David Baxter PhD
Late Founder
Looking for Someone: Sex, love, and loneliness on the Internet
by Nick Paumgarten, The New Yorker
July 4, 2011
For many people in their twenties, Internet dating is no less natural a way to meet than the night-club-bathroom line.
In the fall of 1964, on a visit to the World?s Fair, in Queens, Lewis Altfest, a twenty-five-year-old accountant, came upon an open-air display called the Parker Pen Pavilion, where a giant computer clicked and whirred at the job of selecting foreign pen pals for curious pavilion visitors. You filled out a questionnaire, fed it into the machine, and almost instantly received a card with the name and address of a like-minded participant in some far-flung locale?your ideal match. Altfest thought this was pretty nifty. He called up his friend Robert Ross, a programmer at I.B.M., and they began considering ways to adapt this approach to find matches closer to home. They?d heard about some students at Harvard who?d come up with a program called Operation Match, which used a computer to find dates for people. A year later, Altfest and Ross had a prototype, which they called Project TACT, an acronym for Technical Automated Compatibility Testing?New York City?s first computer-dating service.
Each client paid five dollars and answered more than a hundred multiple-choice questions. One section asked subjects to choose from a list of ?dislikes?: ?1. Affected people. 2. Birth control. 3. Foreigners. 4. Free love. 5. Homosexuals. 6. Interracial marriage,? and so on. Another question, in a section called ?Philosophy of Life Values,? read, ?Had I the ability I would most like to do the work of (choose two): (1) Schweitzer. (2) Einstein. (3) Picasso.? Some of the questions were gender-specific. Men were asked to rank drawings of women?s hair styles: a back-combed updo, a Patty Duke bob. Women were asked to look at a trio of sketches of men in various settings, and to say where they?d prefer to find their ideal man: in camp chopping wood, in a studio painting a canvas, or in a garage working a pillar drill. TACT transferred the answers onto a computer punch card and fed the card into an I.B.M. 1400 Series computer, which then spit out your matches: five blue cards, if you were a woman, or five pink ones, if you were a man.
In the beginning, TACT was restricted to the Upper East Side, an early sexual-revolution testing ground. The demolition of the Third Avenue Elevated subway line set off a building boom and a white-collar influx, most notably of young educated women who suddenly found themselves free of family, opprobrium, and, thanks to birth control, the problem of sexual consequence. Within a year, more than five thousand subscribers had signed on.
Over time, TACT expanded to the rest of New York. It would invite dozens of matched couples to singles parties, knowing that people might be more comfortable in a group setting. Ross and Altfest enjoyed a brief media blitz. They wound up in the pages of the New York Herald Tribune and in Cosmopolitan. The Cosmo correspondent?s first match was with a gym teacher who told her over the phone that his favorite sport was ?indoor wrestling?with girls.? (He stood her up, complaining of a backache.) One of TACT?s print advertisements featured a photograph of a beautiful blond woman. ?Some people think Computer dating services attract only losers,? the copy read, quoting a TACT subscriber. ?This loser happens to be a talented fashion illustrator for one of New York?s largest advertising agencies. She makes Quiche Lorraine, plays chess, and like me she loves to ski. Some loser!?
One day, a woman named Patricia Lahrmer, from 1010 WINS, a local radio station, came to TACT to do an interview. She was the station?s first female reporter, and she had chosen, as her d?but feature, a three-part story on how New York couples meet. (A previous installment had been about a singles bar?Maxwell?s Plum, on the Upper East Side, one of the first that so-called ?respectable? single women could patronize on their own.) She had planned to interview Altfest, but he was out of the office, and she ended up talking to Ross. The batteries died on her tape recorder, so they made a date to finish the interview later that week, which turned into dinner for two. They started seeing each other, and two years afterward they were married. Ross had hoped that TACT would help him meet someone, and, in a way, it had.
After a couple of years, Ross grew bored with TACT and went into finance instead. He and Lahrmer moved to London. Looking back now, he says that he considered computer dating to be little more than a gimmick and a fad.
The process of selecting and securing a partner, whether for conceiving and rearing children, or for enhancing one?s socioeconomic standing, or for attempting motel-room acrobatics, or merely for finding companionship in a cold and lonely universe, is as consequential as it can be inefficient or irresolute. Lives hang in the balance, and yet we have typically relied for our choices on happenstance?offhand referrals, late nights at the office, or the dream of meeting cute.
Online dating sites, whatever their more mercenary motives, draw on the premise that there has got to be a better way. They approach the primeval mystery of human attraction with a systematic and almost Promethean hand. They rely on algorithms, those often proprietary mathematical equations and processes which make it possible to perform computational feats beyond the reach of the naked brain. Some add an extra layer of projection and interpretation; they adhere to a certain theory of compatibility, rooted in psychology or brain chemistry or genetic coding, or they define themselves by other, more readily obvious indicators of similitude, such as race, religion, sexual predilection, sense of humor, or musical taste. There are those which basically allow you to browse through profiles as you would boxes of cereal on a shelf in the store. Others choose for you; they bring five boxes of cereal to your door, ask you to select one, and then return to the warehouse with the four others. Or else they leave you with all five.
It is tempting to think of online dating as a sophisticated way to address the ancient and fundamental problem of sorting humans into pairs, except that the problem isn?t very old. Civilization, in its various guises, had it pretty much worked out. Society?family, tribe, caste, church, village, probate court?established and enforced its connubial protocols for the presumed good of everyone, except maybe for the couples themselves. The criteria for compatibility had little to do with mutual affection or a shared enthusiasm for spicy food and Fleetwood Mac. Happiness, self-fulfillment, ?me time,? a woman?s needs: these didn?t rate. As for romantic love, it was an almost mutually exclusive category of human experience. As much as it may have evolved, in the human animal, as a motivation system for mate-finding, it was rarely given great consideration in the final reckoning of conjugal choice.
The twentieth century reduced it all to smithereens. The Pill, women in the workforce, widespread deferment of marriage, rising divorce rates, gay rights?these set off a prolonged but erratic improvisation on a replacement. In a fractured and bewildered landscape of fern bars, ladies? nights, Plato?s Retreat, ?The Bachelor,? sexting, and the concept of the ?cougar,? the Internet promised reconnection, profusion, and processing power.
The obvious advantage of online dating is that it provides a wider pool of possibility and choice. In some respects, for the masses of grownups seeking mates, either for a night or for life, dating is an attempt to approximate the collegiate condition?that surfeit both of supply and demand, of information and authentication. A college campus is a habitat of abundance and access, with a fluid and fairly ruthless vetting apparatus. A city also has abundance and access, especially for the young, but as people pair off, and as they corral themselves, through profession, geography, and taste, into cliques and castes, the range of available mates shrinks. We run out of friends of friends and friends of friends of friends. You can get to thinking that the single ones are single for a reason.
