March 2025 Fintwit Book Club: Diary of a Very Bad Year with Byne Hobart from The Diff
The book’s core lesson is that genuine expertise provides a probabilistic edge, not immunity from consequential error. The anonymous hedge fund manager is incisive across markets yet says in March 2008 that “the worst has passed,” “subprime looks contained,” and Bear Stearns lacks a solvency problem. Byrne Hobart’s corrective to hindsight bias: “on a dollar-weighted basis almost nobody saw a financial crisis like that coming”—otherwise positioning would have defused it before it became a crisis.
The decisive 2008 failure was not merely mortgage losses but an information shock that seized the financial plumbing. Once AAA could mean either “money good” or “probably worth 95 cents on the dollar,” paper financed with 3 cents of collateral suddenly demanded far more, forcing deleveraging across apparently unrelated strategies. The book’s tap-water analogy carries the mechanism: systemic failure begins when a foundational assumption stops being true.
Credit bubbles can grow from tiny pricing errors because scalable funding attracts precisely the borrowers who should pay more. Equities can trade at 10 times fair value or 30 times forward revenue; credit may only be mispriced from 7% to 6.5%, but scalable funding turns that half-point error into a concentrated book. As Walker puts it, the scariest financial institution is a fast-growing one: losses appear years later or all at once in a downturn.
The most clarifying underwriting question is “what economic activity is being funded here?” Hobart’s specimen is 2021 DeFi yield farming: a dollar-pegged asset paying 20% was ultimately the riskiest layer of a leveraged margin-lending stack, not a productive source of 20% returns. In housing, actual credit funded construction, wages, remittances, and consumption abroad—so unlike vanished equity market capitalization, the money went somewhere real and often irrecoverable.
Uncertainty can damage an economy before conventional fundamentals reveal the break. Home prices roughly stalled in 2005, the first Bear Stearns hedge fund collapsed in 2007, and the full crisis arrived later; Walker wonders whether the March “tariffs on, tariffs off” regime could similarly freeze investment with effects visible only 15 months afterward. Hobart refuses false precision: Q4 2018 also felt ominous, yet going entirely to cash would have been “catastrophically bad.”
The AI boom has obvious misallocation risk, but its asset duration is materially shorter than housing or WeWork’s long-lease structure. Walker says CoreWeave used to depreciate GPUs over three years and, he thinks, was using five years in its IPO, while Hobart expects almost all peak capex to be fully depreciated by 2030 or 2031. Extra generation and cheaper power could remain useful even if demand forecasts disappoint, though Walker warns that prolonged exuberance could create follow-on distortions. Hobart’s balanced verdict is blunt: “current economics do not support current capex,” yet AI research already compresses hours of source-finding into roughly 10 minutes.
AI adoption in investing may divide flexible thinkers from rigid ones more than young analysts from older portfolio managers. A 23-year-old may resist LLM transcript summaries because meticulous manual reading is the only process he knows, while a 52-year-old accustomed to delegation may adopt them immediately. The investable edge remains judgment: recognize when to verify deeply, when a shortcut is enough, and when “something has switched” and it is time to “go for the jugular.”
1. Expertise can be real and still fail at the decisive moment
Byrne Hobart reads Diary of a Bad Year: Confessions of an Anonymous Hedge Fund Manager as a meditation on expertise, its limits, and action under uncertainty. The anonymous manager is “clearly very smart,” intellectually wide-ranging, and consistently focused on underlying systems—yet his accurate calls coexist with conspicuous errors.
Walker’s sharpest evidence comes from March 2008: the manager thinks “the worst has passed,” expects things to be fine, calls subprime contained, and says Bear Stearns has no solvency problem. The diary format preserves those mistakes before hindsight can sanitize them.
Walker asks whether readers are unfairly Monday-morning-quarterbacking Bear. Hobart’s market-based response is that sophisticated counterparties, including firms with explicit crisis-risk policies, still entrusted Bear with money and prime-broker relationships. Walker notes that plenty of people were predicting failures across Bear, Lehman, Bank of America, and other major institutions, but that does not make the manager’s call obviously irrational in the moment.
2. The crisis lived in financial plumbing, not just mortgage losses
Hobart argues that an obvious, gradually recognized housing slowdown would not have produced the same crisis. Ordinarily, weakening credit causes capital inflows to slow and risk to come off incrementally; a crisis requires a feedback loop that keeps financing alive until the system breaks abruptly.
The second-order shock was informational: holders no longer knew what complicated securities contained. AAA might still mean full repayment, or it might mean “probably worth 95 cents on the dollar”—a distinction that becomes existential when financing requires only 3 cents of collateral per dollar.
The anonymous manager’s memorable rule is that major blowups begin when an assumption proves false. The interviewer’s analogy is ordinary tap water: people organize life around water appearing when the tap turns, so discovering it is unavailable disrupts far more than the immediate transaction.
Hobart’s pushback on crisis revisionism: many celebrated subprime bears, including people discussed in The Big Short, identified bad mortgages without fully tracing the dollar-liquidity crunch that followed. What converted periodic bank write-downs into systemic failure was that “the financial plumbing seized up,” making short-term dollars and adequate collateral suddenly difficult to source.
3. Tiny credit-pricing errors scale into dangerous books
Hobart contrasts equity excess with credit excess. A company might trade at ten times what it should be worth or at 30 times revenue two years forward; in lending, the initiating mistake may be merely charging 6.5% when the risk deserved 7%.
Cheap credit scales that modest error. A borrower suddenly able to finance an otherwise marginal asset has strong incentives to do a lot more of it, while a bond fund naturally attracts issuers amazed they can borrow so cheaply. In consumer finance, the analogous borrower sees a $5,000 credit limit as “free money.”
Walker’s formulation is worth keeping: “the scariest thing in finance is a fast-growing financial institution.” Mispriced loans can look excellent for three years because defaults lag origination; only a downturn reveals that 9%, rather than 6%, was the appropriate price.
Recent parallels are mechanisms, not repetitions. Walker points to SVB, First Republic, and Flagstar/NYCB: securities assumed to be money-good can still create balance-sheet trouble if held in size and prices fall, while regulated rents and rising operating costs can undermine loans that once appeared durable.
4. Asking where the money went exposes the real misallocation
The anonymous manager repeatedly asks what underlying activity a boom finances. Hobart applies that test to friends earning 20% on dollar-pegged DeFi assets in 2021: the yield ultimately represented the highest-risk slice of a giant decentralized margin-lending stack, not productive activity capable of turning $1,000 into $1,200.
“Where did the money go?” is often wrong for equities because market capitalization is last price multiplied by shares; much of a collapse is vanished paper wealth. Credit is different: proceeds bought assets, paid workers, or caused assets to be constructed, so actual cash necessarily traveled through the economy.
Housing credit therefore became lumber, houses, construction wages, remittances, and consumption elsewhere. Hobart’s vivid endpoint is a recent immigrant sending earnings home, where parents finally buy a used car; recovering the defaulted bond would notionally require seizing that car in rural Mexico—neither practical nor morally persuasive.
This lens also explains the manager’s interest in emerging-market sovereign credit. Buying a Brazilian bond requires asking whether government spending will expand GDP and the future tax base enough to service it, or whether something else will happen to the borrowed money.
5. The manager’s breadth revealed correlations, sequencing, and creditor power
Both speakers speculate, without identifying him, about the anonymous manager. Walker initially suspects a macro trader, then considers someone from prop trading or a value-oriented emerging-markets background. Hobart guesses the fund may have started in the 1980s or 1990s with convertible arbitrage before expanding into a sprawling mix of strategies, including emerging markets, sovereign credit, black-box trading, and private lending.
Hobart treats that breadth as risk management: an apparently unrelated market may become a bottleneck for the strategy paying one’s bonus. In August 2007, statistical arbitrage suffered “a nightmarish couple weeks” when funds losing money in mortgage securities de-grossed other books, revealing correlations created by shared ownership and leverage.
