Category: The Arts
Podcast with Salvador Duarte
Salvador is 17, and is an EV winner from Portugal. Here is the transcript. Here is the list of discussed topics:
0:00 – We’re discovering talent quicker than ever 5:14 – Being in San Francisco is more important than ever 8:01 – There is such a thing like a winning organization 11:43 – Talent and conformity on startup and big businesses 19:17 – Giving money to poor people vs talented people 22:18 – EA is fragmenting 25:44 – Longtermism and existential risks 33:24 – Religious conformity is weaker than secular conformity 36:38 – GMU Econ professors religious beliefs 39:34 – The west would be better off with more religion 43:05 – What makes you a philosopher 45:25 – CEOs are becoming more generalists 49:06 – Traveling and eating 53:25 – Technology drives the growth of government? 56:08 – Blogging and writing 58:18 – Takes on @Aella_Girl, @slatestarcodex, @Noahpinion, @mattyglesias, , @tszzl, @razibkhan, @RichardHanania, @SamoBurja, @TheZvi and more 1:02:51 – The future of Portugal 1:06:27 – New aesthetics program with @patrickc.
Self-recommending, here is Salvador’s podcast and Substack more generally.
My excellent Conversation with Brendan Foody
Here is the audio, video, and transcript. Here is the episode summary:
At 22, Brendan Foody is both the youngest Conversations with Tyler guest ever and the youngest unicorn founder on record. His company Mercor hires the experts who train frontier AI models—from poets grading verse to economists building evaluation frameworks—and has become one of the fastest-growing startups in history.
Tyler and Brendan discuss why Mercor pays poets $150 an hour, why AI labs need rubrics more than raw text, whether we should enshrine the aesthetic standards of past eras rather than current ones, how quickly models are improving at economically valuable tasks, how long until AI can stump Cass Sunstein, the coming shift toward knowledge workers building RL environments instead of doing repetitive analysis, how to interview without falling for vibes, why nepotism might make a comeback as AI optimizes everyone’s cover letters, scaling the Thiel Fellowship 100,000X, what his 8th-grade donut empire taught him about driving out competition, the link between dyslexia and entrepreneurship, dining out and dating in San Francisco, Mercor’s next steps, and more.
And an excerpt:
COWEN: Now, I saw an ad online not too long ago from Mercor, and it said $150 an hour for a poet. Why would you pay a poet $150 an hour?
FOODY: That’s a phenomenal place to start. For background on what the company does — we hire all of the experts that teach the leading AI models. When one of the AI labs wants to teach their models how to be better at poetry, we’ll find some of the best poets in the world that can help to measure success via creating evals and examples of how the model should behave.
One of the reasons that we’re able to pay so well to attract the best talent is that when we have these phenomenal poets that teach the models how to do things once, they’re then able to apply those skills and that knowledge across billions of users, hence allowing us to pay $150 an hour for some of the best poets in the world.
COWEN: The poets grade the poetry of the models or they grade the writing? What is it they’re grading?
FOODY: It could be some combination depending on the project. An example might be similar to how a professor in English class would create a rubric to grade an essay or a poem that they might have for the students. We could have a poet that creates a rubric to grade how well is the model creating whatever poetry you would like, and a response that would be desirable to a given user.
COWEN: How do you know when you have a good poet, or a great poet?
FOODY: That’s so much of the challenge of it, especially with these very subjective domains in the liberal arts. So much of it is this question of taste, where you want some degree of consensus of different exceptional people believing that they’re each doing a good job, but you probably don’t want too much consensus because you also want to get all of these edge case scenarios of what are the models doing that might deviate a little bit from what the norm is.
COWEN: So, you want your poet graders to disagree with each other some amount.
FOODY: Some amount, exactly, but still a response that is conducive with what most users would want to see in their model responses.
COWEN: Are you ever tempted to ask the AI models, “How good are the poet graders?”
[laughter]
FOODY: We often are. We do a lot of this. It’s where we’ll have the humans create a rubric or some eval to measure success, and then have the models say their perspective. You actually can get a little bit of signal from that, especially if you have an expert — we have tens of thousands of people that are working on our platform at any given time. Oftentimes, there’ll be someone that is tired or not putting a lot of effort into their work, and the models are able to help us with catching that.
And:
COWEN: Let’s say it’s poetry. Let’s say you can get it for free, grab what you want from the known universe. What’s the data that’s going to make the models, working through your company, better at poetry?