If your herd is larger, your top choice is likely to be better, in theory, anyway. This can cause problems. When there is something better out there, you can?t help trying to find it. You fall prey to the tyranny of choice?the idea that people, when faced with too many options, find it harder to make a selection. If you are trying to choose a boyfriend out of a herd of thousands, you may choose none of them. Or you see someone until someone better comes along. The term for this is ?trading up.? It can lead you to think that your opportunities are virtually infinite, and therefore to question what you have. It can turn people into products.
For some, of course, there is no end game; Internet dating can be sport, an end in itself. One guy told me he regarded it as ?target practice??a way to sharpen his skills. If you?re looking only to get laid, the industry?s algorithmic-matching pretense is of little account; you merely want to be cut loose in the corral. The Internet can arrange this for you.
But if you really are eager, to say nothing of desperate, for a long-term partner you may have to contend with something else?the tyranny of unwitting compromise. Often the people who go on the sites that promise you a match are so primed to find one that they jump at the first or the second or the third who comes along. The people who are looking may not be the people you are looking for. ?It?s a selection problem when you round up a bunch of people who want to settle down,? Chris Coyne, one of the founders of a site called OK Cupid, told me. Some people are too picky, and others aren?t picky enough. Some hitters swing at every first pitch, and others always strike out looking. Many sites, either because of their methods or because of their reputations, tend to attract one or the other.
?Internet dating? is a bit of a misnomer. You don?t date online, you meet people online. It?s a search mechanism. The question is, is it a better one than, say, taking up hot yoga, attending a lot of book parties, or hitting happy hour at Tony Roma?s?
Match.com, one of the first Internet dating sites, went live in 1995. It is now the biggest dating site in the world and is itself the biggest aggregator of other dating sites; under the name Match, it owns thirty in all, and accounts for about a quarter of the revenues of its parent company, I.A.C., Barry Diller?s collection of media properties. In 2010, fee-based dating Web sites grossed over a billion dollars. According to a recent study commissioned by Match.com, online is now the third most common way for people to meet. (The most common are ?through work/school? and ?through friends/family.?) One in six new marriages is the result of meetings on Internet dating sites. (Nobody?s counting one-night stands.) For many people in their twenties, accustomed to conducting much of their social life online, it is no less natural a way to hook up than the church social or the night-club-bathroom line.
There are thousands of dating sites; the big ones, such as Match.com and eHarmony (among the fee-based services) and PlentyOfFish and OK Cupid (among the free ones), hog most of the traffic. Pay sites make money through monthly subscriptions; you can?t send or receive a message without one. Free sites rely on advertising. Mark Brooks, the editor of the trade magazine Online Personals Watch, said, ?Starting a site is like starting a restaurant. It?s a sexy business, looks like fun, yet it?s hard to make money.? There is, as yet, a disconnect between success and profit. ?The way these companies make money is not directly correlated to the utility that users get from the product,? Harj Taggar, a partner at the Silicon Valley seed fund Y Combinator, told me. ?What they really should be doing is making money if they match you with people you like.?
Some sites proceed from a simple gimmick. ScientificMatch attempts to pair people according to their DNA, and claims that this approach leads to a higher rate of female orgasms. A site called Ashley Madison notoriously connects cheating spouses. Howaboutwe.com asks only that you complete a sentence that begins ?How about we . . .? with a suggestion for a first date, be it a Martini at the Carlyle or a canoe trip on the Gowanus Canal. (Your suggestion should theoretically be a sufficient signal of your taste and imagination, and an impetus for getting off-line as soon as possible. Apparently, a big winner has been a ride on the Staten Island Ferry.) The cutting edge is in mobile and location-based technology, such as Grindr, a smartphone app for gay men that tells subscribers when there are other willing subscribers in their vicinity. Many Internet dating companies, including Grindr, are trying to devise ways to make this kind of thing work for straight people, which means making it work for straight women, who may not need an app to know that they are surrounded by willing straight men.
Most of the Internet dating sites still rely, as TACT did, on the questionnaire. The raw material, in the matching process, is a mass of stated preference: your desire or intolerance for certain traits and characteristics. Many of the sites make do with that alone. The more sophisticated ones attempt to identify and exploit the dissonance between what you say you want and what you really appear to want, through the choices you make online.
?What you do is more important than what you say,? Greg Blatt, who is the C.E.O. of I.A.C., and a former C.E.O. of Match.com, told me. (Blatt not only runs the company; he?s also a client. He is one of those guys who say they enjoy dating.) You may specify that you?d like your date to be blond or tall or Jewish or a non-smoking Democrat, but you may have a habit of reaching out to pot-smoking South Asian Republicans. This is called ?revealed preference,? and it is the essential element in Match?s algorithmic process. Match knows what?s right for you?even if it doesn?t really know you. After taking stock of your stated and revealed preferences, the software finds people on the site who have similar dissonances between the two, and uses their experiences to approximate what yours should be. You may have sent introductory messages to only two people, and marked a few others with a wink?a nonverbal expression of interest?but Match will have hundreds of people in its database who have done a lot more on the site, and whose behavior yours seems to resemble. From them, depending on the degree of correlation, the software extrapolates about you.
The trick is in weighting each variable. How significant is hair-color dissonance? Do political views, or fan allegiances, matter? The weightings can change over time, as nuances or tendencies emerge. The algorithms learn. And sometimes behavior changes?political opinion matters more in an election year, for example?and the algorithms scramble to keep up.
An engineer named Amarnath Thombre oversees Match?s base algorithm, which takes into account fifteen hundred variables: whether you smoke, whether you can go out with a smoker, whether your behavior says otherwise. These are compared with the variables of others, creating a series of so-called ?interactions.? Each interaction has a score: a numerical expression of shared trait-tolerance. The closest analogy, Thombre told me, is to Netflix, which uses a similar process to suggest movies you might like??except that the movie doesn?t have to like you back.?
I?ve been on two real dates in my life, both of them in my freshman year of college, nearly a quarter century ago. The first, as it happens, was with the eldest daughter of Robert Ross, the founder of TACT. We met at a party and took up with each other for a while. The date itself came later, on the first night of Christmas vacation. We went to ?Burn This? on Broadway. I remember John Malkovich stomping around onstage and then my date catching a train back to Scarsdale. She remembers that we went to a Chinese restaurant and (this hurts) that I ordered a tequila sunrise. That night, anyway, was the end of it for us.
