1 00:00:09,240 --> 00:00:12,680 Speaker 1: Hello, and welcome to another episode of The Thoughts Podcast. 2 00:00:12,760 --> 00:00:16,439 Speaker 1: I'm Tracy Alloway and I'm Joe Wisenthal. So, Joe, I 3 00:00:16,440 --> 00:00:19,120 Speaker 1: think we have to come clean about this particular episode. 4 00:00:19,520 --> 00:00:22,880 Speaker 1: We do have to come clean before we get into 5 00:00:22,920 --> 00:00:26,799 Speaker 1: the discussion. There's a big, what pound gorilla in the 6 00:00:26,880 --> 00:00:30,520 Speaker 1: room that we have to address. Yeah, So the gorilla 7 00:00:30,800 --> 00:00:34,800 Speaker 1: is that last week we recorded an amazing podcast all 8 00:00:34,840 --> 00:00:39,120 Speaker 1: about technology and its role in finance and the broader world, 9 00:00:39,840 --> 00:00:44,200 Speaker 1: and uh, then we were hit by our own technological snaffoo. 10 00:00:44,600 --> 00:00:47,760 Speaker 1: It's right, So we recorded the greatest episode in the 11 00:00:47,880 --> 00:00:52,600 Speaker 1: history of the entire podcast. It was amazing, one of 12 00:00:52,640 --> 00:00:55,680 Speaker 1: a kind, the kind of conversation that you dream of, 13 00:00:56,240 --> 00:01:00,520 Speaker 1: and then unfortunately the audio was bad and the entire 14 00:01:00,560 --> 00:01:03,920 Speaker 1: thing was ruined. Yes, something happened with the computer. The 15 00:01:03,960 --> 00:01:06,440 Speaker 1: computer said no, we're still trying to figure out what 16 00:01:06,600 --> 00:01:10,959 Speaker 1: the exact issue was. But we learned an important lesson 17 00:01:11,120 --> 00:01:14,959 Speaker 1: about the pitfalls of technology, which gives us an excuse 18 00:01:15,080 --> 00:01:18,080 Speaker 1: to have our guest come on and try to have 19 00:01:18,200 --> 00:01:22,640 Speaker 1: the conversation all over again. So here we go. And 20 00:01:22,680 --> 00:01:26,720 Speaker 1: since this episode is kind of about the relationship between 21 00:01:26,760 --> 00:01:30,559 Speaker 1: technology and finance, we can at least pretend that there's 22 00:01:30,640 --> 00:01:33,800 Speaker 1: some lesson here and what happened to us that's relevant 23 00:01:33,840 --> 00:01:36,440 Speaker 1: to the episode. But really, like as we, I wasn't 24 00:01:36,440 --> 00:01:38,760 Speaker 1: really exaggerated when I said it was a great conversation, 25 00:01:38,959 --> 00:01:41,440 Speaker 1: and it would have been so hard to It would 26 00:01:41,480 --> 00:01:44,000 Speaker 1: have been very hard to try to replicate that or 27 00:01:44,120 --> 00:01:46,119 Speaker 1: to try to pretend we were just doing it again 28 00:01:46,160 --> 00:01:48,560 Speaker 1: for the first time. So just in the spirit of 29 00:01:48,960 --> 00:01:52,840 Speaker 1: honesty and recreating spontaneity, we wanted to get it out 30 00:01:52,840 --> 00:01:55,320 Speaker 1: of the way and be honest with our listeners that 31 00:01:55,760 --> 00:01:58,559 Speaker 1: this is a take two of that conversation. Who knows, 32 00:01:58,640 --> 00:02:02,040 Speaker 1: maybe it will be even better the second time around. 33 00:02:02,240 --> 00:02:05,240 Speaker 1: The important thing is we learned a lesson about backup 34 00:02:05,320 --> 00:02:08,680 Speaker 1: systems and tech. All right, So so here goes, well, 35 00:02:08,680 --> 00:02:10,800 Speaker 1: we can't yes, but we can't now pretend to do 36 00:02:10,840 --> 00:02:12,400 Speaker 1: our stick where we don't know. We're like, what are 37 00:02:12,440 --> 00:02:14,560 Speaker 1: we going to talk about this time? Is that would 38 00:02:14,600 --> 00:02:17,760 Speaker 1: really be contrived? After that? No, No, I wasn't going 39 00:02:17,800 --> 00:02:19,720 Speaker 1: to Okay, I'm going to just bring our guests on. 40 00:02:19,800 --> 00:02:24,240 Speaker 1: Our guest for today for the second time is Alfred Specter. 41 00:02:24,560 --> 00:02:28,680 Speaker 1: He's the chief technology officer of Two Sigma. He's also 42 00:02:28,800 --> 00:02:31,960 Speaker 1: a former engineer at Google, and he was also at 43 00:02:32,000 --> 00:02:35,640 Speaker 1: IBM for a very long time. He's an extremely well 44 00:02:35,720 --> 00:02:39,360 Speaker 1: known name in the realm of technology and also in 45 00:02:39,520 --> 00:02:43,120 Speaker 1: quant driven finance, and he's been nice enough to join 46 00:02:43,280 --> 00:02:47,120 Speaker 1: us yet again on odd lots. So thank you, Alfred, 47 00:02:47,160 --> 00:02:50,040 Speaker 1: really appreciate it. It's my pleasure to be here. And 48 00:02:50,080 --> 00:02:52,680 Speaker 1: by the way, the probability that there's a failure and 49 00:02:52,720 --> 00:02:56,200 Speaker 1: a technology system is somehow proportional to the seniority of 50 00:02:56,240 --> 00:02:58,720 Speaker 1: the person that's involved, So if we ever give a 51 00:02:58,760 --> 00:03:01,320 Speaker 1: demo to like a really seenior person, it's much more 52 00:03:01,440 --> 00:03:05,080 Speaker 1: likely to fail. I'm afraid I engendered the failure. No, 53 00:03:05,320 --> 00:03:08,440 Speaker 1: not at all. But you know, Tracy introduced you as 54 00:03:09,080 --> 00:03:11,280 Speaker 1: you know, the CTO at Too Sigma. You don't seem 55 00:03:11,320 --> 00:03:13,400 Speaker 1: like a guy very who's very busy or anything. So 56 00:03:13,440 --> 00:03:15,360 Speaker 1: I'm sure it was very easy for you to reschedule 57 00:03:15,440 --> 00:03:17,359 Speaker 1: your time just come back in for a second day, 58 00:03:17,760 --> 00:03:21,959 Speaker 1: very easy, indeed. But no, seriously, thank you very much 59 00:03:22,000 --> 00:03:27,560 Speaker 1: for coming back in and recreating last last week. The 60 00:03:27,560 --> 00:03:30,440 Speaker 1: way we started our conversation last week, and really the 61 00:03:30,480 --> 00:03:34,120 Speaker 1: first thing we discussed is that your firm, to Sigma, 62 00:03:34,200 --> 00:03:38,040 Speaker 1: it's a very well known Quantitative Hedge Fund is known 63 00:03:38,200 --> 00:03:42,720 Speaker 1: for having a game. You've created a video game and 64 00:03:42,920 --> 00:03:47,120 Speaker 1: created a competition for people all around the world to 65 00:03:47,280 --> 00:03:52,760 Speaker 1: