Longreads + Open Thread

Longreads

• "Language models transmit behavioral traits via hidden signals in data": an incredibly fun concept. Train a model to have some trait, like having a particular favorite animal or being evil. Then, ask the model to generate some random data with no specific associations, like a list of random numbers. Train another model with those random numbers as inputs, and the new model will tend to have the same traits. It's a very strange trait for a computer program to show but a very ordinary trait for an intelligent thing to show—memories triggered by smells, fetishes, and OCD are cases where comparatively tiny or irrelevant stimuli provoke a disproportionate response in natural intelligence. That kind of weird disproportionality might be an inevitable result of anything complicated enough and adaptive enough to be considered smart.
• Tyler Cowen interviews Helen Castor, a scholar of Early Modern England. It's interesting that England was the most centralized state in Europe at that time, given that in a few hundred years the Anglosphere model was closer to decentralized free trade instead. That's easier to reconcile if building up baseline certainty about enforceable property rights is the minimum requirement for an efficient decentralized system. It's sort of Marxist historiography applied to capitalism: you need a sufficiently powerful and effective government to start effectively outsourcing things back to the market.
• Elad Gil on the current state of the AI market. One point he makes is that the winning move for wrapper companies is to ship something and get customers locked in before the product is great, on the assumption that the underlying models will keep improving. If LLMs are almost-useful in some category, it's a good bet that they'll be adequate soon and amazing in a year. Another result of this is that basically every wrapper company will look overhyped at first.
• This essay on the decline of quality is interesting because it's wrapping two phenomena pushing in opposite directions to tell the same story. It is true that in some categories, like clothes, the average quality of what we buy has declined, in part because the cost has dropped even faster and clothes have become disposable. But then it says things like this: "Psychologist Albert Vinyals, author of El consumidor tarado (The Disordered Consumer) (2019), recalls that years ago, the first thing car ads highlighted was their longevity. “Now we don’t even consider it,” he notes over the phone." The reason car ads don't mention durability is the same reason that Osborne Computer ads from the 80s talk about how the computer only weighs 24 pounds and can fit under an airline seat, while modern laptop ads tend not to say much about weight—they're light enough that it doesn't matter! Similarly, the reason car ads don't talk as much about durability is that modern cars last so much longer than they used to. In some categories, quality is rising. In other categories, the decline of quality works in a different way: there used to be a really nice version of some product, available only to the very rich. The version that's good enough for everyone is not as good, but it's incredibly cheap. As with many other such complaints, when you say "People used to be able to afford X, and now they can't," what you mean is "My parents were relatively well-off, which I assumed was normal, and I've reverted to the mean faster than the mean has gone up."
• Park MacDougald does white-collar gonzo journalism, describing the subjective effects of various stimulants on his work. Biology is reluctant to offer free lunches, but in this domain modern medicine at least presents some interesting tradeoffs.
• In Capital Gains this week, at some point, everything is a relative value trade. It's a good idea to evaluate an investment by asking how it's positioned in the supply chin and what dependencies look like on either side. But past a certain point, there's no option but to assume an efficient market.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Are there categories where quality has noticeably declined and it doesn't just reflect the product being more widely-available? Surely there are some, but in that case the question becomes: what leads people to provide a higher-quality product or service than what the market would let them get away with?

Diff Jobs

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

• Well funded, Ex-Stripe founders are building the agentic back-office automation platform that turns business processes into self-directed, self-improving workflows which know when to ask humans for input. They are initially focused on making ERP workflows (invoice management, accounting, financial close, etc.) in the enterprise more accurate/complete and are looking for FDEs and Platform Engineers. If you enjoy working with the C-suite at some of the largest enterprises to drive operational efficiency with AI and have 3+ YOE as a SWE, this is for you. (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 a technical degree (e.g., CS/EE/Engineering/Math) or comparable experience and deal/client-facing experience in top-tier consulting, product management, PE, IB, etc. this is for you. (Remote)
• 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.