Invest Like the Best with Patrick O'Shaughnessy - Tyler Cowen & Daniel Gross - Identifying Talent - [Invest Like the Best, EP. 277]

Starting point is 00:00:00 This episode of Invest Like the Best is sponsored by Canalyst. Canalyst is the leading destination for public company data and analysis. Founded by a former byside analyst who encountered friction sourcing, building, and updating models, canalyst is now used by over 400 institutions, including the largest money managers globally, and by a number of guests on the show. With detailed company-specific models and data on virtually every public company, panelists clients are able to ramp up faster, update models instantly, and incorporate the highest quality fundamental data into any workflow.

Starting point is 00:00:30 If you're a professional equity investor and haven't talked to Canalyst recently, you should give them a shout. Learn more and try Canalyst for yourself at canalyst.com slash Patrick. That's C-A-N-A-L-Y-S-T dot com slash Patrick. Stay tuned after the episode for my conversation with Canales customer, Giuseppe Coco of L-K advisors. We talk about how Giuseppe has built Canales into his process as an international investor and much more. Today's episode is sponsored by Brex, the integrated financial platform trusted by the world's most innovative entrepreneurs and fastest growing companies. With Brex, you can move money fast for instant impact with high limit corporate cards, payments, venture debt, and spend management

Starting point is 00:01:09 software all in one place. Ready to accelerate your business, learn more at brex.com slash best. That's BREX.com slash best. Hello and welcome, everyone. I'm Patrick O'Shaughnessy and this is Invest Like the Best. This show is an open-ended exploration of markets, ideas, stories, and strategies that will help you better invest both your time and your money. Invest like the best is part of the Colossus family of podcasts, and you can access all our podcasts, including edited transcripts, show notes, and other resources to keep learning at join colossus.com. Patrick O'Shaughnessy is the CEO of O'Shaughnessy Asset Management. All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of O'Shaunacy

Starting point is 00:01:58 asset management. This podcast is for informational purposes only. and should not be relied upon as a basis for investment decisions. Clients of O'Shaughnessy asset management may maintain positions and the securities discussed in this podcast. My guests today are Tyler Cowan and Daniel Gross. Tyler is an economics professor and creator of one of the most popular economics blogs on the internet. Daniel is the founder of Startup Accelerator Pioneer,

Starting point is 00:02:24 having previously been a director at Apple and a partner at Y Combinator. Both Daniel and Tyler are prolific talent spotters, and that is the focus of our discussion and their new book, which is called Talent. Please enjoy this conversation with Tyler Cowan and Daniel Gross. Tyler and Daniel, I just a few days ago finished your fantastic new book on talent. I think it's a topic that everyone is implicitly interested in and is incredibly important in the modern world. And I think we have to begin our conversation since we'll go into lots of facets of this topic with some basic ground leveling definitions. So I'd love for each of you, maybe Tyler,

Starting point is 00:02:59 starting with you, to just give what your definition is of talent, having now investigated this topic so much so that you're able to write a book about it. We treat talent as having a creative energy that can alter outcomes and make projects better in ways that matter. Daniel, would you define it any way different? I think it's very related and thank you, of course, for having us on. We're honored to be here. I think it is the ability ultimately to positively manipulate whatever current environment you're in to achieve your desired outcome. There's many different ways of doing that. Some people do that by intellect. Some people do that by charisma. Some people do that by physical strength. But that is ultimately, I think, the basic definition of talent.

Starting point is 00:03:37 How do you think about the difference between means and ends when it comes to talent specifically if both of you mentioned something kind of around the ability to shift the environment in which you're in or affect some end, some goals, some outcomes, some achievement? Maybe others are just really good at doing something. They're talented at something without a goal in mind. What did you learn or think about means versus ends when assessing talent? I look for people who really value the means. So if someone says to you, I want to be.

Starting point is 00:04:04 be the best drummer of all time. Well, that's definitely better than no ambition at all. But I actually want to hear someone saying, I love to drum. And that's the person who's more likely to become the best drummer of all time. Daniel, I'm curious what you think about this idea of Diet Coke. I asked for strange theories or little things that you discovered as ideas. What is this Diet Coke idea? And why is that relevant? I was just saying, I think it's a kind of interesting and abstract that you have Elon Musk, Bill Gates, Lawrence Summers, John Carmack, all these different kinds of conditions. across very different spectrums who all have at some point in their life, and some of them, even to this day, seem to be guzzling Diet Coke. Like if I told you in abstract, there was a secret

Starting point is 00:04:44 black potion that no one knew the recipe of that was made in Mississippi that Titans of Industry seem to drink. You'd say, well, maybe I should drink that too. Now, I know everyone's dietitian told them don't have Diet Coke. It has aspartame and whatnot, and it's extremely chemical, but I don't know. It seems to be working really well for these people. Well, you might say the rest of the country also drinks Diet Coke, too, so I'm not sure how valuable that is as a metric. The more important, interesting thing we were talking about before that the Diet Coke idea is getting at is just raw potential energy. You come into our book and its podcast and more broadly wondering how to assess talent, what to look for in people, how to get better at that. There's all sorts of frameworks

Starting point is 00:05:21 one can throw at with all sorts of fancy words and acronyms, Myers-Braves, MBTI, the Big Five Aspect Scale. And those are important. We can maybe talk about those if we want. But the real thing to look for is raw potential energy that someone has. There's something I find interesting about vigorousness and vitality of people that seem to be consuming all sorts of chemicals, compounds, whatnot, and it seem to be fine regardless versus someone that's very careful and specific about what they do, don't want to have caffeine before bed, whatnot. The flip side, you meet the top salespeople at a certain company and whatnot. And they're just pushing forward. They're having a double espresso after dinner. And I find that sign of physical health and stamina to be something to look for that I think

Starting point is 00:06:00 people don't focus on that much because it's not that intellectually clever or cute. But I think it's just a very basic thing. If you recruiters I work with, first thing I ask people to do, they bring me a long list of candidates and they're going to have to go through all sorts of the different signaling mechanisms you get, like who went to what school and whatnot and who has a better get-up profile. I just asked some of your top three people just by energy, who is the most energetic person to talk to?

Starting point is 00:06:22 And I think if we take our previous definition of talent and reapply it here, someone who just has energy to try in many different ways to positively manipulate the universe to do their bidding is just going to have more shots on goal. I was going to ask you to quantify or discuss the nuance around energy because there's ram down the wall energy. There's subtle ever-present energy. Tyler, in your mind, what type of energy is valuable as you think about talent? People who always want to get to the next destination and achieve something.

Starting point is 00:06:50 And they're very focused on that. And one reason why I tend to agree with Daniel on the Diet Coke question, Diet Coke is something you can almost always get a hold of in the Western world. So if that's your fixation for your nervous energy, you're not going to be frustrated that often. You can always get a Diet Coke. So at some mix of extreme habit formation, a kind of impulse within yourself plus caffeine is how to model that. And if you think, well, how many substances fit all of those three markers? Actually, Diet Coke probably would seem to win. When I used to play chess when I was much younger, The thing I would look at is, how does the person pick up the chest pieces? Do they seem energetic in their moves? Do they seem focused? Do they seem like they give a damn? In most areas that are quite non-physical, you can find physical parallels that you can look at to get a sense of energy, attentiveness, and focus. Even the non-physical world is intensely physical, as I understand it.

Starting point is 00:07:41 I think it's obvious why talent's an important topic, because especially in the symbol manipulation world that we live in in the world of business today, talent is outcomes. In most companies, companies, especially modern technology companies. So I think it's obvious why we would want to search for talent. My hesitance to go straight into the quantitative is important because I come from a quantitative background. I try to jam everything through this lens. But I'm curious what you think about the sort of moneyball-like approach to talent. Having investigated, Daniel, you listed off some of the famous personality inventories. Can this be quantified? What is your view here? Are there useful tools for us to say quantitatively, this person's more talented than this person? Or do we really need to stick in the qualitative realm? I would say three things. First, I think the field of social sciences

Starting point is 00:08:24 has a lot to be discovered. I really think we are probably at day one. Our understanding today, using the frameworks that people talk about in research and literature, as far as I can tell, and Tyler, we'll know much more about this and we'll have some color in a moment, but we're really staring at 1920s film. This is not 4K Marvel cinematic universe stuff. This is like really low, low resolution stuff. In fact, we're trying to take an incredibly high resolution image which is someone's personality and predicting how that actor will work in different environments through a very, very, very now of constraining lossy compression filter called the Big Five Aspect Scale. So it really does not capture a lot. It is what we have. And I know that statisticians will come in

Starting point is 00:09:04 and say, well, you know, we have all these surveys that show that there are a big five personality traits and they don't correlate, but I really tend to not believe that. So I take one, the tools that we have are incredibly coarse. You can kind of get a sense at the end of the day that it's much more subtle because if you look at what machine learning does in other fields, like say, with TikTok, you realize there's more than five graphs on TikTok. So there are more than five different macro personality types. And that's just a bad way of thinking about things. Second thing I'd say is outcomes measuring people by the outcome in the previous work they did because our tools are so coarse, I think is a far better way to do it. So feel the software engineering

