Why Bubbles are Good | Byrne Hobart

What we talk about in the book is no, there are actually some things where they don't just get built sooner because there was a bubble. The only reason they exist was a bubble. The only reason that they exist was that for a brief period in time, so many people were so confident about some very specific change in the way the world works, some deployment of particular technologies or changes in organizational structure. They were so confident in this that everyone knew that if they make a bet that only makes sense if this actually happens, that bet has a much higher chance of paying off. But only if you make that bet right now.

Welcome to Bankless, where today we explore the frontier of why bubbles are good. This is particularly relevant in crypto, I would say. This is how to get started, how to get better, and how to front run the opportunity. I'm Ryan Sean Adams. I'm here with David Hoffman, and we're here to help you become bankless. Our guest today wrote a book on why bubbles are good. Basically, the thesis is we are in an era of great stagnation. Productivity, the economy, our culture of risk aversion has made it such that we are underperforming relative to where we need to be. And he says bubbles, speculative booms are actually the way up. A lot of material here that I think is super relevant to crypto.

I think this audience is going to be a little bit more accepting of some of the theses uh presented by Byr in his book, the idea that bubbles are actually conduits for progress. Not all bubbles, but some bubbles. And I think we all live through these experiences going through the many crypto cycles. It's weird to live through one bubble, I would say. But in crypto, we live through bubbles every four years. So we are intimately aware of like how these things work and what they do and how they impact progress. But Byrne also advocates for like, well, not all bubbles are good. Some bubbles are actually downstream of the fact that society is stagnating. We didn't get to talk about it in the episode with Byrne, but there's a website called WTF Happened in 1971. And I think Byrne is approaching the same idea from a different perspective. It's like society has really started to stagnate. In his book, he gives three reasons, but I'll just leave that for the book and for the listener. So just a really interesting conversation connecting the idea of bubbles as a lens for progress while also society becoming more risk averse. And maybe that kind of illustrates the social gap between crypto people and the rest of society that I think we all kind of know exists.

Byrne also has some really good insights in crypto. He was in crypto before both you and I were, David. So he's got some takes from the very early eras, as well as one of my favorite parts was kind of where we concluded in finding the next bubble and positioning ourselves to take advantage of that. Byrne actually says FOMO is good. You should embrace the FOMO. And so you'll have to hear what he means there on that. Guys, we'll get right to the episode. Bangless Nation, very excited to introduce you to Bern Hobart. He's a financial analyst. He's a writer. He's well known for his newsletter. It's called The Diff. I love this. It's one of my favorite publications on the internet right now. Covers tech, trends, economics. It's like perfect for the well rounded generalist. I'm a bit more of a crypto native, but when I want to be a generalist, I go read the diff. He's written this book along with his co author. It's a book called Boom. We're going to talk about that today. Byrne, welcome to Bankless.

Hey, great to be here.

So I think we're going to spend most of this podcast talking about your book, talking about the thesis of it, because that's super interesting, very applicable to crypto. Maybe you could tell us why crypto bubbles are good in the course of today's episode. But before we do, I want to get maybe your crypto background. You have some crypto pedigree, I believe. Am I understanding correctly that once upon a time you worked for Balaji Srinivasan back in, you know, before let's say the bankless podcast existed, sometime in the 2015 time range? Can you tell us your background on crypto, like what you've done inside of that?

Yeah, yeah, absolutely. So I had paid kind of vague attention to crypto adjusting things for a really long time.

I was actually looking back at some of Nick Sabo's blog posts on Bitgold and realized that I'd actually left comments in like the 2008 timeframe. And the specific thing that I was commenting about was I was complaining that if you have a bunch of computers that are all competing to solve puzzles and that's how you allocate your currency, well, that just leads to this really wasteful, you know, race to spend as much compute as possible, et cetera. So yeah, I was a little bit early to the crypto mining environmental impact thesis. And I guess one of the things I hadn't thought about is, you know, there's there's an environmental impact to other stuff and you know to the fiat system as well. Although I think that that argument is also it is a little bit fuzzy. Like you still, it's not like the need for all security goes away, and like it's not like the demand for physical infrastructure that supports abstract financial infrastructure all goes away, et cetera. So like I still think it's an interesting question to ask whether or not that's worth it, etc. But yeah, so I've been paying attention to crypto for a while. I did indeed work at 21.co, which was just a really, really fun experience. Like lots of incredibly smart people who are looking at this technology and asking, saying basically, we read the white paper and fell in love, and now we should figure out how this actually gets deployed widely and how it gets used and.

I think through all that stuff. And so like the company, it was this combination of like economic think tank and technological think tank and also sort of growth hacking lab of trying to figure out all the different ways to extend the crypto ecosystem. I was not there for an incredibly long time. So I did miss some of the story, but I did see a lot of stuff in my short tenure there. And before that, I'd worked at a hedge fund and then decided to move back to New York and ended up working at some companies that do research, like data intensive research for hedge funds. So I guess I've kind of straddled the tech and finance worlds for a while.

Maybe we'll talk about crypto in Bitcoin a little bit later because there is an entire chapter in your book, you know, talking about Bitcoin and talking about it from the perspective of why the Bitcoin bubble and I guess repeated bubbles have been actually good, have been one of the things that you point to as kind of the solution, the way out of our stagnation. I want to ask you this question too, while we're on the subject. Did you get Bitcoin the very first time you heard it? Or like how long did it take to click for you? Because a lot of people, you know, miss it the first time or two, and it takes repeated exposure to really have it sink in. At what moment did you recognize Bitcoin for the kind of the innovation and the productive bubble that it actually is?

