EPISODE 1646

[EPISODE]

[0:00:00] ANNOUNCER: Chess.com started in 2007 and grew steadily in the years following. The platform exploded in popularity during the pandemic, to the point that their servers struggled with the traffic. It was a great problem to have. Chess.com was instrumental in helping to elevate chess to its current height of mainstream popularity. But how did chess.com come to be? And how was the platform created?

Our guest today is Jay Severson, who co-founded chess.com and was a CTO. He joins the podcast to talk about the origins of the site, its development, scaling the platform, and what he's working on today. 

This episode is hosted by Lee Atchison. Lee Atchison is a software architect, author, and thought leader on cloud computing and application modernization. His bestselling book, Architecting for Scale, is an essential resource for technical teams looking to maintain high availability and manage risk in their cloud environments. Lee is the host of his podcast, Modern Digital Business, produced for people looking to build and grow their digital business. Listen mdb.fm. Follow Lee at softwarearchitectureinsights.com and see all his content at leeatchison.com.

[INTERVIEW]

[0:01:23] LA: Jay, welcome to Software Engineering Daily.

[0:01:25] JS: Thanks for having me, Lee.

[0:01:26] LA: So, chess.com wasn't your first founding success in the gaming space, right? You created your first Internet gaming company, I think, back while you were still in college, is that correct? 

[0:01:37] JS: Yes. Back when I was still a computer science major at San Jose State. I started another company called World Gaming Federation. My goal was to provide ladders and tournaments for people that were serious about gaming. So, the big games that I was playing back then were things like Warcraft 2, and then even a lot of the Yahoo classic games, the card games, and stuff like that. So, I just wanted a tournament system for that, so I created World Gaming Federation, and worked on that, kind of out of my dorm room, while I was still going to school.

[0:02:06] LA: Cool. So, now that company you ended up selling, is that correct?

[0:02:10] JS: Yes, correct. After about a year, we had some pretty decent tournament technology. We were running tournaments, and the other big player in the space at the time was called Case’s Ladder or Internet Gaming League, IGL. They were very big on the ladders and leagues but had no tournament technology. So, they kind of approached me and said, “We'd love to use your technology and integrate into our platform.” I was excited about getting an actual paycheck and some consistent revenue stream while I was going to school to pay for bills. So, I sold it. It wasn't a huge payday. But it was a nice confirmation of an affirmation of what I was building. So then, I ended up working for Case’s Ladder for a couple years, or I think, it was three years, and building out their tournament technology there.

[0:02:51] LA: Then, was it after that? Is that when you started chess.com? Or was there something else in between there?

[0:02:58] JS: No. It was a pretty big break. I worked with Case’s Ladder, like I said, for a few years, finished up my degree in computer science. Case’s Ladder ended up getting bought by a private investor and they wanted to move everybody up to Seattle. I was in San Jose at the time, did not want to make the move. So, I decided to leave Case’s Ladder, got a job as a web developer at a mortgage technology company here in San Jose, and worked for them for four years. Their name was Myers Internet. It wasn't until quite a number of years later, Myers Internet got purchased. I've got a very small paycheck for my four years of work there on my equity, and so I was pretty disappointed and was kind of motivated to go out and do something on my own. And at the same time, my old friend from college had reached out to me and said, “Hey, I happen to meet the guys who own chess.com domain. Let's go meet with them and see if we can do something interesting with that platform.”

We had already done a smaller venture together, wholesalechess.com, that he had started, and wanting me to build the shopping cart system. So, I had already worked with him a little bit. He sold that business, came out here to the Bay Area, was going to Stanford to get his master's degree, and that's kind of when he reached out at the same time that I was a disgruntled employee at Myers. So, I said, “Sure, let's do it.” I started writing some code.

[0:04:11] LA: Cool. So, did you buy the domain? Or did you partner with the people who own the domain then?

[0:04:17] JS: Yes, it was kind of a funny story. When he first asked me if I was interested in doing chess.com, I just said, “No, I'm not interested.” I had seen a lot of chess companies come and go. There was a kasparovchess.com that had kind of raised a bunch of money, built a platform, and then gone under. I knew that there was not a lot of money in this space. People want to play chess online, but they don't want to pay anybody to do it. There was Yahoo chess, which was a free place to play. So, I wasn't super excited about spending a lot of years of my career, building anything in the chess industry. I told him, “I wasn't interested.” He said, “Let's just go meet the guys and see how the meeting goes.” We met him in Palo Alto, had some dinner and he said, “Jay, why don't you tell these guys why you're so excited to build the platform.”

So, kind of put me on the spot. I talked about the different features we were thinking about adding. They had the domain, but they had built a desktop software, and they were using it as a marketing platform, and they were several million dollars, I think, in debt, from all the money they had spent building their desktop software. So, they were looking to use the valuable asset that they had, the chess.com domain, and partner with some guys who knew how to build tech, which was my partner, Erik, and I. Originally, we were just going to lease it from them. It was going to be a co-owned company. They would get some equity. We would get some equity. And there'd be certain proof points along the way to make sure we were growing the business.

