Longreads + Open Thread

Longreads

• There's a tradition sometimes when a company gets a little too enthusiastic about hokey corporate propaganda, and also gets a little sloppy with the quality of its products, the time has come to deploy a verbal assassin. This time, it's Rachel Handler at Grub Street, who wants to know why chocolate, particularly from Hershey's, is so bad these days. Many companies go through a frog-boiling exercise where they slightly tweak how they make their products in order to cut costs; prices and brand names are sticky, so the input costs provide disproportionate wiggle-room. And any one of these changes will often be imperceptible in a blind taste-test. But if they're all in the same direction, then the cumulative effect can be pretty bad, and a company that's engineered around these incremental tweaks is not one that's well-equipped to handle big changes.
• Perry Mehrling (RIP) wrote about the inherent hierarchy of money: many of the things we treat as money are really claims on more money-like money. In everyday life, it's completely normal to think of the balance in a checking account as "money," but this is just a number in a database maintained by your bank; to the extent that it functions as money, it's because there's a (completely reasonable) social convention to treat the bank's deposits as equivalent to dollars, and as exchangeable on fairly predictable terms for goods and services. But one way to describe credit expansion is to think of it in terms of expanding the scope of what counts as money—the term "money-good" should always be a little alarming.
• An anonymous-for-now contributor to Astral Codex Ten reviews Storm of Steel. Some books drop you into a completely alien culture: you might enjoy The Iliad or Emma, but you will never be a bronze-age warrior or a bored aristocrat looking for a husband. Storm of Steel is special in that it describes a particular worldview that 1) used to be a lot more common, and 2) had a very high correlation with dying in a glorious bayonet charge somewhere in France in the 1910s. Ernst Jünger somehow survived this, and produced a book that reflects the peak of a particular kind of martial culture, right at the moment when those virtues had been eclipsed by engineering and accounting as the means of winning wars.
• Don Fox of Plimco Journal has an informative and also pretty catty response to this piece on the Victorians, linked a few weeks ago in The Diff. The basic argument is that 1) the Victorians were not just the upper class Oxbridge Latin-and-sherry circuit, and 2) also, the Victorians didn't do an especially good job. Some of it's fair, some goes too far: empires peak, and then they decline, and even if that decline can't be reversed, there are good and bad ways to handle it. The fact that some Victorian elites spent so much of their time doing what we'd now call networking is good evidence that running a political entity that big was an all-consuming task, and obviously no one knew how to do it since it had never been done before.
• Conor Fitzgerald of The Fitzstack argues that there are high returns on being disagreeable: "Calling someone a nazi, or a transphobe, or a misogynist works very well as a rhetorical tactic and will kill most people’s willingness to talk about certain things where such an accusation can be a risk - though it doesn’t kill their interest in talking about them. What you are also doing here is incentivising the social importance of people who don’t care about being accused of indecency, who will proceed in spite of those accusations, or who (for whatever reason) kind of like it." This does something interesting to the distribution of credibility: it creates a cohort of people who are slightly distrusted by the average person, because they say crazy things, but are very trusted by their fans, who like how they tell it how it is regardless of consequences. Some of these people will just be naturally indifferent to criticism, and some of them will lean into it. But the result is more polarization, because most of us will know of a bunch of fairly crazy commentators whose views we dismiss by default, but those commentators will all have fans who treat them as a system of record for opinions. (Via The Browser.)
• In Capital Gains this week, the story of credit default swaps as told through a bizarre case where a hotel burned down and then the owner bought insurance. CDS are both a way to basically sell bonds short and a way to separate the business of originating loans from that of taking credit risk. This turns out to be uniquely useful now, when CDS spreads are blowing out for a group of companies that might go bankrupt in a few years, but might also be borrowing an order of magnitude more by then.
• A Read.Haus reader asks: is it really such a big deal that AI is disrupting a few academic disciplines whose practitioners are all really smart? They'll be able to get other jobs—probably jobs that are, from a financial perspective at least, much better than what they had before. This raises an important point about the impact of AI: many of the biggest effects won't show up in statistical aggregates. If we look back, we might find that AI's impact on math PhDs is that it doubled their earnings, because they all went to work for AI labs and prop trading firms. But we won't be able to measure the existential despair they might have felt from first defining themselves in terms of their mathematical ability, and then finding out that there's a way to outdo that and it just involves spending a lot of money and periodically telling Claude "keep going, I know you can do it." This will show up everywhere: AI is trained on data produced by humans, so it's very anthropomorphic; it's better at writing wedding vows than op-eds about tariffs, because more of the world model required for the first task is embedded in the tokens. We just live at a very weird time.

Books

D: Heavy Water: In The Onion's parody history book, there's a fun bit in the 1930s about how people are distracting themselves from the Great Depression by reading light escapist literature by H.P. Lovecraft. As weird and sometimes disturbing as the present is, it's good to look back and remember that the 1930s were a lurid fever dream during which it felt like our whole civilization was coming apart—totalitarian states gearing up for war, global economic misery, and early inklings about fantastically powerful weapons.

Historical fiction is a fun genre because the real world has so much obscure weirdness. At one point in the book, there was a character who was so cartoonishly weird that I just had to know who, in the real world, had pulled off what their character did in the book. But that character turned out, improbably enough, to be a real historical figure (this means Wikipedia is a source of spoilers, and I now know that in the sequel, he's going to die will defusing a bomb).

The core of the story is the removal of Norway's stockpile of heavy water before the Nazis could seize it for their nuclear weapons program. As is common in Stephenson's novels, there are parallel bits-and-atoms challenges: on the informational side, we end up with various plot-driven reasons for characters to drop some exposition on the physics of atomic weapons. But there's also a very interesting section (particularly interesting in light of Covid and other gradually-then-suddenly disasters) of how long it took ordinary people in France, and even French leadership, to realize that they were losing the war, and fast. This is the kind of historical event where it's incredibly convenient to have fictional characters act as stand-ins for the reader, to narrate just how weird things are getting.

And this time period was, indeed, weird. There really was an anti-fascist alternative Olympics in 1936, and it really did get canceled in the middle of Pablo Casals conducting a last-minute rehearsal of Beethoven's 9th because of the outbreak of the Spanish Civil War. In a work of pure fiction, this is just too heavy-handed a metaphor for the sudden unraveling of civilization, but history repeatedly gets away with things no novelist could pull off.

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:

• Deerfield-backed Series A company building agents for end-to-end healthcare administration (prior authorization, eligibility checks, patient scheduling) is looking for a senior software/AI engineer to build backend services and stateful agents. Experience building and monitoring production-quality ML and AI systems preferred. (Remote)
• Well-funded, frontier AI neolab working on video pretraining and computer action models as the path to general intelligence is looking for researchers who are excited about creating machines that learn from experience, not text. Ideally you have zero-to-one pre-training experience and/or are a high-slope generalist who’s frustrated that the big labs aren't doing this. (SF)
• 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)
• 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. (Remote)
• Ex-Palantir founders building the meta-harness for all the lucrative agents need full-stack engineers that understand that turning AI into economically valuable solutions means a system that includes deterministic function calls and database queries. This is the company that will power CPU earnings for some time to come. (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.

And: we're now actively deploying capital into early-stage companies through Anomaly. Our focus is on defense, logistics, robotics, and energy. If you'd like to chat, please reach out.