Who's Telling You The Truth About Dating Algorithms?

The online dating industry is a $4 billion business. And everyone from popular author Dr. Pepper Schwartz to mathletic OkCupid cofounder Sam Yagan is trying to crack the code for success. Business success.

You've seen the digital-age versions of self-help gurus, the ones with official titles suggesting they've cracked the code of human compatibility. Relationship Scientist. Behavioral Expert. They hold doctoral degrees. In labs, they reproduce the conditions of relationships, study interactions, generate conclusions. People of type A are compatible with people of type B. Here's why they worked. Here's why they failed. Here, read this new book.

As the global market for online dating surpasses $4 billion, the lucky ones get hired as consultants by online-dating companies. They write personality-profiling tests, tweak the algorithms. They speak at online-dating conferences, describe their unique matching approaches, and promote their books. But the ones most likely to be telling you the truth are the ones that admit that their dating algorithms are also powerful marketing tools. And they're probably math geeks. Married math geeks. 

Dr. Pepper Schwartz is no geek. But she is the author of The Love Test, The Great Sex Weekend, Everything You Know About Love and Sex is Wrong, and more. And she's a professor of sociology at the University of Washington in Seattle who mugs on behalf of the dating site PerfectMatch.com, where she co-developed The Duet® Total Compatibility System. On the final day of this week's annual online-dating industry conference in Miami, Schwartz, a pocket-sized woman with a calming smile, told the audience her system is based on the Myers-Briggs Type Indicator, a questionnaire designed to measure how people perceive the world and make decisions. You can read about The Duet® Total Compatibility System in her book, Finding Your Perfect Match: 8 Keys to Finding Lasting Love Through True Compatibility.

Joining her was Dr. Eli Finkel (Northwestern University), Schwartz's younger colleague in the behavioral sciences, who staked out his own territory as industry-scold, denouncing eHarmony, one of the largest dating sites in the world and the first to market a scientific approach to matching. EHarmony has refused to reveal its algorithm, Finkel said, and therefore the company should not advertise a scientific approach to matching until it can show, publicly, that its system works according to the standards of scientific rigor. An up-and-comer hoping to make a splash, Dr. Finkel spoke with imploring volume and speed, as if an elaborate show of authority might convert the crowd to his cause. He does not yet work for an online-dating company.

For nearly 50 years, ever since computers were first used to help college kids hook up, people assumed, or hoped, that the fact of technology as mediator would mean not just more dates but better dates. The Great God Computer must know something we don't, the thinking went. It just must. The notion became a wonderful marketing tool—red meat for the media. As such, the math geeks who ran the first computer-dating services at Harvard in the 1960s were happy to perpetuate it. Oh yes, of course we're always refining our codes, optimizing our algorithms.

In 1966, the inventor of computer dating, a Harvard math major named Jeff Tarr, joked to a reporter: "If there's some chick I'm dying to go out with, I can drop her a note in my capacity as president of Operation Match and say, 'Dear Joan, You have been selected by a highly personal process called Random Sampling to be interviewed extensively by myself ... '" The industry's second-comer, another Harvard math geek named David Dewan, remembered: "There was a lot of randomness to it. 'Do you like pizza?' 'Me too!' 'What kind of movies do you like?' 'Romantic comedies?' 'Me too!' Then you go meet her and most of the time you put your head in your hands because she was so ugly—and she was thinking the same about you." (This was the pre-Internet era, mind you, when computers had 12K of memory. Match sheets arrived without photos.)

Their doubts aside, the young men still boasted publicly of doing it better than the competition. From the company's perspective, claiming a superior "scientific matching system" or "personality profiling test" could distinguish you from the field. In 1965, Dewan told the Harvard Crimson that his competitor's questionnaire was "less sophisticated, appealing to the big, Mid-west universities."

