WEBVTT - Why One Of The Most Successful Quant Funds Decided To Create Its Own Video Game 0:00:09.240 --> 0:00:12.680 Hello, and welcome to another episode of The Thoughts Podcast. 0:00:12.760 --> 0:00:16.439 I'm Tracy Alloway and I'm Joe Wisenthal. So, Joe, I 0:00:16.440 --> 0:00:19.120 think we have to come clean about this particular episode. 0:00:19.520 --> 0:00:22.880 We do have to come clean before we get into 0:00:22.920 --> 0:00:26.799 the discussion. There's a big, what pound gorilla in the 0:00:26.880 --> 0:00:30.520 room that we have to address. Yeah, So the gorilla 0:00:30.800 --> 0:00:34.800 is that last week we recorded an amazing podcast all 0:00:34.840 --> 0:00:39.120 about technology and its role in finance and the broader world, 0:00:39.840 --> 0:00:44.200 and uh, then we were hit by our own technological snaffoo. 0:00:44.600 --> 0:00:47.760 It's right, So we recorded the greatest episode in the 0:00:47.880 --> 0:00:52.600 history of the entire podcast. It was amazing, one of 0:00:52.640 --> 0:00:55.680 a kind, the kind of conversation that you dream of, 0:00:56.240 --> 0:01:00.520 and then unfortunately the audio was bad and the entire 0:01:00.560 --> 0:01:03.920 thing was ruined. Yes, something happened with the computer. The 0:01:03.960 --> 0:01:06.440 computer said no, we're still trying to figure out what 0:01:06.600 --> 0:01:10.959 the exact issue was. But we learned an important lesson 0:01:11.120 --> 0:01:14.959 about the pitfalls of technology, which gives us an excuse 0:01:15.080 --> 0:01:18.080 to have our guest come on and try to have 0:01:18.200 --> 0:01:22.640 the conversation all over again. So here we go. And 0:01:22.680 --> 0:01:26.720 since this episode is kind of about the relationship between 0:01:26.760 --> 0:01:30.559 technology and finance, we can at least pretend that there's 0:01:30.640 --> 0:01:33.800 some lesson here and what happened to us that's relevant 0:01:33.840 --> 0:01:36.440 to the episode. But really, like as we, I wasn't 0:01:36.440 --> 0:01:38.760 really exaggerated when I said it was a great conversation, 0:01:38.959 --> 0:01:41.440 and it would have been so hard to It would 0:01:41.480 --> 0:01:44.000 have been very hard to try to replicate that or 0:01:44.120 --> 0:01:46.119 to try to pretend we were just doing it again 0:01:46.160 --> 0:01:48.560 for the first time. So just in the spirit of 0:01:48.960 --> 0:01:52.840 honesty and recreating spontaneity, we wanted to get it out 0:01:52.840 --> 0:01:55.320 of the way and be honest with our listeners that 0:01:55.760 --> 0:01:58.559 this is a take two of that conversation. Who knows, 0:01:58.640 --> 0:02:02.040 maybe it will be even better the second time around. 0:02:02.240 --> 0:02:05.240 The important thing is we learned a lesson about backup 0:02:05.320 --> 0:02:08.680 systems and tech. All right, So so here goes, well, 0:02:08.680 --> 0:02:10.800 we can't yes, but we can't now pretend to do 0:02:10.840 --> 0:02:12.400 our stick where we don't know. We're like, what are 0:02:12.440 --> 0:02:14.560 we going to talk about this time? Is that would 0:02:14.600 --> 0:02:17.760 really be contrived? After that? No, No, I wasn't going 0:02:17.800 --> 0:02:19.720 to Okay, I'm going to just bring our guests on. 0:02:19.800 --> 0:02:24.240 Our guest for today for the second time is Alfred Specter. 0:02:24.560 --> 0:02:28.680 He's the chief technology officer of Two Sigma. He's also 0:02:28.800 --> 0:02:31.960 a former engineer at Google, and he was also at 0:02:32.000 --> 0:02:35.640 IBM for a very long time. He's an extremely well 0:02:35.720 --> 0:02:39.360 known name in the realm of technology and also in 0:02:39.520 --> 0:02:43.120 quant driven finance, and he's been nice enough to join 0:02:43.280 --> 0:02:47.120 us yet again on odd lots. So thank you, Alfred, 0:02:47.160 --> 0:02:50.040 really appreciate it. It's my pleasure to be here. And 0:02:50.080 --> 0:02:52.680 by the way, the probability that there's a failure and 0:02:52.720 --> 0:02:56.200 a technology system is somehow proportional to the seniority of 0:02:56.240 --> 0:02:58.720 the person that's involved, So if we ever give a 0:02:58.760 --> 0:03:01.320 demo to like a really seenior person, it's much more 0:03:01.440 --> 0:03:05.080 likely to fail. I'm afraid I engendered the failure. No, 0:03:05.320 --> 0:03:08.440 not at all. But you know, Tracy introduced you as 0:03:09.080 --> 0:03:11.280 you know, the CTO at Too Sigma. You don't seem 0:03:11.320 --> 0:03:13.400 like a guy very who's very busy or anything. So 0:03:13.440 --> 0:03:15.360 I'm sure it was very easy for you to reschedule 0:03:15.440 --> 0:03:17.359 your time just come back in for a second day, 0:03:17.760 --> 0:03:21.959 very easy, indeed. But no, seriously, thank you very much 0:03:22.000 --> 0:03:27.560 for coming back in and recreating last last week. The 0:03:27.560 --> 0:03:30.440 way we started our conversation last week, and really the 0:03:30.480 --> 0:03:34.120 first thing we discussed is that your firm, to Sigma, 0:03:34.200 --> 0:03:38.040 it's a very well known Quantitative Hedge Fund is known 0:03:38.200 --> 0:03:42.720 for having a game. You've created a video game and 0:03:42.920 --> 0:03:47.120 created a competition for people all around the world to 0:03:47.280 --> 0:03:52.760 come and design programs to master the game. So tell 0:03:52.840 --> 0:03:56.000 us what is this game that you have people do 0:03:56.160 --> 0:03:58.520 and why do you have people tried to upbeat it? 0:03:59.440 --> 0:04:02.600 So a couple of years ago we introduced a game, 0:04:02.840 --> 0:04:07.680 a programming competition game where first we within the company 0:04:07.720 --> 0:04:10.520 and then eventually members of the general public got a 0:04:10.640 --> 0:04:14.800 chance to write computer programs that would try to win 0:04:15.320 --> 0:04:18.240 some strategy game. So in fact, it isn't really a 0:04:18.279 --> 0:04:21.160 game of people, but it's a game of programming where 0:04:21.200 --> 0:04:24.359 you program something to try to win. The game was 0:04:24.440 --> 0:04:27.800 really successful internally and excited our engineers and got them 0:04:27.839 --> 0:04:31.279 to think really deeply about algorithms and about how to 0:04:31.320 --> 0:04:35.560 structure situations and game theoretic ways. And we decided to 0:04:35.640 --> 0:04:39.080 launch it thinking that it would attract many programmers that 0:04:39.080 --> 0:04:41.640 would then hear about two Sigma. Some of them might 0:04:41.680 --> 0:04:44.440 actually decide they want to work with us. It would 0:04:44.440 --> 0:04:48.279 also educate people because it requires very sophisticated and clever 0:04:48.360 --> 0:04:51.359 programming to win these games, and we're really interested in 0:04:51.480 --> 0:04:54.760 educating more and more people in tech it was sufficiently 0:04:54.800 --> 0:04:57.320 successful the first year that we did it again, and 0:04:57.400 --> 0:05:00.960 this year there were about six thousand players that wrote 0:05:01.160 --> 0:05:04.359 bots as we call them, to play from about a 0:05:04.440 --> 0:05:08.760 thousand organizations a hundred countries. In the top ten, there 0:05:08.800 --> 0:05:12.279 were six nations represented, and in the top ten winners 0:05:12.400 --> 0:05:15.520 of this two of them were high school students, amazingly enough, 0:05:15.600 --> 0:05:19.279 one of them from Brooklyn and one from Argentina. So um, 0:05:19.320 --> 0:05:22.039 I'm trying to rethink all my questions from last week. No, 0:05:22.480 --> 0:05:26.880 no new questions, okay, fresh questions. We we hear a 0:05:26.920 --> 0:05:32.200 lot about the competition for talent in technology. You obviously 0:05:32.240 --> 0:05:35.880 have all these financial firms that want programmers, um coders, 0:05:35.960 --> 0:05:39.240 people like that, and they're competing with tech firms in 0:05:39.320 --> 0:05:44.800 Silicon Valley. How intense is that competition And what's the 0:05:44.880 --> 0:05:49.440 benefit of trying to attract competition through something like this 0:05:49.600 --> 0:05:55.080 game versus more traditional enticements to the financial industry, like 0:05:55.360 --> 0:05:58.839 just offering people say a lot of money. Well, I think, 0:05:58.880 --> 0:06:02.200 first and foremost, what