--:-- Dwarkesh Patel: All right. Today, I have the pleasure of speaking with Byrne Hobart, who's a writer, consultant, and investor who writes at diff.substack.com. That's D-I-F-F.substack.com. Here's my first question, Byrne. You wrote an article called "Foxes and Hedgehogs," and here's the final line in the article: "If it looks like somebody doesn't have a single big idea, they probably do, and it's a good one." Now, you are somebody who writes every single weekday, and you're writing about all kinds of things from finance, technology, and so on. And it might seem like you don't have one big idea, but that's exactly why I should expect you to have one big idea. So here's my guess where your big idea is, and you tell me if I'm wrong or right. Basically, most human decisions, whether they're made by individuals or by institutions, can be boiled down to some simple financial concepts like expected value, optionality, volatility. And because other people are missing this, they're not reporting on the important trends, or the reporting about them misses the long-term impact these trends will have. How far off am I?
--:-- Byrne Hobart: I think that's a good mental model, and it's one that I've used a whole lot. I do think that it's definitely true that financial concepts can be usefully applied in a lot of different contexts, but it's like any other model. You want to use the model, but you also want to be aware of the deficiencies in the model. Half the point of the model is to make predictions. Half the point of the model is to say, here's a list of assumptions you have to make in order to make a reasonable prediction. And if you can't make all those assumptions, then the model does not actually apply, or you should at least not be surprised to be surprised.
--:-- So that is definitely a big part of how I think. I would say maybe the big idea is more of the big question, which is just how do people coordinate when they're solving complicated problems? Because a lot of the interesting problems in the world, you can't have one lone genius solve them. There are various institutions that try to solve them, but a lot of times, the institutional mandate is not to solve that problem. It's something else. And maybe there's a real mandate and there's a fictional one. A lot of the mission-driven public companies out there will have both the mandate of, say, SpaceX: we're going to go to Mars, but also we're trying to maximize shareholder value. And it's never clear which one is actually the external story that they're just telling you so that they can really accomplish their internal goals. It's actually not clear which of those is which.
--:-- Maybe even within SpaceX, there are people who think, okay, the Mars stuff is how we recruit good engineers, how we raise a bunch of money when we go public, why the stock price will be really high. But really, SpaceX is trying to maximize earnings per share on a 10 or 20 year time frame. And then there may be other people at SpaceX who are like, yeah, everyone thinks that SpaceX is just this company that's trying to make money, and of course we will make money, but the real point is to make humans an interplanetary species. So coordination is just a really hard problem to solve. It's a hard problem to think about.
--:-- Even when you think about how these different institutions have different goals, different stated goals, they may have internal goals that are unstated but exist. A lot of the time, those goals are just that whoever is there wants to stay there, wants to get raises, and wants to feel important. And if they fail to accomplish their goals but the problem they're working on keeps getting bigger, they could stay important for a really long time. So given that there's naturally just a lot of double talk about this, and given that if you're coordinating between different groups of people who have different goals themselves, you sort of need to do some double talk. It's just a really hard problem to study. And it ends up being a problem that shows up in a lot of different domains. I've talked about it in finance. It shows up in tech companies all the time. It shows up in politics, both in the sense of who gets elected and which bills get passed, and then in the much broader sense of just how do humans with intractable desires resolve them to one party's favor or another, or just figure out some way that they can all come to an accommodation.
--:-- Dwarkesh Patel: Okay. So then just to boil that down in terms of the question, is the big idea that you are trying to figure out what the secret goals of institutions are that are coordinating on a mass scale?
--:-- Byrne Hobart: Yeah, I guess if you're thinking about what problem should be solved, the problem of productivity stagnation is really important. It affects everybody. It's dire, and it tends to compound over time because different technologies cause other technologies to accelerate. Different social technologies are kind of invisible until they go away. You don't really know that you were working in an effective company, for example, until either it becomes ineffective or you go somewhere else and realize, wait, I can't actually trust that if I email someone and ask them a question, they'll actually get back to me with a good answer. Those kinds of things are only visible when you don't have them.
--:-- So between technological stagnation, which is pretty visible in the productivity stats, and then social technology stagnation, which is a lot harder to measure, but you do get a sense of it just from looking at how the U.S. government functioned in the 30s, 40s, 50s, and 60s versus how it functions today, it's clear that there's some decline happening there. Both of those involve a lot of coordination problems. So studying how people collaborate to accomplish goals is a way to figure out how we can all collaborate to accomplish the goal of getting productivity growth back up to where it used to be.
--:-- Dwarkesh Patel: Okay, so let's talk about that. I was planning on asking you about that anyway, so it's very good that you brought that up. You wrote an article where you are bullish on the claims that the great stagnation is over based on anecdotal data. I think the title of the article was "Stagnation" or something, and I'll link it in the show notes. From what I remember, one of your claims is that even though you have these sort of individual developments like mRNA vaccines, if the regulatory apparatus is not available for them to get approved and scaled quickly, then it doesn't matter as much. I'm wondering if we can quantify that. So if we look from 1971, how much higher would total factor productivity be right now if all the regulations were perfectly in place to enable productivity as high as possible?
--:-- Byrne Hobart: Well, I think it's a mistake to take a time period and say, if we kept things the same. It's more like if we'd adapted in better ways, we'd have more productivity growth. If you just take the gap in measured total factor productivity growth, which is around 1% a year, you compound that over a 50-year time period. So you probably get to something like 60 or 70% higher GDP per capita in the US. It's kind of hard to imagine exactly what that would look like, because it's not like we would all just live in 70% bigger houses and have 70% more cars per household. It's more that there would be other additional products, or that people would have longer lifespans, or hopefully longer health spans, which I think is probably more solvable and definitely more pressing in terms of total human disutility than the lifespan issue for now.
--:-- So you can sort of ballpark what things would look like if that mid-20th century productivity boom had gone on forever. But you're looking at the data, which can tell you the gap between this hypothetical possibility and reality in this aggregate sense, but it can't really tell you what kind of amazing world we would have. To figure that out, you have to look at specific examples, because that whole sweep of productivity growth wasn't just that there's a productivity dial and someone was turning it up slightly and then turning it back down. It was this set of specific inventions that got developed and deployed and continued to get more efficient over many decades. So you'd have to think about the various specific innovations that could have spread around.
