Introduction

Patrick: [00:00:46] My guest today is Christian Rudder, who is the first of hopefully many guests who are not investors per se. Christian's career arc seems too interesting to be true. He's best known as a founder and CEO of OKCupid, the popular dating site but I'll let him walk you through the rest. He's a data and math junkie, a musician, and very interesting thinker and business person. We discussed among other things interesting trends in OKCupid's data, artificial intelligence, the NSA, and great books on the Civil War.

Christian's Career

Patrick: [00:01:30] Good morning, Christian. Thanks very much for doing this with me today. Maybe we'll start with a couple origin stories. One very short, how I came to be in your living room this morning, and then a little bit longer one kind of your backstory since a lot of people won't be familiar. Some will be familiar with your work and your businesses but maybe not you personally. So just so people know since this is kind of an odd connection, I met Christian through Jeff Gramm who is the first episode guest on the podcast. Friends I think for a long time and he recommended that I reach out to Christian and it turned out in the course of our conversation that I was a huge fan of Dataclysm, Christian's book, so tiny little world here. I'm really excited to be here with you today.

Christian: [00:02:11] Well, thanks for having me. I'm glad to be here.

Patrick: [00:02:12] Maybe we could start with your story, kind of a quick narrative, kind of the highlights of your career mainly because it's kind of been all over the place. You've done a lot of interesting things and I'm particularly most interested in the transitions and how sometimes a career flows very naturally. It seems like you've jumped from entire different arenas to new arenas. Maybe how you got to each spot.

Christian: [00:02:37] Yeah. I mean it'll be hard to kind of connect it into a whole theory or a whole story. I mean I guess I was a founder of OKCupid, which is a dating site. That's probably the kind of tent pole of my life I guess. I was also the main songwriter in a band called Bishop Allen and I also played in the band. I was one of the people who helped make SparkNotes. I wasn’t a founder but joined very early on, which SparkNotes are like study guides, kind of modern day CliffNotes I guess. I wrote Dataclysm as you know and I guess I did that not chronologically.

Patrick: [00:03:11] I guess what was the first major, was it the band? When you came out of school, what did you do out of school?

Christian: [00:03:18] Well, when I came out of school I was actually working, I had started this, I wouldn’t even call it a software company, but I wrote this program with my friend called Report Engine which is for stockbrokers. It was like a random idea. My girlfriend's dad at the time was a stockbroker and he needed something and I made it. It was-

Patrick: [00:03:36] What did it do? Did it pull-

Christian: [00:03:37] It didn’t even pull… We weren’t even that ahead of the game. This was '98. Yeah. You entered and from statements what was going on in the account and it made all this really nice reports and charts and staff. Actually, a lot like how Dataclysm ended up to kind of tie the room together so to speak.We did that. We really didn’t know what we were doing. We sold a few copies of it but unquestionably a failure. Didn’t really go anywhere. Then, I moved to Austin where I was there for like three months and my main job was baking bread at this place Texas French Bread, but while I was there, this was spring of '99. This was like the heyday of internet boomtown. I applied for a bunch of random dotcom jobs and one of them was for the spark.com aka SparkNotes. I got that job. I sent in some writing samples. They hired me. It turns out they were also Harvard guys and I knew them. One of them from math class. I was a math major, but I was hired to be a comedy writer essentially and make basically viral content. The Spark was one of the few places that was making viral stuff back then. We had the first online purity test. I mean this was like Paleolithic internet times basically.

Patrick: [00:04:46] How did it work? Because I remember using SparkNotes in college but basically just going to book's page and using that content-

Christian: [00:04:56] I mean you're 31, right? This was way before you ever would have used it. It used to be the Spark and SparkNotes was two separate URLs. The Spark was like the stuff that brought people to SparkNotes. It was kind of a genius plan, which I didn’t think of, so I can say that. I was the Spark had tons of time wasting stuff like things to read, dumb test to take, all this kind of thing, toys, web toys and that would appeal to high school and college students and so they would pass all these things around waste their time not study while they're playing around in the Spark lo and behold, hey, student who hasn't study, check out our new product SparkNotes. Which when I joined I think only had Hamlet and maybe Macbeth so only had two guides.

We offered SparkNotes to kind of like solve the problem that we had created and SparkNotes took off and later they shutdown the Spark after I left because it was kind of one of these things, me and my friend and OKCupid co-founder, Chris Coyne, we wrote everything and as soon as we were gone there was kind of no point to continue. SparkNotes was big enough on its own at that point. Anyway, so I left SparkNotes in 2002. We started OKCupid in 2003. Really around the same time that I started the band Bishop Allen, kind of did both of those things at once. The first few years were more Bishop Allen than OKCupid, and then once I started writing the blog for OKCupid where we kind of took all our data and analyzed it, had some kind of cool graphs about race and sex and attraction and all of that stuff. That was about 2009, then I just kind of was full time OKCupid until I left OKCupid in 2015.

Patrick: [00:06:23] In the band, were you touring a lot? Were you on the road?

Christian: [00:06:25] Yeah. Those years from 2004 to say 2008 or '09 was maybe three months out of the year I was on tour. Yeah. We weren’t paying ourselves for OKCupid at that point so it was like my full-time, my only real job at that point.

Patrick: [00:06:40] It sounds like the strategy at OKCupid mimicked or took some of what worked at Spark/SparkNotes

Christian: [00:06:46] Totally, and my role was-

Patrick: [00:06:47] … in through really interesting content.

Christian: [00:06:49] Exactly. Actually, I don’t think it was… At least those first, from say my first half of my OKCupid time was very similar where I was just writing stuff that would make people realize that OKCupid existed in the same way I was writing stuff at SparkNotes that made people realize SparkNotes existed. I would work on the product and do a little bit of programming here and there but nothing serious.

Dataclysm

Patrick: [00:07:10] It's a good transition into the book, which is at least at first I think an extension of what you were doing on the blog but then morphs into a lot more. Including I think some of the best chapters of the book are the later ones where you kind of riff on things like the NSA and data in general and sort of the quantified self and the collection of data on people and sort of identity which is great. It's a book for everyone listening that I'm shocked. I read a lot of books, this is a book that should be Gladwellian-type sales. It is that entertaining and interesting-

Christian: [00:07:48] Thanks.

