Casper, Learn to Code, and Dengism with American Characteristics

Plus! Marketing, China, Peak California update, unweirdness

The Latest

Casper filed to go public last Friday. I was impressed, in the limited sense that they’ve articulated a pivot from a bad business (mattresses) to a better one (pretty much anything else related to sleep). Selling long-lived durable products is hard, because people forget about your brand long before they make a repeat purchase—there’s a reason autos are the single biggest category of local TV ads. But they’re well below the scale where they can break even, unless every other direct-to-consumer mattress company flops in short order. And the best catalyst for that is a bad IPO for Casper.

There’s a popular narrative that tech people are ignorant of the humanities, but I think there’s a better argument that it’s the other way around. I am not the first person to make this argument; several people informed me, with varying degrees of hostility, that I really should have cited C. P. Snow. Guilty as charged, although you’ll notice Snow’s article doesn’t have as many sick Twitter-sized burns.

And, today, a meditation on American productivity growth and business culture: are consulting and private equity Dengism with American characteristics? It’s quite hard to measure, but bad news if so.


Fred Wilson draws attention to an important point: in the early 2000s, marketing online was close to free, and paid marketing was a negative signal (“If you don’t rank well on Google, does anyone actually like your product? If they’re not telling all their Facebook friends to use it, do they even use it?”). In the last five years, the narrative has shifted, and lots of late stage rounds are raised based on an elaborate customer acquisition cost vs customer lifetime value narrative, with CAC as the part of the model that can absorb arbitrarily large, Softbank-sized sums of money. (Back in 2011, someone tried to make “A Romney” slang for $10,000, after his infamous bet. Maybe we’ll start using “A Masa” the same way. Wag raised 3 Masas at a 3.5-Masa pre.)

The market for marketing dollars has gotten more efficient, and not in a way advertisers like: Google, Facebook, and Amazon have monopolies on three categories of marketing dollars (Google: general purchase intent; Facebook: branding; Amazon: immediate purchase intent). Now they’re charging accordingly.

Some good China updates. First, Huawei is bad news. And overrated, too, with fewer meaningful 5G patents than you’d think. And, on a more measured note, China’s economic model borrows from many countries, including the early 19th century US. The American habit of massive banking panics followed by catastrophic depressions turns out to have been a competitive advantage.

A nice update to the Peak California thesis: Silicon Valley’s comparative advantage has shifted from tiny, risky startups to colossal monopolies with data- and distribution-based economies of scale. The political coalitions haven’t gelled around this yet: on the left, you have people who are skeptical of big business, concentrated power, and the potential risk to democracy; on the right, the concern is big business run by the kinds of people who go to Burning Man and don’t have kids (or who have kids but go to great lengths to raise them pseudo-Amish). On trade, significant portions of the right have decided that it’s best to have no industrial policy, but if we must have one it should be written in DC and not Beijing; they may come to a similar conclusion on tech, keeping their friends close and their enemies closer.

Finally, Tyler Cowen interviews Reid Hoffman and asks the question I’ve wondered about: why isn’t Reid weird?

The Book

Regular readers may recall that I mentioned collaborating on a book a while ago, on the subject of secular stagnation and counterexamples thereto. The book is part of the broader “progress studies” movement, and we are, indeed, making progress. If you look at the last fifty years of economic history, the question is always “Why did productivity growth in the rich world slow down so much?” but if you look back 5,000 years instead, you ask “What’s the deal with that spike in productivity growth at the very end?” So far, the history of scientific progress feels a lot like science itself: something happens for completely coincidental reasons, and you try to work backwards to figure out what you can learn from it.

A quick reader request (not the first; won’t be the last): does anyone know where to find data on college graduates by major in the early twentieth century? One working theory of why physics was so fecund around then is that Gödel scared all the math geniuses away from working in pure math, but that theory would work much better with actual data.