Longreads + Open Thread

Longreads

• In Shakos Metaheuristics, a writeup of a very fun tool that's basically the Google Books n-gram explorer, but for concepts rather than specific works. It's a really fun tool. If you want to track the growth of some ideology, for example, you might try just counting the number of times it's mentioned. But often mentions peak after influence—neoconservatism as a political force got a lot of attention in the mid-2000s, but its big impact was a few years earlier, and you need a fuzzier match to detect that. LLMs speed up this process, so instead of identifying key thinkers and terminology, and avoiding the false-negative effect from different strains of the belief system dying off before you're aware of them, you have an LLM come up with a broad mix of indicators and track those. It's especially fun for pop culture, like nostalgic TV and movies taking off in the early 1970s as Baby Boomers realized they weren't kids anymore. This piece is especially fun because you can go back to your formative years and figure out which big intellectual developments were fairly unique to you, and which were just the result of what you were reading and watching at the time that your worldview gelled. It's been exciting to see LLMs applied to history—many of the records we have take the form of unstructured text, and programs that are especially good at analyzing such text will probably yield fruitful insights. (It would be neat to go back and answer some of the "who knew what when?"-style questions by using huge volumes of data to find circumstantial evidence.) Also, part of the way through the piece, Shako drops the incredible observation that a book is a higher-order function that you use as an input when you're interpreting reality through that book's lens. A very good way to think about it!
• The Dan Wang annual letter is back! There's a tradeoff in thinking about the way things work in different places, where it's helpful to be an insider to understand DC, Silicon Valley, New York, etc., but that also leads you to parochialism. You can also easily run into the opposite view-from-nowhere problem, where you construct a model of how that cluster works derived entirely from first principles. Wang is very good at giving the Beijing-from-SV perspective, the SV-from-the-rest-of-America perspective, etc.; he's one of the great cultural translators of our age.
• Andrew Odlyzko, who has written many excellent pieces about financial history, looks at J.W.M. Turner's career arbitraging UK government securities. This is a fun look at a polymath, and a reminder that markets used to be much less efficient. But it may be less of an arbitrage and more of a risk premium: Odlyzko notes that two different British perpetuities traded at prices that weren't internally consistent, and that one of them, consols, traded at a premium. But that might have actually been because of liquidity: there were more consols, and they get mentioned in literature from time to time (Dickens and Thackeray both use them as a general term for the kind of safe asset rich people would own), so they might have been the modern equivalent of on-the-run treasury bonds—more liquid, and thus worth a little more to a liquidity-sensitive trader. If that's what was going on, then Turner is an interesting financial figure in another sense: someone who took advantage of having a big balance sheet to do a trade that a less-capitalized trader couldn't risk exploiting at scale.
• Philip Trammell and Dwarkesh Patel on the inequality impact of AI, particularly the possibility that capital will become fungible with labor and the owners of most capital will compound their wealth to basically infinity. It's possible, but one of the problems with Piketty's work has always been that we don't know the full distribution of returns, and we also have better historical data on markets that didn't go to zero. It's entirely possible that the holders of capital are basically levered-long the overall economy and civilization, and will take the biggest hit when it falls apart. But, more importantly, we'll probably have a completely different economic model by then. A medieval economist's factor model would probably include people, livestock, and land, but the broader concept of capital wouldn't have made much sense to them. Anyone who's been part of a fast-growing organization (or a rapidly-growing family) knows that 2x growth is never uniformly 2x—it usually means that one thing is 1.1x as hard, something else is 0.9x as hard, something completely unexpected is 10x as hard, etc. There are lots of unpredictable bottlenecks in both the deployment of hardware to support current AI usage and in the deployment of AI models themselves inside organizations.
• Bartosz Ciechanowski has an excellent interactive tutorial on how cameras work. The piece is interspersed with interactive widget that show you what happens when you play with lens thickness, camera angle, exposure, etc. Experienced photographers will end up with very good intuitions about this, even if they don't understand the mechanics, but isolating every component is a great way to understand the whole system better. This post is also just a great format for connecting mathematical abstractions like the cosine-fourth-power law to the practical question of how to make a picture look good.
• On read.haus, a user shared a fun question I've wondered about: why do governments offer bonds with a definite term, rather than perpetuities? One reason is that it's a decent marketing gimmick: some people want ongoing income, but some of them want to save over a discrete time period and then get their money back. That deal is easier to understand than walking someone through the math of net present value, or explaining to them why their bond went down so much when rates went down just a little. A finite maturity would have been especially reasonable at times when governments operated on a gold standard and would potentially expect to temporarily borrow money during wars, and to repay all of it during peacetime. Today, governments could theoretically get by with issuing just two kinds of debt: a perpetuity paying, say, $1/year, and a floating-rate note that continuously pays the risk-free rate continuously. Between these two, you can construct whatever duration you like. But it would be hard to move in that direction: a US-issued perpetuity will start out less liquid than other treasury contracts, and those other contracts already give professionals the ability to buy and sell as much duration as they care to. Meanwhile, some institutions have mandates that specify the duration or maturity of the bond they buy. Given the network effect of liquidity, a small advantage for term debt compounded. Perpetuities would be a nice addition to the rates toolkit, though.

Books

American Zion: A New History of Mormonism: If you ask a member of a particular sect to draw a religious family tree showing who branched off from whom, they draw a straight line from themselves down to whoever founded the broader religion in question, and then label every other sect as a diverging branch. So, Catholics can talk about when the Orthodox church broke with Rome, and the Orthodox can talk about when the Patriarch of Rome decided to start his own thing. But this case is easier to make in some places than in others, and in the case of the Church of Jesus Christ of Latter-Day Saints, it's safe, from a historical standpoint, to regard the starting point of the faith as the day Joseph Smith reported that he'd uncovered a buried book of golden plates constituting the Book of Mormon. (As the author notes, calling them "Mormons" was originally a label adopted by outsiders in an often pejorative sense, later reclaimed by the church, later somewhat disclaimed by it, but also still in common use.)