Patrick: [00:07:48] … and the reason is because so many books today, this kind of pop-side type books are regurgitations or replications of 10 academic studies with little quaint examples, little stories to illustrate those studies and it just feels kind of all the same. It's almost always with someone else's data and what's unique about this book is that it's your data. No one else really has access to this or had access to this so it's trends within a dataset that's very unique.

Christian: [00:08:17] Yeah. I really tried to get away from that model of here's an academic finding and here's this kind of quirky example of it and practice that we found in that one factory in Idaho. Because I knew that I had my own data to mine and I understood it probably better than anybody in the world, and especially writing about online data in general but specifically data from a dating site. It's just so inherently interesting. I feel like I had to take advantage of that myself.

Patrick: [00:08:47] This is probably our closest area of direct overlap where both of us spend a lot of time dealing with data. Obviously very different data. Financial statements versus dating information but could you talk a little bit about your background in working with data? What tools you use? How you learned things like that the people might be able to use themselves if they're interested in this field?

Christian: [00:09:08] Sure. Like I said earlier, I was a math major. I didn't actually take a statistics class. I was mostly into algebra and more abstract stuff, but a lot of the statistics and data analysis is not rocket science. In fact, the more rocket sciency it gets the kind of like, that usually implies the worse your data actually is. So at OKCupid, we never had to bring those tools to bear, but I use SQL. OKCupid uses SQL to store the data. I would use a little bit of Python or I would have first more sophisticated things. I would have programmers write a little script for me or I would just take a raw data and put it into Excel. Excel can hold, well, it can hold about a million rows, but 500,000 with some room to breathe. I did a lot of that in Excel just because I've been using Excel, the that very first software program report engine was originally built as like really arcane set of Excel macros. I love that program. I did a lot of it in Excel. Everything else is pretty much home-spun. I mean I just kind of crunched everything just using basically the raw materials. I don't ever use things like Chartbeat or Mixpanel or any of that stuff. I don’t know.

Patrick: [00:10:26] One of the notable things about the book is the charts and at the end you credit Tufte, which is sort of the gold standard of visualization of information. How did you build those? It doesn’t look-alike they were built in Excel. It looks-

Christian: [00:10:38] Oh, they're all in Excel.

Patrick: [00:10:39] They are in Excel.

Christian: [00:10:40] Yeah.

Patrick: [00:10:40] Is there some sort of Tufte package or something that I don’t know about or did you just did it?

Christian: [00:10:44] No. I just did it. Yeah.

Patrick: [00:10:45] You just kind of made it work.

Christian: [00:10:45] Yeah. You just get in there. Yeah. I mean some of those things I used using the actual grid rather than a chart. I would use the grid in there. They're kind of like key mapping ability for the different cells. Yeah. I spend a lot of time in general with that stuff. All the graphs on the blog were done in Excel. Sometimes with like JavaScript overlays that I would write so they would animate and stuff. This was 2009, '10. There were basically no charting tools, kind of off-the-shelf charting tools or out of the box charting tools. I just made everything up. Yeah. We're doing the blog and certainly the book in the second credit Tufte for I really want to get away from the infographic feel where everything is like super designed and there's just tons of illustrations and everything looks very nice in that kind of generic way. I just wanted it to seem as informational as possible or keep it as information forward as possible.

Patrick: [00:11:32] This might be a peculiar question, maybe one you've never been asked, but why did you choose the color red in contrast-

Christian: [00:11:37] Oh, in the book?

Patrick: [00:11:38] … with black in the book?

Christian: [00:11:38] Well, I had one color to work with, just that's what they told me. Given that, it almost chooses itself. Blue is too close to black. Green, it's just kind of weird. Again, it's still dark among the kind of warmer colors. Yellow is too light. Orange, I mean the red is a little bit orangey but, it's surprisingly easy to settle on red. It just kind of like, I mean also Tufte is an inspiration here. He uses the one color for endpoints on his sparklines and stuff like this in some of his books. I was like I'll just use this red basically.

Patrick: [00:12:11] It worked great. The book looks great. Let's dive into some of the data in the book and some of the trends.

Christian: [00:12:16] Sure.

Wooderson's Law & The Pratfall Effect

Patrick: [00:12:17] Maybe a fun one to start with, maybe it's the low-hanging fruit, but it's a good one. I think that people appreciate is this idea of Wooderson's Law. Can you describe what that is?

Christian: [00:12:24] Oh, man, Wooderson's Law. Yeah. Okay. Well, the Wooderson in question here is Wooderson from Dazed and Confused, which is actually the reason I moved to Austin for those few months anyway because of that movie. He says at some point in Dazed and Confused. That's what I love about high school girls, man. I keep getting older but they stay the same age. That viewpoint is basically, at least in the data, that the viewpoint of collected maleness in general. Guys no matter the age always pursue the youngest woman in the group of women that they're presented with. 20 year-old guys think 20-year-old women are the hottest that makes sense. 25-year-old guys think 20-year-old women are the hottest. Okay. 30, 40, 45 year-old guys think 20-year-old women are the hottest. It's not exactly an earth-shattering conclusion. You can open a magazine and see all the models are very young. Movie stars, certainly female movie stars, there's a huge youth SKU there. But it is interesting to see it play out in the real world in terms of the decisions like actual online daters are making in their ratings of people and how they contact. Guys basically go after the youngest women period.

Patrick: [00:13:38] It's different for women, right? The same sort of scatter plot is quite different.

Christian: [00:13:43] Yeah. The women one is way more sane. In fact, reflects what women tell us they're after, which coincidentally is the same thing men tell us that they're after. They want someone near their same age. 25 year-old woman will say she's looking for a 25 year-old guy or she'll say she's looking for a 25 year-old guy, and then lo and behold that's the age of guys she thinks is most attractive. 30 year-old woman thinks 30 year-old guys are most attractive and that's also who she told us she would think was most attractive and so on kind of up that diagonal, 45 year-old women maybe a little bit younger, but still right around their same age, whereas, again, guys a 35 year-old guy will tell us, "Oh, yeah. I would love a woman who's 30, 32, 33, 34." That's what he says on the site, goes through it's all 20 year-olds. This is something that you kind of have to wrestle with. One of the many sort of challenges of running a dating site. There's all these kind of warts on the, especially the male, but in general the human psyche that you have to grapple with and kind of try your best to rectify or smooth out so that the site will function, because obviously these 20 year-old women don't want to hear from any 45 year-old guys. We've got to make everybody happy running OKCupid so that's just one of the many kind of sociologically interesting and sort of professionally frustrating aspects of human behavior.