come and design programs to master the game. So tell 66 00:03:52,840 --> 00:03:56,000 Speaker 1: us what is this game that you have people do 67 00:03:56,160 --> 00:03:58,520 Speaker 1: and why do you have people tried to upbeat it? 68 00:03:59,440 --> 00:04:02,600 Speaker 1: So a couple of years ago we introduced a game, 69 00:04:02,840 --> 00:04:07,680 Speaker 1: a programming competition game where first we within the company 70 00:04:07,720 --> 00:04:10,520 Speaker 1: and then eventually members of the general public got a 71 00:04:10,640 --> 00:04:14,800 Speaker 1: chance to write computer programs that would try to win 72 00:04:15,320 --> 00:04:18,240 Speaker 1: some strategy game. So in fact, it isn't really a 73 00:04:18,279 --> 00:04:21,160 Speaker 1: game of people, but it's a game of programming where 74 00:04:21,200 --> 00:04:24,359 Speaker 1: you program something to try to win. The game was 75 00:04:24,440 --> 00:04:27,800 Speaker 1: really successful internally and excited our engineers and got them 76 00:04:27,839 --> 00:04:31,279 Speaker 1: to think really deeply about algorithms and about how to 77 00:04:31,320 --> 00:04:35,560 Speaker 1: structure situations and game theoretic ways. And we decided to 78 00:04:35,640 --> 00:04:39,080 Speaker 1: launch it thinking that it would attract many programmers that 79 00:04:39,080 --> 00:04:41,640 Speaker 1: would then hear about two Sigma. Some of them might 80 00:04:41,680 --> 00:04:44,440 Speaker 1: actually decide they want to work with us. It would 81 00:04:44,440 --> 00:04:48,279 Speaker 1: also educate people because it requires very sophisticated and clever 82 00:04:48,360 --> 00:04:51,359 Speaker 1: programming to win these games, and we're really interested in 83 00:04:51,480 --> 00:04:54,760 Speaker 1: educating more and more people in tech it was sufficiently 84 00:04:54,800 --> 00:04:57,320 Speaker 1: successful the first year that we did it again, and 85 00:04:57,400 --> 00:05:00,960 Speaker 1: this year there were about six thousand players that wrote 86 00:05:01,160 --> 00:05:04,359 Speaker 1: bots as we call them, to play from about a 87 00:05:04,440 --> 00:05:08,760 Speaker 1: thousand organizations a hundred countries. In the top ten, there 88 00:05:08,800 --> 00:05:12,279 Speaker 1: were six nations represented, and in the top ten winners 89 00:05:12,400 --> 00:05:15,520 Speaker 1: of this two of them were high school students, amazingly enough, 90 00:05:15,600 --> 00:05:19,279 Speaker 1: one of them from Brooklyn and one from Argentina. So um, 91 00:05:19,320 --> 00:05:22,039 Speaker 1: I'm trying to rethink all my questions from last week. No, 92 00:05:22,480 --> 00:05:26,880 Speaker 1: no new questions, okay, fresh questions. We we hear a 93 00:05:26,920 --> 00:05:32,200 Speaker 1: lot about the competition for talent in technology. You obviously 94 00:05:32,240 --> 00:05:35,880 Speaker 1: have all these financial firms that want programmers, um coders, 95 00:05:35,960 --> 00:05:39,240 Speaker 1: people like that, and they're competing with tech firms in 96 00:05:39,320 --> 00:05:44,800 Speaker 1: Silicon Valley. How intense is that competition And what's the 97 00:05:44,880 --> 00:05:49,440 Speaker 1: benefit of trying to attract competition through something like this 98 00:05:49,600 --> 00:05:55,080 Speaker 1: game versus more traditional enticements to the financial industry, like 99 00:05:55,360 --> 00:05:58,839 Speaker 1: just offering people say a lot of money. Well, I think, 100 00:05:58,880 --> 00:06:02,200 Speaker 1: first and foremost, what we're seeing is technology playing a 101 00:06:02,240 --> 00:06:05,599 Speaker 1: bigger and bigger role in almost every industry. I refer 102 00:06:05,680 --> 00:06:09,839 Speaker 1: to that as CS plus X for all X. So 103 00:06:10,480 --> 00:06:14,720 Speaker 1: the innovation is occurring at that intersection of computing and X. 104 00:06:15,400 --> 00:06:18,520 Speaker 1: It's certainly happening now in finance, but I think what 105 00:06:18,640 --> 00:06:23,800 Speaker 1: comes first is technological excellence. So we see ourselves as 106 00:06:23,880 --> 00:06:28,760 Speaker 1: having to play in exactly the same markets for talent 107 00:06:29,320 --> 00:06:31,960 Speaker 1: then tech companies in many domains, and I think that 108 00:06:32,000 --> 00:06:35,800 Speaker 1: will occur even beyond finance and healthcare and education, etcetera 109 00:06:36,120 --> 00:06:39,680 Speaker 1: in the future, this kind of a global technology community. 110 00:06:39,800 --> 00:06:42,600 Speaker 1: We try to appeal to that in having a culture 111 00:06:42,720 --> 00:06:48,400 Speaker 1: internally that values technology, that values algorithms, and values careful thinking, 112 00:06:49,120 --> 00:06:53,680 Speaker 1: values terrific engineering, and we try to portray that externally 113 00:06:54,560 --> 00:06:57,480 Speaker 1: so that people know that's the kind of firm they're joining. 114 00:06:58,400 --> 00:07:02,240 Speaker 1: So tell us about the game specifically, what kind of 115 00:07:02,240 --> 00:07:05,920 Speaker 1: game is it? So the game is a turn based 116 00:07:06,040 --> 00:07:10,440 Speaker 1: strategy game. So there this year somewhere either two or 117 00:07:10,600 --> 00:07:14,280 Speaker 1: four players on the game. When the game starts, the 118 00:07:14,320 --> 00:07:18,800 Speaker 1: players have three ships each and outer space, and the 119 00:07:18,880 --> 00:07:22,920 Speaker 1: goal is to have the ships take over a large 120 00:07:23,000 --> 00:07:27,160 Speaker 1: number of planets and basically take over the galaxy that 121 00:07:27,240 --> 00:07:30,240 Speaker 1: they're part of. Uh. It's really simple in a way 122 00:07:30,280 --> 00:07:32,800 Speaker 1: that the ships can really do only three things. They 123 00:07:32,800 --> 00:07:36,760 Speaker 1: can move a certain number of positions, they can land 124 00:07:36,800 --> 00:07:39,560 Speaker 1: on a planet, and they can take off from the planet, 125 00:07:39,880 --> 00:07:42,840 Speaker 1: and there's some things that happen when they encounter other ships, 126 00:07:42,840 --> 00:07:45,400 Speaker 1: and when they get on the planet, how they