- Samanth Subramanian has a lengthy profile of Michael Lewis in The Guardian. Lewis makes the useful point that writing a biography of a living person involves a tradeoff: if you get close enough to understand them well, you also spend a lot of time with them, and they naturally spend that time trying to persuade you that they're a good, upstanding person. This clearly happened with SBF, in that the new book has lots of revealing anecdotes about Bankman-Fried's character, the upshot of which seems to be that he's vocally committed to proper behavior in the abstract but incapable of treating any particular people especially well. The book is most enjoyable if you come to it with a strong view of what happened and want to mine Lewis' anecdotes for more information on why it happened.
- And another contribution to the annals of careful coverage of fraud: here's Gideon Lewis-Kraus in The New Yorker on Francesca Gino, Dan Ariely, Data Colada, and the scandal of fabricated research on the nature of dishonesty. As with the SBF saga, there will be few surprises on the fakery itself to readers who followed this story as it unfolded, but lots of details on the people behind it and how it happened. There are some good thoughts on the psychology of academics, and why it's so hard for them to call out dishonesty: one researcher who doubted some of Gino's work was told by her thesis advisor that "Academic research is like a conversation at a cocktail party. You are storming in, shouting 'You suck!'" That may be true, but if there's a norm against telling anyone they suck, the party in question will be full of people who do suck, and the party will, too. The piece also notes that the researchers who fabricated data were actually smart and hardworking, in addition to being dishonest. This is not an uncommon pattern: if you are, say, the 300th-best in the world at something, you might be able to cheat your way to #100. Very few people will care (though the new #101 will). But if you're #3 in the world, and you cheat, you can hit #1. That kind of cheating will often involve covering up bad performance rather than faking good performance, nudging the data closer to the p-value cutoff and the like. But once the habit is formed, it's hard to quit.
- This paper by Kalash Jain on the impact of industry classification on companies' behavior and stock price performance is fun: the goal of any classification schema is to group similar companies together, but some companies are dissimilar or have shifting businesses; sometimes a mail-order company becomes a streaming video company, or a starts operating outsourced datacenters. Industry groupings are a blunt instrument, but they enable a more fine-grained one: since analysts cover companies within an industry, and the questions they ask company A give an indication of what they've heard from companies B and C, the better the grouping, the more accurate that information transmission will be. And analysts will often cover industries based on the Global Industry Classification Standard, because GICS is what indexes use and indices are how analysts are benchmarked. This ends up being a fun look at the difficulty of creating any kind of taxonomy that everyone will be satisfied with, as well as the feedback loop between automated classification and human judgment.
- Eugene Wei has thoughts on what makes a good graph. This is really two stories in one: it's a piece about using graphic design to accurately convey information despite Excel's best efforts to make charts hideous. And it's also about how effective organizations use the information they're gathering. The trend today is to collect lots of data that can be applied through automated means to continuously deliver marginal improvements. But that's descended from an ethos of gathering data and regularly presenting it to human beings who will spot patterns, identify larger potential improvements, and ideally get a better mental model of whatever system they're collecting data on. TK Via The Interesting Times
- Is college getting relentlessly more expensive? The discourse and the sticker prices say yes, but Dan Currell writes in National Affairs that the answer is no: high prices are a form of price discrimination, and schools give massive "scholarships" (i.e. standard discounts) in order to get their prices more aligned with reality. One pathological feature of this system is that it creates a form of class discrimination, too: middle-class parents may have a sense that colleges offer discounts to students, and that they're more affordable than they look. Poorer families won't be as aware of this, and may discourage their kids from even pursuing higher education because of the sticker price. Which can still work out nicely! Even at a lower price, the opportunity cost of college is rising as alternatives improve. The piece is a good reminder that it's important to look at actual, rather than stated, preferences. At least in higher education, a sticker price is something colleges say, and the effective tuition they charge is what they do.
- In this week's Capital Gains, we look at why "Tech-Media-Telecom" became a standard grouping of companies, what those companies do and don't have in common, and how active hedge funds slowly got dominated by TMT groups while "TMT" described a growing share of the underlying economy.
- Going Infinite: The Rise and Fall of a New Tycoon: Michael Lewis' book is both more and less than meets the eye. For someone who had more or less unfettered access to the principal players in what turned out to be a fairly large fraud trial, Lewis uncovered less evidence than one would have expected both before and after FTX's problems came to light. On the other hand, as a fun book about what it was like to be in the eye of the crypto storm at a crucial period, it really is great. Lewis is great at characterizing people memorably, though for the central player in the book it's hard to say he's characterized him accurately. (Disclosure: The Diff is quoted briefly in the book.)
- Orality and Literacy: humans spoke long before we started writing, but writing has evolved into a very different medium from speech. This book is a collection of essays by Walter Ong exploring all of those differences, full of fascinating observations. Some of these are more on the theoretical side, like the idea that speech is dynamic. You can't "pause" speech on a single word; if you're paused, you're listening to silence rather than speech! But text is intrinsically paused, and, per Plato's complaint, you can't interrogate a written work and force it to correct or explain itself. The written word always gets the last word. And some of it's more historical: he notes that until quite recently, written works implicitly assumed an audience for a spoken work. Even when Chaucer wrote the Canterbury Tales, he had to set a frame in which each of them was a spoken work. This book is worthwhile now as online speech has evolved—some of it's clearly part of the literary tradition, and much of it is better understood as speech that happens to be transcribed.
Companies in the Diff network are actively looking for talent. A sampling of current open roles:
- A private credit fund denominated in Bitcoin needs a credit analyst that can negotiate derivatives pricing. Experience with low-risk crypto lending preferred (i.e. to large miners, prop-trading firms in safe jurisdictions). (Remote)
- The leading provider of advanced options analytics — “the ASML of options trading” — is growing rapidly, very profitable, and looking for an early-career generalist with prior options experience. (Connecticut, NYC)
- A company building ML-powered tools to accelerate developer productivity is looking for software engineers. (Washington DC area)
- A concentrated crossover fund is looking for an intellectually curious data scientist with demonstrated mastery in analytics. Experience with alt data, web scraping, and financial modeling preferred. (SF)
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.
- Drop in any links or comments of interest to Diff readers.
- What are some other good mental models of different media, like speech versus text, that were explained before the Internet reached its current form but are more applicable in an Extremely Online age?