FOODY: I think that it’s people that have phenomenal taste of what would users of the end products, users of these frontier models want to see. Someone that understands that when a prompt is given to the model, what is the type of response that people are going to be amazed with? How we define the characteristics of those responses is imperative.
Probably more than just poets that have spent a lot of time in school, we would want people that know how to write work that gets a lot of traction from readers, that gains broad popularity and interest, drives the impact, so to speak, in whatever dimension that we define it within poetry.
COWEN: But what’s the data you want concretely? Is it a tape of them sitting around a table, students come, bring their poems, the person says, “I like this one, here’s why, here’s why not.” Is it that tape or is it written reports? What’s the thing that would come in the mail when you get your wish?
FOODY: The best analog is a rubric. If you have some —
COWEN: A rubric for how to grade?
FOODY: A rubric for how to grade. If the poem evokes this idea that is inevitably going to come up in this prompt or is a characteristic of a really good response, we’ll reward the model a certain amount. If it says this thing, we’ll penalize the model. If it styles the response in this way, we’ll reward it. Those are the types of things, in many ways, very similar to the way that a professor might create a rubric to grade an essay or a poem.
Poetry is definitely a more difficult one because I feel like it’s very unbounded. With a lot of essays that you might grade from your students, it’s a relatively well-scoped prompt where you can probably create a rubric that’s easy to apply to all of them, versus I can only imagine in poetry classes how difficult it is to both create an accurate rubric as well as apply it. The people that are able to do that the best are certainly extremely valuable and exciting.
COWEN: To get all nerdy here, Immanuel Kant in his third critique, Critique of Judgment, said, in essence, taste is that which cannot be captured in a rubric. If the data you want is a rubric and taste is really important, maybe Kant was wrong, but how do I square that whole picture? Is it, by invoking taste, you’re being circular and wishing for a free lunch that comes from outside the model, in a sense?
FOODY: There are other kinds of data they could do if it can’t be captured in a rubric. Another kind is RLHF, where you could have the model generate two responses similar to what you might see in ChatGPT, and then have these people with a lot of taste choose which response they prefer, and do that many times until the model is able to understand their preferences. That could be one way of going about it as well.
Interesting throughout, and definitely recommended. Note the conversation was recorded in October (we have had a long queue), so a few parts of it sound slightly out of date. And here is Hollis Robbins on LLMs and poetry.
Stories Beyond Demographics
The representation theory of stories, where the protagonist must mirror my gender, race, or sexuality for me to find myself in the story, offers a cramped view of what fiction can do and a shallow account of how it actually works. Stories succeed not through mirroring but by revealing human patterns that cut across identity. Archetypes like Hero, Caregiver, Explorer, and Artist, and structures like Tragedy, Romance, and Quest are available to everyone. That is why a Japanese salaryman can love Star Wars despite never having been to space or met a Wookie and why an American teenager can recognize herself in a nineteenth-century Russian novel.
Tom Bogle makes this point well in a post on Facebook:
I have no issue with people wanting representation of historically marginalized people in stories. I understand that people want to “see themselves” in the story.
But it is more important to see the stories in ourselves than to see ourselves in the stories.
When we focus on the representation model, we recreate a character to be an outward representation of physical traits. Then the internal character traits of that individual become associated with the outward physical appearance of the character and we pigeonhole ourselves into thinking that we are supposed to relate only to the character that looks like us. Movies and TV shows have adopted the Homer Simpson model of the aloof, detached, and even imbecilic father, and I, as a middle-aged cis het white guy with seven kids could easily fall into the trap of thinking that is the only character to whom I can relate. It also forces us to change the stories and their underlying imagery in order to fit our own narrative preferences, which sort of undermines the purpose for retelling an old story in the first place.
The archetypal model, however, shifts our way of thinking. Instead of needing to adapt the story of Little Red-Cap (Red Riding Hood) to my own social and cultural norms so that I can see myself in the story, I am tasked with seeing the story play out in myself. How am I Riding Hood? How am I the Wolf? How does the grandmother figure appear in me from time to time? Who has been the Woodsman in my life? How have I been the Woodsman to myself or others? Even the themes of the story must be applied to my patterns of behavior or belief systems, not simply the characters. This model also enables us to retain the integrity of the versions of these stories that have withstood the test of time.
So if your goal is actually to affect real social change through stories, I would encourage you to consider how the archetypal approach may actually be more effective at accomplishing your aims than the representational approach alone (as they are not necessarily in conflict with one another).
What should I ask Henry Oliver?
