Top Economist on Academia's Flawed System & How AI Will Transform It | Tyler Cowen

There's plenty of truth, mistruth, and speculation about tariffs. Now, from the perspective of an economist in the United States— by the way, I'm from Toronto, so this affects both of us— what do you make of the proposed tariffs?

Well, Trump has inconsistent words on the matter of tariffs. He says he loves tariffs, and if you listen to him going back to the 1980s, he's quite keen to have higher tariffs. So at first, as you know, he threatened tariffs on Canada and Mexico, and then he pulled back the threat. As we're recording today, this morning in the middle of February 2025, he threatened reciprocal tariffs on all countries putting tariffs on American goods. Virtually all economists, myself included, think that tariffs are a mistake. They do not help the distribution of wealth. They make your economy less efficient. They make prices higher. They hurt innovation. And consumers have somewhat lower standards of living. We would all make exceptions for what are called cases of national security. So if a nation needs to build its own ships or drones, if they would have some kind of protectionism to make that possible with domestic industry, at least in principle, there's an argument for that. But otherwise, typically economists believe in zero tariffs, and so do I. So I'm against all of those Trump trade policies.

So the Wall Street Journal said that it also has, or will have, job creation inefficiency. So for instance, the machine washing tariffs cost something like $815,000 per job created, something like that. And then there'll be a domino price effect where even untariffed goods will increase in price. So do you broadly agree?

Very much agree. Keep in mind, we're putting tariffs on intermediate goods that that are inputs into our own production. So to that extent, it's a tax on US producers, and it's just bad for US job creation, US wages, US business.

Well, what will the consequences if these tariffs do get implemented? Now, there are a variety of proposed tariffs, so there are a variety of potential futures, but make some prediction over the near term and then the next 4 years.

Well, there's so many proposals floating around early in the Trump term. I do think if the United States puts tariffs on goods from Canada, most of the losses are borne by Canadians, not Americans. So to me, that doesn't make it any less bad. But in many of these situations, the larger country in essence has more bargaining power, and it's the prices in the smaller country that adjust more. It's still a bad idea, and the United States gains nothing from it. And that's not even getting into the questions of wanting to be nice to one of your very best allies and neighbors. In terms of specific predictions, I don't think it's the end of the world, but you have a stock market that's somewhat lower in the short run, an inflation rate that's somewhat higher. Real wages go down by some fraction of a percentage point, but there's just some costs and no real gain is the best way to put it.

What are the costs and what are the gains of Canada becoming a 51st state?

Well, that's a huge question. It seems to me that there's more than one state in our neighbor to the north, but I don't think Canadians at all want it, and I don't think Americans want it. So why do it? I think Trump has a perfectly legitimate point when he says Canada is not meeting the NATO-agreed target of spending a certain percentage of GDP on defense. And Canada should change that. I agree with that one point. I don't really agree with any other part of it. Most Americans don't want Canadian voters, and most Canadians don't want American voters. So let's be separate countries, compete with each other in a friendly way, cooperate. That's how it's been for a long, long time. And I see no good reason to change that.

How serious do you take those statements? Of Canada becoming another part of the United States?

I don't know. I've never met Trump. So I think one possibility, I'm not saying it's true, but it's a hypothesis, is that he's trying to scare Canada into doing more to spend on its own defense. And the only way to scare Canadians is just to be totally hostile to them. And furthermore, he's expecting Canada to take a larger share of the burden of defending the Arctic And he's decided that just being nice to Canada isn't enough. So you just have to make them think the US wants to conquer them and won't in any way help them. I would be surprised if that were actually good tactics. I'll just say it's a hypothesis. I don't know if it's true or not.

Now let's suppose it is true. Let's suppose you're correct. What would the consequences of it just being a, a tactic be?

Well, one possibility is that Canada, Canadian public opinion turns more against the United States and it's harder for any Canadian leader to get Canada to do what the US has been telling it to do. I'm no expert on Canada, but my intuition suggests that's very often what happens between Canada and the United States. An alternative equilibrium, which again is possible, though seems to me less likely, Does Canada actually get scared and militarizes the economy somewhat more and takes an aggressive stance in the Arctic and does what Trump is requesting? So either is possible, but I think backlash effects, especially when it's like bigger, smaller nation and their neighbors, backlash effects are pretty often quite strong.

Now, all of these are deleterious effects. Is there anything positive that people are missing about this issue?

Well, I think around the world, people are observing the United States being nasty to Canada, its neighbor, close friend, and ally. Uh, our other allies observe this and probably they infer they're more on their own than they used to think. So some of them will spend more on their own defense, which I view generally as a positive. Some of them will simply view the United States as less reliable. Which I view as a negative. So there's some upside to that. But in general, the US is the country, like, good at cooperating and not that good at threatening. So to all of a sudden decide our comparative advantage is in threatening, I would be surprised if that worked out well.

Now, people should know this is a podcast where I use my background in mathematical physics and interview people on various theories of reality called Theories of Everything. And the question is, well, why am I interviewing an economist all of a sudden when I ordinarily speak to people in theoretical physics and math, computer science, consciousness studies, and philosophers? Well, Tyler, you've mentioned that you're a philosopher who happens to study the economy rather than an economist.

That was half tongue in cheek, but yes, I meant it also. My PhD is in economics, just to be clear.

Explain what you mean.

I would be considered an economic generalist. So I've blogged every day for almost 22 years. I wrote for the New York Times for 10 years. I've written for Bloomberg Opinion for almost 8 years. So if you write on such a large number of topics, which is what that in practice means, you just think more and more about the whole world and what is really going on, and also how non-economic forces shape people's behavior. And you're forced to think and rethink matters at a very fundamental level every day of what you're doing, basically for your entire life. And that describes myself. I've also traveled a good deal to over 100 countries. I've been to Canada many, many times, even to some of the less well-known parts. And it makes one somewhat philosophical.

Has tenure made you more or less risk-averse?

I received tenure in the second year of my first job, so I barely know life without tenure. Uh, I think I've been quite risk-loving as tenured professors go, but I'm not sure that stems from having tenure.

Right. And generally speaking, does tenure make people more risk-averse?

When I look at the data, there are some papers on this. It's surprising to me how small an effect there is after tenure. The people's behavior doesn't seem to change much. Also, their productivity doesn't seem to go down, at least in economics. My hypothesis is that what's going on with tenure is a strong selection effect. So you select for people who are risk-averse to begin with, and then you train them for, depends how many years, but at least 10 years to follow the dictates of a relatively small circle of peers that makes them all the more risk-averse. And the actual granting of tenure, I don't think is that relevant in the risk aversion equation.

Okay.

The subjects here, they're so risk-averse to begin with, it's not even easy to make them more risk-averse.

So Freeman Dyson said about the Institute for Advanced Study that you would think getting a tenure or a tenure job there and then not having to teach and just focusing on your research would allow you to be more creative, innovative. But he found that people who, if you look at their productivity after they get accepted, that it's their innovation or their, their, like some creative index after they get accepted, that it's, it's It's less than what they had to do to get into the IAS.

I haven't read that paper, but some of that could be regression to the mean, right? If you have to be super innovative to get in and then you're just hit by further random shocks, maybe you just go back to what you were on average to begin with.

Wouldn't it also be that people are more creative when they're younger, or is that a myth?

Uh, there's a lot of data on that. It depends on the field. In fields like yours, it's very strongly the case that people are most creative when they're younger. For economics, I think people are most creative in their sort of 30 to 45, which is young, but not like being a brilliant 22-year-old. For literature and philosophy, it's a much more complicated story. Someone like John Rawls, who's really not so young when he publishes Theory of Justice. So it depends on the field. I think the more directly G-loaded something is, like math, astrophysics, the more likely it's to be the province of the young. Chess players also, you see this. Gukesh is world champion and he's 18. I think he's 19 now, but he was 18 when he won.

Do you think it's also, it's not just being young, but it's the, it's your entry into a field? So for instance, I believe Schrödinger was something like 40 or 50 when he started making contributions to biology. And some would consider that innovative, but he was older. However, his entry into the field could be what allowed the innovation.