Starting point is 00:09:40 where I have the most experience, just looking at the code someone's written is going to be far more predictive, I think of any type of IQ test or psychometric tests are going to get them. And the really nice thing in the software world, by the way, is a lot of the code happens out there in the open. You can actually figure out who the best software engineers are in the world in open source, quantitatively, actually, just looking by their code. So I'd say if there was an interesting quantitative thing to look at, it would actually be the proof of a work. Now, not everything works this way. Obviously, if you're a physicist at Stanford, it's not the same as looking at your C++ code. But assessing the prior body of work, I think, is far richer a thing to look at, and

Starting point is 00:10:14 pine on, then the results on the test. And the last thing, the final thing I'd say is fundamental constraint, I think, of all personality tests, not IQ tests, is that they're all self-report at the end of the day. These are people assessing their own personality. Obviously, you have gaming issues where someone's assessing the personality they want to be as opposed to where they are. You have also people just repeating the same test throughout the day, delivering you completely different results. So I think that field is interesting. And I know I said three points, but maybe one final fourth point. I think they are probably helpful in the way that they provide vocabulary for a team. It might be true that you never want to use psychometrics really to measure anyone in your team or forget who to interview, but it's probably also helpful to have everyone on the team understand when you say openness exactly what you mean. When you say conscientiousness, exactly what you mean. They provide a language, and that is quite powerful. I think language was probably the original technology of our species in scare quotes. Tyler, I don't know if you have any related color to add on that or maybe disagree entirely. I would follow up on the money ball example. I'm a big believer in the best metrics. And sports has a lot of the very best metrics.

Starting point is 00:11:13 But the problem is, other people learn them. So Moneyball only worked for a few years or in the NBA for a while. Three-point shots were underrated. That hasn't been the case now for decades. So you need to learn those things that your competitors aren't learning. And over time, those become more like a form of appreciation, almost like art or music appreciation. And the NBA, the frontier question is, which of these tremendous athletes is actually going to be a knucklehead and which will be a team player? And that's what you want to learn, because you've measured everything possible. You have all the tape. So if you just stick with metrics, you will end up behind your sector because those are easier for people to copy. I'd love me to riff, Tyler, on the pricing

Starting point is 00:11:53 of talent. If you think about it as a market, as you said in the NBA, the whole idea here was to get talent for cheaper than what should be its market price. And there's an arms race there. Just like in quantitative investing in public markets, signals have a half-life, generally speaking, with maybe some exceptions, but in general, the signals have a half-life. So how do you think about the market for talent. And if that is the goal to be able to hire people at the best price or at the right price, how this should sit inside of a company. If a company is mostly going to be driven and thrive or die based on its talent, how should a company think about the pricing of talent specifically? In the talent literature, there's often a big confusion between what is good in absolute

Starting point is 00:12:31 terms, smarts, work ethic, conscientiousness, of course, and what is undervalued in the marketplace. And if you simply pursue those things that sound good, like smarts and conscientiousness, unless you have a comparative advantage at getting the smartest people to work for you. So maybe Harvard has that comparative advantage. But most of the rest of us do not. So you have to look for the differential factors that you can either spot better or recruit better. And those again tend to be more subjective. So does the question, how readily can a person figure out his or her role on a team?

Starting point is 00:13:05 Much harder to measure than smarts or conscientious. It's an area where you might retain some kind of enduring comparative advantage. So not to confuse absolute importance with what's underrated. What do you think is most overpriced as a dimension of talent in today's labor marketplace? Depends on the job. For lower level jobs, everyone's looking for conscientiousness. It obviously is important. But if you're trying to hire away conscientious people from other organizations, the organization

Starting point is 00:13:34 knows they're conscientious. They want to keep the good ones. They let the lemons come to you. So I would just say, be very careful when chasing after conscientiousness for its own sake. Get without at all denying its absolute importance. Daniel, what would you say? One thing I've observed in both founders that I've had the honor and pleasure I guess I've working with as well as just hiring people on my team is that you have a consensus mechanism

Starting point is 00:13:56 for hiring talent. You end up optimizing for very even-keel people. The most important thing to pay attention to the most kind of mispriced people, so to speak, are ones where people on the team vehemently love them and vehemently hate them, because you often get outliers in more than one dimension. You'll have people that are a typical example, like incredibly productive but difficult to deal with, people that are incredibly extroverted and fun to deal with, but very, very bad at dotting the eyes and crossing the T's. Consensus mechanisms don't end up hiring those people because it's a bit painful to push through,

Starting point is 00:14:27 but I think those are the most important people. The best way, I think, to go about that is to have an organization where you can structure it such that the Vaughnoyman is only doing math and it is not doing any HR. Really, really focusing on specialization is good. And that really broadens your lens because you can now start to think of, who can I hire that somewhat actually disfigured in some ways and just have them focus on one specific thing. When you find those people and you can get them out of the general big organization that they're in, if they're stuck there, it's usually a great blessing for them too. Your prototypical example is the engineer that's really, really good, but stuck in the cogs of a large company and you just let them go at a five-person team and they cut

Starting point is 00:15:02 through red tape and it's really amazing. Those people often again, when that company grows, they don't work anymore because there's much interfacing with other people and they need to find the next small team. Just searching for unevenness I think is really helpful and people don't do that. I feel much better about a hire when I know what's wrong with the person. Exactly. And part of the paradox, the people bringing me the hire just want to sell me on the person and that's the wrong attitude. I try to beat that out of them a bit. I'm like, well, tell me why they're a good match for us. How were they screwed up? How were they uneven? And then possibly the mental picture can click. Once hired someone who on paper was perfect, in the interview was perfect. He showed up,

Starting point is 00:15:38 really just did no work, period, did nothing, had no ambition, really, had multiple advanced degrees from prestige schools. It was a mistake. And ex-ante, I couldn't answer the question. What's wrong with this person? Why does he or she want to work with us? What do you think the relationship is between experience and or age and talent? It's going to depend on the sector. So if it's chess players, you can be washed up at 35 or 40. But if it's novelists or philosophers or, in fact, people hiring talent, you tend to get better at that with age up until some extreme point.

Starting point is 00:16:12 So a lot of the advice we give, it's highly context specific. You might just want boring semi-low energy people if your product, say, is life insurance. Can't take too much risk. You're highly regulated. Maybe you have some kind of favored market position, just to keep the ship moving, in essence. So it's going to depend. If you think about talent, back to that absolute versus relative, so you could be talented at a small, mid-level job. But I think the talent that we're all interested in is people that are going to make some sort of strange, large, orthogonal impact on a domain around them that they're interested in.

Starting point is 00:16:46 What about then? And do you think that age for someone that's going to make a massive impact, is there any evidence that you want them to start younger, that you want to back them when they're younger and less experience because they're less tainted by the normal ways of doing things in a given field? Is there anything there if we're thinking just about talent as measured by like huge impact in building something new? Of course, people live too at this point say most founders are young and then Juniper Networks and Jim Simons from Rentech and all these people that started their career actually relatively later on in life. So I think there are examples both ways. I think it is a fair statement that by and large, if you had to average the age of successful tech founders, you would school young. I mean, that is just the truth. Now, part of it is that those people don't actually fit into any other part of society. So it's not something innate with the youngness, although I think there's definitely something there. But also, these are outcasts that got fired from large companies,

Starting point is 00:17:37 didn't fit into college left. God bless the internet, because it provides not just a great employment opportunity for these misfits, but also a way for them to make incredibly successful companies and obviously impact on the world. You definitely have the flip side, though, in areas of management, say engineering management as an example, a lot of the job is pattern recognition, both in the type of person that you're dealing with hiring, I've seen that type of person before, seen this type of situation before, and infrastructure decisions. The early stage companies with the early 20-year-old founder or whatnot, I'm saying this as someone who was. I was very young when I started my first company. You tend to succeed in spite of the age sometimes, not because of, because you make terrible infrastructure

Starting point is 00:18:15 decisions. You have no infrastructure taste. So you're just trying things left and right. You go through nine different decisions where someone who's seen something before usually just helps you go to the best path. Sometimes the best teams are one where you have a relatively young founder and one or two people on the executive team that have seen things before now common. Pitfalls here is obviously you don't want a bunch of suits in the room. The old Steve Jobs quotes, everyone's heard, whatnot. But it's helpful, I think, to have a balance just because there are roles where taste matters. I used to work at Apple, used to run searching machine learning at Apple. And one thing Apple thought of that I thought was interesting, there's this little pamphlet I got at some point of basically why I'm not a senior

Starting point is 00:18:49 vice president. And the responsibilities people have all the way from the lowest rung of the organization to the CEO. And the way it was constructed is the cost of making an erroneous decision, how long of a time would take Apple to recover from that? A engineer, rank and file engineers, maybe three to four hours, slightly more senior engineers, maybe a week. Managers, maybe two to three weeks. A director is maybe a month or two. A vice president is two or three quarters. And all the way up to senior vice president's two or three years. And if Tim makes a strategic error could be a decade until Apple recovers. I think that's a good sense of also when you want a care of age. How large is the penalty of making a mistake? You would certainly not want your

Starting point is 00:19:30 neurologist or anesthesiologist to tell you, hey, I just got into Ycombinator and I read a few good Wikipedia articles and I'm ready to go. That's another helpful way to think about age, is basically the cost of an error. For all the talk about youth, I still think that something like 14 to 19 year olds remain underrated. And in particular, I'm thinking about the realm I work in, which is the world of ideas, where if you get something wrong, it can just be ignored. You're not throwing all of Apple, of course, and that if you want to shape ideas that you should be talking to, writing for, spending time with, people in that age bracket, 14 to 19, they're so smart, they see things afresh, they're willing to take chances, can be just full of energy.