gain strength, 127 00:07:45,440 --> 00:07:48,000 Speaker 1: and when more ships are created. But there are only 128 00:07:48,040 --> 00:07:50,960 Speaker 1: three commands to do it. On the other hand, there 129 00:07:50,960 --> 00:07:54,000 Speaker 1: are many, many possible positions in the galaxy, and that's 130 00:07:54,000 --> 00:07:59,160 Speaker 1: what makes the game interesting. There is a huge combinatorial explosion, 131 00:07:59,320 --> 00:08:02,040 Speaker 1: as we say, of moves that you can make at 132 00:08:02,080 --> 00:08:05,640 Speaker 1: any given time. So, but it's extremely challenging to write 133 00:08:05,640 --> 00:08:10,200 Speaker 1: a program to win in this galaxy. Compare the complexity 134 00:08:10,320 --> 00:08:13,560 Speaker 1: of this game to a sort of move based game 135 00:08:13,640 --> 00:08:18,040 Speaker 1: like we would like chess, for example. So in chess, 136 00:08:18,680 --> 00:08:21,840 Speaker 1: the thing we think about, despite all the complexity of 137 00:08:21,880 --> 00:08:24,400 Speaker 1: doing it, is that there's only one piece you move 138 00:08:24,440 --> 00:08:28,119 Speaker 1: at a time, and that piece, depending upon the piece, 139 00:08:28,120 --> 00:08:30,600 Speaker 1: can do different kinds of things. But we call it 140 00:08:30,640 --> 00:08:33,200 Speaker 1: a branching factor of thirty five. At each move in 141 00:08:33,240 --> 00:08:37,120 Speaker 1: the game, you can do about thirty five things, and Haylight, 142 00:08:37,240 --> 00:08:45,680 Speaker 1: the branching factor is ten, followed by zeros, so a 143 00:08:45,880 --> 00:08:49,640 Speaker 1: very very large number of moves. So it's essentially impossible 144 00:08:49,720 --> 00:08:52,640 Speaker 1: for human to play, but a bot can play it 145 00:08:52,800 --> 00:08:55,679 Speaker 1: really well because computers, as we know, are pretty fast. 146 00:08:56,720 --> 00:09:00,319 Speaker 1: So people are playing this game, which bots have been 147 00:09:00,520 --> 00:09:04,040 Speaker 1: most successful and what types of strategies have they been pursuing. 148 00:09:04,640 --> 00:09:08,040 Speaker 1: This is a really interesting question. In the game. You 149 00:09:08,120 --> 00:09:12,160 Speaker 1: might think that the approach should be that people should 150 00:09:12,200 --> 00:09:15,120 Speaker 1: sit down. Players should sit down and think hard about 151 00:09:15,559 --> 00:09:18,120 Speaker 1: should they go to a near planet that's very large, 152 00:09:18,160 --> 00:09:20,920 Speaker 1: should they go to a distant planet that's smaller. Should 153 00:09:20,960 --> 00:09:23,080 Speaker 1: they hide out in a corner and wait for other 154 00:09:23,160 --> 00:09:27,120 Speaker 1: players to interfere with each other and the like. That's 155 00:09:27,160 --> 00:09:30,319 Speaker 1: an algorithmic approach to the game. Or there's the question 156 00:09:30,360 --> 00:09:33,760 Speaker 1: of should we be doing what say the deep mind 157 00:09:33,880 --> 00:09:37,160 Speaker 1: people in that Google subsidiary in London are doing and 158 00:09:37,200 --> 00:09:40,840 Speaker 1: building AI programs that play the game against each other 159 00:09:41,520 --> 00:09:45,280 Speaker 1: and learn the right approaches by essentially trial and error 160 00:09:45,280 --> 00:09:48,760 Speaker 1: and by seeing which wins both approaches are used in 161 00:09:48,800 --> 00:09:52,760 Speaker 1: the game. The top players, the top say thirty or 162 00:09:52,800 --> 00:09:58,560 Speaker 1: forty players used algorithmic approaches where they really thought things through. However, 163 00:09:59,000 --> 00:10:02,120 Speaker 1: now this year some of the top players in the 164 00:10:02,160 --> 00:10:05,840 Speaker 1: top fifty or sixty actually built very simple bots with 165 00:10:06,000 --> 00:10:10,000 Speaker 1: very small amounts of code that actually learned by playing 166 00:10:10,000 --> 00:10:13,480 Speaker 1: the game. Millions and millions of times, and it's quite 167 00:10:13,480 --> 00:10:16,400 Speaker 1: interesting that that actually is working in a world which 168 00:10:16,480 --> 00:10:19,920 Speaker 1: is this difficult. And of course you mentioned the Google 169 00:10:20,000 --> 00:10:24,400 Speaker 1: deep Mind endeavor. It's important in the history of chess computers. 170 00:10:24,440 --> 00:10:26,440 Speaker 1: This is the two different approaches. So back in the 171 00:10:26,559 --> 00:10:28,960 Speaker 1: nineties when we think of Cass pro Verse deep Blue 172 00:10:29,559 --> 00:10:32,640 Speaker 1: Deep Blue at the whole library of games and all 173 00:10:32,640 --> 00:10:36,400 Speaker 1: these grand masters training it, and the new generation just 174 00:10:36,520 --> 00:10:39,920 Speaker 1: learns chess from day one and it teaches itself without 175 00:10:39,920 --> 00:10:43,400 Speaker 1: any GMS or anything. And these days that new approach 176 00:10:43,400 --> 00:10:45,839 Speaker 1: is what works. But well, you're saying in this game, 177 00:10:46,400 --> 00:10:50,160 Speaker 1: you've seen some success from both approaches. That's right. In 178 00:10:50,640 --> 00:10:53,959 Speaker 1: the recent Alpha Go program that that deep Mind did, 179 00:10:54,000 --> 00:10:56,200 Speaker 1: they learned to be a world champion in chess in 180 00:10:56,240 --> 00:11:00,320 Speaker 1: four hours of play without much background. Really remarkable. This 181 00:11:00,360 --> 00:11:03,920 Speaker 1: game is considerably harder. So if we think about artificial intelligence, 182 00:11:04,280 --> 00:11:07,199 Speaker 1: some artificial intelligence is just to try to duplicate what 183 00:11:07,240 --> 00:11:09,719 Speaker 1: people do. So like an early problem in AI was 184 00:11:09,800 --> 00:11:12,880 Speaker 1: digit recognition. Could you read, say, the numbers on a 185 00:11:13,000 --> 00:11:16,760 Speaker 1: check automatically? That was AI just a few years back. 186 00:11:16,840 --> 00:11:20,120 Speaker 1: That was a very hard problem. Now then another problem 187 00:11:20,120 --> 00:11:22,000 Speaker 1: in AI is to do something that humans do, but 188 00:11:22,080 --> 00:11:24,719 Speaker 1: do it better. So that's like self driving