The manager was early to name-check Huawei and correctly anticipated a U.S. credit-rating downgrade, but thought the downgrade would be a much bigger event than it was. Walker’s sequencing explanation is that it happened after recapitalization, when Europe had become more of the problem area and the U.S. remained a safe haven; he suggests it may even have been bullish for the dollar as investors sought dollar liquidity.
Hobart adds a counterfactual: had the downgrade occurred before 2008, regulators might have been slow to adjust risk weights, forcing banks to raise capital against Treasuries while also covering subprime losses and potentially worsening the crisis.
When tradable-credit spreads became “pennies in front of a steamroller,” the manager moved toward private credit and used every available pressure point. He called a delinquent borrower’s customers and suppliers—“Did you know that this company doesn’t pay its bills?”—and used the resulting pressure to force repayment.
6. Capital allocation can magnify talent and rationalize self-interest
The manager raises the concern that PhDs, doctors, and other scarce specialists migrated into pre-crisis finance. Walker updates the tension: a drug researcher might change lives inside Pfizer, yet potentially earn 100 times more investing in the companies developing the best drugs.
Hobart holds the libertarian counterargument alongside the discomfort. Complex economies need information routed to the right decision-makers, and financial prices might guide capital indirectly—even telling a CFO whose stock rose from 20 to 50 times earnings to stop repurchasing shares and reinvest.
Promotion turns specialists into allocators everywhere. Mark Zuckerberg mostly directs layers of people rather than writing code; an exceptional FDA reviewer may come to supervise reviewers; a McKinsey analyst doing quantitative work can become a partner focused on clients, sales, and team direction.
The scale argument is powerful: a doctor treats perhaps eight patients daily, while financing a drug might affect 800,000. Hobart’s warning is sharper: be suspicious whenever careful reflection reveals that “the most lucrative thing” is also conveniently the most moral thing one could do.
7. Uncertainty can freeze activity before fundamentals visibly break
On March 26, Walker connects the diary to markets enduring “tariffs on, tariffs off” and shifting definitions of friend and enemy. Earlier in the month, the Russell had fallen for “like 12 weeks in a row,” turning friends’ opportunity emails into what he calls therapy sessions.
The book’s cleanest self-inflicted uncertainty comes from automakers and dealers warning Congress that nobody would buy a car from a bankrupt manufacturer. The manager thinks the industry created the fear it described: consumers already bought airline tickets from bankrupt companies, but public rhetoric taught them cars were different.
Hobart resists turning present anxiety into a mechanical crash call. Q4 2018’s Nasdaq decline made worsening sentiment and slower corporate spending plausible, yet an investor who declared the growth cycle finished and moved entirely into cash would have made a catastrophically bad call.
His underwriting approach is to examine the final activity. Lending is safer when the marginal dollar funds an asset whose income services the debt; danger begins when new liquidity merely raises collateral values supporting older loans, with “income plus expected price appreciation” disguising the circularity.
8. AI’s short asset duration limits losses, but infrastructure can outlast hype
Walker frames the allegedly March 27 CoreWeave IPO as a possible WeWork-style comparison: bulls see technological growth, bears see fragile structure. Hobart’s distinction is duration—WeWork took long-term leases and resold office space on the spot market, making it first to absorb a collapse in demand.
CoreWeave’s GPUs depreciate much faster. Walker says the company moved from roughly three-year to five-year depreciation, while Hobart expects almost all peak AI capex to be fully depreciated by 2030 or 2031. Short duration cannot prevent mistakes, but it can cover more of them than a decades-long asset structure.
Walker’s pushback is that the boom is reshaping longer-lived infrastructure: Microsoft and Constellation were entering a deal to restart Three Mile Island, while the investment case contemplated U.S. power-demand growth rising from roughly 0% over the prior 20 years toward 5% annually. If AI disappoints after two more exuberant years, excess generation and related infrastructure could create follow-on distortions, even if consumers ultimately benefit from surplus power.
Hobart accepts that “current economics do not support current capex,” but argues cheap electricity is broadly useful, particularly if manufacturing returns to a high-labor-cost United States. His Y2K analogy: the dot-com boom funded waste, yet also financed the unglamorous software replacement that helped keep systems running after January 1, 2000.
9. AI rewards flexible judgment, not youth by default
Existing utility is already visible. Hobart gave Deep Research questions he had previously investigated and received largely the same sources and summaries in 10 minutes rather than many hours. That outsources the initial screen—whether something merits deeper human work—and expands the range intelligence can examine.
Walker keeps asking people he discusses finance and AI with how a 45-year-old portfolio manager differs from a 25-year-old analyst, without yet receiving satisfying answers. He expects clarity when the current cohort of college seniors reaches a second year on the desk or a first year at a private-equity firm after treating LLMs as naturally as earlier cohorts treated Google.
Hobart complicates the age stereotype. New entrants can harden instantly around one method—such as defining value as six times earnings—while veterans who survived multiple regimes learned to replace frameworks. A 23-year-old may insist on reading every transcript; a 52-year-old already comfortable delegating may summarize first.
The extreme shortcut is Druckenmiller reportedly asking ChatGPT for the five largest U.S.-traded Argentine ADRs and buying all five without further research. Hobart’s own libertarian enthusiasm trade used an Argentine ETF, but shallow work produced shallow conviction: “It was quick. It was lazy,” and he exited early.
10. Newspapers show how structural decline becomes cyclical collapse
N+1’s chapter introductions mark the crisis through newspaper layoffs and closures, with Michael Jackson’s death as the only celebrity landmark Walker recalls. Walker finds the choice jarring only 17 years later: institutions once central enough to mark social and economic deterioration are now largely absent from daily life.
Hobart reconstructs the economics. From the 1950s through the 1980s, two-paper towns consolidated into one-paper monopolies, especially in classifieds; cable television and AM talk radio weakened local advertising power before Craigslist separated classifieds into businesses resembling Zillow, eBay, and Backpage.
Newspapers protected near-term cash flow by serving older, higher-spending readers rather than cultivating younger ones. That made the audience “literally dying”; when durable-goods advertising collapsed in the recession, the structural decline became terminal. Hobart’s grim metric: the more valuable the obituary section, the more subscribers the product is documenting on their way out.
Full transcript
With me today is my co-host from The Diff, one of my favorite newsletters, Byrne Hobart. Byrne, how’s it going?
Hey, it’s going great.
I’m really excited to talk to you about the book we did today. It’s a book club. It’s not a talking-our-book club.
I like that. That’s a really good line.
Byrne, the book we’re talking about is on your recommendation, as so many of these are—and I don’t mean that in a bad way. You’ve made some great recommendations. It’s Diary of a Bad Year: Confessions of an Anonymous Hedge Fund Manager.
This is a really interesting book. It’s written over the course of 2 years. An interviewer from n+1 magazine interviews a hedge fund manager leading into and coming out of the financial crisis. I’d encourage anyone to read it. I had a really interesting time with it.
But I’ll pause there. Overall thoughts on the book? I’ve got lots of thoughts and questions, but I’ll turn it over to you to start.
I read the book as this really interesting meditation on expertise, the limits of expertise, and operating under uncertainty. You can read it and see that this anonymous hedge fund manager is clearly very smart and very thoughtful—not just in his domain, but he’s willing to range over a pretty wide variety of topics. He’s very much a systems thinker who’s always asking, “What is the underlying reality here?”
He gets a lot of things right about the crisis, and then he gets a lot of things wrong, too. I think it’s just a good reminder that there are a lot of people who are really well informed. They do know what’s going on. They do actually make a bunch of accurate predictions, and even then, they’re going to have egg on their faces over a lot of different details.
You hit on the exact same thing that was my first takeaway from the book. You read it, and this guy is smart. There’s no doubt about it when you’re reading it. He’s pulling from all different sorts of sources. He says during the book—I think he says he had a humanities background—and it’s no surprise because he kind of pulls from himself.
But you can’t help but read it and notice—and it was actually my first question to you—that he’s got egg all over his face from a lot of predictions. He’s getting interviewed in March 2008, and I just pulled some quotes, if I can find them really quickly. He says, “Look, I think the worst has passed. I think things will be fine. Subprime looks contained. Bear doesn’t have a solvency issue.”