Patrick: [00:15:04] One of the other things I found most interesting in the book, which I felt was very broadly applicable pretty much no matter what you do. If you're in the business of offering a product or service this seems to be applicable. In this case the product or service being yourself, but the idea of-

Christian: [00:15:20] Hopefully not a service.

Patrick: [00:15:21] Right. The idea of higher variance and this kind of idea of the pratfall effect that you talk about, maybe touch on the role of variance in the success of people dating.

Christian: [00:15:33] Sure. I mean generally speaking the better looking you are the better you do on a dating site for sure I assume with a high degree of confidence that's also true in real life, right? The better you do romantically. But within say a group of people who all are the same amount of good-lookingness, a kind of iso-attractiveness band. The people who exhibit the highest variance in their appeal do by far better. By variance I mean you have, say you're a 5 out of 10. It is better to have half the people think you're a nine, and half the people think you're a one, than it is to have everyone think you're a five. That's obviously the most extreme amount of variance. Have half the people think you're extremely good looking and half the people think you're repulsive, because what we found is that it's almost like an activation energy required in terms of someone approaching you or wanting to actually meet you in person presumably and date you or whatever. The more variance you have in your look the better you do. An example of a thing that people, and our users have kind of realized this. I think this is something that people intuitively know in general. Tattoos is kind of very easy example. There's lots of people who think tattoos are super hot. There are lots of people who think tattoos are just like not for them. They don't want to date somebody with a ton of tattoos.

That works super well. Nose rings, piercings that kind of thing, crazy looking hair, blue hair, whatever, a buzz cut on a woman for example or really long hair on a guy maybe. It's almost like you want the people who are predisposed to like you to really like you and everybody else you don't even care. They might as well hate you essentially to use kind of extreme terms in terms of your dating outcomes. We found that say a 5 out of 10 that has a high degree of variance gets the same amount of attention as say a conventionally attractive seven or an eight. You kind of move up a few leagues so to speak just by being strange-looking.

Patrick: [00:17:27] That makes sense on the top end for sure, obviously, you're getting more people rating you highly. I'm sure there's a high correlation between a high ranking and wanting to meet the person, right? High variance makes sense there, but in things like book sales, for example, they've shown that actually what the best-selling, many of the best-selling books have a lot of one-star reviews and a lot of five-star reviews. Maybe there's an element of controversy or something new and different. Tying it back to the idea of like products and services, if you're just offering like an iterative generic version of what everyone else is doing, probably not going to incite a lot of interest. If you're doing something much more new and different, some people will hate it some people will like it, but maybe talk about the negative side. It makes sense more people rank high going to reach out, but is there also an effect from like a boosting effect from the negative ratings?

Christian: [00:18:17] We did find that. I mean it's definitely lower than just the power of having a lot of people really like you, but we did find that just everything else being equal having more people dislike you, strongly, slightly helps the amount of attention you get. I have no idea why. My operating theory would be that like everybody has a sense of what everybody else is what the general perception of a person is going to be. I can see somebody and I might not find her really attractive, but I have a good sense that like, "Oh, I bet a bunch of guys think she's hot or whatever." I think what happens is if you see somebody that you're into and you think that there are a lot of people that are not for a certain type of person, it's like, well, it's a turn on. It's kind of like a white knight effect I guess where you're like a lot of people aren't going to be into this girl. My message is going to be the one, I'm going to find that diamond in the rough. I'm going to come in there and say, "I really think you're beautiful. You probably don't hear…" Implied you don't hear this that often.

I think that's what's going on and that's not the best explanation of it. It's hard to talk about this stuff without speaking in kind of broad and generally macho terms. Just for your listeners, I talk like this because online dating is driven by guys approaching women. 90% of the activity on the site is driven by men, first contacts, ratings, all that stuff. Men are the initiators offline as well, but I think especially online where for whatever reason women just approach online dating a lot more passively than guys do. You know what I'm talking about, guys hitting on girls and all this stuff. The straight male drives like 80% to 90% of the traffic on a dating site.

Patrick: [00:19:56] You mentioned earlier that every guy every age likes the same. The formula for good looking girls is the number 22, right?

Christian: [00:20:05] Yeah.

Patrick: [00:20:06] Much different for women. There's this issue where surveys people will be asked what they want and very often that doesn't actually correlate with what they do or what they act on. Do you find those gaps kind of everywhere?

Christian: [00:20:21] I wouldn't say everywhere, but certainly, well, with that example for sure as I was saying, you ask guys what they want and they say they're looking for women roughly their same age, maybe a few years on either side, and then they go out and do what you just described. For race, people almost universally say race is not a factor in who they think is attractive or interracial dating is great if you ask people, but they go out and act… It's like a not in my backyard type of attitude with interracial dating, for example, where people will profess to say that it's fine, but they themselves do not want to do it generally. Yeah. You do see that quite a bit. I mean people don't know themselves. I think there's also social desirability bias where they don't want to look bad to even a computer questionnaire. Again, this kind of gap in what people say they want, and then what they go out and do trying to close that gap or negotiate that gap is a big part of running a dating site.

Patrick: [00:21:21] For me, one of the most surprising and actually kind of upsetting big bummer of a chapter in the book about trends in race in dating. Maybe touch on what you found kind of the high-level punch line and maybe we could get into that a little bit.

Christian: [00:21:35] Sure. Yeah. I mean I like this that chapter because it speaks to the ability of data to make people talk about something that they might not otherwise talk about. What's in the chapter is it basically shows that in online dating pervasively whether it's OKCupid, Match, any dating site whose data I've ever looked at, there's a very clear racial bias against black people and Asian men. When I say against black people, I'd say from Asians and from Latinos and from whites. It's not just whites versus everyone versus what are minorities, and essentially those three groups, black men, Asian men, and black women, they have essentially 75% the experience of everyone else. They get 75% the likes, 75% the messages, 75% the replies. They just have a kind of lesser experience and facing this data, working at OKCupid, I get asked and we even ask ourselves like, "What can we do about this? It doesn't feel good to see this kind of thing happening."