we're seeing is technology playing a 0:06:02.240 --> 0:06:05.599 bigger and bigger role in almost every industry. I refer 0:06:05.680 --> 0:06:09.839 to that as CS plus X for all X. So 0:06:10.480 --> 0:06:14.720 the innovation is occurring at that intersection of computing and X. 0:06:15.400 --> 0:06:18.520 It's certainly happening now in finance, but I think what 0:06:18.640 --> 0:06:23.800 comes first is technological excellence. So we see ourselves as 0:06:23.880 --> 0:06:28.760 having to play in exactly the same markets for talent 0:06:29.320 --> 0:06:31.960 then tech companies in many domains, and I think that 0:06:32.000 --> 0:06:35.800 will occur even beyond finance and healthcare and education, etcetera 0:06:36.120 --> 0:06:39.680 in the future, this kind of a global technology community. 0:06:39.800 --> 0:06:42.600 We try to appeal to that in having a culture 0:06:42.720 --> 0:06:48.400 internally that values technology, that values algorithms, and values careful thinking, 0:06:49.120 --> 0:06:53.680 values terrific engineering, and we try to portray that externally 0:06:54.560 --> 0:06:57.480 so that people know that's the kind of firm they're joining. 0:06:58.400 --> 0:07:02.240 So tell us about the game specifically, what kind of 0:07:02.240 --> 0:07:05.920 game is it? So the game is a turn based 0:07:06.040 --> 0:07:10.440 strategy game. So there this year somewhere either two or 0:07:10.600 --> 0:07:14.280 four players on the game. When the game starts, the 0:07:14.320 --> 0:07:18.800 players have three ships each and outer space, and the 0:07:18.880 --> 0:07:22.920 goal is to have the ships take over a large 0:07:23.000 --> 0:07:27.160 number of planets and basically take over the galaxy that 0:07:27.240 --> 0:07:30.240 they're part of. Uh. It's really simple in a way 0:07:30.280 --> 0:07:32.800 that the ships can really do only three things. They 0:07:32.800 --> 0:07:36.760 can move a certain number of positions, they can land 0:07:36.800 --> 0:07:39.560 on a planet, and they can take off from the planet, 0:07:39.880 --> 0:07:42.840 and there's some things that happen when they encounter other ships, 0:07:42.840 --> 0:07:45.400 and when they get on the planet, how they gain strength, 0:07:45.440 --> 0:07:48.000 and when more ships are created. But there are only 0:07:48.040 --> 0:07:50.960 three commands to do it. On the other hand, there 0:07:50.960 --> 0:07:54.000 are many, many possible positions in the galaxy, and that's 0:07:54.000 --> 0:07:59.160 what makes the game interesting. There is a huge combinatorial explosion, 0:07:59.320 --> 0:08:02.040 as we say, of moves that you can make at 0:08:02.080 --> 0:08:05.640 any given time. So, but it's extremely challenging to write 0:08:05.640 --> 0:08:10.200 a program to win in this galaxy. Compare the complexity 0:08:10.320 --> 0:08:13.560 of this game to a sort of move based game 0:08:13.640 --> 0:08:18.040 like we would like chess, for example. So in chess, 0:08:18.680 --> 0:08:21.840 the thing we think about, despite all the complexity of 0:08:21.880 --> 0:08:24.400 doing it, is that there's only one piece you move 0:08:24.440 --> 0:08:28.119 at a time, and that piece, depending upon the piece, 0:08:28.120 --> 0:08:30.600 can do different kinds of things. But we call it 0:08:30.640 --> 0:08:33.200 a branching factor of thirty five. At each move in 0:08:33.240 --> 0:08:37.120 the game, you can do about thirty five things, and Haylight, 0:08:37.240 --> 0:08:45.680 the branching factor is ten, followed by zeros, so a 0:08:45.880 --> 0:08:49.640 very very large number of moves. So it's essentially impossible 0:08:49.720 --> 0:08:52.640 for human to play, but a bot can play it 0:08:52.800 --> 0:08:55.679 really well because computers, as we know, are pretty fast. 0:08:56.720 --> 0:09:00.319 So people are playing this game, which bots have been 0:09:00.520 --> 0:09:04.040 most successful and what types of strategies have they been pursuing. 