So you have this really smart guy in the moment. When you read this—and this is one of the fun things about reading in-the-moment diaries, especially if they’re wide-ranging—do you think that when we read this and say, “Bear Stearns clearly was insolvent,” that’s all of us having hindsight bias? Or do you think this is a really smart guy doing great work who just made one wrong call?
I knew people—I didn’t know them personally, but there were plenty of people in March 2008 who were saying not just Bear Stearns, but Lehman, Bank of America, and, really, going up and down the list, that everything was insolvent. I just wanted to ask how much of what you read here do you think is us judging this person because we know the answers, versus this being a really smart guy doing great work who just made one wrong call?
One of the tests of that is that Bear, like all of the other big investment banks, had lots of very short-term capital and lots of counterparty relationships. Bear was a prime broker. A lot of very sophisticated financial-market participants—including people who would have some kind of counterparty-risk policy, where they were not just looking at the credit rating but actually asking themselves, “How do these guys perform in a crisis?”—were still very willing to put their money where their mouth was and say that Bear was trustworthy.
In some ways, a financial crisis doesn’t make that much sense if it was obvious and everyone should have seen it coming. On a dollar-weighted basis, almost nobody saw a financial crisis like that coming. That is what makes it a crisis.
It’s not a crisis if we have this gradual understanding that residential real estate is getting a little overextended, credit is weakening, and therefore we’re all going to take some risk off the books. That kind of thing happens all the time, but it doesn’t show up as a crisis. What happens is that the industry grows for a while, a lot of money flows in, and then the flow of money slows down because people decide that it’s no longer worth the risk.
That is the modal outcome for any scenario with this kind of setup. Occasionally, you get a case where there’s some kind of feedback loop that keeps money flowing in for longer than it should. In that case, we often don’t really understand it until the very end.
If you go back and do some kind of crisis revisionism, looking at some of the books that people love to cite—The Big Short, for example—a lot of the people in that book got some things right, but they weren’t talking very much about what actually led to a real financial crisis, as opposed to banks taking one of their periodic writedowns on some category of lending that got out of hand.
The actual crisis was that the financial plumbing seized up. There was a liquidity crunch, and it was very hard to source very short-term dollars. That was a second-order effect of not knowing what was in all of these complicated securities.
He does actually talk about that in the book. He talks about how it becomes an information problem: You have a bunch of AAA-rated paper, but now you’re not sure whether AAA means “money good” or means it’s probably going to be worth 95 cents on the dollar. If you’re financing that and you’re only putting up 3 cents of collateral per dollar, suddenly you can’t do that for anything. You have to put up much more collateral.
That dynamic—I’m not sure if any of the people who called the subprime part of it correctly figured out that entire implication. I’m sure there was somebody out there who connected all those dots, but that was part of what made it such an extreme event, and not just another case where the credit market overheated for a while and then fixed itself.
What made the book really interesting was that the interviewer—maybe he’s dumbing himself down, but in my mind, he’s not that financially sophisticated—asked things like, “Could you explain the hedge fund manager’s carried interest?” At one point, the hedge fund manager says that they got carried away, and the interviewer asks, “Carried away with what?” I was kind of thinking, “I feel like you would know that even if you were in finance.”
But it was interesting because sometimes he’d back up and have the manager explain things. One of the things that really stuck with me was when the hedge fund manager said, “One of the ways big blowups happen is when there’s an assumption, and that assumption gets proven incorrect.”
In this case, it was the assumption that when Moody’s rates something AAA, you’re getting your money back. That turned out to be incorrect, and everything imploded around that.
The interviewer says, “It’s kind of like assuming that the water will run when you turn on the tap. When the water isn’t there, all of a sudden, your entire world is out of whack.”
One thing that was interesting to me, on that assumption point, is that so many of his worries and concerns play so hard into what we’ve seen—or what we might be seeing—over the past 2–3 years.
We could start with an easy one: the assumption that AAA-rated paper is good. You start thinking about SVB, First Republic, and the Flagstar-NYCB disaster. There was an assumption that AAA-rated paper was money-good and that we could just hold it on our books until it matured. It turns out that if you hold enough of it and it goes down enough—or, in NYCB’s case, if you have these loans and OpEx keeps going up while rent regulations are limiting your rent—you could have a lot of issues there.
I’ll pause there, because there are a lot of other ones I want to discuss.
A lot of credit cycles—you never run through exactly the same credit cycle twice, but in retrospect, you do find a lot of commonalities between different credit cycles.
Often, what’s going on is that people are slightly mispricing risk. That’s, I think, one of the interesting things about credit versus equity market extremes: with equity, when equities get mispriced, it’s really wild mispricings. You look at a bunch of companies at their peak 2021 valuations and say either, “This company was trading at literally 10 times what it should have been trading at,” or, “This company was worth billions of dollars.”
I’d add a zero to that 10 times there, Byrne.
Right. But even for some of the companies where they’re growing really fast, there is a point where, if you’re paying 30 times revenue two years out or something, you just need a lot of things to go very right for that to be even remotely possible.
With credit, it’s often that people make fairly marginal mistakes. They should have lent at 7; they did it at 6.5. What happens with those is they just compound really fast, because when you’re in a credit market, it really scales once there is access to capital. So if there’s some category of borrower who, through some indirect means, is able to borrow at a little bit less than they should, and they’re able to put that into some asset that they think has some return that is suddenly at the acceptability threshold because of this cheap credit, then they do a lot of it.
Your credit book is always going to be skewing toward the people you should not be lending to. If you’re running a large bond fund, the people who most want to issue bonds are the ones who can’t believe that they’re able to borrow money at such a low rate. In consumer-facing financing, it’s always people who get their first credit card or BNPL and they’re like, “Wow, this is free money. This is a cash windfall. The credit limit is 5K, so I’m going to spend another 5K.” You always have that automatic selection for that.
The whole business is trying to mitigate that kind of selection effect, trying to find the signs that this borrower is just not suitable, even though they seem to be in some statistical sense.
What you said is exactly the reason why the scariest thing in finance is a fast-growing financial institution, right? Because you start lending to people at 6, and they should be at 7. You’re going to have literally unlimited demand for that. The real issues with that aren’t going to show up until 3 years later, when you start seeing the defaults—or, more likely, when you have a downturn and then all of a sudden, “Oh, that’s why we should have been charging them probably 9 instead of 6,” and everything’s just imploding all around you.
One really interesting thing I was—not shocked, but a throughline of the book is the hedge fund manager talking about how the financial crisis is a misallocation of resources and talking about how the bubble is a misallocation of resources. All of us know—I mean, you’ve written a book on booms and busts and what they can do—that a financial crisis is a misallocation of resources.
But I was just surprised that someone who’s trading in the moment—he says, “Most of my trades are done on a 6- to 12-month time horizon”—focused so much on the consequences of the misallocation of resources. A house that shouldn’t have been built getting built here, metal that shouldn’t have been mined getting mined there. I was just a little surprised by that. I don’t disagree with it; I think that’s a pretty common worldview of booms, busts, and euphoria, but I was surprised that he was so focused on that. I wanted to get your thoughts on that.
Sometimes, just asking yourself, “What is actually the underlying economic activity that’s being funded here?” is a really clarifying question. Back in 2021, in a different corner of financial markets, I had some friends who got really into yield farming in DeFi, and I kept asking them, “What economic activity is being funded by the 20% yield you’re earning on something that is a dollar-pegged asset? What is actually going on such that the person who borrows $1,000 from you has $1,200 in a year?”
It turned out the answer was that this was the highest-risk slice of some giant stack of leverage that was all being used for margin lending—decentralized margin lending. Once you see that’s what’s actually going on, you just know that this can’t actually go on forever. But if you do find that all the money is going toward some kind of productive thing, then you feel a little bit more comfortable. At least you have some kind of underlying economics that you can start trying to underwrite.
I think part of it is also this thing where sometimes it’s just an interesting question to ask: Where did the money go? But if you ask that about equities, where the equity market was worth this many tens of billions and now it’s worth some smaller number of tens of billions—or trillions, sorry, tens of trillions—“Where did the money go?” is kind of the wrong question, because it’s not like people actually had that cash on hand. That’s just the number you get when you multiply the last trading price by shares outstanding.
