The Diff - Longreads + Open Thread

Longreads

• Adam Iscoe writes about the lively secondary market for reservations at popular New York restaurants. There's the obvious outrage angle about how scalpers are making a full-time living transferring reservations from people who lucked into getting one to richer people who just bought one. But the piece also raises the meta question of why restaurants don't monetize the reservations directly or charge enough that there's rarely enough demand to put them at 100% capacity. If someone's going to be price-gouging, it might as well be whoever is responsible for that delicious entree or the delightful amuse bouche. But restaurants, particularly high-end ones, are selling a combination of food and atmosphere; charge full price, and diners are subjected to the awful experience of being surrounded by merely rich people, instead of having the dining room sprinkled with at least a few younger guests who are out for an annual splurge instead of a weekly ritual. So this is really a story about technological changes overtaking social norms; underpricing the food and making reservations hard to get is a good way to inject some healthy randomness into who gets to dine there. Arbitraging that away eliminates a market inefficiency, but it was an inefficiency that people liked.
• Dan Schulz interviews Sebastian Mallaby, author of The Man Who Knew, More Money than God, and The Power Law (I strongly recommend all three). There are some good bits on the craft of writing biographies and profiles: showing people drafts of what you're writing about them gives them an opportunity to make corrections, but also creates an opportunity for follow-up interviews. There are also fun thoughts on the Fed, and the tradeoff between transparent communications of future policy and the ability for that policy to be taken seriously—forward guidance is only credible to the extent that central banks stick with it even when it's exactly the opposite of what makes sense under new circumstances, which is to say that it can't be relied upon at all. And the interview closes with a very fun synthesis on Marx and Carlyle: changes in technology are an important driver of history, but the outliers in that model are all willful and unusual individuals.
• Constance Grady at Vox looks at the ecosystem of garbage ebooks, which used to be dominated by books written on-demand by whichever writer had the lowest bid, and which are increasingly written by AI instead. Historically, one of the convenient features of spam was that scalably sending a message meant sending exactly the same text to numerous people, which was easy to detect. AI has made it harder, since responses to the same prompt will vary. But there are also better tools for automatically measuring text quality. Unfortunately, that runs into two problems: first, the exact problems that AI-generated text has—hallucinations, and an inability to follow a thread consistently—will be limitations in the models used to detect it. And second, there is a long tail of content that looks like garbage to almost everyone, but that still has an audience that likes it. The extremely weird Kindle Singles romance microgenres might be supplied by not-especially-talented writers or by AIs, but they exist because there is, in fact, demand for that stuff.
• Philo at MD&A explains why stock buybacks are fine, with more than a little merited frustration. It's all there: buybacks are like dividends, but with fewer perverse incentives; they aren't big enough to meaningfully distort share prices over the long term; and if it's bad for big companies to return capital to shareholders, the biggest companies get even bigger. A tax on buybacks is a tax on Apple and Google not aiming to be $5 trillion companies. Maybe that's optimal (I like both companies' products a lot, personally, and use them all the time), but it should be part of the debate.
• Viola Zhou in Rest of World covers the culture clash when American engineers started working for TSMC. In some ways, this is an incredibly bizarre piece—it's a wonderful privilege to grow up in America, a wealthy and stable country whose primary language is by far the world's most popular second language, and it's petty to complain that decisions at the world's best chip manufacturing company get made in Mandarin and Taiwanese Chinese merely because Mandarin- and Taiwanese-speaking engineers figured out the chip-manufacturing process faster and better than English-speaking ones did. Setting that aside, it is notable that TSMC has such a strong culture of long hours and tight deadlines. On average, you should expect productivity per hour to drop with more working hours; you're a lot less focused after twelve hours in the office. But in knowledge work, that's not necessarily true, because your productivity is a function of how much information you have at hand and can readily access. For chip fabrication in particular, a field that's heavy on both theoretical and tacit knowledge, one person working eighty-hour weeks is more valuable than two people working forty-hour weeks, because one of those two people can't telepathically communicate important knowledge to the other. That is unpleasant, and means that the job is unsuitable for many people. Fortunately, there are plenty of other places to go if you're okay with working at the Nth-best rather than best in the industry.
• In Capital Gains this week, we look at whether investors should think more about GAAP earnings or free cash flow. Free cash flow is what's actually available to shareholders, and GAAP numbers can be gamed—but free cash flow can also be gamed! In the long run, that's hard to do, but such is the case with GAAP as well. And in the very long run, they sum to the same amount: accrual accounting just represents expectations about future inflows and outflows of cash.
• On this week's episode of The Riff: income-sharing agreements, the privacy/security/usability efficient frontier, and how Meta ended up with all their GPUs. Listen with Twitter/Substack/Spotify/Apple/YouTube.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• We’d like some more ideas for Capital Gains. Any burning questions about finance, economics, and corporate strategy that need the Cap Gains treatment?

Diff Jobs

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

• A startup building a new financial market within a multi-trillion dollar asset class is looking for a junior ML engineer, especially someone interested in using LLMs to make unstructured data more tractable. (US, Remote)
• A company building the new pension of the 21st century and building universal basic capital is looking to add a legal ops hire. (NYC)
• A well-funded startup is building a platform to identify compliance risks associated with both human- and AI-generated outputs. They are looking for a cloud infrastructure engineer to join their team of world-class researchers. (NYC)
• A private credit fund denominated in Bitcoin needs a credit analyst that can negotiate derivatives pricing. Experience with low-risk crypto lending preferred (i.e. to large miners, prop-trading firms in safe jurisdictions). (Remote)
• We’re looking for generalists with nontraditional backgrounds and some programming experience for a variety of roles. 

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.