All these years (and all this behavior science) later, it's not the professor-backed dating sites but the ones run by math geeks that seem to be on top. At the conference, Sam Yagan, a cofounder of the free dating site OkCupid.com, strutted around, collected multiple awards (for the second year in a row), and gave a talk on how he sold OkCupid to Match.com last year for $90 million, an incredible sum for an advertising-based business model that is thought by many in the business to bring in little revenue.

Playing on their admiration and jealousy, Yagan, a Harvard grad who wears jeans and OkCupid T-shirts beneath a blue blazer, encouraged his colleagues one minute, and provoked them into fits of rage the next. At the awards dinner, Yagan seemed to privately commend a cofounder of HowAboutWe.com for the company's recent completion of a venture capital financing, in which it raised $15 million. The next day, he told the conference that historical valuations of online-dating companies don't justify large venture investments, and therefore "woe be to those who take money."

But on the algorithm panel, Yagan, the lone dating site owner invited to speak on the topic, was relatively subdued. He responded to Drs. Finkel and Schwartz with an occasional eye roll. When it was his turn, he dispatched a very brief, slide-assisted explanation of OkCupid's matching process.

The user, let's call him John Dater, is not required to answer any questions. But the site is premised on the idea that the more questions John D. answers, the better OkCupid works for him. Questions range from the abstract (Would you prefer that good things happened, or interesting things?) to the specific (How often do you feel the need to get really drunk?); from penal policy (Which of the following is the more appropriate penalty for rape: death, castration, prison, community service?) to trust issues (Would you be okay with your significant other spending a lot of time with his/her ex?); from the practical (Ideally, how often would you have sex?) to the downright intimate (How do you feel about kissing your partner after he/she goes down on you?).

For each question, John provides three answers: (1) his own answer, (2) the answers he's willing to accept from a match, and (3) the level of importance he attaches to the question: irrelevant, a little important, somewhat important, very important, or mandatory. "Our objective," explained Yagan's partner, Chris Coyne, "is to figure out what you want, rather than figure out what's best for you. So to accomplish that we play a giant question-and-answer session with everyone on the site at once." When John clicks on a new profile he's shown a "match percentage" that accounts for all the questions that both he and the potential match have both answered. If they've both answered 1,000 questions, then OkCupid's algorithm generates a match percentage based on 6,000 answers—the product of 1,000 questions times 3 answers per question times 2 daters.

The innovation of OkCupid—and what distinguishes it from other "matching" sites such as eHarmony and PerfectMatch—lies in its pliability. Coyne believes he can give members the power to sort through matches online the same way they would offline—assuming a world of perfect information. "Suppose your buddy tells you he has a girl he wants to set you up with," Coyne said. "No matter how much you trust him, you're not going to say 'yes' without asking questions first. In your mind you have a filtering process that's built in, which is different from another guy's filtering process. EHarmony would say they know all our filtering processes. We say we don't. But we can give you the tools to express your filtering process. We can show you how everyone stacks up against your filter, and how you stack up against theirs. Then you're on your own."

As for Yagan, this side of the business is not his priority, and he distances himself from the notion that they're relationship gurus. "We're a bunch of math guys," he told the Boston Globe in 2007. "We don't know anything about dating." This is typical Yagan: We don't even know what we're doing and we're still doing it better than you.

So it could be false modesty. Or it could be that there's only so much technology can accomplish when it comes to predicting compatibility. That one of the most successful sites is run by four young nerds, all of whom are married, seems meaningful.

"I don't care what matching system I have," Yagan told a reporter after the algorithm panel. "I just know I have to have one."

Dan Slater is writing a book about the online-dating business and what technology means for the future of relationships, to be published by Penguin (Portfolio) next year. Slater is the founder and editor of The LongForum, a website that promotes the best of long-form journalism.