With older faiths, a lack of records means that we have to speculate about what their early history was like, what sorts of internal debates over doctrine and organization occurred, etc. But in the case of the Mormons, there's plenty of contemporary coverage. That makes them fairly unique relative to the most-followed religions: there is no debate about the historicity of Joseph Smith, and not only do we have the Mormons' side of the story of how they were treated early on, but we have the other side of the story, too—the Mormons were a lively political issue for decades of US history. Early on, this was because of conflicts over how closely the Utah Territory government org chart corresponded to that of the LDS church—the US used to have a de facto independent neighboring country, with a radically different ideology, right next door, just like Taiwan and China today.

That status didn't last especially long, and Utah ended up converging on the governing and legal norms of the rest of the US. But not in every respect, or all at once: as it turns out, polygamy was a huge culture war issue in the last half of the 19th century, and created all sorts of bizarre situations. For example: Utah was the second territory to give women the right to vote, after Wyoming. But part of the political calculus was that this was basically gerrymandering: they expected women to vote for devout members of the church, who would then enforce polygamy (and it was a way to signal that, despite the polygamy, women in Utah had rights that women almost everywhere lacked). There was even a lobbying push, by an organization whose founders included at least one prominent suffragist, Julia Ward Howe, to have the federal government overrule Utah's granting women the vote. And this debate was going on for decades. Towards the end, in 1896, there was an election for state senators in which one Angus Cannon lost to Martha Hughes Cannon, one of his six wives (she was a strong advocate for women's suffrage, as well as polygamy.

The LDS church, like the CCP, had a long journey so arduous and definitive that they were ruled almost exclusively by people who'd participated in it until basically all of them were dead. That tends to make it hard for organizations to adapt as things change. So the overall story is one of punctuated equilibrium: there’s a norm, external norms drift, at some point there’s a crisis, and then there’s a resolution that typically involves some guarded assimilation. One thing that’s striking in this book is just how assimilated modern Mormons are—they’re less hyphenated-Americans and more hyper-American, and trends like Dry January and 

One part of Mormon history that's made its way into popular culture is the church's relative reluctance to get with the program on racial tolerance. What's especially odd about that is that in the earliest days of the church, they seem unusually tolerant. But in a way, this makes sense: in the early days of some group, it's all hands on deck; ten converts are better than nine, and if you can convert a couple and they both proselytize, you get a two-for-one deal. But later, when things stabilize, it becomes more adaptive to have some consistent norms. And you can end up with these norms falling behind shifts that happen elsewhere. (The same thing happens in language; American English is, in a limited sense, closer to Shakespeare than modern UK English—if you live on the periphery of people speaking a language, you tend to be conservative about changes in language lest you sound like an uneducated hick, but if you were living in London you'd quickly notice that English was suddenly non-rhotic.)

And yet, they did change; the church long had a nominal opposition to intoxicants, but only gradually ramped up enforcement of it until the temperance movement picked up steam. Prior to that, their attitude was a little looser: Brigham Young celebrated his selection as Joseph Smith's successor with what one attendee, Thomas Bullock, described as "Jerusalem Wine & delightful Strawberry wine," and used some salty language, too. But surprisingly late adoption of what seems like a strong cultural norm happens in other faiths, too; the rosary got popular during the Reformation. 

But all of this is partly a function of the fact that the early history of the LDS church is so well-documented, and by people who have much more modern sensibilities than the ones writing in Medina in the 7th century, Judea in the 0th, or Ur in the negative-twentieth. So a big share of the textual evidence we have about religious founders and early leaders comes from their fans. It's unique that we get a haters-and-all look at a faith from the very beginning, and that's probably a more realistic one—just one that’s missing from the historical record of plenty of older faiths! Every major religion started out high-variance enough to win converts and then settled down into a stable enough form to keep them around for generations. When the faith in question is new enough, we can see more of that process.

The story of the LDS in the US is one of deliberate and careful assimilation—adapting enough local norms to be compatible with the rest of the country, but retaining enough distinctions that there’s a meaningful group identity, too. And, for all his faults, Joseph Smith really did lead his followers to a promised land of SaaS, skiing, and great performance on pretty much any qualitative metric of human well-being you care to look at.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• Does anyone have some good examples of pages like Ciechanowski's camera tutorial, that walk through a complicated concept with lots of simulations. Kevin Simler's outbreak comes to mind—any others?

Diff Jobs

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

• A well-funded, Series C startup building the platform and agent primitives to drive operational transformation at large, complex institutions (starting with higher education) is hiring platform engineers. The work spans distributed systems, applied AI, and full-stack infrastructure, focused on deploying reliable agents that meaningfully bend institutional cost curves. (Remote)
• A software PE firm with an exceptional long-term track record is looking for consultants, ops executives, generalists, etc. with experience in procurement and vendor management. (Remote, US).
• A first-of-its-kind PE firm applying blockchain technology to transform legacy businesses is seeking an Investment Partner to lead deal execution and portfolio management. Ideal candidates have 12+ years of transaction and portfolio experience at leading tech-focused buyout funds (Silver Lake, Vista, Thoma Bravo, Francisco Partners, TA Associates, HG, Thrive Capital, or similar). No crypto experience required—they're looking for elite PE operators who are curious about frontier technology. (NYC)
• 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)

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.