OkCupid and the impending AI takeover
Remember those OkCupid blog posts about the mathematics of beauty, the lies people tell in online dating, and how OkCupid experiments on its users? Those were the work of Christian Rudder.
Christian Rudder is the co-founder of OkCupid. He is also a founding member of celebrated indie-rock band Bishop Allen. He has his own IMDb page from roles in Nick and Norah’s Infinite Playlist and the mumblecore film Funny Haha. He is cited on the “Love” wikipedia page, and he’s written massively popular blog posts at OkTrends. He’s a pretty interesting fellow.
Two years ago he authored Dataclysm: Love, Sex, Race, and Identity–What Our Online Lives Tell Us about Our Offline Selves, the New York Times best-selling book that delved into what data tells us about who we really are.
The Harvard-trained mathematician got his start in tech in 1999 when he joined SparkNotes to popularize the site through a humor column. It’s a role that we’d call “content marketing” today.
Recently I chatted with Christian, and the conversation was about as interesting and eclectic as his resume. We spoke about his history at SparkNotes and OkCupid, how tech companies are using customer data, the difference between handing over information and having it taken from you, and why his next book will be about the impending A.I. takeover.
How did you first get involved with SparkNotes?
I knew one of the founders from math class. I heard they started the company through a friend. It was Chris Coyne, Max Krohn, and Sam Yagan. And then I think I was the second employee. The job was to be a humor writer. I sent in some random samples of stuff, and ended up writing about three-quarters of the stuff on the company blog, The Spark.
There were no front-end programmers; we were both the content team and the front-end team. We would just literally open up index.html and make changes. And that experience of having to create something new every couple of day was extremely useful.
SparkNotes became popular by the tests and random articles about bullshit we’d make for the Spark that high school kids and college kids forwarded around. In fact, while I was there, that part was way more popular than the study guides.
You’ve got to remember this was before anything even remotely similar was being done. This is 1999. It was me and Chris, and we were the only people doing that kind of stuff. We did a lot of random crap that the internet has since become saturated with. And we got a lot of press.
After SparkNotes was acquired, you moved on to create OkCupid right with the SparkNotes founders just as social networks were starting to take off.
Chris really had the idea of doing a dating site. We actually made community features on SparkNotes, like SparkMatch, which were a precursor to OkCupid. Out of all the founders, Chris has always been the dude with the feature ideas, and I’m usually the person who fleshes out the editorial.
OkCupid was started right around the same time as Facebook. I remember in 2004 being in the office when someone pointed out Friendster to me, which people don’t remember was way more popular than Facebook.
But we were building a social network in the dating space, and the dating space was pretty shitty. It was basically Match.com and eHarmony, and they were both super expensive and had bad user experience. So there was a lot of opportunity there for a site that worked in a more sophisticated way. And OkCupid was free to join, so that was a huge competitive advantage as well.
OkCupid had a distinctly different feel than eHarmony. eHarmony felt like it was for people looking to get married.
For sure. OkCupid was more casual, and a lot younger. It’s what online dating has become now. We more-or-less pretended that online dating had no stigma and that it was just a thing that people should be doing. Then it eventually became that way. Back in the early aughts, I’d never been on an online date. I remember when we started OkCupid, I was like, “I don’t know who would ever use this thing.”
As a less technical founder, what was your role?
Chris and Max built a site that I was able to popularize through the blog. If they hadn’t made OkCupid as awesome as it is, even the funniest blog would’ve been useless. And I may have become the public face of OkCupid through the blog, but they actually made the real nuts and bolts of the site that people like.
In those six years before I started writing the OkTrends articles, I didn’t have as much to do, but I did work on the editorial. I worked a little bit on the product because, even at the time Match bought us, we only had 11 employees. I did a little bit of everything except back-end coding. We still did lots of tests and all this other kind of stuff.
All of the various features had to be figured out. You used to be able to, like, Wiki-edit people’s profiles, and you used to be able to earn points for stuff. We tried everything. Tons of them didn’t stick, but some, like Quickmatch, have.
What led to the OkTrends blog?
We knew we had a lot of interesting data. Dating data is exceptionally interesting, even compared to Twitter’s data, and certainly way more interesting than the kind of data sets that you see at a lot of places, like Alexa site rankings.
Other than perhaps Facebook and to some degree Google, what website collects this type of personal data with any plausible amount of accuracy and thoroughness? Think about it: on dating sites we need your age, where you live, your gender. That’s the bare minimum. Those things we can get with relatively good accuracy because the user has the incentive to be accurate with those things. OkCupid gets data points that no other website has an excuse to collect.
At first we tried hiring a PR person to package up some of this stuff, but those attempts were pretty lame. So when we finally decide to just write it up ourselves–as it has through the history of me working with those three guys–it just naturally fell to me.
How did you come up with the topics for those OkTrends posts?
At first, you come upon this idea and think, “We should have a data blog. We have great data. Nobody’s really writing about this stuff!” And at that point, the ideation process was actually very similar to what me and Chris did at The Spark where it was like, “Shit, we’ve got to come up with a new test this month.”
It was a constant search for a top-level topic, and then a struggle to make that topic interesting and differentiate it from similar other posts that you yourself have written or that is already out there.
Then we’ve already written about sex, we’ve written about gender. We’re just scratching off topics, you know? So then you might have an idea like, “Hey, we haven’t talked about wealth or income or any of that stuff,” but then you look at the data and realize that it’s all garbage. Can’t do that, so what else?