But shortly after we made that deal, and we started writing some code, they actually filed for bankruptcy, because I think, they were trying to get all the debt off their books. So, my partner and his friend went on to the auction phone call and ended up buying the domain and all their assets for a little over $50,000.

[0:05:57] LA: Wow. That was relatively cheap for something that – for chess.com, which is, at that point in time, this would have been, what about, 2010 or something like that?

[0:06:08] JS: No, no. Way earlier, this is like 2004, 2005.

[0:06:12] LA: 2004, 2005. Okay, but still, even back then, the well-known domain names like chess.com and those sorts of things would have been very, very, very valuable assets.

[0:06:23] JS: Yes. It was definitely a bargain, considering the fact that they also got all the tech. Now, the tech wasn't super useful in its current form. We had to kind of retool it to make it more web-ready and all that. But $50,000, for something they had spent $2 million developing, it was definitely pennies on the dollar. We got really fortunate to get those assets. I don't think we'd be anywhere close to where we are today if we hadn’t built this platform on top of chess.com, the domain. I mean, it was a huge advantage for us in growing the business as far as SEO. Because back then, if you typed chess into your browser, the browsers back then would just autofill .com for you. So, anybody who was going to their browser and search for chess kind of naturally ended up on chess.com. So, we were able to leverage that advantage in growing traffic and acquiring new users and eventually, kind of getting us to where we are today.

[0:07:14] LA: Right. I'll tell you, I used to play chess a lot as a kid. I rarely play now. I just don't have the time. But when preparing for this interview, I started fooling around the chess.com. I'll tell you, one of the things I really like about what you do isn't just the chess games. It's the puzzles. I love going through the puzzles and doing those. I love that sort of a mental stimulation. I don't have to invest in an entire game. I don't have to interact with anyone at the other end either. So, I love the puzzles. What do you find most people are interested in? Are they interested more in the interactive games with a person that they don't know on the other end? Or are they mostly interested in the puzzles, or playing a computer? What's the primary driver for why people use you?

[0:08:02] JS: I definitely think the majority of people that come to our platform come to play chess. I mean, that's definitely what pulls them in. It's funny, you say puzzles, because when we first built the platform, it’s not even close to what it looks like today. I mean, our goals initially were very, very small. We just wanted to basically route traffic to some other destination and capture eyeballs along the way, because we were, basically, a one dev shop. My partner didn't know how to code. So, he was kind of the visionary and I was the builder.

Originally, it was like, “Okay, let's outsource some of this work. Let's find some off-the-shelf third-party software we can use to just get like forums up and running and get like an email solution, and we’ll offer people like an ad chess.com domain name.” Initially, when we launched this, literally what we had, we had like a directory of other chess websites that you could go and find places to play, find places do tactics, find places to read about chess, to buy books. We're just a directory of other chess websites. You could sign up and create an account and get @chess.com domain name. So, you could have like lee@chess.com and people loved that.

It wasn't very ambitious. We didn't have grand plans to becoming the worldwide chess platform of all things. It's like, the more we build, we were kind of like, “All right, instead of sending the people off to this website to go play tactics, what if we built our own tactics trainer, our own puzzle solver?” I was like, “I think I could build that.” So then, I'd start putting something together, or we would talk to the other websites, see if they were interested in working with us or partnering.

I would say 9 times out of 10, we would talk to people and they’d say, they had no interest in working with us. So, we would just be like, “All right, we’ll just try to build it on our own.” So, we spent one day, I mean, literally, I spent one day throwing up like a Java applet where people could play against the computer and it was very simple. It was off the shelf, third-party, kind of open source deal. Then, we spent months building this kind of correspondence email chess where you could actually make moves in your browser and it would send the move off to somebody else who would then they would take their turn. It's what we called daily chess or email chess now.

We spent a lot time on different things and then we kind of came back and we look at traffic. It was like 80% to 90% of the users that came to our platform, we're just playing on this really terrible play versus the computer Java applet. So, you're right, and that I think a lot of people that come to the platform, I think, initially are very intimidated by the idea of playing against other real-life people. So, they want to do something that's a little bit more of a solo experience. They want to do a puzzle or they want to just play against the computer.

Now, chess.com has spent years building out the solo player experience, where we have lots of different personalities, and you can play against famous people. You can play against Magnus Carlsen, when he's 10 or when he's 13. For a while, you can play against Beth Harmon from Queen’s Gambit. I'm not sure we still do that, because there was some copyright issues there, whatever, with Netflix. But I think that's a big draw for people that’s coming to the platform, play chess, but do in a way that is a very kind of individual experience and not intimidating, where it's like, “Oh, man, I don't want to go up against some real person and get killed.” For some reason, it just feels a little more intimidating.

[0:11:05] LA: Right. You didn't set out to build the SaaS application. You set out basically to build kind of like a Yahoo for chess. It’s kind of what you were trying to do initially, but it evolved over time. Essentially now, you're more of a SaaS gaming platform than anything.