Uh, that's an interesting hypothesis. It could be you can carry ideas from your first field into the new field. Uh, it's worth asking how many people are there with major innovations in two quite different fields. I mean, Linus Pauling has two Nobel Prizes, but I would think those are the exceptions and not the regularity.

Now I want to know from your point of view, what are the nourishing aspects of the academy? So the university system, and then we're going to get to the problems. Let me make that clear. What does the academy do right that is difficult to do outside the academy?

If you choose to exercise it, you do have true freedom of speech in the United States, not everywhere. I think Canada is somewhat different. You can spend much of your day hanging around with other smart people. Having to teach forces you to rethink things every year, every semester in a very useful way. Just having to explain it to people. Some are super smart, some really don't care. Your teaching experience should be diverse. Having that experience all the time, I think, is very powerful as an engine for improving your own thought. Uh, Just having the opportunity to mentor other individuals or maybe be mentored by them is a strong positive. You wouldn't say the pay is great relative to people's human capital, but it's certainly something you can live on in most parts of the United States. So that's a bit mixed, but I would overall say it's a positive. And there's the opportunity to earn outside income. The positives are quite significant.

So what are the problems?

The incentives are to be more and more specialized, to be too narrow, to court the approval of a relatively small number of peers. Rewards and productivity, they're correlated across some margins, but at many other margins, not. The data on current graduate students indicate they have pretty serious mental health problems to really a shockingly high degree. Like as many as a third of them in some studies. University bureaucracies are getting worse, really much worse at a fairly rapid rate that shows no sign of slowing down. You spend more time in committee work, more time doing referee reports, more time writing non-useful letters, more time just dealing with process in a way that is stultifying and I think morale-sapping. I think morale in the US university system has been fairly low., and it's lost a lot of prestige and status. Some of that deserved it. So the negatives are pretty real too.

What about the grant system? Now, I understand that's a broad question, and there are various fields which may have different systems, quote unquote, so feel free to delineate.

It really depends on your field. So a lot of what economists do, uh, does not in a major way require significant grants. So we are somewhat isolated from it. But some economists say you do randomized control trials, then you need to pull in some fairly large grants. It takes quite a long time to apply for the grant, do reporting for the grant, cultivate the people who might be funding you, and that's all getting more bureaucratized. So that's another big problem. But I would, as a side observation, just say quite a few economists don't face it so much. People in biomedical and a number of other fields would face it much more.

So explain to people who are not in the university system currently, or maybe they are, but they're students or graduate students, they're not employed by the university. Why do professors need grants when they're already paid by the university? What is it about grants that are required and why is there a pressure to attain them? And what's the downside of trying to attain them?

Some studies are quite large. You need to assemble a lot of data. There's one study, I visited it once. They had a whole operation set up in India with a reasonable number of people helping out, even at India-level wages. There were people to collect data, there were people to take a control group and then a group where you vary their treatment and then measure results of giving one group microcredit, the other group no microcredit years later and keep track of those people, make sure the data are not corrupted. The costs of that are pretty high and you can't come close to paying for that with the salary of a professor, even 10x the salary, you would not be able to pay for that happening. So you need to raise grants if that is what you do. The cost of the grant system, first it just takes a lot of time to raise the grants, but it also skews your research over time. You end up doing the kinds of work that it's easier to raise the grant money for. That doesn't have to be bad, but I think on average, I think the judgment of researchers is better than the judgment of the grant system. So that would be a significant cost.

There's a trope that people say, and they think they're being, they think they're going against the grain when they say this, even though everyone says it, the system is broken. And it's clearly a nonspecific statement. Now, would you say the academic system is broken? Would you go that far?

Well, as an economist, I have a theory of everything, and part of that theory of everything says it's all comparative. Broken compared to what? I think the scientific community is either at or close to its all-time peak in terms of quality. Still, there's plenty wrong with it, some of which we've talked about already. And that's a better way to view the matter than, oh, it is broken, it isn't broken. To me, those words don't mean very much.

So if you were to improve it, how would you improve it? Like, what's a, what's a straightforward manner? The lowest hanging fruit.

Well, here's a simple example. We have something called the National Institutes of Health that supports biomedical research. And when a COVID pandemic came along, The NIH, for a long time, they were not able to speed up their procedures, and people wanting to do research on COVID might have to wait 6 months or more to get an answer and get the money. Now, when a pandemic is killing people every day, I think that's just crazy. I mean, that particular thing you could say was broken, and that would be very easy to fix, but they just kept the same old bureaucratic procedures and how an application had to look and had to be done. And they should have swept that all away overnight and just set up a very fast system where in essence they give people a decision and money, say, in 2 weeks. That's just a very vivid, obvious example. But I think more insidiously, a lot of the costs are invisible. And there have been some studies of the arms of the NIH that was designed to give out, you know, more, more innovative, risk-loving grants., and that branch of the NIH was in fact unable to do that by any metric people could come up with. So that would be a case where the system is so sclerotic, we can't get it to take more risk, even when we deliberately want to choose to do so.

You said a lot of the costs are hidden.

I don't think many people see that cost. And the whole research programs of academics, they evolve to fit into the system. So a lot of people end up not so unhappy. They've learned how to play the system. They get money from the grants. They're the Darwinian survivors, you could say. They don't have a huge incentive to complain too much about the system. They might wish their paperwork burden were lower, but they're the winners. So the whole thing is populated by the winners, and there's a dangerous selection of opinions that goes on when that's the case.

What do you mean by paperwork burden? Like these administrative tasks? Mm-hmm.

Uh, there was a survey Patrick Collison and I did, this is now about 4 years ago, where we asked principal investigators, uh, how much of your time do you spend having to do grants? Not do the research, just try to get the grant. And it was somewhat over a third, if I recall. Uh, it seems to me that's far too much of very valuable time. Uh, the system should be easier and simpler. A grant application can be hundreds or even thousands of pages. Uh, I, I simply do not believe that improves the quality of the final decisions. So it all ought to be changed.

There's new AI agents coming out each day, or at least every week, and some can apply for grants for you. Do you see this helping? As you know, on Theories of Everything, we delve into some of the most reality-spiraling concepts, from theoretical physics and consciousness to AI and emerging technologies. To stay informed in an ever-evolving landscape, I see The Economist as a wellspring of insightful analysis and in-depth reporting on the various topics we explore here and beyond. The Economist's commitment to rigorous journalism means you get a clear picture of the world's most significant developments Whether it's in scientific innovation or the shifting tectonic plates of global politics, The Economist provides comprehensive coverage that goes beyond the headlines. What sets The Economist apart is their ability to make complex issues accessible and engaging, much like we strive to do in this podcast. If you're passionate about expanding your knowledge and gaining a deeper understanding of the forces that shape our world, then I highly recommend subscribing to The Economist. It's an investment into intellectual growth. One that you won't regret. As a listener of TOE, you get a special 20% off discount. Now you can enjoy The Economist and all it has to offer for less. Head over to their website, www.economist.com/toe, T-O-E, to get started. Thanks for tuning in, and now back to our explorations of the mysteries of the universe. And some can apply for grants for you. Do you see this helping?

I don't know. I think we're going to have many more applications. And on the judging side, people will not be willing to use AI in a sufficiently clever manner. So my best guess is at first it will overload the system more. Across some time horizon, I'm not sure when, but I would think in less than 5 years, the AIs can referee better than the humans can. How long will it take our bureaucracies to make that switch? That could be a very long time, but that at some point will be the answer, is to have AIs do the refereeing and have them do it quickly. But it's not a thing that can happen right away.

When you say that you're not sure how long it will take, quote unquote, our bureaucracies, how long it'll take them to make this switch, you're referring to the grant agencies?

Ah, yes, but you could talk about academic journals, universities, many other institutions do forms of refereeing and judgment, and I observe no change in the process. Now you could say AI isn't quite good enough to referee. I'm not sure. I think L1 Pro referees quite well. I just did a referee report and I sent in my report and I sent in the L1 report. They didn't use the L1 report. Which I find interesting. And the O1 report was not worse than mine. It was different. But I think all papers should be refereed, at least in part, with O1 or something like it. O1 Pro, that is.