0:09:04.640 --> 0:09:08.040 This is a really interesting question. In the game. You 0:09:08.120 --> 0:09:12.160 might think that the approach should be that people should 0:09:12.200 --> 0:09:15.120 sit down. Players should sit down and think hard about 0:09:15.559 --> 0:09:18.120 should they go to a near planet that's very large, 0:09:18.160 --> 0:09:20.920 should they go to a distant planet that's smaller. Should 0:09:20.960 --> 0:09:23.080 they hide out in a corner and wait for other 0:09:23.160 --> 0:09:27.120 players to interfere with each other and the like. That's 0:09:27.160 --> 0:09:30.319 an algorithmic approach to the game. Or there's the question 0:09:30.360 --> 0:09:33.760 of should we be doing what say the deep mind 0:09:33.880 --> 0:09:37.160 people in that Google subsidiary in London are doing and 0:09:37.200 --> 0:09:40.840 building AI programs that play the game against each other 0:09:41.520 --> 0:09:45.280 and learn the right approaches by essentially trial and error 0:09:45.280 --> 0:09:48.760 and by seeing which wins both approaches are used in 0:09:48.800 --> 0:09:52.760 the game. The top players, the top say thirty or 0:09:52.800 --> 0:09:58.560 forty players used algorithmic approaches where they really thought things through. However, 0:09:59.000 --> 0:10:02.120 now this year some of the top players in the 0:10:02.160 --> 0:10:05.840 top fifty or sixty actually built very simple bots with 0:10:06.000 --> 0:10:10.000 very small amounts of code that actually learned by playing 0:10:10.000 --> 0:10:13.480 the game. Millions and millions of times, and it's quite 0:10:13.480 --> 0:10:16.400 interesting that that actually is working in a world which 0:10:16.480 --> 0:10:19.920 is this difficult. And of course you mentioned the Google 0:10:20.000 --> 0:10:24.400 deep Mind endeavor. It's important in the history of chess computers. 0:10:24.440 --> 0:10:26.440 This is the two different approaches. So back in the 0:10:26.559 --> 0:10:28.960 nineties when we think of Cass pro Verse deep Blue 0:10:29.559 --> 0:10:32.640 Deep Blue at the whole library of games and all 0:10:32.640 --> 0:10:36.400 these grand masters training it, and the new generation just 0:10:36.520 --> 0:10:39.920 learns chess from day one and it teaches itself without 0:10:39.920 --> 0:10:43.400 any GMS or anything. And these days that new approach 0:10:43.400 --> 0:10:45.839 is what works. But well, you're saying in this game, 0:10:46.400 --> 0:10:50.160 you've seen some success from both approaches. That's right. In 0:10:50.640 --> 0:10:53.959 the recent Alpha Go program that that deep Mind did, 0:10:54.000 --> 0:10:56.200 they learned to be a world champion in chess in 0:10:56.240 --> 0:11:00.320 four hours of play without much background. Really remarkable. This 0:11:00.360 --> 0:11:03.920 game is considerably harder. So if we think about artificial intelligence, 0:11:04.280 --> 0:11:07.199 some artificial intelligence is just to try to duplicate what 0:11:07.240 --> 0:11:09.719 people do. So like an early problem in AI was 0:11:09.800 --> 0:11:12.880 digit recognition. Could you read, say, the numbers on a 0:11:13.000 --> 0:11:16.760 check automatically? That was AI just a few years back. 0:11:16.840 --> 0:11:20.120 That was a very hard problem. Now then another problem 0:11:20.120 --> 0:11:22.000 in AI is to do something that humans do, but 0:11:22.080 --> 0:11:24.719 do it better. So that's like self driving cars. You 0:11:24.760 --> 0:11:28.120 can easily imagine that it should be possible. Maybe it's 0:11:28.120 --> 0:11:30.719 hard to build a self driving car because we can 0:11:30.720 --> 0:11:33.320 do it pretty well. Then there are these questions of 0:11:33.360 --> 0:11:36.440 things which we can't even do, and that's a game 0:11:36.559 --> 0:11:38.760 like hay Light. Can we get a I s to 0:11:38.880 --> 0:11:41.560 do that? And there are implications of course in financial 0:11:41.600 --> 0:11:46.199 markets were all kind of challenged by predictions and optimization 0:11:46.200 --> 0:11:49.240 and financial markets. Maybe it's very much the case that 0:11:49.320 --> 0:11:52.040 these AI systems, in the fullness of time, will do 0:11:52.120 --> 0:11:55.520 things we ourselves can't even think of doing today, and 0:11:55.960 --> 0:11:59.360 in making a better economic system. So I'm always curious 0:11:59.360 --> 0:12:03.040 when it comes to these bots that are essentially self 0:12:03.240 --> 0:12:07.600 learning the game, how good are they dealing with spontaneity 0:12:07.720 --> 0:12:12.760 or the unpredictability of other people's decisions or say, you know, 0:12:12.880 --> 0:12:16.160 just a human playing the game who might make a mistake. 