The Diff

This is an audio version of the free newsletter. Read the newsletter at thediff.co, and subscribe for an additional 3 posts (and podcasts!) per week.
  • (00:00) - Money and Legitimacy
  • (06:53) - Moats
  • (07:36) - Big Tech Sees Like a State
  • (08:43) - Shifting Costs
  • (09:31) - Managing Macro
  • (10:11) - Experimentation

What is The Diff?

The Diff is a newsletter exploring the technologies, companies, and trends that are making the future high-variance. Posts range from in-depth company profiles, applied financial theory, strategy breakdowns and macroeconomics.

## Money and Legitimacy

In Coup d'État: A Practical Handbook, Edward Luttwak notes that one of the first things successful coups do is to seize radio stations, TV stations, or newspapers. For the plotters, the ideal way for everyone to hear about the coup is to hear that it already happened, and that someone new is definitively in charge. (Specifically, the narrative they often use is that someone else was plotting a coup, and that a few quick-thinking colonels have arrested the perpetrators and have generously agreed to stay in charge until things settle down.)

This raises the question of what seizing power looked like before mass media. And there's an interesting answer: the first stop for a successful coup was wherever the reigning monarch lived, but the second stop was often the nearest mint, to start creating coins in the new ruler's image. would reach its apogee with the sultanate of Mehmet III in 1595, when nineteen coffins containing the bodies of his brothers were carried out of the palace.] William the Conqueror was putting his face on coins two years after the Battle of Hastings, Charlemagne did it after getting promoted to emperor in 800, and one of the reasons Caesar upset the Roman Senate was that he'd made the then-unprecedented decision to mint a coin with the portrait of a living person (him).

This makes plenty of sense. Internally, in an environment with low literacy and few kinds of physical media, it's a good way to get the news of a new ruler to as many people as possible. Externally, it tells other countries a) who's in charge, and b) that there is someone sufficiently in-charge to mint currency that can be used to buy goods elsewhere. This is a big deal; in the fiat currency era, a good working definition of an economically-independent government is that it can create a currency that's accepted as payment outside that country, whether that's in the form of paying for imports with it or borrowing from external investors in local currency.

Fiat currency still provides that same extension-of-legitimacy effect. Every time bilateral trade is denominated in dollars, or global companies headquartered outside the US report their financials in USD, or worried savers convert legal deposits in their rapidly-inflating home currency into gray-market deposits in dollars, it burnishes the US's status as a superpower. One irony of this is that one contributor to the dollar's power is that the US has, historically, exercised less control over money flows than other countries. (The mark was an important currency, but it didn't get much traction as a reserve asset or tool for global trade, because Germany was slow to liberalize currency flows as fast as other big economies. The mark was more stable, which made it an attractive reserve asset, but German policymakers recognized that if the value of the currency were to rise enough to reflect this, it would hurt German exports.)

The dollar sometimes gets used as a policy tool because of its centrality to the global financial system. Sometimes, the logic works like this:

1. A given transaction happened in jurisdictions where it's legal. But it would be illegal in the US, and the US doesn't like it.
2. One of the parties to that transaction was a global bank. That bank naturally does lots of business in dollars.
3. The US threatens this banks's access to the dollar system if it doesn't adhere to US rules.

This would be called "imperialism" if the prototypical victims weren't French investment bankers doing business with various unsavory dictatorships. It's very hard to tell what the limit is for sanctions-based diplomacy, because the wealth and power of the targets is so hard to measure. It's even hard to categorize this—it's not quite soft power (people learning your country's language because they like K-Pop, anime, or reruns of Friends), and it's not really hard power (the capacity to blow stuff up), but some intermediate thing.

This money/legitimacy collection will get more important over time if network states start to take off. The shortest possible definition of a network state doesn't involve issuing currency, but the one-paragraph description does. It's a useful concept. The optimal size of the entities that best correspond to our modern idea of "nation-state" is not a constant, for both technological and social reasons—if territory is easy to defend or costly to attack, you tend to get a larger number of countries, and if country A invading country B tends to cause more reprisal from country C, that also leads to more fragmentation. The changes that make network states possible are that 1) it's easier to reach critical mass with narrow affinity groups, and 2) it's possible for these groups to run a trade surplus with the rest of the world by exporting services, there being a high overlap between people with interests so specific and extreme that they seriously consider starting a country together and people who have valuable technical skills.