Particularly in the dot-com era, a lot of these companies had pretty high insider ownership. They were fairly new companies, and so they’d sold a big chunk to VCs, while founders and employees had a bunch of money. A lot of that money disappeared, but it was paper wealth. It wasn’t money that was being spent.
In credit, you can’t actually say that. If there is a subprime-backed CDO and it raises however many hundreds of millions of dollars, that money did actually flow into some kind of real-world activity. It bought actual specific assets, or it led to those assets being constructed in the first place. The money did have to go somewhere.
You still have the paper-wealth dynamic where the question of where the money went is partly, “This house used to be funded with 10% equity, but now the house is worth 30% less.” That’s where the money went. That’s why the bond is not actually able to pay off at the price, or at the principal value, that you expected it to.
But you still want to ask: Someone had money, that money got to someone else, it went somewhere else and somewhere else and somewhere else, and so on, and now it’s gone. I thought some of his answers to that were actually pretty good: Some of the money turned into what would have been productive activity if there had been more real underlying demand for it, whether it was cutting down the trees, building the house, or whatever.
Then where did that money go? If the marginal worker was a recent immigrant who was sending a lot of money back home, then the credit availability got recycled into investment in housing, some of that got recycled into remittances, and then into consumer spending in poorer countries around the world. Then you start to realize, “Okay, that’s why we’re not getting the money back.” We’d have to go somewhere in rural Mexico and seize the used car that someone’s mom and dad were finally able to buy because their son worked really hard building houses in suburban Phoenix for a while and was able to send money home. We’re not doing that. I don’t think there’s actually a very good moral case for doing that.
You can be annoyed that there were defaults on the bonds and things, but I think that’s part of what he’s going for. This was actually one of the interesting things that I wondered about when I wrote in the newsletter a couple of weeks back that I had reread the book ahead of this call. One of my readers asked me if I could figure out who this guy was, because there has to be a pretty short list of people.
Because he does a bunch of different things and talks about a bunch of different things his fund does, you just have the vibe that this is one of those funds that probably started in the ’80s or ’90s doing one strategy—probably convertible arbitrage, which is what a lot of them started doing—and then by 2007 it was a huge, sprawling fund with lots and lots of different strategies. He seems like the kind of person who’d be getting lots of questions from lots of different people about what’s going on and be able to go back to, “What is the underlying economics?” That’s a really good use case, I think.
He talked about doing emerging-market stuff, including emerging-market sovereign credit, and just asking yourself: If you buy a Brazilian bond, what is the Brazilian government doing with this money? Is it a useful thing to do? Is it going to make their GDP grow so that they have a larger tax base, so they actually pay the interest on this bond, or is something else going to happen?
No, you hit one of the questions I was wondering about as I was reading it: Who is the hedge fund manager? He talks about so many different things. I initially thought this guy for sure was a macro trader, right? By the end of the book, when he talks about his different strategies, he even says it seems like he’s a real value investor.
So, probably the early days of the prop trades or something. He's a real value emerging-market player, which does make some sense: he's very broadly read, and he understands how financial crises go through. But I was surprised. It's just a really curious question: what firm was he at when he talks at one point about, “Hey, we have to shut down our black-box trade”? He seems to have his fingers in a lot of different pies and his pulse on a lot of different markets, so it was really curious. Did you have any guesses?
I certainly do not. If I had really good guesses, I probably wouldn't want to share that. I would probably try to email this person and say, “Hey, is that you? I love your book.”
Did you feel very seen? Because I think you could narrow it down if your life's work is to figure out who HFM is. I think you could narrow it down based on the descriptions he gives and the fact that he moves to Austin at the end. But my question to you was: did you feel seen where he's like, “Hey, I'm burnt out. I'm going to Austin”? He's an early comer. But did you feel seen by it?
It was something I thought about a little bit. I think it has a different connotation pre- and post-pandemic, and these things always shift a little bit. But certainly, if you're in an environment where the way your life has changed over the last 10 years is that you're trading different asset classes, you might be at a slightly different part of Midtown while you're doing it, and then you do move to Austin, that is a huge shift.
But for me, the Austin shift was that I'd spent several months mostly inside what was increasingly feeling like a very, very small apartment, and I could go somewhere with a backyard. So that's what I ended up doing. Different set of trade-offs. I did feel seen.
Yeah, we chose this book because, especially at the beginning of March, I think things were feeling really dire. The Russell was down for 12 weeks in a row, and I was starting to get—I feel like I need to create a timer based on when I'm starting to get what I call my therapy-session emails from some of my friends, versus, “Hey, man, are you seeing any opportunities?” Just my therapy sessions.
But I was feeling very seen because, man, this is really stressful. Markets keep going down. And then I've got a kid, and I was like, “Hey, Alicia, taxes are high.” I did a podcast where I was like, “We're thinking about moving.”
Then I read this book, and the guy's like, “It was stressful running through the GFC,” which today is not the GFC, but it was stressful. New York City taxes are high, and he's getting married, whereas I have a kid, so he's moving to Austin. I was like, “Bingo. I'm thinking about doing it.” I was just feeling very seen by it.
At the beginning, they mention that he talks very fast and very passionately. He's hard to keep up with, which I think we know two people who do the same thing there as well.
Yeah, one of the things that makes it hard to track this person down is that there is this archetype of someone who is in the market. They have their specialty, but they're generally interested in the ways that people make money.
Some of that is just this practical thing of, okay, if all of my income stems, basically, from whether emerging markets are doing well relative to the U.S. or not—which, in a lot of cases, you can hedge, but often the expectation is that if emerging markets are really hot, you're probably getting a good bonus that year if that's the main thing you trade—there could be this practical argument of, okay, all of my income is based on this thing that I'm probably not going to hedge, so I should at least understand how other things work. Maybe I should invest in a fund my friend is raising or something, and I want to know how they make money and so on.
But it also pays off to be generally curious about things that are not directly related to the way you make money, because they often end up being the bottlenecks or the exogenous risks to the things that actually do make money for you. In fact, part of the financial crisis—part of what kicked it off—was that it turned out a lot of strategies were more correlated than people thought because you had the same diversified fund running all these different strategies. Because they were uncorrelated, the fund could lever up.
Then suddenly you find out that when mortgage-backed securities start to weaken, stat arb goes through this nightmarish couple of weeks in August of 2007. As far as I know, no one's been able to trace that to anything other than the fact that there were funds doing a lot of stat arb and also doing a lot of mortgage-backed-securities trading. When one started going badly, they had to de-gross, and that forced everyone to de-gross. Since it was August and a bunch of people were on vacation, that de-grossing was a little choppier than it had to be.
There were a lot of things in the book that I read and thought, “Damn, this is really striking me as a parallel to something that we've seen over the past few years,” or a parallel to some of the worries that people are having in the market right now. I'm happy to lob a few your way, but I was wondering if any one or two jumped out particularly to you.
I should look at my notes because I had a bunch of highlights in the book, and I had not realized that when I read it. When I originally read it, I was one of the people experiencing the very slow recovery in employment and wages after the financial crisis, but I happened to be working at a company that, for noneconomic reasons, had decided to get an office right on Union Square.
So every day at lunch, I would just go downstairs, eat a really quick lunch, and then go to the Barnes & Noble two doors down and read for half an hour. This was one of the books that I read in a series of Barnes & Noble lunch-hooky-playing escapades.
But then, when I was reading it on Kindle, I realized, “Wait, I have highlights in here.” At some point, I had downloaded it on Kindle and reread it years ago. The highlights were often things where I thought, “Hey, that sounds really, really familiar.”
One of the things that actually stood out to me, which is kind of adjacent to this, is that sometimes he would talk about things where it feels like he's really, really forward-looking. At one point, he name-checks Huawei as one of China's higher-value-added export businesses. It was a big company back then, certainly, but it was not a company that I think everyone had heard of to nearly the same extent that they have now.
So that was one. Let's see—he actually predicted a downgrade of the U.S. credit rating.