0:12:16.679 --> 0:12:19.319 Do they always assume that the other players are a 0:12:19.520 --> 0:12:24.080 rational or can they react in some some way to 0:12:24.600 --> 0:12:27.920 the unexpected. I guess I think it's a really good question. 0:12:28.160 --> 0:12:29.880 I don't know the answer, and I think it's a 0:12:29.920 --> 0:12:33.360 subject of research now. To understand that. Two things come 0:12:33.400 --> 0:12:35.920 to mind. One is I saw some of the early 0:12:36.840 --> 0:12:41.640 newscasts on the early go playing programs, and people thought 0:12:41.679 --> 0:12:44.160 they were really creative and doing things that hadn't been 0:12:44.160 --> 0:12:47.320 seen before. I'm not a go aficionado, but I believe 0:12:47.360 --> 0:12:50.120 that to be true. The second is, it's certainly the 0:12:50.160 --> 0:12:53.000 case that many think that great creativity is kind of 0:12:53.000 --> 0:12:56.640 serendipity or almost a kind of randomness that happens. And 0:12:56.679 --> 0:12:58.600 of course, if we think that, and we think that 0:12:58.720 --> 0:13:02.240 great creativity comes of kind of the random ideas that 0:13:02.320 --> 0:13:04.959 maybe one of our strange colleagues might have some days 0:13:05.360 --> 0:13:10.640 that can be programmed. So obviously, humans can't play this game, heylite, 0:13:10.679 --> 0:13:14.600 It's way too complicated. Can humans appreciate the game like 0:13:14.679 --> 0:13:18.200 in the same way like if you watching two bods 0:13:18.240 --> 0:13:21.480 play against each other? Is it understandable enough so that 0:13:21.679 --> 0:13:24.319 someone could look at the game and sort of grasp 0:13:24.480 --> 0:13:28.240 what they're doing? Absolutely. A couple of things about that. 0:13:28.320 --> 0:13:30.760 Number one is that if humans are going to want 0:13:30.800 --> 0:13:33.760 to program bots for the game, they have to find 0:13:33.800 --> 0:13:36.640 it entertaining. So it has to be an interesting objective 0:13:36.679 --> 0:13:39.080 that they're trying to achieve, and they have to be 0:13:39.120 --> 0:13:41.640 able to watch and understand what their body is doing, 0:13:41.960 --> 0:13:45.120 and it's quite exciting, so you need that for the game. 0:13:45.559 --> 0:13:48.559 And then secondly, we in fact saw that in reality, 0:13:48.880 --> 0:13:51.160 many people have put up plays of the game on 0:13:51.200 --> 0:13:54.840 YouTube and other places where you can watch really interesting 0:13:54.880 --> 0:13:57.520 games and how they unfold, and you get to see 0:13:57.559 --> 0:14:01.160 the strategy. For example, what may happen is a player 0:14:01.600 --> 0:14:03.599 a players bought have to be careful. How I say this, 0:14:03.679 --> 0:14:06.760 A players bought may realize that it has very little 0:14:06.840 --> 0:14:09.840 chance of winning, but perhaps if it goes hides in 0:14:09.840 --> 0:14:12.920 a corner, the other players may defeed each other and 0:14:12.960 --> 0:14:15.960 it might come in second. And that's a strategy that 0:14:16.040 --> 0:14:20.600 happened in the first round of the game. That bot Yeah, well, 0:14:20.880 --> 0:14:23.680 the in fact you do. We do tend to personify 0:14:23.720 --> 0:14:26.600 these things over time, which is another interesting aspect of 0:14:26.640 --> 0:14:30.200 how humans deal with computers. But we we didn't see 0:14:30.240 --> 0:14:33.840 this behavior until the last week or so of Halight 0:14:33.920 --> 0:14:36.320 version one, and then all of a sudden, we call 0:14:36.360 --> 0:14:39.480 it an emergent behavior. It emerged from the game. We 0:14:39.560 --> 0:14:43.320 never anticipated that that happened. And there are other other 0:14:43.480 --> 0:14:46.960 kinds of strategies that also occur as well in the game. 0:14:47.520 --> 0:14:50.680 So you've been running, uh, two