[Patrick notes: I'm delighted to be joined today by Byrne Hobart. It is my usual practice to add notes in this format to transcripts.]

Byrne is a writer and investor, best known for his newsletter the Diff, on the intersection of tech and finance. He also writes Capital Gains, explaining the finance concepts one's education may not have covered. Byrne is publishing a book, Boom: Bubbles and the End of Stagnation, which is coming from Stripe Press in October, and which you can pre-order on Amazon.

Timestamps:
(00:00) Intro
(00:25) The 30-year mortgage is an intrinsically toxic product
(04:46) Young households are the socially optimal holders of equities risk
(10:19) The structure of private equity returns
(14:18) Sponsor: Check
(15:32) Meta-analysis of the finance newsletter space
(19:54) Byrne’s aspirations for The Diff
(25:01) The origins of names
(27:19) The epistemics of a hedge fund
(34:26) Venture capital vs hedge funds
(38:13) Understanding scrapers
(41:20) How to learn about an industry from scratch
(45:37) The business of online travel agencies
(49:21) Wrap

Sponsor: This podcast is sponsored by Check, the leading payroll infrastructure provider and pioneer of embedded payroll. Check makes it easy for any SaaS platform to build a payroll business, and already powers 60+ popular platforms. Head to checkhq.com/complex and tell them patio11 sent you.

Transcript

Patrick McKenzie: Hideho everybody. I'm Patrick McKenzie, better known as patio11 on the internets, and I'm here with my – I would have called you an internet buddy once, but we've been upgraded to friends, Byrne.

[Patrick notes: In all businesses since time began, sometimes coworkers or customers graduate to friends. In the weird world that is working for the Internet, sometimes those of us who rub elbows online for years eventually discover we enjoy each others’ company in addition to enjoying each others’ ideas. Sometimes this friendship is entirely online; sometimes it extends into the other real world. Families get introduced to each other, food and drink get consumed, etc.] 

Byrne Hobart: Hey, great to be here.

Patrick McKenzie: We have an interesting way in which we found each other back in the day, but when I try to introduce people to you – you should definitely be following Byrne Hobart in all the places – I tell them to read your magnificent tome The 30-Year Mortgage is an Intrinsically Toxic Product.

Can you give people the brief version of that thesis?

Byrne Hobart: Yes. It is very, very unnatural for a bank that has demand deposits to lend someone money for 30 years. But we make them do it; there was a long struggle to do this, and there's a book called The Dead Pledge that goes into great detail on how that happened on the policy side.

What I was interested in was, what are the consequences of having these long term loans? Now, the consequence for homeowners is pretty good. You have this asset that does tend to appreciate over time; it may not be the best thing to put your money into, but it does go up and you can borrow a whole lot of the money that you invest in that asset.

So that is good. If it's a 30-year loan and you're paying it down continuously, you know exactly what you're paying, so you've simplified a lot of your finances. 

You’ve basically locked into, paying some amount of interest and making some amount of investment in capital over a long period. 

The standard 30-year mortgage in the US has this wonderful feature for you, where if rates decline, then you can refinance at the new low rates. 

So you have this free option – but it's not free, because options are never free; there's always a spread between what you would pay on a mortgage and what you would get on a 30-year treasury bond and that spread basically reflects the cost of the option.

But what you end up with is a really interesting situation. There are a lot of cases where in a financial market you'll have people who are buying underlying assets, and they're often buying just the stock, the bond, whatever – they're buying it to hold it for a very long time. They call them “real money.” Then there are more speculative market participants who are often doing a trade in and hedging some of their risk, and they're making money on the spread between what is the thing they're trading and then what is the cost of hedging it. 

With the 30-year mortgage, what happens is there is this large interest rate-options position, and one side of that trade – the person who is buying the 30-year mortgage – they do generally want to hedge some of the risk associated with that option. 

But someone who has a mortgage generally does not do that; they generally don't calculate, “what is the value of this option, and how many contracts of treasury futures do I need to be long or short at any given time to hedge out my risk?”

If they did that, things would be a little bit different, but what happens is they don't. 

So anytime interest rates go down, the optimal hedging strategy is to buy more treasuries; anytime interest rates go up, the optimal strategy is to sell more of those treasuries. The existence of this huge amount of refinanceable mortgage debt does actually cause long term rates to be more volatile than they otherwise would be, and the only consequence of that is that that is just the benchmark that everyone uses for the cost of money.

[Patrick notes: Let me repeat that because it is important: sometimes the tail wags the dog, and here it is the tail wagging the entire world. Hedging mortgage rate interest exposure for the most common way to purchase homes in specifically the United States implicitly dominos into the cost of capital for almost all businesses worldwide, exchange rates for most major currency pairs, the competitiveness of Japanese cars in overseas markets, etc etc.] 

What does the Treasury bond of a given maturity yield? That is the cost of money over that time period. The dollar is just the universal denominator for everything, and this is dollars, dollars in time zero versus dollars at time t.

Patrick McKenzie: To draw out an implication of this, some people say that “well, other countries choose to socialize medicine, but make the housing market a relatively free market; the United States chooses to socialize the housing market, but make medicine go to the free market.” But it's actually much more impactful from that.

The price of literally everything in the world is downstream of this peculiar choice for the microstructure of the American housing market. 

Byrne Hobart: There are other elements of toxicity in this asset that we can get into.

Patrick McKenzie: Oh, I think we could go for a very long time on it. 

I have a special interest in real estate because my father raised me reading the Wall Street Journal and telling me stories about it.

I told him at age seven – not due to toxicity, the risk-free rate or the consequences of hedging strategies, but just because of the conversations he was having around the dinner table about the type of people who turned up in Chicago real estate commercial real estate transactions, – I cried to dad and I said, “I can't go into the family business, I'm going to have to be an engineer or something.”

Actually, part of that anecdote is not true – at age seven, I wanted to be the Commissioner of Weights and Measures. But you can tell that I'm definitely still the same person. 

You said an interesting thing to me before .

I think that one of the positive consequences of the decision to securitize substantially all housing loans in the United States was that it moved this asset and the asset liability mismatch from the balance sheet of banks to holders that could be more structurally aligned with having very, very long term assets. 

I think probably the optimal holder for a 30-year mortgage note is something like a pension fund, which, on the one hand they know they need cash in 30 years, and if interest rates go up, the notional value of the bonds they are holding goes down – but that isn’t actually any skin off their nose because their cash needs are still as they modeled them previously, and now the net present value of their funding shortfall is smaller.