So I think there was a period, and I guess this ties into the bubble question because there was a period where I thought this is a really interesting idea, and it's very cool that you can create the design for a decentralized currency. And then the next question is how do you make it an actual currency? Because if you have a system where all the rails are in place and you can transfer these tokens around and anyone can participate and you could be very, very certain that you don't need some central trusted intermediary, like you have total control, et cetera. Like all of that is really, really cool. And then the question is, what are these worth and what can you buy with them? And so for a while I thought it was like an interesting proof of concept. And I guess my basic attitude in retrospect was if somebody does something with this, that would be pretty neat. And what I didn't realize was that the value could be just an emergent property of people interacting on the network. First, they're sort of sending crypto around for fun and a proof of concept, and then you have the famous pizza transaction. But what the pizza transaction at least did was it set some kind of anchor where there is some kind of real exchange rate and you are actually using the crypto system to interact with the rest of the economy in semi-useful way. Like obviously, the retrospective dollar cost of that pizza purchase was really, really high. But also the transaction cost of just everyone had to figure out okay, how do I actually send someone this money? Like, do I have the address right? You know, lots of nerve wracking things. And there has been this process where, on one end, the actual tokens get more valuable, which means there are more transactions that you could conceive of doing with them. But also, there's just a much better interface for actually transacting with tokens. So I didn't buy any crypto for a long time. And actually, the first time I bought any was when I realized that Y Combinator had funded Coinbase, and I thought, look.

The crypto ecosystem is going to select for people who either are like really technological, but maybe don't know a lot about economics or like have very peculiar ideas about economics, or people who are just very ideological, like people who are libertarians and archo-capitalists, like they like the idea that this is money without central banking. And so that's why they're going to do it. And also, if you have a new bearer instrument that can be easily hidden, easily transferred, etc., it's going to attract a lot of criminals. And I felt like if I'm just some guy on my laptop looking for a way to buy Bitcoin, the criminals are looking for me. And that it's going to be very, very straightforward for me to transfer the fiat money to someone. And then will I actually get crypto? Maybe, maybe not. So yeah, for a while, I just I didn't look into it because I just wasn't sure that there was any straightforward way for me to buy any. And then once Coinbase existed, my basic thought process was why Combinator will be humiliated if this is a scam. They probably have the resources to make customers whole. Like I think at the time that I signed up for Coinbase, the daily transfer limit was 100 USD, at least for me.

So I was like, you know, they can't take that much of my money. So yeah, I bought a little bit then and then kind of forgot about it. And then I actually was so this was like late 2012. I bought some and then I was at a hedge fund. And at one point someone in the hedge fund sent this email around saying, what is Bitcoin? And he had this chart. And the chart was a year-to-date chart where it started around the price that I'd paid and then it had gone up like 10x in the ensuing couple months. And so uh that was pretty exciting. That was pretty fun. Like it was an immaterial amount of money, but it was still cool to see number go up. And so yeah, I spent some more time on it. There was a lot of drama in the crypto world. So 2012 to 2013, you started to have price appreciation. You had the whole Silk Road news cycle. And you know, Silk Road was it was a proof of concept of people will actually use this. They will actually use this to participate in the economy. It turned out it was participating in not the illicit legal economy, a different part of the economy, but still, and it was actually a use case where you definitely don't trust the person on the other side of that transaction. Like that whole industry, the whole supply chain is defined by the fact that no one trusts anyone and there's no recourse other than just, you know, writing off your losses or doing something really horrible. Like there, you can't go to court over a bad drug deal. And so Bitcoin was actually a pretty good tool for that. I think the US dollar, like physical dollars, are still a better tool if you are in the drug business. There's still a way better way to transfer money around, preserve your purchasing power. And of course, there's not an audit trail for where did this particular hundred dollar bill go? Where has it been, etc.? Whereas with Bitcoin, yeah, you can audit it right back to uh when it was first mined. So yeah, I'd been paying attention to it for a while. And that I think, like looking at that evolution, I think one way to look at it was you could say, okay, someone invented this technology, it was a concept looking for a use case. Use case turned out to be buying and selling illegal drugs. US government does not like that, they're not going to let people do that. Therefore, it was a fun experiment, and then that's the end. But the other hypothesis was well, no, there's more of a liquid market here, and people are actually using this. And you know, it went down. I remember looking at the intraday chart, the day that Silk Road got seized and seeing, wait, it went down, but it didn't go to zero. It wasn't like the only reason people owned crypto was to buy and sell drugs, even though, like, you can look back at really, really early charts of Bitcoin. And I think there's there's one year where you can actually see this rally coming into Burning Man, and then that crypto actually dropped after Burning Man. Yeah, which that also is that is like one of the indicators of just it's a pretty early, pretty primitive financial system. Like real world swings in supply and demand for particular products actually affect the exchange rate. And there are real world antecedents to that. So there's this story I've never bothered to check because it's too good to check, but there was apparently a weather vein in front of the Bank of England, and they knew that if the wind is blowing in a way that is favorable to ships coming in, that you actually need to inject a little bit more liquidity in the financial system. And then when the wind is blowing the other way, fewer cargo is coming in, so you don't actually need as much liquidity in the system. So this is a case where literally how strong the wind is blowing is determining the day-to-day trade balance between, say, England and the Netherlands, and that tells the central bank what to do. And since it's a central bank, you know, under a more fiat system, they can just inject liquidity into the local financial system. And crypto at the time did not have that. And that was another thing that just looking at the evolution of crypto really shocked me was so we're zooming way ahead now, but in March of 2020, I had actually in 2019, I'd been doing some more quantity research into the question of what kind of asset is Bitcoin? Like at that point, we'd established.