This trend has been present since I first started looking at the data. Every site I've ever seen, it's just a universal deep and broad trend. What can we do about this? After a lot of thought the answer for us was like, "Nothing. We can't make a change to the site that's going to make people less racist any more than like OKCupid could ever change society." If there's a Latino man who doesn't want to talk to black women, we can't just show him more black women. He's just going to say, "Well, OKCupid is not showing me what I want. I'm gone." I'm not going to change his behavior and he probably doesn't even realize he's voting or acting the way he is anyway. That was my belief. The only thing we figured out we could do would be to publicize this data and say, "Hey…" There's often a discussion of like, "Are we a post-racial society? Is racism really a thing anymore?"

Certainly a few years ago before all of the shootings were more made as public as they've been, this was a common perception where it's just like that's so 1960 to think about racial bias. We published the stuff and said, "No. This is really what's going on." That actually was our most popular blog post. It's kind of I think the best part of the book to show that there really are, there are types of people in America that have just a worse experience. Obviously, I'm looking at specifically online dating or dating in general, because I don't think it's just an online thing, right?

Patrick: [00:23:57] What's striking about it is that it's not as you say just one race with one other race. It's that when you look at other minorities, which are well represented in the sample, you don't see the same thing happen.

Christian: [00:24:11] Yeah. No. I mean Latinos are for all intents and purposes in the data the same as white people. They have the same biases against the same groups and they suffer no biases themselves. That's obviously different than the way it works in society at large here or Asian women, they are part of the majority in this sense. Yeah.

Patrick: [00:24:32] The chapter, I think it was one chapter one section, for me, was the most and you say it's one of your favorites, which I agree with even though it's kind of a negative. It seems to be the best example I've ever come across that sort of puts data behind this open question of post-racial America. How bad is it really? This is sort of a definitive, it's literally in red in the book, a red stamp saying it is this bad and it's everywhere.

Christian: [00:24:58] Well, the thing that I enjoyed the most about writing that chapter, I agree with you, it's depressing. But the thing that I guess I like the most about that chapter is that there's very little data about person-to-person interactions and how race affects those. Tons of data on how this race or that race does on SAT or graduates from high school or income or where people live and the loans that they're getting or not getting. This class of person does better than this class of person on some third party thing, a test or a loan application or whatever. There's very, very little, in fact, almost no data on how this one group of people interacts directly with this other group of people. Online dating, because we collect race and it's very reliable information and the people have to talk to each other unlike on Facebook or Twitter where you kind of can just talk to your friends who are probably like you, and the people are all strangers so there's no kind of pre-existing network laid over this whole thing, you get to look at person-to-person, how do white men interact with Asian women or how do white women interact with Asian men or black men interact with Asian women or whatever and how did those Asian women react in turn? Then, you get to ask, and then answer a lot of really interesting questions in a way you can't really, it's a lens that you don't see very often, a way of looking at race that you don't see very often.

Patrick: [00:26:22] Well, it's a great chapter, an amazing book that I highly recommend everybody check out. It's not overly long. It's probably 250 pages something like that.

Christian: [00:26:31] Yeah. It's 260.

Patrick: [00:26:32] It is to my mind one of the books that I'm surprised that not everyone has read primarily because, again, it's dealing with data that's not just, here's an academic study and here's a silly little five paragraph example of this study in the real world. It's a real, I don't want to say proprietary, but custom set of data from which we can learn a lot about one of the main aspects of our life, which is kind of how we deal, interact with other people, specifically in dating, but even in broader trends than that. Maybe we could talk a little bit about books now and some of your favorites as a reader. Obviously, you've written a book and have aspirations maybe for more books, but what are some of your all-time favorites, maybe even extra points for under the radar ones that people would not have heard of?

Book Recommendations

Christian: [00:27:19] Sure. My all-time favorite book is pretty easy. That would be Shelby Foote's three volume history of the Civil War. I've read it three times. When you're in a band, you have a lot of time just driving around. I got really into very long non-fiction. I read the three-part biography of Theodore Roosevelt, the three-part biography of Winston Churchill, tons of stuff like that, but I just love Shelby Foote's Civil War. It's easily my favorite book. Let's see. I mean I love Dune. I don't read that much science fiction or fantasy, but that's the way that Frank Herbert gets that kind of like interior-ness and makes this world that's new, but also plausible. I don't know. That book's also amazing. More recent stuff, I really liked Look Who's Back. Have you read this book?

Patrick: [00:28:07] No.

Christian: [00:28:07] It's a German book about Hitler not dying in World War II. He wakes up, it's like whatever 2010 or something like this in Berlin. He's just kind of like nobody really believes that he's really Adolf Hitler, he is. It's kind of like a farce. It's kind of amazing. It's also frighteningly appropriate in the age of… He's very Trumpian. Let me put it that way in the book. Of course, unintentionally, this was before this election cycle. That book is great. I really liked Age of Miracles. Have you read that?

Patrick: [00:28:38] Yeah.

Christian: [00:28:38] It's a sci-fi book. Read it a couple times. It's written from the perspective of, maybe she's 13 or 14 year-old girl. The world rotation is slowing down and it just has this amazing tenderness that the science in it is actually kind of cool too, but it's just a great story. Have you read Station 11?

Patrick: [00:28:58] Yes.

Christian: [00:28:58] It's similar in tone. It's also an apocalyptic book written by a Brooklyn woman, but subjects are very different, but it has that kind of feeling to it. I love that book. I love Age of Miracles. I love Station 11 too. It's a great book. I think I mentioned H is for Hawk earlier. I really liked that book as well.

Patrick: [00:29:17] It sounds like a lot of non-fiction biography

Christian: [00:29:20] Yeah. I love non-fiction. Yeah. The non-fiction biography, definitely, I've spent the most time with. I've read probably way too many biographies of Napoleon. I don't know. I kind of would go deep dives. I was like Civil War for a while, then I was Napoleon. I know it's also kind of a martial topic, which isn't really my thing, but just got into it as a thing and there's lots of stuff written about these people. The Civil War I like so much, because it touches on issues that are still just so relevant today with reconstruction went down and state's rights and Americans fighting each other and, yeah. Just race, obviously, it's just amazing. Just an amazing time in American history.

Patrick: [00:29:59] How do you find new books?

