A new study has released it’s findings on what sort of information a computer learn about a person based on likes on the social media site Facebook. The researchers found that although a computer can’t make perfect predictions about someone’s personality, it can accurately judge personality traits about as well as spouses can.

In fact romantic partners are the only type of relationship that get as close to matching a computer’s skills in getting to know a person. Coworkers, roommates, friends, and family members are close contenders, but the spouses are the only ones that can rival the computer’s knowledge.

Strangely, it only takes a computer a few seconds to gather up all the information that it needs to make assumptions about a user’s personality. Even when humans are great at reading other humans, there is always more to learn about someone’s personality.

The study gathered up information using the myPersonality project, which is an application designed by David Stillwell at the University of Cambridge. Over a four year period from 2008 until 2012, six million users of the app were being researched based on how they used their Facebook account, as well as through psychometrics tests that revealed information about their personality traits. The sort of things that people “like” on Facebook include musicians, blogs, movies, politicians, celebrities, places, etc.

In return for letting the app gather up their data, the participants would get some explanation of where they fell within the five major personality traits of neuroticism, openness, conscientiousness, agreeableness, and extraversion. The app also collected data from friends of the participants to find out how other people who knew them viewed their personalities.

For the purposes of the study that was published in the Proceedings of the National Academy of Sciences, only 86,220 people were chosen out of the six million app users. The computer used a created model to see how Facebook likes correspond with personality traits, and the researchers found that the program only needed ten likes to assess a personality type better than a coworker could.

With 70 likes a computer could judge someone better than a roommate, it took 150 to pass a family member’s judgements, and only 300 to one up a spouse. This last one surprised researchers, since of course, people tend to know their spouses pretty well. Or at least hope that they do.

The computer program was not perfect in all aspects however. It had a much better time discovering openness in people than neuroticism for example, due to the fact that it is just more challenging to see neuroticism based on the type of things people click like on.

Like groupings of things like “Beer Pong” and “Snooki” were read as outgoing characteristics, while “Wikipedia” and “Doctor Who” would be read as more on the reserved side. Liking “the Bible” points toward cooperative personality types, while “CHANEL” screams a little more competitive.

Psychologist Michal Krosinski and the paper author Youyou Wu understand if these findings unnerve people a little due to the potential of how this sort of information could be used without one’s knowing. This is the sort of information after all, that advertising dreams are made of.