For the next date, on the advice of a classmate from Staten Island, who claimed to have dating experience, I took a sophomore I liked to a T.G.I. Friday?s, in a shopping center on Route 1 in New Jersey. On the drive there, a fuse blew, knocking out the car stereo, and so I pulled over, removed the fuse box, fashioned a fuse out of some aluminum foil from a pack of cigarettes, and got the cassette deck going again. My companion could not have known that this would hold up as the lone MacGyver moment in a lifetime of my standing around uselessly while other people fix stuff, but she can attest to it now, as she has usually been the one, since then, doing the fixing. We?ve been together for twenty-three years. Needless to say, we had no idea that anything we were saying or doing that night, or even that year, would lead us to where we are today, which is married, with children, a mortgage, and a budding fear of the inevitable moment when one of us will die before the other.
So, for the purposes of this story, I didn?t do any online dating of my own. Instead, I went out for coffee or drinks with various women who, according to their friends, had had extraordinary or, at least, numerous adventures dating online. To the extent that a date can sometimes feel like an interview, these interviews often felt a little like dates. We sized each other up. We doled out tidbits of immoderate disclosure.
I talked to men, too, of course, but there is something simultaneously reductive and disingenuous in most men?s assessments of their requirements and conquests. Some research has suggested that it is men, more than women, who yearn for marriage, but this may be merely a case of stated preference. Men want someone who will take care of them, make them look good, and have sex with them?not necessarily in that order. It may be that this is all that women really want, too, but they are better at disguising or obscuring it. They deal in calculus, while men, for the most part, traffic in simple sums.
A common observation, about both the Internet dating world and the world at large, is that there is an apparent surplus of available women, especially in their thirties and beyond, and a shortage of recommendable men. The explanation for this asymmetry, which isn?t exactly news, is that men can and usually do pursue younger women, and that often the men who are single are exactly the ones who prefer them. For women surveying a landscape of banished husbands or perpetual boys, the biological rationale offers little solace. Neither does the Internet.
Everyone these days seems to have an online-dating story or a friend with online-dating stories. Pervasiveness has helped to chip away at the stigma; people no longer think of online dating as a last resort for desperadoes and creeps. The success story is a standard of the genre. But anyone who has spent a lot of time dating online, and not just dabbling, has his or her share of horror stories, too.
Earlier this year, a Los Angeles filmmaker named Carole Markin sued Match.com in California state court after she was allegedly raped by a man she met on the site; he turned out to be a convicted sex offender. (Twenty years ago, Markin published a book called ?Bad Dates,? for which she solicited anecdotes from the likes of Johnny Bench, Vincent Price, Lyle Alzado, Isaac Asimov, and Minnesota Fats. They suggest that all good dates may be alike but that each bad one is bad in its own way.) Markin?s suit asked not for money but for an injunction against Match.com to prevent it from signing up any new members until it institutes a system for background checks. (A few days later, the company announced that it would start checking subscribers against the national registry of sex offenders.) To some extent, such incidents, as terrible as they are, merely reflect the frequency of such transactional hazards in the wider world. Bars don?t do background checks, either.
Most bad dates aren?t that kind of bad. They are just awkward, or excruciating. One woman, a forty-six-year-old divorced mother of two, likened them to airplane crashes: the trouble usually occurs during takeoff and landing?the minute you meet and the minute you leave. You can often tell right away if this person who?s been so charming in his e-mails is a creep or a bore. If not, it becomes clear at the end of the evening, when he sticks his tongue down your throat. Or doesn?t. One woman who has dated fifty-eight men since her divorce, a few years ago, told me that she maintains a chart, both to keep the men straight and to try to discern patterns?as though there might be a unified-field theory of why men are dogs.
The dating profile, like the Facebook or Myspace profile, is a vehicle for projecting a curated and stylized version of oneself into the world. In a way, the online persona, with its lists of favorite bands and books, its roster of essential values and tourist destinations, represents a cheaper and more direct way of signalling one?s worth and taste than the kinds of affect that people have relied on for centuries?headgear, jewelry, perfume, tattoos. Demonstrating the ability, and the inclination, to write well is a rough equivalent to showing up in a black Mercedes. And yet a sentiment I heard again and again, from women who instinctively prized nothing so much as a well-written profile, was that, as rare as it may be, ?good writing is only a sign of good writing.? Graceful prose does not a gentleman make.
The fact that you can?t get away with lying in your profile for long doesn?t prevent a lot of people from doing it. They post old photographs of themselves, or photos of other people, or click on ?athletic? rather than ?could lose a few pounds,? or identify themselves as single when they are anything but. Sometimes the man says he?s straight but the profile reads gay. Sometimes he neglects to mention that he is a convicted felon. OK Cupid, in an analysis of its own data, has confirmed what I heard anecdotally: that men exaggerate their income (by twenty per cent) and their height (by two inches), perhaps intuiting that women pay closer attention to these data points than to any others. But women lie about these things, too. A date is an exercise in adjustment.
It is an axiom of Internet dating that everyone allegedly has a sense of humor, even if evidence of it is infrequently on display. You don?t have to prove that you love to curl up with the Sunday Times or take walks on the beach (a very crowded beach, to judge by daters? profiles), but, if you say you are funny, then you should probably show it. Demonstrating funniness can be fraught. Irony isn?t for everyone. But everyone isn?t for everyone, either.
I had a talk-about-dating date with a freelance researcher named Julia Kamin, who, over twelve years as a dater on various sites, has boiled down all the competing compatibility criteria to the question of, as she put it, ?Are we laughing at the same ****?? This epiphany inspired her to build a site?makeeachotherlaugh.com?on which you rate cartoons and videos, and the algorithms match you up. As she has gone around telling people about her idea, she says, ?women get instantly excited. Men are, like, ?Um, O.K., maybe.? ? It might be that women want to be amused while men want to be considered amusing. ?I really should have two sites,? Kamin said. ?Hemakesmelaugh.com and shelaughsatmyjokes.com.? (She bought both URLs.)
Good writing on Internet dating sites may be rare because males know that the best way to get laid is to send messages to as many females as possible. To be efficient, they put very little work into each message and therefore pay scant attention to each woman?s profile. The come-on becomes spam and gums up the works, or scares women away, which in turn can lead to a different kind of gender disparity: a room full of dudes. ?There is a fundamental imbalance in the social dynamic,? Harj Taggar, the investor at Y Combinator, told me. ?The most valuable asset is attractive females. As soon as you get them, you get loads of creepy guys.?
The online dating sites are themselves a little like online-dating-site suitors. They want you. They exaggerate their height and salary. They hide their bald spots and back fat. Each has a distinct personality and a carefully curated profile?a look, a strong side, and, to borrow from TACT, a philosophy of life values. Nothing determines the atmosphere and experience of an Internet dating service more than the people who use it, but sometimes the sites reflect the personalities or predilections of their founders.
