Longreads + Open Thread

Longreads

• Samuel Hughes in Works in Progress on the Victorian way of work, which in some ways feels very modern: Victorian elites had short formal working hours, and spent some of them at home rather than at the office, but they constantly socialized with one another in a way that either reinforced professional relationships or constituted after-hours dealmaking. So, eerily similar to someone who might get into a Twitter argument while at the office but also answer work emails at 10pm. It's one of those cases where the definitions of "networking" overlap: in those bandwidth-constrained days, constantly talking to people was the only way to keep track of what was happening in parliament, the exchange, the colonies, the universities, etc., all at once. Those coordination costs still exist, but they're lower, and there are more impersonal means of accessing information. One way to look at Victorian England is that it was actually operating at the limit of what human institutions could handle at that time; the marginal cost of coordination kept rising as the empire got more complicated, so it took an all-out effort just to maintain it.
• Via Kevin Roose, here's Dario Amodei's remarkably prescient 2017 essay on the Big Blob of Compute model of intelligence. This is, in some ways, an exciting piece to read: he called the industry's shot almost a decade ago, and every major company in the industry is following the playbook he laid out. On the other hand, it's deflating if machine intelligence is not the discovery of some tidy model of the universe, but instead a messy process that somehow manufactures correct answers by a clever variant on brute force. This piece also helps explain the somewhat mystical tone AI researchers sometimes use: they've been rewarded for having faith (in scale) but not for trying to understand the mysteries of specific kinds of intelligence.
• Nicholas Lemann in the New Yorker on the often shambolic process by which California's wealth tax got on the ballot. It's a fun look at the guts of the political process, particularly in California (using plebiscites feels more democratic, but they tend to get executed by people who have the same job function as legislators, but far less transparency). At one point, a think tanker talking about evolving attitudes towards tech companies makes a revealing comment, that every Tesla owner he knows is ashamed. This is not normal! There are certainly some people who like EVs and don't like the direction Elon Musk's politics have taken recently, but the vast majority of car buyers aren't trying to make some kind of political statement. (If that were the popular perception, it would be great for Ford: you could be in favor of anything from the editorial line of the Dearborn Independent to the typical grant recipient of the Ford Foundation.) Also notable is the discussion of how California's propositions are often introduced as a bargaining maneuver, where the plan is to suggest something popular but extremely impractical in order to get someone to agree to a boring but valuable concession instead: "The system allows those who propose initiatives to withdraw them before Election Day. This year, for example, Regan dropped an initiative that would have restricted hospital executives’ pay, in exchange for the California Hospital Association not imposing new administrative hurdles on unions pursuing political campaigns." The system was really not intended to make everyone who was good at gathering signatures a temporary, off-the-books legislator, but that's what we've ended up with.
• Hollis Robbins on appreciating architecture without being able to mentally visualize things. Aphantasia is a very fun topic because it's very hard for either side to imagine what the other experiences: it's impossible to imagine being able to imagine things you can't, and almost as hard to imagine being unable to imagine things. Which means that it's a great jumping-off point for talking about aesthetics in general, and the question of what kinds of experience is the right resolution: is beauty something you can appreciate if you see something only once, or should it be judged by how you tolerate seeing it every day?
• From Alethios: thoughts on institutional decay. People sometimes use the term "scar tissue" to describe processes put in place after a disaster to prevent it from happening again. It's a good analogy, because it reduces flexibility and requires constant maintenance. One valuable piece of the framing is that it describes the alternative to endless bureaucracy: mid-career employees with a long tenure at the organization, who have the reputational capital to make big decisions and enough experience that they're usually good ones. This is a very hard theory to prove or disprove, since the effect happens on such a long delay, but it is a useful way to think about state capacity—it's not just a question of who chooses to work for the government, but who sticks around for a long time.
• In Capital Gains this week, we look at why some companies can have a high market value but still be terrible credits, and some thoughts on who the optimal borrowers in AI are.
• A Read.Haus reader wants to know if AI agents will make the market more efficient and force banks to pay higher interest rates. Consumer inertia is an incredibly powerful force, but it's up against a force that's compounding fast. And in general, when people grant their AI agents broader permissions, they do it in big chunks, so even if they're not specifically optimizing their budget, if they're optimizing anything money-constrained, that's an obvious suggestion for their agent to make. But it will take a while to get there, which means banks will have some time to come up with ways to increase switching costs or ways to recapture that elsewhere. For people below the FDIC limit, a bank account is a completely commoditized financial product: they're all just brand names for the same ultimate borrower, who isn't going to default. So the economics of agentic comparison shopping between banks will say a lot about where margins end up for products with low differentiation and high search costs.

