Longreads + Open Thread

Longreads

• First, some fun literary criticism performance art. Alex Tabarrok notes that Marxism lives on as a framework for literary criticism, long after most economics departments have moved on, which is a weird little bit of path-dependence that sometimes produces some procrustean analyses of characters and their motivations. A class can't have its heart broken, or swear revenge, or recover from a mental breakdown; Marx is just not the right guy to start with if you want to understand a story. Neither is Ludwig von Mises, but that didn't stop Paul Cantor from reviewing Thomas Mann's Disorder and Early Sorrow from an Austrian economics perspective. This piece is a lot more fun if read tongue-in-cheek, and what it really illustrates is that schools of analysis inject some randomness into your analysis. Even if these schools of thought have completely random fixations, they at least mean that different people zero in on different things and detect different patterns. And—to the Marxists' credit—they are not making this stuff up. Marx was a big deal intellectually and influenced many people, and not always in ways that were obvious to them. So, sometimes an ideological close reading actually puts a work into a context that the author wasn't fully aware of.
• Jack Adamović Davies in Bloomberg on the Isle of Man as a hub for illegal gambling, and worse. There are some jurisdictions that compete by having a simple, streamlined regulatory system, often one where it's easy for corporations to handle the paperwork around taxes because they don't owe them. But if a company gets regulatory cover from one country's government, allowing it to operate in other countries, and that company is also producing more revenue than the government, it will sometimes end up running a kind of stealth buyout, hiring former government officials and crafting policies to its benefit.
• Ken Shirriff, whose Righto is required reading for anyone who loves old computers, looks at an older-than-usual one: IBM's 604 electronic calculator. Manufactured in 1948, it used vacuum tubes to perform more than sixty operations per second. What's fun about this one is that you can see hints of the early history of scaling hardware manufacturing: IBM found a more modular way to use the (flaky) vacuum tubes, which gave them a reason to mass-produce more of them, leading to more complex devices that eventually found a computing substrate that could really scale.
• Erik Hoel has some novel thoughts on AI and consciousness. Agents keep getting more capable, but one thing they seem to lack is the continuous thread of consciousness that we expect from intelligent beings. Among other things, if their internal state gets manipulated, they can't tell. And he notes that (some) video games represent a great evaluation because it's hard for computers to intelligently explore enough of the possibility space to get anything useful done, whereas human players can usually get to the point where they get a reinforcement learning loop going.
• Hollis Robbins on the scourge of college students having mostly fake discussions on message boards in order to get credit. Like many other stories about modern higher education, it's mostly sad that everyone is wasting their time here. If college is rigorous, you have to at least make some kind of effort to keep up, or you'll be forced to admit that you've wasted time and tuition. But if more parts of education can get replaced with a trivially ChatGPTable checkbox, you don't have to reckon with wasting your time until years of it have gone by.
• This week in Capital Gains: the long, sad history of getting the technology theme right and getting the winning companies wrong. In some ways, this is inevitable: if there are different components to some new technology supply chain, then some part of the supply chain ends up subsidizing another, and it's often the most over-capitalized part that performs this role. So the average dollar invested in some theme—the second industrial revolution, the car, the chip, the Internet, etc.—probably goes to what turns out to be a low-return part of it.
• A Read.Haus reader asks if a recession causing a decrease in ad spending could pop the AI bubble. The Diff has written about how hyperscalers that monetize with ads have strongly implied that they're already realizing a great return ($), and early in Covid, speculated about the possibility of an adpocalypse, which did not actually happen ($). So they're both part of the beat. Right now, more marginal token demand comes from software development, so there isn't a direct link to ad demand. If you look at the history of online advertising as an analog, then a recession leading to a decline in ad spending would lead to share shift to them (the big platforms were able to growth through the financial crisis, because theirs were the high-ROI ads people couldn't afford to cut. But in this case, the advertisers who'd be taking share are also a big chunk of the existing market. A demand shortfall that hurts ad spending would be worse for AI a year from now than it would be today, but it's also not especially plausible today, in part because AI spending has stimulated so much GDP growth.

Books

Yuppies: The Bankers, Lawyers, Joggers, and Gourmands Who Conquered New York: I encountered the concept of yuppies the same way many kids who grew up in the 90s did, as a punching bag for humorists like Dave Barry. Somewhat later, that picture got fleshed out more: yuppies were a symptom of the collapse of the postwar economic order, and the rise of a more market-oriented, harsher, more meritocratic system (and, like all meritocratic systems, participants overestimated just how meritocratic it was). This book is an excellent retelling of that story, packed with details that put it in a fuller context. The only thing that spoils it is the blood libel in the middle.