The crypto is an asset class. Bitcoin is kind of the blue chip of this asset class. And people incorporate it into financial portfolios that have other just normal person financial assets like treasury bonds and SP at index funds and whatever. And so I was asking myself, like, is this

In one sense, it's a risk-on asset. Like it's a cool technology thing. And so maybe it's like a hyper-levered version of the NASDAQ. And when people want to take risk, they put some more money into NASDAQ and way more money into crypto. And then my other hypothesis was no, this is an asset where you have a fixed supply, it's easy to move around. And so you know that pretty much no matter how bad things get, as long as you and your loved ones and a USB drive can get exfiltrated from the country, then like some of your assets are safe. They're as safe as they're going to be. And so maybe crypto is actually the most apocalyptic asset, and you should compare it to things like gold and the Swiss franc and the yen and other things where when people panic, that's what they end up buying more of. And so I looked at a bunch of correlations and what I found was just noise. Like there, there wasn't anything it really correlated with. There wasn't any way to really make sense of the price movements in light of other stuff. And then I kind of set that aside as okay, that's my understanding of crypto in a financial context, is that it's this uncorrelated thing. And if you think it's going to be a major reserve asset, then the risk-adjusted returns, you know, if you think like 10 years from now it's going to be part of a lot of the FX reserves of a lot of different countries, and maybe it starts to supplant gold, then it's a pretty good deal. And if you don't think so, then it's not. And you know, I looked at just what are the risk-adjusted returns that are different scenarios for timelines and probability of that happening, etc. And then March of 2020 happens and Bitcoin crashes along with everything else. And I was just really confused. I was like, why is it crashing? Like, shouldn't you be panicking and selling everything else you own and putting that money in crypto so you can fly off to like, you know, be the last flight into New Zealand or smuggle yourself into some isolated place where you won't get sick. And then, you know, when things stabilized, you re-emerge. So I'd kind of thought maybe crypto goes up during the most apocalyptic scenario and it went way down. And that's when I realized that no, there's actually a lot of leverage in the crypto ecosystem. Like, this is already like the most volatile asset class that is an actual asset class and not a derivative on something else. And what do people do? They wanted to borrow against it and make it even more volatile.

So I thought that was actually a notable evolution because when you look at the historical gold standard system, they're just not a lot of times and places where people are transacting primarily in physical gold. What's usually happening is that there are financial institutions, they have gold reserves, those reserves are used to balance long-term trade and investment flows between countries. But what people transact with day to day is some kind of token that is implicitly backed by that. And maybe it's fully backed, maybe it's partially backed, and you just trust that, you know, England is probably going to be around for a long time, that if the Bank of England doesn't have all your gold right now, they will eventually be able to have enough gold to redeem whatever thing you have. So you can still treat it as money good, money equivalent, and just way easier to transport, etc., even if it's not fully backed by gold. And so.

It started to look like the crypto ecosystem was developing some of that, but developing it in a maybe more cycle-inducing way than other financial systems, because what you typically see in a fiat system is both bankers and regulators realize that banks have this amazing privilege that they can essentially create money or they can create something that is, as far as the average person is concerned, it is money. That's an incredibly valuable privilege. And you sort of want to make sure that they are using that correctly. Because if you don't, then the responsibility of your entire banking system is determined by whatever the least responsible bank is, because they're the one who creates the most credit. And so they're the ones who define how much credit gets created.

And crypto just doesn't have any way to do that. Like there is, because it's decentralized and permissionless, et cetera, like you can't just say the maximum leverage for a DeFi lending protocol is X, and you know, we're not allowing it if it's above X. Like the protocol doesn't know, the protocol doesn't care. And the underlying Rails don't know, they don't care. And so you actually have a system where by default credit is going to be very pro-cyclical. So a lot of people will want to make loans, and a lot of people want to borrow when crypto assets are appreciating. That creates a feedback loop. This is like a classic Hyman Minsky thing. So Hyman Minsky, very interesting economist, like said some things that were absolutely brilliant, some things I strongly disagree with. But his description of credit driven financial cycles is, I think, really powerful and particularly powerful for people in the crypto world. Because what he says is that there is this cycle where

When money is tight, almost any investment you can make actually has really good returns. There just isn't any money to go around to make those investments. And over time, people start making investments. Those investments do well, even though the credit was expensive, and lenders start to realize that, hey, we're making pretty good returns and we can lend more and you know we'll continue to make good returns. And maybe the rates at which they lend start to go down. And so they're making less money in percentage terms, but they're willing to lend a lot more. So in absolute profit terms, they're doing okay. But while they're doing that, what they're also doing is just increasing the amount of credit in the economy. So they're increasing people's buying power. And if that has a feedback loop into real world supply and demand, what happens is it actually causes economic growth. And that economic growth means that the loans that you underwrote.