Wait, did they know that you sent in an O1 report or did you just—

Of course, I was fully transparent. I said, here's my report. And then I said, here's the O1 Pro report. And the author who didn't know I was the referee, he actually wrote a blog post about his experience. Relating it to AI, and he didn't mention getting a referee report from O1 Pro, so I'm pretty sure they never passed it along to him. That's a very small thing, but to me it's just a sign of the system being far too inertial and unwilling to countenance that kind of change. It was a free lunch, right? They got a human report.

Why don't you test that by not telling them that this is an O1 report, and then later you can tell them?

I think that would be dishonest on my part. Uh, it would be a violation of some kind of professional ethics. You know, you could try that as a systematic experiment, see if you could get IRB approval. I hope someone does that. Uh, I think the receivers would know, but they would know because the O1 report in some ways would be better. So I don't think you could trick people, but I think the O1 report would be fine.

Well, what about using the O1 report and then say adding 20% to it or changing 20%? So then it's an amalgam.

The O1 report is pretty long. It tends to state a bunch of obvious things that are fully correct, but a human would not bother to write down, at least as of this moment. That's what I call the dead giveaway. And the phrasing is quite uniform and perfect in a way that a human would not usually produce. You could change all that. It's a fair amount of work. Again, someone should do this and just see what happens. I think we'll, we'll learn what I feel I already know, which is that these things referee pretty well right now. And when O3 comes out, which is very soon, sometime in 2025, that's going to be much better. And you know, there's another Claude coming out soon. So as you well know, progress is rapid here.

O1 report, do you mean Deep Research or something else?

No, O1 Pro. So Deep Research is part of O1 Pro. Deep Research is amazing, but I don't think it's best for referee reports. It's best for generating 10-page summary papers and answer to questions you ask it. You can't upload a PDF into it and ask it, tell me what you think. Now, OpenAI tells us that's on the way. I believe them., but right now it's not there.

The Deep Research actually uses O3.

That's right. But it's not pure O3 in the sense that you can just ask it questions the way you can ask O1 Pro. But we're told that's coming very soon. Again, February 2025. Sam said weeks or months, a literal quote from Twitter. Mm-hmm.

Okay. So as of February 14th, Valentine's Day, 10 days ago, you wrote the following, you said, about Deep Research. You said, I've had it write a number of 10 pages for me, 10-page papers, and each of them were outstanding. I think of the quality as comparable to having a good PhD-level research assistant and sending that person away for a week or two.

That's right.

Okay. I'm going to link that, your blog and also your podcast, just on screen and in the description for people to, to take a look at. Can you tell me more about your thoughts on deep research?

Well, it's interesting to look at the, the comparative variation. So it's much better in some fields than others. If I ask it about history of economic thought, where it's read everything, it's all open source, it's all online, it's just superb. But I tried it on something this morning. I had a question about the literature on the returns to mentoring, like, does it improve people's career prospects? And I thought the answer was not superb. I thought it was as good as what humans have done. I just think that's somewhat mediocre. And Deep Research reflected that. So there, there are problems with it in, in some critical ways. It still does not go beyond what it's trained on. Uh, people exaggerate that problem. Like I said, the humans are not doing better. But the areas I happen to work in, I would say, are areas where O1 Pro and Deep Research are especially strong. It's very strong in economics.

There was an economist, sorry, The Economist report or article about this, about Deep Research saying that it wrongly estimates the average amount of money that some American household, let's say 25 to 35 or to 34, whatever the bracket is, spends on whiskey in some year, even though you can look that up, that exact number up. And that also there's this tyranny of the majority where if you ask it, is income inequality rising? Most people think yes, and it tends to say yes, even though the experts think, well, income inequality is either staying level or just, or increasing a small amount.

Well, anything in the media about AI, you want to try to replicate. So I see claims all the time, what it can do, what it can't do, especially what it can't do. And then I find it can do it for me. Or there was another claim recently in an actual research paper, you know, which says like the life of a Pakistani is 10 times more valuable than the life of an American. I can't get it to say that. And I put in the same prompt. So I'm not saying that column is wrong, but some of it may have been a problem with prompting. Uh, but it, it is correct that it's still imperfect and I wouldn't use them for mathematics. It's not what they're there for. There's other tools for math that do great. You know them better than I do, but that's not what Deep Research is for. It's for deep research and generating something like an annotated high-quality literature survey on a topic you need to learn quickly. And again, for that, it's fantastic.

You're known for being a prolific writer. So we've been talking about research and, well, I, it's, it's difficult to call what deep research does research because research is a broad term. Research can mean you just investigate and, and do an overview of a subject or get an overview of a subject, but it can also mean producing new insights in some field. Like a professor does research. They're not just in a library, just reading. So I'm going to ask you a question about that, but first, is writing thinking?

I find that when I write, it forces me to think through the idea. So for me, writing would be thinking, but I would not want to issue that as a blanket statement.

What's the difference between the writing that you do that is thinking And then writing that is non-thinking. So obviously there are various forms of writing, like texting someone on WhatsApp. Few people would call that the type that elicits new ideas, or, or it would be more banal than beneficial.

Again, there's a lot of floating concepts in the background here. The people texting, uh, I strongly suspect that they're mainly thinking. So sometimes I'll do something like when I type in my credit card number, That's a form of writing, but I couldn't even tell you the number I'm typing. It's the muscle memory. So when I type/write that, am I thinking? I don't know. It sounds like a semantics issue. I tend to see writing and thinking as pretty closely bound together.

What role does developing and expanding a worldview play in your thinking?

If you set out to read or listen to all the people who have worldviews, I think you'll learn a great deal. Whether it's Plato or Hegel or Einstein, Shakespeare, those are some of the very best people to study. It forces you to think big. There'll almost invariably be parts of it you disagree with, and you've got to figure out why., it heightens your ambition. It gets you more excited about the world because you see how different pieces may or may not fit together. So it's one of the most wonderful intellectual tonics.

Now, if your worldview is truthful, do you think that all worldviews should be converging on the same truth?

I doubt if that's the case. There's something plural about reality and truth, and it might be irreducibly plural, but from a practical point of view, I suspect it's irreducibly plural. So there's general relativity, there's quantum mechanics, there's the laws of economics, there's Freudian psychology, there's the humanities. And even if they don't contradict each other, I don't know that there's some broader consilience through which they converge. I would be very surprised if that were the case.

I don't know if you've heard of Nancy Cartwright, the philosopher.

Yes.

She has a patchwork theory of reality.

Yes, I've thought about this a great deal. I sometimes wonder if your field, physics, isn't ultimately just a patchwork and that there is no final theory of everything. I don't feel able to judge that on technical grounds, I would just say when I hear people in your fields talk about what they do, Nancy Cartwright sounds more plausible to me. I had dinner, what, 3 nights ago with 2 people who do astrophysics at Harvard. They were obviously super smart from age, seemed to be tenured full professors. They just sounded so unconvincing, you know. So, yeah, I think about Nancy Cartwright frequently.

So the theory of everything in that case doesn't have to be one patch that covers the whole universe. It could still be the quilt. So the way that the ways that the different patches overlap somewhat, or their relation to one another.

And many quilts in different places, right?

Yeah. That's still a theory of everything. If you can, if you have some all-encompassing manner of conceptualizing the world.

Like an economics theory of everything would be something like, we'll start with the idea of gains from trade and ask how people are trying to reap them. And just by asking that question, your analytical machinery will get you further in understanding the world. That's not all of our theory of everything, but that would be one starting point. And that's not going to be made easily commensurable with most other disciplines.

I was working on a documentary a few years ago, and I was auditioning people for the role of an editor. And then one of the editors asked a question that was seemingly unrelated to the documentary. He said, Kurt, actually, if the documentary had to do with what is extremism on each side, like when does political extremism happen on the left and the right, something like that. And then he said, Kurt, I want to know why is it that Intelligent people who have access to the same data, who are well-intentioned, disagree. And then I remember thinking that has nothing to do with the documentary or that it's not even an interesting question. And for once a month or so, I think about that question. And I think that guy, I think that's a deeper question than the guy thought or that I thought at the time.

I wrote a whole paper on that question with Robin Hanson called Are Disagreements Honest? And Robert Aumann won a Nobel Prize in economics in part for his work on that question. It's a very difficult issue, very hard to make progress on, I found. My co-author and I could not even agree on what our own paper meant. We were each satisfied with the draft, but we disagreed as to the implications. And that's a paper on disagreement.

