不只是数字游戏：会计简史（Fintwit Book Club 2025年1月）

All right, hello. Today I’m Andrew Walker, host of Get Another Value Blog. Today I’m happy to have on one of my favorite authors, Byrne Hobart. I say “have on”—this is just a test take. Byrne Hobart from The Diff.

We’re thinking about doing a monthly Finch Wood Book Club. We read, on Byrne’s suggestion, *More Than a Numbers Game*. We’ll include a link in the show notes, but Byrne, I’ll toss it over to you. You chose *More Than a Numbers Game*, so why don’t you quickly—I love having the hard copy—say why you chose it, your thoughts on the book, and everything?

Part of the reason I read this book was that I was reading *Supermoney*, and it mentioned it. That book was written in the early 1970s and was by the same guy who wrote *The Money Game*. It’s a really fun book. Among other things, it has a really early interview with Warren Buffett.

It talks about how investors vary in how sophisticated they are. One of the things it says is that some investors are so anxious to do the work, do the research, and know their companies that they don’t just read the annual report; they literally go to the SEC and read this weird thing called a 10-K. It has the stuff you won’t find in the annual report.

The other thing it said was something to the effect that some investors don’t even wait for the annual earnings number; they literally look for the quarterly numbers instead. To me, this is like saying some investors calculate what they think the stock is worth before buying it, or some investors use limit orders. This is really, really obvious stuff.

But if you go back far enough—if you’re reading 19th-century trading anecdotes—it’s totally reasonable to think, “Okay, these people did not really know what earnings were.” Did they know what depreciation was? Maybe, maybe not.

I realized we’ve made a lot of progress in accounting. Even though there’s probably more progress to be made, we’re in a much better situation than we were in the distant past. We’re in a much better situation than we were when I first started looking at financial statements in the early 2000s, and I wanted to know where that had come from and how it happened.

I also had this general sense that a lot of accounting judgments seemed superficially wrong until you try to come up with a better alternative. Then you decide, “Okay, this is maybe not great. Maybe we could have done it another way, but this is actually a pretty good idea.”

I think the first accounting change that I remember reading about was the end of goodwill amortization. There was a Buffett letter, when I first started reading Buffett letters, that talked about this change, and Buffett thought it made sense. If you were to acquire all of Coca-Cola, it wouldn’t make sense to depreciate the Coke brand down to 0.

But it also struck me that this stuff is kind of arbitrary. The financial statements—the cash flows are not changed, but the numbers that you as an investor see do change. You and I talked about this a little bit over email just before the show: sometimes the numbers investors see, even if those numbers don’t really correspond to cash flows, affect their decisions. They affect how they value the company, and that affects how the competitors of that company behave.

Good accounting is actually very, very socially useful because it’s not just a scorecard for any one firm. You can take the libertarian view, and I think it’s fine: if you run your own business, do your accounting however you want. If you happen to run out of money and you didn’t realize that was going to happen, that’s your loss, and maybe you learn why some accounting principles are generally accepted and some aren’t.

But if you run a public company, I think there is some obligation that, if people are going to look at your financial statements and respond to them, it’s very important to society that these financial statements accurately reflect reality. This book is just the story of how people made that happen.

I’m just going to yes-and a few things there, but first I’m going to divulge something. Sometimes I’ll get emails from listeners that are like, “Hey, I love the podcast, but when you talk, I need to listen at 1× speed, and when your guests talk, I speed up to 1.5× speed.”

I think you and I might be the first example of 2 people who are so passionate and fast-spoken that people might need to turn this down to 0.75× speed if they’re going to do it. To yes-and everything you said, I really liked everything you said there.

I love the point you made about people who used to go and read the 10-K. He’d be like, “These crazy people will go and read the 10-K instead of just the annual report.” It reminds me of how, in the 1930s, before Ben Graham came around, everybody was just buying stocks like crazy. Then Ben Graham said, “Let’s calculate the asset value.”

Today, if you’re like, “I’m buying this stock at 8× price-to-earnings,” I’ll always say, “Unless you’ve got more insight behind that, you’re probably going to get your face ripped off.”

But let’s go with the last thing you talked about. I love that point because the anecdote—I think my favorite anecdote in the whole book—is WorldCom. It’s toward the end. WorldCom is capitalizing its line costs, and because of that, it looks much more profitable than AT&T, its competitor.

I believe the book says AT&T looks like a lazy dinosaur or something. AT&T, which is expensing its line costs, looks staid. If you capitalize something, you basically divide by 10, 20, whatever, and you’re going to report a much higher margin than someone who’s expensing.

AT&T fires 20,000 people and buys a ton of cable companies for over $100 billion and almost destroys itself. I’ve never heard of a company almost getting destroyed by a competitor’s fraud before, so I thought it was really interesting.

I call management teams all the time. I’m like, “Hey, your margins are 5%. Your competitors’ margins are 8%. Why the discrepancy? You need to get up there. You’re lazy. You need to fire people.” It really drove home that fraudulent accounting statements can mess up a company, but they can really mess with other companies, competitors, everything. It has a real impact.

I’ll pause there. Anything you want to say on that?

Yeah, I think that’s true. There’s this weird dynamic. I recently read this anecdote about the story of the Mechanical Turk—not the Amazon project, but the thing it’s named after—which was a fake chess-playing robot. There was really just a guy under the robot moving the chess pieces.

Apparently, there was a British inventor, Edmund Cartwright, who saw the Mechanical Turk and fell for it. But he thought to himself, “If it’s possible to build a machine that is sophisticated enough to play chess, then surely you should be able to build a machine that can weave.” So he built an actual power loom, an early prototype of a power loom.

Sometimes fraud does force people to step up their game. They’re like, “How can we possibly beat these people? We have to work much harder.” They do work much harder. It turns out that the fraud was very slight.

That is also something just worth noting about a lot of accounting frauds. There are a handful of frauds where the plan from the beginning is, “We’re going to lie about our numbers and rip people off.” But the way a lot of them happen is that, just this one time, just this quarter, we’re going to front-load just this one transaction and hit our numbers.

We know we’re capable of hitting our numbers, so it’s not really a big deal. In fact, some of the 1990s accounting frauds started with hiding profits. Enron did some of this, too, because its trading desk was so wildly profitable. First, they wanted the cookie jar so they could offset future write-offs.

Second, they just really didn’t like the optics: California is in the dark because we’re milking them for all they’re worth.

Absolutely, yeah. Yeah.

Yeah, so I think generally it is corrosive to have different yardsticks that look like the same yardstick from the outside but are actually measuring different things. I think that is part of the purpose of accounting, and the book talks about this: you want to be able to compare different companies.