Yes, I will be doing a Conversation with him. We will focus on our mutual readings of Shakespearer’s Measure for Measure, with Henry taking the lead. But I also will ask him about the value of literature, Jane Austen, Adam Smith, Bleak House, his book on late bloomers, and more.
Here is Henry’s (free) Substack. Here is Henry on Twitter.
So what should I ask him?
What is the greatest artwork of the century so far?
That question is taken from a recent Spectator poll. Their experts offer varied answers, so I thought at the near quarter-century mark I would put together my own list, relying mostly on a seat of the pants perspective rather than comprehensiveness. Here goes:
Cinema
Uncle Boonmee, In the Mood for Love, Ceylan’s Winter Sleep, Yi Yi, Artificial Intelligence, Her, Y Tu Mama Tambien, Four Months Three Weeks Two Days, from Iran A Separation, Oldboy, Silent Light (Reygadas), The Three Burials of Melquiades Estrada, Get Back, The Act of Killing, Master and Commander, Apocalypto, and New World would be a few of my picks. Incendies anyone?
Classical music (a bad term these days, but you know what I mean):
Georg Friedrich Haas, 11,000 Strings, Golijov’s Passion, John Adams Transmigration of Souls, The Dharma at Big Sur, Caroline Shaw, and Stockhausen’s Licht operas perhaps. Typically such works need to be seen live, as streaming is no substitute. As for recordings, recorded versions of almost every classic work are better than before, opera being excluded from that generalization. So the highest realizations of most classical music compositions have come in the last quarter century.
Fiction
Ferrante, the first two volumes of Knausgaard, Submission, Philip Pullman, and The Three-Body Problem. The Marquez memoir and his kidnapping book, both better than his magic realism. The Savage Detectives. Sonia and Sunny maybe?
Visual Arts
Bill Viola’s video art, Twombly’s Lepanto series, Cai Guo-Qiang and Chinese contemporary art more generally (noting it now seems to be in decline), the large Jennifer Bartlett installation that was in MOMA, Robert Gober. Late Hockney and Richter works. The best of Kara Walker. The second floor of MOMA and so much of what has been shown there.
Jazz
There is so much here, as perhaps the last twenty-five years have been a new peak for jazz, even as it fades in general popularity. One could mention Craig Taborn, Chris Potter, and Marcus Gilmore, but there are dozens of top tier creators. Cecile McLorin Salvant on the vocal side. Is she really worse than Ella Fitzgerald? I don’t think so.
Popular music (also a bad term)
The best of Wilco, Kanye, D’angelo, Frank Ocean, Bob Dylan’s Love and Theft. How about Sunn O)))? No slight intended to those listed, but I had been hoping this category would turn out a bit stronger?
Television
The Sopranos, the first two seasons of Battlestar Galactica, Srugim, Borgen, and Curb Your Enthusiasm.
Assorted
Hamilton, and there is plenty more in theater I have not seen. At the very least one can cite Stoppard’s Coast of Utopia and Leopoldstadt. There is games and gaming. People around the world, overall, look much better than ever before. The Museum of Islamic Art in Doha and the reoopened Great Egyptian Museum in Cairo. The new wing at MOMA. Architecture might need a post of its own, but I’ll start by citing the works of Peter Zumthor. (Here is one broader list, it strikes me as too derivative in style, in any case it is hard to get around and see all these creations, same problem as with judging theatre.) I do not follow poetry much, but Louise Glück and Seamus Heaney are two picks, both with many works in the new century. The top LLMs, starting (but not ending) with GPT-4. They are indeed things of beauty.
Overall, this list seems pretty amazing to me. We are hardly a culture in decline.
J. M. W. Turner, financial arbitrageur
Abstract. J. M. W. Turner is famous for his achievements in graphic arts. What is not known is that he engaged in some pioneering market arbitrage, a profitable and risk-free swapping of British government securities. His activities lead to interesting insights into British markets of the 19th century. Financial innovation frequently created profitable arbitrage opportunities. However, among regular investors it seems that it was mostly mavericks like Turner who took advantage of them. Apparently there were strong cultural factors that inhibited most people from imitating him, which allowed obvious pricing anomalies to persist for extended periods.
That is from a recent paper by Andrew Odlyzko. Via Colin.
A Call for New Aesthetics
We, Patrick and Tyler, have differing views of the artistic merits of Bauhaus, but we are both very impressed by the movement’s success: they sought to define an aesthetic for the twentieth century, and basically did. Bauhaus obviously sits as part of broader tides—functionalism, constructivism, De Stijl, etc.—but the project also shows how intentional artistic ambition can succeed. Everything from modern offices to modern tech hardware is in some sense downstream of Bauhaus.