Okay, so I imagine one route one explanation is, well, reality is plural, this patchwork idea of reality, but I think that's getting too close to the metal. So what would be the other reason? What would be other reasons for disagreement among intelligent people with the access to the same data and they're well-intentioned?

Well, I think in practice, people simply overrate their own ability to seek truth, that they self-deceive quite a bit, and they think they're more intellectually honest than the others. And they dismiss or downgrade the opinions of the others for a long list of reasons, most of which translate into the other person is not as smart or as honest as I am. And they overweight their own opinions. It's not the only reason for disagreement, but I think it's the most common.

Do you suffer from that as well?

I think everyone does. So if I meet an equivalently smart economist of whatever similar demographic background, whatever I might demand from the person as a credential, and that person and I disagree, I don't automatically converge to them or split the difference or whatever you think I ought to do. I don't do it. I would say something like, well, I knew people like that already were out there and I've already done my adjustment. I might adjust a little bit extra more. Now there's other parts of life where you adjust completely. So I was in Philadelphia walking around downtown. I asked someone how to get to a particular building. What they told me, I immediately acceded to their superior knowledge of the truth and I followed their directions like I can do it. Yet in other areas, who's the epistemic peer or expert is itself up for grabs. And you want, you might say, well, always defer to your epistemic superior, but whether or not someone's your epistemic superior, you judge that by the content of their views. And you're in this circular trap. You can start talking about fixed point theorems, Bayesian updating. There's a lot of math you can bring to bear on this. But the end of it all, I'm not sure any of that has really solved the problem.

So this is related, I assume, to meta-rationality.

Absolutely.

Can you please explain what metarationality is? And then I have some questions about it.

Well, my notion of metarationality is you're aware of your own limits, so you know when to defer to other people and, uh, you're not too arrogant and you make a lot of very good decisions by listening to others. I wouldn't say that's a completely, it's not a fully complete formal definition of metarationality, but it's a kind of working definition. So someone like Elon Musk, when he runs his companies, uh, there seems to be a lot of evidence he's very metarational. He'll defer to engineers, to people doing various tasks. Uh, when he thinks about certain issues in the public sphere, whether or not one agrees with him to whatever percentage, it seems far less obvious that he's metarational.

So when you're not metarational, do you tend to overestimate or underestimate your ability?

In politics, I almost always observe overestimating, but there might be areas where people underestimate their abilities. I think those are exceptional. Experimentally, I think there've been some papers on this. I just don't remember what those papers found. So maybe I think like I'm not as good a cook as I am, wouldn't possibly You know, there might be some things.

If you were perfectly meta-rational, would you not recognize the limitations of your own rationality and then be paralyzed by uncertainty?

I don't see why you would have to be paralyzed. You would do some kind of broadly Bayesian calculation and then just decide what to do. I think you'd be very effective. Now, you might be less motivated. Whether self-deception motivates us by instilling overconfidence, That's my worry, that high motivation is more or less optimal, but there's this side effect that we all end up not sufficiently meta-rational.

If we were to just take it in the Bayesian direction, then your priors aren't given to you. Now you have, you have rules for updating priors, but your initial conditions are not given. Like you somehow come up with that and it's either given to you by your culture or evolution or what have you. And if you then you see that and you see the arbitrariness of the priors, how can that not paralyze you? Hi everyone. Hope you're enjoying today's episode. If you're hungry for deeper dives into physics, AI, consciousness, philosophy, along with my personal reflections, you'll find it all on my Substack. Subscribers get first access to new episodes, new posts as well, behind-the-scenes insights, and the chance to be a part of a thriving community of like-minded pilgrimrs. By joining, you'll directly be supporting my work and helping keep these conversations at the cutting edge. So click the link on screen here, hit subscribe, and let's keep pushing the boundaries of knowledge together. Thank you and enjoy the show. Just so you know, if you're listening, it's curtjaimungal.org, curtjaimungal.org.

This gets at exactly my disagreement with Robin Hanson on the paper he and I wrote together So he views priors as an actual thing where you can ask literal questions about the prior, like the ones you just ask. I view priors in a kind of Quinean framework. They're not a thing, they're an analytic category. So there's not a fact of the matter as to where your prior comes from. So the questions you try to ask don't have a directly factual answer. Prior's just within a theoretical construct., and you'll arrive at different answers depending on which of those theoretical frameworks you're using.

Have you heard Raymond Smullyan's argument about how a perfectly rational agent will become inconsistent?

I, I don't recognize his name, but I might have heard the argument. What is it?

So Raymond Smullyan is a mathematician, or was a mathematician, and he was a playful mathematician. He was a logician. And he used Loeb's theorem, Loeb's theorem, which broadly says something like, if you can prove that, it's quite hairy to say, if you can prove that if something is provable, then it's true, then that something must already be provable. And then if you assume soundness, then that means that something is already true. And by the way, the actual proof that uses Loeb's theorem is much longer. It's difficult to say with words. It can be conveyed symbolically fairly straightforwardly. So I'll put it on screen, but I'll also say it to you aloud. So consider, let's say, a perfectly rational agent. Let's call them A, and this person can reason about their own beliefs using formal logic. Then by Loeb's theorem, if you can prove again that quote unquote, if something is provable, then it's true, end quote, then that something must already be provable. So the next step is to say that being rational, what does being rational mean? It means if you can prove something, if A can prove something, then A must believe it. And then also, furthermore, since A is rational, A must also have as a belief that if something is provable, then it is true. Then you go to step number 5, which is if they believe something, then it's provable because a perfectly rational agent isn't going to believe something that just is false, is not provable. Then you apply Lobe's theorem to the agent, implying that A can prove any statement, whether it's A, whether it's not A, making them inconsistent. The core idea is that perfect rationality forces you to believe quote unquote everything because there's a sneaky self-referential loop in the formal logic. I'll also put on screen a link to LaVeyre's theorem, which is related to this because I did a write-up on that. You're either arrogant or you're inconsistent. Okay, why is that the case? Because you have to be a rational agent. So one is that let's say you list out all your beliefs. So you have belief one, comma, dot, dot, dot, belief N. Okay, it could be a huge number, but you have all these beliefs. So you're either arrogant in that you believe all of these to be true, or you have some humility and you say, okay, well, one of these is incorrect. But then you all, but then that means that you hold some belief here, K from 1 to N, some K from 1 to N that is true and not true at the same time. So you're inconsistent. You're either arrogant or you're inconsistent.

I'm not sure inconsistent is the right word for that, but I do see the logic. There's imperfection in beliefs no matter what you try to do. I would agree with that.

It would be inconsistent if, like, inconsistent means you believe both A and not A.

I don't know. There's some kind of sense reference distinction in the background since you don't know which one is the wrong view. So I, I would be disinclined to so readily boil it down to that. Just, you know, some of your beliefs are wrong. You're not sure what to do about that. And so your beliefs are screwed up. Oh, but does that make you inconsistent? I don't know. I think it's a funny definition of inconsistency.

How do you cultivate meta-rationality?

The people I know who talk about it a lot, are often pretty arrogant. So maybe cultivating it is the problem. Uh, maybe failing and being aware that you failed and being in an activity where there's a lot of immediate feedback is what's useful. So I know a fair number of people who do things that they're like Wall Street traders, or they were Wall Street traders. Uh, they're the most meta-rational people I've met. When they're wrong, they know it. I think chess players are pretty meta-rational. When you lose a game, you don't have many excuses. You can say bad luck, but basically you made some bad moves, right? So I think repeated practice of something with immediate objective feedback makes you more meta-rational. Trying to cultivate meta-rationality probably makes you less so.

Now, you also have a distinction between stamina and grit, and I would like you to expound on that, please. 'cause they sound the same.

That's in my book on talent with Daniel Gross. You know, there's a literature on grit, which is a kind of doggedness. It's a formal psychological concept. It's not just a word in Webster's Dictionary. I think of stamina as a little different. It's the ability to put more and more on your plate without burning out, without wearing down. And to do it for decades on end. There are plenty of people who like hang in there and stick with the problem and they have grit, and that's a, a good work quality to have usually, but they don't necessarily have stamina in the way that say early Bill Gates had incredible stamina. Elon Musk right now has incredible stamina. I think they're somewhat different concepts.