Books

A Fabulous Debt: The Epic Story of How Bonds Built the Modern World: I'm very high on the list of people who'd like to read a history of bonds, but even allowing for that, this was a very good book. Bonds represent an interesting evolution in the history of finance, turning a promise from a bilateral agreement to one that could be separated from the original lender and treated as an asset. These portable promises made capital more mobile, especially when that capital was used to scale some already-proven process. Bonds are a financial instrument that is designed for reversion-to-the-mean bets, which means they're uniquely revealing about what institutions people think are durable at any given point in time.

Reading financial history means learning that some tricks are older than you'd think. For example, the leveraged buyout conventionally dates back to the Bear Stearns "bootstrap deals" of the 60s, or perhaps to the recapitalizations of companies like Du Pont and Carnegie/US Steel. But arguably the first leveraged buyout also represented the birth of the bond: it was the independence of Venice, when wealthy residents were forced to give the state wealth (to fund its fleet), in exchange for tradable bonds yielding 5%.

Bonds are useful as a liability and as an asset. It's helpful to have some kind of standard savings medium and default form of interest-bearing collateral to put other transactions together. So bonds end up creating both a bigger government and a bigger private sector: they were historically a good way to fund wars (borrow during the fighting, run a surplus after to slowly pay down the debt), and they're also a good building block for a financial system—banks, insurance companies, and endowments are going to end up with piles of assets that they won't need to touch for a long time, but that they also don't want to draw down to zero, and a safe asset is a way to accomplish both.

Bonds are also useful to historians. At specific points in time, they tell you which countries were trusted more than others—the 17th through 19th century interest rate gap between what the English and Dutch governments paid traces the story of reserve currency transition nicely. And over time, rates can tell you how big a deal certain events were (one nice touch in Andrew Roberts' Napoleon is that after major battles, he'll tell you French bonds reacted so you know whether or not the outcome was a big deal.)

One interesting dynamic with bonds is that they enable governments to make longer-term decisions, and for political leaders at a particular point in history to bind their successors to their choices. One example the book cites is that Lincoln was able to get the transcontinental railroad through Congress because the legislators who would have blocked it had all defected to the Confederacy. After the war, the debts stuck around. One side effect of this is that it means their successors, constrained by these obligations, are weaker—unless they, too, expand the scope of the government and borrow to fund it.

More financial technologies ended up getting built on bonds: the book has chapters about the rise of rating agencies (which displaced underwriters as the main tool for lenders to assess borrowers), the eurodollar market, mortgage-backed securities, credit default swaps, currency and rates derivatives, and more. Bonds are a surprisingly useful building block, and they and their derivatives let investors express very precise views about the world.

Bonds look boring compared to stocks. Skill issue! As Michael Milken once pointed out, the nice thing about bonds is that if you're right, you get paid no matter what anyone else thinks. (The book talks about Bill Gross using this insight: he'd look at high-yield bonds that were about to mature, but still carried high interest rates that implied ridiculous odds of the company running into financial trouble in the next few months.) The bond market is a compendium of promises made by institutions, with the credibility of those promises marked to market in real time. Given enough leverage, a bond portfolio can be just as zippy as one focused on equities, and in some ways it's simpler to isolate the various risks involved in bonds—duration, credit, prepayment—in order to take just the one you want. It's a triumph of civilization that we can all be so collectively certain about promises made by strangers.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• A while back, The Diff wrote a few pieces on logistics—like looking at how warehouses are priced in two dimensions but monetized in four or how logistics networks are similar to computers. What’s new since then? Especially interested in talking to startups; fragmented industries plus changes in information technology often yield interesting new companies.

Diff Jobs

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

• General Catalyst backed founder is looking for technical generalists who are excited about creating a fully autonomous firm. If you want to own the entire process of identifying, instrumenting, steering, and monitoring post-PMF businesses that run autonomously, and study how machine intelligence actually behaves in the real economy, this one is for you. (Remote, SF)
• a16z, Sequoia, Thrive-backed next generation pharma company that instruments the best drug hunters in the world with deterministic software, data and AI to accelerate the process of acquiring, developing, and monetizing promising therapeutics (three active immunology programs) is looking for senior software engineers. If you are excited about sublinear headcount scaling and building technology that drives therapeutic abundance, please reach out. (Boston, NYC, SF)
• High-growth startup building dev tools for wrangling and debugging complex codebases is looking for someone who can personally execute the SaaS bear case: review the third-party software they use and figure out what to keep, what to drop, and what to implement in-house. (SF, DC)
• A top prop trading firm is looking for people who combine exceptional quantitative ability with the judgment and relationship instincts to build critical financial partnerships and lead high-stakes negotiations. Candidates may come from markets, finance, or further afield; what matters is this combination of abilities, strengthened by the perspective that comes with experience. (NYC)
• A top prop trading firm is looking for people with exceptional strategic thinking and quantitative skills to help the firm model and manage its financing risk and strategy. Open to candidates from a range of analytically demanding fields. (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.

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.