We’re more than a quarter way through the new century and we can now ask: what is the aesthetic of the twenty-first century? Which are the important secessionist movements of today? Which will be the most important great works?
Today, futuristic aesthetics often mean retrofuturistic aesthetics. So, what should the future actually look like? There will not be a singular answer to that, but we are very interested in attempts to answer the questions.
In particular, we would like to fund some artists who are thinking about this.
Tyler: Circa 2026, beauty can be found in strange and unusual places. It does not always announce itself as such. It can violate our expectations in unreasonable ways. It often springs from an understanding that the world has changed fundamentally, and matters of the aesthetic need to recognize and respond to that. What can you do that that will surprise and inspire me?
Patrick: In 1925, Ortega y Gasset said “modern art, on the other hand, has the masses against it, and this will always be so since it is unpopular in essence; even more, it is antipopular.” Sagmeister and Walsh argue that we’ve stopped trying to produce beautiful work, and Nicholas Boys-Smith shows empirically that modern buildings are substantially less favored than designs that respect the specific character of the place. So, what are new directions forward? What is new and also beautiful?
Grant. We are seeking to fund artists, architects, and designers who are consciously working to define New Aesthetics.
- Our primary interest is in visual arts and architecture, but open to all mediums.
- AI presumably opens interesting new opportunities, though we haven’t seen much great work that only uses AI. (If jazz didn’t exist, could you prompt Suno to create it? This seems like an open problem.) As such, we’re neutral on the use of AI.
- We will not fund work that is already ubiquitous today.
Mechanics
- Grant size: $5k – $250k.
- Deadline: Applications are open until March 31, 2026.
- Selection: Reviewed on a rolling basis. Applying sooner is probably better than later, because we might at some stage decide that we’ve funded enough.
- Application: Include some examples of your work along with a rough description of what you’d use the grant to do. We don’t know how many applications we’ll receive, so please make your work as easy to assess as possible. Whether or not English is your first language, we much prefer applications written by humans over those written by AIs.
- Declines: To keep logistics simple, we will not inform declined proposals of our decision.
To apply, just email [email protected].
TC again: Here is the website.
Art as Data in Political History
From Valentine Figuroa of MIT:
Ongoing advances in machine learning are expanding opportunities to analyze large-scale visual data. In historical political economy, paintings from museums and private collections represent an untapped source of information. Before computational methods can be applied, however, it is essential to establish a framework for assessing what information paintings encode and under what assumptions it can be interpreted. This article develops such a framework, drawing on the enduring concerns of the traditional humanities. I describe three applications using a database of 25,000 European paintings from 1000CE to the First World War. Each application targets a distinct type of information conveyed in paintings (depicted content, communicative intent, and incidental information) and a cultural transformation of the early-modern period. The first revisits the notion of a European “civilizing process”—the internalization of stricter norms of behavior that occurred in tandem with the growth of state power—by examining whether paintings of meals show increasingly complex etiquette. The second analyzes portraits to study how political elites shaped their public image, highlighting a long-term shift from chivalric to more rational-bureaucratic representations of men. The third documents a long-term process of secularization, measured by the share of religious paintings, which began prior to the Reformation and accelerated afterward.
Here is the link, via the excellent Kevin Lewis.
My Conversation with the excellent Gaurav Kapadia
Here is the audio, video, and transcript. Here is the episode summary:
Gaurav Kapadia has deliberately avoided publicity throughout his career in investing, which makes this conversation a rare window into how he thinks. He now runs XN, a firm built around concentrated bets on a small number of companies with long holding periods. However, his education in judgment began much earlier, in a two-family house in Flushing that his parents converted into a four-family house. It was there where a young Gaurav served as de facto landlord, collecting rent and negotiating late payments at age 10. That grounding now expresses itself across an unusual range of domains: Tyler invited him on the show not just as an investor, but as someone with a rare ability to judge quality in cities, talent, art, and more with equal fluency.