Yeah, yeah, he's saying a downgrade, just a downgrade.
Yeah, yeah, yeah. He was right that the U.S. could get a downgrade. He thought this would be a really huge deal. It turned out not to be. But I think that was just a sequencing thing.
If the U.S. credit rating had gotten downgraded in June of 2008, if Moody's had said, “Look, you have this huge economy. It's an oil importer. Clearly, demand is not responsive to higher prices. Therefore, the U.S. has just this gaping oil liability, and we don't think it'll be creditworthy long term,” et cetera, maybe that would have actually been what sparked the financial crisis.
But since it happened afterward, when everything had already been recapitalized and Europe had become more of the problem area, the U.S. was still a safe haven. The U.S. credit downgrade was certainly really embarrassing, but it was one of those things where, at that point, it was probably at least bullish on USD in the sense that everyone decided, “Hey, the world is even more chaotic than I thought. I better make sure I have dollars so that I can deal with my dollar-denominated liabilities in the future,” or just because dollars are safe in a case like that.
I think another interesting thing, just to play off what you said there, is that because it was post-2008, the banks and regulators were probably a little bit easier to work with each other on risk capital and how to account for things.
Whereas if it happened before 2008, I could imagine a world where bank regulators are really slow to respond. You get a U.S. downgrade, and banks say, “Oh, crap. We had zero risk weighting on our U.S. Treasury holdings. Now we have to go raise capital to cover the U.S. Treasuries,” and regulators say, “Our hands are tied. It's not AAA anymore.”
Then you have, as you said, the financial crisis spiraling even more out of control because you have to raise capital to cover your subprime losses and you have to raise capital to cover your Treasuries. I could imagine that world. So in some ways, maybe it happening after the GFC actually helped with it a little bit. I don't know.
Yeah. So another one, as I skimmed through my notes on this, was direct lending and private credit, where he talked about being more in that space. I do get the impression that maybe the spreads in the tradable stuff had gotten so narrow that he realized this was just not a good way to make money.
They switched from picking up nickels in front of a steamroller to picking up pennies in front of a steamroller, and at some point you just—it’s not worth it.
So he switches to private credit and then seems to be a very early adopter of really sharp-elbow tactics in private credit. He has this whole extended riff about how he lent to some company, and they don't pay him back. He knows they could, so he starts calling up their customers and suppliers and saying, “Did you know that this company doesn't pay its bills?” He uses that to force them to actually give him his money back. That felt a little bit forward-looking. I don't know if more of that happens now, but it's at least better known that private-credit people do not mess around.
No, I thought that was exactly it. And you could tell the interviewer was probably a little less financially sophisticated, but they were surprised by this. They're like, “What do you mean? If you're a lender—if somebody borrows from you and they won't pay you back—what do you mean you can do things outside of the courts?” I think they were a little taken aback when they said, “Yeah, we'll leak to the press: Hey, those guys aren't paying their bonuses.” The pressure on them from a bunch of different fronts starts getting higher and higher. I thought that was very interesting.
And you know, the more things change, the more they stay the same. One other thing—some stuff that I wasn't sure was forward-looking or was, as I just said, the more things change, the more they stay the same. The first one was that he mentions, again, that he's very focused on the misallocation of resources. He asks whether it's a shame that all of these PhDs, all of these doctors, all of these people are coming into finance and the huge funds pre-GFC, rather than doing the work they were trained to do. If you're a trained PhD, it probably makes more sense for you to be doing math work; if you're a doctor, treating patients.
I was really interested in that. I mean, you've written about this before, and it turns out that a lot of finance is—if you're the best quant modeler—where a lot of the money can be made. If you can decrease the latency or whatever, it's just interesting how finance is where a lot of these skills can be applied the best. Doctors, it turns out—if you're really good at researching drugs, you can make a lot of money and change the world at Pfizer. But you could make 100× more going and investing at a hedge fund and saying, “I'm going to read the biology and invest in the best companies that have these drugs.”
So I just thought—I don't know if he was early to that or if that's always been a concern. I'm sure it's probably a combination of the 2, but I'll toss it over to you.
Yeah, it is something that I go back and forth on a lot because I think you have this very obvious sense in which someone who gets a PhD in particle physics or something probably has a lot of contributions to our understanding of reality. Maybe those contributions are actually only comprehensible to 20 other people who share exactly their niche specialization. But still, advancing those frontiers is important.
Then I also try to hold in my head the sort of Cato Institute, very standard libertarian counterargument to that, which would be something to the effect of: We produce a whole lot of stuff, and there's just a lot of information that needs to be allocated in the economy. As the economy gets more complicated, more of what has to be done is getting the right information to the right people. Financial markets are just a tool for doing that, and they essentially pay you for making sure everyone's informed about where capital should be directed.
You can definitely push back on that and point out things like large-cap U.S. companies are net returning capital. So it's not like the market gives them a lot of signals on capital allocation. Maybe it is a capital-allocation signal if the market takes a company that was trading at 20 times earnings and pushes that multiple up to 50 times earnings. Maybe the company's CFO says, “Okay, the market is really telling us we should not be buying back stock. We should actually be reinvesting in our business.” So maybe it does perform this indirect capital-allocation role that way.
That problem does reach a pretty large scope when you have a complex economy with lots of different moving parts. And I think you can use exactly that to rationalize why people work on ads: if they're not allocating stuff to the right person, then they're basically in the business of producing more stuff. But if we have so much stuff that the ad business is a really lucrative business, then maybe we actually have a surplus of stuff and a shortage of ability to match that stuff to the people who really want it. So I kind of go back and forth on that.
One of the ways that you can sort of escape this dilemma is to point out that even within the more real-economy areas, the better you are at a given job function, the more likely it is over time that you could get promoted to the point where you're not actually doing that function. You're supervising a bunch of people who do that function. So Mark Zuckerberg doesn't write a lot of code anymore. From his online writings, it sounds like he still does write some and tries to stay a little bit sharp on that. But he's mostly telling people to tell people to tell people to tell people, and so on, what code to write. That is actually a higher-value use of his skills, including the coding skills, than actually writing the code himself.
Sometimes you do have people, and you can think of that internally as a capital-allocation job, and exactly the same thing exists even in the government. If you are, say, someone who is at the FDA and your job is to review the data, decide whether or not this drug is viable, and so on, at some point, if you're really, really good at that, you're a standout performer there. Maybe your job becomes supervising a group of people who do the thing that you used to do as your day job, and you still have to be good at it to understand what they're doing and how to prioritize and judge their work. But you're not actually directly doing that. You are once again a capital allocator, just allocating a less visible form of capital.
So maybe it is just everyone's fate to feel kind of disconcerted by the fact that we invest a lot of resources as a society in capital allocation, very broadly defined, and that the better you are at the real stuff, the more likely it is that your job will become the allocation stuff instead.
When I was at McKinsey, there was a joke among all the analysts that McKinsey was the only job—and this was just high-level management consulting—where, at the lower levels, all you're doing is quantitative analysis and strategic analysis. Then, as you get higher and higher up, a McKinsey partner or principal isn't spending any time on the deck. Maybe they'll do a review right before, but they're spending all their time meeting with clients and selling. In between, they're managing a team of people who are doing the analysis.
That used to be the joke: it's the only place where you go from doing quant at the lower level to doing sales at the high level. Now that I'm out of McKinsey, a little bit older, and have a few more grays in my beard, I'm like, that's the dumbest thing I've ever heard. Almost every job at the entry level is going to involve something more quantitative, and then, as you get higher, you're managing a team or managing capital allocation. You're spreading the knowledge that you hopefully gained or have.
I think the difference with finance and all of these is, hey, if you're a doctor, you can treat 8 patients a day, right? An hour per patient—8 patients a day. If you're doing it in a capital-allocation finance model, you're financing drugs that could treat 800,000 patients a day or something. So the returns to scale there are really interesting. I hear you. It's kind of a shame that the best doctor is not out here telling Andrew how to fix his sprained ankle or something, but it's probably better in the long run if they're allocating to the best drugs that are going to save thousands and thousands of years of people's lives.