cars. You 189 00:11:24,760 --> 00:11:28,120 Speaker 1: can easily imagine that it should be possible. Maybe it's 190 00:11:28,120 --> 00:11:30,719 Speaker 1: hard to build a self driving car because we can 191 00:11:30,720 --> 00:11:33,320 Speaker 1: do it pretty well. Then there are these questions of 192 00:11:33,360 --> 00:11:36,440 Speaker 1: things which we can't even do, and that's a game 193 00:11:36,559 --> 00:11:38,760 Speaker 1: like hay Light. Can we get a I s to 194 00:11:38,880 --> 00:11:41,560 Speaker 1: do that? And there are implications of course in financial 195 00:11:41,600 --> 00:11:46,199 Speaker 1: markets were all kind of challenged by predictions and optimization 196 00:11:46,200 --> 00:11:49,240 Speaker 1: and financial markets. Maybe it's very much the case that 197 00:11:49,320 --> 00:11:52,040 Speaker 1: these AI systems, in the fullness of time, will do 198 00:11:52,120 --> 00:11:55,520 Speaker 1: things we ourselves can't even think of doing today, and 199 00:11:55,960 --> 00:11:59,360 Speaker 1: in making a better economic system. So I'm always curious 200 00:11:59,360 --> 00:12:03,040 Speaker 1: when it comes to these bots that are essentially self 201 00:12:03,240 --> 00:12:07,600 Speaker 1: learning the game, how good are they dealing with spontaneity 202 00:12:07,720 --> 00:12:12,760 Speaker 1: or the unpredictability of other people's decisions or say, you know, 203 00:12:12,880 --> 00:12:16,160 Speaker 1: just a human playing the game who might make a mistake. 204 00:12:16,679 --> 00:12:19,319 Speaker 1: Do they always assume that the other players are a 205 00:12:19,520 --> 00:12:24,080 Speaker 1: rational or can they react in some some way to 206 00:12:24,600 --> 00:12:27,920 Speaker 1: the unexpected. I guess I think it's a really good question. 207 00:12:28,160 --> 00:12:29,880 Speaker 1: I don't know the answer, and I think it's a 208 00:12:29,920 --> 00:12:33,360 Speaker 1: subject of research now. To understand that. Two things come 209 00:12:33,400 --> 00:12:35,920 Speaker 1: to mind. One is I saw some of the early 210 00:12:36,840 --> 00:12:41,640 Speaker 1: newscasts on the early go playing programs, and people thought 211 00:12:41,679 --> 00:12:44,160 Speaker 1: they were really creative and doing things that hadn't been 212 00:12:44,160 --> 00:12:47,320 Speaker 1: seen before. I'm not a go aficionado, but I believe 213 00:12:47,360 --> 00:12:50,120 Speaker 1: that to be true. The second is, it's certainly the 214 00:12:50,160 --> 00:12:53,000 Speaker 1: case that many think that great creativity is kind of 215 00:12:53,000 --> 00:12:56,640 Speaker 1: serendipity or almost a kind of randomness that happens. And 216 00:12:56,679 --> 00:12:58,600 Speaker 1: of course, if we think that, and we think that 217 00:12:58,720 --> 00:13:02,240 Speaker 1: great creativity comes of kind of the random ideas that 218 00:13:02,320 --> 00:13:04,959 Speaker 1: maybe one of our strange colleagues might have some days 219 00:13:05,360 --> 00:13:10,640 Speaker 1: that can be programmed. So obviously, humans can't play this game, heylite, 220 00:13:10,679 --> 00:13:14,600 Speaker 1: It's way too complicated. Can humans appreciate the game like 221 00:13:14,679 --> 00:13:18,200 Speaker 1: in the same way like if you watching two bods 222 00:13:18,240 --> 00:13:21,480 Speaker 1: play against each other? Is it understandable enough so that 223 00:13:21,679 --> 00:13:24,319 Speaker 1: someone could look at the game and sort of grasp 224 00:13:24,480 --> 00:13:28,240 Speaker 1: what they're doing? Absolutely. A couple of things about that. 225 00:13:28,320 --> 00:13:30,760 Speaker 1: Number one is that if humans are going to want 226 00:13:30,800 --> 00:13:33,760 Speaker 1: to program bots for the game, they have to find 227 00:13:33,800 --> 00:13:36,640 Speaker 1: it entertaining. So it has to be an interesting objective 228 00:13:36,679 --> 00:13:39,080 Speaker 1: that they're trying to achieve, and they have to be 229 00:13:39,120 --> 00:13:41,640 Speaker 1: able to watch and understand what their body is doing, 230 00:13:41,960 --> 00:13:45,120 Speaker 1: and it's quite exciting, so you need that for the game. 231 00:13:45,559 --> 00:13:48,559 Speaker 1: And then secondly, we in fact saw that in reality, 232 00:13:48,880 --> 00:13:51,160 Speaker 1: many people have put up plays of the game on 233 00:13:51,200 --> 00:13:54,840 Speaker 1: YouTube and other places where you can watch really interesting 234 00:13:54,880 --> 00:13:57,520 Speaker 1: games and how they unfold, and you get to see 235 00:13:57,559 --> 00:14:01,160 Speaker 1: the strategy. For example, what may happen is a player 236 00:14:01,600 --> 00:14:03,599 Speaker 1: a players bought have to be careful. How I say this, 237 00:14:03,679 --> 00:14:06,760 Speaker 1: A players bought may realize that it has very little 238 00:14:06,840 --> 00:14:09,840 Speaker 1: chance of winning, but perhaps if it goes hides in 239 00:14:09,840 --> 00:14:12,920 Speaker 1: a corner, the other players may defeed each other and 240 00:14:12,960 --> 00:14:15,960 Speaker 1: it might come in second. And that's a strategy that 241 00:14:16,040 --> 00:14:20,600 Speaker 1: happened in the first round of the game. That bot Yeah, well, 242 00:14:20,880 --> 00:14:23,680 Speaker 1: the in fact you do. We do tend to personify 243 00:14:23,720 --> 00:14:26,600 Speaker 1: these things over time, which is another interesting aspect of 244 00:14:26,640 --> 00:14:30,200 Speaker 1: how humans deal with computers. But we we didn't see 245 00:14:30,240 --> 00:14:33,840 Speaker 1: this behavior until the last week or so of Halight 246 00:14:33,920 --> 00:14:36,320 Speaker 1: version one, and then all of a sudden, we call 247 00:14:36,360 --> 00:14:39,480 Speaker 1: it an emergent behavior. It emerged from the game. We 248 00:14:39,560 --> 00:14:43,320 Speaker 1: never anticipated that that happened. And there are other other 249 00:14:43,480 --> 00:14:46,960 Speaker 1: kinds of strategies that also occur as well in the game. 250 00:14:47,520 --> 00:14:50,680 Speaker 1: So you've been running, uh, two rounds of this game 251 00:14:50,720 --> 00:14:54,960 Speaker 1: now Halight right, Um, you're in your second iteration. Have 252 00:14:55,160 --> 00:14:58,160 Speaker 1: you ever recruited anyone that was playing the game? Has 253 00:14:58,160 --> 00:15:03,640 Speaker 1: it actually translated into tangible recruitment benefits for you? Yes, 254 00:15:04,160 --> 00:15:07,120 Speaker 1: happy to say that we have a tremendous employee that 255 00:15:07,160 --> 00:15:10,000 Speaker 1: came out of Halight, one who's working with us on 256 00:15:10,040 --> 00:15:14,440 Speaker 1: our London office, and we have many more people that 257 00:15:14,520 --> 00:15:17,720 Speaker 1: remind us that they know about two Sigma because of Halight. 