Starting point is 00:20:09 And I think the rest of the world still doesn't see that. They put out their papers or their books for someone who's 47 years old and might review their book. I mean, that's fine. I'm not that those people don't matter. But they don't quite grasp that some kid who's 16 is smarter than they are. And that's the person who's really going to matter. So I think we're still underrating that in many areas. And finance and tech, in particular, music, chess, those are mostly the domains of youth.

Starting point is 00:20:36 All of those make me think of intelligence, just raw intelligence. And so I'm curious what you learned about the utility of IQ or other measures. I'm not terribly familiar with measures beyond IQ and their relationship to talent. I'm thinking of David Rubinstein at Carlisle Group, who told me a couple weeks ago, very often the biggest problems were the highest IQ hires that they made. So there's an interesting question of should hire IQ be founders versus employees or managers. But what did you learn about IQ and its usefulness or where it may steer you wrong? I think it's very often overrated.

Starting point is 00:21:08 So for many jobs, you have to be at a certain level of smarts, but above that level, it doesn't correlate well with success. Let me just give you two statistical measurements. First is a study of Swedish CEOs of the larger Swedish firms. The median Swedish CEO is at the 83rd percentile of IQ, which is pretty good, smarter than 83% of the rest of the Swedish people on average. But it's not exceptional. Before doing all the work for this book, I might have thought, well, that's going to be 95,

Starting point is 00:21:38 95, 96, 97, for a lot of them 99. But 83, it's a little underwhelming. Here's another number. Lower in the distribution. If you move someone from the 25th percentile of IQ up to the 75th, that's a big leap, 50 percentile points, they will on average earn 10 to 16 percent more. Okay, I'd love to earn 16 percent more, but that's it. So smart people in particular, I think they significantly overrate IQ with the caveat that for a whole, whole bunch of jobs. Yes, of course, you have to be above a certain level to do it at all. I think that's right. I mean, these things tend to go in waves. If you listen to the Radio Lab podcast of the early 2000s, their G, IQ is like a big deal. And now I think we've been

Starting point is 00:22:22 talking about it so much. We're in this countercyclical wave. We're kind of realizing there are limits to the returns of focus on IQ. And I wholeheartedly agree with everything Tyler said, with the caveat that I think anyone certainly in this podcast recording that we're on, but also anyone probably listening to the podcast exists in a very special world where most of the people they're surrounded by are very intelligent. When you leave that world, you realize there is a baseline of IQ that matters quite a bit for the roles that we're talking about. Productive employment, I think is possible at many different ranges of IQ. But for top tier talent, there is definitely a baseline that matters that if you don't have, it's going to be difficult. I think it is probably

Starting point is 00:22:57 on the margins to something that people might focus on maybe too much instead of too little. I think there are also some jobs that fit what I would call the multiplicity. model. Take someone like a Bill Gates. Someone like that has to be in the very top tier of at least seven or eight different qualities. One would be energy, another would be work ethic, another would be IQ. And if even a single one of those falls away, Bill Gates can't be Bill Gates anymore. Those typically are very, very high-level jobs. We've heard of the people who hold them. And there, I think, a super high IQ is just an absolute prerequisite. But so are seven or eight other things. In the marginal sense, it's not only IQ that matters absolutely. You've referenced this idea of

Starting point is 00:23:38 outsider or outcast or somebody that rejected some institution that they were in. How important is all of that? What's beneath that as an interesting variable? Is it some set of early experiences? Is it a personality characteristic? I guess it's sort of a nature and nurture question on outsiders and why they're important. If you're an outsider, you take more chances. You're not surrounded by peers who want you to do a specific thing. I think that's super important. If I think of the chances I took in my own career that worked out very well, a large number, maybe all of them stemmed from the fact that I'm at George Mason University and not Harvard. And what's expected of me is simply something totally different. I'm allowed to define that for myself. No matter how independent-minded you are,

Starting point is 00:24:20 it's very hard to avoid conformism and getting locked into playing and winning a status game that maybe isn't the game that can matter. So if you're at Harvard, you want to publish a great number of papers and high-impact journals. If you're talented, you can do very well at that. But for a lot of people, it's a dead end in terms of actual influence or improving the world. And people from other places, on average, they're going to do much worse. But some number of them will just do totally different things.

Starting point is 00:24:44 And they're not worried about, hey, what will my buddies at Harvard think of me? I'd love to turn to the notion of evaluating talent. Obviously, you've both done quite a lot of this and studied how others do it. I've got a few topics that I'd love to hear what you learned as you investigated them. Maybe the most interesting to me is this investigation of downtime, like what people do in their downtime. I'd love you to outline why that's interesting and very specifically how you pull it out of somebody. If you're interviewing them or doing a reference check or something, talk to me about the utility of this variable. It's kind of similar to this idea of why looking at someone's prior

Starting point is 00:25:17 body of work is a bit better than a test, meaning given the universe of available options, the real test is what you decide to do every given second moment and how you weigh your opportunity costs. It's exceptionally interesting, of course, to ask what people do when they don't have any particular obligations. So they're downtime. The movies they watch, the activities they do, the books they read. You can just ask people what kind of movies, TV shows, books, subreddits, links you find interesting. Tyler taught me which I think is a wonderful idea. There's many layers to thinking about the answer. But the first layer that's quite important is looking for specific answers instead of just generalizations. You ask someone about their favorite movies.

Starting point is 00:25:55 It's really helpful if they say, I really like The Matrix, I really like Pulp Fiction, I really like Catch Me If You Can, versus action movies and stuff. There's something about that, the person who's able to articulate specific proper nouns. So that's definitely one baseline thing to work on. Tyler, what do you think? This is really an area where you're the master of this domain. If the key thing you're trying to learn is how does the person work together in teams, small teams, everyone is prepped with their examples of how great they are or the mistakes

Starting point is 00:26:23 they learned from, but ask them about a sports team, if they know basketball, the Los Angeles Lakers, why didn't they do well this year? Now, you're not testing them about the knowledge of that. You want to pick something you know they know, and they feel confident about. But if you get vague generalities, like their morale was low, well, okay, it might be true, but what exactly went wrong? Or if they know movies, well, in Star Wars, Luke Leia and Hahn, they seem to make a lot of mistakes. What did they do wrong? Did they listen to Yoda too much? Like, talk me through that. they can't prep for that kind of question, and you get their understanding of personal dynamics. If they read a lot, ask them about their favorite Shakespeare play, Lady Macbeth, how evil is she?

Starting point is 00:27:02 What was her big mistake? Whatever you think they can speak to and see how well they can speak to it. And you want to look for the questions. How well do they understand the hierarchies at work in the situations they're perceiving? What kind of resentments come out when they're talking? What level of detail can they show? and how varied and complex is their understanding of other human beings' psychological mechanisms. It's very hard for someone to fool you with that kind of test.

Starting point is 00:27:27 Is this all a study of their curiosity? Is that what it boils down to? That too. So you can test their curiosity by seeing how many different areas they can talk intelligently about. But surely they ought to have one. Like you want to let them pick the area. You have to know something about it. But you're not testing their knowledge base.

Starting point is 00:27:44 You're testing how well they can think about what they know best. Tyler, do you have a favorite on-ramp question to get into this part of an interview with somebody, like a way to get down this path that you ask often? I'll look at their age and where they're from and whatever clues I have. And it could be books or movies or a travel spot or a sports team. I'm comfortable with any of those. Again, I'm trying to cater to their expertise. It can't be something I know nothing about.

Starting point is 00:28:08 I can't say, well, who's the greatest player in Canadian curling and how does he train? Right. It's harder for me to draw information from that. but I've hardly ever met a case where I can't ask them anything. Sometimes some other cultures it's harder, Nigeria. But even there, I'll talk about the Kootty family, who were the generationally best performers and why, and how did they organize their groups? There's always something. When learning about somebody, it's become very popular to say that interviews are maybe necessary, but not all that useful, whereas references can be incredibly useful at getting to know the underlying

Starting point is 00:28:40 person. To what extent do you both agree with that concept? Do you value references in the is very much if you're trying to back somebody or invest in them or hire them. It's a specific context. But do you think that references is an art that's way more important than actually interviewing and getting to know the person directly? And if so, by how much? I think the correct answer is definitely references are useful. One of the things Tyler, I think, asks people, not necessarily in interviews, but in general is what one's equivalent of playing scales is. So if you're pianist, you can sit in front of the piano every single day and play scales and just get better and better and better. And so finally, you go to Carnegie Hall, you step in on the Steinway, and you play Moonlight Sonata and it all works.

