The Diff - Longreads + Open Thread

Today's issue of The Diff is brought to you by our sponsor, Warp.

Longreads

• This 2008 paper by Slava Gerovitch is a look at the USSR's abortive effort to create a national network of computers in order to operate a more efficient planned economy. As someone who earns all of and spends much of his money using such a network, it's interesting to see what could have been, and why it mostly happened here and not over there. The use case for that Soviet network was a valid one, as evidenced by the fact that this is a big part of the modern Internet's social utility—the entire ad ecosystem is a decentralized economic planning tool that's constantly gathering incremental data on consumer preferences, and trying to shift those preferences in more profitable directions, while plenty of supply chain planning and management also assumes the existence of a network that receives data in real time and dynamically updates plans accordingly. So, it was a good thing to build, but it was the wrong place to start; the US Internet began with a completely different purpose, and only grudgingly adopted commercialization. So it's a double irony: the US got an Internet because we deferred to the intelligentsia and built a cool communications tool for universities, instead of focusing on bourgeois concerns like how many toasters to produce this month.
• Dan Williams on the inherent tensions of open societies. A culture of vigorous political debate contributing to a democratic process has an obvious free-rider problem: voting is a whole lot more fun if you're voting for the greatest person ever, in the most important election in history, in order to fight off the forces of evil. And every incremental piece of information you get makes that less of what you're doing. If one person mostly ignores politics and votes for fun, they free-ride on the politically-informed electorate, but if enough people do it, the entire system starts making bad decisions. (You, the reader, are welcome to pick the year in which we hit this tipping point—but not before you read the "Politically motivated cognition" section!) What's pleasantly surprising about all of this is that the system works so well; being politically informed is a norm, like honesty or tipping, and in many societies people are perfectly willing to conform to norms despite being constantly presented with enticing opportunities to defect.
• Ben Kuhn on the nuances of trust as a limiting factor for scaling teams. This isn't "trust" in the sense of thinking people are dishonest, but "trust" in the sense of thinking that, if you ask someone to do X or explain Y, you'll get roughly the outcome you had in mind. One solution he offers: "At Anthropic, Dario gives biweekly-ish “informal vision updates” (hour-long talks on important updates to parts of company strategy) that I think of as the canonical example of this. Just about everyone at Anthropic is trying to build an internal “Dario simulator” who they can consult when the real one is too busy (i.e. ~always)."
• In Fortune, Leo Schwartz covers the rise and fall of Jump Trading's crypto business. There's a lot here: Jump was already very active in equities, and that meant that there was a high opportunity cost to putting new traders on their existing markets and strategies. Crypto was lower-stakes for them, which meant that good crypto traders could rise faster in the organization. Crypto market-making economics were very different from other assets, though: in most markets, a market-maker is essentially working for other market participants, by competing to sell them liquidity at the best price. But in crypto, tokens would give options to market makers in order to incentivize them to create volume (and, of course, an incentive to increase the price). This creates a perverse incentive: liquidity is subsidized for new tokens, but once the option gets exercised and the market-maker's tokens get sold, they don't have an incentive to provide liquidity. So it helps explain, not just why crypto trading was so frenetic, but why that activity consisted of seemingly random tokens rising and falling. Getting rugged, one way or another, was built into the structure of the market.
• In Wired, Lauren Smiley profiles a prolific Uber/DoorDash scammer, whose scam consisted of a little bit of gaming their signup incentives and a lot of identity-borrowing and identity theft allowing people who weren't authorized to work in the US to work for these companies. This phenomenon gets talked about a lot, but it's naturally hard to track—no one in a position to know really wants to talk, so what you get instead is a dated snapshot of what worked and how well as of a few years ago. (The subject of the article ended up serving prison time.) It has an echo of The Real North Korea: illegal immigrants could work in the gig economy, but only by paying a commission to someone who could get them a valid ID and then get past authentication. One of the most striking things is just how little effort this used to require: the subject of the piece got pretty far with techniques like making up random social security numbers and photoshopping new faces into photos of IDs. She was running the operation for over a year before she switched to using a VPN. Growing companies are often very slow to patch this kind of thing, but one of the problems with taking advantage of lax security is that, once they do pay attention, there's quite a paper trail.
• In Capital Gains (our free spinoff newsletter covering topics in finance, economics, and strategy), we look at the purpose of the stock market if public companies return more capital than they raise. There are still plenty of positive externalities from being able to roughly quantify the performance of thousands of different businesses.
• And in this week's episode of The Riff, we talk about algorithmic collusion, the 4chan of crypto, and why I was wrong to think that 2024 would be the most boring presidential campaign in history. (Note that this was recorded on Monday, when Trump's odds were ~70%; they've since dropped to the low 60s.) Listen on Spotify/YouTube/Apple. And, as a bonus, I joined Patrick McKenzie/patio11 on his new podcast, where we talked about mortgages, writing the first draft of financial history, measuring skill, and much much more.

Books

There are some fun history books that cover a well-known era with a narrow focus—The Prize is basically a history of the world from ~1860-1990, told entirely through the lens of the production, transportation, refinement, and consumption of oil. Mark Kurlansky does this for a few other commodities, like cod and salt. History is easier to read when the protagonist is a human being rather than a product, so, in approximately that tradition, Sean McMeekin's Stalin's War retells the story of the Second World War, with a different main antagonist. The central thesis of the book is that the conflict was as much Stalin's as Hitler's idea, and, that some otherwise-confusing parts of the story make more sense in light of that. A major subplot here is either the US's bizarre deference to Stalin's wishes or Stalin's masterful ability to manipulate his allies.

