Longreads + Open Thread

Longreads

• Adam Strandberg does some admirable detective work to trace the origin of the myth that chess grandmasters burn 6,000 calories per day. It seems to result from a one-man game of telephone, where Robert Sapolsky conflated a few metrics (peak breathing rate while playing chess versus average breathing rate in other cases became average energy consumption while playing chess), and then rounded everything up, including the chess players' skills (none of the players in the study were grandmasters). Part of the job in academia is to popularize work, and one way to do that is to pull out a few fun facts. But part of the responsibility of the job is to triple-check your work, and resist the normal process by which memes evolve into a more virulent form and loses vestigial features like factual access.
• Dwarkesh has a collection of his notes on Joel Kotkin's Stalin biographies, ahead of an interview. It's a good preview for the interview, but also interesting to see the same research process presented in a different medium—lots of semi-disjointed notes instead of a cleanly-edited video interview and transcript. It's also a good illustration of how much knowledge about history compounds: there was a famous period of catch-up growth in East Asia in the last half of the twentieth century, but that's an echo of the catch-up/keep-up growth cycle Europe went through over the preceding century. Pre-revolutionary Russia makes more sense in comparison to 1960s Korea or 1980s China, as a politically-uncertain place trying to maintain government control while embracing economic growth.
• Anton Howes on the long, strange history of lower-smoke coal. One surprising detail is that the idea that conservation is economically beneficial&dmash;that pollution is a proxy for the inefficiency of some process&mdsash;is much older than previously thought: from the 1790s: "the enormous waste of fuel in London may be estimated by the vast dark cloud which continually hangs over that great metropolis, and frequently overshadows the whole country, far and wide; for this dense cloud is certainly composed almost entirely of unconsumed coal." And finally, this piece illustrates that the deployment of some technologies is contingent on natural resource endowments and labor costs, and that good ideas promoted before their time can be forgotten and rediscovered many times over.
• José Luis Ricón on the concept of agency, and why it's a useful thing to have but a hard thing to observe. This is probably why it generates so much discourse: "You can just do things" can mean "You can drop out of school and embark on a life of adventure" or "you can really just follow the rules and, in the developed world, lead an incredibly nice life." These both show a certain kind of agency.
• Philo on "thesis drift," a concept that comes from markets but, like many other such concepts, is clearest there but present in many other places. It's a sort of streamlined version of the idea that the purpose of a system is what it does—if your reasoning changes a lot but your behavior never does, it's a sign that you have no idea what you're doing.
• In this week's Capital Gains, we consider what makes companies cyclical, how it happens, and how it occasionally reverses.

Books

Softwar: An Intimate Portrait of Larry Ellison and Oracle: Software companies that sell to businesses can have crazy nonlinear changes in valuation, where VCs decide that they're worth 50x or 100x revenue. That kind of bet can work when investors have a good model of where software fits into an existing stack, how it can be priced, how it gets sold, etc. Softwar is, basically, a book about how that knowledge was won.

Like many novels, the book opens in media res and then tells the rest of the story in flashbacks. We begin with Larry Ellison rushing from one meeting to the next, flogging his latest software suite, while tech spending collapses in the wake of the Y2K spending burst and the end of the dot-com bubble. It sounds stressful: at one point, two executives at an Oracle client company realize Ellison is traveling in their general area, so they hop on the company jet to tell him in person how peeved they are about the quality of Oracle's latest release. Ellison's messaging in this tour is interesting, because Ellison's big themes are 1) you should not be hiring lots of systems integrators to customize Oracle's products (and, coincidentally, take money out of your budget for buying Oracle), and on a related note 2) you shouldn't buy software customized to your business practices, but should instead customize your business processes around the software. Which is all incredibly self-serving, but also in a sense true: if you can squash the marginal cost of sending information or executing simple rules to roughly zero, that completely changes what your company ought to be good at.

The backstory is one of a company that was both technically strong and incredibly good at sales, but occasionally not great at managing cash flow; they had a near-death experience in the early 90s where they almost ran out of cash (and, as a nicely literary premonition, this started right around when one of their brand new incentive programs was paying some sales commissions in gold). But they managed to survive that, get back to growth, and expand from databases to ERP to CRM to everything else. The stretches that take place in the 90s ought to be a relief to anyone at a high-growth company right now, because Oracle is described as completely chaotic, full of corporate infighting fueled by weird grievances, and run by a CEO who's occasionally checked out to focus on yachts, mansions, etc. And yet, everyone does well; even the people who lose out in corporate knife-fights end up with a big pile of stock options, even if they have the strange habit of getting suddenly fired right before another tranche vests.

(Conspiracy-minded readers will be delighted to know that not only was Oracle's first customer the CIA, and "Oracle" was actually the code name for an internal CIA project. They also mention that the first big customer category was other intelligence agencies, including the NSA. The less conspiratorial read is that of course the government needs database software, and during the Cold War they tended to be more tech-forward. And if you're trying to create a company with state-compatible backdoors, you probably want to give it funding through front companies and you definitely want to coach the CEO about the importance of not blabbing about all of his intelligence ties to friendly corporate biographers.)

Ellison does produce some notable wisdom about selling complex software packages to big customers. One of the interesting ones is his claim that analysts and the media tend to like whichever product is growing fastest, and if you're dependent on client demand and on integrations with other products, you need that endorsement. So, better sales and marketing can actually make a product better for customers even if the product itself doesn't change. This has some explanatory power, for better or for worse.

The book's style is unique. Ellison is clearly a fan of deals as well as technology, and his arrangement for the book was that he cooperated with the author but got the right to include his own footnotes. This means he gets the last word every time one of his former colleagues complains about his behavior. And it gives him the opportunity to vociferously deny that he would ever wear a pink tank top ("This is very important," he explains). So you're getting a rough draft of text history, annotated by one of its most important players. A lot has changed since the book was published in 2004 (though Ellison doesn't seem to have aged much). Softwar shows how that change happened.

Open Thread

• Drop in any links or comments of interest to Diff readers.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• Ex-Optiver quants with over a decade of experience in HFT and AI are rethinking time series prediction from first principles. They are building a research lab, initially monetized via derivatives trading, with the goal of building a better representation of the future. The team is hiring a founding engineer, with experience building real systems: distributed compute, ML pipelines, etc. Python, C++, or Rust preferred.
• Ex-Citadel/D.E. Shaw team building AI-native infrastructure that turns lots of insurance data—structured and unstructured—into decision-grade plumbing that helps casualty risk and insurance liabilities move is looking for a data scientist with classical and generative ML experience. (NYC, Boston)
• An OpenAI backed startup that’s applying advanced reasoning techniques to reinvent investment analysis from first principles and build the IDE for financial research is looking for software engineers and a fundamental analyst. Experience at a Tiger Cub a plus. (NYC)
• A Google Ventures-backed startup founded by SpaceX engineers that’s building data infrastructure and tooling for hardware companies is looking for a product manager with 3+ years experience building product at high-growth enterprise SaaS businesses. Technical background preferred. (LA, Hybrid)
• Deerfield-backed, Series A company building agents for healthcare administration (prior authorization, eligibility checks, patient scheduling) is looking for a senior software/AI engineer to build backend services and LLM agents. Experience building and monitoring production-quality ML and AI systems preferred. (NYC, Hybrid)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.