You don’t just want the CEO of the New York Central saying, “We’re the best railroad. Look at our dividend,” and the head of the Penn Central saying the same thing. You actually want to have some way to quantify who owns what, what return they’re getting, and what is driving that return.

I think it does come from a really good place, but accounting is an attempt to take a really messy world of abstractions and turn it into something concrete and rules-driven. You want a situation where, if you had 2 accountants prepare the same firm’s financial statements, they would have exactly the same number.

I don’t know if anyone has ever tried that, but I assume that there’s a very small limit of company size after which you basically never get the same number from 2 different people.

It's funny because one of the through lines throughout the book is that Arthur Andersen is mentioned 5 times, and it's mentioned tons of times. But 5 times, they say Arthur Andersen was the place where they most wanted every accountant to give the exact same number: no discretion; everyone plays by and follows the exact same rules. And when you said that, I think that might have been in the '90s, that was the place where everyone was most likely to get to the same rules.

And part of that—and there was obviously 100 other stuff—but it clearly seems like that desire and that culture is ultimately one of the many things that led to their demise. I put it at, like, if there were 100 issues, I'd say it's probably number 6 on the list or something.

Yeah. I think what happens is that any accountant—any good accountant—would say that accounting approximates economic reality. It does not perfectly correspond to economic reality, so there will always be a gap. And if they say that, then part of their discretion is potentially giving the company more credit than the strict application of the rules would do.

And some of it is saying, "Look, guys, it is very obvious that if you own just enough of this special-purpose entity to not consolidate it on your balance sheet, and then you're doing business with that same entity and it's collateral for the derivative rights—you have a derivative on your own stock that you're also explicitly designing so you don't have to report that—you'd have to look at that and say, 'This vehicle has no purpose other than moving risk off the balance sheet, but the risk is still there.'"

But if you are sufficiently rules-driven and it complies with every single one of the rules, and Enron has hired really good accountants of its own who know exactly what the limits of those rules are, then maybe you are stuck saying that, just per our ethics statement, we absolutely must approve this particular instance of lying to your shareholders.

You know, when you say that, it's like they hire really good accountants who know how to take the rules to the exact limits. If they had come along 20 years later, they probably would have been so rich trading crypto or prop trading.

For a very specific example, I think about the question, "Will the U.S. government shut down?" on Polymarket. When they reached a deal, I think it dropped toward "no" for a second, and then enterprising traders read the fine print and said, "Oh my gosh, the U.S. government will technically shut down for an hour, and this is a yes." Enron's accountants would have been so perfect at that. It's like, "No, in one life it's fraud; in the next life, they would have been great prop traders."

Yeah, although I think that you can almost view prop trading as the prosocial application of the same skill that allows you to commit accounting fraud. Some prop trading is that you find an interesting pattern and you find the most cost-effective way to exploit it. Sometimes it is that you reason really carefully about how the asset you're trading is structured, or how you yourself could structure this trade, and you find the one detail that other people are not thinking about.

The one thing where, "Hey, if you don't pay attention to the correlation between these 2 assets and a derivative that touches both of them is worth X, and if you do pay attention to that correlation—or you pay attention to how that correlation would change in different scenarios—then the value of the derivative is something totally different."

So they are sort of doing this distributed bug-bounty program where, if someone has put together an asset that just doesn't make very much financial sense and is inevitably going to blow up, the prop traders are the ones putting some of the downward pressure on that asset's price, and they do keep things in a slightly better equilibrium. So, in that adversarial scenario, that skill set is actually really valuable and does make the market more efficient and does keep people from making dumb mistakes, but in a scenario where you're cooperative, then it's tricky.

And actually, the book talks a little bit about this question of who's going to pay for the accountants, who's going to pay for the audit, and how that question has evolved, and there are just no good answers. If investors pay for the audit, you either have 1 audit per investor and then you have a ton of duplicated work, or you have 1 investor pay for the audit and then everyone uses the audit, in which case the biggest investor is basically paying a tax to keep the rest of the market informed.

And so maybe the least bad option is you have the company pay for the audit, and you make sure that, over time, investors have just learned that there are auditing firms that are a really good signal, there are auditing firms that are kind of a neutral signal, and there are auditing firms where you actually want to have an alert. You want to scrape or subscribe to the RSS feed of new S-1s and have an alert that triggers every time a particular auditor is involved in a company disclosing publicly, and then you immediately add it to one of those watch lists that displays the borrow cost right there. You know what's going to happen.

Yeah. I'm laughing because, in your scenario where the top investor pays for the audit, you could have the scenario where Byrne owns 100,000 shares, I own 100,001 shares, and then I'm like, "I sell down to 99,000"; you sell down to 98,000 because we're trying to pass that bill back and forth.

I like also your audit costing, right? It's one of the 3 themes of this book and one I really want to talk to you about: how cyclical finance can be. And audit cost—hey, who should pay, the investor or the company? Obviously, the company pays; there can be mismanagement.

This book was released in early 2008. How much does that remind you of the rating agencies' issues that would spring up literally as this book is getting published, where people were debating, "Oh, isn't it kind of strange that the companies pay rating agencies for their ratings, and then investors, especially credit funds, rely on it?"

And so we can talk about that, or I was just struck by there are 1,000 other examples of, "Hey, in the '20s, people are relying on dividend payments to estimate earnings." That sounds a lot like yieldcos in the 2010s to me. In the '70s, one of my favorites is that banks come to the accounting board and say, "Hey, if we've got a loan that's had credit issues, but we can modify it so we still expect full repayment, can we not take a write-off because it would destroy our capital base?"

It had so many rhymes with the held-to-maturity things that destroyed Silicon Valley Bank, or even some of the mark-to-market issues in the 2000 crisis, the 2008 crisis—wherever you want to go. So I want to ask you: what did you think about those through lines and that cyclicality? And I'll have a counterexample to that to follow up on.

Yeah, and I think another thing that was not structured this way but was economically equivalent to this was sell-side analyst compensation before the 2003 settlement. In effect, the companies were paying for coverage because the sell-side researchers get bonuses based on the underwriting that their bank does, and that underwriting is partly a function of whether they put a buy rating on a company that issues lots of stock and lots of convertibles.

And so, in effect, if you look at just the flow of funds, the flow of incentives, it is pretty economically equivalent to companies paying for coverage, and they pay for good coverage. And that didn't work, but what it did mean was that there was a lot of coverage and it was pretty widely available. And now we have the situation where it's much harder for firms to engineer things so they get really good coverage.

To the extent that they can, they're actually paying in kind by doing management meetings at conferences hosted by the banks whose analysts they get along with, and it's just really easy to get along with someone who thinks you're a swell CEO whose stock is thought to be undervalued. It's just pretty nice to have something like that in common with someone.