--:-- A really good book to start with there is *Where's My Flying Car*. It's basically this extended, very detailed, very thoughtful, very well-footnoted rant from someone who really wants to have a flying car. He's interested in them, he's flown things that are basically flying cars, he's done all the math on how they would work, he's looked at the safety concerns, he has looked at how they would affect things like the layout of cities and the nature of travel. And he makes a really strong case that we should have them and that they have been not specifically illegalized, but the regulatory noose has tightened. It's just gotten a lot harder to fly anything, and much harder than that to build something totally new and fly it without getting in a whole lot of trouble.
--:-- Of course, the trade-off there is we don't have very many flying car accidents because we don't have any flying cars. So there would be downsides, and the early innovators will tend to have more accidents. I'm reading a bit about the early history of the steam engine. Lots of explosions, lots of times where we basically learned things about the physical limits of different kinds of materials by building higher capacity boilers and then having them explode. So there's definitely some downside that we are avoiding, but there's also a lot of upside we're missing. That downside is easier to see because it is weighted to the very beginning when the new technology doesn't work especially well but it can kill people. And we don't see what things would look like in 20 years when it's working really well, it's extremely safe, and you have to try pretty hard to actually kill yourself using it properly.
--:-- Dwarkesh Patel: Right. He especially blames the Green Movement and all the government activity it sponsored as being a big culprit there in that book. One of the questions that comes up when we talk about how regulation has been slowing progress is that there are more than 200 countries in the world. You could develop flying cars in any one of them. But what are the odds that you have a correlation that's so exact between countries that you managed to kill off the avenues to flying cars? Is it just that there are very few countries where something like this is possible and their regulations are correlated? What's going on?
--:-- Byrne Hobart: There are a couple of things going on. One is that the U.S. is a very rich country, and that allows us to indulge in a lot of weird hobbies like building flying cars, and before that, building heavier-than-air flying machines in general, or messing around with the earliest regular wheeled cars. It's easier to do that stuff when you're able to eat three meals a day and you can have shelter. So it doesn't really matter if it's legal to build something like this in a country where they don't have the spare wealth to build it.
--:-- For aviation in particular, the barrier to entry for countries is pretty high. Planes have really complicated supply chains. They're very sensitive to the quality of a lot of different components. For larger planes, like wide body commercial planes, you basically have two entities in the world that make them competitively: Boeing and Airbus. China has been trying for a really long time to catch up in that business and has not been able to do so just yet. So that is an indicator that it's pretty hard, or that China's GDP per capita is not quite at the level that can support an industry like that.
--:-- So it does matter if regulations are strict in a handful of countries that are pretty rich and that also have developed capital markets. That's another thing you really need for something like this. Once someone has the product, once they've done the proof of concept, once they've shown that it can work, it has to scale up and it has network effects. That is actually one of the exciting things on flying cars that has started to happen since I read *Where's My Flying Car* a couple of months ago. United Airlines has actually been putting in orders for basically flying cars. They seem to be pretty serious about messing around with that, and messing around with supersonic travel, and really rethinking the range of speeds at which human beings should fly and the range of distances at which they should fly. So things could be looking better there, but that's also maybe a sign of how difficult it is to expand this industry. You need one of the big four airlines to back you up if you're even going to have a hope of selling these products and actually selling tickets to passengers who will then fly in their flying cars.
--:-- Dwarkesh Patel: One of the mysteries, though, is why don't Western companies with their capital and their technologies, if they're getting stuck in the regulatory process in the West, why don't they just do their experiments in Honduras or Senegal or something? You can imagine this, for example, with human challenge trials with COVID. You could have some poor country completely inoculated to COVID by March, and we're still dealing with the aftermath. It's mysterious why that didn't happen. I suspect it'll happen next time. If there's another big pandemic, I do think that you'll have some people who just say, we're going to do human challenge trials. I'm going to whatever this country is, I'm going to bring all the research as I can, and we're going to develop a vaccine, we're going to test it. Hopefully we do a good enough job that it's politically untenable to get us in legal trouble afterwards.
--:-- But there is this weird coordination between media on the one hand and regulators on the other hand. Whenever someone is breaking these rules, if you try to go around FDA rules, it's pretty clear that that's what you're trying to do. It's pretty clear that you were saying, I don't actually think these rules should apply, so I'm not going to follow them. And that gets a lot of attention at the FDA, and they have a lot of power to make sure that you are never able to sell whatever the result of that experiment is in the US. It's not like the FDA is going to say, if someone in a different country invents a miracle drug and it works really well, and then they try to sell it in the US, it's not like the FDA will just flat out refuse to approve it, although they will take a long time. It's more that if they know that you are specifically trying to rat around the regulations, they take that pretty seriously.
--:-- That seems to be just a feature of big institutions in general. If they set up some kind of rule and you thwart it and that gets their attention, then they're very serious about making sure that you don't thwart it again. It's not just a public sector thing. With Disney, if you try to make knockoff Disney characters, you will hear from Disney's lawyers. If you try to host pirated video games, you hear from the game publishers. A lot of big institutions are very serious about keeping their rules enforced as much as they can, both because they have those rules because it benefits them and because it's a source of legitimacy that they can write the rules and everybody follows them.
--:-- Dwarkesh Patel: And it gets pretty sinister. I read a book by, I forgot her name, but she worked at a pharma company and she wrote a book called *Death by Regulation*. One of the points she made was, if anybody actually talked about how many deaths the FDA causes by just not approving drugs, if the company says, listen, here's the approval process, here's how much it sucked, that really has an effect on whether future drugs that they make get approved. So there's really a silence on that.
--:-- But speaking of the FDA, here's just one technology I've been interested in. What are the odds that polygenic scores for embryo selection get approved by the FDA?
--:-- Byrne Hobart: I suspect that... I'm definitely not an FDA expert, so I'm probably the wrong person to have a strong opinion on this. My general sense would be if you are testing for genetic diseases, and maybe you can broaden the definition of what constitutes a genetic disease over time, that stuff seems to be pretty safe. But if you're doing things like, I want my kid to be tall, not just normal tall, but like can play in the NBA tall, and I would pay a million dollars to select the embryos that give me an NBA qualified kid, I suspect that that is going to be very hard to approve.