In a weaker economy, they're retroactively better loans because now people have more money to spend, they're more credit worthy, et cetera. And what could eventually happen is that you start to have this overshoot where people are borrowing money for things that really only make sense if the economy continues to grow and that credit expansion is the only reason the economy is continuing to grow. And so you have a bunch of loans where it's the exact inverse of what happened early on. Like early on, the loans looked risky because the economy was weak, the economy gets better, and so the loans turn out to be less risky. But now the loans look less risky only because the economy is really strong, and it's only strong because there's a lot of credit flowing through it. And at some point, somebody just doesn't want to make the next loan, or there is some hiccup somewhere in the system, and you start to see the whole process on wind, and you find out that these last loans were underwritten based on assumptions that only made sense given that expansion of credit that did not actually happen. So, this is, you know, another way to tell the story of the housing crisis, for example, is that the marginal buyer of a house in 2006, very likely that their income, like the reason that they were not buying a house in 2005, was that they couldn't get the loan. Maybe they work in construction, maybe they are a mortgage broker themselves, they're in some part of the country that is seeing lots of economic growth that is disproportionately residential real estate investment. So, like you have a lot of these loans where they look good on paper, like this person could actually pay back the money they've borrowed, but only because lots and lots of loans are being made. And then when that reverses, I mean it is a very painful reversal, it overshoots, and then we kind of get back to the beginning. So the annoying thing with the crypto cycle is that at that cyclical peak, a lot of the crypto economy activity is just speculation. Like there's a lot of money in helping people gamble on crypto. There's a lot of money in making markets in various weird crypto assets or in operating exchanges or in making large margin loans or offering derivatives, et cetera. And what that tends to do is it actually tends to suck a lot of the attention away from fundamental improvements in crypto and from user-facing, like real economy facing versus financial economy facing technologies. Like that stuff gets crowded out. It's simply less lucrative. In 2021, you don't want to be doing a remittance business that is going to help people more cost effectively send money back home. You want to be doing the business where instead of that person working a job in a rich country, sending back money back home to their family, like you make a lot more money from that person if they have a job in the US and they put all of their profits into Binance or FTX or something and are just constantly trading.

At the time it was NFTs, really.

Yeah, yeah. The NFTs, the NFTs thing also went crazy. So crypto, it has these really extreme cycles. Now a lot of those cycles involve some credit creation and then credit destruction. So it ends up being just a really interesting lab for looking at how bubbles happen and how they work. And the thing underlying all these bubbles, like the assets themselves, have this very bubble-like dynamic, which is just intrinsic to currencies, which is that you don't buy currencies based on your expectation of future cash flows. You really buy currencies because you know someone else will buy them from you later on and you have confidence in what that price will be. That sounds like a really speculative thing. Like if I said, you know, I'm buying Tesla shares, not because I know anything about electric vehicles or whatever, but because I know someone else is going to pay five percent more for them next week. That's very, very bubbly kind of talk. But with currencies, like that is why I have US dollars, is that I know that people will treat the dollar as valuable.

It's like currencies are priced, they're not valued, right? Yeah. That's the difference here.

Yeah.

Yeah, and like you can sort of view taxation as this sort of way that the government either you know runs an ongoing buy and burn program in crypto terms or just steps in to always be the greater fool that like you may not think dollars are valuable, but there is this organization out there where once a year they give you this amazing deal, which is give us these worthless pieces of paper and you don't have to go to jail. Like, what a great deal! Like, I can't believe this person did pay such an irrationally high price for these pieces of paper, but hey, I'll take it. I think it's a great trade. And yeah, crypto doesn't have that, so it has to like float around based on the sentiment of people transacting in crypto. And that just, you know, much like gold, like gold has value in for exactly the same reason that people think that someone else will buy their gold, and in particular, that when the economy goes crazy and when the stock market crashes, or when there's a wave of money printing, there's high inflation, like the gold is still going to be there. You know, one ounce of gold is still one ounce of gold. And so it preserves your purchasing power in cases like that.

One interesting thing though, Byrne, on that is like there is an element where crypto networks are somewhat like the US government with a taxation system in that block space itself is denominated in the crypto asset. So to the extent that there is block space demand, that creates some reservation demand for the underlying cryptocurrency asset. And there are some economic designs, for instance, in Ethereum, where actually a portion of that your taxation call it is burnt. We see that to a lesser extent in Bitcoin, right? There's not as much, you know, exogenous block space demand in Bitcoin. It's all because of Bitcoin. But you also see this design in crypto networks as well, kind of like thinking as block space is kind of the thing that you buy in the taxation system of the crypto network.

Yeah, the so the economics of BlockSpa get really interesting because what you can do with BlockSpace is you can basically free ride on the total value of the network and the total value of all these transactions and basically say that as long as the specific transaction I'm encoding within this does not have some independent value that swamps the value of the overall, say, Ethereum ecosystem, like it's not worth it to do something that weakens Ethereum in order to get after my transaction. So, like it does have value in that sense, but it's just very, very hard to put a dollar price on that block space. Like you have to sort of look at what is the global demand for very economically significant transactions. And that's a deliberately vague term because that significance has to exist in the head of whoever's doing the transaction, and then try to figure out okay, what would someone pay for that to be hosted on this blockchain versus some other blockchain versus just we're going to do this transaction, you know, we're going to actually hire some lawyers, we're gonna put together a Word document, it's going to describe things in legal language, and if we don't like the outcome, we'll take each other to court. And it's just, it's really like you have many, many orders of magnitude of potential value for that. So it is something, but it's also very hard to put a value on. And then

Another issue with that is that if you are trying to design an economic system, like if you do have to tax something, one of the questions I ask is like, what is the least damaging thing to tax? And what are some disproportionately damaging things to tax? And I have the kind of semi-ideological view that taxing transactions that enable price discovery is a generally value-destructive thing. And unfortunately, it is inevitable. Like the main thing that a modern country taxes is personal income or personal consumption, you know, some combination of those. And both of those are price discovery functions. You know, income, income is the market's attempt to figure out what you personally are worth and to figure out which company you are best suited to work for by determining which one is the high bidder. And that's a very economically valuable thing to do. That is finding people's life's work and then rewarding them appropriately for their contributions to society. So, you know, maybe that is the most important thing the economy does. And so if you put a tax on that, you are saying we're going to accept that this gets done a little bit worse than it otherwise would be because we do need the revenue for other stuff.

