Data Skeptics: Just How Much Can We Trust Dating Algorithms?

Image courtesy of L. Whittaker via Flickr, modified by Curiousmatic. 

Are the complex algorithms used by dating websites really the key to finding love?

Data Skeptics is a Curiousmatic series that covers intriguing Internet and data topics worthy of close and careful examination. The series explores various spaces within the vast and mysterious world of data,  from which we derive insight for curious minds through a lens of informed skepticism.

View part one, Data Skeptics and the Deep Web, here.

Many of us have been there: the website says he’s 72 percent enemy, but he’s the love of your life, or she’s a 90 percent match in spite of blatant red flags. And yet, by either accuracy or chance, many couples wind up extremely happy due to algorithmic match-making.

But can advanced datasets, mathematics, and science ever truly match the intricate yearnings of both the heart and mind? It’s certainly a tall order — and one many companies claim to serve with accuracy.

One in five relationships do start online now, according to intel — though some doubt the credibility of this claim, and predict that number to be closer to one in 12. Whatever the case, these sites are doing something right.

Let’s take a look at the formulas of the most popular dating website and examine how well they hold up.


Notorious for heaps of embarrassing and crude bombardments of message propositions, usually by male members, OkCupid uses what they call a “unique” and technical methodology to calculate and crunch compatibility.

[contextly_auto_sidebar id=”NM00XppD1LVVqbRAmkESQjvvfkQpuDUk”]It works like this: members provide answers to specific questions, along with how they would like their perfect match to answer, and ranking of how important the question is. Computers then assign numeric value to these answers, divide the totals of two parties’ numbers to find the “satisfactory” percentage, then multiply said percentages and take the square root for the final compatibility percent.

Let’s not forget that OkCupid has admitted openly to experimenting on its users to test interactions — for example, by taking away pictures and measuring outcomes, or taking away profile text and seeing what happens.

The results? Profile text hardly mattered at all, with value assigned almost totally based on pictures. This just goes to show that regardless of data crunching, physical attraction spurs and impacts a majority of interactions, rendering all that math nearly useless.

Match’s algorithmic approach is an evolutionary one that develops and uses hundreds of equations to determine potential compatibility.

By examining what members say (through an extensive survey), tracking what they do (sometimes at odds with what they say), observing how similar people behave, and applying mass amounts of data (16 years worth) to predict patterns, Match doubled “yes” matches after their first round of changes, Mashable reports.


Popular dating site eHarmony has shouted claims across the mountains that their computerized algorithms will fetch you a soulmate, no problem. Unsurprisingly, this promise has been met with skepticism.

But how does it work? Unlike other dating sites, eHarmony doesn’t let users browse for partners on their own as one would a food buffet. Instead, for a mere $60 monthly, matches are derived from a 200 item questionnaire.

The formula is secret, however eHarmony social psychologists have revealed that the newest algorithm (as of 2013) focused on six factors: levels of agreeability, preference for closeness, degree of sexual and romantic passion, levels of extroversion, importance of spirituality, and optimism/happiness.

The most similar scores, eHarmony says, yield the best matches. But critics argue rightfully that similarities do not a perfect pairing make.

The skeptic’s approach

Certainly, the psychologists, scientists, and mathematicians behind dating websites’ algorithms know that even the most theoretically accurate equations come with a margin of error, as physical and emotional attraction remain subjective and tricky to predict.

Websites can only “know” so much about one person with all of the data in the world, and less about the chemistry of two, even when deploying impeccably neat guessing techniques.

Relative newbie dating app Tinder may be revelatory for just this reason: simplicity. Swiping yes or no based on appearance, without the burden of background data and pressure of personality scores, at the very least is as simple as deciding to talk to a stranger based on the limited social cues available. And that’s been working for centuries.

According to a report commissioned by the Association for Psychological Science, the claim that algorithms are a better judge than human reason is, in fact, untrue. Scientists concluded that the success of calculating compatibility online is roughly the same as finding compatibility with a stranger in a public space.

Further, the studies by individual websites that claim otherwise are by definition subjective and unscientific, and the data used negligent of crucial face-to-face factors, the report says.

The takeaway

Online dating algorithms are not infallible, and in spite of improvements, may never perfectly predict the compatibility of human beings in the flesh.

Some aspects that cause even more unreliability, underscoring algorithmic credibility, are:

  • Lies: People often inflate their personality and other aspects of self (81 percent misrepresent their height, weight, or age online, the New York Times says)
  • Attraction: Impulses are still largely based on attraction, which there is no formula for, and which overrides the importance of personality online
  • Fluidity: Mass amounts of data can’t predict a person’s actions, as human personalities, preferences, and beliefs are not static.
  • Opportunity overload: The “shopping market” approach of most online dating sites makes it less personal. Overloads of data and a lack of accountability or vulnerability may cause users to dehumanize interactions, while others may shut down from either bombardment (typically women) or a lack of response (typically men).

Regardless, dating websites continue to succeed at a growing rate in connecting people in innovative and exciting ways. And though at the end of the day, if you’re looking for love, it may be better to be open minded than skeptical, it’s foolish to mistake a mathematical equation for the whimsy of love.

We measure success by the understanding we deliver. If you could express it as a percentage, how much fresh understanding did we provide?
Jennifer Markert