--:-- But China's government seems a lot more open to genetic stuff. They were very early buyers of a lot of genome sequencing equipment. And they also... Yao Ming is literally the result of them telling two very tall people to get married and have a kid, and he turned out to be tall. So we know this genetic stuff can work some of the time. That government seems much more open to it. So I suspect that if that kind of thing does happen and it's legal, it's probably going to happen in China.
--:-- Dwarkesh Patel: Right. I didn't think about it that way, that you could just kind of sneak it in through other tests. You can imagine, oh, we're just testing for certain nervous system diseases, and then, oh, wow, would you look at that, they correlate with intelligence.
--:-- The next question I have is, Robin Hanson has written recently, and we talked about this, the idea of just patient longtermism, which is the idea that you park some money in a compounding fund, and within a few centuries you will be the wealthiest person in the world, if you just assume some sort of compounding return. In fact, one of the things I think you said at some point about your blog is, "I'm trying to make my great grandchildren rich by just identifying these long-term trends." As far as we know, this hasn't happened. It happened in odd cases like Ben Franklin, but on a large scale, this really hasn't been tried. It might be in attempt right now, but what gives? Why is this not a strategy people are more familiar with using?
--:-- Byrne Hobart: I think it wasn't Piketty's main point, but Piketty does have a point here that one of the things that resets this is war, famine, pestilence, revolution. Those things tend to knock everyone's net worth down pretty substantially. And if your net worth is in financial assets that are relatively easy to seize, then it goes down a lot faster. So that is probably part of what's going on.
--:-- There was that study a while ago looking at the wills of Florentines who died in the 15th century and how the last names that were more prominent then are actually still more prominent today. The last names that were more associated with being a doctor in 14th or 15th century Florence, those people are still more likely to be doctors. So there is some persistence that actually outlives the direct loss of wealth.
--:-- But I think if you're trying to compound money for a really long time, I don't think that the main thing that you want to focus on is your annual returns. I think the thing you want to focus on is what are the once-a-decade, once-a-century, and if you're really thinking about it, once-a-millennium risks that will turn up and that can actually push that return to zero. I don't know of any financial asset that you can actually trust in that sense. The US is a really weird case because we actually haven't had any of these massive catastrophic drawdowns in everybody's wealth that basically every other country has experienced. So if you look at it from a US context, you can actually imagine someone who put some money in a trust and they're just buying mostly treasury bonds and also stocks, and it compounds over time and ends up being a colossal amount of money.
--:-- But there's actually a story in Adam Smith's *Money Game*, the pseudonymous financial commentator, not the economist. He talks about a guy who did that in the 19th century. He bought a lot of very reliable blue chip stocks, and I think he set up his will so that his heirs just could not get the money for several generations. Smith says that there were years and years of legal battles, and finally the heirs were able to get access to the trust. And everything was worthless. It was all like American Alarm Clock Manufacturing Co. that had been bankrupt for 20 years by that point. So if you're just picking a set of stocks over a long time, they'll all go to zero. Over more intermediate time periods, you can rebalance, and that keeps the portfolio value above zero. But if your country goes through a war or there's a revolution, then there's a good chance that all of that value gets wiped out.
--:-- Dwarkesh Patel: Interesting. This is related to a question that I was talking to with my last podcast guest. You've written an article called "In Praise of Filter Bubbles." This is somewhat related to it would be good if there were more cults and more secret societies. I was just wondering, we wouldn't know this because they're secret societies, but what would you guess, how many secret societies do you think there are that have assets in excess of $1 billion? If any?
--:-- Byrne Hobart: Well, that's a tough question because you would need some kind of legal structure to own the assets. And if it's going to persist for a long time, you need some system for figuring out who's going to be in charge and for picking people who are going to keep that secret society going. I wouldn't be surprised if there were organizations that might have started out trying to be a secret society that accumulates a bunch of money, and eventually they end up being run by people who are just in it for the money.
--:-- Maybe a fair number of universities. People who started universities often had peculiar ideas about how human beings should behave and how to educate people. If you have a bunch of money and you're thinking about what to do with your life, and you say, you know what I'm going to do, I am going to exert maximal influence on people at impressionable ages and also make these people really important so that they have a big impact on society, you're already doing a secret society conspiracy. You're just calling it Harvard or Yale or something. So you could actually look at modern universities as having maybe evolved from something closer to a secret society. And then they have their own little secret societies inside of them, which have also apparently gotten really boring and have just become this kind of careerist thing that you do if you want to get a good job at an investment bank.
--:-- It's hard to keep secret societies alive for a long time. I think part of the reason for that is you do have to keep secrets, but you also have to reveal secrets. If people join it because of one narrative and then you tell them once they're pretty high up, okay, that narrative is totally fake, here's what's really going on, some of them are real believers. So you can actually end up with your secret society failing because it recruited people who thought they were the right people for it and they turned out to be the wrong people for it.
--:-- There's actually a kind of weird example of this happening in reverse with Al-Qaeda. The book *The Looming Tower* talks about bin Laden's early life and his early campaigns as a terrorist. The impression I got from that was that a lot of the people who worked with bin Laden in the 90s thought he was a loser, but that he had money. So they were sort of scamming him and just telling him, oh, yeah, next year we're definitely doing a terrorist attack. And then somehow he actually organized something and was able to do several successful terrorist attacks. But for a long time, it looked like a lot of the people involved just did not think that Al-Qaeda was anything other than a way to mooch off the bin Laden fortune. So you have a lot of randomness in what institutions try to accomplish, how they try to accomplish it, and who they end up recruiting.
--:-- One way around that is families. With families, if you have some familial goal and it's going to take multiple generations, you can be more honest with your kids than with someone you're trying to recruit from the outside. You can sort of work up to telling them what the actual plans are. I think that with multi-generational wealth, especially families that got rich first and then got into politics, there's probably some of that going on. It's not like the Kennedy family was some huge, weird secret society. It was more like there was a conspiracy, and the conspiracy was get one of Joseph Kennedy's kids elected president. And then they did it. Mission accomplished. That might be the more viable model.
--:-- So if you're trying to count up secret societies that have secretly accumulated money, what I would look at is family fortunes where there's someone who made a bunch of money and then their kids are doing things that are more prestigious and more power-seeking but are not as financially rewarding. Probably some of those basically count as secret societies in some approximate sense. I don't have a good way to speculate on how many secret societies there are, how powerful they are, etc. Either there aren't very many or they're really good at being secretive. I guess one way to say that the population of them is probably pretty low is that we don't have a lot of dark matter to explain. There are not a lot of times where we look at what happens in the world and we have to think, wow, there are really dark forces at work. There are a lot more times where you can look at the world and say, wow, there are pretty incompetent forces at work and nobody knows what they're doing and there's nobody in charge. You rarely see all the chess pieces moving in a really clever way. You just sort of see the chess pieces moving at random.