258 00:15:18,320 --> 00:15:22,040 Speaker 1: So it's valuable from a marketing perspective as well. So 259 00:15:22,080 --> 00:15:24,760 Speaker 1: I think it's something that will be around and helping 260 00:15:24,840 --> 00:15:27,920 Speaker 1: us for a long term. In fact, I met a 261 00:15:27,960 --> 00:15:31,080 Speaker 1: college intern who did Halight as a high school student 262 00:15:31,120 --> 00:15:33,640 Speaker 1: and said that she knew about two Sigma because she 263 00:15:33,680 --> 00:15:35,880 Speaker 1: did it as a high school student. I want to 264 00:15:36,240 --> 00:15:40,840 Speaker 1: turn to more just the you know, talk about UH 265 00:15:40,960 --> 00:15:43,760 Speaker 1: quantitative finance and some of the lessons you've learned before 266 00:15:43,800 --> 00:15:45,760 Speaker 1: we do though, and before we move off the game. 267 00:15:46,040 --> 00:15:48,720 Speaker 1: When we in our first attempt at recording this episode, 268 00:15:48,720 --> 00:15:50,920 Speaker 1: at the end, you said, oh, you wanted to talk 269 00:15:50,920 --> 00:15:52,960 Speaker 1: a little bit more about some of the high school 270 00:15:53,000 --> 00:15:57,000 Speaker 1: students who had done so well in the game, and 271 00:15:57,040 --> 00:15:58,920 Speaker 1: so I don't want to forget to do that this time. 272 00:15:58,960 --> 00:16:00,880 Speaker 1: Tell us a little bit more out how high school 273 00:16:00,880 --> 00:16:03,320 Speaker 1: students are able to who they are, how can they 274 00:16:03,360 --> 00:16:07,320 Speaker 1: compete with the top computer scientists in programming at So 275 00:16:07,760 --> 00:16:10,200 Speaker 1: let's just start with one thing first. So you have 276 00:16:10,280 --> 00:16:13,320 Speaker 1: to design a game so that it's easy to get 277 00:16:13,360 --> 00:16:16,040 Speaker 1: started with. Right. That's a nice thing about Checkers, right 278 00:16:16,080 --> 00:16:18,080 Speaker 1: for little kids. You can learn the rules quickly, and 279 00:16:18,160 --> 00:16:21,040 Speaker 1: yet it's pretty sophisticated to play it. Well, the same 280 00:16:21,080 --> 00:16:23,840 Speaker 1: thing happens here. You want to build a game that's 281 00:16:23,880 --> 00:16:26,800 Speaker 1: easy to get started with, but that has a really 282 00:16:26,880 --> 00:16:31,440 Speaker 1: really long path, maybe in essentially an infinite path towards perfection, 283 00:16:31,520 --> 00:16:33,760 Speaker 1: so maybe there can be no absolute perfection. You can 284 00:16:33,760 --> 00:16:36,160 Speaker 1: play a very very long time, then it's a much 285 00:16:36,200 --> 00:16:39,960 Speaker 1: better game. So we even wrote a paper about how 286 00:16:39,960 --> 00:16:44,000 Speaker 1: to design these games, called the Design and Implementation of 287 00:16:44,080 --> 00:16:47,760 Speaker 1: Modern Online Programming Competitions. So, again going back to the 288 00:16:47,840 --> 00:16:51,360 Speaker 1: ease of starting, we realized that since they're easy to 289 00:16:51,360 --> 00:16:55,400 Speaker 1: start playing, they're accessible to high school students. So we 290 00:16:55,440 --> 00:16:57,840 Speaker 1: went out and did a bunch of hackathons around the 291 00:16:57,880 --> 00:17:00,720 Speaker 1: New York City area and some other places and had 292 00:17:00,920 --> 00:17:03,040 Speaker 1: quite a bit of acceptance. We had almost a thousand 293 00:17:03,120 --> 00:17:06,680 Speaker 1: high school students doing this worldwide, and we learned about 294 00:17:06,680 --> 00:17:09,879 Speaker 1: it because a teacher in Texas initially wrote to us 295 00:17:09,880 --> 00:17:12,960 Speaker 1: and said that it was a great opportunity for members 296 00:17:13,080 --> 00:17:16,280 Speaker 1: of his class to start programming. And we think that 297 00:17:16,320 --> 00:17:18,919 Speaker 1: early outreach is very important. It's also a core value 298 00:17:18,960 --> 00:17:21,400 Speaker 1: of the firm because the co chairs of the firm 299 00:17:21,440 --> 00:17:25,359 Speaker 1: are very involved in mathematics education for young kids and 300 00:17:25,400 --> 00:17:29,280 Speaker 1: also for programming educations via the M I T. Scratch 301 00:17:29,320 --> 00:17:33,840 Speaker 1: initiative for middle schoolers and uh and high school students. 302 00:17:34,359 --> 00:17:36,720 Speaker 1: So just one last thing on that one. I gotta 303 00:17:36,760 --> 00:17:39,960 Speaker 1: just mentioned. So this, this this kid in Brooklyn actually 304 00:17:39,960 --> 00:17:42,679 Speaker 1: had an article written about him in the Brooklyn newspaper. 305 00:17:43,080 --> 00:17:45,440 Speaker 1: So that was very exciting. I was called Brooklyn High 306 00:17:45,480 --> 00:17:48,439 Speaker 1: Schooler takes on the World. We'll have to check that 307 00:17:48,480 --> 00:17:52,320 Speaker 1: one up well, link to it when we post this. Yeah. 308 00:17:52,520 --> 00:17:56,919 Speaker 1: Uh So, widening the conversation out to finance and tech, 309 00:17:57,520 --> 00:18:00,000 Speaker 1: we were referring to Two Sigma earlier as a very 310 00:18:00,000 --> 00:18:04,600 Speaker 1: all known quant fund. I'm wondering what makes a quant 311 00:18:04,680 --> 00:18:08,480 Speaker 1: fund a quant fund, given that nowadays it feels like 312 00:18:08,520 --> 00:18:11,960 Speaker 1: pretty much every fund has some sort of systematic or 313 00:18:12,040 --> 00:18:16,600 Speaker 1: programmatic trading actually happening. Right, So two SIGMAS a tech 314 00:18:16,680 --> 00:18:19,359 Speaker 1: firm that looks at many places where we can