If it is, it’s not the only one. As the Internet is subsuming offline interaction, it’s becoming less of a written medium and more of an oral medium expressed through text. Optimistically, this is because people have some baseline desire to spend a certain amount of their time reading and writing, and a larger amount of time chatting, and the reading-and-writing time went online first.

But behavior doesn’t necessarily stay constant in response to new technology, and some sites provide an unprecedented environment for the evolution of written-but-not-literate memes. This is probably not something worth preventing, but it’s something worth avoiding.

What’s a written-but-not-literate meme? There’s a giant cultural gap between them, but two great examples are LOLCats and character-epithets from The Odyssey. In The Odyssey, Homer constantly repeats himself—the man of craft; the wine-dark sea; the flexible, wily king; Dawn with her rose-red fingers; etc. One theory is that these catchphrases are a way for the singer to mark his place in the story; if you can’t write anything down, you need to keep track of your mental flashcards somehow.

These lines are repetitive, but they serve a purpose for a writer who can’t just flip back a page for context. We don’t see them in purely written works because a writer can remind himself of what he just wrote, or refer to notes. (Writing itself is pretty repetitive, in that it uses the same 26 symbols over and over and over, but that repetition fades into the background.)

The Internet is reliving this evolution, only backwards. Blogs are, by default, permanent: post once, and the piece can be read (and linked) forever. Social media is harder to search, and harder to share broadly. So instead of explicitly linking to a piece they want to reference, social media users tend to make a broadly-understandable reference to the concept. Those references, and snowclones thereof, form the vocabulary of memes.

On any given day on Facebook, I’ll see a couple references to feels guy or doge.

And it has to be every day.

Because the advantage of memes is that they give everyone a vocabulary to respond to nearly anything, but the disadvantage is that they’re easy to forget. You can write a blog post and cite it a couple times a year, when it’s really relevant; a meme that only gets used a couple times a year is just nonsense.

That’s why deliberately ephemeral sites incubate so many memes. They go through an artificially accelerated evolution—there’s not really a concept of linking to an old 4chan thread or Snapchat joke, so most of those die; but the ones that live on are insanely catchy, and once they escape the ‘chans or ‘chat they spread fast. Most memes are selected for memorability, but these are selected almost entirely for repeatability.

And that’s not necessarily bad! The entire curriculum of every school is a list of facts and ideas compelling enough to repeat. But there’s a finite amount of time in the world, and that time is easy to fill—especially when there’s a population of time-sinks that have gone through a Darwinian process that roots out all but the most provocative thoughts, regardless of truth value.

That’s not a disaster. It might even be a steady state. But the next time I’m talking to someone online and I’m tempted to pick a response from a menu of options—the day’s memes, autocorrect suggestions, just the right emoji—I’d rather say something new or say nothing at all.

The book straddles economic history and history-history, with lots of tables of wool production and birth/death rates, interspersed with commentary from diplomats’ letters (the ancient ancestor of snarky blog posts).

The biggest surprise to me was how often historically bad news turned out, for survivors, to be pretty good news:

• The bubonic plague devastated populations, particularly near port cities. But that increased the ratio of fixed assets to population, which improved the average laborer’s bargaining power. And that doesn’t just mean a higher average standard of living—it means that laborers captured a bigger share of any improvement in their own productive capacity.
• The Dutch had both a dearth of natural resources and a pretty inconvenient location vis-a-vis proximity to larger and bloodthirstier countries. Because they didn’t have domestic resource competition, they pursued a free-trade policy and ended up with a big share of manufacturing (especially of the high-volume/low-price variety) and shipping. They also got a large, well-timed influx of human capital when their neighbors were invaded.
• The English deforested a huge proportion of their country, mostly for fuel. By the mid-1600s, after a few centuries of population growth and improved standards of living, they more or less ran out of burnable trees and were forced to switch to inconvenient, but energy-dense, coal. More coal mines raised the odds that someone would find a way to use coal-power to get more coal. The rest is (additional) history.
• Northern Europe lacked Italy’s strong craft guilds, which had enormous institutional knowledge. But craft guilds were slow to adapt, and skeptical of labor-saving technologies and cheaper goods. So as soon as labor-saving devices got cheap enough, the places without guilds flourished.
• The worst thing that ever happened to Spain was the discovery of gold and silver in their overseas colonies. This basically de-industrialized Spain, and eventually gave them a gigantic bureaucratic/religious infrastructure that basically existed as a welfare program—which worked in the short term, but stopped making sense as more gold and silver got diverted to smugglers.