So now we have a system where the institutions are paying for the research, and it turns out that they're perfectly happy to pay for research, most of which is distributed internally. So there's still a lot of really great equity research, but it's not very far out there. It gets forwarded to half a dozen people within one firm; they make the trade; that's how the research is monetized.

So, in some ways, when you have a setup for paying for information and the information being widely available, if you align incentives really well such that there's a really strong incentive for that information to be accurate, you're also creating an incentive for that information to not be widely distributed. And I think in the case of the firms just paying their auditor to audit it, to have some stamp of approval, part of what that does is it means that the auditors know that their value to every other company is contingent on their willingness to say no to whatever client is closest to the line.

And it seems like the structure of the industry doesn't really support there being multiple firms where you can actually stack-rank their reputation. It's just really hard to be the second-most reputable of the big firms. It seems more feasible for there to be this cluster of big companies, where it just means you've got to be a big company, and then smaller companies, where it means some combination of, "Yeah, you just can't afford to pay EY that much money right now; it would not be a good use of shareholder funds."

But, 2, maybe you can’t afford to have someone from EY looking at your books and asking themselves whether or not it would be good for their firm’s franchise if your firm got a clean opinion.

I think you wrote up Supermicro. Did you see Supermicro? That was one of the most fascinating things. For those who weren’t following, Supermicro was a growth darling. I think I calculated it: the Russell was up 10% last year, and 1.5% of that, I think, was Supermicro. From January to March, Supermicro’s stock increased 5× or something, and then it graduated from the Russell 2000 to the S&P 500.

The Russell 2000 kicked it out, and then Supermicro had all these accounting issues—tons of accounting issues, insider trading, a short report, all this sort of stuff. The stock dropped 80%. It’s funny, because it had me wondering: if it had been in the Russell 2000 for the entire year, I think it would have dragged the Russell 2000’s returns from 10% last year to 7.5%. But neither here nor there.

The interesting thing for the discussion you just had was that Supermicro’s auditor resigned in kind of fiery auditor fashion. It’s one of those things where you read it, and if you didn’t know anything about accounting, you’d be like, “That’s normal.” But if you had read a hundred audit statements, you’d be like, “This is the one where they’re saying, ‘We think you can rely on paragraphs 1 through 7, and also paragraph 9, and paragraphs 15 through 20,’” or something. They were just saying, point by point, “This is the part you should not trust.” That’s pretty bad.

The other thing was that it was a Big Four firm that resigned. Supermicro is still a $50 billion enterprise-value company, but a Big Four audit firm resigned while the company was on fire and with that dramatic letter. Who can step in? If you’re the firm that steps in, is it because you think you’ll take more risk around the edges than the other firm? Are you flat-out saying the other firm was wrong? Are you creating a reputation as a monetizable place—“Pay us a little extra and we’ll let you skirt around the edges”? That might be a disaster for the rest of your business.

The solution probably was obvious in hindsight, but it was a crazy example of everything you’re saying. If there’s the Big Four and then there’s the Mid-Tier 10, one of the lower-rung Mid-Tier 10 firms, I think, took it. I’ll toss it over to you.

I think in a situation like that, if there are that many outstanding accounting questions, the thing to do to make the business a viable large-cap again is to hire the new auditor to confirm that everything the previous auditor said was right, do a massive restatement, and pay that new auditor a lot of money. You are going to drop them for a Big Four firm as soon as you have a chance to do so.

If both sides understand that, and if that Mid-Tier 10 company knows that, for the duration of that client relationship, it at least gets to say in its presentations to other clients, “We audit 1 member of the S&P 500, and we’re hoping to expand that franchise,” then you hope you don’t get too many follow-up questions. But it is really tough to get out of accounting messes, and I think a lot of them do start out with just honest mistakes.

Part of what the book is describing is what companies did internally. That’s another point to bring up about the book: part of what makes modern accounting so annoyingly complicated is that there are actually 2 schools of thought that got merged into 1.

One school of thought was, “I am a London-based investor, and because I like growth and excitement, I am lending money to the hot tech stocks of the day, which are American railroads. Because I don’t want to be sailing across the ocean every so often, I need accurate financial statements. I need to know what my money is getting.” Since I’m a bond investor, what I care about is: I put $1 million into this company; do I have $1 million worth of assets? I know that, as long as that’s true, my investment is pretty secure. Yes, there will be cash flows, but I don’t have access to the upside of those cash flows. I just get the money that I’m contractually guaranteed to get. So they cared about that.

The other piece was controlling costs and understanding product mix, usually from the perspective of a manufacturer. You are General Motors; you make a bunch of different kinds of cars, they all use different components, and you just want to understand: Should this assembly line stop making this model and make the next model? What is the payoff if I make it out of this material versus that material?

Those are really different questions. One of them is looking at a stock of physical assets, a set list of assets, and the other is looking at a flow and the interrelation between those flows. But you do need to answer both, because the stock is the accumulation of flows over time, especially if the company is not paying out all of its earnings as a dividend.

They had to merge those 2 schools of thought, but it’s just 2 different groups thinking very differently. One group is thinking about maximizing profit and upside, and the other is thinking about controlling downside and having some liquidatable collateral that backs some loans.

If you produce a series of balance sheets, you have to link them with P&L and cash flow statements. You kind of have to do both, but if they start with different assumptions and different goals, then it’s tricky to tie them together. That is to say that, when you look through this history, there are a lot of cases where people had to make judgment calls and think about those judgment calls.

One of the pieces I really liked was the anecdote about when the US government was giving companies a tax credit for capital investment. There were actually really good arguments for the 2 different accounting treatments. One was, “You bought a $1 million machine, you get a $60,000 tax credit, therefore you actually spent $940,000 on this machine. Depreciate it as if that’s what you had paid.” The other school of thought was, “No, you got a $60,000 check in response to something you did for your business. That’s revenue.” You can see it both ways.

It was also true that one of the reasons accountants do tend to choose the more cautious of 2 equally defensible viewpoints is that you never want to give someone an incentive to do something more aggressive than they really want to do. You never want to give them financial statements that reward maximum risk-taking and reward it similarly to the outcomes of more prudent decisions.

I think that goes nicely to my next question. My overall takeaway from the book was that, throughout the book, almost every chapter has an example of companies or investors screaming at their accountants: “Hey, you can’t do this.”

There’s debt that used to be off-balance-sheet. Then you’ll have an example of an accounting regulation that would require companies to bring it on, and they’ll be like, “You can’t do this. Our investors will go crazy.” Or expensing stock comp—all these examples.

