Longreads + Open Thread

Longreads

• Sophie Schmidt of Rest of World reflects on visiting North Korea a decade ago. In many fields, it's illuminating to strip things down to a simplified model, figure out how it works, and then add complications back in, one by one. We can model North Korea the same way: relationships between countries are complicated and always require some give-and-take, but North Korea doesn’t have much to offer aside from their ability to do harm to other countries. So North Korea’s behavior is very different from that of places that have broader interests and more of a formal export sector, but their policies still have their own internal logic. Per one expert quoted in the piece, "They've weaponized their weirdness."
• In The Atavist, Jessica Camille Aguirre writes about a wild set of interlocking scams involving emissions credits and tax evasion. Part of the magic of value-added taxes is that they're self-auditing: the end seller of a product pays a full tax, then gets reimbursed based on the value-added tax paid by their supplier, so each level of the supply chain wants the next level to report accurate numbers. This breaks down a bit when the entities are a chain of related parties, some of which are faking their numbers. In this case, the scam collided with value-added taxes on emissions credits, an entirely abstract product. But Peto's Paradox strikes again: scammers who did business together also scammed one another, limiting the size and increasing the visibility of the original scheme.
• Andrew Marantz profiles the AI safety subculture. This subculture's norms are very conducive to exploring contrarian theses; the more you emphasize substance over persuasive skill, the more variance you tolerate. But you can also think of the difficulty of buying into unusual ideas (like: AI has a nonzero chance of ending human civilization) as part of the collective memetic immune system: at least in our ancestral environment, most of the weird ideas were bad ideas, and that prior is still pretty strong. These contrarian theses are worth paying attention to in part because they have a long head start in thinking through the implications of more powerful computers, and in part because that same early interest in AI means they're well-represented at existing AI labs. Of course, AI research is never aggressive enough for the accelerationists and never quite cautious enough for the safetyists, so that statement can be quibbled with. But it's worthwhile, especially for AI optimists, to seriously engage with what the other side thinks might go wrong, if only to add to one's own library of dangerous mental biases.
• This 2010 post is a look at what was going wrong with Blackberry at that time, but also a good guide to what a declining hardware business looks like in the early stages of its decline. In an ideal scenario, a company is shipping more units and adding new features fast enough that it can expand gross margins over time—if you were happy with your first iPhone or two, you're more willing to splurge on a higher-end model for your next upgrade. If that doesn't work, cutting prices can be the next-best alternative, but unless there's either a plan to resume price increases later or to introduce some recurring software product that offsets that margin weakness, those price cuts are the beginning of the end.Via Two Natural in the comments from last week's piece. 
• Gabe Fleisher in Politico profiles a formerly-pseudonymous 20-year-old college student in the UK whose Twitter account, @ringwiss, is known for expertise on US congressional procedures. At an object level, it's a very DC-y kind of post, with wonks going crazy over how wonky someone is. But the most striking part is that @ringwiss only started paying attention to American politics in 2020. At least in this case, four years of part-time work is all it takes to go from no reputation whatsoever to being a recognized expert in a field. There are many, many such opportunities. 
• In this week's episode of The Riff, we tak about personal hedging, the economics of outages, data feedback loops, and why alpha is rarely for sale. Listen with Twitter/Substack/Spotify.
• In Capital Gains this week, we look at graceful corporate aging: why it's often a bad sign that a growth company is profitable early on, what corporate maturity looks like, and why it's good for companies to die.

Books

Funny Money, published in 1985, details the rise and fall of Penn Square Bank. It's a now-obscure company, but in the late 70s and early 80s, Penn Square was one of the US's fastest-growing banks. They'd tapped into two related trends: first, high energy prices meant that dollars were piling up in oil-exporting countries, and found their way back to big US banks, which then sought out borrowers so they could put those funds to work. One growing set of borrowers: independent oil and gas exploration companies in Oklahoma, where Penn Square was based. So Penn Square originated loans, passed most of them on to larger banks, and kept a small slice of the risk in order to align incentives.

The broad outline of that story is familiar to anyone who lived through or read about the great financial crisis. One difference in the 1980s was that banks were less levered, less dependent on wholesale funding, and generally less interconnected, so a credit problem in one industry didn't lead to liquidity seizing up everywhere else. The way financial crises get bad is when some asset or strategy that looked money-good suddenly turns out to have some risk; shrinking the set of assets that people treat as money is equivalent to suddenly, sharply reducing the money supply. But the root cause of a change in the perception of moneyness is a change in the perception of bankiness: Penn Square's larger bank counterparties assumed that the bank was underwriting carefully, but it mostly grew fast by getting sloppy with paperwork, and, for a while, kept its financial performance looking good by lending more to borrowers so they could service their older loans. Not all medium-risk loans to small businesses are created equally.

The book has some good asides and nice character portraits; it's quite fun to read about a teenage oil entrepreneur whose oil play is ruined by flooding—a banker tells him that, since he's under 18, he isn't legally on the hook for the debts he owes. The entrepreneur explains that he just doesn't do business that way, and the banker gives him a loan which gets him back on his feet. It's a great story about either personal honor or incredible risk tolerance. (As it turns out, mostly the latter; here's what that founder did later in life.)

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Are there any other older financial crises and scandals that are worth revisiting today?

Diff Jobs

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

• A seed-stage startup is using blockchains to enforce commitments and is in need of a fullstack developer with Solidity experience. (Remote)
• A company building ML-powered tools to accelerate developer productivity is looking for ML researchers with a knack for converting research into Github repos. (Washington DC area)
• A fintech company using AI to craft new investment strategies seeks a portfolio management associate with 2+ years of experience in trading or operations for equities or crypto. This is a technical role—FIX proficiency required, as well as Python, C#, and SQL. (NYC)
• An alt data provider is using LLMs to scrape and is in need of a backend engineer—experience building with LLMs required, web scraping experience preferred. (Remote)
• A well funded seed stage startup founded by former SpaceX engineers is looking for full stack engineers previously employed by Anduril or Palantir. (LA)

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.