Tyler and Gaurav discuss how Queens has thrived without new infrastructure, what he’d change as “dictator” of Flushing, whether Robert Moses should rise or fall in status, who’s the most underrated NYC mayor, what’s needed to attract better mayoral candidates, the weirdest place in NYC, why he initially turned down opportunities in investment banking for consulting, bonding with Rishi Sunak over railroads, XN’s investment philosophy, maintaining founder energy in investment firms and how he hires to prevent complacency, AI’s impact on investing, the differences between New York and London finance, the most common fundraising mistake art museums make, why he collects only American artists within 20 years of his own age, what makes Kara Walker and Rashid Johnson and Salman Toor special, whether buying art makes you a better investor, his new magazine Totei celebrating craft and craftsmanship, and much more.
Excerpt:
COWEN: Now, I don’t intend this as commentary on any particular individual, but what is it that could be done to attract a higher quality of candidate for being mayor of New York? It’s a super important job. It’s one of the world’s greatest cities, arguably the greatest. Why isn’t there more talent running after it?
KAPADIA: It is something that I’ve thought about a great deal. I think there’s a bunch of little things that accumulate, but the main thing that happens in New York City is, people automatically assume they can’t win because it’s such a big and great city. Actually, the last few presidential elections and also the current mayoral election have taught people that anyone could win. I think that, in and of itself, is going to draw more candidates as we go forward.
What happened as an example, this time, people just assumed that one candidate had the race locked up, so a lot of good candidates, even that I know, decided not even to run. It turns out that that ended up not being the case at all. Now that people put that into their mental model, the new Bayesian analysis of that would be, “Oh, more people should run.”
The second thing: New York has a bunch of very peculiar dynamics. It’s an off-year election, and the primaries are at very awkward times. I believe there’s a history of why the primary shifted to basically the third week of June, in which there’s a very low turnout. The third week of June in New York City, when the private schools are out and an off-year election. You’re able to win the Democratic nomination and therefore the mayoral election with tens of thousands of votes in a city this big. That is absolutely insane.
A couple of things that I would probably do would be to make the primary more normal, change the election timing to make it on-cycle, even number of years. You’d have to figure out how to do that. Potentially have an open primary as well.
COWEN: If we apply the Gaurav Kapadia judgment algorithm to mayoral candidates, what’s the non-obvious quality you’re looking for?
KAPADIA: Optimism.
COWEN: Optimism.
KAPADIA: Optimism.
COWEN: Is it scarce?
KAPADIA: Extraordinarily scarce. I think there’s much more doomerism everywhere than optimism. At the end of the day, people are attracted to optimism. If you think about the machinery of the city and the state, having a clear plan — of course, you need all the basics. You need to be able to govern. It’s a very complicated city. There’re many constituents.
But I think beyond that, you have to have the ability to inspire. For some reason, almost all of the candidates, over the last couple of cycles, have really not had that — with the exception of probably one — the ability to inspire. I think that is the most underrated quality that one will need.
COWEN: I have my own answer to this question, but I’m curious to see what you say. What is, for you, the weirdest part of New York City that you know of that doesn’t really feel like it belongs to New York City at all?
Definitely recommended.
Ford Madox Ford on Joseph Conrad
And, above all things else, as the writer has somewhere pointed out, Conrad was a politician. He loved the contemplation of humanity pulling away at the tangled skeins of parties or of alliances. Until, suddenly a strand gave, a position cleared up, a ministry was solidly formed, a dynasty emerged. He was, that is to say a student of politics, without prescription, without dogma, and, as a Papist, with a profound disbelief in the perfectibility of human institutions. The writer never saw Conrad read any book of memoirs except those of Maxime Ducamp and the Correspondence of Flaubert; those we read daily together over a space of years. But somewhere in the past Conrad had read every imaginable and unimaginable volume of politicians’ memoirs, Mme de Campan, the Duc d’Audiffret Pasquier, Benjamin Constant, Karoline Bauer, Sir Horace Rumbold, Napoleon the Great, Napoleon III, Benjamin Franklin, Assheton Smith, Pitt, Chatham, Palmerston, Parnell, the late Queen Victoria, Dilke, Morley…. There was no memoir of all these that he had missed or forgotten—down to Il Principe or the letters of Thomas Cromwell. He could suddenly produce an incident from the life of Lord Shaftesbury and work it into Nostromo: which was the political history of an imagined South American Republic. That was one of the secrets of his greatness.
But certainly he had no prescription. Revolutions were to him always anathema since, he was accustomed to declare, all revolutions always have been, always must be, nothing more in the end than palace intrigues: intrigues either for power within, or for the occupancy of, a palace. The journalists’ bar in the palace of the Luxemburg where sits the present Senate of the Third Republic was once the bedchamber of Marie de Medicis. That is not to say that Conrad actively desired the restoration of the Bourbons: he would have preferred the journalists to remain where[Pg 60] they were rather than have any revolution at all. All revolutions are an interruption of the processes of thought and of the discovery of a New Form … for the novel.