Christian: [00:30:00] I usually just go to the bookstore, which my bookstore's WORD in Greenpoint, and I just randomly browse and grab stuff. I had never read anything about Abraham Lincoln or the Civil War or anything and one day I walked in… Actually, a different bookstore. I just grabbed a couple of things off the shelf and read them and loved it and got really into it or same with Age of Miracles, just pull it off the shelf, maybe had a little card because it's a Brooklyn book. Grabbed it, and seemed good. I'm kind of willing to give anything a try. I finished most things I read though, obviously, not everything. I think the book I'm reading right now is… I say this I think because I left it at my parent's house so it's unclear if I'm still reading it, but SPQR, about the history of ancient Rome.

Patrick: [00:30:40] Cool. I've seen that on book stands. Yeah.

Christian: [00:30:43] Yeah. It's good. It's good. It's something I didn't really know anything about. I like books about stuff I don't know what… I like when I'm unfamiliar with the topic.

Patrick: [00:30:53] We talked a little bit offline about what may be your next project. I would love to combat that a little bit indirectly by first asking for you to describe what Math 25 and Math 55 are?

Christian: [00:31:05] Oh. Okay. Yeah. I mean Math 25, 55 is essentially a pair of math classes. 25 is the one that you can elect to take as an incoming freshman math major at Harvard. It's extremely hard. I didn't take it and maybe 40, 45 people do. If you do well enough in the first few weeks or you have such radiant talent that they identified in you that quickly, you get invited to take 55, which is that you're basically invited by the Harvard Math Department to take this special class, which is for people who will eventually become professional mathematicians, essentially the best. The 10 people that they pick for Math 55 are probably 10 of the best 100 their age in the world in math.

Patrick: [00:31:52] This is like the Navy Seals of math?

The NSA & Artificial Intelligence

Christian: [00:31:55] Exactly. Right. Certainly for any other math major the people who took 55 as freshmen, you have this aura about you your entire career. I mentioned the book because the NSA recruits heavily from Math 55 for its employees, from that class of people who took 55. Government workers have this reputation of being kind of lazy or whatever you think of the Post Office. You think sloth, you think indifference. I mentioned this in the book, because the NSA is not like that. They're the best of the best. They're on top of it to say the least. You can't really talk about data without talking about government surveillance, the NSA, privacy and I make the point and it still holds that what the NSA is doing is powerful. You can pretty much, even if you don't know exactly what it is, you can be guaranteed that it's being done very well.

Patrick: [00:32:53] To your points earlier about collecting data on people and how we are increasingly quantified through all the major services that we use all the time that get us coming back through these sort of dopamine snacks that Twitter retweet provides or something like that that gets us engaged to use the industry term. I guess the next step will be even smarter methods of getting us engaged including artificial intelligence. Could you kind of riff on AI a bit, concerns, interests in that field?

Christian: [00:33:23] Sure. I guess let me start with just kind of finishing that thought about the NSA. When I say what they're doing, they're doing well. I mean they're doing it efficiently and with focus, not necessarily what they're doing is good, because I think the idea of corporate data gathering and government data gathering just usually those two very separate things get subsumed to this one idea of all these guys have all this data about us and it's violating our privacy. I actually think kind of from the inside I see it as two very different issues. The corporate side has its own problems. I mean they're gathering data to sell you things, which is annoying. I don't like seeing ads any more than anyone else. I don't like spam based on the emails that I've sent. However, the way they're doing it is aggregated. Advertising sells a product to a consumer, which could be all people for Coca-Cola, but could also be for Axe body spray, 18 to 25 year-old guys or even a more narrow slice of the world, but it's never to a particular person.

No company, OKCupid, Facebook, Twitter, no company is pulling out individual profiles, individual people and asking, "What is this person been up to? What are they into? What are they like? What are they going to do next? Who are they talking to?" That never happens. It's one thing not worth the time. There's all kinds of rules in place to keep that from happening. It's just not a thing. That's the corporate side of what corporations do with your data. The government has sort of the reverse relationship. Nobody at the NSA is looking at what's going on and saying, "You know what? Let's go arrest 25 year-old guys." It just doesn't happen. They're looking at a specific 25 year-old guy. Who he's talking to? What he's doing? What he likes? What he might do next? With a name and an address and a family and it's far more pernicious and I think invasive for that specificity. I think it's important when people are thinking about privacy to separate the actual use case for the data, and therefore, the implications for any individual.

Anyhow, I think as people employ algorithms to make decisions about data or anything else. Computers are obviously getting smarter and I think I'm in that camp that thinks eventually in 20 years there will be a kind of general computer intelligence that will be as smart as a human being, and therefore will quickly become way smarter than a human being and that could have grave implications for the way the world works, the way life exists on the planet. I think you'll essentially be creating a god, a real god and that just is a frightening idea.

Patrick: [00:36:00] Is there anything we can do, I'm thinking about like Asimov's rules of robotics or something. Are there things that we can do before that happens that make that god benevolent or remain under our control? What are some of the potential solutions here or is this just, are we just on a path that we can't get off of?

Christian: [00:36:22] Yeah. I mean I haven't thought as deeply or as well on this thing as Nick Bostrom who's wrote a book called Superintelligence-

Patrick: [00:36:28] Superintelligence. Yeah.

Christian: [00:36:29] … which is great. Maybe you've already, your listeners maybe already know about it through you, but I think first of all, any superintelligence that exists, it's going to be a role of the dice no matter how well thought out it is by us. In the same way that you don’t know what's going to happen when you have a kid. No matter your genes or how you raise it or whatever, who knows? But even more so with this. But that said, I mean we obviously owe it to our children, certainly to ourselves to give a superintelligence the best odds of being altruistic and having it benefit all mankind rather than destroying it. I think the best way to do that is some kind of Los Alamos level, Manhattan Project level national, probably global coordinated effort to have everyone agree and work on together relatively in the open on this project, because I think whether it's a single government agency kind of doing it in secret, the NSA, the Chinese government, the Russian government. I mean who knows? Any government, because it will be the security apparatus or defense apparatus that's going to create this thing or a corporate entity.

