. So our family and friends are likely to have a fuller picture of our personality than a stranger. Well, that’s until Facebook came along.

The classiest life events make a prominent appearance on social networks. The latest glamorous holiday, or the promotion at work, is more likely to feature in our feed than the latest TV and sofa binge. However, rather like our loved romantic partners, our loved social networks are becoming a heavily connected part of our lives; almost an extra limb. This makes us more likely to share a controversial or traditionally private aspect of our personalities online. We become a little less perfect and a little more honest when we post about our latest health hazard, frustration at the supermarket, or political opinion. Our online personas are becoming more like extensions of our real selves rather than idealized virtual identities. In some ways, it’s exciting to achieve this level of openness in online communication. In other ways, especially in relation to privacy, it’s plain scary.

Sometimes, it’s funny to hear our loved ones describe a specific pattern in our behavior that we had not noticed. We exclaim “You know me better than I know myself!” with a smile. But what if Facebook can do the same thing? And what if Facebook knows you better than even your closest relative or loved one?

Our history of Facebook likes is a diamond mine for data scientists. Those likes hit a wide range of material including musical tastes, humorous memes, and political anger. So when they’re analyzed, they present some surprisingly accurate insights into our character.

In 2013, researchers tested how well a computer could understand you as an individual by working through your Facebook likes. Their database covered more than 58,000 volunteers, and their statistical models aimed to predict personal characteristics by analyzing the content of each volunteer’s likes. The models could predict whether someone was homosexual or heterosexual with 88% accuracy, whether they were African American or Caucasian American with 95% accuracy, and whether they were Democrat or Republican with 85% accuracy. In 73% of cases, they could even tell whether someone was a smoker or not.

Using a similar statistical model in 2015, researchers from the University of Cambridge and Stanford decided to examine whether Facebook likes could also accurately predict a user’s personality. The gold-standard personality assessment in academic psychology is known as the OCEAN model which targets the “Big Five” personality traits: openness, conscientiousness, extraversion, agreeableness, and neuroticism. Facebook users completed a questionnaire to assess these traits, and the researchers then compared how well a computer could predict personalities next to the predictions of human acquaintances. Shockingly, the computer models did a better job of predicting users’ personalities than close friends and family. In fact, the computer’s performance was only matched by the best human judges of our character: our spouses.

So Facebook has the capacity to read us as well as our most intimate companions. The thought is as amazing as it is unsettling. The laptop that I type on right now, in some respects, understands me just as well as my wife does. But beyond simply understanding us, could our social networks directly change our moods and behaviors?

It’s no exaggeration to say that one particular study in 2014 unleashed a media firestorm. Researchers were testing how emotions spread across online networks. For over half a million Facebook users, the experimenters manipulated the likelihood of seeing positive or negative emotional information in personal news feeds. In one condition, the chance of a positive emotional message appearing was reduced to only 10%, while a negative emotional message was set to 90%. The numbers were reversed in the alternative condition.

When participants in the experiment saw less positive emotional information in their feed, their own status updates also contained less positive information and more negative information. When they saw less negative emotional information, their posts showed the opposite pattern. In other words, this was a clear case of rapid online emotional contagion. Our emotions are vulnerable not only to the events in our own life, but also to the events in other people’s lives when they share those events with us. In the physical world, the consequences of this are somewhat attenuated; we have less frequent and less transparent access into other people’s emotional experiences. In the world of the social internet, these barriers fall down. We have direct and immediate access into what our networks are telling us in real-time.

The reason the study above was so controversial is because it felt too invasive. It was always obvious that Facebook was running experiments on us to test their product — perfectly normal practice for tech companies — but there was something especially weird about learning that they could curate our news feeds to manipulate our emotions on a mass scale. When we normally talk to someone in a personal conversation, there is no external editor in our brain adjusting which information we hear and which information is blocked. We hear everything we are told and can decide how we feel about it. We can even walk away from the person if we don’t want to hear what they are telling us. But when our conversations take place on somebody else’s platform, suddenly we have less control over what we see and hear. Even though we signed up to the terms and conditions of the service, we don’t like to find out that our communications could be manipulated to make us feel a certain way.

When we learn about the results of controversial studies, it’s not unusual for our first thought to be one of panic or anger. But with all technological developments, it’s important to get past the most immediate dramatic reactions and look objectively at the possible advantages moving forward. Most recently in October 2018, researchers analyzed the language that hospital patients posted on Facebook, and used it to predict a depression diagnosis with over 70% accuracy. And that is only through analyzing language before a formal diagnosis by a doctor. Our interactions on social networks present a large clean window through which we reveal the qualities of our mental health. Once you get past the eerie feeling that may settle in your stomach after learning this, it’s easy to see the potential benefits to our health services. We currently struggle to understand, predict, and treat the large and diverse sphere of mental health problems. Every additional reliable insight can help to strengthen our medical frameworks.

The most practical applications of detailed online data relate to psychological profiling. A business that understands our likes and dislikes can serve us more effectively. Remember the study I described showing how Facebook likes predict our personalities? If a company understands our personality through this data, they can tailor their adverts and messaging in a way that appeals to us. The extraverts among us are likely to prefer colorful, outgoing, party-related images in adverts, while the introverts will lean toward quieter and more personal images. So what happens when we get what we want?

A study at the end of 2017 showed that messages tailored to our personalities and digital footprints were more persuasive than non-tailored messages. When adverts matched our profiles, for example our level of extraversion, they produced 40% more clicks and 50% more purchases than mismatched adverts. Companies that decide to use this approach will bother us less often with irrelevant nonsense that we don’t want to see. That’s a good thing. But their ultra-persuasive messages might also give us an extra push toward making a purchase we can’t afford or toward an activity that is not necessarily the healthiest personal choice. Perhaps not such a good thing.

The evidence all falls into a mixed bag of costs and benefits. With a reliable insight into our unique characters and perceptions of the world, internet products can provide better personalized services, and doctors can improve the timing and accuracy of their diagnoses. But the sheer extent and quality of the personal data tumbling clumsily across the web also presents opportunities for our data to be used against our own interests.

The number one defensive weapon against data misuse is awareness. That’s why experiments are so important. By explicitly teaching us how our data could be used and how it impacts our behavior, advancing knowledge provides a way to counteract possible negative consequences and speak up against anything we don’t like. I doubt we’re any more likely to read terms and conditions before accepting them, but at least we can be more conscious of how to use a service once we are signed up. If you have data that you desperately want to keep private, do not post it. If you feel the disadvantages of a social network outweigh the advantages in your own life, change how you use it. The stronger our awareness, the better we can make those decisions. At least for now, our choices are still our own.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store