The short book, online and free, is interesting throughout. Ford knew Conrad well, and appreciated him at a deep level.
Colors of growth
This looks pretty tremendous:
We develop a novel approach to measuring long-run economic growth by exploiting systematic variation in the use of color in European paintings. Drawing inspiration from the literature on nighttime lights as a proxy for income, we extract hue, saturation, and brightness from millions of pixels to construct annual indices for Great Britain, Holland, France, Italy, and Germany between 1600 and 1820. These indices track broad trends in existing GDP reconstructions while revealing higher frequency fluctuations – such as those associated with wars, political instability, and climatic shocks – that traditional series smooth over. Our findings demonstrate that light, decomposed into color and brightness components, provides a credible and independent source of information on early modern economic activity.
That is new research by Lars Boerner, Tim Reinicke, Samad Sarferaz, and Battista Severgnini. Via Ethan Mollick.
My Conversation with the excellent Dan Wang
Here is the audio, video, and transcript. Here is part of the episode summary:
Tyler and Dan debate whether American infrastructure is actually broken or just differently optimized, why health care spending should reach 35% of GDP, how lawyerly influences shaped East Asian development differently than China, China’s lack of a liberal tradition and why it won’t democratize like South Korea or Taiwan did, its economic dysfunction despite its manufacturing superstars, Chinese pragmatism and bureaucratic incentives, a 10-day itinerary for Yunnan, James C. Scott’s work on Zomia, whether Beijing or Shanghai is the better city, Liu Cixin and why volume one of The Three-Body Problem is the best, why contemporary Chinese music and film have declined under Xi, Chinese marriage markets and what it’s like to be elderly in China, the Dan Wang production function, why Stendhal is his favorite novelist and Rossini’s Comte Ory moves him, what Dan wants to learn next, whether LLMs will make Tyler’s hyper-specific podcast questions obsolete, what flavor of drama their conversation turned out to be, and more.
Excerpt:
COWEN: When will Chinese suburbs be really attractive?
WANG: What are Chinese suburbs? You use this term, Tyler, and I’m not sure what exactly they mean.
COWEN: You have a yard and a dog and a car, right?
WANG: Yes.
COWEN: You control your school district with the other parents. That’s a suburb.
WANG: How about never? I’m not expecting that China will have American-style suburbs anytime soon, in part because of the social engineering projects that are pretty extensive in China. I think there is a sense in which Chinese cities are not especially dense. Indian cities are much, much more dense. I think that Chinese cities, the streets are not necessarily terribly full of people all the time. They just sprawl quite extensively.
They sprawl in ways that I think the edges of the city still look somewhat like the center of the city, which there’s too many high-rises. There’s probably fewer parks. There’s probably fewer restaurants. Almost nobody has a yard and a dog in their home. That’s in part because the Communist Party has organized most people to live in apartment compounds in which it is much easier to control them.
We saw this really extensively in the pandemic, in which people were unable to leave their Shanghai apartment compounds for anything other than getting their noses and mouths swabbed. I write a little bit about how, if you take the rail outside of major cities like Beijing and Shanghai, you hit farmland really, really quickly. That is in part because the Communist Party assesses governors as well as mayors on their degree of food self-sufficiency.
Cities like Shanghai and Beijing have to produce a lot of their own crops, both grains as well as vegetables, as well as fruits, as well as livestock, within a certain radius so that in case there’s ever a major devastating war, they don’t have to rely on strawberries from Mexico or strawberries from Cambodia, or Thailand. There’s a lot of farmland allocated outside of major cities. I think that will prevent suburban sprawl. You can’t control people if they all have a yard as well as a dog. I think the Communist Party will not allow it.
COWEN: Whether the variable of engineers matters, I went and I looked at the history of other East Asian economies, which have done very well in manufacturing, built out generally excellent infrastructure. None of these problems with the Second Avenue line in New York. Taiwan, like the presidents, at least if we believe GPT-5, three of them were lawyers and none of them were engineers. South Korea, you have actually some economists, a lot of bureaucrats.
WANG: Wow. Imagine that. Economists in charge, Tyler.