[0:11:22] JS: Correct. Yes, that's right. I think we just kind of went to where the traffic and the users drove us to go. I mean, we just kind of listen to the community. I think, initially, we were a very content-heavy application. So, one of the first features we launched was the ability to blog. You could have chest.com/blog/lee, and it would have like your chest.com blogs, and you could write about anything and everything chess. We had a lot of users in the early days who came in and they wrote all kinds of amazing chess articles about the history of chess, about famous chess players, famous chess games. They weren't paid contributors. They just wanted to write about chess, and they loved that we were giving them an online space to do that.

What that meant for us was that over the years, we became very SEO-heavy for all longtail searching. So, no matter what you're searching in chess, if you want to look up like Bobby Fisher's greatest games, there'll be an article about that on chess.com. And so, you’d end up in our platform reading about it.

Being a content platform first, really allowed us to win the SEO race against other chess websites that were out there. Then, I think we focused on the play features later. Then, once we kind of had all that stuff done, we were like, “Well, how are we going to monetize this?” Because we were getting a lot of traffic. We're getting a lot of users. Ads don't pay that well. So, we knew at some point, we needed features that people would be willing to pay for and some sort of freemium service, so that we would provide 90% of the stuff was free, but for our power users, they might give us a monthly or yearly subscription paycheck.

So, we decided that people were willing to pay for training tools, that if we provided all the play tools for free, but said, “Hey, if you want to get really good, you can subscribe.” So, we took the old desktop software that we had purchased, turned it into lessons, because it was basically a database of a lot of different positions with Grandmaster content and comments on top of it, and we rolled that out into a package that was included as part of the subscription. So, if you subscribe, you got unlimited puzzles and tactics per day. Whereas a free user might only get three puzzles per day. You might get access to the first five lessons, but if you wanted more lessons, then you would subscribe.

It was kind of like play for free, and then subscribe to get better. So, we would hope that people would get the chess bug, have a desire to get better at the game, and then eventually be like, “You know what, $100 a year is a pretty good value for the access to all of this amazing content.”

[0:13:52] LA: So, is most of your revenue now coming from those premium experiences, versus advertisement?

[0:13:57] JS: Yes. I don't know the numbers. I mean, just so your listeners know, I was there from the very beginning till 2017 as the CTO. I have not been there since 2017. Although, I do keep in touch with all the guys there. Still good friends with all the head VPs and co-founder, Erik, and all those guys, and I do go to the company meetups every year. I get an update occasionally. But I think that balls got throw out a number, I would guess that 80% of chess’s revenue comes from the subscriptions and 20% comes from advertising.

[0:14:27] LA: For a site that started with content first, that ratio is really, really good. One of the problems with content sites is you're very limited on your ability to monetize beyond advertisement. Because a lot of people don't really like to pay for premium content that they believe they can get for free. But it sounds like you found a good niche where it's not just content you're giving away, it's actual training and experience and that's got a lot more value.

[0:14:58] JS: Yes. I think too, for anybody who has gone down the competitive chess road to any degree, meaning you decided to play in a tournament, or there's tournaments all over the board, over the board tournaments that you can go play. I would say, a very small percentage of our users have done that. But for somebody like myself, who has gone down that road before, you very quickly find that you're spending a lot of money on books. That was the old traditional way of getting better at chess, is you would buy chess books. One chess book might be 20 bucks. And before you know it, you spent a couple 100 bucks on chess books.

So, if you look at our platform, in that sense, we're giving away volumes and volumes of Grandmaster content and videos, lessons, tactics, puzzles. They're structured in such a way, I think, that it's much easier to consume them, too. We're going to look at your games, and we're going to say, “Okay, he doesn't recognize knight forks, or he doesn't recognize bishop pins, or he doesn't recognize pawn structure. So, we'll actually recognize certain patterns in the reasons of why you're losing your games, and will recommend certain lessons to you, or certain puzzles to you, based upon the way that you play. So, it's a much more interactive AI, if you will, approach to actually making a player, a better chess player.

I think people see the value add. It's like 100 bucks for them to analyze my games, tell me what I'm doing wrong, suggest lessons to me, suggest puzzles to me, suggest videos to me. And all those things people realize, are not cheap to create. We have grandmasters that work in the chess industry for a living and produce content. So, by them subscribing, they're keeping chess professionals. People who have dedicated their whole lives to the game, employed, and giving them work. So, I think people understand it is a virtuous cycle, and that they're helping to grow the chess ecosystem and make it a viable living for people that actually have spent their lives dedicated to the game.

[0:16:51] LA: So today, in today's implementation, how much of the strategy and the training, like you we're talking about different ways of seeing what type of training is best for certain individuals, et cetera. How much of that is the traditional database-driven approach that the original, the old-style chess AI is used to do, versus modern artificial intelligence techniques that are available today, up to and including Generative AI?

[0:17:21] JS: It's starting to get a little outside. Like I said, I'm five to six years behind where they're at now. But I do know, it's changed a lot. I mean, if you were to rewind the clock back to Chess Master and the early chess software that was around. There was even like Sargon. There was old chess algorithms out there for quite a while, and then you kind of fast forward, and then there was like Deep Blue, and it could process millions of moves per second. Then, fast forward to today, there's completely new algorithms that are figuring out better ways to correctly evaluate chess positions.