It's likely that most network states will find that, even if they have their own currency, and even if they like it, they won't be able to import many products with it, except in cases where the seller is humoring them by pricing things in their currency with the plan to immediately export it back. Small, open economies tend to find that the best currency to use is one that's already in use by their trading partners: Ireland, Dubai, and Hong Kong all either adopted an external currency or pegged their currency to one. Singapore is an exception, but this is partly a historical artifact: in the beginning, they were doing plenty of business in British pounds, and keeping reserves in pounds was part of how they maintained relations with their former mother country. Now, they essentially benchmark their currency to a basket of other countries' currencies. This makes a lot of sense for a financial center and entrepôt; they'll often be the intermediary between, for example, one party that transacts only in yen and another that denominates everything in dollars, so keeping their currency stable with respect to their trading partners in the aggregate is a safe bet.

Issuing a currency makes sense for a country that has enough domestic economic activity that there's fairly predictable demand for that currency. And many countries don't qualify—for a lot of them, the financial trap that they run into is that they lack the state capacity to implement a currency peg, and lack the even higher level of state capacity required to issue and manage a useful currency. Which loops right back to the argument about currency and legitimacy: given money's network effects, the world arguably has too many nation-state-issued currencies, but perhaps too few company-issued pseudo-currencies. The latter works when it can be managed well and exploited ruthlessly, but for most purposes, just opting into an existing currency system saves them a lot of trouble. As transaction costs decline, that could easily change—the network effects of currencies are weaker if translating between them is cheap and happens invisibly to the end user. On the other hand, translating between currencies is a classic area where payments companies know they have more pricing power, because it's not as if their customers are all incredibly excited to individually figure out the most cost-effective way to turn zloty into kyat or vice-versa. Everyone needs currency, but if you're going to have one that you call your own, you need a very good excuse.

## Elsewhere

### Moats

One of the most irritating sources of competitive advantage is doing work that nobody thought was worth doing, and that turns out to be worth doing exactly once. This shows up in a surprising range of fields, from commercializing moonshot research to building a distribution network to keep blue collar workers' uniforms clean. For Google, one of those is building a presence in India, a country with a low GDP per capita and high linguistic diversity. This was an expensive prospect early on, and involved lots of the kind of hard-to-scale activity that Google is notoriously averse to. But the net result is that advances in translation and speech recognition mean that Google's AI tools can offer a more universal interface.

### Big Tech Sees Like a State

A fun question to ask MBA students is: consider a feature for a product that doesn't make the product any better, doesn't affect consumers' decisions to buy it—but does make it a more attractive target for thieves. Something like the set of features that make the Kia challenge a possibility. Does this produce shareholder value? When does it not?

That's obviously an extreme example, and most people are not cynical enough to deliberately engineer a subpar product just so customers will have to buy more replacements. But businesses vary in how much it pays to keep products safe from customers. All else being equal, higher theft rates mean more purchases, offset by an eventually tarnished brand. But for products that monetize through usage, the seller's incentives in this respect are closely aligned with the buyers': the thieves are stealing their customer lifetime value, too! So, Google has added additional anti-theft countermeasures to Android. They're creating a positive externality by making phones—which are lightweight and have a readily-ascertained resale value—harder to steal. But they're also doing that because they have a direct economic incentive.

### Shifting Costs

An interesting interview with the head of AWS, touching on, among other things, AI. One notable bit: Amazon's AI coding tool was designed for use cases beyond writing new code, like updating legacy code. At one level, legacy code is a subsidy for AWS, because it means the same product running in a predictable way, and if customers don't want to pay to update the code, but also don't want to shut it off entirely, they're implying high willingness to pay. But making it financially cheap to rewrite old code makes it cognitively cheap to sign up for yet another recurring AWS expense—if you're confident that the unit economics will improve, you can be more willing to start spending, and Amazon's bet is that enough people who start spending on this basis won't stop. (And that if their code runs faster and is cheaper to maintain, over time they'll spend more.)

Disclosure: Long AMZN.

### Managing Macro

A recent Capital Gains piece talks about one reason central banks move slowly—every economic cycle has slightly different rules from the last one, so historical data isn't a perfect guide to how the economy will respond to different interventions. China is going through a fun example of this right now: the rise of deep discount shopping services has led to deflation in consumer goods prices. Platforms and merchants have different incentives: merchants want to make a profit on the specific products they sell, but platforms are fine with any given merchant having negative gross margins if that leads to more overall purchase volume and fees. So the platforms will, sometimes sneakily, push some sellers to keep lowering prices.

### Experimentation

With a given set of rules, the bigger a free trade zone is, the better: there will be more value-creating transactions available to more parties. But crafting rules gets harder when there are more interest groups, so these arrangements tend to experience a slow decline in utility after they've been around for a while. The EU is exploring a workaround, where instead of trying to integrate all members' capital markets, they'll do experimental integrations with a subset of them. The tradeoff is that it slightly defeats the purpose of a common rulemaking organization if there are multiple sets of rules covering different members. But it also helps address the uncertainty-driven stalemate.