Longreads + Open Thread

Longreads

• Ruth Michaelson in BusinessWeek writes about bank robbers in Lebanon who are robbing banks to get access to their own funds. It's a depressing story, and the background is a bit of a throwback: Lebanon was able to achieve economic growth by pegging its currency to the US dollar and then allowing credit expansion at home. In the short term, especially in a credit-constrained economy, that's actually a winning strategy: the economy gets more financial firepower to support the currency peg. But if it goes on long enough, the banking system ends up undercapitalized, dollars get scarce, and the financial system collapses.
• Alex Tabarrok has a good paper on Baumol's cost disease and its connection to the "Linder Theorem," the observation that the richer people get, the busier they seem to be. This is noticeable in big cities; of course people in New York walk fast, since there are so many places to go! It's a generative set of ideas, helping to explain everything from US healthcare costs to affordable haircuts in the developing world to video on demand. And it's a great example of a pedagogic observation about economics: even economics students can struggle to explain the concept of opportunity cost—but the average person definitely acts as if opportunity cost is real.
• Pradyumna Prasad has a good piece on how the Economic Freedom Index ignores some important factors, most notably the government's ownership of the economy and land-use regulations, giving misleading numbers. There are two parts to this: the land use point is absolutely correct; limiting housing supply in San Francisco, Hong Kong, and London is a government policy that reduces economic freedom by raising prices. On government ownership of land, and of stakes in businesses, the answer is trickier. In one sense, equity ownership operates equivalently to corporate taxes: either way, in the long run, the government collects some share of a company's profits. But marking the value of future tax receipts to market, which is effectively what happens when you replace a 20% corporate income tax with a 20% stake in the business, does have value: it provides a real-time view of the market's estimated impact from policies. It also makes some things more commensurable: it's hard to price the importance of domestic control of some strategic company, but putting a literal price on it, where when the government cedes control of some company it gets a cash windfall, helps to make the calculation easier.
• William Broad in the NYT has a surreal but fun article on how annoying the number three is. In many systems, one participant is trivial to model, two are pretty straightforward, and three—whether it's three stars in a solar system, three magnets affecting an object, or three superpowers fighting for dominance—have nonlinear effects. (What about four or more? At some point, you get to graduate to being able to look at statistical averages where one-on-one interactions mostly cancel each other out.) It's adjacent to the phenomenon that for most phenomena, there are roughly zero, roughly one, or pretty much infinitely many instances; there are other numbers, sure, but they get tricky fast.
• Tyler Cowen interviews Reid Hoffman, mostly about AI. Hoffman talks about AI regulations and the idea that every AI should have some legal person or company associated with it, and Cowen makes some good points about how hard that is to implement in practice. (Will every cloud company verify that every process run on their machines is not, by some definition, AI, and that if it is AI it has a responsible owner?) There are also good thoughts on AI in education—biting the bullet and having ChatGPT generate mediocre essays that students then edit into excellence. And the best point from the piece, about both the limits and complementarity of AI tools: LLMs are much better at answering questions than at asking them. Which makes sense; an answer fits the "predict the most likely next token" model very well, whereas a good question starts with a model of the world and awareness that this model is incomplete. One way to frame this: AI can answer questions, but so far can only simulate asking them.
• In this week's Capital Gains, we look into how one should read an S-1. (A confession here: yesterday's Diff post about an AI/makeup IPO ($) looks like it's following the advice here, by finding a case where the company buries an important disclosure about their business model in the "Risks" section while portraying it differently elsewhere. But what actually happened in this case was that I asked my wife about the company's products and got a short course in what people find frustrating about DTC subscription economics.)

Books

• American Republics: This book covers American history from 1783 to 1850 (part of a longer series). The reading of this period that the book argues for is a twist on more purely slavery-centric narratives—the driving historical force in the book is a longstanding effort to avoid dealing with the question of slavery. American expansion ended up being a sort of ponzi scheme where every newly-claimed territory a) gave the pro- or anti-slavery side more influence in elections, and b) provided a safety valve where people dissatisfied with life in a more settled part of the country could move somewhere cheaper with more opportunities. This was a meta-instability on top of a system that was already fairly chaotic. It's a wonder it ever worked.
  
  Via Scholars Stage.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Some companies do some of their technical hiring by posting puzzles on their site and interviewing people who provide a good solution. But they tend to deprecate and eventually remove this as they grow. (Google, for example, once bought a billboard inviting people to go to [first ten-digit prime found in the digits of e] dot com in order to apply for a job.) Are there any companies out there that got a disproportionate share of their employees from this kind of source?

Diff Jobs

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

• A well funded seed stage startup founded by former SpaceX engineers is building software tools for hardware engineering. They're looking for a full stack engineer interested in developing highly scalable mission-critical tools for satellites, rockets, and other complex machines. (Los Angeles)
• A vertically integrated PE-backed company applying a rigorous investment/operations approach to a high-growth industry is looking for an analyst who has banking experience. (Little Rock, AR—no remote, but relocation assistance is possible)
• A fintech startup that gives companies with complicated financials a single source of truth for managing their cash flows and understanding their unit economics is looking for a founding engineer with JS, Typescript, Node.js, and React experience. (Bay Area, Hybrid)
• A new health startup that gives customers affordable access to preventative care and lifestyle interventions seeks a founding engineer. 7+ years of JavaScript experience preferred (TypeScript is ideal), and payments experience is a plus. A great opportunity for anyone excited to make healthcare better by treating problems cost-effectively before they're catastrophic. (US, remote; Austin preferred)
• A firm using machine learning to customize investments is looking for a data engineer. (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.