It ends up being a trade-off just worth taking. And basically every civilization ends up taking that trade-off. But it does have a cost. And so if you look at block space as the underlying intrinsic value of a crypto ecosystem, you are saying that we're going to weight our implicit taxation of participants in this system even more so towards things that entail price discovery. And that's maybe not ideal. Like you could actually potentially view Ethereum in particular and Bitcoin in a hypothetical different world than the one we live in today as kind of analogous to the China model, where part of the government's revenue source is that it owns a lot of property and it owns the right to decide what gets used for what purposes. And by taking a farm outside of a city and saying this can now be industrial property, you get a large markup on it, and you can fund a lot of the state's operations by just slowly selling off those assets. And

As long as you're like liquidating a percentage of assets that roughly matches the long-term price appreciation of those assets, you do have an effectively unlimited runway to just start with a large base of assets and slowly sell it off in order to subsidize positive externalities. So, you know, there's the hypothetical version of this world where Satoshi says, I am going to sell off, I don't know, 5% of my stash every year, and I'm donating that to core developers and I'm making angel investments in crypto companies that have good use cases, et cetera. And Satoshi didn't do that. And we talked about this a little bit in the book. Like the fact that Satoshi was there for a while and then very much split and is no longer participating, or at least not participating under the Satoshi moniker, that was very important to Bitcoin's development. But

I think that that is a different version of the future where you don't have direct taxation. You have basically quasi inflationary taxation of there is someone dumping this asset on the market periodically, but they are dumping it on the market in order to spend the money on things that create positive externalities. And so as long as those externalities exceed the market impact of the sales, then it creates net value for the entire ecosystem. And this is the model that I think a lot of um we'll call them altcoins.

There are other words for these, but like a lot of them try to have something like that. But I think it's just very hard for them to calibrate how much do you sell, at what pace do you sell, and what do you actually invest in that makes this a winning ecosystem? Those are just really, really hard questions to answer.

Byrne, in your book, there's an entire chapter dedicated to the story of Bitcoin and the bubble of the crypto world. And it's just a really strong anecdote to I think prove or just illustrate some of your larger points that I really want to actually kind of take a step back and define. Your book is called Boom. It's on, I'll call it, there's like two entrees in your book. There's the story of bubbles and the story of stagnation. And you also bridge these two things together. I think in the crypto world, we live for bubbles. Bubbles are fun for us. We enjoy them. Money's flying around, money's changing minds. It's kind of, yeah, it's where like all of the excitement is really made in the crypto world. And then also previously on the Bankless Podcast, we've touched on the recent like accelerationist movement coming out of Silicon Valley and Beth Jesus. So I think our audience is maybe primed for these two different components, but no guest so far has really been able to like link these two things together and talk about the thesis that you've talked about in your boom. So maybe we can kind of like just set up the context of which you've written Boom and kind of prepare our listeners for the two entrees that we're going to bring them.

Okay. Yeah. So I guess one way to do it is set up like thesis, antithesis, and synthesis. So the standard way to think about bubbles, the way that I thought about bubbles from you know the first time that I heard about what happened in the 1920s, or like I started paying attention to financial markets in the 1990s and then was paying very close attention as that whole thing unwound. So like the initial thesis is okay, sometimes investors go pretty crazy. Sometimes they stop thinking about the fundamentals of what they're investing in and start thinking about other people's behavior. And as they do that, once everyone imagines that there is someone slightly dumber than them who's just waiting in the wings to buy shares of broadcast.com or you know buy my large, you know, massive levered position in RCA in the summer of 1929, et cetera. Like once everyone is operating on the assumption that someone else is going to step in and buy, then at some point you just run out of dumb people. Like everyone is fully tapped out, they've all bought as much as they're going to buy, they're all borrowing a ton of money, and then a stiff breeze can just knock the market over. And then there's this antithesis point to this, which is hey, if we look at the long term economic impact of these bubbles, what we actually see is that.

Some of this is like it's actually kind of a Robin Hood thing going on. You have these really rich people who invest a ton of money in some project or set of projects, and they do lose a lot of their money, but things get built. So look at say railroad bubbles in the 19th century. A lot of people lost a lot of money speculating in railroad stocks. But the US has a really, really good freight rail network. It is very, very cheap. If you're transporting things over land, if you can transport them along railroad, you are paying a fraction of the cost of transporting them on a truck. And trucks are also pretty efficient. So, you know, it's a big deal. The UK similarly had a very well-developed railroad network, and that ended up being a nice subsidy to their industrialization. So they had railway bubbles in a smallish one in the 30s that felt like a big one at the time, and then a really big one in the 1840s. And that one, you know, they were investing a substantial chunk of GDP, like I think it was like a single-digit chunk of GDP each year in just building new railways. And what that meant was as their economy further industrialized, as they were importing more raw materials from the rest of the world and manufacturing them, building finished goods domestically, and then shipping those to the rest of the world again to keep their trade system in balance. Having an internal system of railroads was actually a really, really good way to have an efficient manufacturing footprint within that country. You could get the coal where it needs to be, you could get equipment where it needs to be, you could get the iron to the steel mill, you could get the steel to the factory that's turning it into whatever the end product is. Like this all ended up being really valuable.

But when you think about that, okay, so maybe just from the perspective of economic it like minimizing long-term economic inequality, maybe having these periodic booms where the bust tends to hit the richest people a little bit harder because they just have more money to lose, and then the infrastructure is beneficial to everybody. Maybe that's good. On the other hand, if the infrastructure would have gotten built anyway and we were just really in a rush, we built it really fast, we were kind of sloppy, we didn't know exactly which cities needed railroad connections, so we just threw them out wherever we could. And you know, sometimes that leads to a new city coming into existence, like Omaha, pretty much a creation of railroads. But sometimes it just leads to a railroad to nowhere where we dug up all that iron and dug up all that coal and shot down all those trees and built this thing. And you know, a lot of people worked really hard, and some of them probably died because 19th century, you know, workplace safety was not quite up to modern standards. And yeah, then we ended up with just a railroad to nowhere. So maybe that was pointless. Maybe we should have done a slow roll on this, but then.