rounds of this game 0:14:50.720 --> 0:14:54.960 now Halight right, Um, you're in your second iteration. Have 0:14:55.160 --> 0:14:58.160 you ever recruited anyone that was playing the game? Has 0:14:58.160 --> 0:15:03.640 it actually translated into tangible recruitment benefits for you? Yes, 0:15:04.160 --> 0:15:07.120 happy to say that we have a tremendous employee that 0:15:07.160 --> 0:15:10.000 came out of Halight, one who's working with us on 0:15:10.040 --> 0:15:14.440 our London office, and we have many more people that 0:15:14.520 --> 0:15:17.720 remind us that they know about two Sigma because of Halight. 0:15:18.320 --> 0:15:22.040 So it's valuable from a marketing perspective as well. So 0:15:22.080 --> 0:15:24.760 I think it's something that will be around and helping 0:15:24.840 --> 0:15:27.920 us for a long term. In fact, I met a 0:15:27.960 --> 0:15:31.080 college intern who did Halight as a high school student 0:15:31.120 --> 0:15:33.640 and said that she knew about two Sigma because she 0:15:33.680 --> 0:15:35.880 did it as a high school student. I want to 0:15:36.240 --> 0:15:40.840 turn to more just the you know, talk about UH 0:15:40.960 --> 0:15:43.760 quantitative finance and some of the lessons you've learned before 0:15:43.800 --> 0:15:45.760 we do though, and before we move off the game. 0:15:46.040 --> 0:15:48.720 When we in our first attempt at recording this episode, 0:15:48.720 --> 0:15:50.920 at the end, you said, oh, you wanted to talk 0:15:50.920 --> 0:15:52.960 a little bit more about some of the high school 0:15:53.000 --> 0:15:57.000 students who had done so well in the game, and 0:15:57.040 --> 0:15:58.920 so I don't want to forget to do that this time. 0:15:58.960 --> 0:16:00.880 Tell us a little bit more out how high school 0:16:00.880 --> 0:16:03.320 students are able to who they are, how can they 0:16:03.360 --> 0:16:07.320 compete with the top computer scientists in programming at So 0:16:07.760 --> 0:16:10.200 let's just start with one thing first. So you have 0:16:10.280 --> 0:16:13.320 to design a game so that it's easy to get 0:16:13.360 --> 0:16:16.040 started with. Right. That's a nice thing about Checkers, right 0:16:16.080 --> 0:16:18.080 for little kids. You can learn the rules quickly, and 0:16:18.160 --> 0:16:21.040 yet it's pretty sophisticated to play it. Well, the same 0:16:21.080 --> 0:16:23.840 thing happens here. You want to build a game that's 0:16:23.880 --> 0:16:26.800 easy to get started with, but that has a really 0:16:26.880 --> 0:16:31.440 really long path, maybe in essentially an infinite path towards perfection, 0:16:31.520 --> 0:16:33.760 so maybe there can be no absolute perfection. You can 0:16:33.760 --> 0:16:36.160 play a very very long time, then it's a much 0:16:36.200 --> 0:16:39.960 better game. So we even wrote a paper about how 0:16:39.960 --> 0:16:44.000 to design these games, called the Design and Implementation of 0:16:44.080 --> 0:16:47.760 Modern Online Programming Competitions. So, again going back to the 0:16:47.840 --> 0:16:51.360 ease of starting, we realized that since they're easy to 0:16:51.360 --> 0:16:55.400 start playing, they're accessible to high school students. So we 0:16:55.440 --> 0:16:57.840 went out and did a bunch of hackathons around the 0:16:57.880 --> 0:17:00.720 New York City area and some other places and had 0:17:00.920 --> 0:17:03.040 quite a bit of acceptance. We had almost a thousand 0:17:03.120 --> 0:17:06.680 high school students doing this worldwide, and we learned about 0:17:06.680 --> 0:17:09.879 it because a teacher in Texas initially wrote to us 0:17:09.880 --> 0:17:12.960 and said that it was a great opportunity for members 0:17:13.080 --> 0:17:16.280 of his class to start programming. And we think that 0:17:16.320 --> 0:17:18.919 early outreach is very important. It's also a core value 0:17:18.960 --> 0:17:21.400 of the firm because the co chairs of the firm 0:17:21.440 --> 0:17:25.359 are very involved in mathematics education for young kids and 0:17:25.400 --> 0:17:29.280 also for programming educations via the M I T. Scratch 0:17:29.320 --> 0:17:33.840 initiative for middle schoolers and uh and high school students. 