Longreads

• Richard Dewey et. al. trained a model to play a simplified version of liar’s poker via self-play, and then pitted it against experienced human player. They also had it play against LLMs (one interesting note there is that the LLMs tend to play cautiously; they speculate that part of what’s happening is that so much poker advice for beginners suggests folding more often, and that’s carrying over to this domain). Liar’s poker turns out to be a surprisingly complicated game with a vast number end states, so playing it means doing a tiny bit of deterministic reasoning and accumulating an arsenal of nested heuristics—which is a good description of a lot of machine learning.
• Alison Killing in the FT on the incredible scale and rapid retrenchment of Neom, especially of The Line. It’s sometimes worthwhile to meditate on just how big even conventional infrastructure projects are. One World Trade Center’s steel consumption was about half the Roman empire’s annual iron ore production. And that kind of comparison is even more extreme when the project in question is so enormous: executives at The Line talked about how they were building the largest occupied structure ever, and then plan was to make twenty copies of it. The plans are more modest now. Neom is in one sense a prudent effort to diversify the economy of a one-industry country. But that industry is cyclical, and that means the amount of capital available for diversification sometimes bobs up and down.
• Boaz Barak on the counterintuitive economics of AI. This is a very good piece, that thinks clearly about bottlenecks: if we automate lots of labor, and that makes us richer, that makes the remaining labor much more valuable. But AI messes up the classic growth equation, because it’s a case where capital is increasingly fungible with labor. We’ll need a whole new formula to even describe what growth looks like in an AI-heavy economy.
• Amir Fischer interviews Susquehanna founder Jeff Yass. Especially fun given the full-throated defense of prediction markets, in which he argues that they could have hypothetically prevented the Iraq war and more theoretically prevented the Civil War. The basic concept is that they’re a live ticker of public figures’ lies: if the President says that a war will be cheap, or a policy will prevent inflation, or whatever, it’s very nice to have a live market betting on whether or not that will happen. But markets are only one mechanism for establishing consensus, and not necessarily a popular one. Prediction markets might end up producing accurate assessments that a given policy will be disastrous, but that could very well give populist advocates of these policies more political energy.
• Dan Kagan-Kans on writing for AI, and the writers who explicitly aim to do it. If your writing is incorporated into models, it will make some tiny incremental change to how those models work. If your writing happens to include a uniquely good answer to a popular question, your impact will be that much bigger. It’s still small in the scheme of things, but one way to frame this is to look at the most influential people who ever lived, and ask what fraction of people’s actively-used beliefs Abraham, Mohammed, Jesus Christ, or the Buddha can claim direct responsibility for. The answer is orders of magnitude higher than the human average, but it’s hard to argue that any single historical figure has cracked 10% on that metric. Writing for AI can have a similar expected value to writing for humans, albeit with a different input: an incredibly successful thinker today might have a life-changing impact on 0.01% of people, whereas someone writing for AI would change everyone’s behavior by 0.01% instead.
• This week in Capital Gains, we ask: why does volatility matter? And, if it matters, who benefits from being in a position to care about it less?
• On ReadHaus, users were asking for my opinion on prediction markets. The Diff party line is that they’re a very interesting product whose positive externalities are hard to capture by their owners. However, the owners are able to take a vig from imposing a negative externality on society by enabling lots of gambling. The ones who decide to moderate more aggressively will probably be ahead of the ones who get regulated into doing so.

Books

Who Knew: Barry Diller was once pretty young to be running an entire movie studio on his own, and is now surprisingly old for a dot-com dealmaker (his corporate mothership, IAC, has owned and subsequently spun off brands like Expedia, TripAdvisor, LendingTree, and Match/Tinder). The first half of his autobiography feels like a Gore Vidal novel: he grew up feeling deeply ashamed of his sexual orientation, and morbidly afraid that his parents would find out (especially since they had their hands full with his drug-abusing older brother). Early on, he seems pretty dysfunctional, and once he joins the workforce at the William Morris Agency through a friendship—the early chapter is peppered with references to partying at Lew Wasserman’s house, being friends with Doris Day’s son, etc.—he looks like a different archetype, the well-connected but unambitious entertainment-adjacent underachiever.