Starting point is 00:29:18 I think a different field is an interesting question. What is the thing I can do every day, where I can just compound? And one, in I think the particular area of talent is actually building a network in your field. That is a compounding activity because that increases the value of the reference that you're going to get. The signal that you're going to get from a reference is strictly a function of just how close you are to that person. And obviously, the closer you are, the more honest to read you'll get. And there's all sorts of things you could do. You could force them to stack rank the individual somehow. On the scale of 1 to 10, how good at Bob was at software engineering is a good question.

Starting point is 00:29:50 And then when they say eight and a half out of 10, you can say, well, why is 8 and a half the right number for them? And that kind of gets a little bit of alpha that's kind of helpful because they might say things that aren't just effusively positive. You'll just get more signal if you know the person better. It's actually a very helpful thing is to have you or the recruiter and your team have a very, very strong network somehow. Because references, yeah, are earlier we made the case that their prior body of work is really the best way to evaluate someone. And us neurons have like code you can look at. But in all honesty, all roles are interpersonal. References are just their body of work in quotes that they've done, how fun they are to interact with.

Starting point is 00:30:24 The other final little anecdote I'll say about references is it's always a great opportunity to use that as a way to source candidates too. People don't take advantage of this. But let's imagine you're looking to hire someone to be a engineering manager of folks on your quant fund. You're now on the phone with that person's manager, say. And so it's an amazing time to both, A, try and hire them if they seem good, but B, just ask them, who else would we be talking to? It's a small trick I've used to grade advantage over the years. I think references are essential for jobs requiring a lot of conscientiousness that are mid-level.

Starting point is 00:30:56 And then they absolutely rule. But especially if you're looking at very young people, I find references not useless, but why should you trust their judgment over yours? say someone who's a teenager. Who is that you think they know that has better judgment than you? I think also for creativity, do their references have a high enough level of creativity to see this person's creativity? I mean, they might, but you certainly shouldn't assume they do, again, especially for the young. A simple thought question, say Mark Zuckerberg at age 18 or Peter Thiel at that age, what are their references like? They're probably not negative,

Starting point is 00:31:30 but the people they know at that time, are those people good enough to see how good like Peter and Mark would become, highly unclear. So there's some irreducible side of the equation, especially for creativity, especially for the young, where you have to use your own judgment. And if nothing else, you're ending up having to assess the referring person. So why not just assess the candidate himself or herself in those cases, but do both? Do you think that there are any beyond this idea of prior work and using the reference as the best way to understand prior work for a lot of people? Is there anything else that you recommend people listening, try to get out of a reference call as an objective?

Starting point is 00:32:06 Making the referer feel safe to criticize, which is very hard to do. I wouldn't say you need to trick them, but you have to go overboard and ask something like, would you work with this person or could you, in a ranked quantitative sense, compare them to other people I might know? You have to really force their hand to get a slight downgrade of comment and then take that seriously when you hear that downgrade. And the amount you have to work is strictly a function of how well you know that person or how well, you know, the friend that referred you to that person. I think there's some type of composite function there. One of the interesting chapters in the book

Starting point is 00:32:44 is about disability. I find this area really interesting as it pertains to talent. We talked about specialization a little bit already. What did you learn about different kinds of disability and their relationship to talent, maybe to start with a simple one? The concept of ADHD comes up. I don't know if you would call that a disability or not, but it's something that comes up a lot, certainly in my world. And very often, some of the most successful people have this condition and they've figured out how to control it or hone it or pointed at something specific. What did you learn about disability and what led you to write that chapter specifically? I would say what we call disabilities very often or not, maybe sometimes they would be

Starting point is 00:33:20 for your entire life. But in the workplace with division of labor, you can be more obsessive have a very specialized talent. People with ADHD, often they hone it and they use it to propel themselves forward. So one version of ADHD might be, well, you can ever finish a book or finish a thought. I think that's less likely the case. If you use ADHD with, well, I'm done with this diet Coke. I want another. I'm done with this page. I'm going to turn the page. I'm getting bored with this book. So yes, I'm going to finish it, but I want to make sure I get to another book. It becomes a positive, propulsive mechanism. And it's a high variance thing to have. But I think, again, On average, it's correlated with success, not failure.

Starting point is 00:33:59 What about more extreme forms of disability, whether that's something like autism? And again, maybe the same answer that you gave for ADHD in some ways. But what other forms of disability are interesting as it relates to talent that you studied? Extreme is a hard word to you. So Elon Musk gets on Saturday Night Live and says he's Asperger's, which is essentially the same as autism. He's the most successful, greatest entrepreneur of our generation. Again, I don't know how happy he is. but you need to take that seriously.

Starting point is 00:34:26 My goodness, look at what he's done. His command of engineering is phenomenal, his ability to achieve, be out there in the public arena, create memes, everything. My goodness, how many major companies has you been a part of? So I think high variance is the model. It's not that everyone with the so-called disability is actually a superperson, but it's assigned to look more closely. What could this person do for me?

Starting point is 00:34:48 And because they might be weird or come off as weird, the chance you can hire them is actually probably much higher. If you understand what they're about, they may need some kind of special accommodation. Well, we understand you, we'll give you that. Your chance of recruiting them is higher too. Do you think there's a relationship between happiness and talent? Do you think very talented people are less or more happy? That classification of happiness, type 1, 2, 3.

Starting point is 00:35:12 Type 1 is happy in the moment, going down a roller coaster ride for those leg roller coasters. Type 2 is you're on a hike. It's tough. In hindsight, you reflect positively on it. But in the moment you wish you could stop, type 3 is just, well, just sucks both ways. My sense is certainly successful and I think talented people are more of the type two happiness. I actually don't think Elon has in the moment an amazing life. I think he's going to have years and decades and probably centuries of positive reflection on it. Look how cool it was.

Starting point is 00:35:42 But I think when you're at that last stage, what is it, the tending step on Everest and you're just trying to make it to the top and not die. I don't think that's fun at all. I think the average day in Elon's life is, there's a lot going on. I don't know how pleasurable it is. So maybe there's a difference between happiness and pleasure in that sense. It's also not clear to me, and this is more of a personal observation than one speaking that on behalf of literature, how much striving for happiness is the exclusive thing one should be seeking. I think that's a very kind of Buddhist idea. I'm not sure that actually gets you to happiness, so to speak, meaning I've often, if you really strive for like constant pleasure happy, you do get a lot of type one happiness, but I'm not actually

Starting point is 00:36:22 sure how long that lasts, the circuitry in the brain is very much triggered not to want that 24-7. So even if you're on a roller coaster every day, by the 10th day, it's not that exciting. It's a bit of a meandering answer. One final thing I will say, which is I do find talented people's default state, many, not all, but it is something to look for. Their default state giving, say, the same set of inputs, if you could say the same set of stimulus to like 10 different people and whatever that you want to define stimulus as a series of Gmail messages, some good, some bad, hike, whatever.

Starting point is 00:36:53 The default optimism I find talent people have is a bit greater. The belief that things will be great, things are great, things will continue to be great, that does seem to be a bit more present. What I've always wondered is to the extent to which that is a biomechanical mechanism, as opposed to it being something psychological, I really do wonder if it's just a sense of liveliness and vitality. And now we're really going off the rails. But to what extent are folks that are just not happy by default,

Starting point is 00:37:19 to what extent is that something physical, some type of vagus nerve biome thing. And you certainly know as an extreme if one is actually sick, if you have the flu, you're not happy. That's an extreme. If you sleep really well and it's a great day and there's sunshine outside, you are happy. Maybe that's not a zero to one. Maybe it's not a binary continuum. And so maybe you can be like one percent of just sick. Maybe some people are like that a lot. There's all sorts of wacky evidence to indicate the fact. But that's often what I really wonder when I meet people that are kind of morose all the time is the extent to which I'm not a psychologist, but I really do wonder, is that really some childhood trauma, whatever thing? Or is it something physical in the body that's driving the

Starting point is 00:38:00 mind to predict lower odds of good scenarios and higher odds of bad scenarios? I would put it this way. I look for a positive sense of agency, ability to experience and enjoy flow, and the person not being that content, that trifecta together, I think. think is very powerful. I would second Daniel's remarks on the physical. And I feel this myself. If I'm trying to write something, my body has to be able to squirm in the proper way for me to get the writing done. And it doesn't make any sense. And I'm not saying it's the same setup for everyone. But each person has his or her physical requirements. Are those satisfied in the person? Can they be satisfied? Does the person understand how to mobilize them? Those are good questions.