Our goal is to take chess engines, and figure out how to make that interesting to a human being. So, it's one thing for a computer to say, why it is better here by four and a half points. But for it to actually explain to a human being why it's better, is a completely different puzzle to solve. It's not easy. But that's, I think, the key to unlocking a very good computer analysis tool for you in your games. Because if we can actually say the reason why it is better here is because your king is weak and your pawn structure is bad. Your pieces aren't developed or whatever, that actually is meaningful to humans.

For a long time, people would say, “Well, computers just play better.” I mean, computers have played better than humans for quite a while now and it's not even close. I mean, the best players in the world, Magnus Carlsen cannot beat a computer at all, ever, not even one game out of 100. So, they've gotten to the point now where they're playing on a much different level. So, unlocking the ability for those top players and even for average players, to actually take advantage of what the computer is doing, is kind of what we're at the precipice of doing a chess.com. It’s like how can we take computer analysis, computer AI, and make it meaningful to humans, so humans can actually get better at the game. I think that's where there's a lot of value add. That's part of the reasons why people do go premium, is because they see that we have at the end of every game, you can do a game analysis, and it'll actually summarize and say, “Here's how many blunders you made. Here's how many mistakes, dubious moves, great moves.” And some people, they get thrilled when they say, “Oh, my gosh, in this game I had four great moves, one brilliancy and people love brilliancies.”

So, all of those stats come from our ability to actually pick apart a position and say, “How hard was it to find this move and to make this move? Were there other equally good moves? If not, then that move qualifies as like a great move.” Just trying to pick apart chess positions and figure out how to actually score them and give that feedback to the user, is what we've kind of been after. It's definitely kind of the holy grail of chess, I think, going forward, is using computer AI to actually make the game even more interesting to human beings. Because I think when Deep Blue beat Kasparov, and it was the beginning of the end for humans in chess. Everybody thought the game was going to die.

What we've seen instead is computers have actually made the game way more interesting, because now you almost have like a cheat sheet of perfect moves of like, in this position, what was the best move according to a computer? And that's also how all our cheating algorithms work, as we kind of look at human moves and say, “What's the likelihood that they're going to make a perfect move, 5 to 10 moves in a row?”

So, computers and chess have really made the game very, very interesting. And there's lots of different engines out there that analyze positions differently. There's a whole computer chess league where engines play against each other all the time, which is fascinating. So yes, I could ramble on that for a long time.

[0:21:00] LA: No, no, that's great. In fact, I was going to ask about the tournament aspect, given your initial early background in gaming tournament, was there ever any thought of bringing a tournament platform into chess.com, or was it just enough of that out there, and the training aspect was better?

[0:21:18] JS: That was one of the projects I still built while I was there. I built a tournament system. So, tournaments are hugely popular on chess.com today. It's a big piece of the platform. You can join tournaments of all different shapes and sizes, different players, different formats. That was a big piece that I built while I was still there. There's amateur, kind of user-run tournaments, where you as a user can go in and set up a tournament. You can invite all of your friends. You can invite strangers, and there's tournament director status, where you get to pick and choose a format you administer, and you kind of run it yourself. There's official chess.com-run tournaments. Then, outside of that, there's actually like the professional tournaments, where we run tournaments exclusively for the top players in the world.

Speed Chess Championship would be an example of that, where we take like the top 16 players, and they a play particular format, and they're playing for, I don't know, $10,000 or $25,000. I'm not sure what the stakes are now. But they're playing for pretty decent prize pools. So yes, tournaments are a huge part of it, which kind of leads me to talk about. eSports is a real thing in the chess space now. You have streamers who are playing chess, and we have real live commentary that's going on. So, every time we hold these events, it's a big deal on Twitch with a lot of viewership, and great commentary, top-notch budgets. So, it doesn't just look like it's somebody's YouTube channel. I mean, these are professional events, with professional commentary, with big prize funds, and lots of viewers, sponsors, the whole deal.

So, I wasn't a believer that chess would ever be really an eSport, but I think with the right format, shorter games, and great commentary, it's become a real thing. The other thing that's really cool about the way we've done eSports is, while you're watching two grandmasters play, we have an eval bar, an evaluation bar, that shows the viewers exactly like who is winning the game. I think, rewind 20, 30 years ago, imagine you're watching Garry Kasparov or Bobby Fischer, or whatever play a game. Not only are they six-hour games, which are extremely long and boring for a lot of people who aren't into chess, but you would have no idea what's going on in the game. You could look at a position and if you're not a grandmaster, you really have no idea who's winning or why.

But now, you kind of have this computer eval bar, which is saying, “Hey, white is better here by a little bit.” And then white will make a move, and all of a sudden, you'll see the eval, bar drop and black is now all of a sudden winning. So, the commentators can react to that and be like, “Oh, my gosh, white just made a terrible move. He must have missed something. Now, black is better.” So, the real-time computer feedback of any given position, makes watching two grandmasters play a lot more interesting from a spectator point of view.

