Longreads
- What did high-flying tech stocks look like a century ago? This post has share price data and some fundamentals for RCA, which was the default-metonym hot stock of the 1920s. One notable feature of RCA is that it doesn't look so much like a growth stock as like a deep tech stock: there's a long period where the company was in the wilderness, still proving out its technology and looking for a business model. Once the network effect kicked in, where more radios meant more money invested in radio production, which means more appealing shows, it suddenly took off. Even if the details are different, some of the economic forces are consistent. Via Marginal Revolution.
- Ron Garret has a retrospective on starting six different ventures, each of which failed. (In between two of them, he took a job at pre-IPO Google.) One of the most illuminating is that he started a ride-hailing company, iCab, right around the same time that Uber started. What didn't work there was that he couldn't find a good way to get drivers onto the platform. In one sense, Uber's an incredibly obvious-in-retrospect idea: once we all have a combined communications system, GPS, and point-of-sale terminal in our pockets, using those to create a better taxi network makes sense. But going from the infrastructure to a network where drivers are confident they'll find passengers and vice-versa means solving a chicken-and-egg problem, and probably spending a lot in each market to get the flywheel spinning.
- Benjamin Breen says AI works well as a tool for assisting historians. One thing historians seem to do a lot of is taking advantage of adjacent knowledge: if you're trying to understand what was happening in a given country at a given time, knowing what was happening elsewhere at the same time is useful context. If you want to know how some trend evolved, looking at what it started as and what it turned into later does the same. AI is very good at this kind of thing, because that's what it's literally trained to do: token-by-token, it's assembling text that's near what it started with in some multi-dimensional space. But the ways AIs do this and the ways historians do are different, and seem complementary.
- Patrick McKenzie on the bizarre Chicago Bally's fundraise, where the casino offers 100:1 leverage (at a steep interest rate), but only to select investors. It's an oddly multi-layered compromise: Bally's promised to raise capital from investors with certain demographics, but wasn't especially excited to go out and find people who'd cut big checks. They could, however, embed enough leverage that their investors were making token payments—and then provide that leverage at an interest rate high enough, and an overall valuation high enough, that they'd probably recapture most of the upside that way.
- Physics Forum, a decades-old site that is exactly what it sounds like, has been posting LLM-generated answers to old questions, using the handles of users who haven't logged in in years. It turns out that the site was experimenting with using LLMs for one of their popular use cases, answering obscure queries, but in a deceptive way. This gets framed as part of "dead Internet theory," the idea that most of the content online isn't being produced by people at all. (But it's worth noting that dead internet theory predates LLMs—making it, in a way, another piece of science fiction that predicted the future.)
- In Capital Gains this week: when you're gathering data, always start with a model of what you think is going on. You run the risk of confirmation bias, but it also gives you a way to update your views if you're disciplined about it.
- This week's Riff covers the DeepSeek panic, Threads, Elon, and national champions. Listen with Spotify/Apple/YouTube.
Books
In last week's Longreads, we reviewed Monday Starts on Saturday, a book in which magic is real—by reciting the right incantations, particularly skilled wizards can literally manipulate the state of reality and bend it to their wishes, a power many such wizards abuse. This week: The Caesars Palace Coup, which is a story of a case where that's literally true: distressed debt.
The book tells the story of Caesars, the casino conglomerate, which turned into a money-printing machine by carefully tracking its customers, identifying the exchange rate between free perks and gambling losses, and then exploiting the gap aggressively. This made them a big, profitable business, with a theoretical growth path. (They're also a very rare case outside of biotech, non-bio tech, and finance, where an academic researcher's work is so impressive that he's literally put in charge of the company: the CEO who led this initiative was Gary Loveman, who'd gotten connected to them through consulting projects he undertook while teaching at Harvard Business School.)
The basic plot is that Caesars got taken private, and the economy almost immediately went into a recession that made it unlikely that they'd be able to meet their obligations. The PE firms that owned the company did a series of transactions that 1) gave the main business a bit more liquidity and breathing room, and 2) moved more of their key assets outside of the main entity where bondholders could get at them. There were some pretty egregious-looking deals, but the company was also pretty distressed at the time, and when a company doesn't have many alternative sources of capital, it will end up signing some lopsided deals with whoever is actually willing to inject more cash.
Compared to other finance books, this one spends a lot more time with lawyers than bankers and operating executives. A company is more than a big pile of contracts, but in some legal sense it is a big pile of contracts. There's alpha in interpreting those creatively, though that alpha often comes at the cost of all of the intangible, non-contractual assets that make a business worth more than whatever capital has been put into it and whichever assets that capital acquired.
Open Thread
- Drop in any links or comments of interest to Diff readers.
- It looks like tariffs will really happen. Place your bets: what are the interesting second- and third-order consequences of them?
Diff Jobs
Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:
- YC-backed AI company that’s turning body cam footage into complete police reports is looking for a tech lead/CTO who can build scalable backend systems and maintain best practices for the engineering org. (SF)
- 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 product-oriented software engineers. Experience at a high growth unicorn a plus. (NYC)
- A hyper-growth startup that’s turning customers’ sales and marketing data into revenue is looking for a forward deployed engineer who is excited to work closely with customers to make the product work for them. (SF, NYC)
- A growing pod at a multi-manager platform is looking for new quantitative researchers, no prior finance experience necessary, 250k+ (NYC)
- The treasury management arm of an established stablecoin project is looking for a research economist to bring a macro perspective at the intersection of traditional monetary theory and digital currency innovation. Advanced degree and portfolio of high quality research output preferred; no crypto experience necessary. (Remote)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.