[Patrick notes: Explaining the joke: I am assuming that any pension fund large enough to talk about is underreserved because the incentives for looting pension funds and then kicking that can down the road are extremely attractive. This thesis was mildly terrifying when I encountered it 20+ years ago at the Pension Tsunami blog (now offline, sadly). The size of the shortfall is, broadly, larger these days, and a topic of much conversation here in Chicago, because both Chicago and Illinois have deeply dysfunctional polities which were unwilling to tell politically influential stakeholders No. ]

You've told me something interesting about who the socially optimal holder for equities risk is. Who do you think that is and why?

Byrne Hobart: That is households saving for retirement, particularly when they're fairly young. 

The reasoning there is that there's an efficient frontier where, if you're allocating between different asset classes, you can generally trade more volatility for more returns in roughly similar ratios – starting with, instead of buying one month T-bills, you buy one year T-bills; now your interest rate went up,  from negligible to trivial, but it's still pretty low. 

Then you can go all along that line, out to equities – and then you can go past that line, you can also go to asset classes that probably have negative expected value and are not really investable in the same way, but equity is a useful way to finance a lot of a lot of organizations, especially large ones or ones that are trying to get much larger very fast.

But equities are also quite volatile compared to other asset classes. If you are a large investor, you do care about that volatility.

Let’s say you're a college endowment. You are spending some amount of money every year; if your portfolio loses a third of its value, either the budget for the school goes down by a third, or the percentage of your assets that you spend every year goes up by 50 percent.

[Patrick notes: As Byrne elides here, because it is common knowledge among professional money managers: if the college’s endowment goes down 30% under your watch, that will certainly have consequences for the college, but the most immediate consequence for the investment manager will be them being fired and then disgraced. This is one of a few ways to have “career risk” as a professional money manager, and is a fairly aligned form of career risk. Many forms of career risk cause a principal/agent problem because the money manager should do something but won’t because of career risk. If the constituents of the pot of money understood the situation in the same way the money manager does and had simpler preferences than they do, there wouldn’t be career risk, but if wishes were fishes then sushi would be a lot cheaper than it is.] 

Byrne Hobart: Both of those are very scary things to have to deal with, and it just doesn't really make sense that [when asking] the question of whether or not there should be a major in X or whether or not we should hire this particular professor – is asking what the S&P did over the last six months really relevant to the decision? Usually not; [that] probably should not be the main consideration.

So you have a lot of kinds of investors who are afraid of volatility. 

Then there's a set of investors who really like volatility in particular ways: some of them are gamblers and a lot of them are founders. 

I think it is socially optimal for founders to own a lot of stock because they should have a large monetary incentive to do a good job – particularly, the more of their net worth is in the company, the more averse they are to completely blowing it up.

[Patrick notes: This has historically been an argument advanced against secondary sales by founders by e.g. venture capitalists. This argument has lost in recent years, partly as companies have taken longer to go public and partly as money was abundant in the zero-interest environment. I’ve personally not cared for it much in that context; the founders you want to back would feel far more psychic damage from no longer being the CEO of a high-flying company than they would feel from living in diminished material conditions. But outside of that context, this is a game of making numbers sum up to 100%, and allocating tens of percent to founders versus to teachers’ pension funds sounds eminently fair to me. This, however, does imply a natural consequence: when you sort the population by wealth descending, the beginning of the list will be dominated by people who, told they’d be zeroed out in the event of a business failure, said “Bet.”]

That's a healthy incentive to have, because they have a lot of people who are paying their bills, paying their mortgages, sending their kids to school, etc. with the money from this company. It's good for them to really viscerally feel the risks that they take.

But then there are just not enough super rich founders to buy all of the equities in the US. You need someone else to hold them. There are some investors who hold equities, but are not net long – a lot of hedge funds do that.

So who is this mysterious class of investors who could want to own a really, really volatile asset and should mind least if it drops in value? I think the answer is households. 

Households are saving if they're buying the most volatile asset class of the major asset classes, because it's also the asset class with the highest returns. If they're putting in a consistent amount of money every month or over set time periods, then a lot of that volatility washes out. It doesn't make a big difference to your retirement if, when you were 24 and you had $2,000 in your brokerage account, you lost half your money. It's just not a big deal. 

As you get older and as the number goes up – and as the question of “how much of this money can I spend every year if I'm not earning money anymore,” as it gets more salient – then you do want more of that portfolio to be in bonds. 

But as long as you have a large cohort of young people who are earning more money every year, we have a natural buyer for equities. We have a natural group that should be taking equity risk. I think the simplest way to do that is to buy index funds. That is my default suggestion to people: if you are asking how to make money, buy index funds. 

If you're asking how to potentially make more, but also more likely lose money in informative ways, picking individual stocks is certainly a great education – but it's  bad advice to give someone without direct context about their situation.

Patrick McKenzie: Yep. I have spent a little bit of time playing many video games. I've played World of Warcraft and I've also played in the stock market. (I've probably lost more in the stock market – I don't actually know if I cleared the implicit cost of capital in my public investing career, but be that as it may.)

[Patrick notes: This was, on reflection, slightly more modest than reality would have it, but not by much. You would definitely not hire me to be your money manager on the basis of my 2004-2024 investment performance in public equities. I’d put an IRR here but calculating it rigorously would take a week of work. Among other things, there is an external-to-the-accounts tax drag caused by tax-advantaged accounts in America not being tax-advantaged for investors with Japanese tax residency and so much of the apparent gain in my accounts has been dutifully offset by me paying out cash-on-hand to Tokyo for nearly 20 years.]

Why isn't the natural holder for equities, say, sovereign wealth funds, or perpetual institutions like an endowment, for example?

Byrne Hobart: They do have some of their money in equities, but what they usually find is that the risk-adjusted returns – the extra return they get from that is not worth the fact that if you have a large drawdown you either have to make tough decisions – or if you don't have to make tough decisions, it means the endowment was too big for whatever purpose it was trying to serve. It's really socially suboptimal if there is a nice public park with a $3 billion endowment and they spend a basis point of it every year on keeping the park nice and clean. That's just not the ideal way for that wealth to be allocated. 

Endowments do tend to have a public equities allocation that used to be much higher; it's gone down over time. A lot of that has been that they moved into private equity. 

[Patrick notes: Yale University’s endowment is something of a bellwether for the smart money among the smart money in endowment management. They happily publish their investing strategy, including a graph which matches Byrne’s claim here.]

It's a tricky argument there, because private equity on a risk-adjusted basis looks really good. But with public equities, the way you get a risk-adjusted number is you look at where the stock closed and that is a clear number, that is a price at which someone transacted. With private equity, they're private – there's not a last price; the last price at which someone transacted was when the private equity firm bought it.

So it would be inaccurate for them to just say, “we bought this company five years ago for a billion dollars, therefore it's on the books at a billion dollars.” Like, no, it’s either worth more or worth less. It turns out if you run a regression on private equity returns and you look at not just stock market returns, but the returns for companies in roughly the market cap range that PE likes to buy (and companies that are statistically a little bit cheaper, like low price-to-book, etc.) and you adjust for their leverage, you can generally predict quarter-to-quarter aggregate returns for PE really well, except in the worst quarters where mysteriously private equity does really, really well compared to public markets.