The Second World War is easier to understand as a conflict between two coalitions that had some narrow interests in common but also had deep incompatibilities, but those coalitions were never comprehensive and gelled surprisingly late: the Soviet Union was allied to Germany right up until Germany invaded them, and actually conquered more of Poland by land area than Germany did in 1939. (The book also makes the surprising-to-me claim that after the German invasion in 1939, the Polish military was regrouping in the Eastern part of that country, preparing a longer defense—only to be attacked from the opposite direction by the Soviets!) Russia shared Germany's habit of invading or annexing neighbors, including an attack on Finland that prompted Britain and France to consider invading the USSR in 1940. Other details also complicate the two-coalition model, like the fact that after Pearl Harbor, the US did not declare war on Germany, and had a stroke of luck when Germany declared war first. Later in the war, the USSR violated a second neutrality agreement with a former ally when, in August 1945, it invaded Manchuria. There was, in other words, no point in the entirety of the Second World War in which all of the main victors were fighting all of the main losers.

In defense of the usual story, one reason for that was that, starting in June 1941, the war in Europe was almost exclusively a land war between the USSR and Germany. Stalin's War argues that Stalin was preparing for a preemptive invasion of Germany, knew that the Germans also planned to invade, and preferred to be attacked first in order to get sympathy from allies. He also knew that defense is easier than offense, and that Russia vastly outnumbered and even more vastly outgunned the Germans. (At the start of Operation Barbarossa, Russia had 15,000 tanks on the front Germany was attacking, compared to 3,300 for Germany; they had a similar number of aircraft, compared to 2,250 for Germany.) But the Soviet military had also sent a fair number of senior commanders to labor camps or executed them, and, at least at the outset, would be defending recently-conquered land rather than home territory. So the Germany invasion, despite being outnumbered, moved much faster than Russia expected, and emphasized capturing cities with significant military-related manufacturing, or critical raw materials. The result of that was that the one country most actively fighting the Axis couldn't produce guns, bullets, boots, food, tanks, planes, or fuel in the quantities necessary to do anything but gracefully retreat. Fortunately for them, the US could produce these products, and did so. Getting those to Russia was tricky, so one understandable reason for Stalin's neutrality with Japan was that it allowed the US to ship supplies over the Pacific. This explains the otherwise implausible fact that Russia won a war of attrition after losing a chunk of its industrial capacity to an invader.

The book does have some weak points. The author has a quite reasonable antipathy to Stalin, but sometimes attributes to malice what can be explained by either incompetence or weirdness. Was Stalin indifferent to allied loss of life in transporting equipment to Russia? Probably. Did he, as McMeekin argues, deliberately choose a supply route that led to maximum allied casualties? Probably not—those casualties also deprived him of military equipment! Similarly, Stalin's insistence on having conferences in out-of-the-way places under Soviet control gets cast as a mind game meant to intimidate the by-then-ailing FDR, but it can also be explained by Stalin being incredibly paranoid, and assuming that his allies would spy on him the way he did on them.

Overall, this book is a good reminder of two ideas that are always in tension. On the one hand, history is incredibly contingent, and there's a near-possible world where the "Second World War" describes a conflict between a US/British/French alliance and a German/Russian axis, with Japan as perhaps a neutral player that joins the fight against that axis at the last minute, just before the first atomic bomb is dropped on Vladivostok. The other idea is that as much as the alliances, interests, and conflicts are random outcomes, the actual outcome of twentieth-century conflicts was heavily determined by which country could produce the most equipment and secure the largest stocks of metals and fuel. So it's relatively straightforward to predict which side wins, but very hard to predict who ends up on which side.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• In preparation for a near-future post: are there some good case studies of industries that totally changed how they charge for what they offer, but still ended up having pretty similar economics? The enterprise software business got a lot better as it moved to subscriptions, and ad agencies had worse economics after they stopped charging a set cut of ad expenditures. Did anyone just mean-revert?

A Word From Our Sponsors

Still running payroll the old way?

Warp is built for founders and startups who have less time on their hands. Warp bundles payroll, compliance, and benefits into a single dashboard, automating annoying registrations that come with hiring across states or offshore.

Unlike bloated platforms pushing unnecessary features like applicant tracking systems and performance reviews for your employees, Warp is laser-focused on the essentials. Get in, get it done, get out.

Say goodbye to endless payroll reminders and cryptic notices from tax departments. Reclaim your time and focus on what truly matters – growing your startup.

Get started now at joinwarp.com/thediff.

Diff Jobs

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

• A company building high-performance simulation tools for smart contracts and other applications is looking for a DevOps/MLOps engineer with experience in reinforcement learning, and an interest in AI agents. (Remote, European Time)
• A well-funded company building AI-powered compliance tools is recruiting for sales engineers, who will assist with closing and onboarding new customers. Experience with AI, regulatory/compliance, or both is ideal. (NYC)
• A top prop trading firm is looking for an intellectually curious, mathy generalist to work on projects spanning business strategy, technology, and markets. (NYC)
• A well funded early stage startup founded by two SpaceX engineers is building the software stack for hardware companies, and is looking for a senior infrastructure engineer. (LA)
• A startup building tools to enable people build communities and own their own data is looking for a site reliability engineer with deep Kubernetes experience. (Remote, SF preferred)
• A startup building a new financial market within a multi-trillion dollar asset class is looking for generalists with insurance and analytics experience. (US, Remote)

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.