A long time ago, I read Robert Shiller’s New Financial Order, which advocated the creation of GDP futures (of many permutations) as a way for individuals and governments to hedge against macroeconomic risk, and measure market insanity. Instead of just saying “Current market values imply above-average GDP growth for the foreseeable future,” you could just say “I shorted the S&P and bought GDP futures” instead. This seemed like such a good idea that I assumed it would take off as soon as those futures were tradable.

Which, it turns out, was in the fourteenth century, when the Genoese government issued securities called luoghi, which paid the owner a fraction of taxes collected. If you assume government taxes at the Laffer maximum, this is basically the same concept.

It was clever financial engineering, but it didn’t save Genoa from guild-driven stagnation and decline when technology advanced. So, score one for engineering-engineering over financial engineering.

Science Fiction, broadly defined, means thinking hard about reality, tweaking a few variables, then running it in fast-forward to see if anything interest happens. Defined that way, it’s not merely influenced by current tech trends—startup pitches and internal strategy emails in fact are a sub-genre of applied science fiction. As long as software is eating the world, the easiest science fiction to write is all about the implications of newer and cooler software. It’s easy to just assume that hardware will keep advancing, too. That assumption used to be merely simplifying, but now it’s been explored thoroughly, so it’s boring, too.

Seveneves is not boring.

I recommend reading it right now for two reasons:

1. The reality Seveneves tweaks is very much late-2014/early-2015. The book feels ridiculously current just now, down to details like the ubiquity of clickbait slideshows to how many Twitter followers a famous pop-scientist would have. It’ll still be a good book in 2025, but it’ll feel distinctly dated.
2. Seveneves is a clarion call for doing cool stuff with atoms rather than bits. And one of the reasons people hate atoms is that they have such a long lead time from idea to workable execution. On the off chance that Seveneves inspires someone build something cool, they’d better get cracking.

For a book that’s all about awesome hardware in a world where lots of the action is in software, Seveneves does not exactly discount the importance of being able to code. I’d estimate that two thirds of the dramatic peaks are in some way precipitated by the surprising output of a computer model. But Stephenson treats software the way lots of software people naturally treat hardware. Of course it will be there when we need it. But not in an interesting way.

A lot of the conflict, besides your traditional “humanity versus the howling void of inner space” stuff, actually boils down to nerds versus narratives. In this case, the nerds are constructing deterministic solutions in the face of risks that are understood in general but pretty random in particular. Meteors and radiation are, in different ways, very statistical ways to die. And—not just because we didn’t evolve in orbit—humans are not well-equipped to intuitively interpret risks. We are pretty well-equipped to form factions and allocate blame, so as risks get more statistical you can expect leadership, in general, to get worse.

The usual interpretation of disaster fiction is that disasters are surmountable but human beings are incorrigible. In this book, the disaster is bad enough that my money wouldn’t be on humanity, regardless of how many technical tricks we have up our sleeves. But it’s hard to keep up a good fictional pace when your dramatic arc is literally defined by a segment of an orbit.

But if anything is going to actually help human beings prepare for disaster, it’s going to be recruiting people into the nerd faction and away from the narrative faction. And Seveneves is definitely masterful pro-nerd agitprop.

• This is a feature, not a product.
• Google/Facebook/Amazon could easily do this.
• It’s not scalable. It won’t catch on. (These go on the same line because anything that is scalable and hasn’t caught on yet might be something that will never catch on, and anything that is catching on is not scalable in its current form.)
• They’ll never justify that valuation. Their business model must be terrible if that’s all they’re worth. (Another good matched set.)
• That’s not secure.

Paul Graham, News.YC’s moderator emeritus, says that “The defining quality of the middlebrow dismissal is that it’s a cache dump of the writer’s prejudices…” This probably explains why they’re so common. It’s natural to see any news story as confirmation of your existing prejudices—but it’s easy to read confirmation where it’s not there. So it’s easy to see Groupon or Zynga as a “failure” for being worth merely a couple billion dollars when they were once worth ten billion plus apiece. (You could get pretty rich failing like that a few times.)