For every example except for 1—maybe EPS shares in the ’90s is the 1 example—companies say, “You’re going to destroy the capital markets. You’ll destroy us.” Then he kind of dryly follows it up with, “Capital markets didn’t care. There have been studies. Capital markets didn’t care: off-balance-sheet, on-balance-sheet, expense, capitalize.”

Capital markets get there. Maybe 1 company can fool the market, but on the whole, it doesn’t really change. I was really impressed by that, and I thought about it. I want to ask you about that, and then I want to ask you about the present day. The book stopped in 2008, so I do wonder about that.

I think part of that is actually to the credit of these companies when they are arguing with their accountants. They are saying, “This change is going to destroy us.” For that not to happen, it has to be the case that they were actually behaving pretty economically rationally. If they were gaming it, then it absolutely would destroy them.

This comes up with stock-based comp. There was this funny tweet a couple of weeks ago from someone saying, “Sophisticated investors don’t debate this.” I couldn’t tell which side it was taking. I couldn’t tell if it was taking the side of just ignoring stock-based comp entirely, or just treating stock-based comp as if it were a cash expense and pretending that the company continues to pay employee salaries if it’s paying those salaries in stock, because those are economically equivalent.

There are some companies where the price makes a lot more sense if you pretend that stock-based compensation isn’t real. But then you look at the long-term chart for those companies and you’re like, “Well, if I had bought this because the valuation made sense because I ignored stock-based comp, I would not be in a happy place right now.”

Like Snap. Snap has been very generous with its shares, and the shares have not really appreciated over time. But the thing is, you talk to these companies—because it's generally tech companies that you're talking about this with, right?—and you talk to Snap, or Twitter was very much in this place as well. It's like, “Look, man, we're competing against Facebook and Google. You're the average engineer versus the super-above-average engineer or the above-average one—there's a huge difference. We compete with them, and what do you want us to do?”

I've literally had them ask me, “What do you want us to do? Yeah, we're spending 8% of sales on stock. Our stock goes down all the time, so it's not like it ever comes out. And if we don't do that, then we're never going to hire an engineer again.” They're kind of damned if you do, damned if you don't.

And I guess the correct answer might be: short them, long them at 7, and go sit on a beach.

Yeah, I think that is often the answer. Because if someone like Meta or Zoom—both of them have moved much more toward cash. They're saying, “We're doing cash comp,” because there's too much confusion about our economics. Also, we can afford it because we have cash flow.

If you're competing against a company like Meta and you have an engineer you really want to hire, you have to pay $800,000 to beat the Meta offer. Meta can do $800,000 in cash or stock, and you can only do $800,000 in stock. To the extent that that works, it has to be the case that either someone has to be misvaluing the stock, or, if employees are going to overvalue the stock and assume that it always goes up, maybe that is actually economically rational. But if it's economically rational to spend money this way, you should still be willing to disclose it.

My thought experiment is always that if you imagine a company that just switches from half cash and half stock-based compensation for its employees to all-cash compensation, and we're going to issue enough stock to pay all that cash, that has no change in the economic value of the company other than, one, if employees are more incentivized by the stock, and two, if the underwriting fees are higher than whatever the administrative cost of just giving employees stock options is. So if there's any change in how you value a company if they say that they're doing that, then something is wrong with your accounting.

Well, the answer might be, “Hey, do you like accounting or do you like making money?” That's how it feels the answer is sometimes. I've never heard someone say that during a bear market. That's a great point. I think that discussion of stock comp is something we could have had in the 1990s. I mean, there's a big section about 1990s stock comp. We could have had it post-stock-comp crash, or we could have it today, which I think speaks to how cyclical these things can be.

Let me fast-forward that to today, because one of the debates I have a lot with my companies and with other investors is—I'll just give a specific example—there is one company in an industry that recently switched its financing from a way that would finance its inventory off-balance-sheet to on-balance-sheet. So it brought on a decent bit of debt. This is a pretty asset- and inventory-heavy industry, and most of the other companies in their industry have that sitting off-balance-sheet.

There's no funny business here. It's just the terms and everything; that's how it works. I know a lot of investors, including myself, were like, “Dude, this is bad for this company. How are they ever going to get out of this doom loop?” They screen with more debt—disaster for quants, disaster for screening in a world of passive. How is this good?

The company's like, “Hey, we can save 50 basis points a year on hundreds of millions of dollars. That's pretty meaningful.” So economically, it's better for them. But I guess my question to you—and I will admit I kind of lean toward the economics, so maybe I'm biasing the witness—is, in a world where today it's all quants and all indexers, do you think that the screaming that companies have put up with for 100 years—“You destroy your optics, you destroy a company”—is more meaningful today than it was throughout this book's life cycle?

I don't actually think so. I think a lot of those quant strategies—yes, it is true that if you change something about your economics such that you do actually produce more free cash flow, but it also looks like you have more leverage even though you actually had that leverage all along, there will be some systematic strategies that are selling. But those strategies are long 600 stocks, short 800 stocks, or something, so it's incremental. It does add up because there are a lot of them, and a lot of them are using very similar signals, for the good reason that those signals do tend to work.

But if what the company is doing is actually engineering this turnover in its shareholder base, proportionally more of its investors are going to be people who care about cash flow and economics, and fewer of its investors are going to be the quants and indexers.

Wait, are those people with a higher cost of capital? Let's say I run a 10-stock fund and all of it is deeply researched. My cost of capital is going to be much higher than the quant strategy that's long 500, short 500, and just playing the statistics. Isn't that right? Just because I need some return on my time, isn't my cost of capital higher than that?

That's actually a tricky question, because I think the quants would think about capital a little differently from the way discretionary investors do. Typically, I think at least my perspective starting as a discretionary investor was that the default portfolio is 100% net long, 100% gross long, and then anything you do beyond that is something extra you do. So I mentally benchmark everything to: am I doing more or less? Now that I've gotten more quantitative, I think about it: am I trying to have more volatility than the S&P 500 or less? What's my beta versus the S&P? And what's the incremental contribution of this position to my overall volatility and structure?

If you're running one of those quant strategies where you're massively diversified, using lots of different signals, and you have a position in basically everything liquid, then because of all that diversification, you can lever up. You wouldn't want to think of the cost of capital in terms of what the prime broker is charging for the incremental capital. You do sort of want to think of the cost of capital in terms of: if they take on this much more risk, how much more equity do they need? How much collateral do they need to send over to their prime broker? I think that in that case, the difference in cost of capital is probably not that high.

