Longreads + Open Thread

Longreads

• In Astral Codex Ten, a lengthy review of Alpha School by an anonymous parent who sends kids there. The most straightforward intuition pump for Alpha is that white-collar work has changed a lot in the last fifty years, in terms of how people receive and process information, how they're incentivized, what the team structure is, etc. It's a bit suspicious that another massive information-processing service sector, education, looks as familiar as it does. Of course, managing a group of kids learning algebra is a different task from managing a group of adults who are coming up with next year's hiring plan or whatever. You wouldn't want to blindly impose modern corporate norms on a bunch of eight-year-olds. On the other hand, you probably would want to do what the private sector does (albeit in a distributed way): run as many experiments as you can manage in parallel, quickly drop things that don't work, and scale up what does. (Conflict-of-interest disclosure: some of my kids attend, they've advertised in The Diff, and my wife works for them. But, conflict-mitigating note: my wife skipped one grade, I didn't skip any, our oldest is chronologically about to start fourth grade but in terms of coursework is mostly wrapping up sixth. Yes, selection effects, but also, where else can you finish learning one topic and then immediately move on to the next without waiting for everyone else?)
• Ross Douthat interviews Peter Thiel on politics, technology, and the apocalypse. One fairly minor quote that's worth thinking about: "I would define the middle class as the people who expect their kids to do better than themselves." This is actually a great working definition: a dividing line between middle and lower class is the assumption that the system basically functions well, and that if you work hard and follow the rules you'll come out ahead. If you don't believe that—for good reasons or bad ones—you'll behave quite differently. On the other end of the spectrum, the richer people get the less they focus on how much richer their kids will get—it probably won't affect their standard of living much, and someone who grows up wealthy probably sees more of the downsides to extreme wealth (or at least the process of getting it, which tends to be stressful and time-consuming). A pretty standard very-rich-person ambition for their kids is for those kids to do something that doesn't pay very well at all, but is high-status, like journalism, the arts, nonprofits, or academia. But parents in the 80th percentile of the income distribution get nervous thinking about their kids regressing to the 60th, even if that's objectively a perfectly nice life.
• Joshua Rothman with a balanced take on the future of reading: we're reading fewer long books, but we also have more ways to make knowledge accessible—he includes an example of how Claude rewrites some of the now-infamous opening passages of Bleak House in modern English, without changing the meaning. (Dickens may be hard because he uses unfamiliar vocabulary and an elliptical sentence structure, and it's hard to keep track of the grammatical role a word plays if you're also trying to figure out if "divers" means "diverse" and what "spongey fields" are supposed to be.) Personally, I don't read AI summaries of anything I expect to have artistic merit, or of anything that requires a lot of context. AI is a helpful copilot for putting things in context, and fact-checking, but it's not a direct substitute. As LLMs offer an increasingly compelling version of the average, aggregated view, there's more complementary value in the traditional reader-book relationship—a one-on-one conversation with a particular person.
• Michael Hopkins in Works in Progress writes about how cruise ships keep getting bigger and, in real terms, cheaper. There are two regulatory arbitrages here, but as with many regulatory arb stories, the initial opportunity to take advantage of gaps in laws gives way to a service that's valuable on its own. The first arb is that since the ships are traveling internationally, they can pick and choose regulatory regimes, and tend to choose the ones that let them pay people less. But they're still paying a premium to what these workers would earn in their home countries, so what it really amounts to is a way to make a tiny subset of the service sector more exportable than it otherwise would be. The next arbitrage is land use: since they're not dealing with artificially scarce real estate, they can keep on building in response to demand. (The decline in inflation-adjusted costs for cruise ships shows you what housing economics could look like if building were unconstrained, or if people were completely indifferent to location: we'd consume a lot more housing, but because so much more of it would get built, the cost to building more would keep declining as workers got better.)
• Yuxi Liu: Structure and Interpretation of the Chinese Economy: a very good overview of the different stages of China's economy. Because it's such a complex topic, a tempting way to think of their policies is to imagine a dial that can turn between "state control" and "anarcho-capitalism": Mao spun it as far is it would go in one direction, Deng started inching it back the other way, Xi moved it back, etc. But it's closer to a story of periodic total overhauls of different economic sectors, so parts of their development model spring into existence for a while, drive a ton of growth, and then disappear. One minor quibble is that the piece notes that China has basically zero long-term productivity growth. The stat is true, but the way I'd interpret it is that China reached a tipping point where incremental investment was unproductive, but that the workers who are doing actually productive work are steadily getting better at it (as they generally do in other places). So low productivity growth is a side effect of the capital in the denominator being marked at cost rather than written down to its actual, much lower value.
• On the Yet Another Value Podcast FinTwit book club, we covered Ray Dalio's How Countries Go Broke. A fun one: Dalio's model is actually very clarifying, but it only covers some drivers of economic growth. And there's always the classic question for bears: the scarier an apocalypse you imagine, the more you want to tilt your safe asset portfolio away from gold and towards guns, ammunition, food, etc.
• It’s great to see how much people have enjoyed the ReadHaus Diff chat. (Disclosure: I’m an advisor.) If you use it while logged in, there’s a sharing option, and in the future Longreads will highlight some of the more interesting discussions that get shared. 