If a PE firm has another dollar of debt in the aggregate in their portfolio for every dollar of equity, in a year where the market goes up 10%, they would typically go up 20%, and in a year where the market is down 5%, they should be down 10%. That is often true, but in a year where the market is down 40%, you'd expect them to be down 80%.

You'd expect them to be very, very financially distressed. But they turn out to have picked companies that were all very very recession-resistant, and their managers saw around the corners and took action early – so mysteriously, they mark it down less. And so you get these really good risk-adjusted returns, but nobody is voluntarily exiting a private equity investment at the bottom of the market.

People do exit – people do sell companies in very distressed times, but it's really hard to do and you will get destroyed. A good example of that is SiriusXM, which did a deal at the bottom in 2009 with the Liberty Media complex of companies: it was a very, very good deal for the buyer and a very tough deal for Sirius, because Sirius had, I think, days of cash left before they'd have to file. Pretty much every day Liberty would offer a number that was (a) much lower than Sirius wanted to take, but (b) lower than yesterday's number. They just kept doing that until the other player at the poker table folded. 

But yeah, you don't have accurate marks at those points. I think if someone has to take the risk of, “this asset class doesn't do that great once you adjust for its leverage, and once you adjust for just a little bit of chicanery, and how they mark up their assets sometimes.”

If there is a class of investor who can take that risk, who's sophisticated enough that they could do the due diligence, they probably have some sense of what's going on.

Large endowment funds that are patient capital, that know what their cash outflows over the next 10 years will look like right now and can plan around the cycle of these 10 year funds – they're probably the best investor for that. 

 I think the other thing in who's the best investor is, “is this a kind of investment that is uniquely tax inefficient, and therefore should it mostly be sold to entities that do not pay taxes?”

That is another case where the tax inefficient stuff – it should be going to endowments.

Patrick McKenzie: So what's a strikingly tax inefficient thing? Something which is buying and selling frequently? Like, timber? Timber is not buying and selling that frequently.

[Patrick notes: I used timber as an example here because timber goes through periodic phases of being very hot with endowments. Also, my accountant has previously chided me that (at least in Japan) timber is the one complicated tax filing that I’ve never needed him to file and so he expects me to show up owning trees one day to spite him. And so this is me working in real time through confusion about whether timber would be a tax-efficient or tax-inefficient investment.]

Byrne Hobart: A very tax inefficient strategy that you could run as a public market investor is, “pick stocks on the basis of one to four week catalysts such that you are constantly realizing short term capital gains instead of long term capital gains, and to just never pay attention to the tax consequences.” 

But if you're doing this through an offshore entity and all of its economic owners are non-profits or are not US nationals, then that entity earns money, it pays money to a management company – that management company does pay money to people in very high tax states, but the actual return stream it's getting is pre-tax.

[Patrick notes: There exist many hedge funds, and at least some VC funds, which have Bermuda/Bahamas/etc vehicles for those limited partners who are not U.S. tax residents. In most cases this is aboveboard, as long as they’re complying with the laws in their particular jurisdiction.]

Patrick McKenzie: Ah, got it. 

So switching gears for a moment, you write a newsletter (The Diff), I write a newsletter (Bits About Money), Matt Levine writes a newsletter (Money Stuff); the three of us are often considered either competitors, or “clearly they must be colluding somehow, there must be a group chat,” because in some cases we've all hit “publish” on strikingly similar topics within, like, a multi-minute window.

[Patrick notes: An example, where Matt (in New York), Marc Rubenstein (in London), and I (in Tokyo) all managed to land newsletters worldwide in a seven minute window. This happens far more often than you’d expect.]

So, do  you want to talk about our deep, shadowy backroom shenanigans?

[Patrick notes: We in fact do not have a group chat, do not routinely preview our work with each other, and do not coordinate beats or schedules. We do occasionally swap emails/Twitter DMs/etc and have mutual high regard. If we were to all be in the same city, there would certainly be merriment and diversion, and then someone might complete the Adam Smith observation (Wealth of Nations, chapter 10) and introduce some contrivance to raise prices.]

Byrne Hobart: I think if we were cynical dollar maximizers, we would have a group chat, and we would decide which of us writes the main story about which event. 

There will sometimes be things that happen where I realize, Matt is going to do a really good job on this one, and I'm probably not going to do as good a job; he'll probably see something I don't. A lot of the stuff that intersects with the legal system, for example – he's a lawyer, I'm not. He went to Harvard Law, I dropped out of a much worse undergraduate school.

[Patrick notes: Harvard undergrad, Yale Law.]

Then, things that touch the innards of financial infrastructure – you have worked on that,  you have personally touched the innards of some of the financial infrastructure. 

Patrick McKenzie: I might've written a few lines of code in my day, although I hope they've been ripped out by now, but yep.

[Patrick notes: Of a few hundred commits over the years, I think only about two were reasonably characterized as directly touching financial infrastructure.]

[Patrick notes, on the general topic: I generally view this question less in terms of avoiding competition and more in terms of “beats.” My writing (and the underlying cohesion-to-reality) tends to be best and do best when I’m on my beat. With some exceptions, when I stray from that, it gets more handwavy. My self-assigned beat for Bits about Money is financial infrastructure, with a subfocus on direct movement of money. I also intentionally avoid commenting on current events unless it is absolutely unavoidable. As a result, insider trading or wonky bond structuring will almost always be written by Matt first and best. Byrne is a steamroller with respect to wide-ranging philosophy of capitalism and individual firm-specific investment cases. And I write about the plumbing, ideally in a way which both continues to be useful reference material for years and occasionally successfully predicts the future.

I also, idiosyncratically, try to limit how much BAM covers crypto, partly because I’m a crypto skeptic, partly because I’d like to keep my skepticism to a dull roar, and partly because I feel crypto gets coverage grossly out of proportion to its actual impact on or utility for financial infrastructure. (Why? It incentivized an emergent, distributed boiler room. If you write about crypto, you will attract the attention of a very engaged, very online community and your metrics go up and to the right. Many people have been convinced by those graphs, and by social pressure, that this implies there is a there there.)]

Byrne Hobart: So I think there is a somewhat natural division of labor and to the extent that we could – you know, no collusion, but to the extent that we were implicitly colluding, it would be in the same way that companies collude on earnings calls.

Airlines will sometimes say things like, “We're not really worried about low cost carriers entering our key markets because we know that we can beat them on price if we have to. We know that they know that – 

Patrick McKenzie: (laughs) Oh, I love it.