[0:23:57] LA: I'm thinking the whole industry of sports commentary, and it was related to chess in particular, what type of skills does it take to be a chess commentator? I mean, you obviously have to be an expert at chess. But it takes a lot more than that to make it interesting to take a chess game and turn it into something that commentary makes it interesting for people to watch, et cetera. So, I imagine finding commentators is probably a hard thing to do.

[0:24:22] JS: It is. It's a really unique skill set too, because if you think about your average chess grandmaster, somebody who has spent a considerable portion of their life dedicated to the game, they're not typically extroverts. These are people who have had their head buried in a game where you don't talk to your opponent. A game that's played six hours in silence. You're seeing moves in your head. It's hard to then say to this person who's not used to being outwardly social to say, “Hey, sit in front of this camera and here's a mic. You're going to talk to thousands of people. Make it interesting. Make it exciting. Somehow, become an extrovert, become an announcer, a broadcaster, at the same time, be a grandmaster and understand the nuances of this very complex position.”

It has been very tricky finding great commentators in the chess space. I think not only that, too, but it's like, we have learned that there's a place for the grandmaster and the expert who can accurately break down a position and kind of predict what's about to happen. You need those people commentating on the games. You also need the filler, who can just sit there and talk about what they're thinking, what they're feeling. The history of the backgrounds of the players, any past games that they've played. So, you kind of need that personality. But then, you almost also need a personality that can relate to the viewer.

So, we've found that actually having somebody who is commentating, who isn't a grandmaster, or an international master, but can ask a question that a novice or an average chess player might ask in a given position. Why did he put his bishop there? Why hasn't he castled his king yet? Sometimes, a spectator is watching and even if you're listening to a grandmaster commentary, the grandmasters commentating but he's doing it from a completely different viewpoint than your average spectator is. It really does take a blend of different personalities on the show to pull off a really successful broadcast. 

[0:26:16] LA: That's actually – this itself is quite interesting. I think that whole space of sports commentary and what works and what doesn't work, is itself quite interesting. Seeing it apply to chess, like you say, it's not a sport, I would have historically thought would have been conducive to eSports. 

[0:26:34] JS: Yes, exactly. I agree. I mean, I think when I was first pitched the idea, I think Danny Rensch was are a big proponent of it. He's an international master chess player, kind of became the face of chess.com early on doing a bunch of live streaming and things like that. And he was like pushing eSports and chess as a spectator sport very early on. I was always like, “I don't want to waste time and resources and money doing this, because I just don't think people are going to be interested in it.” I'm glad I was proven wrong, because it's actually become a pretty big deal, and it's a big reason why the game and the platform have continued to grow.

[0:27:11] LA: And sponsorship is really the thing that keeps that industry going. Is that correct?

[0:27:16] JS: Yes, sponsors are big. But really, I think it's more about branding, recognition, and just new user acquisition. I think, when you have an eSports event, and it's being streamed on Twitch, you're starting to get cross-demographic promotion to other users on Twitch that aren't necessarily chess players. So, one of the big events that we have run in the past that's been successful is an event called PogChamps. Now, pog is like Internet slang for like a beginner. So, we had pog champs where basically you took a bunch of famous Internet celebrities, famous streamers, who played things like Fortnight or PUBG, or other games, League of Legends, and they were well known on Twitch already, but they weren't chess players.

We took all those people, put them into a chess tournament, and they kind of had grandmasters as their coaches. They would go through some training. Then, we would pit them against another famous celebrity, and they would basically play two beginner games of chess. So, you're watching blunders, and horrible moves, and mistakes. But you're also watching their grandmasters kind of facial reactions to the moves they're making. So, you are kind of seeing the coach's reactions, and you're getting good entertaining commentary, because people are like, “Oh, my gosh, these people are terrible chess players.” But it's fun to watch normal people play chess, and not always grandmasters.

What that did is it really introduced the game to a whole new demographic of people. It made the game more accessible, because instead of being a game for geeks, it was now a game for the cool kids, because all the cool streamers were playing it. So, this was right around COVID, and also, kind of right around Queen’s Gambit. So, we kind of had a trifecta of a perfect storm, where it was like people are stuck at home, people have nothing to do. They're learning a new language. They’re learning how to play chess. Then, Queen’s Gambit comes, out and everybody wants to learn how to play chess because of that show. And then, PogChamps happens. So, everybody's seeing all their famous celebrity personalities playing chess. We even had famous football players, and athletes, and actually real-life Hollywood celebrities playing. I think, the Mountain from Game of Thrones was in one of our tournaments, people like that.

So, all of that happened. And then before I knew it, my son who was a teenager at the time, and I couldn't get him to even look at a chessboard. All of a sudden, he's coming home and he's playing chess on his phone because all his friends are playing chess on their phones. All the kids on the basketball team and the baseball team are playing chess. So, it was a big change culturally, I think, almost worldwide, to where all of a sudden, it was like chess is now a game for the cool kids. That's when we really kind of hockey sticked in a big way.

