The Washington PostDemocracy Dies in Darkness

For the love of data

Talking romance and algorithms with OKCupid co-founder and president Christian Rudder.

The co-founder and president of OKCupid talks with the Post's Lillian Cunningham about the role of serendipity in love and business. (Video: Lillian Cunningham, Kyle Barss, Julio Negron and Jayne Orenstein/The Washington Post)

Though a former math major at Harvard and the main data detective for OKCupid, Christian Rudder is oddly skeptical about algorithms and the science of prediction. An appreciation for serendipity, he says, can be even more valuable — in love and in business.

Rudder, 39, is president and a co-founder of OKCupid, which purchased in 2011 for $50 million. He has a penchant for analyzing numbers (his new book, “Dataclysm,” digs into the behavioral patterns and biases that years of online dating data unearth). Yet he’s quick to point out that while a computer can capture human behavior, it can’t always predict it.

In this interview, edited for length and clarity, Rudder talks about what trends he sees in the online dating industry, the most interesting OKCupid statistics and how his approach to romance spills over into the way he manages his company.

(You can also listen the On Leadership podcast with Christian Rudder here or on iTunes.)

Q. For your book, you pored over tons of data from OKCupid and other sources about preferences and behaviors. What was the most interesting insight? 

A. Definitely the thing that surprised me the most — because it’s from a data set that I was way less familiar with — was this finding about Facebook “likes.” Just by looking at the articles and albums that you clicked on, these researchers could predict your race with about 95 percent accuracy, your gender with 88 percent accuracy, your orientation with 85 percent accuracy. At, like, 60 percent they could guess whether your parents got divorced before you were 21. That’s not even a demographic detail, that’s just a thing that happened to you in the past. That’s a pretty intense level of accuracy.

Q. What have you learned about human behavior from your own data from OKCupid?

A. The more time you spend with big data, the more it confirms your cynical intuition about people, for the most part. For OKCupid, the most sensational thing is the way men and women in a straight dating case look at age so differently. Women want a guy to be roughly as old as they are — so a 30-year-old woman wants a 30-year-old guy, and so forth. But for guys, it’s just a straight ticket vote for the youngest possible age in the data set, which is 20. 50-year-olds, 40-year-olds — they all think 20-year-old women are the hottest. Which of course is not surprising, but it’s still weird to be able to see it with that level of clarity and precision.

Q. The dating site eHarmony is planning to start a matchmaking service for employers and jobseekers. Is that something you’ve thought about for OKCupid?

A. That idea has come up at OKCupid, because there are lots of places where interpersonal compatibility seems like a relevant thing — whether it’s roommates or employers or mentors. We’ve always rejected it, because I think you’re going to have a really hard time getting high-quality data from people. It’s one thing to offer your opinions with the end goal of finding someone to share your life with, or even finding someone to sleep with or whatever. But if HR comes to you and is like, “Here, fill out this thing,” you’ll do it because you’ll get fired but the incentives to be truthful and really project yourself into the data are pretty slim. People are going to be guarded.

Q. Is it true that neither you nor any of the other co-founders have ever gone on an online date?

A. That’s true — to my knowledge, at least. I never have.

Q. Even just for customer research, did you never think it was important to experience what someone goes through when using a site like OKCupid? 

A. I’m not going to sit here and tell you it wouldn’t be useful. But at the same time, we had all been on dates. I think the dates that arise online aren’t that much different than the dates that arise when you’re talking with someone at a party and you agree to go get dinner a couple days later.

The distance, if you want to think metaphorically, between people starting to talk on OKCupid and exchanging a few messages and then actually going out and getting a cup of coffee or a beer is very short. There’s not a lot of romance on OKCupid, per se. That all happens afterwards. It all happens in real life.

But we spend a lot of time looking at and thinking about the interface. That’s a place where metrics are invaluable. You can have an intuition about how people are going to use a page or service, but it’s effectiveness or people’s willingness to use it? It’s very hard to guess those things in advance. So we launch and test things, and people use it or they don’t. Then we move on. The audience, so to speak, guides us a lot.

Q. What’s the worst date you’ve ever been on?

A. When I was 15 or 16 I went on some really awkward dates. I mean I don’t even know if they were dates. We would go to get a burger and my parents would drive us, and then we would just sit there in the back of the car, and be like, “Okay, see you at school.” Those were 100 percent the most awkward.

Q. Any lesson from dating that influenced how you run OKCupid?

A. In the past, online dating was laden, or plagued, with the expectation of finding your soul mate. EHarmony’s entire existence is predicated on the idea that you’re going to find “the one,” that they’re going to sort the crazy world out for you with an algorithm. We rejected that with OKCupid.

I met my wife when we were both dating other people. In 1999, her boyfriend’s band was playing, as was my band. Then she went to go do something else, and I went to do something else. My roommate ran into her randomly in New York City three years later. So we reconnected, and we eventually started dating. There’s so much randomness in love, you know? That kind of story isn’t really something that an algorithm can predict — that this one person I met and talked to years ago, I would later marry. That’s such an insanely hard thing to fold into a computer.

Somebody’s pictures or profile could be amazing, but they could have the most obnoxious laugh in the world. There’s so much about people you just can’t get until you’re actually sitting down with them — and computers don’t get that. We try to really recognize the limits of what we can do as a dating site. And for me, part of that was because I’ve never really been good at planning out my life, in general, but certainly not my love life. It’s always been a thing that kind of happened miraculously, and so we tried to recognize and bow before the power of serendipity in romance when we designed the site.