The cost of equity capital is sort of: what is the return that you're missing out on if you don't do this? That is something the quants can probably calculate a lot more accurately than discretionary investors can. But it's also something where the quant number is probably going to be higher, and it should reasonably be higher. They should target a higher return because they have more historical data and a better sense of what the actual return is on the equity slice of what you're doing from this particular set of strategies.

I also think that if you're thinking about the cost of capital for individual discretionary investors, or for discretionary investors generally, you do want to think about the return-on-time function. It's a little bit tricky because, one, you do get some level of upside from researching one company and understanding more about its overall industry. I'm actually here at a company where part of the thesis is, “Hey, they are getting kicked out of an index.” So that's bad for their stock price temporarily.

The other is that the core business is one of those nice, fairly high-margin industrial companies, but they decided to become a big conglomerate. So they bought a distributor. The distributor added a huge chunk of revenue and a tiny amount of profit. Now, if you screen them, the industry classification is still machinery, but they suddenly look like a kind of average-margin machinery company. But it's actually—or it's not machinery, it's industrial stuff. They're just a Polish manufacturer.

What's the Leonardo DiCaprio line from Django Unchained? “You had my curiosity, but now you have my attention.” I'm going to be hitting you up for that company.

Okay, yeah. No, it's an interesting one where basically part of what I think is, if a rogue asteroid destroyed the distribution company, I think the stock would probably go up over the next year relative to the counterfactual, because suddenly they'd look like a high-margin, more interesting company.

So that's a case where you're asking about companies that screen poorly, and you're saying, “Hey, this is a company that did a weird deal. Now they screen as something that they're not in this case.” And if you just hived that off, the market's inefficient enough to go up. Isn't that proving my point? In my inventory-debt example, if they just took it off-balance-sheet, economically it'd be a poor choice, but their stock might go up because they screen better.

So, yeah, I’m coming to that in the short term, but then I think the thing that offsets that is private equity, where they do care about cash flow. They fixate on cash flow. They’re very good at modeling balance sheets, P&Ls, and cash-flow statements. They’re also very good, at least when they control the company, at making the choice that improves cash flow, even if it makes the accounting numbers look bad.

Pretty much every S-1 I look at for a PE-owned company is reporting a GAAP loss and has been persistently reporting GAAP losses. When you read the letter from the chairman and read their description of the business, they’re clearly describing an actually good, viable business that’s generating cash flow and accruing value over time. So, I think they’re fine with that. It’s a large asset class.

So, I agree with you. Things work out in the infinite; that is the fix, right? But I do think in the short to medium term, the company you described, if private equity bought it today, would have no problem with the margins. The company I described, they’d keep it exactly the same way. But in the short to medium term, when you’re looking for alpha or undervalued stocks, I do think it’s interesting because private equity can’t buy every company.

I know—I have a lot of friends in private equity—they don’t go hostile on companies, right? They need the red carpet rolled out: “Hey, we would like you to buy us.” So, you could have a company that’s trading inefficiently, in my hypothesis, for 3–5 years until an activist comes in and either forces them to sell or makes it so uncomfortable for the board that they sell.

So, agreed, in the long term, private equity will do it. There are all sorts of other reasons why levers will get pulled to make it efficient. But I do think it’s still an interesting question in the short to medium term.

I do. I think the nice thing about some level of diversification is that if you have a diversified set of long-term bets, it’s always somebody’s short term, right? So, there’s always some case where it’s actually working out pretty nicely.

I think it’s very, very hard not to make money if your analysis is correct, you understand why someone is selling a particular kind of company, and you understand why that’s the wrong decision. It’s very hard at scale to actually lose money taking the other side of that transaction. In fact, going back to the prop-trader thing, that’s sometimes what they’re looking for: Who is trading this with other-than-economic motivation, and how can we step in and exploit that?

My favorite example of that is looking at the Sharpe ratios of buy-and-hold strategies for corporate credit by credit rating. The highest-Sharpe thing to do is buy the highest-rated junk bonds, and the lowest-Sharpe thing to do—the very lowest-Sharpe thing to do—is buy CCC-rated things that are just about to default. That’s for lottery-ticket reasons.

The other low-Sharpe thing is buying BBB-rated bonds, because everyone who has an investment-grade mandate knows that’s where all the action in their universe is. Companies know that if they optimize for exactly that credit rating, they’ll be investment grade and be as levered as they can be while remaining investment grade. There’s just infinite appetite for that particular kind of paper.

I presume the reason this gap doesn’t close is that it would be hard to get cheap enough funding to lever up this portfolio. But if you could do a massively levered portfolio of long BB and short BBB, you’d get an interesting return profile. You periodically have upgrades that immediately push something into the overbought rather than oversold category, and you just monetize that by selling.

I suspect you’d also have sizing problems. The size of the BB universe is probably 1/10 the size of the BBB universe. So, it’s like, cool, you want to do this trade, and you’re quickly running into liquidity, sizing, and all that sort of stuff.

It is interesting that you mentioned that, because they mentioned Milken a few times in this book, and they had an anecdote that I had never understood before. They mentioned Milken—obviously, a fascinating guy and a fascinating example. I didn’t realize this was their suggestion, but I believe they suggest that a lot of Milken’s success—and this is true for a lot of people who are successful—came from a few tailwinds.

They specifically point to some tailwinds in the tax code. I believe that in 1981, accelerated depreciation was allowed, and they point to a few other examples that, right when he was getting big, really set the tailwinds on fire for his thing. I just never heard of those specific tailwinds, so I was very interested in that. It relates to the high-yield example you lay out.

Yeah, Milken did actually have some pretty good timing. If you get into fixed income when rates are high and there’s a recession, and then the economy grows and rates go down, that’s pretty good.

There were some other factors, too. I reread Predator’s Ball a couple of months back, and one of the things it pointed out was that his brother was actually really good with taxes. At the time Milken was active and earning a lot of money, there was this provision in the tax code where, if you bought a Treasury bond and then sold different pieces of it—if you sold the principal or the interest-rate component—you could allocate your cost basis as you wished between those.

Buy a Treasury, sell a zero-coupon bond, and you’ve immediately taken a capital loss. Treasuries were yielding 15%, so it’s a pretty substantial capital loss. He was compounding money pretty fast in gross terms and, in after-tax terms, was probably compounding about as fast at that point.

That’s so crazy, because you could do that. That’s basically an infinite money machine. You could be reporting negative short-term income. None of this is tax advice. We should have had a tax lawyer, but looking at the historical tax-code stuff, there’s a lot of stuff where it was just an infinite money machine.