Starting point is 00:38:43 Say more about the lack of contentment as your third thing that you said there. Is this just something that we're unsatisfied at the presence or you're sort of compelled to continue in exploration? Is that the general idea? What is it about contentment that's important or lack there are, rather? For leadership roles, it should be there's never enough. It's partly something that feeds into charisma, but it's also a belief you want for its own sake. If the person can be made happy by some set of achievements, that to me is definitely a negative. Should never be enough, a kind of restlessness, never happy enough with themselves, always questing for the next thing, the new thing, the higher level. It depends on the sector. But it ought to be something

Starting point is 00:39:21 where there's not a plateau. When I was in Beijing as a ninth grader, there was a group of people outside our hotel that were doing Tai Chi every morning. And at the time, I did a bunch of karate and karate you do these cataas where you memorize moves and then perform this thing. I watched them and then I tried to join them and an older woman slap me on the wrist and said, you know, at your age, you shouldn't be doing Tai Chi because it will kill your ambition. You won't want to do anything else. You'll be content and that's a problem. Does that spend? speak to you at all. Do you think that? Absolutely. But that seems like a good thing. I'm against good things. You're against good things. Say more. It's partly in jest, but the notion of a good thing is a little

Starting point is 00:39:58 too complacent for me. It's a challenging thing. And maybe in a deep way that makes you happier, but don't focus on the happiness. Focus on the process. Focus on the thing. Focus on your quest. Maybe you'll end up happy. Don't think too much about happiness and be suspicious of people who think a lot about unhappiness. I think often they're miserable. and they wallow too much in something or rather, and they end up as complainers. Tyler, from the cheap seats, you seem maybe content's the wrong word. You seem extremely enlivened by the things that you do, intellectual investigation, conversations, writing, ideas.

Starting point is 00:40:32 Is that not contentment? What am I missing there? Well, I always want to do more. Typically, I've always been starting some new project, but it helps me maintain a very high energy level. Let's write another book. Let's do another podcast. The old Ernie Banks sang, let's play two, right?

Starting point is 00:40:46 He meant double headers, but it's a general idea. It's part of my life philosophy. Let's play too. It's interesting. Before we have recorded, I said, what's a book that is your favorite looking back and you said the next one. So it stands up pretty cleanly. But I asked Daniel about me. He'll give a better answer than I would or ask me about him. Yeah, what do you think? What do you think, Daniel? Well, I think Tyler has a very unique ability to consistently produce writing output, which I think is a more mature version of the same mind that excelled playing chess early on in life. It's a very taxing cognitive thing to do. Most people have the dread task of, oh, I need to go

Starting point is 00:41:22 write the thing. And maybe I'll get to that tomorrow, not today. Tyler's the opposite mode of that where he's effectively writing every day. So I think that is a mixture of mental stamina and mental ability. It's an interesting question for someone to ask Tyler that I've certainly asked in the past. You have that experience where you're writing something and it starts out good. And then as you're writing, realize halfway through everything I wrote doesn't make sense. And that original point actually didn't ever connect to us. I'll get back to it tomorrow. But he's working through that puzzle every day, which I think is very impressive. It just takes a certain sense of fortitude. In reality, probably the most interesting thing is the fact that I don't think it's a huge kind of

Starting point is 00:41:58 jaco-willing David Gaghan's struggle for him that he's pushing through some type of courage and fortitude and attaboy stuff. It's the thing it kind of happens. And when you meet people who are really excellent in a particular field, I always find it fascinating where some of them are just not trying that hard. They just are. That photo of Usain Bolt running that nine second smiling and making it look easy at Kipchoga, breaking the marathon record, also making it look easy. There's a related thing going on there, which is people that truly excel at something. They're not trying that hard. They just are good at it. Tyler, how do you see Daniel as most talented? Combination of things. First, an extreme intelligence from birth, which is important for him,

Starting point is 00:42:37 but a willingness and desire to see things differently and see and articulate an angle to a problem that the other very smart people around him just do not see. So on the surface, he's relatively quiet in a group, but is thinking more, calculating more than the other people there, and will come up with something they haven't thought of. And then he combines that with the desire and ability to build something from that. So not just, well, I saw that, people like that on Twitter. Oh, I'm so great. But it has to be converted into something quite tangible and physical. So I think the physical is a big part of Daniel's excellence. He's run marathons in Antarctica. That's not an incidental fact about him. What he's trying to do with those smarts, assessing

Starting point is 00:43:21 projects, people, is to turn it into some real-life equivalent of running the marathon in Antarctica and his ability to integrate the software of his mind with the hardware of the concrete ambition. I think Daniel's just an A-plus. How's that, Daniel? Pretty good. It's a great excuse to talk about the physical. Daniel and I have talked a lot about running in the past. Usain Bolt thing is a great example. You mentioned this physical restlessness you need, Tyler. What is the prescription or the RX coming out of this research for how should people think about the physical side of their lives, even if I think most of the book and most of what we're talking about and thinking about here is mental talent? What is the supporting physical infrastructure beneath that that's interesting,

Starting point is 00:44:00 if anything? I think everyone should exercise every day, if only for investment reasons, but at a deeper level to understand your own physicality and how it interacts with your mental output is a great way to become more productive. And there's not going to be any single answer that's going to hold for everyone, but that I need to squirm on the couch in a certain way. Well, kind of unique to me, but I understand that. And it helps me write better. So to figure out what for you is like me squirming on the couch, I strongly recommend.

Starting point is 00:44:29 Daniel, what have you gotten out of running? Why is this so important? It's funny. I'm smiling because the pursuit of maximizing the self-approvedment, quantified self-stuff, I find it's a funny scene where zero interest in the field is a bad sign. 110% interest in the field is just as bad a sign. In reality, I find most productive people have very deep, very private, a kind of secret interest in the field. Would never ask someone all those Tim Ferriss, although I think Tim Ferriss is great,

Starting point is 00:45:02 but that kind of style of what's your daily routine, do you take cold showers, but secretly are wondering if they should be taking cold showers. That's kind of where I sit. Running happens to be the thing I've chosen to self-identify as and do every day, almost every day. It is funny how actually productive people, when you meet them behind the scenes, there's a veneer that they're really not doing much. Some of them are. Some of them just do sleep fend magnificently every night and really don't work out and actually seem very healthy and vibrant, but find a lot secretly in the QS. seen, but don't really want to talk about it too much. And I think that's probably the healthy balance to have. So I'm not here to reveal any daily routines I have. I'm like 1% of successful

Starting point is 00:45:42 of the guests that you've had on your show. So I'm not the person you should be asking. But yeah, if you do manage to get Ken Griffin to talk about his sauna cycling, I think that'd be helpful. Were there frustrating aspects of the research part of this project, meaning things that you wanted to learn that just didn't have answers or hypotheses that were just totally wrong or unsatisfying things? What was frustrating about this book? I'll say one thing. I'll let Tyler comment, not just with this book, but in general, I have been a bit frustrated with the disconnect between the world of social sciences and the most cutting edge machine learning that we have available to us today. I think there's a bit of a selection effect of who goes into what field in

Starting point is 00:46:21 industry, but you read a study that's purportedly showing results. And studies are truth. That's what we learned over COVID, right? This is a study. It's true. Expert checkers and experts agree. It's true. But in reality, these things are, we got 40 people in the room and we asked them to fill out a Google form four times. But there's so many obvious experiments you would look for. If you were running this as a machine learning team at TikTok, you just say, well, can we take the text output people have and put it into different clusters and heaps, not label them, but put it into different clusters and heaps so that we can figure out the different writing styles people have. Maybe you do the same with video. Maybe you do the same with images. The profile of pictures,

Starting point is 00:46:55 If people select for themselves on social media, gosh, that's got to be predictive in different ways. And there are a few things here and there about it, but nothing really that's gone deep. No one's really brought the arsenal of tools that machine learning offers to social science with the correct level of depth. I run a small company that's doing a little bit of this, but I'm disappointed that there's no open AI for psychology, so to speak. Tyler, what's your take? I would second all that. There are just very few papers, research papers, in these areas that I trust. And the ones I trust the most tend to show, well, Factor X is only so important, like IQ is only so

Starting point is 00:47:28 important. Conscientiousness is only so important. Those actually seem to replicate. The papers that make bigger claims about something being very important tend to fall apart, or they're so context-specific. You have to be very careful with them. But more and more, I think of the quest for talent as a kind of art. Now, in the arts, you can understand works of music or painting or movies far better or far worse, but it's hard to give people fixed rules. You wouldn't say, well, always buy the paintings with red in them or paintings with dogs in them. That'd be pretty crazy. But if you look at enough paintings and talk about paintings with other smart people, you'll understand paintings much, much better. And you can do that with talent. There's something ineffable about the