--:-- Dwarkesh Patel: Right. The one thing that shapes my view on that is UFOs. Either they're aliens, or it's a military thing, or just an error. But one of the alternatives slightly nudges your prior on secret societies up, because maybe that's what's confusing the military.
--:-- One example, by the way, of a familiar secret society, it's not even a secret society. I was reading Andrew Roberts' biography of Churchill. His dad was a minister and was planning on becoming Prime Minister. He just died early. Winston Churchill just adopted his dad's entire platform even after his dad died.
--:-- You wrote an essay called "Optionalities for Enumerated Cowards." It really reminded me of this little passage from William Deresiewicz's book *Excellent Sheep*, which he wrote about the caliber of students that are going to college nowadays. I just want to get your reaction to this passage. He writes, "Yale students are like stem cells. They can be anything in the world. So they delay, they try to delay for as long as possible the moment when they have to become just one thing in particular. Possibility paradoxically becomes limitation." I guess I'll get your reaction to that generically, but I'm curious, to the extent that that quote reflects something true, is it post-selection in the sense that the people who get out of Yale just become risk-averse? Or is it that the people who go to Yale in the first place, getting generic degrees, are they risk-averse in the first place? They're looking for an option, basically, in the form of the insurance that Yale University offers. What's going on?
--:-- Byrne Hobart: Well, I would say it's probably a lot of selection effect. I do encounter a lot of really bright people who did not go to elite schools and did really interesting things instead. I don't know that the schools really make people conformist, but definitely if you can get into a really good school and then go straight from there into consulting or banking or, I think at this point, big tech, it is a pretty conformist thing to do. No one is going to think you made a bad decision. Whereas if you have that opportunity and you go do something else, a lot of people are going to wonder what's really going on. So I do think it selects for some risk aversion.
--:-- Some of that is just that it's incredibly competitive to get into these schools. There's a lot of randomness at the top now. That is partly a feature of competition: the fiercer competition gets and the more stratified things are, the more likely it is that any one person is where they are because they were a little bit underslept on test day or got a couple answers right. It's a lot more skewed to randomness at the high end, which actually would make people more risk averse. If they got into a worse school than they really deserved, they're stuck with this feeling that it's hard to attain things, so it is worth going on the safer path. If they get into a really good school that is better than they expected, they have this sense that as long as I stay on this exact path, I am continuing to be 99.9th percentile instead of being stuck at 99.5th. That is the best life I could expect to have.
--:-- It is really hard to force people to have agency. It's really hard to force people to take risks, except in simulated, fake sorts of ways, but it is an important thing for people to do. That essay, I'm very happy with it in part because it was fun to write something about Bastiat, and in part because I think it's true. From a financial modeling perspective, one of the things I realized when I was thinking about optionality in your life—being able to choose a lot of different life paths and not having to be stuck with one thing—is that in financial markets, you can buy optionality very directly. You buy stock options. And it turns out that the converse of that, selling people optionality, is a generally profitable strategy. It does have drawdowns during a market crash, but in general, you make more money over time by systematically selling optionality rather than by systematically buying it.
--:-- I suspect that something like that is probably true in the real world, especially because there are very few people who are explicitly saying, "I'm trying to minimize my optionality. I'm trying to cut off options and not have choices." So given that there are some decisions you can make that do require you to forego other decisions, I suspect that those decisions are going to be good, or they're going to be better for you just because so many people are looking for maximum optionality, so they're not taking them.
--:-- Dwarkesh Patel: Yeah, yeah. One way of reading Tyler Cowen's "complacent class" is basically just that our preferences for optionality have grown a lot. You can think of it in terms of not starting a business, not having a family early at least. That's just a way to preserve your optionality. What is the explanation for that? If it's true that we're a society where we value optionality more than we did before, what explains that?
--:-- Byrne Hobart: Yeah, I think a lot of it is this wealth effect. In general, money is a way to get freedom in individual ways and then in aggregate ways. You have a lot less optionality if you're living in a subsistence society where your choices are either do something that gets you fed or starve. So we definitely have more choices as we get richer. I think one of the issues is that we are partly wired to be more satisfied with actually making choices and pursuing specific goals than with making the choice to not pursue any specific goals.
--:-- I don't really know what the deeper, darker meaning of this pursuit of optionality is. I know a lot of people talk about it explicitly and implicitly. It seems like this search for optionality takes place on a kind of fractal level. Not only do people want to choose a job with lots of different exit opportunities, but they'll try to make their plans to have lots of different plans so that if they decide they don't feel like going out this night, they have an option to go out the next night. You just get less done if you have lots of opportunities and you could easily forego any one of them. There just aren't that many people who succeeded in a memorable way because they kept all of their options open.
--:-- The closest to that you can get is that there have been a number of people who made their money in smart deals where they did manage to control the downside. But to take the optionality example, that is more like being a structured products trader or an options trader who is trying to construct a pretty elaborate payoff structure where you get to control your losses on one side and maximize your payoff on some other side. That is a lot more complicated than just optionality. You're not buying it wholesale. You're actually picking and choosing which options you care about and which ones you don't. That is something worth keeping in mind, and maybe I didn't emphasize enough in the essay that I'm not saying you should always choose the minimum optionality choice. You could read that essay and immediately sign up for the military or join a religious order and have no more choices in your life for a long period. I'm not saying you should immediately seize the opportunity to give up options. I'm saying that you should be very judicious about them. Going back to the markets example, if markets are not perfectly efficient, but they are pretty efficient, you should expect that you can get a superior payoff, but only if you are actually doing pretty good work and only if you're doing something you're actually pretty good at.
--:-- Dwarkesh Patel: Yeah. Some of the risk aversion in society can be explained by optionality alone, not just conformity. One of the examples I like is: think about somebody who becomes a dentist. It takes like 12 years or something, $500,000 or more in student loans. It's a very specific thing to do. You're getting rid of options. What if you just spend those 12 years and that much money trying to start 12 businesses, one each year, see if it fails, and you just move on? In some sense, that choice has more optionality, but people are still less willing to do that for some reason.