apply 315 00:18:19,480 --> 00:18:24,120 Speaker 1: technology to optimize outcomes and finance. So we're also in insurance, 316 00:18:24,160 --> 00:18:27,520 Speaker 1: and we're in venture capital, etcetera. But certainly one of 317 00:18:27,560 --> 00:18:30,440 Speaker 1: the things we do is investment management. As you mentioned, 318 00:18:30,840 --> 00:18:34,320 Speaker 1: I think what differentiates us is number one, the deep 319 00:18:34,440 --> 00:18:38,240 Speaker 1: and long term technical talent that we've had. After all, 320 00:18:38,280 --> 00:18:40,879 Speaker 1: we were started by an m I T pH D 321 00:18:41,000 --> 00:18:44,439 Speaker 1: and AI about fifteen or more years ago. David Siegel 322 00:18:44,760 --> 00:18:48,680 Speaker 1: and John Overdeck, the other co chair, is a real 323 00:18:48,840 --> 00:18:53,439 Speaker 1: expert mathematician, silver math OLYMPIAD and a statistician. So the 324 00:18:53,480 --> 00:18:56,560 Speaker 1: two of them really brought this to the firm quite 325 00:18:56,560 --> 00:18:59,959 Speaker 1: a while back and it's everywhere in the firm. Second 326 00:19:00,080 --> 00:19:02,320 Speaker 1: is we do have scale in this. We've been doing 327 00:19:02,320 --> 00:19:05,000 Speaker 1: it a long time, and I think that scale is 328 00:19:05,240 --> 00:19:09,680 Speaker 1: really something that differentiates us from many of our competitors, 329 00:19:10,320 --> 00:19:13,640 Speaker 1: right because, as we know, we've all heard every bank 330 00:19:13,760 --> 00:19:16,120 Speaker 1: CEO these days or at times they say, oh, we're 331 00:19:16,119 --> 00:19:19,399 Speaker 1: really a software company that does banking, or really a 332 00:19:19,440 --> 00:19:23,360 Speaker 1: tech company. But you have a long experience with companies 333 00:19:23,359 --> 00:19:28,200 Speaker 1: that are undisputably tech companies. Google and IBM, one of 334 00:19:28,280 --> 00:19:31,560 Speaker 1: the biggest differences in terms of culture that you see 335 00:19:31,600 --> 00:19:36,040 Speaker 1: at a place like two Sigma versus your experience at Google, 336 00:19:37,960 --> 00:19:41,800 Speaker 1: I think probably if you could name one, it's that 337 00:19:42,080 --> 00:19:46,040 Speaker 1: technology is viewed at least as the equal, if not 338 00:19:46,160 --> 00:19:51,119 Speaker 1: the driver, of the core business. So at our firm, 339 00:19:51,160 --> 00:19:54,800 Speaker 1: there's no question that those of us that do computer science, 340 00:19:54,880 --> 00:20:00,199 Speaker 1: mathematics and statistics are viewed by almost everyone as the 341 00:20:00,280 --> 00:20:03,040 Speaker 1: basis of the firm's success. Now, of course we need 342 00:20:03,119 --> 00:20:05,000 Speaker 1: and we're very happy to have the folks that do 343 00:20:05,119 --> 00:20:07,840 Speaker 1: compliance and legal and all the other activities that are 344 00:20:07,880 --> 00:20:10,960 Speaker 1: needed in the firm, but it's really a technology and 345 00:20:11,119 --> 00:20:14,520 Speaker 1: math and statistics first operation. I think the same thing 346 00:20:14,640 --> 00:20:17,280 Speaker 1: is true at the really successful tech companies as well, 347 00:20:17,800 --> 00:20:20,679 Speaker 1: and became frankly less true at the tech companies that 348 00:20:20,720 --> 00:20:23,840 Speaker 1: didn't do so well. It is kind of interesting that 349 00:20:23,920 --> 00:20:28,760 Speaker 1: if you think at places where algorithms and programmatic strategies 350 00:20:28,920 --> 00:20:32,560 Speaker 1: might be really interesting to do, uh, the finance companies 351 00:20:32,560 --> 00:20:38,200 Speaker 1: should theoretically be really really intriguing, because banks and insurers 352 00:20:38,200 --> 00:20:41,679 Speaker 1: have these realms and reams of data that should be 353 00:20:41,760 --> 00:20:45,720 Speaker 1: interesting for anyone with the technology background. But it almost 354 00:20:45,760 --> 00:20:48,280 Speaker 1: feels like it's taken a little bit of time for 355 00:20:48,320 --> 00:20:50,440 Speaker 1: people to catch on to that, and it's only now 356 00:20:50,520 --> 00:20:52,800 Speaker 1: that a lot of the financial firms are making this 357 00:20:52,920 --> 00:20:56,879 Speaker 1: really big push. Why do you think it's taken a 358 00:20:56,920 --> 00:21:00,719 Speaker 1: bit of time. So one is, of course, finance use 359 00:21:00,840 --> 00:21:04,200 Speaker 1: technology very early on, right, It was among the earliest 360 00:21:04,280 --> 00:21:07,840 Speaker 1: users just to computerize account records and transfers and such. 361 00:21:08,359 --> 00:21:12,040 Speaker 1: So perhaps it's the case that because finance used a 362 00:21:12,040 --> 00:21:16,919 Speaker 1: lot of technology, there became kind of a installed base 363 00:21:17,080 --> 00:21:21,760 Speaker 1: of old technology that actually acted as an impediment to modernization. 364 00:21:22,400 --> 00:21:25,199 Speaker 1: So I think that is one fact. So those of 365 00:21:25,280 --> 00:21:28,160 Speaker 1: us that are newer in the business have an advantage. 366 00:21:28,200 --> 00:21:31,200 Speaker 1: An example, of course, if you look at say online advertising, 367 00:21:31,560 --> 00:21:33,960 Speaker 1: it didn't exist more than a couple of decades ago. 368 00:21:34,040 --> 00:21:36,640 Speaker 1: That's when it all began, so there can't be an 369 00:21:36,640 --> 00:21:40,040 Speaker 1: installed base from the nineteen sixties. So I think we 370 00:21:40,080 --> 00:21:44,119 Speaker 1: didn't have, if you will, negative inertia in new fields, 371 00:21:44,119 --> 00:21:46,240 Speaker 1: and we we did have some of that in finance. 