Longreads + Open Thread

Longreads

• Calvin French-Owen reflects on working at OpenAI. If there's one thing to take away from this, it's that OpenAI is early, and still has many of the characteristics of young companies that are figuring things out despite being one of the world's most valuable businesses on paper. This shows up throughout the essay, as it talks about fluid team structure, intentional duplication of similar projects, and the like. The conventional wisdom in startups is that every time you triple your headcount, everything breaks. But that also means that any process you can keep using as you scale is something that creates a distinctive company culture. In this case, one of the weird answers is: everything happens on Slack, rather than over email. The categories of email and messaging have bled together a bit, but this is still surprising—long, internally-coherent documents are a pretty natural medium for research, but apparently not the only possibility.
• Kevin Kwok with a more positive look at "HALO," or hire-and-license-out, deals. These are the quasi-acquisitions that happen in AI where they're mostly buying the founders and research team, but cutting a check to the rest of the company and leaving it behind. One point he makes is that we just haven't seen enough of these deals to figure out what standard terms should look like, and that over time employee equity compensation and investor's rights will reflect this as a novel risk factor. One irony of the entire situation is that if HALOs are sometimes the right move, it means that there are fewer extreme upside opportunities, and that to compensate for this, early-stage companies will have to skew their compensation more to salary and less to equity, and have to find some way to either assure people that the company will try to stick around for a while or that it'll be worthwhile to work there for a year or two and then leave. So big tech's demand for certain flavors of talent is so insatiable that it's actually warping startups to make them function a bit more like big tech.
• The NYT has a great piece profiling a sports betting handicapper/social media personality who has obviously tilted his business model more towards the marketing end than towards analytical rigor. This worked: he has 2.5m followers on Instagram, though the piece isn't able to dig up any evidence that his picks are good or that he has any long-term subscribers. They do find a lot of evidence that he's enjoying his handicapping fees. Any time there's a business devoted to selling information, there's a degenerate case where that business spends all of its resources marketing itself as informed and none of them getting informed. Though part of successful sports betting is understanding the psychology of more naive market participants. As the article illustrates, there are other forces that keep this from being taken to too much of an extreme.
• Jonathon P. Sine on Stalin and Russia, as informed by the Kotkin Stalin biographies. One of the arguments against a revolution of any kind is that the incumbent system is simply too complicated for anyone to understand, much less completely replace with a new and better system. And the answer history offers is that successful revolutions change titles but not the fundamental org chart: you confess your sins to a commissar rather than a priest, you lobby the nomenklatura rather than the boyars, and you'd give your life for General Secretary of the People's Republic rather than the Tsar. So in one sense, revolutions don't matter. On the other hand: Tsarist Russia was not on track to achieve a standard of living comparable to Western Europe. Year-by-year, the gap in output growth was small, but it accumulated. For a while, the Soviet system was able to achieve higher growth, but eventually it hit its limit. Revolutions matter, but it takes a long time to see what they really accomplished.
• Paul Bloom writes in The New Yorker about chatbots and loneliness. It’s a balanced piece: an LLM isn’t a complete substitute for talking something over with an actual person, but it can be better than nothing for someone who doesn’t have access to a sympathetic person to talk to. There are second-order tradeoffs if you no longer need to maintain human relationships in order to have a shoulder to cry on, but if nothing else the usual lesson from technological history is that adoption happens based on first-order effects and we all figure out how to manage the second-order ones.
• In Capital Gains this week, we consider the concept of "the number." Sometimes, the performance of a stock almost entirely boils down to some metric, usually a growth-focused one in tech but sometimes something closer to margins. It's an exciting time in a company's life when you can boil everything down to a single metric, but it never lasts forever.

Books

When the Clock Broke: Con Men, Conspiracists, and How America Cracked Up in the Early 1990s: It's always fun read history that you remember happening. Whether your reaction is to remember how seriously people took what was ultimately a minor event, or wondering how something so weird couldn't have completely changed the world the way it felt like it would at the time. Reading history you were slightly too young to experience firsthand is a different feeling, because it turns out that the political history I do remember from the late 90s was very much an aftershock of the early-90s weirdness. In When the Clock Broke, John Ganz offers a tour of this strange era in conservative politics.

That weirdness seems to have been something between disaffected Reagan true believers who felt that they didn't quite get what the wanted during the 80s and saw the country sliding back under Clinton and Bush, economic anxiety among manufacturing workers and people hit by various local economic issues (S&L closures in New England, defense layoffs on the West Coast), and general Last Man at the End of History ennui.

It's tempting to say that this book bounces between fringe figures and the mainstream, but almost everyone profiled in the book achieved some mainstream success. An ex-Klansman (at least, that's what he said at the time) who used his supporters' money for gambling, once wrote a dating guide for women under a pseudonym, etc. sounds like a marginal figure. But David Duke won around 40% of the vote in two statewide elections in Louisiana. Various conservative radio hosts said things that would probably get a Fox News host taken off the air for a while, but still had audiences of millions. If they were fringe, they represented a pretty big fringe.

It's not just a story about outsiders, though; there's plenty of material about the establishment. Ganz has a great chapter on the LAPD, and how they slowly evolved into a politically independent entity that had very different attitudes about vigorously enforcing rules internally versus externally. Ross Perot was a wealthy executive before he got obsessed with prisoners of war in Vietnam and eventually talked himself into making a run for the presidency.

A book like this makes sense right now because many of Donald Trump's critics have bought into a fundamental tenet of Trumpian Epistemology, i.e. that every phenomenon in earth can be best understood in reference to Donald Trump. He makes a few cameo appearances in the narrative, and the closing of the book notes the way Trump's path mirrors that of his constituents: the 80s were good years, fueled by leverage, but somehow they didn't quite make the jump to the 90s economy, and when Clinton talked about low unemployment and a declining deficit, it probably sounded like unseemly and out-of-touch bragging to someone who'd had higher real earnings at their factory job two decades earlier. Making your way into a more service-sector job might be satisfactory, and for Trump it was better than average, but at some point he realized he wasn't the only American who really missed the 80s.

Open Thread

• Drop in any links or comments of interest to Diff readers
• What are some other historical periods that are worth revisiting today?

Diff Jobs

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

• A blockchain company that’s building solutions at the limits of distributed systems and driving 10x performance improvements over other widely adopted L1s is looking for an entrepreneur in residence to spearhead (prototype, launch, grow) application layer projects on their hyper-performant L1 blockchain. Expertise in React/React Native required. Experience as a builder/founder with 5–10 years in consumer tech, gaming, fintech, or crypto preferred. (SF)
• Deerfield-backed Series A startup 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)
• Ex-Citadel/D.E. Shaw team building AI-native infrastructure to turn 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/agentic ML experience. You will develop, refine, and productionize the company’s core models. (NYC, Boston)
• A company that was using ML/AI to improve software development/systems engineering before it was cool—and is now inflecting fast—is looking for a product marketing manager to articulate their value proposition and drive developer adoption. If you started your career in backend engineering or technical product management, but have since transitioned (or want to transition) into a product marketing seat, this is for you. (Washington DC area)
• A Series B startup building regulatory AI agents to help automate compliance for companies in highly regulated industries is looking for legal engineers with financial regulatory experience (SEC, FINRA marketing review, Reg Z, UDAAP). JD required; top law firm experience preferred. (NYC)

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.