OK Cupid, in its profile, comes across as the witty, literate geek-hipster, the math major with the Daft Punk vinyl collection and the mumblecore screenplay in development. Get to know it a little better and you?ll find that it contains multitudes?old folks, squares, more Jews than JDate, the polyamorous crowd. Dating sites have for the most part always had either a squalid or a chain-store ambience. OK Cupid, with a breezy, facetious tone, an intuitive approach, and proprietary matching stratagems, comes close to feeling like a contemporary Internet product, and a pastime for the young. By reputation, it?s where you go if you want to hook up, although perhaps not if you are, as the vulgate has it, ?looking for someone??the phrase that connotes a desire for commitment but a countervailing aversion to compromise. Owing to high traffic and a sprightly character, OK Cupid was also perhaps the most desirable eligible bachelor out there, until February, when it was bought, for fifty million dollars, by Match.
OK Cupid?s founders, who have stayed on since the sale, are four math majors from Harvard. While still in school, in the late nineties, they created a successful company called the Spark, which composed and posted online study guides along the lines of Cliffs Notes. At the time, they experimented with a dating site called SparkMatch. The fodder for their matching apparatus was a handful of personality tests and droll questionnaires that they?d posted on the Spark to lure traffic. They sold the company to Barnes & Noble in 2001 and then reunited in 2003 to revive the dating idea. To solve the chicken-egg conundrum of a dating site?to attract users, you need users?they created a handful of quizzes, chief among them the Dating Persona Test. A man might learn, for example, that he?s a Billy Goat, a Backrubber, a Vapor Trail, a Poolboy, or the Last Man on Earth. The Hornivore (?roaming, sexual, subhuman?) might want to consider the female type Genghis Khunt (?master of man, bringer of pain?) and avoid the Sonnet (?romantic, hopeful, composed?). They also urged people to submit their own quizzes. By now, users have submitted more than forty-three thousand quizzes to the site. Answer this or that pile of questions and you can find out which ?Lost? character/chess piece/chemical element you are.
Essentially, OK Cupid opened a parlor-game emporium and then got down to the business of pairing off the patrons. The quizzes had no bearing on the matching, and at this point they are half-hidden on the site. They were merely bait?a pickup line, a push-up bra. There is a different question regimen for matching. On OK Cupid, the questions are submitted by users. There are three variables to each question: your own answer, the answer you?d like a match to give, and how important you think this answer should be. The questions are ranked in order of how effective they are at sorting people. Some questions might be of utmost importance (?Have you ever murdered anyone??) but of little use, in sorting people. Others that divide well (?Do you like Brussels sprouts??) will not do so meaningfully.
And yet some questions are unpredictably predictive. One of the founders, Christian Rudder, maintains the OK Trends blog, sifting through the mountains of data and composing clever, mathematically sourced synopses of his findings. There are now nearly two hundred and eighty thousand questions on the site; OK Cupid has collected more than eight hundred million answers. (People on the site answer an average of three hundred questions.) Rudder has discovered, for example, that the answer to the question ?Do you like the taste of beer?? is more predictive than any other of whether you?re willing to have sex on a first date. (That is, people on OK Cupid who have answered yes to one are likely to have answered yes to the other.) OK Cupid has also analyzed couples who have met on the site and have since left it. Of the 34,620 couples the site has analyzed, the casual first-date question whose shared answer was most likely to signal a shot at longevity (beyond the purview of OK Cupid, anyway) was ?Do you like horror movies?? When I signed up for the site, some of the first things I was asked were ?Are clams alive?? and ?Which is bigger, the sun or the earth?? It?s hard to discern the significance.
The purpose of the blog is to attract attention: the findings, like the quizzes, are to lure you in. Rudder has written a lot about looks: whether or not it helps to show cleavage (women) or a bare midriff (men)?the answers were Yes, Especially as You Age, and Yes, If You Have Good Abs and Are Not a Congressman. He found that women generally prefer it when in photos men are looking away from the camera (hypothesis: less intimidating), and that men prefer the opposite (they want a woman?s full attention). A user can rate other people?s profiles. The matching algorithms take these ratings into account and show you people who are roughly within your range of attractiveness, according to the opinions of others. The idea behind the matching algorithms, Chris Coyne told me, is to replicate the experience you have off-line. ?We tried to imagine software that would be like your friend in the real world,? Coyne said. ?If I were your friend and I told you that So-and-So would be the perfect date, your response to me would be to start asking me questions. Does she like dancing? Does she smoke pot? Is she a furry? Is she tall? On the Internet, people will ask?and answer?extremely personal questions.?
OK Cupid sends all your answers to its servers, which are housed on Broad Street in New York. The algorithms find the people out there whose answers best correspond to yours?how yours fit their desires and how theirs meet yours, and according to what degree of importance. It?s a Venn diagram. And then the algorithms determine how exceptional those particular correlations are: it?s more statistically significant to share an affection for the Willies than for the Beatles. The match is expressed as a percentage. Each match search requires tens of millions of mathematical operations. To the extent that OK Cupid has any abiding faith, it is in mathematics.
There?s another layer: how to sort the matches. ?You?ve got to make sure certain people don?t get all the attention,? Rudder said. ?In a bar, it?s self-correcting. You see ten guys standing around one woman, maybe you don?t walk over and try to introduce yourself. Online, people have no idea how ?surrounded? a person is. And that creates a shitty situation. Dudes don?t get messages back. Some women get overwhelmed.? And so the attractiveness ratings, as well as the frequency of messaging, are factored in. As on Match.com, the algorithms pay attention to revealed preferences. ?We watch people who don?t know they?re being watched,? Sam Yagan, the company?s C.E.O., said. ?But not in a Big Brother way.? The algorithms learn as they go, changing the weighting for certain variables to adjust to the success or the failure rate of the earlier iterations. The goal is to connect you with someone with whom you have enough in common to want to strike up an e-mail correspondence and then quickly meet in person. It is not OK Cupid?s concern whether you are suited for a lifetime together.
OK Cupid winds up with a lot of data. This enables the researchers to conjure from their database the person you may not realize you have in mind. ?Like that guy in high school with the Camaro and the mustache who bow-hunts on weekends,? Rudder said. ?You can find that guy of the imagination by using statistics.? The database also gives them a vast pool to sell to academics. In no other milieu do so many people, from such a broad demographic swath, willingly answer so many intimate questions. It is a gold mine for social scientists. In the past nine months, OK Cupid has sold its raw data (redacted or made anonymous to protect the privacy of its customers) to half a dozen academics. Gregory Huber and Neil Malhotra, political scientists at Yale and Stanford, respectively, are sifting through OK Cupid data to determine how political opinions factor in to choosing social partners. Rudder, for his part, has determined that Republicans have more in common with Republicans than Democrats have in common with Democrats, which led him to conclude, ?The Democrats are doomed.?
by Nick Paumgarten, The New Yorker
July 4, 2011
For many people in their twenties, Internet dating is no less natural a way to meet than the night-club-bathroom line.