Either people didn’t know about it, or they felt there wasn’t a large enough population of really rich people who were willing to exploit it that way. There were things like oil-depletion allowances, where you could basically write off some of the revenue from oil as the cost of the oil not being in the ground. There was no symmetry where, if you drilled for oil and discovered oil, you had to realize a massive capital gain on all the oil in the ground. No, you just depreciated that capital that you got for free, sort of.

Do you think the reason these existed in the ’70s, ’60s, or whatever—and I don’t, to my knowledge, mean maybe there’s a secret rich cabal that knows a lot of them and there are good tax loopholes and stuff—but the reason they seemed more prevalent then is hindsight bias?

We hear of the 5 people who made $1 billion doing these, and we don’t hear of the 1,000 who didn’t. Or do you think it’s because there was no internet and capital was a lot slower back then? If you were really up-and-coming, you could study these things and take advantage of them. Today, if that happened, there would be a Reddit board saying, “Hey, look at this famous money hack,” and either the government would shut it down because so many people would do it, or—probably that’s what would happen.

Do you think it was a lack of information, or do you think it was just hindsight bias that did that?

I think both of those matter a lot. Control-F is just a wonderful technology. If I wanted to maximize my net worth and could bring 1 computer tool back in time to 1955—Control-F, and then just having a digital version of the tax code—then I could make pretty much infinite money. I’d probably end up in prison somehow, but in the meantime, lots of fun.

I think the other factor is that there’s this coevolution between what the tax code tells you to do and what you end up doing. My favorite example of this is Ronald Reagan destroying the Midtown dining scene because he changed the tax deductibility of dining and entertainment from 100% deductible to 50%.

If you think of 100% deductible entertainment with a top marginal tax rate in the late ’50s of 92%, that means that if you take someone out for drinks for business, it’s a 92%-off happy hour versus just going out and having fun on your own. That’s where the three-martini lunch came from: These were really, really affordable martinis in after-tax terms.

It’s a massive tax arbitrage. At some point, people realized that, except for special occasions, they shouldn’t be going out to eat without some business purpose in mind because it was so much more affordable. Suddenly, a lot of business culture revolved around this, and you started to realize that the tax code is actually this massive cirrhosis subsidy and is causing a lot of people to be pretty dysfunctional starting at 1 p.m.

And also, isn't it the case that it's not collecting that much revenue? We can reallocate some of society's resources away from Midtown dining establishments and toward other socially useful things if we say, “Okay, we're going to cut the top marginal rate, but also all of these games you're playing—you've got to stop playing these games.”

Corporate perks used to be a lot better at that time because those were also fully deductible. I think some of that was just the IRS not asking questions it should have asked, like, “What is the business purpose of this apartment? What is the business purpose of this car?” All of this corporate travel—I’m sure they could have had some awkward conversations had they chosen to.

But in an information-scarce environment, where you're one of the overworked, beleaguered IRS people and you get this massive document from General Electric, are you actually going to go through every line item and figure out if everything's a legitimate expense? Probably not. You're probably going to look at the line that says “real estate,” rather than the list of every apartment that GE rents for executives, and try to figure out which of those are actual things for someone who's temporarily in the city and needs a corporate apartment to stay in versus someone who's just getting their housing paid for with pre-tax or after-tax dollars.

I think it was you who said, “Hey, the three-martini lunch policy basically formed the plot for Mad Men, right? Without that IRS code, we wouldn't have had Mad Men”—the TV show. I want to ask two more things. One of the things, as I'm reading this book, is that you see—and I think it's with the benefit of hindsight—so many crashes coming, right?

One example is, “Hey, in the ’70s, they go for the loan accounting trick I talked about earlier: If we've got a loan that had incredible credit trouble, and we defer interest payments for 1 year and push them out by 2 years, as long as we'll get paid par, we can keep it marked at par.” Obviously, that's a disaster, and they argue that it sets up the savings and loan crisis. There are plenty of others. The other one I use is, “Hey, how many times do you see somebody saying, ‘This market reminds me a lot of 1929’?” You'll never hear about any of them except for the one person who calls it right.

I just want to ask: When you read this book and see all of these parallels to today or things you saw coming, do you think that's because of perfect hindsight bias? Or do you think it's possible, in the moment, to use these historical parallels to actually help you avoid shenanigans? It's like economists calling 9 of the last 5 recessions; hindsight bias calls 10 of the last 1 crashes.

Yeah, I would say if you're looking for just a carbon copy, you won't find it. There's just not going to be another company that starts out owning natural-gas pipelines, creates these special-purpose entities, has a really profitable trading division, and then loses all of its money on dumb stuff and has a run on the bank. You're not going to find that.

But you will find cases where they find a way to follow the letter of the law such that their net income looks good, their cash flows don't seem to correspond to that net income, and they've found some way to make their business look better than it really is. With a lot of these, though, for me it was really useful to try to take the other side of the arguments and figure out, aside from just wanting their P&L to look better or wanting their earnings to be smoother, whether they had a reasonable justification.

Specifically, the chapter where they talk about inflation started making me think about inflation and how you should account for it. They give an example of, let's say you buy a truck and it generates some revenue, and the cost of trucks is rising really fast, such that the economic depreciation is higher. The replacement cost of the new truck means that your trucking business is actually a net destroyer of capital.

What I realized was, one, it's insanely hard to model all of this. The narrower the CPI component is, the more the person you're talking to can say, “That is weighted to a slightly different kind of truck than the kind that we buy, and so we want to use a different number.” That leads to infinite debates unless you just say, “We're going to use nominal values.”

Also, a lot of companies' obligations are in nominal terms. They don't borrow in CPI-linked bonds; they borrow in bonds whose interest rate bakes in some assumptions about inflation. When they pay people, part of the reason economists like a little inflation is that a little bit of inflation means you give everyone a continuous pay cut, and the only people who are maintaining their standard of living are the ones who are getting raises. That's just a lot easier than having a totally flat price level that doesn't rise over time and having to tell people once a year, “Hey, we're cutting your salary by 3% because you're just not as good as you thought you were.”

That truck example, to me, actually felt like all the ways you could do this just using nominal figures and not adjusting for inflation is the one that leads to the fewest arguments. Since everyone knows these are nominal numbers, the higher inflation is, the more the equity analysts are thinking about inflation, and the more the credit analysts are asking themselves, “Okay, what is the replacement capex for this, and what does that do to cash flow over time?”

You have analysts asking good questions about how to actually price this, and then you have accountants asking good questions about what the company actually owns and what the number is that everyone can agree on. In some sense, the accountant's job is to choose an answer that's 85% right, but everyone can understand the logic, versus 99% right, where we can spend forever debating the logic.