Byrne Hobart: “– if Spirit wants to send more flights into Denver, there are plenty of places where Spirit is making money where they could easily stop making money. There's always that possibility.”

Patrick McKenzie: The Japanese salaryman in me is just like, chef's kiss, this is wonderful.

[Patrick notes: In the culture that is being a salaryman, resorting to explicit threats is a failure to use one’s advantages. No true king needs to announce that they are the king, pace Tywin Lannister, and no one who could grind a counterparty’s business into a fine paste and consider that low-intensity conflict resolution needs to engage in boorish behavior like announcing that capability. Implying that capability is either a gross miscalculation or a calculated insult. Your counterparty should be symmetrically intelligent and informed. Therefore the proper salaryman takes note of the long, successful relationship with the counterparty, then announces the price for this year’s renewal, and buys drinks for his counterparty to commiserate after hours.] 

Byrne Hobart: But it’s also, I think someone who is running an efficient business – and is proud of the business, and believes that they have competitive advantages, and wants to articulate what is good about that business to their investors – is also going to say similar things in maybe a less sinister tone. They do make an effort not to sound too much like it's a threat.

Anyway, I think within any space where you have people who have finite time or finite capital and they're competing for opportunities, they will always try to ask themselves, “am I likely to be the winner, to the extent that there are multiple people competing for this?” – and also, “is there a thing where my comparative advantage is better at doing X, and even though Y is this time a bigger story, I will just not do Y?” 

I think we we all end up probably doing some of that by accident, just as a sort of background process; I think it actually in this case creates some consumer surplus because it does mean that we are all trying to write the thing that we would write best and cover the event that we would cover best.

There's some things where none of us will be able to resist the urge to say something. I'm sure there could be some fun discussions, some fun compare and contrast between say, your FTX narrative and Matt's and mine, and how those have evolved, who figured it out first, or who admitted first that it was a fraud.

I think you win that one, or at least you seemed to figure it out quickly.

Patrick McKenzie: [Well,] that's a very close call.

[Patrick notes: I was confused when the FTX collapse began, because FTX’s comms strategy suggested it was factually collapsing but I modeled them as being unlikely to collapse due to incompetence. I had thought they were spinning money from being Tether’s money launderer in chief, and that doing that in tens of billions of dollars in size probably paid for sideline interests like running a crypto exchange, buying Congress, and donating to pandemic relief.

I was also more briefly confused as to whether misappropriating customer assets would be a surprising update to the crypto-sphere. I assumed that everyone knew the reason for Alameda to fund an exchange was to use their customers as a capital source and that protestations to the contrary were known by everyone in crypto to be lies, but the genre of lies that are broadly acceptable in crypto. You know, like bank fraud: illegal, obviously, and you can’t come out and say you’re doing it, obviously, but find me a crypto person who in their heart of hearts actually thinks that bank fraud to enable a crypto exchange is wrong.

It turned out I was simultaneously not cynical enough and too cynical by half: SBF et al were able to successfully project an illusion of wide-ranging competence by using tribal signals and having an actually objectively impressive PR and influence game, and OTOH the crypto community believed that SBF was lying but not to them.]

Byrne Hobart: Although Matt does have the whole, “it sounds like you're saying you're in the Ponzi business” – 

Patrick McKenzie: Yeah, I think Matt was the first person who really said to the Emperor, to his face, “You're stark naked and you're bragging about it. Why is that?”

There was a wonderful, wonderful episode – I love his remark later that it was discovered later that, as things were collapsing around them, there was an internal discussion at FTX with regards to how to manage the communication strategy, and someone suggested, “maybe we should send SBF to be on the Odd Lots podcast again against Matt Levine” and Matt Levine said in a way that only he can, what tremendous content that would have been.

Patrick McKenzie: So, we have differentiated point of views, we have differentiated voices, differentiated beats. 

One thing that I intentionally try not to do: you folks are absolute writing machines and grind this out every day or almost every day, and are often commenting on breaking news that has a finance-adjacent angle to it – I intentionally try to keep aloof from news unless it is so squarely in the financial infrastructure thing that me not commenting on it would be newsworthy. I largely try to write… I hate the term “explainer,” but give people sort of deep dives into various parts.

Anyhow, talking my book – the purpose of talking to you is to have you talk your book!

What do you think is the intellectual center of Diff?

Byrne Hobart: I'm interested in the big questions of, “how do people coordinate? What do people want?” The Diff is kind of just this effort to figure people out, because I do not have a comparative advantage at doing that just one-to-one (like having a really deep conversation with someone about their values and things like that) but I think aggregates are a little bit easier to understand.

Also, characters, unusual people, are in some ways easier to understand because there are going to be some variables that just totally dominate and that make them easier to model. 

That's a lot of what I spend time thinking about – just trying to get a handle on these truths of how human beings interact, why we are the way we are, and how our ability to just push our will out into the world is affected by technological leverage and institutional leverage. 

I also find history really interesting, and history is getting written all the time. Sometimes I think to myself that the person I'm truly writing for is the grad student a hundred years from now who is studying history of the early 21st century.

I want to be the favorite primary source that the nerds about the 2020s are all reading. I think that is a thing worth aspiring to – sometimes I will read older primary sources and just be like, “this person is immensely fun and very funny, and they have a very seemingly modern outlook.”

There's this really really great paper called – okay, I'm blanking on the name, it's by a I think a math professor, Andrew Odlyzko, and he's writing about the British railway bubble.

[Patrick notes: Collective hallucinations and inefficient markets: The British Railway Mania of the 1840s,  appendix 1 “The Glenmutchkin Railway by W. E. Aytoun”, which begins on page 191 of the PDF.]

He has this appendix in his paper where it is a work of fiction that was pseudonymously written by a railway lawyer, and if you read it – if I sent this to you and I said, “someone wrote this about initial coin offerings, but he decided to set it in Scotland in the 1840s –but you'll recognize every character,” like, it's true. 

There's one character who is this very somber, low-church, Protestant business owner and he is just really, really obsessed with being a stickler about all the rules. – he doesn't want to invest in a railway that that runs on Sundays, for example – and then he turns out to be this completely ruthless businessman. 

So he's saying all the right things about ESG, and then ripping people off non-stop; then he gets his comeuppance in a wonderful way. 

There's another character – I have met this person many times at bitcoin conferences – who hasn't read the technical stuff, but read a book about economics. In this one, I think he decided that, having read Adam Smith, he was now a businessman. (Sometimes it's Ayn Rand or Murray Rothbard or something – those can be worth reading, but they're very much not something written by someone from the business world explaining how business actually works – but they feel like that.)

This person also does dumb things and loses money in comical ways. I like reading those kinds of sources, and I like those people enough that I try to be in that cohort.