[0:29:52] LA: Yes, this would have been 2020, maybe ’21, in that time period. Is that correct?

[0:29:56] JS: Correct. Yes.

[0:29:59] LA: So, that was your big hockey stick. Was that your first big growth spike? Or did you have other growth spikes earlier on?

[0:30:05] JS: It's all relative. A big growth spike to us in the first couple years would have been like a hiccup today. So, I think it's just all very relative. I remember there was one time we were getting – we used to measure things kind of in daily user signups. That was our big, exciting metric. I remember the first time we had like a thousand people sign up in a single day, we were just like flabbergasted. We're like, I can't believe we got a thousand people join the platform today, in comparison, and we have upwards of 200,000 people signing up per day now.

So, it's a very big difference. But I remember, I think it was the first year or two after we had launched our iOS application, and there was a whole transition into mobile, which was a whole another story and tech problems to solve. But for a long time, we were just a web platform. Then, it was like, “Okay, I think we need an iOS app.” So, we had to hire like an iOS developer. But a couple years after we launched our iOS application for the iPhones, we noticed, there was like this big spike in users, and it just like wasn't going away. It was like, “Man, every day, we're getting like a like double, triple the number of signups.” Then finally, I think it was after, I felt like four or five days or a week or something, somebody realized that we had been put on to the App Store’s featured apps page. Nobody told us, nobody warned us. But all of a sudden, it was like chess.com was a featured app on the App Store, like the first thing you would see when you go to the App Store. Just little things like that, that are completely out of your control, and all of a sudden, you're getting 3x to 5x the number of signups that you're used to. Back then, it’s like our tools weren't great. So, we didn't recognize it right away.

Anyway, there was definitely spikes like that were just things we didn't even do. There was things like the World Chess Championships would happen. So, all of a sudden, the entire world is tuning in to watch Magnus Carlsen play a big game, or Garry Kasparov, play a big game. Again, another event completely out of our control. All of a sudden, our signups would go through the roof. So, they tended to be events that were outside of our control, and we just had to be sure that our platform, our technology was ready, and that it was grade A. So, that when people typed in chess on their browser, it ended up at our platform that we would actually convert that user to an active user who would come back to the platform because they were impressed with their first-time user experience, and the tools that we had and the games that they could play and things like that. Same with COVID. COVID was definitely out of our control. Queen's Gambit was out of our control. So, I would say the majority of our hockey sticks have been accidental and out of our control, and we just happen to be the beneficiaries of a big wave that kind of surged our direction.

[0:32:46] LA: It's great that you were able to take advantage of those. A lot of companies will get the opportunity to grow, but their platform won't support the growth because it wasn't planned, and they'll end up struggling because of that. But you were able to keep up with demand.

[0:33:01] JS: Yes, exactly. The alter story to that is I've dabbled in lots of different things over the years, but I built a game called Wisecrack for mobile phones. It's basically just like a party game. You write funny questions, you write funny answers. But I didn't spend a lot of time on it. I was like, I hired a couple of people to help me build it, because I just really wanted it during COVID. So, I can play party games with friends over Zoom. All of a sudden, I'm getting tons, like, literally hundreds of complaints are flooding in about I can't play the game. I really want to play. JB is waiting. JB this and that. I was like, “What is going on?” It was literally probably a month after I put the application into the App Store. 

Justin Bieber found it and tweeted about it, wrote it on his Instagram and said, “Hey, come play some Wisecrack with me.” I mean, literally millions of people were trying to get on my Wisecrack platform and play a game with Justin Bieber and it completely crashed. My servers weren't working. The app wasn't working. Nothing was working. So, was a huge missed opportunity. Sometimes a wave comes your way, and it literally just crushes you. That was kind of the other funny story.

[0:34:08] LA: How was chess.com able to meet that demand? I mean, when you started, you didn't build the application to scale. I mean, you built it using technology already had, putting things together, and going from there. But how did you change it to be a scalable architecture that allowed you to take advantage of those spikes?

[0:34:24] JS: You can't ever build a platform or an application on day one to support millions and millions of users. It's just, you'll be paralyzed trying to make it scalable to that degree. I don't care how much they say, “Oh, you got, cloud this and cloud that these days.” It's just never that simple. There's always going to be things where it's like, “Okay, I didn't foresee this or that.” So, there's some times too, where you just build features that you think are really cool on day one. But if you were to fast forward like five years, you'd be like, “That feature is so not scalable. I don't know why we ever built it.” 

A couple examples of that is like we had a map feature in the early days of chess that would show a pin drop for every single user who was online on chess.com. Like an actual map of the globe, and you would see pins everywhere. It was cool when we have like a thousand users. But fast forward is like, this is never going to scale, because we're literally calling a Google API with coordinates and trying to figure out where all of our users are. I mean, it's just like a feature that was cool for like a year. Then, it was like, “Okay, we got to end of life that as soon as possible.”