What we talk about in the book is no, there are actually some things where

They don't just get built sooner because there was a bubble. The only reason they exist was a bubble. The only reason that they exist was that for a brief period in time, so many people were so confident about some very specific change in the way the world works, some deployment of particular technologies or changes in organizational structure. They were so confident in this that everyone knew that if they make a bet that only makes sense if this actually happens, that bet has a much higher chance of paying off. But only if you make that bet right now. So look at AI right now.

Zero to one

Yeah. So AI right now is a fantastic example where NVIDIA is improving their chips at a very, very rapid pace. And it would be completely irresponsible for them if there weren't a use case. If they're just hoping that, hey, someone out there will create a game that is so demanding on graphics that no matter how powerful our GPU is, there will be demand for it, that would be a little bit irresponsible. But in this case, they know that there's basically a market for flops and that if they build more efficient GPUs, there's absolutely demand for them. And then at the labs, what do they think? They think, well,

This model, like the scale of the model we're considering, is so absolutely vast it is just completely computationally infeasible. Like if you described the amount of compute that is needed to train, say, Llama 3.1, if you had described that to someone 20 years ago, you would have been explaining to them why AI will never happen. It just requires a wildly unrealistic amount of compute. But of course, we know that that amount of compute is actually realistic because of NVIDIA's efforts. So each side is basically coasting on the assumption that, you know, coasting in a pretty aggressive way, but they're coasting on the assumption that the other side is actually going to build the thing that makes the thing they're building worthwhile. And so you can have this equilibrium where, you know, Blackwell doesn't get built or doesn't get built for a really long time, and where Llama 3.1 does not get built or does not get built for a really long time, but probably just doesn't get built at all. Like that's one stable equilibrium. And then the other equilibrium is they both get built and neither would really exist without the demand entailed by the other.

So when you have a situation like that, like you can basically radically advance along one branch of the tech tree. And what you're really trying to do is figure out okay, what are the real world fundamental limitations? So maybe the limitation is you can train really, really powerful AI models until you run out of good, fresh data, like high quality tokens. And once you hit that point, maybe you can't train anymore. But we didn't have any way of even speculating about that until we started actually making the big bets. And I don't know for sure that that is the case. I think there are some interesting things going on with synthetic data right now that can maybe not completely demolish the data wall, but maybe help us gradually scale it. But that is the general model is that when you have a boom, you have a lot of capital flowing into a given sector and it's based on a specific vision of what could be different, then everyone who is working on some sub project within that, it's actually a less risky project, conditional on everyone else taking big risks.

Which is just, it's a really cool thing to think about. It means that bubbles are a coordination mechanism. They are a way to tell people that this is the thing to work on, this is where to apply your skills. It also kind of like one another way that we like to think about it, and we talk a little bit about this in the book, is there's this idea of industry clusters where there are particular cities you should be in if you want to be in a particular field. So if you want to work in finance, it's a really good idea to be in New York. If you want to get into movies, you probably want to go to LA. Like if your dream is to work in this industry, you don't want to work in like the third or fourth or fifth best city. Like movies do get made in Atlanta, but it's probably not a great idea to bet everything on your film career by moving to Atlanta and getting an office job. Like you probably want to move to LA, wait tables, do auditions, et cetera. And maybe it doesn't work out, but that's the place where you find out whether or not you've actually got what it takes. And a bubble is it's exactly like that. It is one of those industrial clusters, but it's a cluster in time rather than in space.

bm.

We were saying this is the time when everyone who cares about this thing needs to be building and needs to be building something that is adjacent to things that are already getting built. And you need to just sort of have this faith that you don't actually know what everyone's product roadmap is. You don't know what their plans are. So you don't know for sure if people are building the thing you think is the right thing, but you don't have the freedom to wait and figure it out. By the time you know exactly what else is getting built.

Someone else has probably built the thing that you would have. And this should change how you think about some aspects of bubbles. Like there's people talk about bubbles being driven by fear of missing out, by FOMO. But our view in the book is if you actually have skills that are relevant to a particular bubble, and that bubble starts,

missing out is the thing you should fear.

Like that should actually be just existentially terrifying for you because

It's a real fear, yeah.

yeah, like sometimes you look at what's going on in the world and you're like, I was basically put on this earth to be participating in this thing right now.

Are you gonna do the second best or third best thing? Hopefully not. Hopefully you do the thing that now is is the time to do it. And of course, it's daunting. And of course, it feels terrifying because like bubbles, they feel like bubbles for a lot of the duration of the bubble. So people started talking about a dot-com bubble in 1995, and they had no idea how crazy things would get. And people were talking about AI being overhyped. Like there were companies going public and having AI-related names and AI-related tickers pre-Chat GPT. Just they knew it was kind of a theme. They knew investors kind of liked it. They knew that Google had been talking for a couple of years on investor calls about how they're an AI company. And, you know, there were these periodic stories of like a group of Stanford grad students would just take the work they were doing for their thesis and they'd be like, Well, this isn't a thesis, now it's a startup, and suddenly the startup gets acquired for tens of millions of dollars. And that's a lot better than even the offer you get from wrapping up your PhD and trying to get a job at a big tech company. So, like there were already bubbly things well before the part that we now think of as the true bubble.