0:17:34.359 --> 0:17:36.720 So just one last thing on that one. I gotta 0:17:36.760 --> 0:17:39.960 just mentioned. So this, this this kid in Brooklyn actually 0:17:39.960 --> 0:17:42.679 had an article written about him in the Brooklyn newspaper. 0:17:43.080 --> 0:17:45.440 So that was very exciting. I was called Brooklyn High 0:17:45.480 --> 0:17:48.439 Schooler takes on the World. We'll have to check that 0:17:48.480 --> 0:17:52.320 one up well, link to it when we post this. Yeah. 0:17:52.520 --> 0:17:56.919 Uh So, widening the conversation out to finance and tech, 0:17:57.520 --> 0:18:00.000 we were referring to Two Sigma earlier as a very 0:18:00.000 --> 0:18:04.600 all known quant fund. I'm wondering what makes a quant 0:18:04.680 --> 0:18:08.480 fund a quant fund, given that nowadays it feels like 0:18:08.520 --> 0:18:11.960 pretty much every fund has some sort of systematic or 0:18:12.040 --> 0:18:16.600 programmatic trading actually happening. Right, So two SIGMAS a tech 0:18:16.680 --> 0:18:19.359 firm that looks at many places where we can apply 0:18:19.480 --> 0:18:24.120 technology to optimize outcomes and finance. So we're also in insurance, 0:18:24.160 --> 0:18:27.520 and we're in venture capital, etcetera. But certainly one of 0:18:27.560 --> 0:18:30.440 the things we do is investment management. As you mentioned, 0:18:30.840 --> 0:18:34.320 I think what differentiates us is number one, the deep 0:18:34.440 --> 0:18:38.240 and long term technical talent that we've had. After all, 0:18:38.280 --> 0:18:40.879 we were started by an m I T pH D 0:18:41.000 --> 0:18:44.439 and AI about fifteen or more years ago. David Siegel 0:18:44.760 --> 0:18:48.680 and John Overdeck, the other co chair, is a real 0:18:48.840 --> 0:18:53.439 expert mathematician, silver math OLYMPIAD and a statistician. So the 0:18:53.480 --> 0:18:56.560 two of them really brought this to the firm quite 0:18:56.560 --> 0:18:59.959 a while back and it's everywhere in the firm. Second 0:19:00.080 --> 0:19:02.320 is we do have scale in this. We've been doing 0:19:02.320 --> 0:19:05.000 it a long time, and I think that scale is 0:19:05.240 --> 0:19:09.680 really something that differentiates us from many of our competitors, 0:19:10.320 --> 0:19:13.640 right because, as we know, we've all heard every bank 0:19:13.760 --> 0:19:16.120 CEO these days or at times they say, oh, we're 0:19:16.119 --> 0:19:19.399 really a software company that does banking, or really a 0:19:19.440 --> 0:19:23.360 tech company. But you have a long experience with companies 0:19:23.359 --> 0:19:28.200 that are undisputably tech companies. Google and IBM, one of 0:19:28.280 --> 0:19:31.560 the biggest differences in terms of culture that you see 0:19:31.600 --> 0:19:36.040 at a place like two Sigma versus your experience at Google, 0:19:37.960 --> 0:19:41.800 I think probably if you could name one, it's that 0:19:42.080 --> 0:19:46.040 technology is viewed at least as the equal, if not 0:19:46.160 --> 0:19:51.119 the driver, of the core business. So at our firm, 0:19:51.160 --> 0:19:54.800 there's no question that those of us that do computer science, 0:19:54.880 --> 0:20:00.199 mathematics and statistics are viewed by almost everyone as the 0:20:00.280 --> 0:20:03.040 basis of the firm's success. Now, of course we need 0:20:03.119 --> 0:20:05.000 and we're very happy to have the folks that do 0:20:05.119 --> 0:20:07.840 compliance and legal and all the other activities that are 0:20:07.880 --> 0:20:10.960 needed in the firm, but it's really a technology and 0:20:11.119 --> 0:20:14.520 math and statistics first operation. I think the same thing 0:20:14.640 --> 0:20:17.280 is true at the really successful tech companies as well, 0:20:17.800 --> 0:20:20.679 and became frankly less true at the tech companies that 0:20:20.720 --> 0:20:23.840 didn't do so well. It is kind of interesting that 0:20:23.920 --> 0:20:28.760 if you think at places where algorithms and programmatic strategies 0:20:28.920 --> 0:20:32.560 might be really interesting to do, uh, the finance companies 0:20:32.560 --> 0:20:38.200 should theoretically