Starting point is 00:48:10 success. And if you look at some of the biggest successes, such as Peter Thiel, who has this incredible record, Peter is not a formulaic talent searcher. It's highly intuitive. It's almost a kind of moral judgment. His own background is in the humanities, not in number crunching, and that is itself interesting. When building good models, the size of the dataset, we've learned from modern AI research is incredibly important. You both have huge, what I'll call reference classes of people and talented people, and you both actively are searching for talent all the time, whether that's to invest in or interview or write about or back through grants. This is something that matters greatly to both of you. What has changed, if anything, the most about how you

Starting point is 00:48:49 evaluate talent from before beginning this project to now as a result of the work? I would say where the person has ended up, not where they come from, but geographical location predicts talent more than I had thought, and I find this actually somewhat discomforting, since it could have dangerous political implications that you have talent so concentrated, say, in South England, near New York City, Bay Area, some other places, parts of India, but the people who are really talented actually are able very often to make the efforts to get themselves to those places, more than I had thought, and I already had thought that. For me, the big model update has actually been one that I got from Tyler. I started an organization

Starting point is 00:49:28 that built our own IQ test. We built our own psychometrics test. We went really deep and really climbed all the way up, but now all the way down the Dunning Kruger chart of belief in that stuff. And I've greatly reduced my priors on all of those things, with the exceptions that I mentioned earlier, so as a way to provide taxonomy and language to a team and whatnot, and increase my belief, yeah, to Tyler's point on the prior body of work. You know, it's funny, my sister is an organizational psychologist, and I remember coming up to her after I learned about the Big Five Aspect Scales and started to work on implementing our own version of it. She said, oh, you know, you're in the Big Five Aspect Scales, Halo. Come back to me in a year, let's check in

Starting point is 00:50:01 and see where you are, and she was totally right. I do think when people first learn about it, they have a tendency to believe that it is the ultimate utensil for understanding people and whatnot. Again, it's useful as language, but I rate that stuff far lower because I've realized how the sausage is made and the factory doesn't look great. Until we have better tools, the work is the way to go. Another revision I have had is to focus a bit more, not on whom can I find, but who will find me? And it's thinking about the problem in reverse. You don't want it to be either too hard or too easy for someone to find you, at least for what I'm doing. That is at least as important

Starting point is 00:50:34 as me showing up, talking to a person, meeting them, reading their work, whatever, and making my judgment, who's attracted to me? If you're not going to get that right, it's hard to do well. I'll call that the bat signal effect. What have you learned about doing that well? What represents a good bat signal? When I interview candidates for emergent ventures and they say, oh, I'm your biggest fan, I've read all your this, heard all your podcasts, they are not usually the best candidates. They're usually smart, curious people who don't have that much to do. The very best candidates tend to be, I have a friend who loves your stuff. I listen to one of your podcasts before this. interview. They've learned who I am. They have some mental model, but they're not a fan. And I find

Starting point is 00:51:14 that very interesting. It shows it's about your soft network. You can't rely just on your hard network, but who your hard network will bring in as your soft network, you need to be out there in some way with an image that corresponds to what you're trying to pull in and help fund hire, whatever the case may be. Daniel, have you learned anything about this through Pioneer in terms of throwing up the right signal to attract the right people, the sort of pull versus the push? You definitely come across folks who really admire the individual to the point where you realize the fact that you are filling some type of heroic protagonist role model in their life speaks to something that they're missing more than something they have.

Starting point is 00:51:54 Something I saw at YC2, we've come to hear those people. The best people are trying to use your platform to their advantage. It's a great feeling to feel like I'm being slightly taking advantage of. You are a step on the staircase. I'll be stepping on you now to get over here. And it's great. We'd love to invest in people like that. And as they step on us, we buy a little bit of equity and they continue on.

Starting point is 00:52:15 You're really something that they're using as opposed to being the destination. Probably it is a thing that as we become more successful, we get worse and worse. And you kind of see the extreme of this super successful businessman seem to attract around them, these kind of silver tongue types that really like them and admire them. And it does start to distort the thinking of some of these folks when your entire audience or these court gestures that really needed a hero in life. You are that hero. They're trying to to appease you. I think it's a very unhealthy psychological cycle to fall into. You always have to really make sure, first, when you come across someone like that that compliment you and whatnot,

Starting point is 00:52:48 it takes a while to develop T-cell immunity to that. At some point, I think it's very important to become highly allergic to those dynamics and really focus on people that might know you and respect you, but there's a specific goal that's not you that they are trying to achieve, that you might be helpful to get them to another planet. And you need to be very careful. If there's compliments are true, it can be worse. The people who BS you and say things nice about you that aren't true, there's actually maybe a greater chance they're the ones who want to use you. So if you say to yourself, well, that person flatters me, but the things he or she says are true, so that's a good channel of information. That's a very dangerous path to go down. In fact,

Starting point is 00:53:26 it's maybe more dangerous than having some pure BSers by you. There's a section of the book. I've intentionally avoided the area of these personality inventories. You've referenced them a few times. And I think as you go down this path, you sort of are enamored with them originally because there's this physics envy that exists in the non-physics field. You want stuff to jam into a formula, but still, if they're useful as a language, is there one, whether it's MBTI or Ocean or anyogram or whatever it is, is there one that you recommend people check out just to develop that sense of language, that descriptive language around personality characteristics or types? I think the least bad is the big five aspect scale. And it's only.

Starting point is 00:54:05 five, thank goodness. There's a version of it, the big 16. It may have a higher R squared in statistical terms, but you can't use 16 concepts with your team. So big five, it is with a ton of qualifications. You mentioned this notion of geography, which makes me think a little bit of the frictions that might exist between someone talented and getting to one of these hubs of creativity or cities or whatever that might lead to something really interesting because they're a talented person. Say a little bit about the frictions that you investigated. That could be areas of bias that are big topics today, whether that be gender bias or racial bias or something else, what factor did these sorts of frictions or biases play in talent achieving its highest possible

Starting point is 00:54:45 success rate? The most obvious is immigration law, but the one that maybe you can change that is still way under discussed is it is much, much harder for women, especially younger women, to develop mentors in today's world. There's a number of ways you can put it with different valances, depending on, say, how feminist you are. But one is the risk the woman may be harassed. Another is the risk that the man will feel vulnerable because the woman can charge he was harassing her or just create some kind of negative impression. But no matter how you want to frame it, I'm convinced that is a major, major factor holding back, especially younger women, that we're not doing a very good job of addressing. Arguably, in some ways we're making it worse.

Starting point is 00:55:26 by making the issue more fraught. You do definitely want to protect women against harassment, but I don't feel right now we have very good institutions for men mentoring women. Anything you'd highlight, Daniel, in that area? I think an interesting open question is whether Zoom makes things better or worse. I've heard both sides of this. Tyler, actually, what do you think in terms of, say, mentor development for women, the reality of remote work. Is that a plus or a minus? I hear from a large number that it's a plus. I'm not sure I know enough to have my own judgment, but I take those reports seriously from women. And that in a broader Zoom conversation,

Starting point is 00:56:04 they feel they're cut off less and that it's more egalitarian. I hear that a lot too. Yeah. In a recent conversation with Antonio Gracias, who was the first big institutional backer of Elon's company, SpaceX, and Tesla, we were talking about the sorts of talent, the supply of which are a rate limiter on progress in the world. It's tightened a little bit here, but there's abundant capital, financial capital. That doesn't seem to be a major rate limiter in innovation or great new companies or institutions being formed. So talent seems to be the rate limiter.

Starting point is 00:56:34 And I was interested to hear Antonio say, look, it's actually probably not that we need a hundred more Elon's, that we need the layer of talent beneath that enable the rare genius like an Elon to do more and create more leverage for that type of person. I was surprised by that answer, given that he was Elon's big backer and I think would agree with us all that he's a remarkable entrepreneur. What do you guys think is the most important supply-constrained type of talent today that if you could just snap it into existence would make the world a much better place? On the Elon point, I think it's an interesting question. There's a bit of a reflexive loop there where it's someone with the intellectual gravitas and then subsequent market

Starting point is 00:57:10 success like Elon that then attracts the executive bench underneath him. So you kind of have this chicken and egg dynamic where I think a lot of what's going on for Elon today is he can attract talent, no one else can attract. It's part of the star power moat that he has, along with some other regulatory benefits. I don't know that that would actually be the shortage. I think there's some basic base rate of how many humans there are in the world that we were creating every year that are both very talented in the world of engineering with numbers and shapes and code and with other people. For reasons that are pretty deep that I don't really understand, we're pretty decent at creating outliers in either direction. It's very rare to find someone that has both. That's actually a

Starting point is 00:57:52 real constriction. Not only do we not create enough of them, but we also don't slot most of them, I think, into the right roles. There's a very big inefficiency that will probably only realize once we have it fully figured out. And I don't know if this is going to be some type of whimsical techno punk future where the AI is routing jobs. I'm certainly not signing up for that first. That'd be turned into a professional garbage can collector probably. But I think there's a lot of people that are not sorted into the right jobs. I think that's a raison de etre for the book. At the end of the day, it is not a capital problem, but there's a significantly outdated operating system running on the minds of the different people around the world who are responsible for finding the right person for the right