Now, how do you cultivate the stamina?

Well, I think so many of your traits are at least 60% inheritable. So you, in that sense, to that degree, you cannot cultivate it. But I think to the extent you can, hanging around with people who have stamina, having peers with stamina is the best you can do. They're a role model. There's some mimesis that kicks in and you can learn things from them about how to make this work. So For most traits, if you want more of it, uh, alter your portfolio of friends and acquaintances. This gets back to one of the problems with academia. So many of the people you're hanging around with are risk-averse, and it feeds on itself. It saps the risk-loving nature from a lot of people who could do more and take more chances.

Professors often tell their students who are asking for problems, don't go after the large problems, go after solvable problems. And that it's a fool's errand or beginner's mistake to go after the biggest problems that you can. Those are the most enticing, but those are the ones that are the hardest, and so you can't publish. Is that something that you've also seen? And is that what is instilling this lack of risk-taking?

That's part of it. I think that advice has to be embedded in a larger conversation. So as I think you know, most people do not finish their graduate degrees. In the United States, most people do not finish their undergraduate degrees. So most people drop out of the system for whatever reasons. They get tired, they're fed up, they have better offers, they flunk out. So that's like maybe okay advice, conditional. On knowing that a person will and should stay in. But given that most people will leave, I think a lot of people should engage in what I'd call out-of-equilibrium behavior just to learn more quickly that they should leave. And the advice from these professors doesn't consider that much. It's very arrogant in a way, like, well, of course you belong here with me doing what we're all doing. And if you take that as an absolute, then yeah, it's pretty good practical advice, but I think it's actually misleading people. You're not seeing the bigger picture.

I didn't know that it was half, it was more than half of the people who enter undergrad that leave before they get their degree.

You know, universities are notoriously reluctant to publish these numbers for obvious reasons. Uh, at a typical state university, it's below half. And in the US, state universities are about 80% of the total. If you go to Harvard or MIT, this will sound like it's crazy, but it's not. If you go to community college, you'll say, half? I didn't know it was close to half. Some while ago, this is now a while ago, but the numbers I dug up were in the range of 38 to 40% finishing. That's obsolete. It may be higher, it may be lower. I'm not sure. COVID for a while lowered it. I know that. So it's not what people think.

We'll just use some deep research on it.

Yeah, now there was a while, this is again a while ago, but I think some time ago, maybe a third of people finished their economics PhDs. So you should not be giving advice assuming as an absolute that the person should or will finish.

So then are you saying that, look, given that a large percentage of people are going to leave for whatever reason, anyhow, encourage them to pursue what's risky?

I would at least consider that. I think that the correct advice will depend on the particular case, but even someone who's definitely not going to make it, maybe it's by trying the huge idea that they learn this is all not for them and better to learn that sooner than later.

Hmm.

So I don't necessarily discourage people from doing a lot of things that look like mistakes. I think about like, what's the base rate of success? And I try to do the bigger calculation and I talk them through exactly these numbers. Like, where do you think you're going to end up in all this? And it's a very different sort of advice.

I want to ask you about your interviewing style. And again, your podcast will be on screen and the link will be in the description where you have an idiosyncratic manner of asking questions and you get people to quote unquote, be weird. I, I'll ask you a weird question rather about questions about your answers. So even when I'm asking you a question, like, let me just turn the table. So you're answering. How is it you're thinking of your answers? Like, are they stream of consciousness coming out? Are you thinking of it in terms of a narrative? How is it that you know when to end? Walk me through your answers.

I think it's fairly automatic. So I've done hundreds of podcasts. I've lectured for 40 years of my life, more, given, you know, more talks than I can count, done a large number of ask me anythings. I write every day since I've been actually a teenager. So there's a kind of database that's accessible and I don't, have to deliberate to access it. So people sometimes say, well, I answer questions very quickly. It seems like I haven't thought about it, but I have thought about it. Uh, and, and you hear the answer. So that's how it happens, but I'm not sitting there like thinking, oh, will this be a narrative? Will this be that? It just comes out. It's like a burp almost.

So rather than looking at how quickly you respond to questions as evidence of you not putting thought into the question. It's actually evidence that you've thought about this question for so long, but far prior to this question being asked to you in this moment.

That's right. And most, most questions you would've thought about for 30 or 40 years, maybe longer.

Tell me about your interviewing style.

Well, I do much more preparation than most interviewers do. Uh, maybe not more than you do, but most interviewers start with a question like, tell us about your book, which is a completely lazy question. So I think you should have read the book carefully more than once, read the person's other books, possibly articles depending what they do, uh, listen to their other podcasts, YouTube appearances, maybe it's not always possible, but consulted with people who know them. And maybe thought about it and done that for 2 or 3 months. There's some cases where I've done it for like 9 months. When I did an episode on Irish history, I prepped for that, I think, for 9 months. But that's unusual. And then when you show up at the interview, you ask like very specific questions to signal you want the discussion to be as high level as possible. And you hope to draw out of them their very best side, that you're showing them the respect of, you know, you never ask a hostile question, but just like a demanding question, explain this to me, something that's pretty tough and give them a chance to shine and then pre-select for guests who can do this. Ah, you don't want your guest to fall on his or her face. And very rarely does that happen in my podcast. So the preselection is also important.

In order for you to prepare for weeks or even months in advance, that means that you have to be preparing for multiple guests simultaneously. Otherwise you'd be releasing one podcast a month.

That's right. Right now I think I'm preparing for about 8 guests simultaneously, and 2 of them are a common topic, uh, but mostly they're not.

Okay, so that can look from the outside as you're spinning too many plates, but it's my understanding that you love to spin plates firstly, and it's not too many because some of the plates feed into one another.

Yes, I mean, I do think it's too many in this sense. I would love to record more of my podcasts over Christmas and in August, but no one else wants to. Uh, so I end up with parts of the year where they're too clumped together. And I think it is suboptimal. So if someone says it's too many plates, I would say, of course. And then I'll say, will you record with me, you know, on August 17th? And oh no, I'm away with the kids then. Like, fine. You're doing me a favor by recording at all, but that's just how it works out. It's like traffic, right? There's too much of it at 5 PM and not enough of it at 2 PM.

How far in advance do you put something into the calendar?

Oh, it really depends. So I just heard yesterday from the team of Alex Karp, you know, head of Palantir, and they're like, can you do one in 2 weeks? That's not preferred for me. If I have 3 months, I'm pretty happy. Or there's some people I just know, like Ezra Klein I'm doing. I've known Ezra for 20 years. I've done like 6 podcasts with him already. If that's short notice, it's no big deal. But Alex Karp in 2 weeks, I'll do it. I'll put in the work. Uh, but it's a, it's a challenge, right?

Now, what is the reason for you wanting to put in so much work other than quote unquote respecting your guest and it's your way of thanking them? Like, is it because you are insecure when you go into an interview, you want it to be a great interview? What is the reason?

Well, I'm not paid to do my podcast, right? We don't charge people anything. We don't sell ads. There's no revenue to me. I think it's a very efficient way to learn, to have to prep for all these people and then be confronted with the actual public event where you feel there's something at stake. And that means your ex ante prep, uh, will be more powerful than if you just read books on your own. And I just like having an audience. You could call that like conceit or arrogance, whatever. I, I wouldn't reject You know, the negative terms someone might use, ego, I don't know, whatever you want to call it, uh, definitely operates if you do a well-known podcast. That's one reason why it's free. I want more people to listen.

Is that another word for public intellectual?

Sure, of course. Most of us are like that.

So would you say that you tend to be critical of most public intellectuals?

I know a large number of them. Typically, I like them and respect them, and I try to learn from them. I have a lot in common with them because I'm one of them.

Okay, let's forget about the commonalities. What is it about you that makes you different as a public intellectual?

Depends who it is you're comparing me to, but I would say compared to the median, uh, I can absorb information at a much higher rate. Even like holding IQ or education constant. So that means I have access to a whole bunch of different kind of strategies that they wouldn't have. So odds are I've read more, I've been to more countries, I know more people, I'm at a good enough age, I haven't fallen apart yet, but I'm, you know, I'm young enough to not be slow, but old enough to have a lot of experience. So that's another difference.