I wouldn't want the programmers at Twitter coming up with the ruler for the entire world. That sounds horrible to me. Almost as bad as the NSA because who knows? There are just some jokers in Silicon Valley. I mean I love programmers, in general, certainly, everybody I've worked with. I love it, OKCupid. I would not want the programmers at OKCupid coming up with an AI. That would be horrible. I think kind of secretive fractured effort at creating this thing, everyone moving in parallel, racing towards a goal, you cut corners when you race and it's something that if the wrong corners are cut, it could-

Patrick: [00:38:11] You don’t know what might happen.

Christian: [00:38:12] Yeah. We don't know what might happen and it could destroy the world. I mean I don't want to sound like a lunatic. It's really hard to talk about a kind of AI apocalypse without seeming like a crazy person, because of things like Terminator and I don't know, who knows what else? Ex Machina, which I love both those movies, but the thing is not going to be running around doing karate chops or shooting people with a shotgun or something. That's what a person would do. I think it'll be maybe more painless death for everyone, but everyone will die rather than just a couple extras.

Patrick: [00:38:43] I've read a lot of places about the intermediate step I guess before we get to some sort of superintelligence being, automation of pretty much anything repeatable, right? That if you're 21 and thinking about a career something where you're an automaton in the sort of command and control hierarchy type company, that just doesn't make sense anymore, because machines are going to be able to do it better and more efficiently. It seems like or at least this is what the most common refrain is, okay, so focus instead on creative aspects on being the person at a frontier who creates something new and establishes it, a new niche, a new product, whatever, and then lets machines or automation handle it from there, but it sounds like if there is this level of, if creativity in some way is a reflection of our level of intelligence and I think of creativity is just recognizing patterns, if an AI is smarter than us, it could potentially take that too, take the creative component too. Then, the question is, well, what's left for us to do?

Christian: [00:39:46] I mean I think that question people are going to have to answer that question even before there's a general AI, even before there's kind of specific AIs because… I mean automated cars are going to put millions of people out of work. Self-driving cars, and so I mean these are automation in general is going to be a huge economic challenge, and therefore social challenge I think, which I would be more sanguine about if I thought the government could agree to get stuff done, but I think it's going to take some kind of, it's going to be a shock I think. What do you do when there's five million people who can't feed their kids? I think there's going to have to be some kind of redistribution of… For example, if Uber no longer employs drivers, Uber shareholders are all the richer. I think there's going to have to be essentially a tax on the use of artificial intelligence that will have to end up back in the pockets of the people that it's displaced.

Patrick: [00:40:44] One of the things that I found looking at the data is this kind of rich getting richer, obviously, everyone knows about economic inequality, and actually there's been some recent changes in that data that's at least a little encouraging at the kind of wage level, but what we found is if you look at the, let's say profit margins of the highest profit margin companies, right? There's always a group that's the highest margin, what's the margin of that group? That has been steadily rising. Meaning, more and more is accruing to the winners and there's sort of this magnified power law thing going on. Like you mentioned, you don't want one guy somewhere that created this AI and has a massively disproportionate amount of the power influence. It seems like that's impossible to guard against, because it's so much easier to start a company, to program something, one person, five people, it doesn't take a corporation anymore.

Christian: [00:41:38] Well, I mean it's not hard to guard against it if you… Well, if people, if you make the right laws and people obey them. I mean you could tax that income. Look, I mean I think it'll just be interesting to see where things are in 10 years. If for the kind of 99 or the 0.1% of the world, is it worth it to them to have whatever incremental dollars at the cost of living in fear of social unrest for example? Do we want this to be like a South Africa type situation, not racially necessarily, but just in terms of there's certain people who live under guard all the time and there's people who want what they have. That doesn't sound that attractive to me personally.

Patrick: [00:42:19] You mentioned this as a potential project of yours, exploring this more, maybe writing about it. What's your particular angle? What aspects of this big question are you most interested in and do you feel you'd have the sort of inside track on exploring?

Christian: [00:42:34] I mean I have an inside track on how kind of tech businesses are run. I would like to take my knowledge of how decisions are made in a big tech company, how computer programs really get written essentially and maybe run through a sort of speculative non-fiction walk through essentially of how it could happen. Just to make it more real. I think Nick Bostrom's book is amazing. It's extremely dense as I know you know. Its power and its density also make it somewhat inapproachable for most people. It's like whatever. It's very rigorously argued. I think I would like to honestly sensationalize the idea as a little bit more, but make them not the Terminator the kind of Hollywood version, but this could really, this is exactly what could really happen, for example. Just lay it out and maybe tell the story from a personal level on the effects it would have, what the apocalypse would feel like.

I don't know. When I talk to people about this, there's a lot of people, and certainly have never heard of Nick Bostrom, never really considered the idea that the thing that's going to help them not have to drive their cars anymore given 10 years to grow could have grave implications for the world. I just want to spell it out a little bit more. Yeah. I just got to figure out how to do it.

Patrick: [00:44:03] Are there other writers in this area of Bostrom being sort of impenetrable for me certainly and for a lot of people, that you like?

Christian: [00:44:10] Yeah.

Patrick: [00:44:12] Are there other writers that or pieces or essays or anything that you've read that you think is interesting and good?

Christian: [00:44:18] I mean I thought that Wait But Why, I can't remember the guys-

Patrick: [00:44:21] Tim Urban?

Christian: [00:44:22] Yeah. The Wait But Why, superintelligence explanation, which they talk about Bostrom a lot in that book is actually pretty compelling and very approachable. The Singularity is, of course, a very famous book. I've tried to read it. It's sitting on my nightstand. It's not very exciting to read. It's a lot of charts all kind of pointing up into the right in a logarithmic fashion.

Patrick: [00:44:45] Basically things grow out exponentially. Here's 100 cool technologies.

Christian: [00:44:48] Right. Exactly. Yeah. I mean which is cool. I mean you can't discount the vision that it takes to think of this stuff, but it's not a great read for sure. I'm working through this kind of Stanford AI commission kind of white paper right now. Honestly, I would recommend… The Bostrom is worth the time if you're into kind of slightly more technical reading. The Wait But Why explanation is really great.

Patrick: [00:45:15] You say key inside track is understanding how big tech companies work, if you were giving advice to a 21 year-old interested in, it could be finance, it could be any of these fields that's increasingly data-centric quantitative, what skill sets do you think have the most value, maybe are the most undervalued or even overvalued that people should focus on as they're thinking about their careers?