be really really intriguing, because banks and insurers 0:20:38.200 --> 0:20:41.679 have these realms and reams of data that should be 0:20:41.760 --> 0:20:45.720 interesting for anyone with the technology background. But it almost 0:20:45.760 --> 0:20:48.280 feels like it's taken a little bit of time for 0:20:48.320 --> 0:20:50.440 people to catch on to that, and it's only now 0:20:50.520 --> 0:20:52.800 that a lot of the financial firms are making this 0:20:52.920 --> 0:20:56.879 really big push. Why do you think it's taken a 0:20:56.920 --> 0:21:00.719 bit of time. So one is, of course, finance use 0:21:00.840 --> 0:21:04.200 technology very early on, right, It was among the earliest 0:21:04.280 --> 0:21:07.840 users just to computerize account records and transfers and such. 0:21:08.359 --> 0:21:12.040 So perhaps it's the case that because finance used a 0:21:12.040 --> 0:21:16.919 lot of technology, there became kind of a installed base 0:21:17.080 --> 0:21:21.760 of old technology that actually acted as an impediment to modernization. 0:21:22.400 --> 0:21:25.199 So I think that is one fact. So those of 0:21:25.280 --> 0:21:28.160 us that are newer in the business have an advantage. 0:21:28.200 --> 0:21:31.200 An example, of course, if you look at say online advertising, 0:21:31.560 --> 0:21:33.960 it didn't exist more than a couple of decades ago. 0:21:34.040 --> 0:21:36.640 That's when it all began, so there can't be an 0:21:36.640 --> 0:21:40.040 installed base from the nineteen sixties. So I think we 0:21:40.080 --> 0:21:44.119 didn't have, if you will, negative inertia in new fields, 0:21:44.119 --> 0:21:46.240 and we we did have some of that in finance. 0:21:46.680 --> 0:21:49.760 The second is, I think it's important to understand what 0:21:49.840 --> 0:21:53.720 we should be doing in finance, and that's making financial 0:21:53.800 --> 0:21:58.600 systems economic systems work better. So all of us like capitalism, 0:21:58.600 --> 0:22:02.120 we like decentralization, and we like optimization of the firm, 0:22:02.160 --> 0:22:03.800 and we all hope that it will lead to pray, 0:22:03.880 --> 0:22:07.920 to optimality and an efficient operation of society that produces 0:22:07.960 --> 0:22:10.680 lots of goods and services for all. But we all 0:22:10.720 --> 0:22:14.560 know that if we're not careful, inventories build up, or 0:22:14.600 --> 0:22:17.880 prices get out of whack, or people have irrational exuberance 0:22:17.920 --> 0:22:21.600 and the like. I believe with the proper application of data, 0:22:21.720 --> 0:22:25.119 the proper application of mathematics and statistics, we can do 0:22:25.160 --> 0:22:28.960 a better job of running these economic systems. It's not easy, 0:22:29.000 --> 0:22:31.199 but I think that's really exciting. And I have a 0:22:31.200 --> 0:22:34.720 lot of success in attracting technical people to the firm 0:22:34.800 --> 0:22:37.920 because that's what I think we're doing. Do you proactively 0:22:38.680 --> 0:22:41.520 think about exactly what you said about building up some 0:22:41.600 --> 0:22:44.640 sort of legacy code base or some sort of legacy 0:22:44.680 --> 0:22:47.280 set of systems that ten years from now you'll still 0:22:47.480 --> 0:22:49.359 be hewing too, even if it's not the state of 0:22:49.359 --> 0:22:52.840 the art. I worry about it all the time. All 0:22:52.880 --> 0:22:56.640 of us in technology worry or should be worrying about 0:22:57.000 --> 0:23:00.640 the legacy that we will create. And it's a very 0:23:00.640 --> 0:23:03.639 difficult problem. If you think in the United States, they're literally, 0:23:03.640 --> 0:23:07.119 you know, millions and millions of programmers writing computer code 0:23:07.119 --> 0:23:10.680 all the time. All of that code will someday get old, 0:23:11.119 --> 0:23:14.040 and I'm afraid it will look like the substructure underneath 0:23:14.080 --> 0:23:17.200 Lexington Avenue out here sometime and make it very difficult 0:23:17.240 --> 0:23:19.760 to build the next subway. But in banking, you still 0:23:19.840 --> 0:23:23.400 hear stories about some of the banks having, um, how 0:23:23.400 --> 0:23:27.040