Absorbing the information is different than just contextualizing the information. So the way that I view it is that you can, you can have a variety of nodes, but then not many edges. And you strike me as someone who has plenty of, of both edges and the nodes.

Well, I read very fast is mainly what I meant. I don't listen to YouTube any more quickly than my peers. That's one reason why I don't listen to that much YouTube. 'Cause I can't do it faster. So comparative advantage says I should do less of it.

Ah, when you're reading though, it's still just absorbing information, but it's not the same as contextualizing the information. To contextualize, at least for myself, I need to close the book. I need to think, I need to write, I need to talk to someone like you and ask questions or develop my own thoughts. It's not as simple as just reading and then it becomes contextualized. So how does it work for you?

Either writing does it for me or doing a podcast with someone, but sitting on my bum and thinking is worthless. I mean, there's no point. I don't get anywhere. You could say that's a flaw, but I need to be actively solving a problem with it, writing a piece, or no further happens. I might as well just read another book. Which I will.

Do you believe in free will?

No, I'm not always sure what people mean by the term, but I think there's some theory of the physical universe. I'm not sure that's entirely deterministic, but whatever that theory is, I think that same theory applies to living intelligent beings. If that makes me a determinist, fine. Uh, it may not all be strictly determined for reasons that are well known to you.

Are you an agnostic or an atheist or what?

I call myself a non-believer. I don't like the word atheist. I feel it's been abused by many people who are themselves of actually a religious mindset about their atheism. I do not have active belief in a deity, but I would say I'm more skeptical than a lot of people who just call themselves agnostics. So I say non-believer. But I don't rule it out. I just don't see what should impel me to believe in a particular something. It seems to me physics is some huge mystery, theoretical physics. The more I talk to people who do it, the more skeptical I become. I don't know what to make of that. It's not per se an argument for God, but I think I'm sufficiently confused that I shouldn't rule out any hypothesis that a lot of people have considered.

In this dappled universe where there's patchwork and some laws apply, some don't, is there room where rationality or logic doesn't apply and thus you don't need reasons to believe so-and-so?

It could be those words rationality and logic are begging too many questions. So I don't know what that means. Like, do I think there's a region of the universe where A doesn't equal A? I suppose I don't. But could I imagine there's some bigger metaverse and the other universes in the metaverse, they're so strange we can't even think about them? They have so many dimensions, or what counts as a dimension is something we can't fathom, and it would be so unrecognizable or unimaginable to us. That would not surprise me. I don't know if I could even recognize it if I saw it. But no reason to rule that out.

So what does atheism mean to you?

Well, sociologically, atheism right now means people who are followers of ideas like Richard Dawkins and hostile toward religion. It doesn't mean that in the dictionary sense, but I think the sociological sense is the more important one. And if I meet someone who just says, I'm an atheist, I think I am a little put off on average. I don't necessarily disagree with them, and I'm willing to explore it. That's why I don't use that term.

So something's concomitant with the atheism, and it's these trappings that you dislike, not the atheism per se. Tell me about these trappings.

Well, when they say, I'm an atheist, to me, they're a little bit signaling a reluctance to think about it probabilistically. If they could just wear a little button that gave like, what's the chance I believe in God? And the chance were like 1 in 80. They don't have to call themselves anything. I'd be quite happy. You know, my chance, I once said it's 1 in 20. I think recently it's gone up to 1 in 19. If someone calls me an atheist, I don't try to refute them, but I prefer to, you know, think about it in a somewhat more sophisticated manner.

Okay. It's odd to me that it would, you would have such meta-rationality, let's say, that you're able to discern the difference between a 1 in 20 probability and a 1 in 19 for something like this that flies under the radar. So I don't know how much of that was said facetiously.

I can tell you what nudged me. It's very simple and very concrete. So there are all these reports of UAPs, formerly called UFOs, right? I'm not sure what they are, but I have good reason to believe the mysterious data are real rather than fabricated. It doesn't mean it's alien beings. One possible explanation for those data are some kind of supernatural being. I don't think it's the most likely explanation, but now that we have the mysterious data, your chance has to go up a bit. So from 1 in 20, you go to 1 in 19. Why not?

Okay, let's hang on this for a moment. Why would that be evidence of something supernatural rather than super technological?

I think it's more likely super technological. That's why it doesn't go up to 1 in 3.

I see. So it could be either, and your super technological one went up much more than your supernatural one.

Absolutely. Much more. Yes. And we'll learn more soon, or at least possibly, because the Trump people have said they will disclose the files. I'm not sure I should trust or believe them. I'm not sure. I actually think we'll see the files and still be mystified. But I'll be doing more updating on that number, I think, in the next year, in whichever direction. If it's alien beings, which I'm not saying it is, but let's say it were alien beings, I'd be back to 1 in 20.

Explain about these files.

Well, I have spoken to many people in high positions of authority, typically privately, but there's one that's public, and that's when I did a podcast with former CIA Director John Brennan, who is former director of the CIA, a very smart guy, not crazy, obviously had access to as much available information as there is, I think., and I asked him what he thought. This is the public record. It's online. You can Google to it. And he basically says, I think it's alien beings. Now, I don't have to agree with him, but I feel I shouldn't dismiss him, that basically there are repeated recordings of some kind of things that move very fast, have no visible means of propulsion, and we track them on infrared, radar, and satellite data. And there are some eyewitness reports of pilots, and those 3 or 4 sources of data all seem to be consistent. Now, that to me is a big puzzle, and I'm quite sure I don't know the answer, but I do think about it a lot. And now the Trump administration— this is 2 days ago, I think— promised it will disclose all the files. Now, a promise is one thing. I don't actually think they will, but we'll see.

What was it like when you first encountered information about UAPs, or formerly called UFOs?

I didn't believe it. So until, I don't know, 5, 6 years ago, my probability would have been way, way, way low, below 1%, that there was anything to this. Uh, but I would say in the last 6 years, I've had separate private conversations with 9 or 10 people. Some of them would be people, either their names or their titles you would recognize, and they all tell me the same thing. And they say they don't know what it is, so I take it seriously.

So there wasn't an initial event that opened the door that had previously been shut?

No, no little green man who came down and shook my hand. I've never believed in sort of things like that. I don't believe in ESP or the near-death experience being real or whatever. Else you might want to put on a comparable list, Bigfoot. I've never believed in any of it. I'm not like the type to rush to believe. I was very hesitant at first. I still don't believe. And I still think it's a possible answer that our elite leadership is somehow collectively insane. It's possible.

Okay, when you increase the credence on the supernatural, does that not increase your credence on the near-death experiences and, and other that is associated with the supernatural?

Yes, it, it would, but those were like very, very low to begin with. So I could triple those and maybe I have, but it's still very low.

Who's your strongest, most cogent critic?

No, as a generalist, you don't necessarily have these people who are your critic. I would say the colleagues I have lunch with several times a week, Brian Kaplan, Robin Hanson, and Alex Tabarrok, would be my best critics because we talk the most and we kind of scream, not scream at each other, but the purpose of lunch, you know, is a bit to pick a fight and they do it and I do it and we're critics of each other and we have great fun and learn from it. They're my best critics. Now there's wife. It's a separate matter. Like that's always number one in its own way. But I take, took that not to be what you meant.

Yeah. Well, Larry David said your wife isn't a partner. She's an opponent.

Well, his opponent, uh, went and married RFK Jr., right?

So tell me about a time where your critic, it could be your colleague, was right about something you were wrong about recently?

What counts as recent?

Two months.

Well, I thought the recent Trump talk about Ukraine would not seriously affect the future prospects of Ukraine, that he sounded more pro-Russia, anti-Ukraine than say Biden, but that the talk wouldn't matter. But today, no, yesterday I went online and checked the bond yields. And since Trump was elected, the yield on the 1-year Ukrainian bond has gone up from 33 to 34% to I think 48%. And that's a big change over a small number of months. So my view that Trump was not gonna matter on that, I would say is looking wrong. That's recent.

You've talked about being immune to many common sources of stress and, and unhappiness. Is there anything that reliably triggers a strong emotional— could be negative reaction in you, could be positive.

