Dr. Michal Kosinski leapt into global public attention last year when his team at the University of Cambridge Psychometrics Centre designed an algorithm that could use Facebook Likes alone to reliably determine six million users’ private traits like their sexual orientation, IQ, religious beliefs, life satisfaction, and personality traits—even when the Likes seemingly had nothing to do with the traits in question. Since then, this he has been consulted by the EU parliament about Internet issues, as well as marketing and HR executives, boasting over 40 keynote speeches and lectures.
“My research has actually been used to argue that the Internet is technically illegal in the EU” he informs me, referring to a wry case by Vesselin Popov that because businesses can use snippets of data about customers’ online behavior to accurately infer so much private information about them that it violates current EU laws requiring that subjects explicitly give consent for businesses to collect information about their ethnic origins, religious beliefs, sex lives, etc.
Such information can be extremely valuable. In an interview last year Michal said that it can be used to predict people’s future action with reasonable accuracy “like on Minority Report you can really predict what people will do in the future and where incidents will most likely occur at a given time using archived and real-time data of the environment. You can know that a person will be drunk and rowdy at a given time even before that person knows about it! Moreover you will know people better than they even know themselves- this is scary.”
Facebook certainly knows many of its users quite well. “Facebook knows much more about citizens than most governments do, in some regards.” Michal says to me, noting that previously only very wealthy governments could afford to keep extensive files on their citizens’ religious beliefs, sexual orientations, and other sorts of information that Facebook’s algorithms can determine easily.
On the other hand, however, he trusts large companies with this data more than he trusts governments with it, though, arguing that companies are actually more responsive to their customers’ concerns about use of their data than governments are: “people are very angry about the NSA and, what? Does anything change? The politicians distance themselves from it but they're not stopping it because there are important limitations here like national security. Stopping the NSA may make people happier, but what if next year you have another 9/11? Regardless of whether extensive eavesdropping could have stopped it or not, those politicians worked to stop the NSA would be in big, big trouble." On the contrary, Michal believes that companies like Facebook will toe the line to keep their customers happy not only to avoid losing customers in the short term, but to avoid annoying them enough to inspire regulations on their activities.
Michal hardly thinks that these companies are without their own leverage, though, calling Facebook and Google international monopolies that “have huge power. Imagine Google changing the order of search results slightly to promote negative messages about a presidential candidate and boom! Next election, Google could sway the results by a few percentage points. One or few people working in a private enterprise could sway the results of the election.” Though Michal doesn’t mention it explicitly during our conversation, a study last year by the Senior Research Psychologist at the American Institute for Behavioral Research and Technology indeed provided strong evidence that small alterations in search rankings could strongly influence voting preferences.
He also admits that his first reaction to hearing about Facebook publishing its experiments on altering users’ emotions was discomfort, but says “while this is a perfectly human reaction that I shared with many, we also have to realize that it is completely irrational. And bashing Facebook for publishing this paper is not only irrational, but naive….. Facebook is doing stuff like that anyway all the time. It's legal. Any second there are hundreds and hundreds of experiments run on Facebook; many of them are run automatically. People don't even touch the computer, they let the computer generate a change in interfaces."
Michal says that Facebook should be praised for the transparency of publishing their experiments, for openly doing what governments and companies like Google do quietly. The negative reaction to the publication will harm open and transparent science in the field, he says, but it won’t prevent companies and governments from conducting it secretly.
But he still considers the sheer amount of data that these companies have collected to be concerning, noting "It becomes dangerous that one company knows so much."
He also points out the risk of personal data being available to an increasing number of parties, that if combined with ever more accurate prediction algorithms could effectively eliminate privacy and cause civil unrest. He explains with an analogy “Take nuclear weapons; they are a big issue, but if only a handful of governments have access to them, the problem is somewhat contained. If every John Smith can have a nuclear weapon then then it increases the risk that someone will blow everything up. So for example currently you need a lot of resources to determine one’s sexual orientation, political views or religion, and only large institutions, such as governments can afford it.
“If only governments know you're gay in a country where it's illegal to be gay then maybe they will use it against you, maybe not. It's dangerous but the weapon is in one hand only, but now if everyone knows you're gay when it's illegal then it becomes super dangerous because anyone can pull the trigger.
“If people’s intimate traits and views can be revealed by anyone, this creates huge challenges. In my view, not only democracy and freedom, but social cohesion, are to a large extent based on the ability to withhold such information from others.”
The solution he proposes is to allow everyone total control over their own data, including online browsing behavior. As an alternative, customization could be handled like NetFlix’s system for voluntarily providing information for personalization. He agrees that this might cut into revenues somewhat, but claims that it would be made up for with increased customer trust improving engagement with services like Facebook’s, and decreasing the risk of governments regulating them to their detriment.
Personally I must admit that I was somewhat skeptical of this solution, but then I talked to the leaders of a foundation with not only a plan to put it together, but an impressive team and not insubstantial financial backing. I’ll post excerpts from my interviews with them next week, but first: the architect behind the Pentagon’s entrance into social media.
This interview is one of a series that I am conducting for my forthcoming book. For more excerpts of interviews with scientists, government officials, executives, and other thought leaders shaping the future of social interaction subscribe now.