You want to say anything there? There were a few other interesting places, but there was one other thing I wanted to hit you with.
Yeah. I think that framing is probably true, but I always feel like it's really, really good to be cautious about any time that you can rationalize the most lucrative thing you can do as also the thing that happens to be best for the world. That will actually be, in some sense, true on average unless the economy is completely broken. The economy is this system of bidding for people's talents, and the way that you get the highest bid is to put those talents to the best use. The price that you can get for the outputs of that is some measure of how much society values it.
But on another level, I always try to be really, really cautious about any time that the maximally wealth-maximizing thing for me turns out, once I've given it really careful thought, to also be the most moral thing I could do. The temptation to stop thinking right at that moment is so, so strong. So I always try to press back on that if I can. I don't think you're a basketball fan, but it reminds me of basketball.
You see ex-NBA players saying, “The game back in our day was so much harder. It was so much better. It was so much purer.” Every player thinks the game was perfect in the era they played in. If they played in the 1990s, they think the game was perfect in the 1990s. If they played in the 1970s, they think the game was perfect in the 1970s. In 2010, it was the same thing: everybody thinks their ideal version was the best.
Everybody thinks, “The best government is the one that happens to align with my views.” It all makes sense: whatever you’re doing, you get paid the most, and you think it’s the best for the world.
One interesting thing, just when I was reading it: there are some cultural milestones that, if we have time at the end, I thought were funny. But one interesting thing is that I had forgotten this: home prices kind of stall out in 2005, and you don’t really start to feel the effects of this on the market until that first Bear Stearns hedge fund collapses in 2007. Then, 18 months later—I mean, by October 2008—things are starting to go really wild, and it gets pretty crazy.
I guess the reason I say that is that we’re talking again about the end of March 2006: tariffs on, tariffs off, lots of uncertainty. The thing that comes through in this book is that uncertainty is a killer, right? You see that nobody can invest. There’s a really interesting story in there about how the auto dealers and the car manufacturers go to Congress and say, “You can’t let us go bankrupt. People would be crazy to buy a car from someone in bankruptcy.” He’s just hitting them over the head: “That’s crazy. You killed yourself by saying that, because nobody thought that before. But now that you’ve said it, people buy airline tickets from companies in bankruptcy all the time.” I bought a Spirit ticket recently, and Spirit went through a bankruptcy.
I guess where I’m driving at is this: the past couple of months have been filled with uncertainty—tariffs on, tariffs off, all this type of stuff, who’s our friend, who’s our enemy. Markets, of course, have obviously been a little volatile. But I do wonder whether the unintended consequences of some of this volatility, some of these tariffs, and some of this uncertainty are getting felt today. Everyone says this, but if, in 15 months, we look back and say, “We should have known. Things were starting to freeze up, and we were underestimating the unintended consequences of casually changing around the entire global trading infrastructure,” that would be a pretty big deal. To bring this into a more modern-day view for a second, I just want to ask your thoughts on that.
Yeah, it’s such a tough call to make because you can always look back at some previous pseudo-crisis and say, “What if you went completely into cash in Q4 of 2018? You said, ‘This is it. This 20% or whatever drop in the Nasdaq is just the first leg down, and this is the end of the post-crisis growth cycle.’” It would have been very embarrassing and would have been a catastrophically bad call to make. But people were pretty nervous at that time, and it did feel like every day, sentiment was a little bit worse. You could start to ask yourself: even if economic fundamentals are fine for now, how is corporate sentiment going to hold up? How willing are companies going to be to spend in the next year, given that everything is slowing down and their demand picture is more uncertain?
We did manage to just power through it, which was great. But you’re always going to look back and say either you underreacted or you overreacted, and nobody really calls the top or bottom perfectly.
That uncertainty factor did seem like a big theme in the book. I guess it comes back to when he talks about these models of the underlying economics. Some of that is just trying to mitigate the uncertainty, because if you know that, in the end, the marginal dollar that you lend is actually going toward something that you think is worth more than that dollar and produces enough income to service that debt, you can feel a lot more comfortable.
But if you start to think about it and realize, “Hey, the marginal dollar that I lend is actually pumping up the value of the collateral of the previous dollar somebody else lent,” and neither of us actually had enough underlying income to service this, then there’s a problem. There was income plus expected price appreciation, and between those two, we felt pretty secure. But if the price appreciation is actually coming from the inflow of liquidity, then as soon as that stops happening, everything just starts to collapse.
You can underwrite some of that uncertainty by asking, “Do a lot of the underlying behaviors that derive from this investment make sense, or do they not?” But you still have limits. That was another theme of the crisis: we have all these little interconnections, and you just have no idea which link in the chain is the weak one. You have to bet on the whole chain or not at all.
It’s really interesting because, again, we’re taping this on Wednesday, March 26. Allegedly, the CoreWeave IPO is going to price tomorrow, March 27. It’s interesting to think about this: if I had told you 20 years ago—so this would have been 2005—“People are taking out mortgages to build houses,” nobody would have thought, I don’t think, that this was a massive misallocation of resources. These were willing borrowers who were getting loans underwritten, buying houses, and building houses. Everybody probably would have been quite fine with that. It was creating jobs. You would have thought, “That’s great.” Three years later, it’s, “These borrowers were borrowing too much,” and all that sort of stuff.
I see the CoreWeave IPO—and this is nothing new to anyone who listens—as having a lot of similarities to the entire AI buildout. There’s this huge boom. AI is taking every dollar in, and the markets are signaling that AI stocks are a race. The markets are signaling, “Invest, invest, invest, invest.” You hear Facebook saying, “We will invest. We are going to overinvest. We can’t lose this race.”
Then you have CoreWeave coming up, which I don’t think is something akin to the financial crisis, but it reminds me in many ways of WeWork when they were trying to do the IPO. It was going to come at, I think, a similar number to what CoreWeave is trying to come at. People would say, “Hey, this is the future. It’s real estate, it’s tech, it’s growthy.” Or other people would say, “These are poorly constructed, short-term leases, and it’s a house of cards.” With CoreWeave, you can see a lot of the same things. I don’t know where I’m coming at this from in the book, but I can see the connections between the AI bubble and the growth drivers. I’d love to get your thoughts on that.
A huge difference is just duration. I’m literally in a WeWork right now, and I doubt that it’s profitable. Maybe it’s unit-profitable, but it probably doesn’t cover the corporate overhead. It turned out not to be a great business. There were some cool things about it, and I think there was a case to be made that there was actually a potentially viable business hiding in there.
But one of the problems was that they had this natural, crazy duration mismatch where they were buying into very long-term leases and then selling them on the spot market. If spot prices—the price for an office over the next month—are persistently higher than lease prices, then you can do really well with that. But then you’re the first one exposed to any collapse in demand.
I think WeWork had some kind of narrative because it was a post-crisis company. Part of why it could do well was that sometimes companies would downsize. You have a 20-person company, they lay off most of the staff, and now they have 5 people. They don’t want to lock themselves into a lease. They also don’t even have the liquidity to do that, but they do have enough cash flow to keep those 5 people employed. WeWork is just a natural place to stash your remaining human capital and wait to grow out of it again.
You have a lot of different forces pushing in a lot of different directions. But one thing that gives me a little bit of comfort around the AI boom, AI bubble, or whatever you want to call it, is that the spending happening right now—even at peak capex—is all going to be fully depreciated. Almost all of it will be fully depreciated by the end of this decade. So, by 2030 or 2031, whether the capital was allocated well or misallocated, that capital already doesn’t really matter. You have a shorter duration, which can cover a lot more mistakes.
It’s kind of like the BNPL, or buy now, pay later, stuff. Your traditional consumer lending blows up in part because you have this fairly long-term relationship with someone, and you have to figure out not just whether they’re creditworthy when unemployment is incredibly low and everyone wants to buy a used car because they can finally get a job, or the pay from driving for DoorDash pays them well enough that they can make the payment on that car. You have to think about what this borrower will look like if unemployment hits 7% and the economy is not growing.
With buy now, pay later, your economic outlook has to be 6 weeks into the future. As long as things don’t blow up, that “pay in 4” is going to get paid.