372 00:21:46,680 --> 00:21:49,760 Speaker 1: The second is, I think it's important to understand what 373 00:21:49,840 --> 00:21:53,720 Speaker 1: we should be doing in finance, and that's making financial 374 00:21:53,800 --> 00:21:58,600 Speaker 1: systems economic systems work better. So all of us like capitalism, 375 00:21:58,600 --> 00:22:02,120 Speaker 1: we like decentralization, and we like optimization of the firm, 376 00:22:02,160 --> 00:22:03,800 Speaker 1: and we all hope that it will lead to pray, 377 00:22:03,880 --> 00:22:07,920 Speaker 1: to optimality and an efficient operation of society that produces 378 00:22:07,960 --> 00:22:10,680 Speaker 1: lots of goods and services for all. But we all 379 00:22:10,720 --> 00:22:14,560 Speaker 1: know that if we're not careful, inventories build up, or 380 00:22:14,600 --> 00:22:17,880 Speaker 1: prices get out of whack, or people have irrational exuberance 381 00:22:17,920 --> 00:22:21,600 Speaker 1: and the like. I believe with the proper application of data, 382 00:22:21,720 --> 00:22:25,119 Speaker 1: the proper application of mathematics and statistics, we can do 383 00:22:25,160 --> 00:22:28,960 Speaker 1: a better job of running these economic systems. It's not easy, 384 00:22:29,000 --> 00:22:31,199 Speaker 1: but I think that's really exciting. And I have a 385 00:22:31,200 --> 00:22:34,720 Speaker 1: lot of success in attracting technical people to the firm 386 00:22:34,800 --> 00:22:37,920 Speaker 1: because that's what I think we're doing. Do you proactively 387 00:22:38,680 --> 00:22:41,520 Speaker 1: think about exactly what you said about building up some 388 00:22:41,600 --> 00:22:44,640 Speaker 1: sort of legacy code base or some sort of legacy 389 00:22:44,680 --> 00:22:47,280 Speaker 1: set of systems that ten years from now you'll still 390 00:22:47,480 --> 00:22:49,359 Speaker 1: be hewing too, even if it's not the state of 391 00:22:49,359 --> 00:22:52,840 Speaker 1: the art. I worry about it all the time. All 392 00:22:52,880 --> 00:22:56,640 Speaker 1: of us in technology worry or should be worrying about 393 00:22:57,000 --> 00:23:00,640 Speaker 1: the legacy that we will create. And it's a very 394 00:23:00,640 --> 00:23:03,639 Speaker 1: difficult problem. If you think in the United States, they're literally, 395 00:23:03,640 --> 00:23:07,119 Speaker 1: you know, millions and millions of programmers writing computer code 396 00:23:07,119 --> 00:23:10,680 Speaker 1: all the time. All of that code will someday get old, 397 00:23:11,119 --> 00:23:14,040 Speaker 1: and I'm afraid it will look like the substructure underneath 398 00:23:14,080 --> 00:23:17,200 Speaker 1: Lexington Avenue out here sometime and make it very difficult 399 00:23:17,240 --> 00:23:19,760 Speaker 1: to build the next subway. But in banking, you still 400 00:23:19,840 --> 00:23:23,400 Speaker 1: hear stories about some of the banks having, um, how 401 00:23:23,400 --> 00:23:27,040 Speaker 1: do you say coble or cobble This programming language from 402 00:23:27,080 --> 00:23:29,119 Speaker 1: that stemmed from I think it was World War two, 403 00:23:29,400 --> 00:23:32,840 Speaker 1: basically invented in the nineteen forties and nineteen fifties. And 404 00:23:32,880 --> 00:23:35,680 Speaker 1: if you're one of the programmers who can still actually 405 00:23:35,800 --> 00:23:40,520 Speaker 1: code in this ancient, ancient software language, apparently you can 406 00:23:40,520 --> 00:23:43,080 Speaker 1: earn big money. So it does seem to be something 407 00:23:43,119 --> 00:23:47,920 Speaker 1: of an issue. So common business oriented language COBAL. Yeah, 408 00:23:47,960 --> 00:23:51,720 Speaker 1: I think it comes probably from late fifties and sixties. Uh, 409 00:23:51,880 --> 00:23:54,400 Speaker 1: not World War two, but you're on the right track there. 410 00:23:55,080 --> 00:23:57,959 Speaker 1: And yeah, there's a lot of cobaal code around and 411 00:23:58,160 --> 00:24:02,520 Speaker 1: some of it was written by employees who retired, maintained 412 00:24:02,560 --> 00:24:05,919 Speaker 1: by the employees they trained who have now retired, and 413 00:24:05,960 --> 00:24:08,639 Speaker 1: the next generation is maintaining that. And you can just 414 00:24:08,680 --> 00:24:11,120 Speaker 1: think of the engineering challenge do you rewrite it all? 415 00:24:11,800 --> 00:24:13,760 Speaker 1: But do you even know what it does. It's a 416 00:24:13,840 --> 00:24:16,880 Speaker 1: real challenge for organizations to deal with that. I don't 417 00:24:16,920 --> 00:24:19,080 Speaker 1: believe we have any coball. In fact, I'm certain we 418 00:24:19,119 --> 00:24:24,280 Speaker 1: have no coball co ball free. One of the things 419 00:24:24,320 --> 00:24:27,280 Speaker 1: you hear a lot uh Silicon Valley people talk about 420 00:24:27,359 --> 00:24:30,480 Speaker 1: is the importance of culture as the enduring mode or 421 00:24:30,520 --> 00:24:34,359 Speaker 1: the enduring sustainable advantage, and that with whatever else that 422 00:24:34,400 --> 00:24:36,720 Speaker 1: goes on, as long as they have a superior culture, 423 00:24:37,440 --> 00:24:40,720 Speaker 1: that that allows them to beat the competition. How do 424 00:24:40,760 --> 00:24:44,840 Speaker 1: you guarantee that that's in place at two sigma? And 425 00:24:44,920 --> 00:24:47,440 Speaker 1: when you think about all of these new funds or 426 00:24:47,520 --> 00:24:49,960 Speaker 1: legacy funds that sort of want the new quant unit, 427 00:24:50,119 --> 00:24:52,440 Speaker 1: or banks trying to get into quant stuff, how much 428 00:24:52,440 --> 00:24:55,800 Speaker 1: do you see that as an advantage towards competitors who 429 00:24:55,840 --> 00:24:59,600 Speaker 1: would otherwise want to modify what you're doing. I think 430 00:24:59,600 --> 00:25:03,320 Speaker 1: in all of our organizations, talent is the first and 431 00:25:03,359 --> 00:25:08,480 Speaker 1: most important thing. So the talent today is possessing of 432 00:25:08,520 --> 00:25:13,400 Speaker 1: many opportunities because there's so many applications of advanced computer 433 00:25:13,440 --> 00:25:16,840 Speaker 1: science and machine learning and AI and the like. So 434 00:25:17,080 --> 00:25:20,680 Speaker 1: we really feel that that that culture is really important, 435 00:25:20,680 --> 00:25:23,440 Speaker 1: and the culture is it's hard to pin down exactly 436 00:25:23,440 --> 00:25:26,959 Speaker 1: what it is. Certainly, it's clear objectives for the business. 