Books

Genghis Khan: His Conquests, His Empire, His Legacy: The usual reason I pick up a history book is that I know something about what happened, but would like to know how. The nature of the how varies depending on the event, but for most books about war and conquest that don't focus on a single battle, the running question is: how did the people involved get to their destination, while staying fed and hydrated, and how were they in any condition to fight? Travel was not easy in the 13th century, particularly if it involved crossing deserts, forests, mountains, etc.

So a lot of this book revolves around logistics. A Mongol fighter operated a self-contained logistics system, with multiple backup horses, some sheep, etc. At any given time, he might be carrying around milk in varying stages of fermentation, and between horse and saddle there was often some jerky being continuously pulverized by travel. The steppes tended to reach their Malthusian limit, and stay there (the population of Mongolia today is not too far off from its population during the Genghis Khan years). And anything that disrupted that equilibrium, either famine, internal conflict, or the external kind, tilted the incentive towards war. The book points out at one point that, after a decade of campaigning, by 1206 the Mongols had eaten enough of their spare livestock that they needed loot to get back to a caloric surplus. The story of Alexander the Great has similar beats: after enough wars, he's behind on paying his soldiers and worried that the places he conquered will rebel. The solution to both problems is loot, but the next place he looted would be the next problem soon enough.

Mongol logistics had some other surprises. Because they had lots of horses, they could ride for extended periods. And since so much of their calorie consumption came from jerky and yogurt—or, in a pinch, opening up a vein in one of the horses, drinking the blood, and closing it up again—Mongols could camp without cooking fires. A highly mobile, stealthy, army deep in ketosis was a threat throughout central Asia.

This decentralized setup, where every fighter is an autonomous logistics unit, implies a somewhat anarchic setup, but the Mongols had sophisticated legal systems and their combat relied on close coordination: they'd charge the enemy while shooting arrows (Mongols practiced riding horses from a very early age, and got good at a technique where they'd time the loosing of arrows so all of their horse's hooves were off the ground, to avoid messing up the shot.) They also ran coordinated false retreats, followed by the coordinated slaughter of retreating armies, and they were also experts at disinformation and propaganda. (Also, the mentions that Genghis Khan came up with the idea of an anonymous suggestion box.)

That surprising level of sophistication makes sense because the economic niche of nomads isn't just raising and slaughtering animals and periodically raiding neighbors. They also traded, because they had a comparative advantage in rapidly transporting high-value goods long distances, but didn't have the capacity to make many metal tools, armor, and weapons.

Like most of the great conquerors in history, Genghis Khan was unusually good at the strategy and city-sacking part of things, but also had the good fortune to be born at a time when larger nearby states weren't feeling their best. At one point, a general leading a Jin army lost a battle and, knowing that the penalty for losing was death, promptly raced home to kill first (he was given a promotion to buy him off). This ended up adjusting the org chart of how the empire that would be conquered and deposed by the Mongols. The Shah of the Khwarazmian Empire made an unforced error when Genghis sent a large trade delegation; suspecting a trap, the Shah had them all executed, which was not great for Mongol/Khwarazmi relations. So they got conquered, too.

One of the features of historical luck is that it makes the test of whether or not someone is an important figure more sensitive. Central Asia happened to be unusually conducive to massive conquest in the early 13th century, but of all the assorted sons of lesser nomadic nobility, one of them in particular pulled it off.

(Thanks to Ethan Monreal-Jackson for the recommendation!)

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Subscribers may have noted that I'm a bit confused by the level of US equity prices and volatility. The world seems a lot more uncertain than it was a few years ago. Aside from a weaker dollar helping big tech earnings, what's the bull case right now?

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• Ex-Citadel/D.E. Shaw team building AI-native infrastructure to turn lots of insurance data—structured and unstructured—into decision-grade plumbing that helps casualty risk and insurance liabilities move is looking for a data scientist with classical and generative/agentic ML experience. You will develop, refine, and productionize the company’s core models. (NYC, Boston)
• An AI startup building regulatory agents to help automate compliance for companies in highly regulated industries is looking for an applied AI research engineer to work with large amounts of data and build LLM powered enterprise workflows. 4+ years as a software, ML, or data engineer/scientist required. (NYC)
• A Google Ventures-backed startup founded by SpaceX engineers that’s building data infrastructure and tooling for hardware companies is looking for a software engineering manager with 7+ years experience building large scale distributed systems. (LA, Hybrid)
• Thiel fellow/ex-Ramp founder and team are hiring a high energy, full-stack engineer to help build the automation layer for the US healthcare payor-provider eco-system. (NYC)
• An OpenAI backed startup that’s applying advanced reasoning techniques to reinvent investment analysis from first principles and build the IDE for financial research is looking for software engineers and a fundamental analyst. Experience at a Tiger Cub a plus. (NYC)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.