What do you think about how tech companies like Facebook are collecting and using customer data?
Businesses have always experimented on customers. The very existence of a site like Facebook is an experiment. Nobody knows what they’re doing with any of these things. It’s not like there was a proven recipe for social networks before they went and launched one. All these things are works in progress. Just like any business, right?
A business like Facebook just has a different reach in people’s lives than, say, a grocery store. But in the same way, a grocery store will arrange aisles in a certain way. They’re all laid out in a similar way at this point because they figured out the ideal layout.
The trip through the store is this highly researched and organized thing, and it wasn’t always like that. They got here through iteration and experimentation. It’s the same for Facebook or Twitter, and certainly for OkCupid. That’s just how things work.
And all those things are manipulative of human behavior. Of course, that’s all a user interface is, manipulating human behavior. A user interface is a tool both for the user and on the user.
But at the end of the day, Facebook is not essential to anyone’s lives. People like to say that it is, but in the same way that they’re like, “I could never stop eating chocolate.” Well, yeah you can. If you don’t like what Facebook is doing to you or with your data, then just don’t use it. I think that is a perfectly fair argument. Or don’t tell it anything, just make up a fake account if you want to look up an ex or an uncle or whatever.
How does that make you feel about your own personal data?
I don’t really control it. It’s out there, beyond my control. But I knew when I gave it that it would be this way. Despite the media’s insistence of the contrary, I feel like most people for many, many, many years have realized this. When you tell Facebook that you’re friends with X, Y, or Z, or that you’re born in 1973, it is entered into a version of a permanent record. People have known that for verging on a decade now. Everything on the internet is more or less permanent and more or less public. But people continue to use the services and essentially volunteer this information into those spheres. It’s just a trade that we make, I guess.
But there’s a substantial difference between that kind of world and a world with a drone with a camera at every intersection attached to a facial recognition API in that that data is given.
One way or another, you went to, say, Facebook, and typed in all that information. If you really didn’t think it was going to go anywhere but on to your personal Facebook page once, well, I don’t know what to say. But you did go there and type it in and hit enter. You clicked all the buttons. In 99.999% of the cases, you actively chose to do those things. The world that could come, especially with more companies moving into the hardware world, is a world in which you don’t make a choice.
This gets into the NSA government surveillance picture. You don’t choose whether the NSA has all your information, whether it has taken your picture and knows where you are. You might not choose if your photo has been taken.
Maybe it doesn’t exist now, but the same way that Google catalogs every street, maybe they’re trying to catalog where every person is. I don’t really think that they are, but they could. Somebody could be doing this now. And I’m much less ambivalent about these kinds of data gathering things. I really don’t like them. You can’t go to an airport without people taking pictures of you. This is getting in a little bit of a grey area, but I don’t like that kind of stuff.
Right now, you can’t enter Facebook headquarters – maybe you can’t enter Mixpanel’s – without getting your face scanned. That’s borderline. People need their jobs. And if it is your job to go Facebook, you can’t choose whether to go. I don’t like that kind of stuff. That’s of a different type of thing that’s pretty bullshitty.
A real cynic could say, “Well, you chose to go into public,” but I don’t think that’s a fair argument. You have no choice but to go out of your house at some point.
You’ve wrapped up your promotion for Dataclysm, it’s now out in paperback. So what’s next? Any ideas for another book?
I think I’m going to write it about the imminent A.I. takeover. Basically take Nick Bostrom’s Superintelligence and figure out how to popularize his ideas, essentially. He expresses the concerns very well, but that book is so dense and so rigorous. It’s just impenetrable for your typical reader.
Whether that’s like a quasi-fictional sci-fi ish type of explanation, I don’t know; I’ve got to figure out how that’s going to go. That’s my main challenge. There’s a lot of very plausible things to be worried about, but I’ve got to figure out an interesting and hopefully not too depressing way to put it out there.
What is it that you see in artificial intelligence that worries you?
Computers just have deeper intellectual access to what people are doing because a lot of what people are doing and thinking goes through their phones and the internet. Imagine a computer of the past, like a UNIVAC becoming self-aware and superintelligent. Who cares, right? What’s it’s going to print out punch cards? But computers today have deeper reach into our physical world than they ever have. The physical reach these computers have via drones or via self-driving cars… 10 years down the line, who knows what a computer is going to control?
The general public’s idea of sinister A.I. is the Terminator walking around with a shotgun and shooting people. Because that’s what a sinister person would do, but it’s not what a computer will do. That’s what a single individual could accomplish, but a computer won’t be on the scale of a single individual. This is kind of the idea behind the singularity – it’s going to be an unprecedented rupture with the past.
Is the data collection and the machine learning algorithms behind Facebook and Google an early predecessor to the rise of AI?
Those algorithms are certainly the first seeds, but they’re discontinuous with real intelligence. And I don’t think it’s a series of baby steps from that to something that could dominate the world. I’m not really scared of Facebook’s algorithm. It can alter our realities by getting people to think one way or another or whatever. But, that’s not all that different than the power that an editorial board might have had in previous years at a newspaper.
The news feed can’t defend itself. The news feed can’t stop someone from turning it off. That’s a key difference. Obviously, the same type of thinking that goes into these things now is what will eventually lead to a true A.I., but it will take some kind of insight, some kind of genius break.
On the one hand, that’s reassuring. Maybe that genius inspiration never comes. But on the other hand, it could just as easily come tomorrow as it could come 15 years from now.