Related: Inside The Online Matchmaking Industry's Giant Blind Date

[Image provided by ShutterStock, Thumbnail provided by ShutterStock]

Add New Comment


  • Payal Rana

    Ya, It's a hug conflict of interest, and I knew that, but thought effectiveness would trump it. Nope. They are not as blatant when on the podium or panel members, but the talk in the hallways was very much about being ineffective and making it seem like there is huge success, without really offering any. They have to keep you as long as they can.



  • Dan Walker

    I have an online dating product that has been proven to make online dating more successful. This is not spam, as I'm not going to give you the name of it. But, when I attended iDate Miami last year, I was floored, although I must say, quite naive, in the focus of much of the talk being about how to keep subscribers from leaving with a suitable partner. 

    It's a hug conflict of interest, and I knew that, but thought effectiveness would trump it. Nope. They are not as blatant when on the podium or panel members, but the talk in the hallways was very much about being ineffective and making it seem like there is huge success, without really offering any. They have to keep you as long as they can. 

    The business model is based on retention when the consumer's desire is the opposite. Getting you to go from one month to three months to six months and a year is the goal. That's why they love to send you to their old abandoned profiles. For the free sites, getting you to click through as many pages as possible is the goal. Clicking on beautiful abandoned profiles is perfect to them. Seeing you leave is a failure. 

    I had read research that matching people is impossible with criteria like they all use. OKCupid's is interesting as it gets closer to what you might really be about. My idea was to let them discover a person's sense of humor and rapport and chemistry by experiencing things together in a perfectly safe environment online.

    Oh well, back to the drawing board.

  • FernandoArdenghi

    Actual online dating sites offering compatibility matching methods are only fueled by big marketing budgets and not by serious scientific evidence. No one ( eHarmony/eDarling, Chemistry, PerfectMatch, PlentyOfFish Chemistry Predictor, MeeticAffinity, Be2, RewardingLove, Parship, True, etc) can prove its matching algorithm can match prospective partners who will have more stable and satisfying relationships than couples matched by chance, astrological destiny, personal preferences, searching on one's own, or other technique as the control group in a peer_reviewed Scientific Paper. They are all like placebo, because
    * Actual online dating sites offering compatibility matching methods, when calculating compatibility between prospective mates, have less or at least the same precision as searching on one's own. [in the range of 3 or 4 persons compatible per 1,000 persons screened]
    * That is because they use:
    a) simplified versions of personality traits, instead of the 16PF5 or similar with the complete inventory (16 variables)
    b) inadequate quantitative methods to calculate compatibility between prospective mates, like eHarmony which uses Dyadic Adjustment Scale or other sites which use multivariate linear / logistic regression equations o other equations.
    I had reviewed over 55 compatibility matching engines intended for serious dating since 2003, when I had discovered "the online dating sound barrier" problem.

    Breaking “the online dating sound barrier” is to achieve at least:
    3 most compatible persons in a 100,000 persons database.
    12 most compatible persons in a 1,000,000 persons database.
    48 most compatible persons in a 10,000,000 persons database.
    100 times better than Compatibility Matching Algorithms used by actual online dating sites!

    The only way to achieve that is:
    - using the 16PF5 normative personality test, available in different languages to assess personality of members, or a proprietary test with exactly the same traits of the 16PF5.
    The ensemble of the 16PF5 is: 10E16, big number as All World Population is nearly 6.7 * 10E9
    - expressing compatibility with eight decimals, like The pattern is 92.55033557% +/- 0.00000001% similar to the pattern
    Using a quantized pattern comparison method (part of pattern recognition by cross-correlation) to calculate similarity between prospective mates.

    That is the only way to revolutionize the Online Dating Industry.

    All other proposals are .............. NOISE

  • Fact

    Employees who work for online dating sites might say that's it pretty much like dating in the off-line world.  In other words, 20% of the people are getting 80% of the action. That's life.

    But on-line dating sites might have some unique value if you are looking to match up with somebody with a highly specialized interest. For example, maybe you are vegetarian or something like that.