Right. Okay, so that's I'll call it the first entree of your book. And I think our listeners are more or less primed to really lean into this idea of like, oh, like I understand how bubbles, the bubbles in the crypto space, advance the crypto space. Like they can probably resonate with that. I personally resonate with that. And then the other part of your book that I want to introduce is just this idea that also crypto people tend to really lean into risk. Personally, my own anecdote for getting into crypto was I was faced with I could either go to physical therapy school, get a doctorate, or I could come into crypto. And one, I was faced with like six years of school, a bunch of student debt to go down this like very carved and known path to get like a very carved and known like career path out the other end of it. Or I could go into this like unknown, this crypto future, no academia, no certification, lots of risk, but like lots of fun. And in your book, you illustrate that there's this like growing societal risk aversion towards venturing out on your own unknown, building, innovating, doing something new, doing something that is like, you know, not what the typical societal path is. And so in the crypto world, I think we're all kind of identified. There's like this resistance towards accreditation, but and this acceptance of you know the high school dropout, the college dropout. But then like normal society is just like, no, here's your normal nine to five path. You like work at this one job, you become partner, like that's your key to success. And you make it this big argument that there's this slowing progress, there's a growing stagnation in society. Maybe you can start to introduce this other half of your book to really kind of round out this picture here.

Yeah, you know, it's funny when Tobias and I were first bouncing around ideas for the book, at first we weren't even really going to talk about the stagnation thesis at all because we kind of assumed it's sort of in the water. And I think for a lot of people in some parts of the tech world and parts of the policy world, it definitely is. And then we did realize that this is something like we want to make the case for this. I think there are some really good cases that have been made for this. Tyler Cowan's Great Stagnation really introduced this idea to a lot of people, including me, that if you look at long-term charts of productivity growth, you see things that make sense for a while, which is in a pre-modern economy, there's minimal productivity growth. And it is very unlikely that you, if you are just, you know, a farmer in the year 1000 AD, like the economic world that you're born in is very similar to the economic world that you die in. There might be some slight improvements somewhere else, but it's entirely possible that you could live your entire life at that time and just never encounter some technology that was invented within your lifetime, even though that was happening, but it was happening in different places and kind of gradually. And then

As you get closer to the industrial revolution, you do start to see these long-term uplifts in productivity. That we actually get better at farming, we start manufacturing more things, we start to get better at manufacturing. And manufacturing has these feedback loops where if you are better at building very precise machine tools, you can build a lot of other things, including the next generation of even more precise machine tools. So it tends to recursively improve with you know through a lot of human effort and a lot of capital investment, a lot of risk, et cetera. But it does happen. And then there's this, in retrospect, pretty crazy period in the mid-20th century.

In the US, economic output controlling for hours work to controlling for the total capital invested in the economy, output per hours growing like two percent a year. Which is crazy that you know, with no additional material inputs, people are just getting more and more productive. And then after about 1970, that slows to one percent a year. There's a brief blip in the 90s and early 2000s, and then things kind of degrade again. And then there's potentially another productivity inflection happening right now. We'll we'll have to see because these are these are very long-term trends. Like great stagnation was written about 40 years after that productivity fall off. People were talking about it a bit in say the 1980s. I think it was in the 80s where the line was coined that you can see the computer revolution everywhere except in the productivity statistics. But there was this slowdown. And one explanation for it is that there was some low-hanging fruit, that things like electricity and the internal combustion engine and the transistor, these are generally useful technologies. They can be massively improved over time, and that when we discover them, we get a very long period of productivity growth. And we have to redraw a lot of economic assumptions in order to fully take advantage of that. So, you know, to benefit from industrialization, one of the things you have to do is switch your country from being mostly agrarian with a few cities where people actually don't live especially good lives, like the disease burden's pretty high and quality of life is pretty low. But you know, you at least get to do something other than farm all day. You have to switch from that to a world where actually the default is urbanization and where the default job is not growing food or doing things involving food, but is actually producing other goods. And then we've slowly transitioned to more of a services-oriented world. We're we're all service sector people, and that's what a lot of the economy is today. But those transitions take a long time. And even if the technology exists and works in basically its final form, figuring out everything you have to rearrange in order to get the maximum value from it takes a long time. So electrification, for example, factories, people started using electric motors a couple decades before most factories were fully electrified. And it turned out that you need to just redesign a lot of things about factories, including their physical layout. So if you have a pre-electricity factory, it has some form of mechanical power, and there's like a central unit that is providing all that mechanical power, and everything is just connected with ropes and pulleys to that central power source. And so you have a finite fixed amount of power, and you're trying to design everything around that. And that means that if you want to add anything that consumes power, you have to take something out.

Hmm.

And that anytime you replace any one thing, you have to rebalance everything else. It also means that your factory, it's ideal for it to be a very tall building and to have multiple floors where people are using different pieces of equipment that are all connected to the central power source. And that's just in some ways a pretty inefficient way to design things. You also have to, you know, have your factory near a power source. Once you can electrify, you can put your factory just anywhere that you can hook it up to the grid and it can expand in two dimensions, which means you don't have to redesign it every time you expand things. Like if you had some tool in your factory and there's an upgraded version of that tool, you can just unplug the old one, install the new one, and your electricity bill maybe goes up, but you don't have to redesign everything else. And if you had two assembly lines and you realize there's enough demand for a third one, you don't have to find a new building, build an entire new factory. You can just plop down the additional assembly line next to it. So this changes a lot of things. Like it changes how companies think about.