Starting point is 00:58:30 job. I think if we could route people more appropriately, we would get more SpaceX's and more Teslas and more apples than Googles. The free market has this awesome process of grade of destruction that kind of has that happen occasionally. A different way of saying is, startups should have a much lower failure rate if we were able to route everyone properly to do the right thing. We have high failure rate. People kind of celebrate it because successes are so big, but things can be much better. Many more startups can be successful if you were able to route people to the right jobs. And I think that's true over the whole economy, a middle manager at some Fortune 500 company, as well as a VC picking the next startup. I think there's failures at every level. And it's interesting

Starting point is 00:59:09 for me to look at the pandemic. Some of our public health institutions failed pretty badly. the CDC, the NIH, they failed at every level. They're on a kind of automatic pilot. And we only really saw that when there was a particular discrete emergency. Now, when it comes to talent, while I would say there's an emergency every moment, there's not from one month to the next, some obvious new emergency that pops up. So we don't see what we saw as the failures of, say, the CDC in the world of talent. But imagine the talent allocation process as being like the CDC. And the failures are just always there at multiple levels, and we continue down the path on bureaucratized solutions that are actually good enough for maintaining our current level of GDP. Most years, it doesn't go down. We could do much,

Starting point is 00:59:52 much better. And people who are worried about income inequality as a problem, that is simply the other side of the coin of talent allocation. I would say talent misallocation. What would you redesign, Tyler, about what I'll call the attractors or the filters for talent, whether it's, let's take colleges as an example that are designed to first attract, then evaluate, and then foster talent at different levels. What's most broken about them or said differently? Like, if you were designing one of those institutions, YC is interesting in this regard for founder talent, if you were designing a new talent filter institution, what would it look like relative to the ones that we mostly rely on still today, especially colleges? Well, if you take universities where I work, it's striking to me

Starting point is 01:00:35 how slowly they react. So as we know, right now, there's a crisis in Ukraine. a university should be raising its hands and saying, we're going to open up 10, 50, 100, even just five, even just one slot for a super talented Ukrainian student, pay their way, take care of their visa problems, and bring them in. The number of schools that have actually done that, as opposed to the press releases, which I've read and looked into who's actually doing what, is remarkably small. Schools are not set up to adapt quickly. To make, say, 10% of education at a school online can take schools many years to do.

Starting point is 01:01:08 if they get there ever. That's a way of attracting talent that doesn't want to show up every morning to take classes. In general, flexibility and speed of reaction or poor across most of our larger institutions. If you were designing something as a filter, any ideas there for what might make? I'm always struck by Paul Graham's original, I think it was Russ Roberts that interviewed him back in 2008 or 2009. And he described YC as this assembly line that they were trying to make the throughput of the factory as efficient and fast as possible, as high a throughput as possible. Are there ways that we could do this in a more self-served way? Like I think of the CFA exam and finance is an interesting one of, yeah, if you get through

Starting point is 01:01:46 that thing, you definitely have the ability to work really hard and you're smart enough. It says something about you and it's sort of self-serve in a way. Should we have more of that designations or tests or filters that are more scalable? I'm all for competency testing. But there's so many layers at which the problem has to be fixed, just another simple one from my area universities. Mid-century, it was not so unusual to have American university presidents at top schools in their late 20s.

Starting point is 01:02:12 Now, if you're 60, you might be considered too young. That is crazy. It's only one thing. Fixing that one thing may not change matters. But if you realize there's 30, 40 things like that, and to get on the path of fixing each and every one to make our key institutions more dynamic, faster to change, better at allocating talent. What is the biggest open question for each of you about talent at large?

Starting point is 01:02:35 This is super practical for me selfishly. I'm not saying it's the biggest social problem, but in the context of a brief interview or interaction to assess who has durability, I find that much harder than assessing energy or smarts or even conscientiousness. Durability. They'll stick with it. They'll persist. Very hard for me to get a handle on that. Obviously, older people, you just look at the records, basically going to tell you younger people, there's not a record. How do you know? In physics, we have a pretty good understanding that seems to only change in quotes every 200 years or so about how the universe and the world works. But we have pretty good frameworks now that Plank and Einstein to Newton to some extent put in place.

Starting point is 01:03:18 We don't really have that physics blueprint for assessing someone in an interview. I'd like to have something that's as durable, meaning it's a science that will last at least 100 years. truths are reinvented even in physics every couple hundred years or so. So something to last at least 100 years of just how to give in a set of inputs and say 60 minutes of interaction with a human predict how they would respond to different environments over the course of multiple years. And if we had a greater ability to do that, whatever metric you want, global economy, technological progress, a lot of things would accelerate much faster.

Starting point is 01:03:54 So this particular Torah we don't have yet. And the book is really the beginning of a conversation that hopefully will result in the creation of this science. I think that's the real thing missing from the world. Daniel, do you think that the reason why that's hard is that it's almost like a priced marketplace for talent, that it is more of a relative game than an absolute one? And therefore, that science can never exist because it'll always be changing because it's so competitive versus a ball dropping that can be measured. It's going to be the same every time. Nature doesn't respond to you at the same speed that other humans do. I think it's a fair counterpoint.

Starting point is 01:04:26 But if we look at markets as an example of this, there are quant funds you would know better than me, that outperform the market consistently, some of them for a very long time. So there are ways, even in reflexive dynamics, to have some sense of real truth. I agree with you that it's harder said than done, because Earth is not changing gravity based on whether you're dropping a ball. But there's still, I think, a way we can probably get there. And it does not have to be adversarial like markets, by the way. No one wants to be sorted or selected for a job that they do not enjoy. And people seem to be at their best when they enjoy what they do. So I think there's a way to make it happen in the win-win dynamic. I'll definitely remember the book for this idea of

Starting point is 01:05:07 energy, durability, and curiosity really is the three things that I personally took away from it as things to screen for, think about test for, ask questions around. Highly recommend everyone check out the book. I think this conversation is critical for everyone listening. Daniel, you know my traditional closing question. I've already asked it of you last time we talked. So I'll ask it of Tyler this time. Tyler, what is the kindest thing that anyone's ever done for you? My mother was an incredible and loving mother. And I think that is still underrated in our society. And she drove me around to chess tournaments when I was 10 years old and couldn't get around by myself. That was formative for my later development. And she never asked what was in it for her.

Starting point is 01:05:43 It's got to be mom. Love it. Simple, wonderful answer. Guys, thank you so much for your time today. Thanks for writing the book. I hope there's a sequel. Really appreciate it. Thank you. Thank you for having us. Next, you'll hear my conversation with Canalist customer Giuseppe Coco of LK advisors. We talk about the good, the bad, and the ugly around proprietary models and how Giuseppe has made Canales a key component of his investment process. So, Giuseppe, I think the place to start is with the concept of a deep economic model on a business.

Starting point is 01:06:14 You've got a unique background in banking where I think you spent, God knows how many hours building complex models. And I'd love to just begin there. Talk us through your early experience building models, sort of the good, the bad and the ugly. Yeah, so, you know, we started out with investment banking, which is very much on the private side. And there, obviously, you have a lot more information. And so you can go in a lot more detail. So you would look at the models that we were building for deals were frequently 20, 30, 40, 50 taps, thousands of lines long, only like to get to a very simple output.

Starting point is 01:06:46 And, you know, you would spend hours just changing this, changing this, updating this. It would literally take forever and it was very difficult, almost like to audit. You would find something, okay, you know, this number should be this, this number should be that, right? And you would literally go back and spend hours and hours and nights just trying to reconcile that. Just because most of the times people are just adding more and more complexity to those models and always ask for incremental complexity. What do you think is the most useful and the least useful part of how those complex models are built on the banking side? Obviously, precision is good if you can get to it, but false precision is bad.

Starting point is 01:07:24 What do you think the good and the bad is of that style of model building that's so complicated? I think to a lot of people that provides false comfort because it's more like the more of the merrier, but it's actually not the case. It's more sort of, you know, what are the relevant things? What are the key things that actually make a difference? And frequently, that unfortunately just gets lost in the detail. On the good side, to be frank, I don't think there is actually much because we think of a solution like Canales, which the first time I opened a Canales model, I was amazed by

Starting point is 01:07:54 the level of detail and precision that they could get basically into their one-tap models. I was totally amazed by that, that it was even possible. Until that point, I mean, that hasn't even crossed my mind that it was really possible to build such a detailed and sophisticated yet simple model in a manner that they do. If you think about those early days and what Canalist does or when you first encountered it, What did you like about the service when you first encountered it? Like, what did it replace for you? And because you didn't no longer have to do those things,

Starting point is 01:08:25 what did it open up or unlock for you with your time? When I first started on the by side, we started out by the sort of models manually. My former boss asked me, you know, to build out like models manually into this and that. I mean, obviously, like your work basically piles up. And I mean, it just takes hours. It can easily take a few hours until, you know, if we don't want to, potentially even like three weeks, depending on the degree of complexity,

Starting point is 01:08:45 to build a proper and running a fully integrated. model for any of the companies. What Canales does is basically condense all of that process. So it's as simple as downloading any PDF file just from the internet. And you have the whole model there with all the relevant KPIs, with all the relevant drivers, so you can overlay basically your inputs. I think from all the tools I have been using on the BISD and I'm using today, it is the one that reduces friction the most. Giuseppe, I'm curious, where did you first hear of Canales? Funny enough, I actually heard about Canales on your podcast in an ad. And, you know, it was one of those evenings. It was at home listening to a podcast.