But the real driver is not just that it’s easy to misread a story as confirming your beliefs. It’s that the alternative is to read a story and realize you were wrong. And since a middlebrow dismissal is a dismissal of something, it’s generally a dismissal of something that disagrees with the reader’s bias. This thread dissing Groupon, which is worth four billion dollars four years later, is a pretty great example:

a glorified wordpress blog with 10,000 sales reps propping up the fake market with unsustainable or warranted demand. It’s like watching Boiler Room play out in real life. There is no business process innovation here or technical innovation. The only innovation here is using huge marketing dollars to perpetuate a pump and dump ponzi scheme long enough for an IPO.

10,000 sales reps are a commodity. Being the first company to staff up to 10,000 sales reps is enough to create a real competitive advantage.

Probably the best part of the thread is this explanation of how to clone Twitter. TWTR, after a painful guide-down, is worth around $25bn, up from roughly $7bn at the time. Somehow that Hacker News commenter has not found the time to clone them and split that $25bn 50/50. Which is doubly frustrating because the comment isn’t even wrong! It is fairly cheap to build Twitter, if Twitter is “a database that can store tweets.” But if Twitter is “the first place anyone goes to track news in real-time,” it’s something so difficult only Mike Bloomberg and Ted Turner have been able to build anything similar before.

If it’s easy to be skeptical, that’s a huge warning sign. Skepticism should be harder than credulity, not easier—someone is betting their time and money on the idea, and a commenter is only betting their Internet reputation, which nobody will ever care about. Startups are a compelling case for Chesterton’s Fence. If you can tell a new company is obviously flawed, congratulations! You don’t have a useful opinion. Only when you can articulate why a reasonable person would have built the company that way, and what that reasonable person is missing, can you escape the Curse of the Middlebrow.

A year or two later, Bill and Henry decide they should expand into an adjacent market. Their backers point out that a massive competitor already has a head start. So, while continuing to work on their their main job, Bill and Henry run a separate side project that pushes into the new market. Both the main company and the side project are quite successful, and the fearsome competitor turns out to kind of flail around for a while before exiting the business entirely.

They had the usual startup ups-and-downs, culminating in something all-too-familiar: they were earning lots of revenue from a single major platform owned by a third party. And one day the third party changed its platform policies and gave Bill and Henry the boot.

Both the company and the side project successfully pivoted, and neither Bill Fargo nor Henry Wells would recognize their main gig, American Express, or their side project, Wells Fargo, today. But they definitely demonstrate that a founding team is more important than theoretical competitors. And that if you build your business on someone else’s platform (in this case, package delivery over rail), you face intractable uncertainty (express rail nationalization in 1918). And, most interestingly, that once you have the infrastructure necessary to move bits, the most lucrative bits to move are the ones that tell you who is creditworthy and who is not.

When Wells, Fargo, and company were plying their trade, the most efficient communications network was the stagecoach. Wells Fargo and American Express delivered goods, but the most valuable were hard currency—gold bars—and letters of credit. By the time their express business was nationalized in 1918, Wells Fargo and American Express were both more in the credit business than the package business. By weight, the most valuable thing you can deliver on a regular basis is a piece of paper entitling the recipient to spend money.

These days, our bit-moving infrastructure is a bit improved from stagecoaches, and more resilient to platform risk than railroads. But we’re still in the early days of shifting our wealth-moving and wealth-measuring infrastructure onto the Internet. We’ve been grafting older payment rails onto the Internet since the first e-commerce transaction, but we didn’t have a working, Internet-native payment mechanism until early 2009. And now, the most interesting problem in my two favorite fields (finance and the Internet) is the same problem: how do we turn Bitcoin from a clever proof of concept into the currency of the future?

The answer is unclear, but it’s a pretty cool problem to solve, so I’m moving to San Francisco and joining 21.

Political questions get resolved by technology more often than they get resolved by debates (and anything that seems to be eternally worth debating is probably impossible to figure out). So, I consider debating politics, and holding political views, a vice worth avoiding.

I’m not abandoning you, though! Below, a quick summary of the couple mental models that carry 90%+ of my politically argumentative mental freight.