That happens a lot. There's lots of times where you think, okay, how can I build this? This is going to work great. And then all of a sudden, you're like, “Wait, this does not work at all for a million users.” So, I think there isn't really a great answer as far as like, how do you build something to make it scale? It's always going to be the case that you need to refactor and retool, and optimize the heck out of your code. It's always going to be that way. The faster you grow, the more nerve-racking, the more painful those retools and refactors are going to be.

There was times, we were at MySQL, we were a LAMP stack. So, it was PHP and MySQL, and there's queries that weren't great when you're small, and queries that don't scale at all, once you start to hit that query a lot. And when it's trying to scan millions and millions of rows. So, it was a constant iterative process of like, “Now, this query is our biggest problem, it's taking five seconds. We got to get this query down under a second or down under 100 milliseconds.” We would just have tools that monitored our slowest code, our slowest requests, our slowest API calls, our slowest queries. Occasionally, things would just bubble up and you would be like, “Okay, now this page is taking nine seconds to load.” Or this query is taking seven seconds to execute. So, it's like, that's now the problem of the day. We've got to figure out why is that query slow? Do we have to refactor? Do we have to rearchitect our database? Do we have to denormalize things? Duplicate the data, so it's more readable? Do we have to add more indexes?

It was a constant battle. It was just, literally, a constant battle for years and years and years. I was pulling my hair out. I was screaming expletives in my garage. It was –

[0:37:07] LA: Whack-A-Moled to this extreme –

[0:37:09] JS: Yes. It totally was Whack-A-Mole. It's like, the nice thing about getting bigger and bigger is that you just have more resources and brains and smart people to rely on. You can hire more brilliant data scientists and better MySQL guys, better DevOps guys. For a couple of years, it was just me, and it was very, very stressful and very nerve-wracking. We would hire the best and the brightest consultants in the MySQL space. I swear, it felt like we were always doing something for the first time. I was always like, “Oh, we've never seen that before. We've never seen that before.” There's so many times I felt like, “Are we the only company in the world that's using MySQL? Because it feels like we're always running up against things they've never seen.”

So, picking the tech stack is huge, because you want to be on a tech stack where lots of other big companies have already paved the way for you and solved problems and come out with special builds and things like that. So, picking the right tech stack, I think is a huge deal. Picking the right framework. But there isn't really a panacea for it. It's literally just every day is a new problem, every week is a new problem, every World Chess Championship that happens is a new problem, because it's going to spike something and that something is going to cause problems. So, I’m basically, going through that same thing with Landover right now, my new game, where it's like, every little day is a new problem and a new fire to put out, and we're constantly trying to figure out what's going on.

[0:38:26] LA: Let's move on to what you're doing now. You left chess.com in I think, you said 2017. I'm assuming you left it in good hands, obviously, because it's been growing and expanding since then. But you went off to do something else. First, why was that the right time to leave, and what did you often start doing?

[0:38:43] JS: I think, when we started chess.com, Erik and I, we were floormates in college together. We became very good friends, had a lot of fun adventures our freshman year of college, and we kind of reconnected years later to work on Wholesale Chess together. I left BYU where we were at. So, we didn't have four years of college together. He also went off and served a mission for his church, and so he was gone for two years, and then I left the school. So, we'd kind of like, we weren't super close after our freshman year.

But we were definitely good friends. And I think after you work with somebody for a long time, in a very heated, stressful environment, it just became apparent after 12 years that it was like, CEOs and CTOs, the name of the game is you're going to butt heads a lot. It's a healthy disagreement process. It's a checks and balances process where the CEO says, “Let's build this, this and that.” And the CTO says, “Maybe no, and yes.” So, there's constant pushback between what we're going to build, or what we're not going to build, or how we're going to build it or why we can't build it. If you don't push back on anything, you're going to over engineer and you're going to – it's not going to be scalable. It's not going to be feasible. If the CEO comes and says, “Hey, we want a map of every user on the platform shown on Google Maps.” If I had any foresight, I would have been like, “Look, Erik, when this grows to a million users, this feature is not viable.” So, we could have saved ourselves three months of engineering time never building the map, right?

But sometimes you just say, “Okay, boss. Yes, sir. Let's do it. Let's build it.” You kind of pick and choose your battles. But I think after 12 years, it was apparent, one that it was like, they're going to remain business partners and no longer be friends. Or let's be friends, and let's kind of part ways and do different things. I knew that chess.com was in good hands, like you said. We had a lot of good engineers and smart guys, and it was no longer a startup. I think when I left, we were right around 100 employees. So, there was plenty of guys to pick up where I left off and carry the tech flag forward. 

It felt like a good time. There was lots of other things happening that I wanted to pursue. Honestly, too, a big part of it was chess.com was growing very linearly in 2017. It was like we would always grow every year, but it was always kind of the same pace every year. There wasn't usually like a big spike. It was just kind of like, “Okay, yes.” We'd have like a little jump here for this event or that event. But it was like this is a solid company, it's growing. If COVID or Queen’s Gambit had happened before I left, I would probably still be there. Because once those catalysts happened, and we really started hockey sticking and growing at a very crazy exponential pace, I probably would have been like, “Okay, I don't feel good about leaving, because there are fires to put out everywhere.” So, I'm kind of glad I missed that crazy time. But I'm also a little sad that I wasn't there to actually experience the tidal wave and the tsunami. 