In the fall of 1964, on a visit to the World?s Fair, in Queens, Lewis Altfest, a twenty-five-year-old accountant, came upon an open-air display called the Parker Pen Pavilion, where a giant computer clicked and whirred at the job of selecting foreign pen pals for curious pavilion visitors. You filled out a questionnaire, fed it into the machine, and almost instantly received a card with the name and address of a like-minded participant in some far-flung locale?your ideal match. Altfest thought this was pretty nifty. He called up his friend Robert Ross, a programmer at I.B.M., and they began considering ways to adapt this approach to find matches closer to home. They?d heard about some students at Harvard who?d come up with a program called Operation Match, which used a computer to find dates for people. A year later, Altfest and Ross had a prototype, which they called Project TACT, an acronym for Technical Automated Compatibility Testing?New York City?s first computer-dating service.
Each client paid five dollars and answered more than a hundred multiple-choice questions. One section asked subjects to choose from a list of ?dislikes?: ?1. Affected people. 2. Birth control. 3. Foreigners. 4. Free love. 5. Homosexuals. 6. Interracial marriage,? and so on. Another question, in a section called ?Philosophy of Life Values,? read, ?Had I the ability I would most like to do the work of (choose two): (1) Schweitzer. (2) Einstein. (3) Picasso.? Some of the questions were gender-specific. Men were asked to rank drawings of women?s hair styles: a back-combed updo, a Patty Duke bob. Women were asked to look at a trio of sketches of men in various settings, and to say where they?d prefer to find their ideal man: in camp chopping wood, in a studio painting a canvas, or in a garage working a pillar drill. TACT transferred the answers onto a computer punch card and fed the card into an I.B.M. 1400 Series computer, which then spit out your matches: five blue cards, if you were a woman, or five pink ones, if you were a man.
In the beginning, TACT was restricted to the Upper East Side, an early sexual-revolution testing ground. The demolition of the Third Avenue Elevated subway line set off a building boom and a white-collar influx, most notably of young educated women who suddenly found themselves free of family, opprobrium, and, thanks to birth control, the problem of sexual consequence. Within a year, more than five thousand subscribers had signed on.
Over time, TACT expanded to the rest of New York. It would invite dozens of matched couples to singles parties, knowing that people might be more comfortable in a group setting. Ross and Altfest enjoyed a brief media blitz. They wound up in the pages of the New York Herald Tribune and in Cosmopolitan. The Cosmo correspondent?s first match was with a gym teacher who told her over the phone that his favorite sport was ?indoor wrestling?with girls.? (He stood her up, complaining of a backache.) One of TACT?s print advertisements featured a photograph of a beautiful blond woman. ?Some people think Computer dating services attract only losers,? the copy read, quoting a TACT subscriber. ?This loser happens to be a talented fashion illustrator for one of New York?s largest advertising agencies. She makes Quiche Lorraine, plays chess, and like me she loves to ski. Some loser!?
One day, a woman named Patricia Lahrmer, from 1010 WINS, a local radio station, came to TACT to do an interview. She was the station?s first female reporter, and she had chosen, as her d?but feature, a three-part story on how New York couples meet. (A previous installment had been about a singles bar?Maxwell?s Plum, on the Upper East Side, one of the first that so-called ?respectable? single women could patronize on their own.) She had planned to interview Altfest, but he was out of the office, and she ended up talking to Ross. The batteries died on her tape recorder, so they made a date to finish the interview later that week, which turned into dinner for two. They started seeing each other, and two years afterward they were married. Ross had hoped that TACT would help him meet someone, and, in a way, it had.
After a couple of years, Ross grew bored with TACT and went into finance instead. He and Lahrmer moved to London. Looking back now, he says that he considered computer dating to be little more than a gimmick and a fad.
The process of selecting and securing a partner, whether for conceiving and rearing children, or for enhancing one?s socioeconomic standing, or for attempting motel-room acrobatics, or merely for finding companionship in a cold and lonely universe, is as consequential as it can be inefficient or irresolute. Lives hang in the balance, and yet we have typically relied for our choices on happenstance?offhand referrals, late nights at the office, or the dream of meeting cute.
Online dating sites, whatever their more mercenary motives, draw on the premise that there has got to be a better way. They approach the primeval mystery of human attraction with a systematic and almost Promethean hand. They rely on algorithms, those often proprietary mathematical equations and processes which make it possible to perform computational feats beyond the reach of the naked brain. Some add an extra layer of projection and interpretation; they adhere to a certain theory of compatibility, rooted in psychology or brain chemistry or genetic coding, or they define themselves by other, more readily obvious indicators of similitude, such as race, religion, sexual predilection, sense of humor, or musical taste. There are those which basically allow you to browse through profiles as you would boxes of cereal on a shelf in the store. Others choose for you; they bring five boxes of cereal to your door, ask you to select one, and then return to the warehouse with the four others. Or else they leave you with all five.
It is tempting to think of online dating as a sophisticated way to address the ancient and fundamental problem of sorting humans into pairs, except that the problem isn?t very old. Civilization, in its various guises, had it pretty much worked out. Society?family, tribe, caste, church, village, probate court?established and enforced its connubial protocols for the presumed good of everyone, except maybe for the couples themselves. The criteria for compatibility had little to do with mutual affection or a shared enthusiasm for spicy food and Fleetwood Mac. Happiness, self-fulfillment, ?me time,? a woman?s needs: these didn?t rate. As for romantic love, it was an almost mutually exclusive category of human experience. As much as it may have evolved, in the human animal, as a motivation system for mate-finding, it was rarely given great consideration in the final reckoning of conjugal choice.
The twentieth century reduced it all to smithereens. The Pill, women in the workforce, widespread deferment of marriage, rising divorce rates, gay rights?these set off a prolonged but erratic improvisation on a replacement. In a fractured and bewildered landscape of fern bars, ladies? nights, Plato?s Retreat, ?The Bachelor,? sexting, and the concept of the ?cougar,? the Internet promised reconnection, profusion, and processing power.
The obvious advantage of online dating is that it provides a wider pool of possibility and choice. In some respects, for the masses of grownups seeking mates, either for a night or for life, dating is an attempt to approximate the collegiate condition?that surfeit both of supply and demand, of information and authentication. A college campus is a habitat of abundance and access, with a fluid and fairly ruthless vetting apparatus. A city also has abundance and access, especially for the young, but as people pair off, and as they corral themselves, through profession, geography, and taste, into cliques and castes, the range of available mates shrinks. We run out of friends of friends and friends of friends of friends. You can get to thinking that the single ones are single for a reason.