1. Most beliefs people argue for are false. This is a strong claim, but probably true. There are lots of beliefs that are true or useful, but not worth arguing about, such as the speed of light or the value of 3 + 5. There are lots of things that are somewhat ambiguously true and debatable, and people debate them earnestly. And then there are beliefs that correlate with political, social, and religious identity. Most of these are somewhere between deliberately unverifiable and definitely wrong. But if you espouse them, you mark yourself as a member of an ingroup by virtue of how much you’re disliked by an outgroup.Do people have strong opinions because they like the truth, or because they like affirming their membership in the ingroup? A good question to ask here is whether your friends are more likely to argue about a) the best way to measure the atomic weight of deuterium, or b) the optimal minimum wage. One of these can be answered pretty definitively and, for some purposes, usefully. One of these is basically impossible to accurately answer, and will lead to hour-long, totally useless debates.
2. Most exciting headlines are true but misleading.
3. Most social science is bogus.
4. Hypocrisy is massively overrated as a sin. It’s not a good thing, but the supply of people-pointing-out-hypocrites is way too high. And that’s because you can feel morally superior to a hypocrite basically regardless of which thing they did/said was wrong or which was right. Suppose a politician buys his mistress an abortion. It’s kind of hard to complain about this if one thinks abortion should be legal. But suppose that same politician wanted stricter laws against abortion. Now we’re talking!The awful side effect here is that people are constantly hunting for hypocrisy—the One True Bipartisan Issue—and the best way to find it is to misconstrue people’s opinions in such a way that their actions are inconsistent. There is nothing especially wrong with Russell Brand being an outspoken anti-capitalist who lives a lavish 1%-er lifestyle. There would be something wrong with Peter Singer doing it, sure, but Brand is making the best of a system he doesn’t like. The temptation is to say: don’t listen to him; he says the rich are screwing us, and yet he’s rich! But the correct bullet-biting response is to say: although one should give credit to Brand’s economic opinions given his skill at making lots of money, he is still wrong.
5. On a related note, badness and stupidity do not have a lot of explanatory power. There are probably a couple conservatives out there who completely buy into liberal principles and do such a terrible job at reasoning from those principles that they end up agreeing with everything they read in the National Review. But, not very many. There may be some liberals who are Burkeans at heart but espouse the same beliefs as The Nation‘s editorial board because they want to see the world burn. But true evil and comprehensive stupidity are both a) not especially common, and b) perfectly able to explain basically any observation. Like conspiracy theories or mind-in-a-vat thought experiments, the fact that they can explain every observation, including counterfactual ones, means they don’t explain any.
6. Most personal traits are pretty fixed. Through practice and force of habit, people can perform a little better than their baseline. Most people don’t do this. Pretty much any policy idea that tries to alter how people think or behave, without carefully tweaking their incentives, will probably fail. But policies that surf on their incentives—and respond to technological changes in the ground rules—have a chance to do some good.

Of course, it’s more than a little iffy for someone who says “personality flaws are fixed” to go ahead and try to fix one of his. So I’m going to offer myself an exception: I am willing to entertain any political debate that will be relevant in ten years. I know of only one good source for what really mattered in 2005. It’s here, and it’s pretty funny, but almost none of the issues discussed ever ended up mattering. On the other hand, in 2005 I read What You Can’t Say, and that changed everything.

– Wendy Kaminer, A Civic Duty to Annoy

“Trolling is a art.”

-Anonymous

People argue to learn, and people argue to win. You can’t separate the two: when you win, you learn that your views were persuasive. When you lose, you either learn that you’re wrong or learn what counterarguments to address. And either way, arguing combines a personal good (the feeling of being right) with a public good (other people are less wrong, or at least less willing to spread their incorrect views).

This framework gives the arguer a lattice of positive and negative responsibilities. Do cite sources (you can reuse them). Don’t resort to ad hominem (you’ll end up arguing with a different hominem one of these days). Do practice the principle of charity. Don’t be excessively arrogant.

Or, do. Arrogance is tricky. In one sense, it correlates with being wrong—the best way to be confident is to be so mired in confirmation bias that you can’t imagine any reasonable person would disagree with you. But another way to be arrogant is to be highly confident that you’re right.

And if you’re wrong, being an overconfident jerk is a massive public service to the forces of truth! The most glorious comeback in human history that didn’t involve Winston Churchill or Oscar Wilde was almost certainly this exchange. Normally, this would just be a conversation between two people who each think they’re pretty smart. Only thanks to the obnoxious setup can cperciva tactfully establish himself as the overwhelmingly more credible source.