Retaining earnings. And you can actually see this in some of the dividend payout statistics over time. And it used to be that equity was basically like the most junior creditor. People did not actually expect stocks to appreciate over time. They expected them to have higher returns than bonds, but the higher returns would come in the form of a dividend yield that exceeded the bond yield. And the reason the dividend yield exceeded the bond yield was if the company starts losing money, the dividend yield goes away and the bondholders still get paid. So you were sort of this junior junior claimant on the company's profits. And the assumption was those profits fluctuate around a midpoint. And maybe your company does well and it tends to pay a high dividend, and maybe it does badly and pays a low dividend or no dividend at all. But it doesn't just grow indefinitely. But once your factories can be pretty of more arbitrary sizes, and once they can expand incrementally, and once you can buy new equipment within an existing factory and not have quite the same overhead for actually getting utility out of it, then companies start to retain earnings and they start to just grow over time. And that changes the entire nature of the organization. It actually, one of the things it means is that you can hire people and tell them that your job right now is pretty boring. You're an individual contributor, but our headcount is only going up and the responsibilities are only growing in magnitude and the complexity of our process, it's only going to get more complex over time. So there's room for you as just, you know, the new hire on the Ford assembly line. There's actually room for you to move up within the company. Ford is going to get bigger, and what we do is gonna get more complicated, and what we do is gonna get more lucrative. And so you actually have some upside. It's not just you have a job, that's your job, and one day you will die or just be too sick to work. It's like you have a job, but maybe in five years you're supervising a bunch of people with your old job, and maybe in 10 years you're designing processes for all the supervisors, and maybe in 20 years, 30 or 40 years you're actually running the company. So very different kind of value proposition for the entire question of do I work for a company or do I stay home on the family farm or do I start my own little independent, you know, shop doing something? And so getting all of those social changes worked through society, like getting it to the point where parents actually have good advice for their kids and can tell them here are the perks of working for a big company, here are the perks of working for a small company, here are the perks of starting your own thing. Like that takes a lot of time. And that is part of that productivity growth. Because part of that productivity growth is just matching people to the right opportunities and adapting that matching process to changes in the scope of those opportunities and changes and how well we understand them. And so you had talked earlier about this idea of crypto people don't want to follow this standard path of you go to the right schools, you take the right tests, you get the right job, everyone knows it's the right job, and you know, you just follow this track until you're done. Like that, for one thing, those tracks, they're always based on a previous generation's experience of what went well. So they always tend to overshoot, they over-index on basically the past whatever the best jobs were when your parents were first getting a job. And sometimes those things change kind of slowly, but sometimes they change pretty fast and it's a little bit disorienting. And I think if you step outside of that and you say that whenever there's an obvious path, whenever it's a low risk path, the aggregate risk in the economy hasn't gone down because of the existence of big companies. The way that risk is experienced has changed. And sometimes what happens is

Big companies will riff on a different Tyler Cowan book. He Tyler Cohen wrote this book called Big Business, and it is all about how big companies are actually pretty great. And he makes the point that they are more law-abiding because they just have more at risk. Like if a mom and pop store, I don't know, doesn't pay someone overtime, just pays them their usual hourly wage. Like that's not front page news. But if a Walmart manager, somewhere in the vast Walmart ecosystem, decides to pull a trick like that, the headline is not, you know, Bob, who manages a store out there in Oregon, underpaid one guy by $30 this shift. No, the actual story is Walmart, big company you know and love, is mistreating its workers. It's cheating them. So big companies just they have more at risk if they violate the rules. So they tend to be a little bit more sticklers for those rules. And they do just get more output out of each worker. So when you work for a big company, you are producing more output per hour on average than you would if you were at a small company. And the company has kind of set itself up so that it captures a lot of the upside from that and it gives you a low risk slice of the returns, but it is also truncating some of your personal upside. And for a lot of people, that's a perfectly good deal. Like I think for a lot of people, there's no question of like if I could have some chance of 10xing my net worth and also some chance of having to sleep on a friend's cash because I literally can't pay rent anywhere. Like a lot of people would say, No, I'm I'm actually, I don't want to take this bet. Like this is a bad bet. Like I'm I'm much less excited by the prospect of great wealth than I am just appalled by the prospect of losing everything. And then a lot of people who are perfectly willing to take that bet, yeah, they end up in crypto, they end up starting companies, they end up doing other high risk stuff. So, like the amount of risk does tend to get preserved in the economy over long periods, but it does get shifted around. You always want to ask with those low risk situations like, one, are you giving up a lot of upside? But two, have you traded the short-term lumpiness of sometimes you like for me? I write a newsletter and do a lot of other stuff, you know, investing in other stuff. But like, let's say I'm considering do I write my own newsletter or do I work for a big media company writing a newsletter for them? One way to think about that risk is I can't really get fired from my newsletter. Like I get, you know, a little, you know, fractionally fired every single day because people are constantly unsubscribing, resubscribing, etc. But there's no circumstance in which I'd suddenly lose all of my income. But it's entirely possible for a newsletter to get laid off at a big company. And they don't really have visibility into that. Sometimes it is a strategy change that happens many steps away in the org chart. You have no visibility into it, like you're doing your job just fine. You're still creating wealth for the company. Maybe there's like some internal political power struggle, and someone, you know, you happen to be a pawn who gets sacrificed as part of that, you still have risk, and the risk is just harder to underwrite in that case. So, yeah, there's a lot of things to think about with just proper attitudes towards risk. And I think like the really proper meta attitude towards risk is to always ask who's on the other side of this transaction, what are they benefiting from when they transact with me? And am I looking at the full distribution or am I looking at the narrow slice of the distribution that makes this particular opportunity seem more promising and more fruitful?

And of course, risk-seeking people have exactly the same problem. Like if you look at the people who go to Y Combinator,