But there was a massive rewrite we did in 2014, where we rewrote the entire application of the framework to go from an old framework that I had chosen back in 2005. That was now kind of defunct and unsupported, called [inaudible 0:41:52]. And the main PHP framework of the day was Symphony. And so, we were like, “We really need to rewrite everything into Symphony.” That was a very painful, dark, three-year process that we went through as a company.

I think once that was kind of done, I was like, “I think that's kind of the end of my tenure and we've got it on to a nice stable platform going forward. It should scale nicely, and we can hire talent a lot easier to work on symphony, because it's a very popular framework and platform.” While all that was happening, I was also getting very, very excited about cryptocurrencies. I bought my first Bitcoin right around $100. So, I was really early into the crypto space. I was going to Bitcoin conferences, and I really want to do something in crypto space and just didn't see any way to do that at chess.com. So, another reason I left was just honestly, because I was really excited about going off and exploring crypto.

[0:42:43] LA: So, is that what you're doing now?

[0:42:45] JS: I tried. I tried for a couple years. I started two or three companies in the crypto space, and realized after a couple years that I was like, “Crypto is just not super ready for B2C applications.” Everything I built, I was like, “This would have been a lot better if it wasn't on the blockchain, or didn't have to do with crypto.” I was trying to force a round peg to a square hole in some ways where I was like, “I just really want to build something and I want to be on the blockchain. I want it to be crypto-related. But then I would just be like, “Why is this even on the blockchain? Why am I doing this with crypto?”

One thing I built was a Bitcoin betting platform where you could play skill-based games for crypto that was called gambit.com. So, you could play backgammon, and a monopoly, and dominoes, and different games and you actually wager Bitcoin while doing it. But again, it was kind of like why do this with Bitcoin? You could probably just do this with fiat currencies and it'd be the same thing and actually be a lot easier.

After a year or so doing that, I realized it was a massive legal headache trying to do in the United States. It's only illegal in some states, and not all states. Maybe this will work if I was living offshore, but the legalities behind it, and the cheating that runs rampant when there's money involved, it was not super fun after about a year. So, eventually, kind of closed that down. I tried running a crypto analysis company called Rency for a little bit. That was actually pretty interesting. But the crypto crash of like 2017. There was some banks that were interested in the technology I was building. But once crypto crashed in a big way, it was like everybody just locked their doors and they're like, “We don't want anything to do with crypto.” So I wasn't quite sure how to monetize that. And then yes, there were some of the thing I dabbled a little bit.

But eventually after kind of investigating the crypto space, I got burned out. There was all these altcoins. There was a lot of garbage out there. There was tons of scams. There was tons of pump and dumps. The whole ICO craze happened. The initial coin offerings. And there was just so many bad actors in that space. That even though like from the beginning, I had been a big believer in cryptocurrency changing the world and being this new, awesome global digital currency, and all the amazing things that it was going to do, it never materialized. It was always slow. It was always cumbersome. It was expensive. It just never became what I thought it would and the amount of copycat coins that were out there really kind of polluted everything. It made it all just feel just kind of slimy.

I eventually just sold all my crypto and got out, and was like, “Let's do something back to where I have a lot of passion.” And that led me to Settlers of Catan. It's always been one of my favorite games. Every time I would go online to play, I found myself on these terrible Java applets that were outdated, that I had to like, update my Java version, or I would end up in a Flash website, and Flash is not supported anymore. So, then I went to like Catan Universe, and it was reeled with bugs and terrible user experience and paywalls everywhere. And I just got really frustrated. I’m like, “This is such a beautiful game. It's such a great game. I love the basic concept of it all, and I hate the fact that there's nowhere for me to play this game on my phone, or on a website that is just awesome.”

So, I finally said, “That’s it. That's my project. That's what I'm going to do. I'm going to try to make the best Catan mobile app there is.” So, that's what I've been doing for a little over two years. It’s called Landover. We can't call it Catan, obviously. We don't use any of their branding, I also took a lot of the mechanics and changed them to make the game much more playable. I was very much inspired by Hearthstone, which was basically Magic the Gathering put online, and made simpler for the masses. I was a huge Hearthstone player. And I loved what they did for this idea of collectible card games. I was like, “I think I can do the same thing for Catan.” So, that's what I did. I'm like, “I'm going to make Catan, better, faster, simpler. Make it on my mobile phone. Make it so I can play with all my friends.” So, now I do have, got seven or eight games going. All my friends from chess.com play, and they're all addicted. We've got a growing community on Discord and just really excited about it. It's fun.

[0:46:40] LA: So, in 20 years, it'll be the size of chess.com.

[0:46:44] JS: That's the hope. That's the hope.

[0:46:45] LA: That's great. Thank you very much, Jay. This is great. So, my guest today has been Jay Severson, who is the founder and CTO of chess.com. Jay, thank you for joining me today on Software Engineering Daily.

[0:46:58] JS: Thanks for having me, Lee. It's been fun.

[END]