  • Wize Adz

    "But on-line dating sites might have some unique value if you are looking to match up with somebody with a highly specialized interest. For example, maybe you are vegetarian or something like that."

    That's me.  I'm a particular kind of geek, and I happened to be exactly what my now-wife was looking for.  (And she happened to be exactly who I was looking for).

    The dating pool in an online site isn't as big as I thought it was.  I was on e-harmony for 3 years and I rejected hundreds of matches.  She was on for less than 2 weeks before we started talking.  Five years later, we're comfortably married and the proud parents of a toddler.  Clearly, e-harmony's matching algorithm provided a lot of false positives for me -- but, it only takes one true positive, so whatever.  This matches the paradox of the Internet, though -- there are both more and fewer people out there than you can imagine.

    P.S. My younger sister, a highly-educated lesbian folk singer and environmentalist, found her life-partner more at a younger age and easily than I did.  Also, they didn't have to go online -- a mutual friend introduced them, and then my sister drove over to see her for a sort-of blind date.  The fact that a lesbian folks singer had an easier time dating than my straight/white/conventionally-male self should underscore you how special and awesome my wife is, and how well we match each other, and only each other.  :-)

    P.P.S. It's also worth mentioning that my wife and I were living hundreds of miles apart, and we are both in highly skilled and tightly focused jobs that didn't encourage us to get out much -- so, without some sort of online dating, we would likely still both be single or, at best, moving from one unsatisfying relationship to another.  But, since we were able to span geographical and other divides, we're both happy people!  But let's not give the algorithms too much credit - the dating site just provided a convenient list of people to talk to, and a safe way to start talking.  The rest worked out exactly the same way that youngish people have hooked up since the beginning of time -- long conversations, meeting each others parents, awkward advances that were well received -- you get the picture.

  • barnacle

    Nicely said Nickolas James..I couldn't agree more. Best algorithm = Best Marketing--nothing more.  But also a nice piece of analytical writing  by Slater. Teasing out all the threads here will be quite a challenge. Sounds like you have a good jump on it.

  • Nickolas James

    There is one statistic, that isn't yet proven but could be attested to by millions of people, including myself, who have joined and used a dating site - and that is that there are more failed relationships created from relationships that spawned from a dating site than there is after meeting someone in real life. 

    I'm wondering if anyone would comment and attest to this claim. 

    I say this because I strongly believe, that although an algo could determine if two people like ice-cream and walks at the beach on a warm summer day, work in the same industry and have the same favorite color, the dynamics of life - the easy and hard times will aways come into play - and so the algorithm truly doesn't offer much other than a neat marketing ploy. 

    Let the two people get to know each other and if love is truly in the air, regardless of what an algorithm will say, they'll both make it. It's meant to be. 

  • Wize Adz

    I suppose it depends on what you call a "failed relationship".  I have one friend who probably met 75 different guys off various sites, for one date each.  If the relationship fails early (say, after one date), no harm done.  She eventually married a guy who is apparently a saint, and she's been quite happy for 2 years now. 

    However, if the relationship ends after 10 years (as was the case when my parents divorced), then there was lots of harm done.  Though I'm not complaining, since I'm glad I'm alive.  My parents met in marriage counseling for their previous failed marriages which is about the worst place I can think of to pick up a long-term partner.

    Anyway, that statistic isn't very meaningful without the supporting details about what constitutes a failed relationship.  In my opinion, any time that kids aren't involved, the people in the relationship, married or not, can break up or not -- very little harm done, and I don't really consider it to be a problem.  As soon as there's a kid, though, it's impossible to actually end the relationship, and you have a lifelong commitment (regardless of the formal/legal state of the relationship), so you'd better have gotten it right.

    P.S. Met my wife on e-harmony.  I'm a happily married, and we are the proud parents of a toddler.  Funny thing is that she was on e-harmony for two weeks when she met me; I'd been on for 3 years and not been particularly interested in any of the 300+ "matches".  We were both looking for something very specific in a partner.