Yuppies are an intra-elite phenomenon, which is one reason they were so hatable—if the career pipeline for graduates of elite schools involves either money or status, then the people with status will use some of it to malign the people with money. (There's a similar phenomenon today, where some of the fiercest critics of tech bros are their former classmates.) Within that narrow cohort, it was a big swing: "In 1979, only one in thirty seniors at the University of Pennsylvania headed to Wall Street. By 1987, it was one in three. At Yale, 40 percent of the graduating class of 1985 applied to work at just one investment bank, First Boston." That's pretty incredible! Short of wartime mobilization, it's hard to think of a case where there was so sudden a shift into one field—tech has joined finance and consulting as one of the big default paths, but it took longer and wasn't as extreme (in part because, a generation on, investment banking still pays a lot to polished, very hardworking, and reasonably brainy people who have other impressive brand names attached).

This swing birthed some other trends that surely had to start somewhere, but are hard to pin down. Investment banking in the 50s and 60s was still pretty genteel, and more about relationships than transactions. The shift from late nights dining and drinking with clients to late nights sweating over spreadsheets while eating takeout had to happen somewhere in there. But when, and why? It turns out that there's an easy answer: John H. McArthur, dean of Harvard Business School, wanted more applicants with work experience, but when he'd tell inexperienced people to apply again in a year, he was basically asking them to either find an employer who wanted someone temporary, or to lie. So, he asked a few banks to create a two-year program explicitly as a feeder. Since it was a fixed time period, the banks knew what they were getting, and also knew that they could work their analysts half to death and then let them recover a bit in b-school. So the default expectation for new bank employees flipped from having social connections to being able to make deals happen. So banking got more meritocratic, and boy did bankers take that to heart.

Law went through a similar evolution, complementary with banks. There's an interesting dynamic here where neither side wants to be the slow one in making a transaction happen, but everyone's either waiting for the contract to get revised or waiting for the new numbers to come back, so someone's always going to be in a rush. This period marked the end of WASP social stratification in both banking and law, partly because the WASP firms didn't want to be involved in things like proxy fights and hostile mergers, until they bowed to the inevitable and obliterated the behavioral distinction between them and the rest of the industry. (The watershed event here was 1974, when Morgan Stanley helped Inco with a hostile offer.) As big law firms grew, they flattened out: Skadden had a 2.5:1 ratio of associates to partners, but that grew to 4.3:1 in the 80s. Which meant that joining Skadden was a bit of a dare: "Oh, you're intimidated by having lower odds of making partner? Not a problem; I'm sure Davis Polk is hiring."

So, these newly-affluent, relentlessly-grinding workers paying New York rents that will leave modern New Yorkers white-knuckled with rage all had disposable income. What did they spend it on? There's a whole suite of yuppie businesses and yuppie tastes, often defined by yuppies themselves (the book mentions, three separate times, that Zagat's founders were lawyers). To someone of my generation, it's pretty easy to describe their taste: anything that was legibly fancy to an elementary school kid in the 1990s was probably a yuppie fad a decade earlier (sushi, brie, sun-dried tomatoes, a grocery store with several kinds of olives, etc.). There seems to have been a yuppified copy of the Silicon Valley angel ecosystem, where people would periodically hop off the law/banking track in order to sell marked-up food and apparel to other yuppies. And in their defense, most of these food trends are perfectly fine, still around, and higher-quality than ever. It isn't a huge addition to America's consumer bounty, but it's nice that some formerly specialty foods that started out as status-symbol luxuries have moved sufficiently downmarket that they're widely-accessible.

All of this is very interesting, and then halfway through the author makes the completely ludicrous, unsubstantiated claim that yuppies as a class were responsible for a spate of deadly arson in rent-controlled apartments in 1980s. Two true parts of this story are: there were a number of fires in parts of greater New York that were, a few years later, places higher-earning professionals moved to. And a lot of those professionals moved to New York! But you should not assume that the implication is that some Armani-clad arsonists emptying tasteful Williams-Sonoma gas cans in poor homes in order to drive the locals out. The supply chain was a little more complicated. What happened was that many neighborhoods in New York had strict rent control, rents were so far below market that people would stay even if landlords basically stopped doing maintenance and even harassed tenants, and landlords also apparently had very low odds of getting caught. So, if we're listing guilty parties, the indirectly guilty ones are legislators who, through a series of unintended consequences, made arson an extremely cost-effective way to raise rents, and, if we're talking proximate causes instead, the literal landlords who literally burned down buildings and literally killed people. The yuppies are culpable in the extremely indirect sense that, had they formed a unified bloc and agreed not to accept fantastically lucrative job offers or move to the greatest city on earth, it would have taken slightly longer for the wedge between market rent and controlled rent to get big enough to encourage arson. If anything, the yuppies prevented arson by paying high enough rents that it wasn't worth roasting the golden goose.