Robin Hanson thinks people should bet on beliefs. It’s a little socially awkward to put money down, but it’s comparatively easy to stake reputation. And within a given social group, reputation is fungible: someone good at outwitting the overconfident gets more reputational capital to bet.

But confidence doesn’t just affect the results of a discussion—it also affects who participates, and why. If people are generally humble and cautious about their views, there isn’t a strong incentive to disagree. After all, they’re just throwing an idea out there. If people are excessively sure of themselves, it’s practically a challenge to anybody who disagrees. (Cf. Cunningham’s Law.)

Taking the outside view, it’s good for people to be embarrassed when they’re wrong. In the same way that, taking the outside view, it’s good for poor capital allocators to have less capital.

Well-calibrated arrogance is a good way to correct your own misconceptions, and a good way to ensure that reputation more closely tracks correctness. And that’s the only way for humanity to collectively clear out obsolete ideas so we can advance our collective understanding of the world.

Being a jerk on the Internet isn’t a character flaw. It’s a public good.

The trouble with signaling is that it’s too good. Signaling is always a possible explanation for phenomena that are visible but don’t have an obvious cause, but like lots of other powerful explanations (“market price” or “survival of the fittest”) it’s also strong enough to explain counterfactuals. It’s still valid and meaningful, but signaling is not necessarily a comprehensive model. And in this case there’s another model that can explain some of the gains: accumulating lots of knowledge is disproportionately useful in fields where a) winners win big, and b) growth in that field is has an unusually small impact on GDP growth.

One sensible critique of higher ed is that even in STEM fields, people don’t actually use a ton of the knowledge they accumulate. If you’re really good at esoteric calculus tricks, you are probably going to do a great job teaching calculus to undergrads, but you’re not going to be that much better at, say, building a bridge or a bomb.

Okay, actually, you will be better at building a bomb: in Surely You’re Joking, Mr. Feynman, Richard Feynman notes that he learned One Weird Trick for fiddling with integrals:

That book also showed how to differentiate parameters under the integral sign – it’s a certain operation. It turns out that’s not taught very much in the universities; they don’t emphasize it. But I caught on how to use that method, and I used that one damn tool again and again. So because I was self-taught using that book, I had peculiar methods of doing integrals.

The result was, when guys at MIT or Princeton had trouble doing a certain integral, it was because they couldn’t do it with the standard methods they had learned in school. If it was contour integration, they would have found it; if it was a simple series expansion, they would have found it. Then I come along and try differentiating under the integral sign, and often it worked. So I got a great reputation for doing integrals, only because my box of tools was different from everybody else’s, and they had tried all their tools on it before giving the problem to me.

Then, a few years later, working on the Manhattan Project:

I was sent to Chicago with the instructions to go to each group, tell them I was going to work with them, and have them tell me about a problem in enough detail that I could actually sit down and start to work on it. As soon as I got that far, I was to go to another guy and ask for another problem. That way I would understand the details of everything.

It was a very good idea, but my conscience bothered me a little bit because they would all work so hard to explain things to me, and I’d go away without helping them. But I was very lucky. When one of the guys was explaining a problem, I said, “Why don’t you do it by differentiating under the integral sign?” In half an hour he had it solved, and they’d been working on it for three months.

A little knowledge is a dangerous thing. A lot more knowledge is an incredibly dangerous thing, if the other guy has the knowledge and you’re standing within the blast radius.

“We’re not all Richard Feynman!” you can object. That’s true. Two good reasons for that: one, none of us are that smart. Two: none of us have insisted on accumulating a giant library of techniques-that-might-come-in-handy. One of those is fixable.

(Incidentally, I’m spending some of my between jobs-time reading Apostol’s Calculus. It’s a pain, and will have few practical applications. But one practical application is that math-heavy work is vastly less intimidating. Fluency with calc makes it easier to achieve fluency with stats, and fluency with stats is absolutely essential in finance, even if you’re not a quant.)

An early Facebook growth hacker used the term “surface area” to describe one of Facebook’s valuable assets: the fact that a lot of people are bumping into the site at any given time, and any small change Facebook makes to optimize their experience is going to have a disproportionate impact. Learning more problem-solving techniques is a good way to expand your problem-solving surface area.