Christian: [00:45:44] I mean the skills that have been most important to me and what I've done has been a kind of real world experience or a sense of how people work. I haven't spent my entire life in an academic department, done a lot of different things. I think that is extremely helpful in analyzing data. Because certainly when you're working at OKCupid and I assume also any data analysis in a social setting, Facebook, Twitter aside from their advertising data analysis, but you just really, your goal is to understand human behavior. The behaviors that are generating the data that you see and the psychology that is generating the behaviors that are generating the data that you see and some experience with people. I know this is so cliché and stupid, but you just have to be able to, at least, make intelligent guesses about what people want from your website, what people want from the page, what they want from the button that you're doing, why they're doing things? Some of it is just you're theorizing that you can't prove either way, but you've got to make decisions based on whatever conclusion your math is spitting out.

I just think kind of humanistic education, reading a lot, doing a lot of different things really, really, really, really helps. The math is not that hard. It's what you do with the results of the math, it's always by far the hardest part. To what you said, which hopefully made it in here is picking the right metric to manage against. You've got to do that intelligently just in case it wasn't covered. At OKCupid, in the early days you think of a website, you're thinking engagement, you're thinking time on site, you're thinking page views. Well, it occurred to us very early on that those are crappy metrics for a dating site. It doesn't, what does it say about a dating service if somebody's coming day after day after day after day for two years or something like this? It looks great for Twitter. That's horrible for us. They're not finding a date.

What we ended up having after a lot of thought, trial, and error, we ended up picking this metric called four ways, which essentially conversations. How many separate deep conversations are happening on OKCupid? Because that's what we want to generate. We want people talking to each other. People come to OKCupid to meet people, how do we proxy them meeting someone? They're having a conversation on our site. Because we don't know whether people actually meet up in person or not. That was a really important decision. If we had been managing towards page views or clicks or even just raw messages, our site would have failed. Raw messages doesn't work, because people send, might be sending all those messages just to the same handful, relative handful of attractive people, for example, and ignoring everyone else. They're not getting responses. They're not happy. The recipient isn't happy, et cetera, et cetera. We really had to dig to create the right metric for success for us. I think that's because we thought about it a lot and we were able to kind of work back from what do people want out of our website and figure out a metric that we felt could measure that the best.

Patrick: [00:48:37] It seems like another way of this idea, saying this idea that what really matters is coming up with the right questions, not so much the answers. That we're much better at answering stuff now because we've got tools and data, but getting the right motivations, the right questions, the right measurables is the fact, is the search mission that will yield a lot of fruit for budding entrepreneurs, for young workers.

Christian: [00:49:02] Absolutely, and for writers frankly. I mean that was the hardest part about writing Dataclysm was asking the right questions. The data was all there. I had data from Twitter, Facebook, Google, reams and reams and reams of it from OKCupid. But I had to figure out what questions can I answer, what questions should I ask, what questions would people want to read the answer to, and to a degree that's always going to be the question for any author.

What Makes a Good Founder?

Patrick: [00:49:28] Do you think that there's a certain type of person that's best suited to, having gone through it twice yourself, the kind of entrepreneurial startup type setting? Are there people that maybe should avoid it? Maybe the answer is a negative one that it's not for everyone.

Christian: [00:49:45] I guess I mean I'm sure it's not for everyone I would imagine. I've worked with the same group of three other guys both times and things have turned up very well both times-

Patrick: [00:49:57] Kind of great. Yeah.

Christian: [00:49:57] … and so I don't know the failure case luckily enough. But I think there is a lot of luck involved. I think we, me, my co-founder Chris Coyne talk about all the time. We're smart guys. We worked hard. Things could've gone sideways very easily. I wish more entrepreneurs would recognize the amount of luck that has gone into their success, because there's a tremendous amount. That said I mean somebody who, the way to position yourself best for success and an inherently in a big crap shoot that is starting your own company, I think you have to be willing to try a bunch of stuff. I think people, I mean obviously there's counter examples, but people who doggedly stick to the same idea regardless of whether it works or not, I think that's a recipe for disappointment a lot of the time. I think you have to be flexible. We tried all kinds of garbage at OKCupid through the… We started that company in 2003, we weren't even really a thing until 2008 or '09. There were some real lean years for us.

You should be extremely persistent, because we held on through all the lean years, but you shouldn't be too beholden to a single idea in your persistence. Be committed to the idea of having your company, but don't necessarily be too committed to this one match algorithm that you swear is going to work or that people are going to love or this one idea of bringing people together, just use a dating example. It's one idea of how dating should work in defiance of whether people like it or not. You've got to keep an open mind and you've got to be able to listen to the data that's coming back.

You should be extremely persistent, because we held on through all the lean years, but you shouldn't be too beholden to a single idea in your persistence.

Patrick: [00:51:28] You said that OKCupid's growth was not this kind of classic massive exponential ramp up growth that people hear about in the most famous startups, but was much more kind of linear, slow, and steady and that there were those lean years. How do you know or how did you know, and maybe it was just pure persistence, whether or not something was worth continuing to pursue? I mean were there times in that, in those early years when you guys discussed, "Well, maybe this just isn't going to ramp in the way that we wanted it to?"

Christian: [00:51:56] Sure. Yeah. I mean we didn't know. That's the hardest thing. That is the hardest part about being an entrepreneur I think is when your idea is doing okay. When it sucks, "Hey, let's walk away." Obviously, shut it down. When it's great, when you've created Facebook, only an idiot would not realize that that's going well and that you've hit a home run and would not want to stick with it and hire a bunch of people and really throw themselves into a nascent Facebook, right? But what do you do when things are like, "All right, I don't know if it's really going to be successful. We're stable now, but it's not stable long-term. I might be wasting my time and my money?" That's very hard. It really helped there were four of us that were friends. If it had been one person, even maybe just two people who had started OKCupid, I don't know if it would have lasted through those times, because it just… Certainly for one person, no way, because you're just like, "Eh, I'm just going to go get a job."

Patrick: [00:52:52] What was the division of labor and maybe skill set like amongst the co-founders? You said there was four, right, at OKCupid?

Christian: [00:52:58] There were four. Uh-huh

Patrick: [00:52:59] How much overlap was there in responsibilities and skill set and how did that get, how did that shake out?

