THE FUTURE OF SEX
The week of February 8th, 2014

Why you shouldn’t data-mine your exes

By Greg Stevens

I remember sitting around at a party in my third year of college with a group of people talking about the various “types” we were attracted to. It’s a great exploratory conversation for young people to have, because many are surprised at exactly how diverse sexual attraction can be. In a society that often pushes the illusion that everyone is attracted to the same cookie-cutter standard, it can be enlightening to hear from people who are turned on by those who are older, heavier, shorter, or otherwise different from what we are told to expect.

One woman in the group was bisexual and only attracted to masculine women and effeminate men. There was one 19-year-old student who said that older gentlemen, with a tiny bit of gray in the hair and beard, really got him turned on. A six-foot-four Italian athlete and fitness model said he had a thing for short, glasses-wearing geeky Indian women. And so on. Late night college conversations can be great for breaking people’s preconceptions.

When it was my turn to describe my type, I rambled off the list of traits that I knew I usually go for. Some were very standard (I like skinny people) and some were a little unusual (e.g. I’m more attracted to people much taller or shorter than me, but not people of the same height), but when I was done, a close friend of mine confronted me. He said, “What about brown hair? You left off that you like brown hair.”

“What do you mean?” I objected. “Hair color doesn’t matter at all to me!” But my friend pointed out that he had known me for three full years and seen me on many dates and in many relationships, and almost every single time it was with someone who had brown hair.

“Oh,” I said, thinking about it for a moment, “I guess I must like brown hair.”

That moment has stuck with me—because it was the first time I realized that people might not always be aware of what turns them on. There are some things you know that you like, of course, but there can also be physical cues that you respond to, viscerally and emotionally, without it ever bubbling up to the surface of your consciousness.

Neural network algorithms, and other systems that are designed specifically to find patterns in large data sets, mimic the way our visual centers and even our subconscious minds operate.

So when I heard that Match.com was offering a premium service for $5,000 that would, among other things, use facial recognition software to analyze pictures of your exes to help you determine what your physical type is, the idea made total sense to me. Of course, people might not be consciously aware of all of the physical traits that turn them on—and, of course, it might be helpful to use some kind of mathematical analysis to discover these things!

In fact, many of the current facial recognition and pattern-recognition programs used by people who are in the business of big data analysis are unusually well-suited for exactly this type of thing. Neural network algorithms, and other systems that are designed specifically to find patterns in large data sets, mimic the way our visual centers and even our subconscious minds operate.

One of the most notable recent examples, in fact, is a program that was recently trained to predict whether preteens will end up being binge drinkers before they leave high school. To some people, the fact that a computer program could pick on cues so subtle and obscure that it could predict binge drinking in teens seemed nothing short of miraculous; however, this type of number-crunching is exactly what our own brains are doing when they react to something without us consciously knowing why.

Of course, this isn’t the first time people have tried to apply statistical analysis to sexual attraction and online dating.

Everybody’s “algorithm” for finding the right person is very different.

OkCupid published some very interesting analyses about the relationship between people’s “attractiveness ratings” and how many messages they tend to get from prospective partners. With vast amounts of data come very interesting results, like the fact that people who appeal more narrowly (being rated as “very attractive” by some and “ugly” by others) get more messages than people who are seen as “very attractive” to everyone.

Amy Webb, author of the book Data: A Love Story, gave a brilliant TED Talk about her experience using numbers and statistics to improve her own chances at successful online dating by developing her own scoring system and date-selection algorithm. The biggest take-home message of her talk is that everybody’s “algorithm” for finding the right person is very different and is likely to be different from whatever generic formula dating websites try to  indiscriminately apply to everyone.

But although these analytical approaches are very interesting, they still have a blind spot: When people hand-craft a formula for what traits they are looking for in a prospective partner, they are inherently leaving out any factors that they are not consciously aware of. This is where facial analysis, like the service offered by Match.com, can come into play.

Maybe I’m attracted to wide noses, and I don’t even realize it? Maybe my “type” includes some obscure mathematical relationship. Maybe I’m mostly attracted to people whose head height to body height ratio is exactly 0.125. How would I ever know something like that, without the help of a computer? And I think I would want to find out, wouldn’t you?

Scavenging large data sets to try to find patterns that we previously were not aware of is called data mining. If we could data-mine our exes, wouldn’t that just make it easier for us to find the right match on dating sites? If we could plug in all of the physical details of all of our exes into a database and run a complex large-scale computer program that could uncover every little quirk of our physical attraction—from the inter-ocular gap distance (measured in millimeters) that makes us swoon to the upper-outer lip curvature (measured in radians) that just makes us weak in the knees—wouldn’t that only improve our chances of success in our quest for the best possible dating experience?

It might, but it might not. As appealing as it sounds on the surface, it might actually be a terrible idea.

Maybe I’m mostly attracted to people whose head height to body height ratio is exactly 0.125. How would I ever know something like that, without the help of a computer?

For one thing, physical “types” are notoriously malleable and broadly-defined. Unlike sexual orientations, types are true preferences in the sense that you may be more attracted to someone who is of a particular type, but that doesn’t mean you are never attracted to people who do not fit that mold. Many people have multiple different types at the same time, and often those types are very broadly defined and allow for a lot of leeway and variation.

I asked Jesse Bering, researcher and writer on the topic of sexual psychology and author of the book Perv: The Sexual Deviant in All of Us, about the difference between “types” and “orientations,” and he explained it to me this way: Types can evolve and change over time, and aren’t rigid boundaries that define what can or cannot turn you on; a sexual orientation, on the other hand, is a clear-cut boundary on what is or is not possible for sexual arousal, and usually is set early on in development and does not change after that.

The malleability of types seems to be a common experience. When people go through a bad break-up, they may go through a phase where they are chronically attracted to people who physically resemble the person who jilted them. When they fall in love with someone, physical traits of that person that previously were neutral might now become erotic and sexy. When people start hanging out with a new crowd, that has a different style or aesthetic than their previous social circles, their type will often adapt as they begin to find people in their new social circle more desirable.

Data-mining your exes is completely backward-looking.

A quick search on the Internet provides no end of examples of anxious young people posting questions in anonymous forums about whether it is “normal” for their tastes and desires to change over time. Psychological research has shown that all aspects of our personalities change as we get older, whether we recognize it or not. It would be ridiculous to think that our sexual types don’t change as well.

Experimentally, we even know that people’s response to physical traits can be manipulated in the short term by different stimuli. Research on the psychological effects of porn has shown that you can show a man a few pictures from a porn magazine and it alters what physical traits he considers desirable—at least for a short period of time afterward.

Why does all of this make a difference for the question of data mining your exes? Because data-mining your exes is completely backward-looking. It is a fascinating statistical project that could reveal subtle physical traits that you have been attracted to in the past without even realizing it, but that may or may not have anything to do with that your tastes could evolve into tomorrow, next month, or next year.

Maybe all of your exes had thick eyebrows and a bizygomatic facial width of at least 140 centimeters. If you want to be stuck in that rut, then by all means use facial recognition tests that only pair your with prospective mates that have those same characteristics. Having a type isn’t the same as having a destiny, and it might just be time that you broaden your horizons.

A version of this article was originally published on the Daily Dot on July 7, 2014. 

Photo via _Wiedz/Flickr (CC BY 2.0) | Remix by Jason Reed