--:-- You mentioned that people trying to get accredited by elite universities—there's a group of people who are doing cool things but aren't trying to get that accreditation. Are people acting irrationally by trying to get accredited from elite universities? I mean, you got a well-accredited credential just by being a very excellent SEO writer. In some sense, maybe more people should be trying to do that, right? You're getting a better credential by doing that.
--:-- Byrne Hobart: Yeah, I still view elite higher education as probably a pretty good deal for the individuals who are doing it. I just think it's a bad deal for society that so many people are pursuing it. If you have the opportunity to go to a great school and you're considering just not doing that and figuring out what to do instead, you should probably just go to the school. But if you are going to a really good school and you see a particular opportunity that you want to seize right now and you know that entails dropping out, that is when you should not go, or at least drop out. You still have optionality there. I think last time I looked at this, both Mark Zuckerberg and Bill Gates are technically on some kind of hiatus, sabbatical thing from Harvard. They could just enroll next semester and start taking classes again if they wanted to finish. That is an example of just how far the preference for optionality goes. Even once you get into a good school, they actually won't let you drop out by default. You have to ask them if when you leave, they will not let you back in, or they just will.
--:-- In general, going to a good school is a good credential. It's hard for someone who is 22 years old to do something that they expect to succeed in that is more impressive than getting a bachelor's degree from Harvard. You can do things that are more impressive than that, but your expected outcome is probably lower. Your median outcome is lower. On the other hand, if you do succeed in something else and you also turn down the easy prestige, the easy credential, then that's even more impressive. It's not just externally. I think there's something really important internally about having this experience of having an easy way and choosing to do something else, even though it's much more risky. You know that you have slain the optionality dragon and its other heads are going to grow back at some point, but you have actually fought it and won, and that's going to stick with you.
--:-- Dwarkesh Patel: One of my previous guests, Scott Young, he's famous for doing something called the MIT challenge. Instead of going to MIT, he just used their open courseware. MIT has all their classes online. He said, "I'm going to get a four-year bachelor's education in computer science from MIT by just doing the online coursework. If I get above an F on the final exam, which I self-administer, then I consider myself passing the class." He did this. In some sense, given that it was truncated, he did less work than getting an MIT education would be, but he got tons of job offers from that. Maybe people just aren't creative enough in finding a way to get a good signal to employers. It's not an efficient market. You wouldn't expect 19-year-olds to pick up $100 bills on the floor.
--:-- Byrne Hobart: Yeah, I think one thing to note on that is that a lot of people who try that will find that they've really overestimated their willpower. I think this is actually one of the functions that a school like MIT serves: to consistently motivate you both by surrounding you with peers and by giving you professors who are very smart and whose judgment you respect. If your professor tells you this is C minus work, you actually feel bad and you try to do better. With yourself, it's a lot harder to motivate yourself. You can do it, but you will often have to pick your own topics to study. And then there's no credential. It's not like, "I could have gotten this credential, but I didn't go to this particular school." It's more like, "This is not really a thing, but it's a thing I'm actually really good at." So you have to both do the work and then sell people on treating that work as credible.
--:-- Dwarkesh Patel: Speaking of that, you wrote an essay about how introverts will be more influential in the future. If you think about the Big Five personality traits, introversion is one that will be more valuable. Tyler Cowen makes this point in *Average is Over* that he expects conscientiousness to become a highly valuable trait in the future. In that essay, I think you basically said that conscientiousness might become less important, or at least you didn't emphasize it. Given that you can get more returns by doing clever things and scaling them up to a higher level, maybe conscientiousness is a more important trait in the new economy?
--:-- Byrne Hobart: Conscientiousness is really important if you have goals that are set by somebody else, and it's really what lets you achieve those goals. I would rate myself as not very conscientious at all, because I'm pretty bad at fulfilling totally external goals—almost deliberately bad at it. But I do work a lot on things that I'm interested in. This may be almost a terminological difference on what conscientiousness is. I would view it as: if someone asks you to do something and it's some arbitrary thing that they select, are you likely to get it done? People have low expectations for me on many of those things. But if conscientiousness is just "are you willing to put in effort to accomplish something," then that's something that I am willing to do.
--:-- My view, and this is kind of an *Average is Over* view, is that as the cost of communication gets lower, as transaction costs decline, and as more of the economy gets mediated through software, there are just going to be more specialized jobs. If you are really interested in one of them and you would do it for free, then you don't need to be especially conscientious to succeed in it. There's another side to the conscientiousness thing. Another reason you might think conscientiousness would pay better is that entertainment is getting so much better, more addictive, and better at marketing itself to you. Just being able to resist the siren song of Netflix and video games is becoming a more valuable skill. But there are times when I will just be messing around and waste my time, and I'll realize that there's something more interesting I could be working on that would actually produce something tangible that subscribers would pay for and that I would be proud of. So I hit Apple-Q and get back to work.
--:-- Dwarkesh Patel: That's surprising that you wouldn't consider yourself especially conscientious. The distinction makes sense, but given the fact that you produce a lengthy *Loose Ends* every day... The argument that Tyler Cowen makes is about self-education. Another point he makes is that businesses are bigger now. You have global supply chains. If you have a person who is kind of a rebel without a cause, or even a rebel with a cause, just being unreliable and volatile makes you much more dangerous than you would have been in the past because you can mess up more things. I don't exactly know what that means because I've never interacted with global supply chains myself, but that's just what I remember.
--:-- Byrne Hobart: Well, I wonder about that. I suspect that some of the people at Apple who are managing things like their response to COVID are doing this in part because it's a really fun challenge. I'm sure they're paid well, and that's a big motivator too, but it actually sounds kind of like a fun puzzle. If Shenzhen shuts down and most of your manufacturing is close to Shenzhen, what do you do right now? What do you do in a week? What do you do in a month? And by the way, what is your plan for making sure that we never have this problem again? There could be some general appeal. *Factorio* is a very popular video game, and the whole point of it is managing a complicated supply chain and managing it around both passively increasing production and actively dealing with various crises. I think there are people who are really into that. Maybe they're all playing *Factorio* and not actually getting jobs at Apple, but there's some fun to that. I suspect they're actually pretty proud of what they do. Maybe it's not fun in the moment, but I suspect that when a lot of the Apple supply chain people look back on the first calendar quarter of 2020, they think of that as a really, really cool experience. Not a fun one, but a gratifying one in retrospect.