Christian: [00:53:04] Very little overlap at any given time in terms of the roles people were filling. I think we do have some actual overlap in kind of core skills. I mean, for example, I worked primarily on the editorial and the kind of product and the viral content marketing, aka the blog and a few other things until about 2011, at which point, I basically became the CEO of the company for the last four years and ran. I just dealt with a P&L and did almost nothing traditionally thought of as creative. But among the founders, Chris Coyne and I to a much lesser degree, Chris would work on the product, he would do the front end, he would come up with the kind of ideas of what feature is going to be on OKCupid? What features should we try? Max Krohn is a fantastic programmer. He wrote all the back end, made it scalable, made it work. Without him there would be no code to put a front end or for Chris to put a front end on. Sam Yagan was the CEO before me and he later became the CEO of the entire Match enchilada. He's an amazing negotiator, a business person. He did the things that you think of a CEO is doing. We never really stepped…

Chris and I were most similar in our roles. We both kind of worked on the content and the things that the users would actually interact with. We have a great kind of professional relationship. Max definitely had his own single domain. Sam definitely had his own single domain and that helped a lot, because that way we weren't arguing over who was better at X, Y or Z. You don't know how to do this. We definitely respected those fences between our roles. That was very important.

Patrick: [00:54:43] Obviously, it seems like one of your key advantages is the creative side in writing and music and the blog for both Spark and OKCupid. What was it like being the more traditional CEO in contrast to those more-

Christian: [00:54:58] I mean it was interesting. You know what? I became the CEO of OKCupid kind of after the blog had run its course. I was already working on my book, so I felt like creatively I kind of did what I would ever have wanted to do with OKCupid's data and that whole scene. I had a creative outlet, ie the book. Business is creative in its own way. I mean I had to figure out how to take our P&L, which at the time I took over was, say we were making… I think I took over in mid-11 and I think our final EBITDA for that year was 6 million bucks. Match bought us earlier that year because they wanted a lot more out of it and the year I left I think I made close to 40. It was a cool challenge to take, especially to take the principles of how to run a business that Match knew very well, sometimes to their own detriment, but they knew very well and lay that over OKCupid, get the employees to buy into it and lay it over the website in a way that the users didn't find off-putting.

It was cool. I enjoyed it. It was weird for sure. I mean I definitely not ever discussed any finer points of accounting or any org charts and none of that stuff. I didn't know any of that stuff when I took the role out, which of course I was promoted to because I was a founder and it's just like what I needed to do at the time.

Patrick: [00:56:21] How many employees was at it when you left let's say?

Christian: [00:56:23] When I left was maybe 32, 34, 35 something like that. It's a pretty small organizations.

Patrick: [00:56:27] You know everyone? Yeah?

Christian: [00:56:28] Yeah.

Future Projects

Patrick: [00:56:29] You mentioned a book being an ambition potentially on this kind of non-fiction narrative around AI, any other entrepreneurial seeds kind of brewing in your head?

Christian: [00:56:41] One of the things that we all really liked about OKCupid and one of the reasons that we started that business, as opposed to an ad network or something in 2003, was that it created some kind of positive real world outcome. People are falling in love, having sex, whatever based on what we were doing as a business and the business's goal was to make more of that happen in the real world. We really like that. We could have, whatever. We're all good with computers. We're all math people, we could have started some kind of an ad network or some other kind of thing and probably made a lot more money. I would like to stay true to that idea to create some kind of like positive real world thing. I'm a little tired of the internet at this point, been through it. Started at SparkNotes in '99. That's the first bubble, pre-first bubble. I've kind of like been through all that.

Patrick: [00:57:33] You've done that.

Christian: [00:57:33] Yeah. A little bit. Again, some of it I haven't done very well and there's been a lot of luck through that the whole time, but, yeah. I've done that. Yeah. I don't know. I'm kicking around a few ideas. I'm really into farm animal welfare, animal welfare in general just personally and I would love a business, whatever I work on to reflect that or

Patrick: [00:57:56] What's the background there? How did that become a passion?

Christian: [00:57:58] I don’t know. I mean I've been a vegetarian for a long time, so is my wife, so is my daughter. It's just something I've thought about a lot. Actually, I really liked Jonathan Safran's Eating Animals book. I was kind of already on board before that. I imagine most of his readers probably were, but I thought that was a great way to talk about it. I feel like it's like… It's really hard to talk about this without sounding preachy, but I feel like it's a weird, it's one of the most notable inconsistencies in most people's lives I think is how they treat animals. Lots of people have pets and love them, and then yet pay for people to torture very, very similar animals just right down the road, and then eat them. It's just an interesting thing to me with the way people treat animals. I've since become friends with this guy named David Coman-Hidy. He runs this non-profit called The Humane League. That I really love how they do things and talk to him more. I've just kind of, whatever. I've just bought into it. Everybody's got their ideas but…

Patrick: [00:59:06] Very neat. Are there other entrepreneurs or budding companies, we've talked a lot about writers and books, but other ideas that are being acted on or in the early stages of being acted on that you're following with interest or appreciate today?

Christian: [00:59:22] Well, I mean Chris and Max from the OKCupid team, they started this company called Keybase, which is this kind of attempt to make essentially unbreakable cryptography accessible to everyone. You don't have to understand PGP or one-way functions and hashes and all this stuff to make sure your documents and your communications are completely private from Apple, from Google, from Iran, from the NSA. It's a really hard problem actually because the math and it requires complexity. It's a very hard problem to essentially hide that complexity yet make it actually work. They've been working on that. It's called Keybase. It's really cool. I mean I'm an investor, full disclosure, but I think it's interesting regardless. Yeah. I mean there's a variety of kind of smaller companies that I think are doing cool things.

Yeah, I work a little bit with this company that's trying to make online eye tests so you don't have to go to the ophthalmologist to figure out that you can't see or get your prescription. I think that's like a pretty cool idea. I guess I don't really follow… I'm not really a big entrepreneur scenester. I'm too old. Yeah. I try to spend as little time as possible, though I check my phone as much as anybody else, but I hate myself every time I do it. I try to spend as little time as possible like screwing around on the internet, because it sucks up a lot of time. I want to set a good example for my daughter. I'm not the world's like biggest tech enthusiast I guess.

Patrick: [01:00:57] Right. Thank you so much for all the time today and hope everyone enjoyed the talk.

Christian: [01:01:03] Yeah. It's my pleasure, man. I had a good time.

Created by: Tim Roe