Love is an algorithm

How human emotion is translated into maths.

Illustration by D. Thomas Magee, via Engadget

A research piece for Mind Over Matter’s devised piece A Matter of Love.


As a species so driven by curiosity, it is not surprising we use our advancing technology to try and solve the biggest mysteries of human emotion. The concept of computers facilitating the dating process predates the internet; even in 1966 researchers were putting punch cards through computers hoping to find people their mathematically perfect match. 50 years later, how advanced is the technology, and how close can the maths come to defining love?

How does online dating work?

Online dating is used by millions worldwide and any stigma about desperate lovers is falling away. As of 2015 more than half of U.S. adults say online dating is a good way to meet people, and that’s up 15% in the last ten years.

For such a commonplace service, the inner workings of online dating sites are very obscure. Just like KFC, the websites want to preserve their secret blend of herbs and spices — or, in this case, multiplications and square roots. This makes it difficult to understand specifically how it works, but the general concept is matching like with like.

The process that matches two online dating profiles is remarkably similar (and, in some cases, exactly the same) as processes that find you a top-rated plumber, or the perfect stock photo. It can be applied to anything that can be converted into numbers, and while love might seem more nebulous than searching for cat photos dating websites have found a way.

Who better for a simple explanation than Christian Rudder, one of the founders of OKCupid.

Christian Rudder on the maths behind OKCupid

So, all a budding mathematician needs to do is, as Rudder puts it, “take something mysterious — human attraction — and break it down into components that a computer can work with.” Sounds easy, right?

People are numbers

Well, the method requires a slight removal from human emotions. OKCupid received some backlash in 2014 when the tech team ran some unannounced tests of the algorithm. This involved flipping some peoples percentage points and matching them up with people they were not a high match for; saying there was a 90% match when it might have been 30%. The issue with using maths is experiments need to take place, whether you’re testing bank security or human emotions. Buying into the online system means buying into mathematical thinking that makes it work.

Wait, does it work?

In 2012, mathematician Christopher McKinlay created a host of fake profiles on OKCupid in an effort to gather his own data on the online daters. He ended up writing his own algorithm, using what he learned from UCLA and his dating life to perfect his experience. A year later tech CEO Amy Webb gave a TEDx talk on how she went through a similar process trying to find a decent guy in Philadelphia.

In both of these cases the online algorithm wasn’t enough. They both supplemented it with their own math knowledge and, more importantly, subjective experience. This means a lot of trial and error, which is no different from traditional dating.

In fact a 2012 article from the Association for Psychological Science (APS) makes that exact point: online dating is no better than the alternatives. It could be due to the secrecy, but the platforms have produced no significant data to say their services are better than traditional dating. The researchers even claim the basic premise of these sites — that relationship happiness can be determined by personality traits — is misguided, as other research suggests context and interaction are more important.

Having said this, the scientific community are by no means united. The Telegraph published an article in 2013 on research that found online dating made for 25% happier couples. A year later they published another article, with a study saying the services led to marriages three times more likely to divorce.

Perhaps with the current rate of technology we won’t be able to get valuable data for a few years, once the platforms have worked out their kinks. The soundest theory comes from the APS article: that online dating caters to a specific romantic experience. Interactions emerging from dating services revolve around superficiality and listable qualities, and negate the possibility of opposites attracting. Currently, the dating algorithm accounts for one form of human love, and does not (and maybe will never) catalogue the full complexity of attraction.

Destined to fail?

The APS article points out that while dating platforms profess to trying to couple off users, every successful romantic couple means a loss of two customers. This is a reminder that online dating platforms are, at the core, businesses. They toy with data, experiment with markets and try to satisfy investors, the fact their resources are human emotions is irrelevant. If there is any mistrust and doubt, perhaps it shouldn’t be in the maths, but in the humans behind it.

How does it impact on love?

With the wide range of studies the least we can conclude is we don’t know the full impact of online dating on romance. Certainly, communication over computers and phones is now an established part of romantic life (and that’s not a bad thing). At its best online dating uses that avenue to connect people to others they might not have otherwise. At its worst it promotes superficiality and narrows the field of play.

Perhaps Amy Webb and Christopher McKinlay have the answer: a combination of traditional and technological. Use the maths to your advantage to narrow the field, understand the opposition, facilitate the process. When it comes down to it, we still rely on our blood and brains to tell us what to do on a date. Of course, the algorithms are getting smarter with every match, and perhaps some day blood and brains can take a back seat. For now that’s just a Black Mirror episode waiting to happen.

The services may not be able to fully understand all elements of human attraction yet. Certainly, human beings are as of yet not a collection of data points. But the success of online dating proves that maths can, to some extent, create successful romantic partnerships. Whether that’s better than traditional dating (or random chance) is a separate question, but the very fact we’re asking that question shows that science and experimentation makes a good go at capturing what for years has been seen as uncapturable.