And, actually, wait a second: if you weren’t used to thinking geometrically, would the concept of “surface area” pop into your head? People think about that idea generally all the time, but it’s tough to find a peg to hang the idea on. And then you hear “surface area” and suddenly stuff as disparate as “read more” and “meet more people” all pop into place—they’re all hacks to increase your surface area so you bump into more solvable problems.

Charlie Munger has a lot to say about this:

You’ve got to have models in your head. And you’ve got to array your experience—both vicarious and direct—on this latticework of models. You may have noticed students who just try to remember and pound back what is remembered. Well, they fail in school and in life. You’ve got to hang experience on a latticework of models in your head.

What are the models? Well, the first rule is that you’ve got to have multiple models—because if you just have one or two that you’re using, the nature of human psychology is such that you’ll torture reality so that it fits your models, or at least you’ll think it does. You become the equivalent of a chiropractor who, of course, is the great boob in medicine.

It’s like the old saying, “To the man with only a hammer, every problem looks like a nail.” And of course, that’s the way the chiropractor goes about practicing medicine. But that’s a perfectly disastrous way to think and a perfectly disastrous way to operate in the world. So you’ve got to have multiple models.

And the models have to come from multiple disciplines—because all the wisdom of the world is not to be found in one little academic department. That’s why poetry professors, by and large, are so unwise in a worldly sense. They don’t have enough models in their heads. So you’ve got to have models across a fair array of disciplines.

You may say, “My God, this is already getting way too tough.” But, fortunately, it isn’t that tough—because 80 or 90 important models will carry about 90% of the freight in making you a worldly-wise person. And, of those, only a mere handful really carry very heavy freight.

It’s not an accident that all of the people I’m citing are in either tech or finance. These are two fields where a) constantly accumulating knowledge is hugely valuable, and b) the very biggest winners win quite disproportionately.

And what’s startling is that more education as a way to perform better in tournaments like finance and tech, it would explain both the high individual benefits to more schooling and the low collective benefits. Facebook has made a lot of Facebook employees rich, but that increased net worth produces proportionately little increased GDP. At least, little compared to the net worth vs GDP impact of, say, running a successful retailer.

A similar dynamic applies in finance. Good fundamental stock pickers are constantly implicitly managing their surface area—learning a couple adjacent industries really well so that when Company A’s announcement implies bad news for supplier B who reports earnings right before B’s competitor C, being able to trace the impact of that new bit of information fast is a competitive advantage. And it’s a competitive advantage in a field where really successful people achieve extraordinary levels of net worth while directly creating maybe a couple dozen jobs.

This is not at all a nail in the coffin for the signaling theory. Caplan is broadly correct that we likely overinvest and malinvest in education. But it’s too easy to overfit: at least some of that education pays off in a big way, and in a way that doesn’t contradict the raw stats about the median student benefiting little from education, and the aggregate impact of education being low.

I would go on, but I have a calc problem set to get to.

But there is a fairly trivial way to square that circle. A company’s current value is the net present value of all future cash flows it can generate, or the amount of cash flow it can cause a single prospective purchaser to lose, discounted to the present.

Can Snapchat return ten billion 2014 dollars to investors over its lifetime? Never say never, but probably not.

Can Snapchat cost Facebook, Google, or mobile phone companies ten billion dollars in unwanted capex and forgone income? Much more likely!

Graham and Dodd and Schelling

I am by nature a value investor. I read Lowenstein’s Buffett Hagiography,The Intelligent Investor, and Security Analysis from about mid-2000 to mid-2001. This is sort of like cracking a Bible for the first time right after personally witnessing all of the world’s sinners descend straight into hell. Value investors have this mentality that they are an obscure and virtuous niche, making their way in a financial industry strewn with degenerate gamblers.

So it was a big surprise to find out that everyone in finance reads Graham and Dodd and thinks they knew what they were talking about. There are still degenerate gamblers, but they are degenerately gambling on the output of fairly traditional value formulas. (The reason trading is so frenetic is that, if everyone is a value investor, being a pure value investor is a mug’s game—your goal is not just to find what’s undervalued, but to figure out what is undervalued and what’s going to change that. Which also lets you buy something you think is 10% too cheap but is going to hit the right price in a month, instead of buying something at fifty cents on the dollar but living with the risk that it’ll be fifty cents on the dollar in ten years, too.)

Many investors are not just value investors, but are averse to straying too far from value investing. While people will speculate on whether or not a company could get taken out, or what the valuation would be if they were, that’s often viewed as a pretty qualitative question. If it’s quantified, it’s often in terms of whether or not the most likely acquirer has enough dry powder to finance a deal.