No, look, I'm completely with you. I like the trucking example, but you know what came to mind when I was reading that? If you look at the department stores, department stores are probably the number-one destroyer of value investors' capital over the past 15 years. There might be another, but it would be hard, especially among famous value investors. You think about Sears; you think about J.C. Penney.

I always look at the department stores, and once a month I'll have a friend email me and be like, “Hey, Macy's market cap is about $5 billion, and if you look, everyone agrees their real estate is worth between $8 billion and $14 billion.” So you're buying, and my argument is very much along the lines of the trucking argument: “Hey, yes, that's correct.” But if their market cap is $5 billion, they're reporting a profit of $400 million.

What it is is that you have $10 billion of real estate, which probably would generate $800 million on its own, supporting a negative $400 enterprise value, if you do that math quickly. Your issue is that until you unlock that real estate, the retail business destroys value. The management team there doesn't appear to be in a hurry to do that, because it would put them out of their jobs. They're a retail business; if they sell the real estate, they put themselves out of a job.

Yeah, I think it's a division-of-labor thing. It is probably the case that 100% of Macy's is expensive and 10% of Macy's shares outstanding is cheap, because with 10% you get someone on the board and you tell them, “Yeah.” But I think there are a lot of companies like that where shareholders who have tried it have had disastrous results.

Yeah, and then you start thinking, trying to ask yourself, “Okay, why does this gap persist?” Why hasn't the board ever told management, “Hey, we're going to give you the world's most generous change-in-control provision for your severance. We're going to vest 300% of your equity if you leave to sell the company”? They could do something like that. They could give management an incentive to leave.

Or if the board is just going to resolutely refuse to stop burning $400 million a year in opportunity cost on the retail business, someone out there is willing to buy it. Maybe the actual conclusion you reach is that Amazon would love to have permission to buy Macy's. They would pay double the current share price just to get the real estate. They would turn it all into either Macy's-branded but run by Amazon, or Amazon-branded, and just rule Herald Square.

But if all the buyers can't actually buy it, and if selling it piecemeal means you worry that you get halfway through selling it before there's another downturn in that kind of commercial real estate, then maybe it's actually pretty fairly priced.

And maybe it's fairly priced because that price is the equilibrium between a really high number if you look at the mark-to-market value of the real estate, a negative number if you mark the actual cash flows of the business to market, and then you average those and get the current market cap. I also think you had this with U.S. Steel. If U.S. Steel had been named anything else, I think that deal with Nippon would have gone through.

And Macy's—not the Macy's name itself, but because it is so high-profile—I think if you had Macy's sell to anyone, right? They sell it to a real estate firm who's going to shut down Macy's. They sell it to a real estate firm who's, wink-wink, not going to shut down Macy's. They sell it to Amazon. There's going to be so much political pressure. So I think they're almost locked in. Everyone knows it's inefficient, everyone knows the real estate's worth more, but there's just not really a way to extract it.

So, you know what? This is why Macy's can't mark up the value of their brand name and have it reflect all the brand equity, because right now that brand name is clearly a liability. They would pay billions of dollars not to be Macy's, not to have the star. I found out a while ago that the star was actually a tattoo that the founder, Mr. Macy, had. He was a sailor. Macy's logo is a tattoo.

The other thing is the Macy's Thanksgiving Day Parade. I think, if I remember correctly, now it's getting $200 million to be broadcast by NBC and Peacock. And you think about that: $200 million for, I think, 20 years. That's probably worth $2 billion NPV, something more or slightly less. Macy's as a whole is worth $4.5 billion right now. So just their parade—now, I don't know if they're getting it or who owns the parade or something, but it's interesting.

Last question I wanted to ask you: if you read this book, there are hot-button issues in every decade, right? So, the '70s, we mentioned inflation. Inflation, especially around the oil embargo, is a big one. From the '90s to 2000, Sarbanes-Oxley, the accounting scandals, the stock comp. But if I said, "Hey, as you and I sit here at the beginning of 2025, if there was a book released in 2035, what do you think the hot-button issue in accounting would be that it would point to today?"

So, I think the real answer is an answer that doesn't actually qualify for hot-button treatment. It's just not something people get agitated about. But I think that more big tech companies in particular should capitalize more of their intangibles. It is more realistic to understand their business that way. Every year, it gets a little bit harder to look at return-on-equity numbers, and the market's price-to-book value tends to drift up over time.

There's that nice chart in the book, which feels very nostalgic. It used to be 1 to 2, and then there's this weird aberration where it goes 1 to 6. It goes to 6 in the '90s. And we're back.

I also wrote this piece a while ago about how capital intensity is a feature of where you are in the cycle and not of the industry. If you look at a lot of the great capital-light companies, they end up finding that the best way to expand the capital-light business is to do some pretty heavy capital expenditures.

Like Amazon: there was the debate in the '90s over whether Amazon or eBay was the better e-commerce business. For a while, eBay traded at a premium because they were so asset-light. Their customers are the ones who are storing all of the stuff, and the customers are handling all the logistics. Then it turned out that because of that, they just could not guarantee the same speed of shipment and the same selection that Amazon could, and that speed of shipment and selection mattered a lot more than capital intensity.

So, in some ways, it's a self-correcting problem, but it just feels like if you could look at the true economic balance sheet of Google, it's more like a 10% return-on-equity business. A huge chunk of that equity is the algorithm, the brand name, the internal cohesion of the employees, the culture, and stuff like that. That all has value, and they're getting a return on that value, but it's all capital that they had to accumulate.

No one is going to be furiously testifying before Congress asking, "Why isn't the Apple brand worth $800 billion on the Apple balance sheet?" And, "Why isn't Nvidia capitalizing Jensen's vibe and depreciating that over time as he gets older or whatever?" But when I analyze something, that is part of what I'm looking at. How valuable are the intangible assets? And then can they turn $1 into more than $1 of intangibles?

If they can repeatedly do that, and then they are getting this return on the intangible asset, and that intangible asset raises the return on the tangible asset, then they have a nice formula. They'll keep growing, and they'll outrun their cost of capital. If they can't do that—

I hear you, but I think the book cogently argues that the issue is: how do you put that intangible there?

I'll just give you Google. You said it's a 10% return-on-equity business, but Google two years ago was a $400 billion market cap. Today, it's a $1 trillion market cap. So are you going to argue they threw on $600 billion or $500 billion or whatever of intangible without investment, right? But just because of the market cap, it gets really cyclical, and that's why the accountants—and, you know, if they ever did that, every time we had a 2022 and every big-cap company dropped by 30%, they'd all be like, "That's the issue."

So I definitely hear what you're saying, and it's obviously a shortcoming of accounting, but I just feel like it's settled math, right?