So, the best way to read this book is that it's a fun sociological look at a group that was actually pretty important—modern graduates from elite schools are following a life path that is recognizable to the yuppies but incomprehensible to people from even a decade earlier. And then there's randomly a chapter about how rent control is a worse policy than you think and can get innocent people killed, after which we return to analyzing the cultural role of Gristedes and the impact of Gary Hart's proto-Buttigieg presidential campaign. A rollercoaster of a book.

The Lost Books of the Odyssey: speaking of rollercoasters-of-books! This is a wild one. My specific catalyst for reading it is that I was reviewing some hospital paperwork ahead of surgery (first time) and these instructions suggested that I take a physical book in case of downtime. And I was thinking that after this surgery, if all went well, I would—doctor's orders!—have a few days after where my primary responsibility was to do powerful opiates (also a first for me). So this was a pretty arbitrary choice, just something to keep me busy during a few days of lotus-eating.

Anyway, it's fantastic. The conceit is that before The Odyssey became a single work, it was a collection of fragments, some of which extend, explain, or contradict what we now think of as the main narrative. Since there's archaeological evidence that Troy really existed (and really was sacked), there's a good chance that the main characters are based on real archetypes, and I've always thought that Odysseus, as entertaining as he is as a fictional character, was the last guy you'd want to see sailing up to your city.

The stories all have a dreamlike quality, but in some cases that's true of the characters, too—they're really desperate to wake up. Which also feels thematically appropriate. Ten years of siege and ten years of wandering will include plenty of bad days, not to mention twenty years of waiting.

The book doesn't try to hide that it's a modern one; these are backwards-looking narratives, not an attempt to write something in the style of an ancient Greek bard. And, probably, a thousand years from now if anyone's telling stories, they will be able to tell a story that's recognizably based on The Odyssey and also recognizably modern.

It's easy to describe what this book is, but hard to describe what it's like. If you see it in a bookstore, flip to the chapter called "Blindness," which is the most representative one.

Open Thread

• Drop in any comments or links of interest to Diff readers.
• What are everyone's latest favorite tricks with the most recent models? Share them while they're legal! I didn't get a chance to play around with Fable as much as I would have liked.

Diff Jobs

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

• Ex-Palantir, Citadel founders building the meta-harness (the system that knows what hills to climb, and what the right loss functions are) for all the lucrative agents need full-stack engineers that understand that turning AI into economically valuable solutions means a system that includes deterministic infrastructure. If you’re curious about building a system that finds what the efficient frontier between determinism and stochasticity actually is, this is for you.  (NYC)
• Series A, ex-Navy defense technology firm building AI-enabled drone defense systems is looking for an electrical engineer with range: schematic design, electrical simulation, printed circuit board (PCB) design, and firmware development in high performance languages (C++, Rust, etc.) If you’re interested in power and control systems for mission critical technology, this is for you. (Austin, TX)
• A startup building a new financial market within a multi-trillion dollar asset class is looking for a data scientist with commercial financial experience. (if you’ve been an investor but are newer to the data side, that’s great too.) (NYC)
• Lightspeed-backed team building the engineering services firm of the future is looking for founding members of technical staff excited about working alongside civil engineers to translate their domain expertise into the operating system that powers the next era of great American infrastructure. If you’re an engineer with strong product intuition, who's energized by access to users, and excited by the prospect of transforming how we design and construct our built world with frontier AI, this is for you. (NYC, SF or Remote)
• AI Transformation firm with an ambition to build an economic world model to run swathes of the private, unstructured economy is looking for FDEs, Platform Engineers, and business generalists who understand how to solve problems.  
• Well-funded, frontier AI neolab working on video pretraining and computer action models as the path to general intelligence is looking for researchers who are excited about creating machines that learn from experience, not text. Ideally you have zero-to-one pre-training experience and/or are a high-slope generalist who’s frustrated that the big labs aren't doing this. (SF)

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.

And: we're now actively deploying capital into early-stage companies through Anomaly. Our focus is on defense, logistics, robotics, and energy. If you'd like to chat, please reach out.