Patrick McKenzie: Writing with an eye towards history... I worry sometimes I write with an eye towards Twitter over relatively short time horizons, although sometimes I'm smart enough to have a point of view of, “this thing that is bubbling up on Hacker News or similar will actually be read in the future, maybe I should put some work into it.” 

The one thing I wrote that was probably the most distributed thing I ever wrote – was when John Graham Cummings was complaining about how various web applications didn't accept his name in the Hacker News thread back in the day, and I wrote a comment on, “As a person with a absurdly long name by Japanese standards, I have broken so many computer systems, and I have collected many, many ways in which one can break a computer system by having the wrong kind of name – because every computer system does names incorrectly by the standards of at least some people.”

I was writing this comment up for Hacker News, and then I thought, “well, no one will actually see it after tomorrow if I post it on Hacker News; I'll put it on my blog.” And so, Falsehoods Programmers Believe About Names is copy-pasted into documents at the United Nations, in every engineering department in AppAmaGooFaceSoft, and various other places, banks, etc.

I get CC'd on tickets by people who are like, “despite this bet having been written more than 10 years ago, we have nonetheless managed to do it again.” So I'm the central clearinghouse for every bug that involves a name in a computer system and then get a stupid amount of email that way. Well that was an ancient walk down memory lane.

Byrne Hobart: I try to write things of that nature more with the financial flavor of, “there's a model of the world, and you're instantiating that in some system – then there is the world, and your model interacts with the world. If your model is very, very good, it just rolls right along and everything works fine. But your model is not perfect, and so here are the interesting ways where it breaks down.” 

And sometimes it goes the other way: a lot of names come from the demand for everyone to have a first name and last name at home address. I think Seeing Like a State talks about this and about how there are places in the Philippines where everyone's name starts with the letter A because they just had a list of last names and everyone got one and they got lazy.

Patrick McKenzie: I love Seeing Like a State in so many ways, partly because it gives you this concept of legibility, and it's important to bureaucratic institutions – after you have that concept, I use Seeing Like a State as a pointer to it (rather than because it is a fascinating deep dive into forestry management in Germany in the 1700s).

My funny anecdote about names is that, way back in the day when Japan was going through a similar process of needing to bureaucratize and come up with relatively persistent identities for the purposes of the state, the nobles had two names; the peasants largely had one. 

So the noble names were flowery and poetic and then for the peasant names you ended up with names which were often reflective of the area in which they lived, much like English names – Underhill, for example. Guess what? Someone was once under the hill. 

These are some of the most common names in Japan – Tanaka, “in the middle of the rice field.” If you meet someone named Tanaka-san in the present day, it implies that at least one ancestor of theirs was (1) not in the social class that had a nice flowery name comparing them to e.g. a hawk, and (2) was presumably living in the middle of the rice patties.

[Patrick notes: And here you can listen to a salaryman doing a real-time tapdance about “Oh shoot what is an example here that will not be read as implicit commentary on any influential person heavily identified with their surname.” Tanaka-san, on the other hand, is a socially acceptable Tom, Dick, and Harry.] 

Byrne Hobart: This is an area where Google has decayed slightly, since they do seem to weight clickthroughs pretty highly on search results: it used to be many years ago that if you Googled name origins you got something pretty reasonable. Now they're all really flattering. You can't really find an unflattering name unless the name is literally a foreign word that has a negative connotation.

Byrne Hobart: Hobart now – I don't remember what it is, something like “Mighty impressive warrior” or whatever – but when I first googled it and looked up the origin of the last name it was, “it just means Bart's Hill.”

I guess Bart was somehow a really impressive guy in seventh century England or something, such that even after he was gone they still remembered which hill was his – and he was just a cool enough guy that you would name yourself after not even Bart, but his Hill. 

It is interesting to trace a lot of those last names – then you have the obvious career-based ones. I suppose over time we will eventually start running out of last names, and we'll have to start coming up with ways to invent new ones.

Patrick McKenzie: So as long as we are here in Berkeley, which seems to be in some ways the physical and intellectual center of the rationalist community and perhaps effective altruism – I think both of us are, “I not a member of the club, but I have seen what you've written on the Internet a time or two.”

But let's talk about the epistemology of organizations and how some perceive the world in different ways through institutional culture in vastly different fashions. You've had some experience working with hedge funds and around them; how do hedge funds perceive the world?

Byrne Hobart: Hedge funds in their modern incarnation are machines for looking for deficiencies in other people's model of the world that can be expressed through trades. That model has, has very much evolved – at least for the largest hedge funds, it's evolved towards a setup where, if you look at asset classes, you can see that they have different risk and return characteristics, and then within those asset classes, you can make other judgments. 

You can say things like, let's say, very low-rated bonds are much more sensitive to recession risk, and highly-rated bonds are more sensitive to interest rate risk; you can say that, typically, best-performing stocks will actually continue to outperform, as will worst-performing stocks; that typically statistically cheap companies do a bit better over time than statistically expensive ones; that industries correlate and industry membership explains a large share of a given stock's performance, et cetera.

You can enumerate all of these factors that are just broad statistical explanations for where returns come from, and that allows you to look at someone's investing track record and identify, how much of this was that you picked good stocks? How much of this was that your career happened to span a bull market? How much of this was, not only was it a bull market, but the first job you got happened to be analyst at a tech fund, and tech did unusually well in that bull market? 

We run these regressions and find out, “okay, you, you beat the market by five points a year. It turns out that 5.5 of that was luck, and the other negative 0.5 was skill.” Someone actually did this with George Soros's investment record and found that his skill contributed negative two points, and that following trends in currencies was just a really, really good trade to run at that time. 

I think there's still, there's still something valuable in having implicitly done the regression in your head and actually somehow instinctively identified this systematic signal and executed well.

Patrick McKenzie: I think there's probably a sort of unscored pregame in which you look at every opportunity available in the world and somehow through “luck,” select an opportunity where, go figure, the beta in that opportunity, the returns to the market exceed returns available in other markets during those years. 

Maybe you sub-optimized with respect to how you executed on that opportunity before you, but you picked very well on which opportunity to spend a portion of your professional career going after.

[Patrick notes: I often feel this way about commentary on how geeks are “lucky” to have had their special interest be extremely valuable to e.g. tech industry.]

Byrne Hobart: Yeah, I think that is a reasonable way to look at it, especially in earlier history, but As we have more data now, at least within financial markets…

It is very hard to time these factor performances – very hard to time, “when will this industry do well or worse, when will momentum work unusually well or worse” – I'm sure people try to do it, I'm sure some people are good at it, but if you construct a portfolio where you're netting out exposure to all of those factors, what you have done is you've created a portfolio that is just a measure of someone's skill at identifying the idiosyncratic return drivers of individual stocks. So if they bought NVIDIA, they also had to short a corresponding amount of other large companies, other growth companies, other tech companies, etc. such that, if they make money on NVIDIA after all that hedging, it's because they actually knew something right about NVIDIA.

