Longreads + Open Thread
Epic, Twitter, Industrial Policy, Clubbing, Stripe, Trains
Programming note: The Diff is off Monday for Memorial Day. Back Tuesday the 31st.
Patrick McGee of FT interviews Epic's Tim Sweeney on the Metaverse ($). Part of the metaverse debate is over whether the metaverse is for entertainment or for everything; Epic comes down on the side of entertainment. What's interesting about this is that the more ambitious view should be the more compelling one, but entertainment seeps into other domains quite easily (one big reason computers found their way into homes was for video games, and that was part of the early adoption of smartphones, too). So the right bet might be to start with entertainment and then see if the scope naturally expands from there.
This Vice piece on the collapse of a drug cartel has a fun they're-just-like-us vibe: drug lords, too, can get in trouble for what people say on Twitter. Cheaper technology has reduced some parts of the gap between rich people and poor people; there isn't an even fancier iPhone out there that only the 0.01% can afford, and they're on the same social media sites as the rest of us. That also means that some of the downsides of those technologies are equally distributed, too.
Via Jordan Schneider of ChinaTalk, a good in-depth report on China's industrial policy. A lot of the material revolves around how to define and measure industrial policy; a loan can be anywhere from a market transaction to a guaranteed-to-be-forgiven subsidy depending on the context. One thing that stands out: while China's industrial policy is the biggest both in absolute terms and in comparison to GDP, the US is actually the #2 industrial policy spender globally, mostly through government-supported research and R&D tax credits.
On The Spectrum, On The Guest List is a new Substack delivering exactly as promised: a neurodivergent person's first-person perspective on New York's club scene. It's a very pure market for status, with lots of intermediaries who take a big markup from the fact that nobody wants to directly pay for status but lots of people with money will in fact do exactly that through layers of indirection.
Forbes has a good profile of Stripe. Good quote: "The vision has always been, 'Why can’t we move money around in the cloud the same way that we can move data?' "Because isn’t money just data?" There is a fairly literal answer here: moving money is a more adversarial environment, and anyone who helps deal with that adversariality wants a cut. Over time, you can approach the asymptote where the money/data distinction collapses, but money will always be a somewhat special category of information—the most valuable, but also the hardest to clearly reason about.1
Railroaded: A fascinating look at the making of the transcontinental railroads, with particular emphasis on how corrupt and mismanaged they were. The book is a good illustration of what complex supply chains looked like before the cheap computing: railroads knew that they had plenty of pricing power, and that some loads were more profitable than others, but didn't have a good way to calculate their own profit and loss. (And the statements they gave investors added a layer of imaginative accounting on top of that.)
Railroader: A more recent book on the same industry, this profiles super-successful CEO Hunter Harrison, who was the CEO of four railroads and put together an enviable track record at each. Harrison is a testament to the power of sweating the small stuff: at one point he booked a hotel with a view of the rail yard, and then called the station to ask them why a particular box was in the wrong position. (They fixed it.) Part of Harrison's success stemmed from the fact that railroads are a monopolistic business and he ran them accordingly, dictating terms to customers in order to run a more efficient operation. His operating model has since been adopted by other companies in the industry, and the railroad industry in general has had a strong couple decades after a tough century or so. Industry turnarounds can happen, though sometimes they take a surprisingly long time.
Drop in any links or comments of interest to Diff readers.
If you're reading a business newsletter a generation from now, what industry is most likely to be in the position of North American railroads, where it had a long and bleak decline before finally leveling out and delivering superior returns again?
There are some cases where transmitting data is hard because of the same adversariality. One way big institutions collapse is that as local knowledge gets passed up the chain, the good news gets exaggerated and the bad news gets downplayed. This is part of the story of the Vietnam war, the war on drugs, and the declines of companies like US automakers and IBM. That's an adversarial environment, too; if job security and internal politics are more important than the good of the organization as a whole, information decays as it moves up. In that model, newer tools like live dashboards of KPIs are partly a way to reduce the number of steps information has to take before it reaches the people who make decisions around it. They aren't perfect—choose the wrong KPI, and you'll make the wrong decisions—but they illustrate that solutions are feasible for this category of problem.