--:-- Dwarkesh Patel: Interesting. You wrote an article saying that investing is the "rationality dojo." I'm curious why the subculture of rationality isn't more centered around finance than it is around programming. Maybe debugging is also a rationality dojo, but it would seem that if this community were going to be centered in some industry, it would be finance.
--:-- Byrne Hobart: That's a good question. I have noticed—I would hang out with some of the rationalists in New York, and a lot of them are in finance. But then again, I'm hanging out with people in New York, so of course a lot of them are going to be in finance. I think the rationalist subculture does skew a little more to finance than is representative. But you're right, it's definitely more of a programming thing.
--:-- One possibility is that the rationalists talk about how the center of their subculture was LessWrong. They're clearly not saying "we're going to be perfectly right." They're saying "we're going to identify these flaws one by one and improve them over time." They're not trying to reach some kind of perfect understanding. Maybe this approach of believing that you can solve a lot of these problems, or believing that you can improve on a lot of these problems, is just more of a programmer kind of attitude. I used to program. I don't really do it anymore. It is very much an exercise in rationality, and it's also a more introverted one.
--:-- In finance, part of what you're doing is trying to model other people's behavior. You're trying to figure out not just "why am I right," but "why is the person on the other side of the trade wrong?" That can be sometimes more valuable than knowing you're right, especially if they're wrong in the sense that they have some kind of balance sheet constraint or regulatory constraint that keeps them from doing the obvious reasonable thing. One example of that would be that small cap stocks tend to have less efficient pricing than large stocks. A lot of financial institutions have a size cutoff where they tell analysts, "Please don't look at anything that's worth under $5 billion because we won't be able to take a meaningful position, so it's not worth your time." Whereas you can take a very large position in Google or Amazon, and if it moves 1%, that is more money than you make in the tiny position of the tiny stock. There is a market inefficiency there where a lot of the talent is focused on getting more precise valuations for big companies rather than finding pretty obvious opportunities in small ones.
--:-- Debugging as a rationality exercise is interesting because when you debug, you're trying to outsmart yourself. You're looking at a situation where you were... I forget what the quote is. There's some line about cleverness in code: "Debugging is twice as hard as writing the code in the first place. So if you write the cleverest code you can, then by definition, you are not smart enough to figure out what's wrong with it." That seems largely true. It is an exercise in humility and also in self-knowledge. Maybe finance is the extrovert rationality dojo and debugging is the introvert rationality dojo. Although all the finance people I knew in LessWrong-ish circles were much more likely to be quants. I don't think there was a single investment banker in the entire crowd.
--:-- Dwarkesh Patel: You wrote an article about how rationality saw COVID coming much more so than other subcultures. One of the things you wrote, and I really didn't know what this meant, so I wanted to get a clarification. You said: "It's this attachment to the factual consequences of predictions and detachment from their social consequences that makes rationality relatively good at predicting the future, but comparatively worse at influencing it. They can spend political capital in cost-effective ways, but earning it often requires perversely counter-rational decisions." What did you mean by that?
--:-- Byrne Hobart: Part of this—and this is not something the rationalist subculture strictly advocates, but it is a kind of cultural norm—is just being honest and direct and blunt. It's really refreshing that people will just tell you what they're thinking instead of trying to coddle you or be more polite about it. But it does mean that they are doing something that is orthogonal to politics. If it has a political implication, it's usually that they are making something that in political terms is a mistake.
--:-- It's hard to articulate how exactly this played out with COVID, but I did get the sense that part of the problem is they take ideas seriously. They extrapolate things to their natural endpoint and they're willing to bite a lot of bullets. They're willing to say that if something is weird but factually true, then they believe it. They're also willing to loudly affirm it, which is nice and refreshing. A lot of people just keep their weird beliefs to themselves. Not so with the rationalists. But it also means that the population that was warning about COVID early is also the set of people who are talking about, I don't know, Bitcoin is going to save the world, cryogenics is really important, we live in a simulation. They have a lot of weird beliefs. Some of these beliefs are probably true. Some of them are probably false. They are generally beliefs that make sense if you accept a certain set of premises, and they don't if you don't.
--:-- Believing that COVID was going to be a big problem made perfect sense if you accepted that it has an R0 of two or higher. At the time, we thought the case fatality rate was about 2%. All you had to do was start compounding some numbers in your head and you quickly realized: okay, this is going to be everywhere and it's going to get a lot of people sick. If it broke out in a city right after the Chinese New Year and the city got shut down, that means a bunch of people got exposed and then went all over China. Since flights are still going between China and the U.S. and China and Europe, it's going to be everywhere. They were not making any big inferential leaps.
--:-- But because they've also said things like, "If you think it's theoretically possible that a computer can simulate what feels like the real world, and if you think that there's some chance of that happening, and if you know how computers work, then you think that given some non-zero chance of it happening, the odds of you living in the real world versus a simulated world are vanishingly small." That assumes that there's only one real world, but it's also something you derive from taking ideas seriously. Or with cryogenics: if you think that there is value to living and continuing to live, and you think that there's some chance that you can be brought back to life if your head is frozen, you will do it. Given the upside, it's actually kind of cheap, especially if you do the life insurance thing to pay for it. They're being very straightforwardly reasonable, but it means that anyone who was in the rationalist community and worrying about COVID, you could immediately link it to other crazy things that they had told people that everyone just thought were absurd. The rationalist will argue that saying something is absurd is not the same as saying it's false. It's just saying that if this is true, it's really important. But the normal heuristic is: this is absurd, therefore false, and therefore you are low status for believing in it.
--:-- Dwarkesh Patel: In your essay "Read More," you say that you're more likely to get valuable content that's fitted to your niche because writing has a lower production cost. It's easier for people to produce good content. If you extend this logic, Twitter has an even lower production cost than a book. So should I be reading tweets? Is there a golden spot of fixed costs that a newsletter or a book has that a movie doesn't have on one end, but a tweet doesn't have on the other?
--:-- Byrne Hobart: It is true that if you extrapolate that part of the argument, you do conclude that tweets are good and that watching live streams is even better. There are different kinds of content and there are different production costs. What I was focused on was the kind of content where someone is trying to deeply understand a topic and explain it. In that kind of content, it's hard to do that in a tweet. You can definitely tweet something that is koanic and insightful, but the hit rate on those is pretty low.