What that ends up meaning is that the hedge fund – we've actually made the full circle. People used to knock hedge funds as “a compensation scheme masquerading as an asset class,” and as they've gotten better at building these hedge funds, market-neutral, factor-neutral portfolios, they are increasingly a method of measuring investment skill masquerading as an asset class. 

Because, what do you want? In theory it makes sense that you should be able to charge a lot of money for skill, and you should not be able to charge very much money for, “you happen to get a job analyzing an industry that happened to do well over the time when you were a portfolio manager.”

So it means that as hedge funds have gotten better at just just delivering that idiosyncratic return, and [with] the accumulation of a bunch of different portfolio managers, who are finding a bunch of different ways to extract the “idio” from a bunch of different sets of companies, you can charge a lot more for that – which means you can pay people a lot more, and so you can bring in more talent to the industry. 

That model keeps on growing, but it does become a model where you as an analyst or as the trader or as the portfolio manager, you are constantly asking yourself questions like, “why do I deserve to be right about this?” If you have a reason to think this is a good company, what is the reason that someone else looking at the same evidence didn't think so? 

Sometimes the reason is you looked at more evidence than they did – they talked to five people at private companies that order lots of GPUs, you talked to eight people, you have a slight edge on the person who worked less. 

Sometimes you just have a signal where you identify, not really why didn't someone else exploit this, but why does this happen in the first place? 

Actually, for all its issues, the book Going Infinite does have some good stories about this and about tracing some market aberration back to, “there is a lazy portfolio manager who just habitually rolls into the office and hits the sell button on this particular day of the week.”

[Patrick notes:

You wouldn't have satisfied the Jane Street system, however, because you didn't know why the Japanese stocks were rising in price twelve hours after the South Korean stocks. And so you looked even further into it–as Sam had. And he found the prices of both the South Korean and the Japanese ETFs were being driven by a single trader at a German bank. Every few days, the German bank trader had a bunch of buy orders to execute, in both South Korea and Japan. He'd make his South Korean purchases before calling it a day, passing the Japan orders off to his Asian colleagues to handle when they awakened [sic] in Tokyo. The Jane Street trader could now happily see the pop in the South Korean ETF and buy the Japanese ETF until the German died, retired, for figured out how much his laziness was costing him.

– Going Infinite, pg 66 ]

[Patrick further emphasizes: SBF has a repeating pattern of telling people stories which are narratively rich and just a little too perfect to check. Lewis’ book always repeats them credulously, including when they fail to cohere with other times he has told the same story.

(For example, when SBF told Odd Lots the founding myth of Alameda Research, the Bitcoin arbitrage trade took place in Japan with a Japanese effective altruist making a Japanese company to suborn a regional Japanese bank. Lewis reports it was in South Korea and a South Korean grad student. Someone in the room said they never successfully operationalized it in South Korea, did in Japan, and that it was far smaller than SBF subsequently told people about.)

Now, there do exist pathways in the world to discover the identity of one’s counterparty down to a specific trader at a specific bank, even though typically one’s counterparty is anonymized (recall my conversation with Ricki on this issue). Perhaps the marginal buyer for ETFs in the world’s third largest economy is not the Bank of Japan (balance sheet: trillions) but is instead a single discretionary trader in Europe (balance sheet: tens of millions). On the other hand, another possibility suggests itself: perhaps SBF made this up.

Perhaps Lewis repeated it because he is credulous, has a certain ambivalence about the truth, and does not understand how stock exchanges work. This would be a damning criticism of a financial journalist. I would not make it had he not written a book-length paper trail explaining that he, in fact, does not understand how stock exchanges work.

Pardon the interruption.]

Byrne Hobart: I've heard stories of that kind of thing going back to the nineties. In the nineties, the Japanese pension funds would get the deposit on payday and then they would buy exactly the same treasuries at exactly the same time, and all you had to do was buy those treasuries right before and sell them then.

[Patrick notes: Not in the top hundred leaks of beneficiary assets at a Japanese pension fund, according to this beneficiary of a Japanese pension fund. The Japanese asset management industry very brazenly transfers value to the financial sector, much like the U.S. asset management industry used to in the bad old days.]

So you're always trying to build this model of the world, and of what you know, what you know relative to other people, what mistakes other people might be making, how persistent those mistakes are, how much competition there is to exploit those mistakes, and you're trying to measure the degree to which your returns are being computed away, and the degree to which – let's say there's some suboptimal behavior that is costing some set of investors X amount of dollars, and there's a set of investors trying to exploit that behavior and that in the aggregate they're aiming to make 2X.

Well, now the number of people who are very cleverly buying is twice as big as the number of people who are indiscriminately selling. Suddenly the opposite of the historical strategy is the one that makes sense. 

You're always doing this kind of introspection and always trying to rigorously measure your own skill as an empiricist. It is basically this exercise in just being a rationalist. It is like they are mentally reinventing the entire LessWrong corpus all the time.

Patrick McKenzie: Putting probability values on things, describing the world with numbers. A surprisingly underused way to describe the world.

[Patrick notes: See Nate Silver versus the entire U.S. political commentariat, Deming versus all manufacturers ex-Japan between approximately 1950 and 1990, etc etc.] 

Patrick McKenzie: So, both of us have one foot in finance at various parts – very different parts of finance. 

Finance is a sticky word, much like tech, is where finance is, depending on how you measure it, 10 to 16 percent of the economy. (Tech is eventually going to be any company at which someone writes or speaks, but be that as it may.)

[Patrick notes: People sometimes use 20%ish as the upper bound and if they do that they are including real estate/rental/leasing, for reasons which make a certain amount of sense in an economic modeling perspective, but will misguide you with respect to intuitions.]  

In tech investing, you hear about all this rigor and world modeling that goes into hedge funds and think, “ah, and are our spreadsheets that we keep in venture capital funds equivalently rigorous? Are the equivalent number of user interviews done? Are we steeped in every element of the supply chain?” Et cetera, et cetera –  and I'll say, as someone who thinks that venture backed software companies create a lot of value in the world, “yeeeeah, no, we don't do it that way.”

Byrne Hobart: That's true. Some of that does actually come back to the measurement problem. I wrote a piece on this a long time ago: measuring venture investor skill is one of the hard problems in finance, and may never be solved because, if you are in a power law kind of investing situation, you have these long lags between when you write the check and when you get a wire for a much larger amount to your account. 

Because of that, not only do you have a fairly small sample of successes, but the more successful you are, the more likely it is to be from one really, really big thing you did, which means the more successful someone is, the easier it is to claim that they were lucky. And that just makes it a really frustrating business to analyze and understand.