COWEN: I wouldn’t think it could work. A few lawyers, one engineer. Singapore, Lee Kuan Yew, he’s a lawyer. He thinks in a very lawyerly manner. Singapore has arguably done the best of all those countries. Much richer than China, inspired China. Why should I think engineers rather than just East Asia, and a bunch of other accompanying facts about these places are what matter?
WANG: Japan, a lot of lawyers in the top leadership. What exactly was the leadership of Hong Kong? A bunch of British civil servants.
COWEN: Some of whom are probably lawyers or legal-type minds, right? Not in general engineers.
WANG: PPE grads. I think that we can understand the engineering variable mostly because of how much more China has done relative to Japan and South Korea and Taiwan.
COWEN: It’s much, much poorer. Per capita manufacturing output is gone much better in these other countries.
And:
WANG: Tyler, what does it say about us that you and I have generally a lot of similar interests in terms of, let’s call it books, music, all sorts of things, but when it comes to particular categories of things, we oppose each other diametrically. I much prefer Anna Karenina to War and Peace. I prefer Buddenbrooks to Magic Mountain. Here again, you oppose me. What’s the deal?
COWEN: I don’t think the differences are that big. For instance, if we ask ourselves, what’s the relative ranking of Chengdu plus Chongqing compared to the rest of the world? We’re 98.5% in agreement compared to almost anyone else. When you get to the micro level, the so-called narcissism of petty differences, obviously, you’re born in China. I grew up in New Jersey. It’s going to shape our perspectives.
Anything in China, you have been there in a much more full-time way, and you speak and read Chinese, and none of that applies to me. I’m popping in and out as a tourist. Then, I think the differences make much more sense. It’s possible I would prefer to live in Shanghai for essentially the reasons you mentioned. If I’m somewhere for a week, I’m definitely going to pick Beijing. I’ll go around to the galleries. The things that are terrible about the city just don’t bother me that much, because I know I’ll be gone.
WANG: 98.5% agreement. I’ll take that, Tyler. It’s you and me against the rest of the world, but then we’ll save our best disagreements for each other.
COWEN: Let’s see if you can pass an intellectual Turing test. Why is it that I think Yunnan is the single best place in the world to visit? Just flat out the best if you had to pick one region. Not why you think it is, but why I think it is.
Strongly recommended, Dan and I had so much fun we kept going for about an hour and forty minutes. And of course you should buy and read Dan’s bestselling book Breakneck: China’s Quest to Engineer the Future.
*God’s Grandeur*
— Gerard Manley Hopkins
Do (human) readers prefer AI writers?
It seems so, do read through the whole abstract:
The use of copyrighted books for training AI models has led to numerous lawsuits from authors concerned about AI’s ability to generate derivative content. Yet it’s unclear whether these models can generate high quality literary text while emulating authors’ styles/voices. To answer this we conducted a preregistered study comparing MFA-trained expert writers with three frontier AI models: ChatGPT, Claude, and Gemini in writing up to 450 word excerpts emulating 50 awardwinning authors’ (including Nobel laureates, Booker Prize winners, and young emerging National Book Award finalists) diverse styles. In blind pairwise evaluations by 159 representative expert (MFA-trained writers from top U.S. writing programs) and lay readers (recruited via Prolific), AI-generated text from in-context prompting was strongly disfavored by experts for both stylistic fidelity (odds ratio [OR]=0.16, p < 10^-8) and writing quality (OR=0.13, p< 10^-7) but showed mixed results with lay readers. However, fine-tuning ChatGPT on individual author’s complete works completely reversed these findings: experts now favored AI-generated text for stylistic fidelity (OR=8.16, p < 10^-13) and writing quality (OR=1.87, p=0.010), with lay readers showing similar shifts. These effects are robust under cluster-robust inference and generalize across authors and styles in author-level heterogeneity analyses. The fine-tuned outputs were rarely flagged as AI-generated (3% rate versus 97% for incontext prompting) by state-of-the-art AI detectors. Mediation analysis reveals this reversal occurs because fine-tuning eliminates detectable AI stylistic quirks (e.g., cliché density) that penalize incontext outputs, altering the relationship between AI detectability and reader preference. While we do not account for additional costs of human effort required to transform raw AI output into cohesive, publishable novel length prose, the median fine-tuning and inference cost of $81 per author represents a dramatic 99.7% reduction compared to typical professional writer compensation. Author-specific fine-tuning thus enables non-verbatim AI writing that readers prefer to expert human writing, thereby providing empirical evidence directly relevant to copyright’s fourth fair-use factor, the “effect upon the potential market or value” of the source works.