If your herd is larger, your top choice is likely to be better, in theory, anyway. This can cause problems. When there is something better out there, you can?t help trying to find it. You fall prey to the tyranny of choice?the idea that people, when faced with too many options, find it harder to make a selection. If you are trying to choose a boyfriend out of a herd of thousands, you may choose none of them. Or you see someone until someone better comes along. The term for this is ?trading up.? It can lead you to think that your opportunities are virtually infinite, and therefore to question what you have. It can turn people into products.
For some, of course, there is no end game; Internet dating can be sport, an end in itself. One guy told me he regarded it as ?target practice??a way to sharpen his skills. If you?re looking only to get laid, the industry?s algorithmic-matching pretense is of little account; you merely want to be cut loose in the corral. The Internet can arrange this for you.
But if you really are eager, to say nothing of desperate, for a long-term partner you may have to contend with something else?the tyranny of unwitting compromise. Often the people who go on the sites that promise you a match are so primed to find one that they jump at the first or the second or the third who comes along. The people who are looking may not be the people you are looking for. ?It?s a selection problem when you round up a bunch of people who want to settle down,? Chris Coyne, one of the founders of a site called OK Cupid, told me. Some people are too picky, and others aren?t picky enough. Some hitters swing at every first pitch, and others always strike out looking. Many sites, either because of their methods or because of their reputations, tend to attract one or the other.
?Internet dating? is a bit of a misnomer. You don?t date online, you meet people online. It?s a search mechanism. The question is, is it a better one than, say, taking up hot yoga, attending a lot of book parties, or hitting happy hour at Tony Roma?s?
Match.com, one of the first Internet dating sites, went live in 1995. It is now the biggest dating site in the world and is itself the biggest aggregator of other dating sites; under the name Match, it owns thirty in all, and accounts for about a quarter of the revenues of its parent company, I.A.C., Barry Diller?s collection of media properties. In 2010, fee-based dating Web sites grossed over a billion dollars. According to a recent study commissioned by Match.com, online is now the third most common way for people to meet. (The most common are ?through work/school? and ?through friends/family.?) One in six new marriages is the result of meetings on Internet dating sites. (Nobody?s counting one-night stands.) For many people in their twenties, accustomed to conducting much of their social life online, it is no less natural a way to hook up than the church social or the night-club-bathroom line.
There are thousands of dating sites; the big ones, such as Match.com and eHarmony (among the fee-based services) and PlentyOfFish and OK Cupid (among the free ones), hog most of the traffic. Pay sites make money through monthly subscriptions; you can?t send or receive a message without one. Free sites rely on advertising. Mark Brooks, the editor of the trade magazine Online Personals Watch, said, ?Starting a site is like starting a restaurant. It?s a sexy business, looks like fun, yet it?s hard to make money.? There is, as yet, a disconnect between success and profit. ?The way these companies make money is not directly correlated to the utility that users get from the product,? Harj Taggar, a partner at the Silicon Valley seed fund Y Combinator, told me. ?What they really should be doing is making money if they match you with people you like.?
Some sites proceed from a simple gimmick. ScientificMatch attempts to pair people according to their DNA, and claims that this approach leads to a higher rate of female orgasms. A site called Ashley Madison notoriously connects cheating spouses. Howaboutwe.com asks only that you complete a sentence that begins ?How about we . . .? with a suggestion for a first date, be it a Martini at the Carlyle or a canoe trip on the Gowanus Canal. (Your suggestion should theoretically be a sufficient signal of your taste and imagination, and an impetus for getting off-line as soon as possible. Apparently, a big winner has been a ride on the Staten Island Ferry.) The cutting edge is in mobile and location-based technology, such as Grindr, a smartphone app for gay men that tells subscribers when there are other willing subscribers in their vicinity. Many Internet dating companies, including Grindr, are trying to devise ways to make this kind of thing work for straight people, which means making it work for straight women, who may not need an app to know that they are surrounded by willing straight men.
Most of the Internet dating sites still rely, as TACT did, on the questionnaire. The raw material, in the matching process, is a mass of stated preference: your desire or intolerance for certain traits and characteristics. Many of the sites make do with that alone. The more sophisticated ones attempt to identify and exploit the dissonance between what you say you want and what you really appear to want, through the choices you make online.
?What you do is more important than what you say,? Greg Blatt, who is the C.E.O. of I.A.C., and a former C.E.O. of Match.com, told me. (Blatt not only runs the company; he?s also a client. He is one of those guys who say they enjoy dating.) You may specify that you?d like your date to be blond or tall or Jewish or a non-smoking Democrat, but you may have a habit of reaching out to pot-smoking South Asian Republicans. This is called ?revealed preference,? and it is the essential element in Match?s algorithmic process. Match knows what?s right for you?even if it doesn?t really know you. After taking stock of your stated and revealed preferences, the software finds people on the site who have similar dissonances between the two, and uses their experiences to approximate what yours should be. You may have sent introductory messages to only two people, and marked a few others with a wink?a nonverbal expression of interest?but Match will have hundreds of people in its database who have done a lot more on the site, and whose behavior yours seems to resemble. From them, depending on the degree of correlation, the software extrapolates about you.
The trick is in weighting each variable. How significant is hair-color dissonance? Do political views, or fan allegiances, matter? The weightings can change over time, as nuances or tendencies emerge. The algorithms learn. And sometimes behavior changes?political opinion matters more in an election year, for example?and the algorithms scramble to keep up.
An engineer named Amarnath Thombre oversees Match?s base algorithm, which takes into account fifteen hundred variables: whether you smoke, whether you can go out with a smoker, whether your behavior says otherwise. These are compared with the variables of others, creating a series of so-called ?interactions.? Each interaction has a score: a numerical expression of shared trait-tolerance. The closest analogy, Thombre told me, is to Netflix, which uses a similar process to suggest movies you might like??except that the movie doesn?t have to like you back.?
I?ve been on two real dates in my life, both of them in my freshman year of college, nearly a quarter century ago. The first, as it happens, was with the eldest daughter of Robert Ross, the founder of TACT. We met at a party and took up with each other for a while. The date itself came later, on the first night of Christmas vacation. We went to ?Burn This? on Broadway. I remember John Malkovich stomping around onstage and then my date catching a train back to Scarsdale. She remembers that we went to a Chinese restaurant and (this hurts) that I ordered a tequila sunrise. That night, anyway, was the end of it for us.