If people are late, I don't like it. That would be one thing. I think I have various kinds of impatience, which mostly I can avoid. So the way my life is structured, I just don't have to deal much with various bureaucracies, but that would frustrate me more if the details of my current life were different.

Okay, let me push beyond the abstract and beyond the annoyance. Can you recall a specific instance where you experienced a visceral, uncontrollable, negative response? It could be anger, fear, grief, shame, and what triggered it?

Well, there's been just concrete times when I've been in dangerous situations, like been in a gun shootout in Brazil. Uh, obviously I was very afraid. Huh. I could come up with things like that. I'm not sure that's what you're asking about, but my regular day is really quite normal and even-keeled. And people, maybe at first they're reluctant to believe this about me, but the more they get to know me, they're like, hey, you're really like that. Rick Rubin told me this once about Paul McCartney. He said, when I first met him, I thought, this isn't the real Paul McCartney. So I got to know him and it's like, this is Paul McCartney. And I think he too is pretty I mean, he can give a hard time to the people he works with musically, but every day he wakes up, he's productive, he's happy, he's ready to go, ready to work, wants to do stuff, and other things don't get in the way of that.

What is the reason for that? Is it just genetic?

I strongly suspect that's genetics. I don't think, I've never seen any person who could learn it. But I have no like proof of that.

So it's like being low in neuroticism in the Big Five trait model.

I believe that's also largely genetic, probably much more than 60%.

Do you think it has anything to do with disagreeableness? Is that correlated with, with being even keeled? I imagine it's just neuroticism.

I don't think it's much correlated with disagreeableness.

Are you disagreeable?

Daniel Gross and I covered this a bit in our book. I don't think it's that useful a concept. People are selectively disagreeable. So there's a question, disagreeable about what? So if someone presents me with a new idea, I'm relatively likely to say yes and, in the way you're supposed to in improv. In that sense, I'm not disagreeable. Uh, I'm in a good mood, but I'm disagreeable in the sense that maybe I'm hard to persuade or, or willing to see things differently from some consensus. So I don't think disagreeableness is a quality per se in most people. It's contextual. And that's a problem with a lot of the Big Five personality constructs. Though not neuroticism. I think most highly neurotic people really are just neurotic on most things.

Speaking of the book Talent, you mentioned that, you mentioned something about selecting for contrary and disagreeable ambitious people. I imagine that it's good for corporations or for achieving some goal, but that also seems to go against social cohesion, which you also care about. So helped me understand that, that trade-off.

That's why you want selectively disagreeable people. They also have to know when to cooperate. So you hear Steve Jobs so often cited as the supreme king of disagreeableness, and there are these famous anecdotes in the biographies, which I've heard are true, but he was tremendous working with other people also. So was he so disagreeable? Yes and no. It was selective.

I know that you care plenty about the weird, and it seems to me like if we are selecting for people who are weird, at some point you get people who are appearing weird rather than those who are genuinely original. Like Elizabeth Holmes was appearing to be like Steve Jobs in many respects.

Right.

And so there's, there's a performance to the weirdness.

Well, you might have to have tests for people. Right? You just want to get people talking about something they're not prepared to talk about, and you get a sense of how weird they are.

So you just surprise them on something they're not expecting.

And multiple things that you might need, right? Not just one. And I think pretty quickly that way, especially if you're talking about aesthetics, you can get a sense of how weird the person really is. If they tell you their two favorite bands are Flipper and the Dead Kennedys and they can talk about them at length, I'll start to think this person is weird. If they tell you their favorite musical act is Taylor Swift, it doesn't rule out them being weird at all, but like they've gotta come up with something else.

Now, okay, so that has to do with being conventional or, or not conventional, but there's also context about that. So what's conventional in Silicon Valley isn't conventional in Boston and so on. So how do you disembroil all of that? It's too much of a blanket statement about Taylor Swift.

Well, you have to know a lot. You have to have been to a lot of places and met a lot of people. And there the returns to experience are quite high. I don't think it's that easy.

I want to read to you something about Nassim Taleb. In 2018, There was a blog post by, by you where you shared your thoughts on Taleb's book, Skin in the Game. You wrote, Taleb has some interesting ideas and some that seem, at least on the surface, crazy. I'd like to see them challenged enough so that I could see if there's merit to them. So what are these ideas that you consider crazy?

In Taleb. Well, he has a lot of crazy ideas about Lebanon. I'm not even saying they're all wrong, but they're definitely different from what other smart, educated people tell me. And he's relatively sympathetic to the Shiite side of the conflict, uh, and really very, very unsympathetic to the Sunni side and complicated views on the Christian side and mostly anti-Israel. Again, I don't want to get into what's right and wrong. I've never even been to Lebanon, but I definitely know that a lot of people are surprised when they hear his takes. Now, he's Lebanese, and I think his takes come from his personal experience, but they're definitely highly weird views. They're in a sense not that weird in Lebanon of his demographic, but for an educated, famous public intellectual out in our world, They're definitely strange.

Okay, forget about whatever's political. Is there something about his worldview on behavioral economics or the consequences of fat tails or something like that that you disagree with?

You know, he writes enigmatically and almost in the form of maxims, you could say, but at least at times he seems to suggest there are extra-normal high returns from buying out-of-the-money put options. I would be surprised if that were true. I don't see that in the research literature. I think he's insisting on it. That to me is a weird view. I would like he stated it somewhat more precise, but look, he's Tyler. He's not doing a paper for the Journal of Finance. If he doesn't want to state it more rigorously, that's fine. But still, at the end of the day, I'm more likely to believe what the efficient markets people say than what he said. And I think that's what I had in mind when I wrote that particular sentence, though it's not his weirdest view either.

Now, Nassim has also been critical of you in the past, and I believe this is from, I believe, The Black Swans 2007.

I think that's right.

So you had a blog post, or he had a blog post, sorry, where he was responding to a review of yours. He said that you made several errors. You, I'm just going to say you instead of Cowen. You made several errors in your analysis. Taleb specifically disagreed with your assertion that long shot strategies do not yield profit. Taleb contends that the claim overlooks the impact of unpredictable large-scale events, which he calls black swans, on the financial markets, and that your analysis, Tyler, relied on limited data sets that excluded these critical outliers. So then Taleb expressed interest in debating you about the concept of convexity of trial and error and heuristic learning in relation to Bryan Kaplan's work. So what is this about? And has this debate ever, has this debate occurred?

Well, I've met up with him since then. He and I get along quite well, in fact. He, before COVID invited me to go visit his mother with him in rural Lebanon. He and I were going to do that. But on the substantive question, I still think at the end of the day, the research literature in financial economics sides with me and not him. We never had like a debate. I'm not sure public verbal debates are a good way of resolving a lot of questions, but he and I have talked about plenty of things.

Why? Why are public debates not a good venue or not a fruitful approach to resolving?

Depends on the question, but it's hard for listeners to evaluate claims when they come at such a rapid pace. To have a written exchange which is refereed, I think, is usually better, especially for difficult and technical material.

What about rather than a debate, just having a conversation moderated by someone else and it's just you two speaking?

That could be good.

And it's not set like you're trying to prove one another right or incorrect.

I still think written on average is better, but sometimes that works. I think those conversations are good for learning new things, which is great. I'm not sure they're good for settling the truth.

Do you focus on money?

What do you mean focus on? I don't maximize money. A lot of my outputs I give away for free.

Well, I guess I was getting at a quote. Where you said that you're not focused on money, but then at the same time, this is an odd statement for an economist to make because it seems like ultimately the majority of our world is shaped by economic incentives, whether it's implicit or direct. And so I just want to know about your, your motivations and, and how you think about them.

Not all of those incentives are money in economics, but I would also say I've set up a number of my projects to be sufficiently low cost like the blog, podcast, that I feel I'm always able to fund them. They require some very modest amount of money, but an amount I know I'll more or less always have. So in that sense, I don't have to worry about money. But if I truly were in the position that like I couldn't afford the, what, $220 we pay to WordPress each year, it's probably more than that by now. The fact that I don't know tells you I'm not that worried about it. Uh, then yeah, I'd worry a lot about getting that, that $200, $300.

And again, your podcast will be on screen and link in the description. And do you mind repeating the podcast name?