--:-- It's partly a comparison of learning history or math or learning about a new technology from watching YouTube videos versus learning about it by reading. My argument was that there's a long tail of things where it is actually worth the time and effort to make a book, and it's not worth the time and effort to make a documentary. You can think of it in terms of hours of research required, hours of production required, versus hours of content consumption. Making a documentary requires a lot of people to spend a lot of time, and they're going to cut it down to an hour. With a book, you will usually spend more time reading a book than you would spend watching a movie. There are exceptions on both ends, but that's a generalization. The production time is just typing, and that can happen pretty quickly. There's a lot of reading that goes into it, and that can take an arbitrarily long amount of time depending on how thorough you want to be.
--:-- That reading process is really a process of winnowing ideas down. I'm working on a book with a friend, and I've increasingly come to this view that the writing process happens while I'm reading. What I'm actually doing when I sit down and start typing is more like taking dictation. I have the general thoughts in my head, the general connections between different topics. I've highlighted the things I really want to talk about in the book, and I'm just applying a pretty coherent narrative to it. That part's not super hard. The book is going to be mostly effort in reading and learning, and a relatively small fraction of that is actually producing the end artifact. Whereas with a movie, you can imagine a documentary where there are more person hours spent on the production process—filming it, editing it, managing all the equipment, getting in all the travel—that could actually end up taking more time than the preparatory research to come up with the idea and figure out what the narrative will be. Isn't that the logic there?
--:-- Dwarkesh Patel: Most of the time that goes into the podcast for me is just me preparing and researching my guests. Is the logic there that podcasts are as valid a medium of consuming content?
--:-- Byrne Hobart: Well, podcasts have a totally different dynamic because there's more back and forth. A book is more analogous to a lecture. A lot of documentaries and YouTube tutorials are also basically a lecture. The podcast is a dialogue. There can be tangents, and both sides are contributing things. One person will say something that leads to an interesting question, leads to another tangent, leads to something else. It's creating more connections.
--:-- We both have different clusters of ideas in our heads and different things that we want to talk about. Because they're different clusters, there's some overlap and then some totally separate things. The connections that can happen between ideas in dialogue are going to be a lot broader than the connections that could happen with one person writing a book. But a book will probably get more depth. It's a tradeoff that the different media have.
--:-- It would be hard to do a podcast where you're talking to one person and you're trying to develop a theory of history or decide if inflation is transitory or not. You could have a really interesting discussion and lead to a lot of ideas. But if you wanted to really answer the question of whether we're going to have hyperinflation, you probably want to be writing a book making the case yes or no.
--:-- Dwarkesh Patel: Yeah, that makes sense. A related concept that you've talked about is the convexity of knowledge. There's increasing returns to learning more. Doesn't that imply that the next book you're going to read, you're going to derive more value from it than the next book I'm going to read? It would seem to me pretty clear that I'm probably going to get a lot more value just because I know a lot less. Or maybe this is the wrong way of thinking about it—you gain more knowledge, but since you already have so much, knowledge is worth less to you, so utility is lower.
--:-- Byrne Hobart: We're trying to apply utility functions to things that are hard to quantify in the first place. What I would say is that it's not just about reading the book, it's also about retaining it. That's really where the convexity kicks in. There are things that make more sense in context when you have more context. They also make the context more memorable and make it have more sense.
--:-- Paul Graham actually wrote about this a long time ago, talking about learning history. When you read one history book, you just have a series of dates and names and whatever connections the historian is drawing. There's no broader context that you're putting it in. But when you start to read multiple books about the same time period—maybe one is a political history, one is a biography of one particular person, one is a history of technology changes at that time—you start to see different things connect together.
--:-- One of the things that makes it really memorable is when you see things that you know were important that someone missed. You're actually having a little dialogue with the author. I read a biography of Deng Xiaoping. One of the striking things was that it mentions container ships once, and it's a totally offhand mention—a Chinese delegation went to a German port city and saw containers being loaded on ships. But there have been other things I've read that argued that China's industrialization was massively driven by containerization because it meant you could import intermediate goods, process them, export intermediate goods. You could have these really complicated supply chains because the cost of shipping things on one more hop was a lot lower. It meant that countries with cheap labor had more ways to start slotting themselves into the supply chain.
--:-- The book was still really good, but it was more focused on political history and internal Communist Party deliberations rather than the broad economic sweep. It was memorable to me because I felt like the author knew a lot more than me about China, and there was something I knew about China that the author did not know. That's a really fun feeling, and it made the rest of the book more memorable.
--:-- Dwarkesh Patel: I'm glad you brought that up because it's relevant to a question I want to ask. Consider that I have zero knowledge in comparison to you, which is almost close to true. If you're recommending what kinds of books to read—take that example. Maybe I can read *The Box* about shipping, which is very contemporary and specific. Maybe I can read a biography of a contemporary figure like Deng Xiaoping. Maybe I can read an ancient Chinese classic. Maybe I can read *The Three Body Problem* to understand Chinese fiction. If I'm trying to understand a topic like China, which of those categories is the highest utility, and which are supplementary texts?
--:-- Byrne Hobart: I don't have a good answer. What I usually do is buy a bunch of somewhat random books. It's partly a convexity thing. There are things I'd heard of and knew I would read at some point. When I decided to read about a topic, that's the time to read them. It's hard to give a really good answer other than to try to buy a smattering of different books that take different angles on the same topic.
--:-- Fiction that is written in the time period you're interested in, or that is roughly about the topic, can also be really effective. Authors have an eye for details that other people will miss. The specific tangible things that show up in fiction that don't show up elsewhere add a lot more texture. You should probably do all of the above. You should just have a giant stack of books on whatever topic you're getting into right now.
--:-- Dwarkesh Patel: Gotcha. My favorite essay of yours was the one on the middle income trap, where you compare the process by which countries get stuck in a second-class tier and how a career gets stuck in that place. It's directly relevant to the considerations I have. In the book *How Asia Works*, the author talks about how many of these countries created tariffs to generate internal knowledge about how to produce highly valuable things that were really shitty at first. Is there an analogous strategy where you learn to do things that are not your comparative advantage—you're really shitty at them, but the goal is that this eventually differentiates you? Or is there a better strategy to escape the middle income trap?