And that approach generally makes sense. It’s hard enough to model the behavior of the average investor whose actions determine a stock price day to day. Now you have to model the behavior of acquirers, too?

But it’s necessary. Graham and Dodd wrote for a general audience. They were popularizing techniques, not writing a textbook for full-timers. So they made lots of simplifying, almost Newtonian assumptions, many of which held for a long time. One of which was, implicitly, that competitive dynamics are pretty stable. Yes, there are cyclical companies that are expensive at five times earnings and cheap at fifty. Yes, there are some companies that raise their prices every year. But that rarely changes.

The economy is actually made up of numerous tiny, blurry monopolies, all of which allow owners to extract profits above the cost of capital but under complicated constraints. Sometimes those are easy to intuit (eg patents and trademarks, or owning hard-to-duplicate capex like a railroad). Sometimes they’re fairly counterintuitive (Google’s search engine is sustainably better not because they hired the best engineers, but because their market share is so high that their sample size for testing new improvements is huge).

Across older supply chains, competitive dynamics shift in a fairly understandable range. Slightly more expensive oil can lead to radically more expensive tanker rates and equipment leases, and that doesn’t change. But in newer supply chains, like “attention paid to a mobile device -> information about an individual’s behavior and purchasing habits -> well-targeted ads -> purchases,” supply chains shift qualitatively, too. And the big drivers of qualitative change are product changes and acquisitions.

Graham and Dodd rounded competition down to the point that game theory wasn’t a factor. In many of the industries they looked at, it wasn’t. But in the high market-cap, high-volatility Internet sector, it turns out that game theory is critically important. For the median small tech company, the biggest market-cap impact they can have is negative, for their competitors.

Yes, It’s Speculative, But…

This sounds like a reversal of standard value investing thought. Value investors are supposed to stand away from the crowd and buy unpopular stuff knowing what it’s really worth. Buying something for more than the value of its future cash flows, but selling to somebody who will overpay more, sounds like pure speculation.

But ultimately, the vast majority of investments are going to be sold some day, and if there’s a reason to think the price a) won’t be a DCF, but b) can be analyzed reasonably, it’s totally reasonable and consistent to invest on that basis. Warren Buffett says he buys stocks with a preferred holding period of “forever,” and it’s true that his time horizon is unusually long. But that line itself is a little game theory, too: Buffett buys large stakes in companies that grow free cash flow with little incremental capex, so he can afford to be patient. And by saying he buys now and holds forever, he becomes the buyer of first resort for a huge class of sellers. This is probably not a cynical ploy, but it’s also what a cynical person might do.

For those of us who aren’t trying to acquire family-owned companies at fairly low multiples, a holding period of forever is not so ideal. Outside of big, brand-name consumer products, competitive advantages don’t last forever. (And even when they do, by now everybody knows it, so eventually every single share of Coca-Cola will be owned by Buffett acolytes.)

So we can modify value investing from “buy a stream of cash flows for less than its net present value,” to “buy a stream of cash flows for less than its future value to whichever acquirer is most likely to buy it.” And who is more likely to buy a startup that competes with a big online incumbent? Either that incumbent, to cauterize the damage—or their competitors, to keep it out of the incumbent’s hands. Either way, the exit price of the investment is driven by how much damage it can do to other business models.

Implications

The most significant claim this makes is that speculative startups and value investors are on a continuum, not engaged in separate activities. The value end of the spectrum assumes that all possible future buyers of a given company are homogeneous. The speculative side assumes that buyers are heterogeneous, but that their behavior is still reasonable. (ie when Instagram sold to Facebook, it was selling to a “value investor” who was paying $1bn to avoid far more than $1bn in forgone free cash flow.)

A useful check on this is to see if more monopolistic industries have more high-valuation acquisitions. And it seems so: Facebook and Google (who have near-monopolies on online identity and online demand-harvesting) tend to make splashy and confusing purchases. In more fragmented markets like online travel agencies and general e-commerce, there are fewer such deals. Online real estate: two of the three American pure plays in that market have accepted offers this year, both at surprising premiums.