It also means that it’s very hard for a venture investor to think about the marginal cost of doing one more interview or buying one more data set or something like that – whereas with a hedge fund or a prop trading firm… I don't know that any of them explicitly measure things like “what is the marginal value of this analyst spending the next 30 minutes reading a transcript of an interview or editing the scraper that we're using to track the inventory in this API that the company does not realize is actually exposed to the public internet.”

They don't measure it quite that granulated, but they have a pretty good sense of what is the incremental return on the next action, and they have pretty high confidence in that. 

You don't have that. You don't know if the next call that you take is from someone who's starting the next Stripe – the odds are very, very low, but the odds are non-zero, and you will never actually have enough data to be anything like confident in that. 

Patrick McKenzie: Can I give you a fascinating anecdote for interesting ways in which unexpected data comes in the venture capital process? 

There was once a company which had a direct competitor. The company was close to me; the direct competitor was close to me in a different fashion in that I was a user of theirs. 

The person close to me at the company said, “Hey, we're raising, and do you have any point of view on how big the business of our competitor is? Because that would be a very useful data point to share with VCs.”

And I said, “Would you like the growth graph for the last several years?” 

Now, if I had received that growth graph from them dishonestly, that would obviously be a very inappropriate thing to do. What I actually did was said, “all of my invoices have been sequentially numbered, and I was one of their first customers – I get one invoice every month, and I can tell you exactly, ‘here's the Excel model for how many customers they onboarded in each of the last 36 months.’ Any questions?” (This did not result in a job offer from a VC firm, but again, VCs don't work that way.)

[Patrick notes: I’m a very occasional angel investor. Many early stage startups use e.g. Ruby on Rails, much like I did at three companies I ran. Ruby on Rails promiscuously displays sequentially generated database IDs to unprivileged users. I have not frequently invested in companies which claim 100,000 beta users and 5,000 paying customers then assigned me user_id 1186 when I signed up to kick the tires.] 

Byrne Hobart: There has been a lot of money made looking at the sequential receipts. There has even been money made figuring out what the hashing algorithm is for the no-longer-sequential receipts.

Because often there’s an int being incremented on the backend, and then there is some function being run on that so that the number is obscure. But sometimes if you have enough receipts, you might have enough data to figure out what the pattern is.

Sometimes there's other information embedded in things – a lot of information just leaks by the nature of businesses that are operating online, that have APIs, that are servicing lots of customers. There are just lots of times where they're sharing stuff. and sometimes it's just really brute force, like scraping every open table page and seeing how many tables are available, how many have been taken. and people – people scrape lots of different marketplace-type sites and look for things like that.

Patrick McKenzie: Scrapers are probably only going to get more powerful in the next couple of years. Scrapers are sort of the classic project for programmers. I was building them in college and have done any number over the years. There's always a huge amount of effort put into relatively low value tasks — parsing the DOM, etc. — and then it breaks every time the site that you were trying to scrape upgrades. When you're trying to scrape a live player in economics they might have an engineering team doing updates very frequently.

But now we have computers that can actually read trivially. Sometimes if you give them a document, they might hallucinate, but frequently when you're scraping a very large dataset you don't need to get every fact in that dataset exactly right, you need to really care about what the delta was from yesterday. Fascinating things will happen. 

Scraping can be done for good. We did substantial amounts of scraping at VaccinateCA, for example, to help the nation in porting information about life saving healthcare availability between various IT systems that did not natively speak the same API. But for those in the world who have a proprietary edge because they are better in writing scrapers, that's going to be a tougher business when robots are competing with them, having infinite ability to scrape with their newly-available eyeballs.

[Patrick notes: Simon Willison, who worked with us on VaccinateCA and also invented Django, has written and spoken extensively about practical scraping. One of my favorite presentations of his shows how to effectively generate an auditable time series of data off of many public-facing websites, piggybacking on Github. It takes less than five minutes to set up and is free to do.] 

Byrne Hobart: Yes. I think it was an interview with Ryan Petersen at Flexport, I think at the end of last year, where he said that they have a lot of scrapers because ports have sites. The sites tell you where the container is and sometimes the site changes or something breaks, et cetera, so they have LLMs in the loop where if the scraper has an error, they're automatically doing a call to GPT-4 saying, “I'm trying to find this piece of information on this site and here is the code that I'm using — fix this code so that it does what it's supposed to do.” 

I don't know what the hit rate is on that getting it right the first time, but it really helps. One of the annoying things about scraping is that in some cases you are looking at cumulative things, especially if it's something like order numbers and things like that. But sometimes you're actually looking at some continuous change, you’re counting the number of times something happens. If your scraper breaks, you lose part of your time series and you never get it back. In that context, it is really, really valuable to be quick to fix your scrapers.

Also you need some kind of econometrics-flavored skill, the kind of econometrics that you would do if you read an econometrics textbook and then tried to apply it to real world data, where the actual problem is the data is not good enough for the models — that kind of econometric skill you do need to figure out what would the missing data actually look like.

Then you need the organizational skill of being able to say, “this is real, this data point is supposed to be fake” — and if you need to be looking at real data, you need to exclude it.

Patrick McKenzie: Often people who are trying to get up, up to speed with an industry or up to speed with a particular company for the first time have very different approaches in how they do that. I'd ask you: how do you start learning about fields which are novel to your experience?

Byrne Hobart: Yeah, so I'll start with when I was at a hedge fund, the, the most fun parts were when something really significant changed in the industry and you needed to be an expert in something you had not heard of the week before.

That was just an electrifying experience, to go from not knowing anything about it to, because I was the person who was tasked with learning a lot about it, knowing more than any of my competitors about it in very short order. That was great. I've kind of arranged my life so that I have more excuses to do that. One of those excuses is, sometimes companies will go public, and the numbers look interesting, but I don't know what the company actually does, so I want to figure it out.

I do think that actually reading prospectuses, reading 10Ks is a really good way to start understanding an industry, and as you have seen with the brokerage industry, it is a really good way to find out that many people who cover them for a living have not actually read a 10K and do not understand what their business is.

Patrick McKenzie: This is enormously frustrating to me.

I have a piece which (as we say in the trade) ‘aged well,’ about how discount brokerages make money. I very confidently said in that essay that obviously commissions are going to zero because this is not material for almost any discount brokerage. Then lo and behold, the cartel stopped charging for commissions a few years later, and people were like, “wow, great call.”

I said, “I don't know that it was a great call. I think that anyone who thought about that industry for more than an hour or two saw the handwriting on the wall.”