Yeah, I think it is. I'm exaggerating a bit, but one way to do it is just to capitalize more of the R&D and even marketing expenditures. That is, in effect, how people look at a lot of SaaS companies: they implicitly capitalize the sales cost and just depreciate it over the life of the contract.

In some ways, that does actually line up with how companies make decisions internally and how investors are valuing the outputs of those decisions. Those are your 2 constituencies. What is the company deciding to do? That's the operations-and-control legacy. And then what is the market deciding to do? That's the British-investor-investing-in-an-emerging-market-like-the-United-States philosophy.

If they both agree that this is an asset that you capitalize and depreciate over time, then I think it does make some sense. I don't actually think it's worth doing. The last time there was even a moderate tweak to SaaS accounting, it was just a pain for everyone to update all their models.

Everyone is already taking the models and basically trying to figure out: what return do they get, how are their cohorts evolving, and how fast can they jam money into the adding-new-customer-logos machine such that they continue to add customers who have good net dollar retention and will cause revenue to drift upward. So I guess I don't have a really strong "here's where the accounting is bad for you." Unfortunately, I wish I did.

Maybe SPAC warrants. They should not actually have to mark them to market. That is just the SEC messing with people. They couldn't ban SPACs, so they did the next best thing.

Look, I thought for 15 minutes trying to think of one, and I spent all my time in financial statements and I couldn't think of one. So I was putting you on the spot. There was a softball question that was a 101-mile-per-hour fastball question.

But I will give you one other interesting one, and you can comment or not comment. You mentioned the price-to-book ratio chart, which I really like. For those of you who haven't read the book or seen the chart, it's from around 1920 to 1990. The Dow traded between 1 and 2 times book value pretty regularly. Then in the '90s, it traded up to 6 times book, and that's bubble-ific, but then it comes back down to 4. His argument was, "Hey, it's because we're getting more intangible-heavy companies." Now we're probably way past 6 with Nvidia, Google, and all these things.

I was thinking about that, and everybody loves to quote the Buffett indicator, which was market cap to GDP. For years it was, "If you get over a 1-to-1 market-cap-to-GDP ratio, you're in a bubble." Now we're way past that, and that's because, guess what, a lot of these firms are international.

I was just thinking, there are 2 indicators which, for 50 years, if you had been using them, you could have kind of traded around them. What is the indicator now? These are market-level, not firm-specific, but what is the indicator now? I was thinking, "Hey, maybe price-to-earnings, because we're having so many companies—with AI, they're front-loading so much of their expenses—maybe price-to-earnings is broken these days." I don't know. Ultimately, I am a value investor. I do believe fundamentals will reflect cash flows in the long term, but what are some rules of thumb that investors have always used that might be breaking?

It is an interesting way to think about because, as an investor, if you can find broken rules of thumb, that is where there is a lot of alpha, either on the long side or—for years, people said Netflix subscription businesses couldn't go past $30. That was the rule of thumb.

If you knew Netflix, for whatever reason, was going to break that rule, what is Netflix now? An $800 billion market cap? I can't remember the last time I checked, but you would have made a heck of a lot of money. I'll pause there and let you have the last thoughts on that.

I think you could actually take 2 related rules of thumb and say they are both breaking. One is the earnings thing: especially if you're looking at a high-growth company that has recurring revenue, the higher the quality of the recurring revenue, the more you can predict next year's revenue based on this year's revenue, the less this year's P/E tells you anything.

The other piece is that I think looking at price-to-sales and growth and basically assuming companies will grow into a decent, high-margin business at scale is also breaking down. The model for a ton of SaaS companies is, “We're going to sell something to customers who are growing fast. As they grow, our business with them will grow along with them.” If you are Zoom and you sell to a company that's growing 50% a year, but they're growing their sales 50% a year, they will need 50% more seats in their Zoom license per year.

But that also means that company is not growing into a new margin. They're always paying Zoom some proportionate toll of their business. My line in an older Diff piece was, “One company's net dollar retention is another company's lower steady-state gross margin.”

I think that means that you could have a lot of these companies where, because the software ecosystem has gotten so good, there are so many things that you would have had to build a bad version of internally but can buy a really slick version of externally, their steady-state margins are actually lower. And with AI, people are very aware of this at this point: an AI business may be a software business, but it does not have software margins. There is an incremental cost to every user interaction, and it's pretty high.

That also means that you can't just say, “Well, every software business eventually reaches 90% gross margins and stays there.” Then there's eventually cost leverage on the R&D side and cost leverage on the marketing side, and so you eventually expand to some steady state of 20% or 40% EBITDA margins, depending on how much of a superstar business this is.

Both of those go away. If some software companies end up looking a lot like subscale manufacturers, where they make a dollar of revenue and 20 cents goes to employees and 65 cents goes to suppliers, there's something left over for shareholders, but it's just not that much. It's a lot of work, and there are short product cycles and things. A lot of these companies end up looking, from a balance sheet, P&L, and cash flow statement perspective, a lot more like you're investing in a steel mill than in Microsoft circa 1994.

It is interesting, though. The only thing is, maybe it's because I've had 20 years of it ingrained in me, but it's like, hey, even if that's the case, isn't there somebody who, down the line, is going to buy them and be like, “Oh, we can rip all those costs out”? I guess with steel mills you kind of had that, but with a software company, you can't rip all the costs out of a plant, but you can fire a lot of people in the software company.

Well, but those are the costs where they do expect to get the leverage, and they still will. But I think a lot of the incremental costs—you’re probably not going to build a better Zoom, and Zoom knows it. Maybe Zoom and Google Meet compete a bit, and Teams is there too, but a lot of these companies also recognize that there's some other big software company that sells the same feature set. So what they're always trying to do is get as many integrations as possible.

If you got a Slack bot that pings you 2 minutes before the Zoom meeting, pulls the relevant files from your Dropbox, sees that you haven't logged in, detects that you're still in another meeting, and automatically sends an email to someone, that's simply your Superhuman or something. If you have all these integrations, then switching providers is just this giant technical lift.

That is, again, an intangible asset. The inconvenience of switching off Zoom does not show up as an asset on Zoom's balance sheet, but it is absolutely a source of incremental DCF dollars.

In the book, he worked at IBM with Lotus Notes for a while, and it mentions that as an example of intangibles. I was laughing because, in the 2000s, everybody knew Lotus Notes was terrible, but there are still big firms—even, I believe, to this day—who are still on Lotus Notes, despite the fact that it's a disaster of a system. As you're saying, once you get all those integrations, it's very, very difficult to be like, “We've got 100,000 employees; we're all switching to Gmail.” It just destroys all of those integrations.

Byrne, we are way over an hour. We're going to chat a little bit after this, but this has been great.