Conversations with Tyler, me being Tyler, and there's about 250 episodes. Uh, there must be more by now, but it's a few hundred.

Now you admire ambition, like in your work on, on grit and stamina. That was my my takeaway. But there's, there's a couple of different types of ambition. There's one where it's just you have a grand goal and you want to achieve it, but then there's another where it's an ambition driven by insecurity or a need for external validation. And they could both be highly ambitious. And I'm sure you can even think of current billionaires who are, are intrinsically motivated by insecurity. So is there a difference in your mind between a healthy ambition and an unhealthy one?

I would just say, I think those different kinds of ambition typically are bundled together. They may not even be fully conceptually separable. Um, it comes with the territory. You want people to be impressed by what you do. Insecurity is a legitimate way of describing that, but not the only way. Hmm. Some of it is you just want people to be impressed by what you do, and that's fine. That's ambition.

Now, as we close, I want to talk about grand theories. You said you have a theory of everything, but at the same time, you dislike simplified narratives. So is it just your theory of everything that's running in the background of your mind is sufficiently complex? Are you able to tell that it's, it passes a threshold of sufficient complexity, or how does it work?

I don't think of myself as having a theory of everything. I think I have a, a central mission in what I do. I think some of my tools, especially economics, are very conducive to theory of everything thinking, but they're not my only tools. But when I think like an economist, it's opportunity cost, demand curve slope downwards, and think about how people will try to find the gains from trade. Those are my theories of everything, or the theory of everything within an economic framework. And that's a lot of what I do.

What are you working on right now that excites you?

Well, everything I'm working on excites me, or I wouldn't be doing it. But I've started a new book recently. And the title of the book is simply Mentors. It will be a book on mentoring and also how to be mentored. Hmm. And I feel it's one of the relatively few book topics that will not be made obsolete by AIs, that a human has to write that book. So I'm writing it. A lot of other book projects I will not do because of AI.

Now let's imagine you're speaking to people who could be mentors, and then after I'm going to ask you the same question about but about mentees. So you're speaking to potential mentors. What is it? What is your book about? What is the advice that you have?

I don't want to give away the book very much, and I have just started it. Uh, I've spent a lot of time in my life, you could say, doing mentoring, being mentored. A lot of the best advice is context-specific, and you're not teaching people 3 simple rules, but you're teaching them how to think through context. That's what the book will try to do. There's not really a go-to book on mentoring. If you ask deep research about mentoring, as I mentioned earlier, you get a lot of slop. Now it's slop from the humans, it's the human's fault, but it's not that good. So if I can do just a bit better, I think the returns to mentoring can be pretty high, and I'm hoping that will be useful.

Is it less a book about how to be mentored or how to mentor than it is a book about the significance of mentoring?

It will be both. There's chapters on both of those. Planned chapters, that is. They're not written yet.

What else are you working on that you are happy to talk about because it's not something that is bound by some contract?

Well, there's all the podcasts I'm preparing for. I still need to write another Bloomberg column for this week, which I guess I'll do tomorrow, this being Friday, and I don't have a topic in mind. That's, I wouldn't call it an emergency, but given blogging and writing for Bloomberg, I would say the deadline is always now, and the book projects are longer term. The podcast projects are ideally something like 3-month length. And then the travel project is to get to all these different places, and I'll have a chance to do more of that when the summer starts. And my summer travels are not mapped out yet, but that's the next thing I'll be doing is figuring out where to go and for how long.

So when you're preparing for a guest, there's difficulty in at least two regards. One is that it's plenty of, of information you need to consume. So it's plenty of work, but then it can all, that can also be easy if it's something that you like, like let's say it's history. You love history. Let's just say that.

Yeah. And it's true. I wouldn't say it's easy. I would say it's fun and I don't have to force myself to do it. Uh, but it's not easy in the sense that many things can go wrong. And you have to pay attention to all of them to make sure they don't go wrong. Now I have a lot of experience in doing that. I'm usually, though not always, pretty happy with how it goes.

So then what would you say is the most difficult? Like I mentioned, there's difficult in terms of how much prep does it take, but then that can be mitigated by how excited are you to undertake the tasks in order to complete it. So is there a particular type of guest or, or, or person that is the most challenging?

People who are in sciences that are not that conducive to simple models, and a lot of biology is like this. It's very descriptive. There's all these multiple layers and levels of what's happening. Evolutionary biology is more like how economists think, but a lot of biology is very tough to prep for, I find.

So in the complacent class, You noted a decline in, a declining dynamism and a great reset coming. In, in Big Business, you defend large corporations as engines of innovations, saying that the United States is great due to how it balances talent with the wisdom of old companies. How do you balance these two viewpoints?

Well, big businesses are highly innovative. They pay much higher wages. And they now account for more than half of the jobs in this country. And when I wrote that book, I felt they very much needed a defense. That was the age of Occupy Wall Street and peak Elizabeth Warren, Bernie Sanders. Uh, so that's geared toward a very different debate. Complacent Class is about the fact that for a while, US technological growth slowed down. I would date that roughly 1973 to 2020. As I've written, I think the great stagnation is over. Both biomedicine and AI to me are proof of that. And we're now living in this much more dynamic world, which people somewhat hate because it's more chaotic. And there's also more room in politics from anyone's point of view for things to happen that you don't like. And the stasis is over. So the prediction toward the end of Complacent Class that we would enter this more dynamic era and it would be very unpopular. I think that's coming true.

To close, this podcast is viewed by professors, researchers, and students as well. So I ask people often, the people I'm interviewing, what advice do you have? And you could separate it into these classes. Most of the time professors are reluctant to give advice to other professors or researchers. They're like, that's not my job.

No, no, I'll do it. I'll do it.

Don't worry. Great. Okay, wonderful.

Okay. First piece of advice: at any point in time, whatever is the best quality AI model, if you have to pay something for it, it's probably worth paying for. It's not going to hold true at all future price points, but for the time being, it's definitely true. Pay the price, get it over with, and do it. You'll learn much more than you think, and you'll be asking yourself, how did I ever live without this? Second advice, if you can travel at all, at some point, just pick out a place where you think, like, I don't want to go there, and go there. And you'll be surprised. That's my second piece of advice. Two for the day.

And now your advice for students entering the field. Like I mentioned, this, this podcast is directed toward mathematicians, physicists, computer scientists and philosophers?

First, I would repeat the same two pieces of advice for students, but I would add mentoring is more important than you think. Networking of humans will grow greatly in value as AI expands. So the returns to investing in your network will be 5 to 10x higher than how they look now. And people typically already underinvest in their networks. So overcome your own inertia, put yourself out there and get to meet and know more people.

Why does AI have anything to do with networking? Why would it be more important to network now when AI is expanding?

If you're a person who can do a lot with AI, you can be much more impactful in the very near future. I'd say in the next 2 years, but soon in any case, but you need resources. You might need money. Maybe it's venture capital, maybe it's debt, maybe you need to hire some talent, maybe you need mentors. So you have this potential of being 50x more impactful, but you still need the resources. So the returns to being well networked are just much higher. Now, if you're just literally not going to do anything with your life, I guess they're not higher, but it's conditional on that you might end up being agentic. Network now. The AIs are not going to do it for you.

Professor, it's been a blast. Thank you so much.

Same here. Thank you very much. I've enjoyed it greatly.

Thank you, Professor.

I'm very delighted to have been on this podcast. You know, in the last month or so, 3 or 4 different people, people I don't know, reached out to me and emailed me about it saying, you should check this out. And what's striking it's not that they recommended any single episode, they were recommending the entirety of it and your whole notion of having this project of learning people's theory of everything.

So that wasn't precipitated because I put out a call. I don't know if you know, but I put out a call for questions for you. It wasn't that.

I know. I don't think so.

I don't know anything about timing and the like, but no, 4 days ago I put out that call.

So you said this is earlier. You hadn't asked me yet, and I was looking at your page. And then when you asked me, I said yes because all these people had told me, and I'd been looking at the page and I hadn't done anything with it, but I thought, this looks great, I should do it. And then you invited me.

Yeah, that's wonderful. Yeah. Well, it's an honor that you said yes. I'm— it's a huge honor. Thank you.

A pleasure to do it and keep in touch.