Sparta, LARPs, Ivies

Plus! YC Undoes Something That Doesn't Scale, Why There Aren't Enough Masks, Out-Aggregating Amazon, and more...

This is the once-a-week free edition of The Diff. Paid subscribers get at least three issues each week. Topics in the last week:

Sparta, LARPs, and Ivies

Sparta will be a popular historical topic forever, because part of teaching history is keeping antsy students entertained, and the entire concept of Sparta sounds like it was invented by a twelve-year-old boy. Training from an early age for nonstop war? Awesome. Being fed a starvation diet and told that the only thing wrong with stealing food is getting caught? Rad. Fighting a suicidal last stand outnumbered twenty-to-one? Cool!

We’ve inherited more ideas from Athens (representative government, philosophy, cancel culture, but the idea of Sparta is something we’ll have for a long time.

Which is weird, because the Spartans lost. Not just in a glorious way, at Thermopylae, but in a much more gradual, depressing way. Sparta was important for a while, faded into irrelevance, lost some insignificant wars, and eventually merged with a larger political unit that was later absorbed into Rome. If the Spartan military was the most badass fighting force of all time, surely the nation-state that beat them would have an even more stellar reputation. (It was the Boeotians. I had to Google.)

One reason Sparta stuck around in the collective consciousness: rich kids. Long after Sparta was defeated by Boeotians, and eventually conquered by Rome, upper-class Romans would send their sons to train in Sparta. (This means that when the Founders LARPed as ancient Romans, they were LARPing as ancient LARPers. And when modern writers LARP as founders, they have reached the fourth degree: LARPing as LARPers who LARPed LARPers.)

Usually, failure is overdetermined, but success is not, and Sparta was successful for a long time. It was also static: they had a very well-defined system, and a strict social hierarchy. Unlike other city-states, they didn’t really use money, because there wasn’t much to buy; professional soldiers can only carry so much equipment, and everyone who wasn’t a soldier or a member of their family was some variety of slave with no property rights whatsoever.

One way to look at Sparta was that they were a failure of premature optimization. A deeply unequal system can either collapse when the majority rebels, or get very good at repression; Sparta chose the latter. In a near-subsistence economy, the cost of weapons relative to disposable income is prohibitively high, so a small number of well-armed people can, in fact, keep a large-and-restive population under control. (More recently, although still quite a long time ago, roughly 20,000 British soldiers and administrators maintained Britain’s control over India, a country whose population at the time was roughly 300 million. This didn’t last forever, but went on for a surprisingly long time.)

But a city-state that’s perfectly optimized for one set of circumstances is not well-designed to deal with change. And if part of its meta-optimization is a strong sense of tradition and skepticism of change, then it’s eventually going to be left behind. It’s very hard to piece together ancient economic history, but lead emissions gradually rose in the centuries after Sparta was founded, which indicates more mining activity—and thus cheaper weaponry—which meant that over time, elite militaries mattered less than big ones.

I worry about this story—a country with a narrative but few original ideas, that’s perfectly adapted to particular circumstances and loathe to question them. It seems like something that could happen anywhere.

Clearly, some of the US economy is geared towards true innovation. But a lot is designed around optimization. Banking, consulting, and private equity are all industries built to match up ideas, assets, and people—not to change any of them. And while big tech companies have improved our lives in myriad ways, they, too, spend most of their resources optimizing something that was built a long time ago. Most of Google’s revenue comes from AdWords, launched in 2000; Facebook’s core product, the News Feed, dates back to 2006. These products have changed a lot since launch, but those changes are, comparatively, tweaks. Fundamentally, the idea of AdWords was commercial content related to the implied intent of a search query; the idea of the news feed was a stream of content that could include commercial messages relevant to the user. There are many ways these products can change, but it’s hard to imagine a future where Google or Facebook launches something that replaces them.

Academia is supposed to be a source for new ideas, but there, too, there’s a template: the university-to-grad-school-to-postdoc-to-professorship-to-tenure track forces exceptionally smart people to design their lives around a very particular, hyper-competitive career track after which they can do something original. It makes them impressively bland. It’s not impossible for someone to get tenure and then start exploring bold and risky ideas, but the entire process seems designed to filter those people out. You have to have a Donnie Brasco-level commitment to keeping weirdness under wraps. Going crazy is another option.

It might make sense that a country, like a company, would focus more on optimization than innovation over time. We’re rich enough to waste a lot, which means that reducing waste—in the form of misallocated capital and under-diffused best practices—is valuable. But the parallels are worth meditating on. After all, Sparta was not the only distant land rich kids would visit to get an education.

Covid-19 Updates


Out-Aggregating Amazon

Amazon is somewhat notorious for building its own Basics version of best-selling third-party products. This is, depending on who you ask, a monstrous antitrust violation or exactly what every retailer in history has done. Either way, it’s an opportunity: Thrasio has raised funds to beat Amazon to the punch, buying out companies with top-selling third-party products. It’s a hybrid case of software and PE competing to eat the world.

YC Undoes Something That Doesn’t Scale

Y Combinator has slowly increased how much cash it offers to companies. The original deal was pegged to a grad student stipend, they eventually upped it to $150k, and they later raised a growth fund to invest pro-rata in YC companies' Series A rounds.

They’re now pulling back on that. YC has always understood signaling very well: a big chunk of the wealth they create is providing a better signal of skill than universities. So they’re executing this in a sensible way: they still reserve the right to make follow-on investments, albeit at smaller scale, and they’re only investing once a company gets a term sheet from a lead investor.

This says something interesting about the efficient frontier in venture. The two hard problems in VC are 1) deal flow: getting access to the best companies, and 2) managing convexity: continuously reinvesting in the best of those to maximize the option value of pro-rata rights, founder relationships, and a better understanding of the company in question. YC has basically redefined dealflow, to the point that other early-stage investors have to optimize around YC. But the convexity-management part of the venture equation clearly comes into play early. YC’s alpha from acceptance to demo day is unparalleled.

Once a YC-funded company raises their Series A, the market inefficiency has vanished, and at YC scale it asymptotically approaches a venture index fund. Not terrible; investment management firms have had exclusive high-alpha vehicles and much lower-alpha index-tracking alternatives before, and anyone in a position to write a check to RIEF) knows it’s not Medallion. But YC has the same problem Jack Bogle had in the 70s: the minimum size of an accurate index fund is fairly high, whether it’s buying a market cap-weighted basked of 500 stocks (you try it) or buying 7% of dozens of A rounds.