I do hear you on the quick depreciation, right? You buy a GPU and it's depreciating. At CoreWeave, it used to be depreciated over 3 years. Now I think in their IPO they're depreciating over 5 years, but it's pretty quick.
I think where I might push back is, if you were really worried about the AI boom—and I don't think the AI boom, because people aren't levering it; to my knowledge, they aren't levering it the same way. The thing with housing is you had 80% leverage. It was a mainstay of banks' balance sheets, all this sort of stuff. But if I was going to push back and say, “Hey, the AI boom, especially in Q4 and January of this year, was so hot, and the prices were getting pushed so high, particularly power prices.”
You were having Microsoft and Constellation Energy enter a deal to restart Three Mile Island. There was such a pull for demand that you were having people change around the power grid for this. If the bubble bursts tomorrow, if CoreWeave can't IPO and Nvidia goes down 50%, I think it would be no harm, no foul. But if the bubble ran for another 2 years—and I'm using “bubble” in quotes here; I'm not saying it is a bubble or isn't—you could imagine a world where 4 years from now we say, “Wow, we really got crazy, and the only use case for AI was somebody making some really cool Japanese manga-style images and sharing them on Twitter or something.”
Oh, my God, we've built 7 nukes. We restarted Three Mile Island, and we were basing it on power demand growth going from 0% over the past 20 years in the US to 5% per year. Power demand is going back to 0%, and now we're oversupplied. The nice thing there would probably be consumer benefits, but I could imagine it's one of the ways where a bubble has economic distortions. I believe the HFM in the book says, “Bubbles breed other bubbles.” I could imagine a lot of follow-on issues there.
One of the counterpoints there is just, I think, for electricity in particular, more generation capacity and cheaper power are generally good. It would actually be nice if we had done more of that, especially if you think about the fact that if the US is going to import fewer goods from other countries and we're going to bring back some manufacturing, we do have really expensive labor. We have to be realistic about that, and that means we need some way to compete, some costs to compete on.
Maybe there are cases where it's more straightforward to say, “We should alter policy so it is a lot easier to rapidly build a factory in the US once you've decided to do that.” If you're taking a long and variable period to get everything approved, then it is just harder for the US to step up local manufacturing capacity. Competing on electricity prices is something that I think plausibly the US can do. We do have some cheap sources of energy, and we are actually a really good place—we have a lot of really good places to put solar and a lot of really good places to put wind.
We do have a lot of issues with the grid because we did this build-out before a lot of other places did. We have a grid that was designed for very different use cases, or, you know, a set of grids that were designed for a different balance of power sources and different kinds of use cases than we have today.
But that's kind of like if you look at the late-'90s IT bubble. There was a lot of malinvestment, but I'm personally really, really glad that there was a budget at every financial institution in the world to go through source code that was written in the 1960s and make sure that we were saving dates as 4-digit rather than 2-digit integers. I think Greenspan talked about how he used to feel really proud of himself when he would save 2 bytes of memory by putting in the year as 65 instead of 1965. This was one of his clever performance hacks.
We had to undo all of that stuff, and there was never going to be a time when it was the really cool, trending thing to do until suddenly there was this time when there was a general IT boom. The Fed was kind of worried that we needed this burst of expenditure to get everything ready for Y2K. We had no idea how big the problem was, and there turned out not to be a problem.
But there was also such a huge investment in mitigating that specific problem and in building new systems that just weren't going to have that specific issue. I think it is plausible that, in the end, the counterfactual benefit of the dot-com bubble was that the lights stayed on after the ball dropped on January 1, 2000. That's an interesting way to put the counterfactual.
Yeah. And look, if the worst that an AI boom does, as you said, is that it busts and we have a bunch of extra electricity generation, so consumers benefit from the surplus and our electricity prices go down, then it's really hard at this point to see how AI doesn't have some use cases. I'm sure you use it more than me, but the ChatGPT Deep Research tool is mind-blowing to me sometimes when I put in some of the stuff it finds. It's hard for me to believe that AI is not going to have a lot deeper use cases than that. But even if it's just that, I think we'll come out pretty well.
Yeah. I've tested Deep Research with a couple of things, and what's been interesting is I've tested it on things where I've actually done the research before, and I want to see what it comes up with. It often comes up with basically the same sources I end up finding and summarizes them, but it took 10 minutes and I took many hours.
Yes. But what that means is that you can do that first couple of hours where you figure out if something is worth researching or not. You just outsource that to ChatGPT and go do something else. I think it increases the breadth of things that you can actually apply human intelligence to. So, yeah, I think there are a lot of good use cases right now.
It's also very clear that current economics do not support current capex. That's also true for pretty much any boom, especially a boom where you have a lot of these cross-complementary things, because there haven't been very many companies at all that were founded in this post-AI period where they're not calling themselves an AI company, but they are saying, “We're going to design our org chart and our processes and everything we do around the assumption that LLMs exist and can do human-level work in a set of tasks today and in a growing set of tasks in the future.”
There are companies that implicitly do that because the founders are 19 years old. They've been using this for a large proportion of their economically valuable lives, and they are just very used to the idea that you code mostly in natural language, and if you need to send 1,000 customized emails, that is a job for LLMs and not for you, and so on.
It's one thing I ask them, and to date it might be a little too early, but I always ask them, “Can you tell me the difference between a 45-year-old portfolio manager who's using AI and a 25-year-old analyst who's using AI?” Obviously, they have different jobs and everything, but what are the differences in how they're incorporating AI into their process and how they're using it? What are the differences between the best 25-year-olds who are using AI versus the worst?
So far, maybe I haven't pushed them hard enough, or maybe it's too early. They haven't been able to give me great answers or uses, but in 3 years, when the current breed of college seniors kind of have their second year on the desk or their first year at the PE firm, I think the answers are going to be absolutely fascinating. When someone who's grown up with this—and this is Google for them—I think it's going to be really, really interesting.
Well, there is a weird thing, and I think this kind of touches on HFM, too: In the financial industry, there is this weird relationship between age and mental flexibility. I've just noticed people who are really new to the industry often get very, very set in their ways almost instantly. There is a specific right way to do things, and if you work with interns, you have to kind of reeducate them.
You have to tell them: If you grew up reading Warren Buffett and The Intelligent Investor, it is very hard to break the cycle of thinking that value is finding something trading for 6 times price-to-earnings. It's very hard to break that by saying, “No, a computer can do that. I have to have a deeper insight than that.” It's very hard to break that, and I try to break it in myself all the time.
I'm sure there are going to be some interns who grow up and say, “Hey, explosive-growth crypto works in this market, but it might not work in the next market.” Yes, completely agree.
Okay. But the other side of that is that the people who've been around for a while and who've been through multiple cycles and multiple regime shifts, especially the kind of shift where there's an industry where its multiple used to be X and now the multiple is Y. Whether that is, you're an industrials guy and when you started on the desk, railroads were a declining business by default, and the interesting question is always who goes bankrupt when, et cetera.
Then they change into this business that actually does have really, really good margins, is incredibly hard to compete with, and has gotten really consolidated, et cetera.
The more they've had to actually make those big adjustments—to the business model, how they charge for things, stuff like that—the more it ends up being either that people cultivate the ability to just change their framework, or that the people who can't change the way they think about things just get carried out. Either way, the people I've talked to who've been in the industry for multiple decades tend to be pretty early adopters of a lot of things where you'd expect it to be more of a younger demographic. I would not be surprised if there were cases where the 23-year-old analyst doesn't want to use an LLM to summarize a transcript because he learned to read transcripts a year ago and learned to read them very carefully and highlight everything and so on.
Then you have the 52-year-old portfolio manager who, partly because he's already used to outsourcing this, is already used to saying, “I don't have time for this one. You take it.” Just putting that in a different chat box that goes to a computer rather than a person is an incredibly natural thing to do. Sometimes they do just end up taking shortcuts. When Stanley Druckenmiller had that interview where he was talking about Argentina, he said he just asked ChatGPT for the 5 largest market-cap ADRs of Argentinian companies traded in the U.S., and he bought them all without doing any subsequent research.
