Longreads + Open Thread

This is the last Diff issue of the year. Thank you for reading! We'll be back January 2nd.

Longreads

• The Beeper saga of the last few weeks was a good reason to revisit David Auerbach's essay on working at Microsoft during the IM protocol wars, when Microsoft kept reverse-engineering AIM, and AOL kept changing it to stop them. These kinds of stories are always fun, because after a while the participants develop a sort of friendly rivalry and start leaving Easter eggs where the other side could find them. By the end, AOL's protocol was exploiting buffer overrun errors in the AIM client, which Microsoft's version didn't have. Microsoft did some of its own exploitation, like sending a whistleblower email from a fake external consultant. Zero-sum competition like this is far from the most important thing people do at work, but it's often the most fun. (Disclosure: Long MSFT.)
• Nilay Patel interviews Ryan Petersen of Flexport. There's some corporate politics here, and the obligatory attempts to wish ex-CEO Dave Clark well (Clark, for what it's worth, still omits Flexport from the work experience bit of his LinkedIn profile, but includes it under the "education" section). The best bits: logistics is traditionally a cost center and not a growth area, but Flexport can help customers ship globally once they're established in a single market, which does add some revenue. What it doesn't add much of is pricing power: Petersen argues that the undercut-for-share model simply doesn't work in a fragmented industry where companies compete on cost. They can get customers with cheap prices, but can't retain them when prices increase. Which makes Flexport a marginal cost story. And on the cost side, he has some good remarks on being an AI-second organization ($, Diff): AI hallucinates with complex tasks, but does very well with narrowly-defined ones ("scrape this page to extract this information" versus "track this container"). By trying to break the process down and automate the steps that are automatable, Flexport ended up with the technical infrastructure to run the same procedure on parts of their business that could only be automated with AI.
• Shreeda Segan interviews Emmett Shear on AI. Shear was briefly CEO of OpenAI (I hope he negotiated an hourly vesting schedule for his equity comp). One of his interesting points is that attitudes towards AI are a vector, not a scalar: in a sense, AI safety takes the techno-optimist view that new technologies are powerful enough to be dangerous, while at least some brands of safety-skepticism are implicitly pessimistic about AI's potential. (Of course, you can add another dimension to that, and note that humans have adapted to all sorts of technologies that rendered the world unrecognizable—electricity, internal combustion engines, agriculture, writing.)
• In light of US Steel selling to Nippon Steel, it's useful to revisit the peak of US worries about Japan—by way of this 1985 NYT story tracing the origins of Japan's growth and the US's slow response. Note that in this case, "1985" is a vague date; the story came out a few months before the Plaza Accord, which led to US dollar depreciation and a reduction in the trade deficit. Currency has an impact, but policy still matters, particularly in countries that are willing to craft complex rules to shield their domestic industries from foreign competition. In the end, what stopped Japan from being such a threat to the US was a combination of 1) a financial bubble and collapse, and 2) Japan's working-age population and hours worked per worker both declined ($, The Diff). On an output-per-labor-hour basis, Japan and the US had identical performance over the last three decades. The US's aggregate performance is explained by having more immigrants, having more kids, and working more.
• Here's a great Harvard Crimson piece on Harvard's notorious Math 55 class. This article was written in 2006, and it's fun to Google the then-freshmen and see how they turned out: many became professors (including one who dropped the class as too challenging!), several have research roles at big tech companies, and one ran FTX US. Part of the function of good schools is to gather together a group of people who are each the smartest person in their high school and demonstrate to them that there are much, much smarter people out there.
• In this week's Riff, we talked about OpenAI, macro, and McDonald's as a software company. Listen with Spotify/Apple/YouTube.
• In Capital Gains, we looked at when companies choose to buy out smaller competitors, when they copy, and when they collaborate. Adobe and Figma show one part of the model, but it's not the only approach.

Looking for a last-minute gift with instantaneous free shipping? Want to give someone a present they'll open again and again? Consider a gift subscription to The Diff, for 20% off!

Books

In honor of the holidays, here are the best books I read this year (shipping times vary):

• Aristocracy of Talent: "Meritocracy" has a long and tortured history, both as a practice and as an ideal.
• Investing Amid Low Expected Returns: This book looked poorly-timed when it first came out, because the market was down. Now, stocks are up and a book about what you can expect from the market if you buy at times like this is worth a read.
• The Alchemy of Air: Many different things are tied for most important invention in human history, because so many discoveries (or implementations thereof) are dependent on other discoveries. Still, the Haber-Bosch process is responsible for the fact that the world can feed more than about four billion people.
• Money Men: A Hot Startup, A Billion Dollar Fraud, A Fight for the Truth: When I originally reviewed this book, I mentioned the cliché that it "reads like a spy novel." There is now evidence that the main character in the book was indeed running a Russian spy network!
• Money for Nothing: A history of the South Sea Bubble, but also a story of state-imposed legibility and the growth of financial markets. Very, very worth reading.
• The Great Leveler: Violence and the History of Inequality from the Stone Age to the Twenty-First Century: A good and gloomy look at the various forces that lead to lower economic inequality, all of which are much worse than your neighbor driving a nicer car than yours.
• Trust: The Social Virtues and The Creation of Prosperity: Francis Fukuyama's book ages very well.
• Outsourcing Empire: How Company-States Made the Modern World: Like Money for Nothing, it's about the historical antecedents for things we take for granted today.
• Founder vs Investor: The Honest Truth About Venture Capital from Startup to IPO: A sometimes hostile but always fun exchange of views between a founder and investor on the nature of raising money.
• Project Boing: Genuinely fun novel about working at a big, bureaucratic company, from someone who overlapped with me at the big, bureaucratic company it's based on.
• The Real North Korea: North Korea once identified as Stalinist, is now seen by the outside world as a country ruled by a totalitarian god-emperor, and is, arguably, the world's only functioning anarcho-capitalist economy.
• Orality and Literacy: A growing share of human interaction happens in text form. But are you reading and writing, or are you really talking? This book explores the differences between these communication modes, and the kinds of thinking they engender.
• The Missing Billionaires: Annoyingly, the best way to build dynastic wealth is to run risk-averse strategies whose real returns slightly outstrip your and your descendants' fertility, and to let compound interest work its magic. (Though see The Great Leveler for an exploration of tail risks that complicate this model.)

Open Thread

• Drop in any links or comments of interest to Diff readers.
• One of the fun things about reading lots of 10-Ks and company transcripts is seeing little details of industry cycles that don't line up with the broad sweep of the overall economy. So, readers: how was business this year? What's looking different coming into 2024?

Diff Jobs

Realistically, not a whole lot of hiring gets done between December 23rd and the New Year. However, plans have a tendency to gel during enforced work breaks. But if you’re cooking up new plans for 2024 hiring and job changes, please reach out anyway and we’ll be happy to chat. 

Use the links below to get in touch, and set up a call for next year—if you'd like to have a more general discussion about 2024 career plans, just hit reply.