[Patrick notes: I have received some criticism from describing discount brokers as forming a cartel to continue trades being a priced service, then all abandoning it in effectively a collusive fashion within weeks of each other. On the one hand, as Byrne mentions elsewhere, perhaps this coordination didn’t require a smoky backroom and was instead everyone individually responding to incentives and being unwilling to be the first to blink. On the other hand, putatively serious people have suggested to me that they believe that no one at a discount brokerage seriously considered the question for more than an hour or two at any time in two decades. My estimate of the truth of that is zero.]

Patrick McKenzie: So, I love the idea — not merely the idea, but the practice — of following quarterly reports. When people ask how banking works, I would point them to relatively simple banks and say, just read these, read what seems to be important to them in, what they choose to emphasize, and you will get a great model for it.

I used to point people at First Republic Bank, sadly no longer with us, because it was a relatively straightforward bank, and you could get a sense for what is in a straightforward bank’s report, then work your way up to understanding institutions that are just structurally more complicated.

One thing that I often do is just develop a list of people to chat to. Sometimes it's because they write interesting things on the internet or have been citing interesting things on the internet. A professional skill that I think is extremely valuable to a lot of people — which we don't teach for some reason — is, “how do you identify experts in a field in which you are not, in fact, an expert?” 

One way is to do things like go to YouTube – you might not have attended conferences yet, but a lot of conference talks end up on YouTube. You could spend a lot of time watching talks; if they're in an industry where you don't quite know the lingo yet, that might go quite over your head. But conference speakers are often chosen by an algorithm that is not random: it is the combination of (1) who will actually agree to speak at the conference, but (2) who does the conference organizer reach out to with the expectation that putting this person on the speaker's page will sell more tickets?

They're socially esteemed in the industry, or there's some other reason that people in the industry think that their firm is important (or they've bought their speaking slot which happens a nontrivial amount of the time).

So you can pivot from, okay, the industry's self assessment of important folks includes these folks — do any of them have web presences? Who are the people they cite? Who are the people who seem to be upstream of them in the information flow? Also, many of them have email addresses, sometimes in the same conference talks; maybe you could write them and say, “Hey, I'm interested in your industry. What are the first things I should look at? Would you have a coffee with me about it?”

And that is an extremely repeatable algorithm for learning a little bit about anything. 

Byrne Hobart: I definitely find that that kind of thing of finding someone in the industry I can talk to and talking to them is really helpful, especially because if I've read the 10k or the prospectus, I know how the company presents itself to investors — and the lawyers will make sure the company is not actively lying, but the company wants to make sure that it looks as good as possible. There are many cases where the data-driven, AI-powered customer service platform is 7,000 people in the Philippines, and there's no one actually doing AI research at this company. They may use ChatGPT, but it is not an AI company. It is a business process outsourcing company.

This happens reasonably often, so I like to read enough about a company’s public-facing presentation to investors that I can ask some dumb questions and get my misconceptions out to the expert as quickly as possible; often if you have a misconception that comes straight from the company, they can give you a knowing smile and say, “yes, that is what they like to think, and that is what they like to say — but in fact, here's what they do, here’s how they make their money.” 

There are some industries where the money is just made in a different place than you would think, and it's even not obvious to people in those industries. Online travel agencies apparently, when they started — there was a really good Skift oral history of Booking.com about how, when that industry started, when people started buying travel online, you actually made a lot of money from the plane tickets.

The hotel thing, it was just too hard to get enough hotels on the platform, and it was hard to get them to pay money, they didn't really understand the internet; whereas the airlines did understand selling incremental seats, and their unit economics certainly justify working hard to do that.

So, early on, that was how travel was monetized, and then it completely flipped; now the booking agency, Booking.com or whoever, makes just a token sum on deciding whether it's Delta, United, or American that's going to fly you on this route that all of them do, but they do make a lot of money on the hotels because they have aggregated a lot of hotels. The focus that I heard in that industry a while ago was that the best predictor of the revenue that you can get from a click to a page like a search page is how many listings are there on that page — the more listings you have, the more likely you are to get the business, the more likely you are to find the right hotel for someone, and there's a really good feedback loop there where you're getting more revenue per click, you are buying more clicks from Google; now you have a larger business that can justify a larger salesforce that can sign up more hotels. It just goes on and on and on. 

It also means they want to expand in geographically-adjacent areas and areas where the go-to-market is really similar. So you could work with – you know, large chains are hard, they are very good at negotiating; smaller chains just have less pricing power, so you can go that route. 

Or you can go the exact opposite direction and say, “we are going to sign up every single family-owned bed and breakfast in southern France and then from there we will expand into Portugal and we'll do Italy, etc.” If you do it that way, you're always getting better and better unit economics in the place where you have established that foothold, and then you can just continue to sign up these hotels. 

Then someone else who wants to do that, they know that it will take them forever to get to parity — and meanwhile, the customers know which place to search for first if they want a small, independent hotel. So you could spend an enormous amount of effort getting to something that is 80% as good and earns 20% as much as your competitor, and they probably spent less money doing that because the cost of customer acquisition was just lower back then. 

So it creates these little monopolies, and then they bump into each other — everyone’s going to want to give you a Marriott in a major city, so it has to be on all these platforms.

Patrick McKenzie:

From Marriott's point of view, to bring everything back to credit cards because my life has revolved around credit cards for so long, one of the main reasons that there are rewards programs at all at hotels is to encourage you to book direct with the hotel.

Oh, those magical words — they've appeared in so many boardrooms and strategy presentations, and they're in the physical built environment of hotels. Next time you stay at a hotel, look around for the signs that say, “remember to book direct, book direct, book direct” because they don't want to leak 30 to 50 percent revenue for the room to the OTA. (I don’t know if those are good quotes on the revenue split.)

Interestingly —and this is the thing that people would learn if they read the 10K or quarterly reports for their hotel chain of choice — hotel chains are also essentially travel agents with respect to the individual hotel operators (who are not the chain themselves) because hotel chains want to be an asset-light model where they are a marketing front, and the individuals who are running hotels, hiring people to clean them, etc. are in a different and very real estate-inflected business.

Ironically, bringing things back to tech, many of the hotel chains that Airbnb is competing with are essentially competing on a very similar model. Those chains are a web UI, and some financial infrastructure, which connects independent owner-operators to travelers.

Anyhow, so Byrne, where can people find you on the Internets?

Byrne Hobart: That is a wonderful question. So the best place to go is thediff.co. (It is named after the Unix utility.) I'm also on Twitter, which is theoretically X, but is actually Twitter. Yes, it's my full name @ByrneHobart. So find me on either place and sign up for the newsletter.

[Patrick notes: Agreed here: can pry “Twitter” from my cold, dead hands.]

If there's something interesting that you were working on building, thinking about, speculating about, please send me an email. I enjoy those.

Patrick McKenzie: Thank you very much. All right, Byrne, I will see you around the Internet, and thanks everybody for listening. Bye bye.