--:-- Byrne Hobart: I definitely think that picking up skills that are adjacent to what you do and that you're not good at is really valuable. If nothing else, if you're successful, you will end up delegating some of those tasks to other people. It is very helpful to know roughly how hard they are rather than just having to guess. If you have a job where you're not doing any programming, but you're working with programmers, you should probably learn to program even badly just so you can have some intelligent opinions and know which timelines are totally fictional and which ones are not. It gives you ways to ask better questions.
--:-- One of the escapes from the middle income trap that I think Stilwell talks about is that countries either built branded products that were popular around the world—they could export and sell at a premium price—or they developed indigenous technologies that other places could not match. That's sometimes a process that bounces back and forth. There was a time when it was really hard for U.S. automakers to compete with Japanese automakers. Everyone seems to be a lot closer to parity right now. It was a temporary competitive advantage, but it was a big one for Japan, and it certainly helped their economy continue to grow.
--:-- Then there are all these intermediate industrial goods that Japan produces. There are a lot of supply chains that happen to pass through Japan. The final assembly is in somewhere like China or Vietnam, and the end product is sold in the US or Europe, but there's some essential component that is still only made by a Japanese company. It's hard to match anywhere else. There's a lot of stuff in the semiconductor supply chain where there are 100 different things you have to do to go from silicon to a chip that actually works. Sometimes one or two or five or ten of those steps will be monopolized by some company in Japan.
--:-- That is the escape: to have some set of skills that you are pretty confident nobody can match and that people know you have. That allows you to compete on something other than price. Competing on price is fine when you're young because your cost of living is low and you'll learn a lot by getting thrown into jobs that you were only qualified for in the sense that you were far cheaper than everyone else. But you can't get rich being cheaper than average to hire. At some point, you have to do something different.
--:-- Dwarkesh Patel: Gotcha. My final question is usually: what is one piece of advice you would give to somebody my age? We already talked about some things, so this might be redundant. We talked about avoiding that rationality, reading more, differentiating yourself to avoid the middle income trap. Is there anything else you would recommend?
--:-- Byrne Hobart: A synthesis of a lot of those things is that because books are cheap and the internet is such a wonderful distribution mechanism, it is really not hard if you are determined to be—if you pick a narrow enough topic—to be close to one of the world's leading experts on it in a fairly short time frame. You have to pick a topic that is narrow enough that nobody has written a dissertation about it, which can be hard. The next best thing is you can be the world's public-facing expert on it by reading a bunch of academic papers from people who know more than you, diving through what they cite, reading that stuff too, and consolidating it into something you can put on Substack or somewhere.
--:-- It's a valuable exercise both because it gives you a body of work you could point to and say, "I am willing to work hard on projects. I'm able to find interesting and insightful things and able to learn a lot and synthesize it in a way that helps other people learn too." But it's also useful because one of the things I learned from these various deep dives I've done on different industries, countries, and companies is that the world is just much more complicated than I thought. It's very easy to say that and very hard to internalize it. The only way to internalize it is to realize repeatedly that you were wrong about something because you had oversimplified it.
--:-- Often, it's not just that you were wrong—it's that the first source you read had some theory, and that theory was later overturned. You want to get to the point where you can look back three months and realize you were dumb about something you thought you knew a whole lot about. At some point, when you're no longer realizing you're dumb but starting to just have more open-ended questions that you can't really get answers to, that's a good time to move on to the next big topic.
--:-- There's so much material out there. There are lots of aspects of economic history that I think are worth revisiting right now because we're either at a time that is strange because growth has slowed down and it's going to keep being slow forever, which hasn't really happened before without some kind of big natural disaster. We basically had growth that continuously accelerated, albeit at a slow pace, from pretty much the dawn of time through the 1970s. Then it slowed down. Different countries have had different ebbs and flows of growth, but overall that's been the story: gradual growth, gradual acceleration.
--:-- So we're either at a unique time period because that's stopping. Total factor productivity grew and accelerated for a while, and then it stopped growing so much. That's interesting—it'd be interesting to figure out why that is. Or we're at an interesting time in history because there will be specific technologies that actually do accelerate productivity growth. It's really worth it to know what those are. If you go back and look at the specific technologies that led the first industrial revolution, the second industrial revolution, or the information technology revolution, they are all associated with people making lots and lots of money. Even the people who didn't get extremely rich had really interesting lives. So it is very worth figuring out if we're in year 10 of another one of those 50-year deployment cycles.
--:-- Dwarkesh Patel: Final question. I kind of asked this earlier when I asked what kinds of books to read, but when you're trying to figure out the answer to that question—which I'm very much interested in—other than reading your newsletter, should you be doing a deep dive into how something in the world works today? Should you be trying to understand history? What is the most relevant lens to be looking at here?
--:-- Byrne Hobart: History is a lot easier because we've seen how things played out. If you're looking at contemporary trends, you have a couple of problems. One is that a lot of people writing about them have an agenda because they either want something to happen or they don't want it to happen. There's also the meta agenda thing where the people who are really busy making it happen don't have time to write books, and the people who got left behind one way or another do have time. There are a lot of selection effects that make the data noisier.
--:-- On the other hand, there's a lot more data. If you go back and look at earlier historical events, you may not be able to find much at all, or you'll find a handful of sources and you've exhausted your material. So you do have that trade-off. I go back and forth. It's not a deliberate thing, but I've noticed over time that I will generally spend some time writing about contemporary things and some time writing about historical analogies to those things.
--:-- Right now, I'm doing a bunch of reading on central banks, how those came about. The original reason was just trying to figure out some questions I had about quantitative easing for which I don't have good answers. I'm going back to the beginning and reading about how the Fed was put together, all the debates around that, what the financial system looked like without central banks, and how central banks evolved over time in their powers, mandate, and institutional cultures. That is very much a present-driven but history-focused project.
--:-- Maybe there's not a good answer. History is still happening, so you're technically reading history if you read something that was written yesterday. If you read it with a pseudo-historian's perspective—trying to figure out what people will think about this in 50 or 100 years when the debate is largely settled, or at least when they're still debating this in history departments, what will be the main schools of thought and what will be the strengths and weaknesses of those schools of thought—that is probably the right attitude to have for current events.
--:-- Dwarkesh Patel: Awesome. Byrne, thanks so much for your time.
--:-- Byrne Hobart: Anytime. Thanks.