I'm not sure how true that is, but I feel like I would not have been brave enough to do that at 25, or certainly would not have been brave enough to tell my portfolio manager that that was my entire research process and that's why I got him the right ideas to bet on this theme so quickly. If you've been doing this kind of thing for a long time, maybe you do have the confidence to say, sometimes you know a good shortcut when you see it, and you're going to take it.
First, your LLM plug—putting transcripts into an LLM—has encouraged me. I am always so hesitant because I'm like, “No, I have to read it. I have to listen to it. There might be that one word that switches everything.” You've inspired me. You know what? Just toss it into the LLM, and if it's a big position or if you're really interested, you go back and read it. I'm going to try that.
Your Druckenmiller thing increasingly makes me try to ingrain this in my friends. There are certain moments where your gut is just screaming to you, “Hey, something has switched. This is the moment to go for the jugular,” or, “Hey, something is wrong. We need to get out of this.” Sometimes it's not a quantitative thing. The stock doesn't go from 10 times price-to-earnings to 12 times price-to-earnings and you say, “Okay, I need to get out of it.” Sometimes you hear something.
Druckenmiller, I think what's so good about him is that he's trained himself. He saw Argentina and thought, “Animal spirits are coming. Let's get into this thing.” For me, I would have been like, “All right, I've got to spend 3 months researching the history of Argentinian bond swaps, and I need to deeply research.” I would have come to a no because some of them were just going to nationalize everything again. He saw the ball really clearly.
I actually did just buy an Argentinian ETF, and my thesis was even lamer. I did not make very much money because I didn't hold it for very long. My entire thesis was: libertarians love Milei. Libertarians also love expressing political views by making financial bets. We see this in prediction markets all the time: the odds of the most libertarian candidate tend to be higher.
Ron Paul's odds of winning the Republican nomination in 2008 on TradeSports were always like 8% to 10%, even though the actual odds were, “If every other leading candidate has a heart attack and dies, then maybe Ron Paul somehow squeezes in there.” That was my thesis, and it was quick, lazy, and I had no conviction because I had not actually done all of that in-depth research on figuring out all the structural problems with Argentina and trying to figure out whether the tantric sex guru and libertarian anarcho-capitalist who's also running a country can actually fix all of these problems in time to get GDP up. I just flipped it. It was fun.
You say that about libertarians, and I believe that, right? The overlap between libertarians and online, deregulated, maybe gray-market betting sites has to be extremely, extremely high. I'm laughing at the Ron Paul thing because I remember in 2012 somebody wrote, “Hey, Donald Trump has a 5% chance to win the election on a prediction market.” This was 2012, and he hadn't announced he was running or anything. I think he was teasing with it, like he always did, but people were like, “This is free money. There's no chance Donald Trump is going to win the presidential election.”
In 2012, that worked out well, but in 2016, if you bet 5% against it, you would have had your head ripped off. In 2020, you would have had your head ripped off. I don't know—maybe Donald Trump was libertarians' dream candidate, maybe not—but it's just funny. These long-odds things sometimes work out. The world's crazy. Maybe brand recognition does it.
One last thing I want to mention: one really interesting thing I noted about N+1 is that the intro to everything starts with a little bit of, “Hey, here's what's happening in the world.” I thought it was really interesting. He marked one of the big downturns by talking about newspapers shutting down or firing people. I thought it was really interesting because we sit here today and, outside of maybe The New York Times, newspapers just don't matter, right?
It was also interesting to me because, again, I think even in 2008 and 2009 people had realized, “Hey, newspapers are in for a really tough time in the online world.” I just thought it was interesting that he used that. The other thing that I thought was interesting was that in 2 of the openings, where he says, “What's going on in the world?” he talks about Michael Jackson's death. I just thought it was interesting that the only celebrity he mentions is Michael Jackson. I have no real thoughts here. If you want to talk about the newspapers a little bit, I just thought those were the 2 interesting things. It's interesting just to go back in time 20 years and think about what's big and what doesn't matter.
Yeah. Yeah. I think if you were doing something like that about the COVID crisis, pretty much every chapter would just open with a tweet. Your tweet for January 2020 would be someone tweeting about how this is a ridiculous conspiracy and they can't believe that you're telling every guest at your office to use hand sanitizer or whatever. You'd go through all of the most and least alarmist tweets in each time slice that you're talking about. I think of some crazy things that I said and maybe even believed in January and February of 2020.
Yeah. Yeah. It was a wild time.
The newspaper thing was interesting because they'd had structural challenges. If you go back, it's one of those things where a lot of economic phenomena are either way newer than you thought, and the first time you heard about them was after they had gelled as just an economic fact, or they're way older than you thought, and it's actually a cyclical thing and you saw 1 slice of the cycle. So you either saw a permanent decline or permanent acceleration.
With newspapers, their economics were kind of okay for a long time and then started getting really good from the 1950s through the 1980s because a lot of the 2-paper towns consolidated into 1-paper towns. Suddenly, you had a monopoly in classified ads. Their economics had already started to get chipped away at—I think Buffett writes about this in his late-1980s and early-1990s letters—because of cable TV and AM talk radio. There were more channels, so that monopoly status still mattered for classifieds, but then what ended up happening was they really had to bet their business on classifieds because the local car dealership, furniture store, and so on had a lot of different places to advertise.
Then Craigslist comes out, and every section on Craigslist—I forget which VC had this slide where he has a screenshot of the Craigslist interface, which I guess is still the current Craigslist interface—and he's just circling, “This turned into Zillow, this turned into eBay, and this turned into Backpage,” or whatever. You had that business already getting picked apart, but there was a lag. There were still advertisers who just weren't completely aware of that and had been advertising for a long time. The people who were still reading the newspapers were an older, higher-spending demographic.
That can be a really dangerous trap for a media business in particular. Anytime you have a set of choices where you could either go for a younger demographic—they don't generate much revenue right now, but you will get them and they will potentially be loyal over time—or cater more to your older readers, I think the newspapers went for the older readers. It was better for cash flow, but it did mean that their audience was literally dying. The more important the obituary section becomes to your media outlet's economics, the more you have to be thinking, “Everyone who shows up in this is someone who was a subscriber, and this is how they turn out.”
And then, once there’s an economic shock, the kind of durable goods that often get advertised in local media are exactly the ones where purchasing just craters. It is enough to kill a lot of those businesses. So, yeah, it was an interesting landmark, and I think there’s going to be a time—maybe one of my kids will read that book at some point—and they’ll have to ask me, “Why was this a big deal?” I’ll try to explain that the media environment was like that: following the news, when I was growing up, meant reading the newspaper, and that was just how you knew what was going on in the world.
And now following the news probably means just obsessively checking Twitter and having one or more apps that are named after a newspaper, but you don’t know of any place where you could actually physically buy a copy of that newspaper. You’re still checking that.
I think all the time about my kid, like when and if they go to college. When we used to go out, it was like, “Hey, how are we going to get back?” There were no taxis in the town we were in. There was no Uber. They’re not going to understand the desperation of needing a designated driver or desperately calling someone up and being like, “Come pick us up.”
They’re not going to understand, “Hey, when I was in sixth grade, I went to the mall, and then I couldn’t find my mom. We didn’t have cell phones; I didn’t have a cell phone yet, so I was just running around the mall looking for my mom.” They’re not going to understand any of those things.
So it’s just funny. The newspaper—it was just, again, this was only 17 years ago. They were marking all of these newspapers as, what, a milestone: 500 people laid off. They’re threatening to shut them down, and they’re all gone. They’re just all gone. You can imagine plenty of things like that.
Anyway, Byrne, this was awesome. I think we ran through a lot of things. I enjoyed this book. In some ways, I enjoyed it; in some ways, I learned from it; and in some ways, I was a Monday-morning quarterback, being like, “You are in Paris, about to go down.” But this was a lot of fun.
Byrne Hobart from The Diff, one of my favorite things to read pretty much every morning. Some mornings we don’t get them, but I’m looking forward to it. We’ll have to coordinate on a book, but I’m looking forward to next month’s.
Yes, indeed. Likewise.