437 00:25:27,000 --> 00:25:30,160 Speaker 1: Certainly it's clear understanding of what we do for our 438 00:25:30,200 --> 00:25:32,439 Speaker 1: clients and we have to understand what to do and 439 00:25:32,440 --> 00:25:35,600 Speaker 1: feel good about doing that really well. But it's also 440 00:25:36,000 --> 00:25:38,880 Speaker 1: soft and other things. Just if you think about the 441 00:25:39,040 --> 00:25:42,719 Speaker 1: boards where people were talking about Haylight, um, you know 442 00:25:42,760 --> 00:25:44,720 Speaker 1: you read them if you're an employee and you feel 443 00:25:44,720 --> 00:25:46,960 Speaker 1: good about working at the firm. One of them said, 444 00:25:47,200 --> 00:25:51,119 Speaker 1: it's an absolute blast discussing strategy, sharing replays and getting 445 00:25:51,119 --> 00:25:53,840 Speaker 1: excited about the games with friends. That's a great place 446 00:25:53,880 --> 00:25:57,080 Speaker 1: to work when you're doing that for the world. Last question, 447 00:25:57,160 --> 00:26:00,960 Speaker 1: unless Joe has more, what's your taught up tip when 448 00:26:01,000 --> 00:26:05,120 Speaker 1: it comes to avoiding technological errors such as the one 449 00:26:05,240 --> 00:26:08,800 Speaker 1: we experienced last week. Well, my original career as a 450 00:26:08,840 --> 00:26:13,919 Speaker 1: professor at Carnegie Mellon was in reliable distributed systems, and 451 00:26:14,000 --> 00:26:17,320 Speaker 1: that means that you have to have duplication at many 452 00:26:17,440 --> 00:26:20,320 Speaker 1: levels of the system. So how do you make sure 453 00:26:20,359 --> 00:26:24,000 Speaker 1: you have two of everything in the chain. That's important 454 00:26:24,000 --> 00:26:27,280 Speaker 1: in financial markets, so that we have capacity to keep operating. 455 00:26:27,680 --> 00:26:31,119 Speaker 1: That's probably important in games. We had many servers that 456 00:26:31,160 --> 00:26:33,800 Speaker 1: could run Haylight, so if one of them, god forbid, 457 00:26:33,840 --> 00:26:36,040 Speaker 1: had a problem, another one would keep running. In fact, 458 00:26:36,080 --> 00:26:39,359 Speaker 1: we ran maybe tens or hundreds of servers simultaneously to 459 00:26:39,440 --> 00:26:41,920 Speaker 1: deal with the load. It's probably important in radio and 460 00:26:42,000 --> 00:26:45,880 Speaker 1: podcast two. On that note, a perfect tip for all 461 00:26:45,920 --> 00:26:48,560 Speaker 1: of us to remember in all endeavors of our lives. 462 00:26:49,000 --> 00:26:51,640 Speaker 1: Alfred Spector, thank you very much for joining us. It's 463 00:26:51,680 --> 00:27:05,400 Speaker 1: my pleasure. I enjoyed doing it again, so I really 464 00:27:05,400 --> 00:27:07,439 Speaker 1: hope we don't have to bring Alfred in for a 465 00:27:07,480 --> 00:27:10,640 Speaker 1: third time. But it was really I like, I disagree. 466 00:27:10,680 --> 00:27:13,800 Speaker 1: I really I was gonna say, wait, wait, wait, I 467 00:27:13,840 --> 00:27:16,440 Speaker 1: was gonna say, it was really enjoyable speaking to him 468 00:27:16,520 --> 00:27:19,199 Speaker 1: for another thirty minutes. Agree, And if we have to 469 00:27:19,240 --> 00:27:20,879 Speaker 1: do it a third time, I'm looking forward to that 470 00:27:20,960 --> 00:27:23,960 Speaker 1: as well. But in all seriousness, I think we did 471 00:27:24,000 --> 00:27:27,320 Speaker 1: a pretty good job sort of recreating the magic of 472 00:27:27,359 --> 00:27:29,480 Speaker 1: that first one. No, I really I love that, and 473 00:27:29,560 --> 00:27:31,560 Speaker 1: I love like you know, we talk a lot about 474 00:27:31,920 --> 00:27:35,520 Speaker 1: quantitative finance in our work, and we'll talk about various 475 00:27:35,560 --> 00:27:40,320 Speaker 1: well known strategies, momentum strategies and other uses of alternative data. 476 00:27:40,400 --> 00:27:42,960 Speaker 1: We talked about that all the time and in our reporting, 477 00:27:43,320 --> 00:27:46,080 Speaker 1: but we don't talk about the sort of what it 478 00:27:46,280 --> 00:27:48,639 Speaker 1: what needs to happen for people to come up with 479 00:27:48,680 --> 00:27:51,320 Speaker 1: that stuff, and the idea of that this stuff has 480 00:27:51,359 --> 00:27:54,560 Speaker 1: to happen through recruitment and culture and academic study. So 481 00:27:54,600 --> 00:27:58,399 Speaker 1: I feel like this is a interesting, unexplored facet of 482 00:27:58,400 --> 00:28:00,760 Speaker 1: all this. Yeah, I absolutely agree. And I have to 483 00:28:00,840 --> 00:28:02,960 Speaker 1: say some of the machine learning that we were talking 484 00:28:03,000 --> 00:28:06,560 Speaker 1: about this notion that bots, once they realized that they 485 00:28:06,800 --> 00:28:09,199 Speaker 1: were probably not going to win, or they didn't have 486 00:28:09,240 --> 00:28:11,280 Speaker 1: a good chance of winning, they went and they hid 487 00:28:11,320 --> 00:28:15,480 Speaker 1: in some obscure corner of the haylight galaxy. That kind 488 00:28:15,520 --> 00:28:18,399 Speaker 1: of strategy is just really fascinating, and it's amazing to 489 00:28:18,480 --> 00:28:23,159 Speaker 1: think that high schoolers potentially are coding that kind of 490 00:28:23,240 --> 00:28:25,639 Speaker 1: learning into the system. And the fact that in the 491 00:28:25,720 --> 00:28:28,000 Speaker 1: early rounds of the game they weren't doing that and 492 00:28:28,040 --> 00:28:32,040 Speaker 1: that they learned that sort of adaptive approach over time 493 00:28:32,119 --> 00:28:34,960 Speaker 1: is really fascinating. Does it make you think of Skynet? 494 00:28:35,160 --> 00:28:39,080 Speaker 1: Makes me think of skynet a little bit, definitely. Okay, 495 00:28:39,120 --> 00:28:42,120 Speaker 1: all right, this has been another edition of the Odd 496 00:28:42,160 --> 00:28:45,160 Speaker 1: Lots Podcast. I'm Tracy Alloway. You can follow me on 497 00:28:45,200 --> 00:28:48,280 Speaker 1: Twitter at Tracy Alloway, and I'm Joe Why isn't all. 498 00:28:48,360 --> 00:28:51,080 Speaker 1: You can follow me on Twitter at the Stalwart and 499 00:28:51,120 --> 00:28:55,200 Speaker 1: be sure to follow our hard working producer Topur Foreheads 500 00:28:55,680 --> 00:28:58,640 Speaker 1: at foreheads T, as well as the head of podcast 501 00:28:58,720 --> 00:29:02,959 Speaker 1: at Bloomberg, princesco be at Francesca today. Thanks for listening.