Q. What’s a specific way in which that serendipity infuses the business strategy?

A. We built OKCupid to be very open, where you as a user create your own algorithm, essentially. And definitely that mindset — of recognizing our limitations — is built into how we run OKCupid. We don’t advertise. We see our numbers every month and they’re good so far, but who knows what’s going to happen next month? Since it’s not like we plan to run a Super Bowl ad, or to be on Oprah, it’s hard to anticipate spikes in traffic and registrations. You just wake up every morning and you check the stats, and one day you could wake up and they could be 20 percent off and that just could be the new reality. We try to stay ahead of it, but that’s a hard thing to do. It feels like being a witch doctor sometimes.

Q. What do you think is one of the most interesting or significant trends in this online dating industry?

A. [The mobile matchmaking app] Tinder is an amazing product and it seems to have been catalytic for OKCupid. This is the best year we’ve ever had, and it’s also their second and most robust year. They brought a lot of people into online dating in general and smashed that last stone in the wall of taboo that used to be around it. More or less everybody who’s under 30 is on one of these services, or at least thought about it.

But what a site like OKCupid has to stay ahead of is that people are now coming into online dating with different expectations of what a dating product is. Apps like Tinder are very photo based, there’s no profile, they are super fast, super light, super chatty. How do we conform and react to that? And how do we make sure those people, when they’re ready to do OKCupid, find the site as intuitive and welcoming as Tinder is in the first 15 seconds of downloading it? A lot of sites have to answer that question for themselves. How do they accommodate people who are very used to very lightweight apps?

Q. So do you think that in, say, 10 years, the way we date and meet people will be more or less what it is now, just with better interfaces on these sites?

A. I think so. Sex and romance existed way, way, way before computers. It’s a weird business, because it’s so closely tied to the real world. You can’t really be an OKCupid user until you’re ready to go out and meet people from OKCupid. So I’m sure there will be some kind of innovation of course, and things will get better and all that. But I don’t know that it will be fundamentally different. The idea that you’ll have a watch that could broadcast your dating profile when you walk into bar? I just don’t know if people will get onboard with that.

Q. You’re 39 years old, and Match bought OKCupid in 2011 for $50 million. What is it like to be young and know that, if you wanted to, you could probably not work for the rest of your life?

A. Maybe. New York is expensive. But yeah, probably. I don’t know, that’s a hard question to answer, honestly.

Q. Does it change the way you spend your time or the sort of projects you take on, though?

A. I’d probably be doing exactly the same thing. That’s not to say that money doesn’t matter, because it does if you don’t have it, obviously. But right now my life isn’t any different. It’ll probably be very different in maybe 10 years, but for now I’m still working at OKCupid, I’m still friends with the same friends, I still live in the same apartment, I have all of the same clothes. But of course it’s probably going to open up paths that I don’t even know.

Q. Do you think of yourself as an entrepreneur?

A. I’m definitely not really an entrepreneur, in the sense that I’m not in a rush to get done with OKCupid so I can go start a new business. Also, the three other guys who started OKCupid with me built the business, basically. The success is almost totally due to them. Eventually I’ll probably do something totally different. I’m a very project-oriented person and I like to get things done and complete them, and I’ve done like a very wide and random assortment of things in my life. And I hopefully will work with the guys from OKCupid again. But something about the entrepreneur title just doesn’t seem like a thing to me, personally. It says that you start businesses, but it says nothing about how good the businesses are, how successful.

Q. There was a quote in your book I found particularly interesting “A flaw is a powerful thing. To be universally liked is to be relatively ignored.” You wrote that in the context of relationships, but do you think it carries over to business leadership? 

A. The middle ground is rarely the place that gets the most attention. But that’s actually okay as a manager. You can’t alienate people — and you don’t want super fans, either, because they’re not going to be honestly critical about what you’re doing. Business shouldn’t be driven by people liking you are not, like politics and art and dating often are. It’s about your ability to be fair and do a good job and make the right decisions for the company.

Q. OKCupid and SparkNotes, where you previously worked, both were built largely off quizzes. Why do you think people love personality quizzes so much?

A. People just love to hear about themselves. Even now, in different ways, OKCupid serves that need. That’s how we make money. You can subscribe to find out who’s voted on you and who’s visited your profile. All these tokens gratify your ego a little bit. Quizzes are the same way. You get your results and you’re like, “Oh I knew it! That’s exactly right!” Ego gratification drives so much of the world, and I think it’s a big part of why those tests are popular.

Q. You’re part of the band Bishop Allen. Are your interests in math and music connected?

A. The thing that music most has in common with data analysis is they both require sitting in front of a computer for just tons and tons and tons of time. Sound editing is like: spacebar. Spacebar. Undo. Spacebar. It’s so much of that all day. And data analysis is the same way.

Q. What’s a personal value you hold that has infused your approach to work?

A. I hate being told what to do, in sometimes a childish and unproductive way. I try to give my employees the same leash. You’re not just a pair of hands, where I come up with all the ideas and the plan. I’m like the opposite of a micromanager, sometimes to a fault.

That definitely dates back to when I was in high school. I hated doing homework. It seemed so pointless. So I’d often try to get my teachers to make deals with me, where if I did well on the test they would basically make the homework irrelevant to my grade, which usually worked. But anyway, I try to extend the same sort of latitude that I would expect myself to anybody who works with me.


Podcast: Christian Rudder on leadership