If investors are behaving this way, what are the broader implications? First, monopolies are not as bad as they look in a naive model, because monopolies are more or less required to be the high bidder for direct and indirect competitors. If Facebook really owned personal identity and messaging, that could be problematic. But if owning that means having to periodically write $10bn+ checks for companies like WhatsApp, the net present value of their monopoly is much lower.

And, conversely, the net present value of starting a company purely to be a thorn in the side of monopolists is much higher.

Whether that’s a good thing or not depends on how much you value ridiculously expensive moonshots that can only exist when a giant pile of money is under the control of a single determined owner. Breaking monopolies is great for consumers on the margin, but Facebook and Google, at least, are spinning their monopoly profits into free wifi, subsidized data in the third world, immortality research—all very interesting stuff that they couldn’t pursue if more of their excess profits were diluted.

But this does present a positive vision. An entrepreneur can escape that weird rat race by building something so truly weird that nobody can even describe it well enough to build adjacent competitors. One way to do this is to invent a completely new business, but most completely new businesses are new because they are a bad idea to start. Better to build a search engine when all the other search engines are turning into portals, or a social network when MySpace is a cesspool with a huge head start.

As it turns out, one risk of carry trades in FX is that they are a pretty good deal most of the time, so lots of hedge funds get involved, so when something else blows up, they unwind their carry trades, too. So at any given time: a) the carry trade is earning a positive return from the interest rate differential, b) the trade also has a tailwind from more people making the trade (ie every new “borrow Yen / short USD” transaction, but c) at any given time, there’s a possibility of a sudden and painful reversal.

Fortunately for anyone involved in such a trade, you earn your 2 and 20 on performance at a specific time, not on the expected contours of future performance. So for someone who wants pretty good odds of a pretty good return, a carry trade is a pretty good bet.

In fact, it’s such a good bet from an “expected current profits at any given time” perspective that it might be tempting to get involved in carry trades that have a poor expected value but a high chance of short-term returns. Beware anything that looks like an obvious way to make money; it’s probably worse than it looks.

But why do Yen trades from the Clinton administration matter to an equities guy like me?

Because every trade is a carry trade. In FX and credit, it’s explicit—there’s a difference between the interest you pay on what you borrow, and the interest you earn on what you buy. But in equities, there’s a difference in the return profile for different kinds of equally successful trades.

There’s a lot of market wisdom that sounds like total nonsense about selling short: you need thick skin, you can lose a lot of money, etc. That all sounds like it’s technically true but useless. It is in fact technically true, actually true, useful, and essential.

Because shorts aren’t longs.

Suppose you’re a good stockpicker. You have a sense of which companies are performing a little better than the price indicates, and which are performing a little worse. The natural way to think about this: you research two companies’ growth profiles and find that ABC is 10% too cheap, and competitor XYZ is 10% too expensive. When you buy ABC and sell XYZ short, they both kind of bounce around and gradually converge on the correct value.

Not so!

ABC’s management is happy to go on CNBC, investor relations is loquacious, management is forthright because they have much to be forthright about. ABC stock gradually creeps up. At XYZ, management is less chatty. But lack-of-good-news doesn’t read as bad news, and meanwhile they look cheap compared to ABC. So they move up a little, too. For most of the life of the trade, “long ABC, short XYZ” looks like a loser. It’s only when XYZ reports ok earnings but reduces their guidance that the stock violently moves lower.

This may be why a popular hedge fund model is to be net long, rather than flat and more levered. Although the net-long portfolio is more volatile, it’s mostly losing (less) money when the average investor is losing money—instead of underperforming nearly all the time, and making most of its outstanding returns from a few short positions.

It makes sense for professionals to care about carry, since their incentive structure is not “maximize expected return over an infinite future, while minimizing the risk of going to zero,”; it’s “maximize the value of a ‘snapshot’ of the portfolio when fees are calculated, and minimize the odds of looking so dumb you get fired.” Positive-carry trades trade off a small chance of a wipeout for a great chance to make money continuously.

If smart, sophisticated people with $2.4tr in assets under management all have an institutional incentive to be biased towards X, the natural lazy response is to be biased against X. This creates a weird market inefficiency, where the best stocks are “longs that look like shorts,” ie companies where management doesn’t talk much to investors, whose prices don’t follow a smooth trend, but who have a reasonable chance at surprising to the upside. Those aren’t going to end up in a smart hedge fund manager’s book, because when you check the results frequently enough, “right but early” translates to “wrong.”