That is from a new paper by Tuhin Chakrabarty, Jane C. Ginsburg, and Paramveer Dhillon. For the pointer I thank the excellent Kevin Lewis. I recall an earlier piece showing that LLMs also prefer LLM outputs?
My Conversation with the excellent Jonny Steinberg
Here is the audio, video, and transcript. Here is the episode summary:
Tyler considers Winnie and Nelson: Portrait of a Marriage one of the best books of the last decade, and its author Jonny Steinberg one of the most underrated writers and thinkers—in North America, at least. Steinberg’s particular genius lies in getting uncomfortably close to difficult truths through immersive research—spending 350 hours in police ride-alongs, years studying prison gangs and their century-old oral histories, following a Somali refugee’s journey across East Africa—and then rendering what he finds with a novelist’s emotional insight.
Tyler and Jonny discuss why South African police only feel comfortable responding to domestic violence calls, how to fix policing, the ghettoization of crime, how prison gangs regulate behavior through century-old rituals, how apartheid led to mass incarceration and how it manifested in prisons, why Nelson Mandela never really knew his wife Winnie and the many masks they each wore, what went wrong with the ANC, why the judiciary maintained its independence but not its quality, whether Tyler should buy land in Durban, the art scene in Johannesburg, how COVID gave statism a new lease on life, why the best South African novels may still be ahead, his forthcoming biography of Cecil Rhodes, why English families weren’t foolish to move to Rhodesia in the 1920s, where to take an ideal two-week trip around South Africa, and more.
Excerpt:
COWEN: My favorite book of yours again is Winnie and Nelson, which has won a number of awards. A few questions about that. So, they’re this very charismatic couple. Obviously, they become world-historical famous. For how long were they even together as a pair?
STEINBERG: Very, very briefly. They met in early 1957. They married in ’58. By 1960, Mandela was no longer living at home. He was underground. He was on the run. By 1962, he was in prison. So, they were really only living together under the same roof for two years.
COWEN: And how well do you feel they knew each other?
STEINBERG: Well, that’s an interesting question because Nelson Mandela was very, very in love with his wife, very besotted with his wife. He was 38, she was 20 when they met. She was beautiful. He was a notorious philanderer. He was married with three children when they met. He really was besotted with her. I don’t think that he ever truly came to know her. And when he was in prison, you can see it in his letters. It’s quite remarkable to watch. She more and more becomes the center of meaning in his life, his sense of foundation, his sense of self as everything else is falling away.
And he begins to love her more and more, and even to coronate her more and more so that she doesn’t forget him. His letters grow more romantic, more intense, more emotional. But the person he’s so deeply in love with is really a fiction. She’s living a life on the outside. And you see this very troubling line between fantasy and reality. A man becoming deeply, deeply involved with a woman who is more and more a figment of his imagination.
COWEN: Do you think you learned anything about marriage more generally from writing this book?
STEINBERG: [laughs] One of the sets of documents that I came across in writing the book were the transcripts of their meetings in the last 10 years of his imprisonment. The authorities bugged all of his meetings. They knew they were being bugged, but nonetheless, they were very, very candid with each other. And you very unusually see a marriage in real time and what people are saying to each other. And when I read those lines, 10 different marriages that I know passed through my head: the bickering, the lying, the nasty things that people do to one another, the cruelties. It all seemed very familiar.
COWEN: How is it you think she managed his career from a distance, so to speak?
STEINBERG: Well, she was a really interesting woman. She arrived in Johannesburg, 20 years old in the 1950s, where there was no reason to expect a woman to want a place in public life, particularly not in the prime of public life. And she was absolutely convinced that there was no position she should not occupy because she was a woman. She wanted a place in politics; she wanted to exercise power. But she understood intuitively that in that time and place, the way to do that was through a man. And she went after the most powerful rising political activists available.
I don’t think it was quite as cynical as that. She loved him, but she absolutely wanted to exercise power, and that was a way to do it. Once she became Mrs. Mandela, I think she had an enormously aristocratic sense of politics and of entitlement and legitimacy. She understood herself to be South Africa’s leader by virtue of being married to him, and understood his and her reputations as her projects to endeavor to keep going. And she did so brilliantly. She was unbelievably savvy. She understood the power of image like nobody else did, and at times saved them both from oblivion.
COWEN: This is maybe a delicate question, but from a number of things I read, including your book, I get the impression that Winnie’s just flat out a bad person…
Interesting throughout, this is one of my favorite CWT episodes, noting it does have a South Africa focus.