For the next date, on the advice of a classmate from Staten Island, who claimed to have dating experience, I took a sophomore I liked to a T.G.I. Friday?s, in a shopping center on Route 1 in New Jersey. On the drive there, a fuse blew, knocking out the car stereo, and so I pulled over, removed the fuse box, fashioned a fuse out of some aluminum foil from a pack of cigarettes, and got the cassette deck going again. My companion could not have known that this would hold up as the lone MacGyver moment in a lifetime of my standing around uselessly while other people fix stuff, but she can attest to it now, as she has usually been the one, since then, doing the fixing. We?ve been together for twenty-three years. Needless to say, we had no idea that anything we were saying or doing that night, or even that year, would lead us to where we are today, which is married, with children, a mortgage, and a budding fear of the inevitable moment when one of us will die before the other.
So, for the purposes of this story, I didn?t do any online dating of my own. Instead, I went out for coffee or drinks with various women who, according to their friends, had had extraordinary or, at least, numerous adventures dating online. To the extent that a date can sometimes feel like an interview, these interviews often felt a little like dates. We sized each other up. We doled out tidbits of immoderate disclosure.
I talked to men, too, of course, but there is something simultaneously reductive and disingenuous in most men?s assessments of their requirements and conquests. Some research has suggested that it is men, more than women, who yearn for marriage, but this may be merely a case of stated preference. Men want someone who will take care of them, make them look good, and have sex with them?not necessarily in that order. It may be that this is all that women really want, too, but they are better at disguising or obscuring it. They deal in calculus, while men, for the most part, traffic in simple sums.
A common observation, about both the Internet dating world and the world at large, is that there is an apparent surplus of available women, especially in their thirties and beyond, and a shortage of recommendable men. The explanation for this asymmetry, which isn?t exactly news, is that men can and usually do pursue younger women, and that often the men who are single are exactly the ones who prefer them. For women surveying a landscape of banished husbands or perpetual boys, the biological rationale offers little solace. Neither does the Internet.
Everyone these days seems to have an online-dating story or a friend with online-dating stories. Pervasiveness has helped to chip away at the stigma; people no longer think of online dating as a last resort for desperadoes and creeps. The success story is a standard of the genre. But anyone who has spent a lot of time dating online, and not just dabbling, has his or her share of horror stories, too.
Earlier this year, a Los Angeles filmmaker named Carole Markin sued Match.com in California state court after she was allegedly raped by a man she met on the site; he turned out to be a convicted sex offender. (Twenty years ago, Markin published a book called ?Bad Dates,? for which she solicited anecdotes from the likes of Johnny Bench, Vincent Price, Lyle Alzado, Isaac Asimov, and Minnesota Fats. They suggest that all good dates may be alike but that each bad one is bad in its own way.) Markin?s suit asked not for money but for an injunction against Match.com to prevent it from signing up any new members until it institutes a system for background checks. (A few days later, the company announced that it would start checking subscribers against the national registry of sex offenders.) To some extent, such incidents, as terrible as they are, merely reflect the frequency of such transactional hazards in the wider world. Bars don?t do background checks, either.
Most bad dates aren?t that kind of bad. They are just awkward, or excruciating. One woman, a forty-six-year-old divorced mother of two, likened them to airplane crashes: the trouble usually occurs during takeoff and landing?the minute you meet and the minute you leave. You can often tell right away if this person who?s been so charming in his e-mails is a creep or a bore. If not, it becomes clear at the end of the evening, when he sticks his tongue down your throat. Or doesn?t. One woman who has dated fifty-eight men since her divorce, a few years ago, told me that she maintains a chart, both to keep the men straight and to try to discern patterns?as though there might be a unified-field theory of why men are dogs.
The dating profile, like the Facebook or Myspace profile, is a vehicle for projecting a curated and stylized version of oneself into the world. In a way, the online persona, with its lists of favorite bands and books, its roster of essential values and tourist destinations, represents a cheaper and more direct way of signalling one?s worth and taste than the kinds of affect that people have relied on for centuries?headgear, jewelry, perfume, tattoos. Demonstrating the ability, and the inclination, to write well is a rough equivalent to showing up in a black Mercedes. And yet a sentiment I heard again and again, from women who instinctively prized nothing so much as a well-written profile, was that, as rare as it may be, ?good writing is only a sign of good writing.? Graceful prose does not a gentleman make.
The fact that you can?t get away with lying in your profile for long doesn?t prevent a lot of people from doing it. They post old photographs of themselves, or photos of other people, or click on ?athletic? rather than ?could lose a few pounds,? or identify themselves as single when they are anything but. Sometimes the man says he?s straight but the profile reads gay. Sometimes he neglects to mention that he is a convicted felon. OK Cupid, in an analysis of its own data, has confirmed what I heard anecdotally: that men exaggerate their income (by twenty per cent) and their height (by two inches), perhaps intuiting that women pay closer attention to these data points than to any others. But women lie about these things, too. A date is an exercise in adjustment.
It is an axiom of Internet dating that everyone allegedly has a sense of humor, even if evidence of it is infrequently on display. You don?t have to prove that you love to curl up with the Sunday Times or take walks on the beach (a very crowded beach, to judge by daters? profiles), but, if you say you are funny, then you should probably show it. Demonstrating funniness can be fraught. Irony isn?t for everyone. But everyone isn?t for everyone, either.
I had a talk-about-dating date with a freelance researcher named Julia Kamin, who, over twelve years as a dater on various sites, has boiled down all the competing compatibility criteria to the question of, as she put it, ?Are we laughing at the same ****?? This epiphany inspired her to build a site?makeeachotherlaugh.com?on which you rate cartoons and videos, and the algorithms match you up. As she has gone around telling people about her idea, she says, ?women get instantly excited. Men are, like, ?Um, O.K., maybe.? ? It might be that women want to be amused while men want to be considered amusing. ?I really should have two sites,? Kamin said. ?Hemakesmelaugh.com and shelaughsatmyjokes.com.? (She bought both URLs.)
Good writing on Internet dating sites may be rare because males know that the best way to get laid is to send messages to as many females as possible. To be efficient, they put very little work into each message and therefore pay scant attention to each woman?s profile. The come-on becomes spam and gums up the works, or scares women away, which in turn can lead to a different kind of gender disparity: a room full of dudes. ?There is a fundamental imbalance in the social dynamic,? Harj Taggar, the investor at Y Combinator, told me. ?The most valuable asset is attractive females. As soon as you get them, you get loads of creepy guys.?
The online dating sites are themselves a little like online-dating-site suitors. They want you. They exaggerate their height and salary. They hide their bald spots and back fat. Each has a distinct personality and a carefully curated profile?a look, a strong side, and, to borrow from TACT, a philosophy of life values. Nothing determines the atmosphere and experience of an Internet dating service more than the people who use it, but sometimes the sites reflect the personalities or predilections of their founders.