It’s hard to tell what clicked for him. One minute, he’s dragging his feet on leaving his mailroom job because he knows he won’t make it as an agent. Then he uses yet another connection to get a job at ABC, and suddenly he’s a media dealmaker who more or less invented the made-for-TV movie category.

Soon enough he’s off the Paramount, which he started running when he was 32. This is the most fun part of the book, especially for someone who doesn’t have that much awareness of middlebrow 70s and 80s movies. (The most bizarre concept he mentions is Heaven Can Wait, “a 1978 American sports fantasy comedy-drama” involving a football player who keeps getting reincarnated. It may not be entirely unrelated that Diller finds out his driver, Mario, was also in the business of selling cocaine to Diller’s friends.) These chapters give Diller a lot of time to talk about people he mentored, and also to talk about some of his grievances. He’s still mad that George Lucas didn’t honor his commitment to do a Raiders of the Lost Ark sequel on the terms he agreed to before making the first one a hit.

And that makes sense. The media business is a weirdly fragmented one, where there’s extreme scarcity for some things (there are only so many good weekends to launch a movie, actors, writers, and directors are doing one thing at once, and early in Diller’s career broadcast time slots were also scarce). And it’s a business with unpredictable payoffs. And it’s a business with big information asymmetries, and lots of opportunities to strategically close them. So the whole business is basically structured so lots of incredibly egotistical people have numerous opportunities to betray one another, develop grudges, reconcile when it’s strategically sound, and top it all off with a pretty fun memoir.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Are there any good histories of the broad economic cycle of vices? There are some narrow ones—Last Call on Prohibition, for example—but it would be interesting to see what the long-term pattern of bans, sin taxes, and regulatory capture looks like.

Diff Jobs

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

• A hyper-growth startup that’s turning the fastest growing unicorns’ sales and marketing data into revenue (driven $XXXM incremental customer revenue the last year alone) is looking for a senior/staff-level software engineer with a track record of building large, performant distributed systems and owning customer delivery at high velocity. Experience with AI agents, orchestration frameworks, and contributing to open source AI a plus. (NYC)
• A transformative company that’s bringing AI-powered, personalized education to a billion+ students is looking for elite, AI-native generalists to build and scale the operational systems that will enable 100 schools next year and a 1000 schools the year after that. If you want to design and deploy AI-first operational systems that eliminate manual effort, compress complexity, and drive scalable execution, please reach out. Experience in product, operational, or commercially-oriented roles in the software industry preferred. (Remote)
• A leading AI transformation & PE investment firm (think private equity meets Palantir) that’s been focused on investing in and transforming businesses with AI long before ChatGPT (100+ successful portfolio company AI transformations since 2019) is hiring Associates, VPs, and Principals to lead AI transformations at portfolio companies starting from investment underwriting through AI deployment. If you’re a generalist with deal/client-facing experience in top-tier consulting, product management, PE, IB, etc. and a technical degree (e.g., CS/EE/Engineering/Math) or comparable experience this is for you. (Remote)
• YC-backed, ex-prop trader Founder building the travel-agent for frequent-flyers that actually works is looking for a senior engineer to join as CTO. If you have shipped real, working applications and are passionate about using LLMs to solve for the nuanced, idiosyncratic travel preferences that current search tools can’t handle, please reach out. (SF)
• Ex-Bridgewater, Worldcoin founders using LLMs to generate investment signals, systematize fundamental analysis, and power the superintelligence for investing are looking for machine learning and full-stack software engineers (Typescript/React + Python) who want to build highly-scalable infrastructure that enables previously impossible machine learning results. Experience with large scale data pipelines, applied machine learning, etc. preferred. If you’re a sharp generalist with strong technical skills, please reach out.
• Fast-growing, General Catalyst backed startup building the platform and primitives that power business transformation, starting with an AI-native ERP, is looking for expert generalists to identify critical directives, parachute into the part of the business that needs help and drive results with scalable processes. If you have exceptional judgement across contexts, a taste for high leverage problems and people, and the agency to drive solutions to completion, this is for you. (SF)

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.