Starting point is 01:09:23 And like, you know, I heard automated models out updating. I was like, oh, my God, this is exactly what I need. And I'm curious if you've interacted with others in the investing industry, too, that are using it more and more? Like, are you seeing more colleagues or even competitors or friends using it? Is that part of the growing network of it? Here in the UK, my previous firm, I started using it, and then, you know, a team that was sitting like next to it was like, okay, hey, what are you place doing, you know, how are you doing this so fast? And then they started using it as well. So it became sort of viral. And then when I joined here, so RCIO, funny enough, you know, when we first met, we talked about it. It's like, you know, hey, there's this amazing solution which I'm using as part of

Starting point is 01:10:02 my process. He was like, oh, yeah, he's ex-fidelity. One of the catalyst founders is also ex-fidelity. So he had it very much on the radar. And, you know, it wasn't even a discussion to get up and running when coming here. Maybe just talk about your day-to-day life at LK advisors. What exactly is it that you are doing, what is the daily workflow so that we understand how it slots in? Depends on the time of the year. Currently, you know, we're going into earnings season. So what we're doing right now is lining up our numbers across the models for the companies that we're holding, seeing where our estimates are.

Starting point is 01:10:32 And then obviously, there's just like preparation work at the moment. The rest of the time is screening for real ideas, speaking to management teams, attending conferences, setting up calls. And for all of this, Canada, it's extremely helpful because, you know, you always have a single source of truth, which we can refer to look at the numbers and to get a better sense for where that is and how, you know, something that a management team may say, something that we learn may impact our estimates and where and how they could potentially translate into true value.

Starting point is 01:11:02 That single source of truth thing is interesting. How historically in firms like yours or in your experience, knowing other analysts and PMs, how is ownership of the model typically handled? Because it seems like one nice thing like you said about catalyst is it's a single source of truth. It's almost its own ownership. You don't have to worry about it as much. But how, in the absence of something like Canalist, are models typically shared and responsibility for them shared between teammates? Maybe like even going back to the previous experience, I think generally in finance and I think most people will agree that models are sort of view, the model on a company

Starting point is 01:11:34 on the idea, whatever it may be, is sort of viewed as the holy gray, the numbers that people use based their estimates on of value. And it's sort of like the most things, sort of, you know, what is the impact of fill in the blank get X, Y, Z. So people hold it a very, very high regard, and people are very, I want to say, almost jealous of their model. And everybody thinks that if you own the model, you own the process and you ultimately like have to do. But the model also is, it's usually in pre-canalist type of times,

Starting point is 01:12:04 it is extremely time-consuming and inefficient to maintain. The way it's normally like shared among sort of like teammates is usually it's quite easy for mistakes to sort of sneak in. Canalyst is great because there are no mistakes in their models. If you want to have something added, right, you can just read out to the support team and product analysts and they will mend it to your satisfaction. So thereby using Canales, you don't need to worry about maintaining your single source of true. How would you compare how you use Canalyst from your sort of hedge fund days to what you're doing at LK advisors? Is it different? Is it similar? Is it highlighted anything for you about the product or products?

Starting point is 01:12:45 It's a bit different. I think in my previous role, the coverage universe was a bit more fixed, a bit more Europe focused. So it was more about updating, maintaining, forming a rolling view. I think in today's role, it's very different because our coverage and our universe is basically global. So when I came in, I had to think of, okay, so how can we actually like leverage this? And one of the thinking was, for instance, I was very keen to build a what I would call a quality scorecard, which would allow me basically to, when you have to think about across developed markets, what is what most of what we do, potentially even like some emerging markets. How do you compare, cross-compare companies on a qualitative basis? So we started building out this process, which looks at more than 250 KPIs to help us build sort of a scorecard, which has to score any company along those cases. APIs from one to 10. And this is a process that we found very well working for us. And that without

Starting point is 01:13:46 Cannibalist, I mean, it would have been virtually impossible. Taking years or something. Yeah, it would have taken multiple years, multiple years. What do you think is interesting about where you sit? You know, you're in London. Obviously, a global coverage in universe is probably a little bit more important to you sitting there than if you sat in New York or something. How does that transfer into the use of Canales and the global nature of what you do. Canales over time, since I started first using the product, they have expanded massively, you know, and wider into, especially like European companies as well as EM and developed Asia's companies. So the universe has expanded tremendously. The other great thing is, you know,

Starting point is 01:14:25 we work closely with their product team to make suggestions on sort of, you know, companies that we care about and companies that we know, sort of, you know, people here in Europe care about. And they are extremely reactive to initiating and launching on new models when we ask them to that gets put on sort of like a wait list. So yeah, we continue doing that as we, you know, take an interest in different companies here in Europe. And I think the roadmap is sort of, you know, to get to like sort of 10,000 companies slash models, which is a pretty wide scope. What do you still do that's, I'll call it very manual that you don't think is too high value and you wish could be automated? Another way of asking is like, what do you hope is on candleless products? roadmap. I think it would be nice to have something what I would call a by-site consensus.

Starting point is 01:15:09 If you ask many people in the industry today, by-side consensus is this very elusive concept of whisper, what some people may even call, right, second, sort of unformed expectation, and it may vary. What would be amazing would be to have some sort of panelist user-weighted, anonymized average of what actually the users on the other side thinking and then, you know, sort of providing an opt-in or an opt-out whether you kind of think you want to participate in that. I think that would be amazing. The other thing is they are currently working on this KANIS platform and we have like an internal developer who's working with their team to scale this scoring mechanism that I have just mentioned to you through a Python-enabled web platform

Starting point is 01:15:55 to basically run that even at larger scale through the entirety of their platform. And as that becomes basically more life and more consumer-friendly as their website, I think it could open up very exciting opportunities and use cases down the line. I'm curious, Giuseppe, if there's anything that you think is lost in the process of outsourcing some of this model updating, another way of asking it would be, you know, if you're updating these things manually, does that give you some sort of felt sense for the business that you can't get just by looking at the numbers. And do you think that's worth that at all?

Starting point is 01:16:29 I mean, obviously you're a big catalyst user, so I can guess your answer. But I'm just curious whether there is a downside to, I'll call it, outsourcing some of this manual work around updating the models. I think the first part is that once, you know, when I remember when I open my first candidate, you see all these things and it's more like, okay, how does this work? Right? You look like an introduction. It's like, hmm, you know, I like it, do I trust it? And I think it's more when you have your sort of, you know, the companies that you know and you follow them and you have a sense,

Starting point is 01:16:54 for the history. Obviously, you know, you need to look at the numbers and you just anecdotally get a feel for what it is. But I think the beauty of catalysts is going to mention, right? So you open a cannabis model, there are five tabs and they have these like beautiful summary sheets. And I almost find it a lot easier to just look at those trends and get a sense for how something has performed, what is driving X, what is driving Y. They actually enhance, in my view, that process of understanding what is going on. I had this debate with multiple friends and my view is that it's totally overrated to say, sort of, you know, you need to build the model to entirely understand the business. I think you just need to like look at the numbers,

Starting point is 01:17:37 understand and how they flow, which is, you know, what Canada helps you with and do. I think the other thing that I found super helpful that initially wasn't as intuitive is their custom templates. So Canada's has like standard templates for, an LBO or a DCF, coms, all these usual things. We have our sort of proprietary process of how we look at things, how we value things, the score cut that I've mentioned to you. So we spent, we invested a decent amount of time into building our own templates that correspond to our process that work exclusively on the kind of platform.

Starting point is 01:18:11 Once we scale, we put in that, you know, it incrementally helps us understand and make sense of a business and wise we can, you know, continue to comply with, you know, how we do things and how we think about things. Awesome. Well, Giuseppe, thanks so much for taking the time to do this today. Really interesting career arc that you've obviously done a lot of modeling, so a great set of experience to understand why this is valuable. Thanks so much for your time. Thank you. If you enjoy this episode, check out join colossus.com. There you'll find every episode of this podcast complete with transcripts, show notes, and resources to keep learning. You can also sign up for our

Starting point is 01:18:45 newsletter, Colossus Weekly, where we condense episodes to the big ideas, quotations, and more, as well as share the best content we find on the internet every week.

Invest Like the Best with Patrick O'Shaughnessy - Tyler Cowen & Daniel Gross - Identifying Talent - [Invest Like the Best, EP. 277]

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.