OK Cupid, in its profile, comes across as the witty, literate geek-hipster, the math major with the Daft Punk vinyl collection and the mumblecore screenplay in development. Get to know it a little better and you?ll find that it contains multitudes?old folks, squares, more Jews than JDate, the polyamorous crowd. Dating sites have for the most part always had either a squalid or a chain-store ambience. OK Cupid, with a breezy, facetious tone, an intuitive approach, and proprietary matching stratagems, comes close to feeling like a contemporary Internet product, and a pastime for the young. By reputation, it?s where you go if you want to hook up, although perhaps not if you are, as the vulgate has it, ?looking for someone??the phrase that connotes a desire for commitment but a countervailing aversion to compromise. Owing to high traffic and a sprightly character, OK Cupid was also perhaps the most desirable eligible bachelor out there, until February, when it was bought, for fifty million dollars, by Match.
OK Cupid?s founders, who have stayed on since the sale, are four math majors from Harvard. While still in school, in the late nineties, they created a successful company called the Spark, which composed and posted online study guides along the lines of Cliffs Notes. At the time, they experimented with a dating site called SparkMatch. The fodder for their matching apparatus was a handful of personality tests and droll questionnaires that they?d posted on the Spark to lure traffic. They sold the company to Barnes & Noble in 2001 and then reunited in 2003 to revive the dating idea. To solve the chicken-egg conundrum of a dating site?to attract users, you need users?they created a handful of quizzes, chief among them the Dating Persona Test. A man might learn, for example, that he?s a Billy Goat, a Backrubber, a Vapor Trail, a Poolboy, or the Last Man on Earth. The Hornivore (?roaming, sexual, subhuman?) might want to consider the female type Genghis Khunt (?master of man, bringer of pain?) and avoid the Sonnet (?romantic, hopeful, composed?). They also urged people to submit their own quizzes. By now, users have submitted more than forty-three thousand quizzes to the site. Answer this or that pile of questions and you can find out which ?Lost? character/chess piece/chemical element you are.
Essentially, OK Cupid opened a parlor-game emporium and then got down to the business of pairing off the patrons. The quizzes had no bearing on the matching, and at this point they are half-hidden on the site. They were merely bait?a pickup line, a push-up bra. There is a different question regimen for matching. On OK Cupid, the questions are submitted by users. There are three variables to each question: your own answer, the answer you?d like a match to give, and how important you think this answer should be. The questions are ranked in order of how effective they are at sorting people. Some questions might be of utmost importance (?Have you ever murdered anyone??) but of little use, in sorting people. Others that divide well (?Do you like Brussels sprouts??) will not do so meaningfully.
And yet some questions are unpredictably predictive. One of the founders, Christian Rudder, maintains the OK Trends blog, sifting through the mountains of data and composing clever, mathematically sourced synopses of his findings. There are now nearly two hundred and eighty thousand questions on the site; OK Cupid has collected more than eight hundred million answers. (People on the site answer an average of three hundred questions.) Rudder has discovered, for example, that the answer to the question ?Do you like the taste of beer?? is more predictive than any other of whether you?re willing to have sex on a first date. (That is, people on OK Cupid who have answered yes to one are likely to have answered yes to the other.) OK Cupid has also analyzed couples who have met on the site and have since left it. Of the 34,620 couples the site has analyzed, the casual first-date question whose shared answer was most likely to signal a shot at longevity (beyond the purview of OK Cupid, anyway) was ?Do you like horror movies?? When I signed up for the site, some of the first things I was asked were ?Are clams alive?? and ?Which is bigger, the sun or the earth?? It?s hard to discern the significance.
The purpose of the blog is to attract attention: the findings, like the quizzes, are to lure you in. Rudder has written a lot about looks: whether or not it helps to show cleavage (women) or a bare midriff (men)?the answers were Yes, Especially as You Age, and Yes, If You Have Good Abs and Are Not a Congressman. He found that women generally prefer it when in photos men are looking away from the camera (hypothesis: less intimidating), and that men prefer the opposite (they want a woman?s full attention). A user can rate other people?s profiles. The matching algorithms take these ratings into account and show you people who are roughly within your range of attractiveness, according to the opinions of others. The idea behind the matching algorithms, Chris Coyne told me, is to replicate the experience you have off-line. ?We tried to imagine software that would be like your friend in the real world,? Coyne said. ?If I were your friend and I told you that So-and-So would be the perfect date, your response to me would be to start asking me questions. Does she like dancing? Does she smoke pot? Is she a furry? Is she tall? On the Internet, people will ask?and answer?extremely personal questions.?
OK Cupid sends all your answers to its servers, which are housed on Broad Street in New York. The algorithms find the people out there whose answers best correspond to yours?how yours fit their desires and how theirs meet yours, and according to what degree of importance. It?s a Venn diagram. And then the algorithms determine how exceptional those particular correlations are: it?s more statistically significant to share an affection for the Willies than for the Beatles. The match is expressed as a percentage. Each match search requires tens of millions of mathematical operations. To the extent that OK Cupid has any abiding faith, it is in mathematics.
There?s another layer: how to sort the matches. ?You?ve got to make sure certain people don?t get all the attention,? Rudder said. ?In a bar, it?s self-correcting. You see ten guys standing around one woman, maybe you don?t walk over and try to introduce yourself. Online, people have no idea how ?surrounded? a person is. And that creates a shitty situation. Dudes don?t get messages back. Some women get overwhelmed.? And so the attractiveness ratings, as well as the frequency of messaging, are factored in. As on Match.com, the algorithms pay attention to revealed preferences. ?We watch people who don?t know they?re being watched,? Sam Yagan, the company?s C.E.O., said. ?But not in a Big Brother way.? The algorithms learn as they go, changing the weighting for certain variables to adjust to the success or the failure rate of the earlier iterations. The goal is to connect you with someone with whom you have enough in common to want to strike up an e-mail correspondence and then quickly meet in person. It is not OK Cupid?s concern whether you are suited for a lifetime together.
OK Cupid winds up with a lot of data. This enables the researchers to conjure from their database the person you may not realize you have in mind. ?Like that guy in high school with the Camaro and the mustache who bow-hunts on weekends,? Rudder said. ?You can find that guy of the imagination by using statistics.? The database also gives them a vast pool to sell to academics. In no other milieu do so many people, from such a broad demographic swath, willingly answer so many intimate questions. It is a gold mine for social scientists. In the past nine months, OK Cupid has sold its raw data (redacted or made anonymous to protect the privacy of its customers) to half a dozen academics. Gregory Huber and Neil Malhotra, political scientists at Yale and Stanford, respectively, are sifting through OK Cupid data to determine how political opinions factor in to choosing social partners. Rudder, for his part, has determined that Republicans have more in common with Republicans than Democrats have in common with Democrats, which led him to conclude, ?The Democrats are doomed.?
