What am I up to now?
May, 2026
Contents
Updates
I have no idea where this month went. Some things happened: I got my California Boater Card, I saw two Giants games, I started PT for my knee and my shoulder, and I went on a few nice walks. But our Passover Seder on April 1st feels very recent.
I did accept a new job. After my visit at UC Berkeley, I’ll be spending the rest of the year as a PhD Researcher at Google DeepMind. I’ll be working on big questions around AI’s social and economic impacts, inspired by my existing work on these topics in poor countries.
I’ll be living in London for the role at DeepMind for all of June and July. I’m still looking for housing for those two months, and please reach out with suggestions! The rest of May will be spent focused on getting my non-AI projects in good shape and some fun travel. In early May, I’m spending a week camping in Del Norte and Humboldt counties to see some redwoods. I’ll be visiting DC in a few weeks as well.
My wonderful girlfriend and I will be hosting a few parties in San Francisco in May as well. If you’re in town, please say hi!
Reading
To prepare for my visit at DeepMind, I’ve been reading along two lines. The first is on my prospective research agenda, and the second is about the history and culture of Google and DeepMind. I’ll be writing more about my research agenda in the coming months. In the meanwhile, I read the new biography of Demis Hassabis, the founder of DeepMind and current CEO of Google DeepMind.
The Infinity Machine, Sebastien Mallaby
Hassabis is well-worth a biography. He has an unusually procedural and chapter-based life, from chess prodigy, to video game designer, to computer scientist, to neuroscientist, to entrepreneur, to tech titan. He presents his early ambition of scientific enlightenment is well-attested to, unlike other AI-CEOs who pivoted to the field after success elsewhere. Depending how the next few years go, either this book and the Nobel prize are the end of his obituary, or this is just the first draft.
The first chapters annoyed me – Mallaby is a competent journalist and writer, but is entirely taken by his subject and makes some basic mistakes. On the first page, Hassabis is telling Mallaby about the book which most influenced him: Ender’s Game.
Hassabis was around thirty years old when he discovered Ender, and he was so taken with the story that he asked his wife to read it. She told him she felt sorry for the central character–a boy deprived of childhood and harnessed to a mission chosen for him by adults. But Hassabis identified powerfully with Ender.
He also talks about the influence of Ian Banks’ Culture series.Which I wrote about last month. Mallaby’s summaries of both of these books are wrong.Here’s the summary of Ender’s Game: ‘There, at an intergalactic battle school, Ender is manipulated by adults, bullied by classmates, and put through extreme mental testing, all to discover whether he can shoulder responsibility for the survival of the human race. By dint of grit and talent, Ender rises to the challenge. At the climax of the novel, he outwits an army of alien invaders, destroying their armada and saving planet Earth, though the question of whether he committed genocide in the process hangs over the outcome.’ In order: there’s nothing intergalactic about the school or plot, the aliens are from very nearby, the army he fights is not an invading army, and the question of genocide is very clearly decided in the book. And about the Culture, he writes: ‘AI systems would generate economic abundance, and citizens would lack for nothing. Space travel would be as simple as hopping on a London bus, and people could choose among hundreds of planets to live on. What’s more, the intelligent machines that Banks envisaged would exist peacefully alongside humans; they would be too preoccupied with their own intrigues to pick a fight with mortals.’ The most-recommended books in the Culture series are Player of Games and Use of Weapons. Both of these have, as important plot points, a strong discouragement of humans living on planets (rather, large space stations), and the intelligent machines in these books do nothing but pick fights with mortals. This footnote is very pedantic. But both of these examples come from the first ten pages of the book! Your neighborhood LLM would have done better. The classic example of Gell-Mann Amnesia is this: if you find a mistake in the box scores, you’ll trust the sports section a bit less. But by the time you’re reading the business section, you’ve forgotten about the error, or at least you read as if the rest of the newspaper were somehow more accurate. In this case, my area of expertise isn’t the sports section, it’s 1980s sci-fi, and inaccurately describing the plots of these books set off my Gell-Mann radar. Mallaby’s writing is engaging, and he’s very good at getting interviewees to say exciting and controversial things. But the underlying research is often lacking.
His access to Demis and Mustafa Suleyman, another of DeepMind’s founders, are the gems of the book. The third founder, Shane Legg, provides some interviews but is not as featured. This is too bad, because Legg is the most interesting of the three founders, and remains involved with social sciences research at DeepMind. I’d be excited to read more of his work.
These complaints are minor. This book is the best public telling of how Google acquired DeepMind in 2014, how Google was largely absent in language models from 2017-2020, and the merger of Google Brain and DeepMind in 2023. There’s good characterization of the leading researchers at DeepMind over the years, and a narrative for why Google’s AI efforts have lagged behind the other big companies. It’s a useful book for understanding the company’s culture, but is generally more focused on Demis’ genius.
Tau Ceti
Early in April, I saw Project Hail Mary, great movie. When I re-read Ursula LeGuin’s The Dispossessed a few weeks later, I was very surprised to find that it, like Project Hail Mary, was set in the Tau Ceti solar system.
It turns out that Tau Ceti is a very common setting for sci-fi writers. Besides Weir and LeGuin, Asimov, Kim Stanley Robinson, Dan Simmons, and Arthur C. Clarke all set books around that star.Of those, I’ve read Project Hail Mary, The Dispossessed, Rama Revealed, Hyperion, and one of the Aurora books. It’s very close and very similar to our sun, and it’s got a kickass name.
The Dispossessed is my favorite of LeGuin’s books, and one of the best books on the psychology of social and economic extremes. It’s also the only book I’ve read which makes interwoven narrative streams work.Wikipedia has a useful chronological list of chapters: 2, 4, 6, 8, 10, 12, 1, 3, 5, 7, 9, 11, 13. The book is about a physicist, Shevek, who travels from his anarchocommunist planet to the planet next door, which has a normal mix of countries and political systems.
Much like how “The Ones Who Walk Away from Omelas” is not about torturing a child, The Dispossessed is not really an “ambiguous utopia”. There’s really nothing utopic about the anarchocommunist moon: there’s a drought and a famine; there are bandits raiding food trains and killing the conductors. There’s even prison!Although they call it an asylum, and mention repeatedly that people only go there if they request to. But it’s definitely a prison. It’s on an island! The only island on the planet. It’s not about utopia, but it’s about a weak, social version of the Sapir-Whorf hypothesis, that the world you’re raised in, the language you speak, the obligations you’re given, drive your perception of the world.
And here’s a spoiler – the happy ending surprised me. I’ve read this book many times; the first time, more than 20 years ago. But this time, the happy ending surprised me. There’s a sense from the omniscient narrator that Shevek deserves this. But the actual ending, I think, comes in the second-to-last chapter, though the midpoint of the narrative, from an un-named character speaking to the non-governing government of the anarchocommunists. And it is also about Desert.
For we each of us deserve everything, every luxury that was ever piled in the tombs of the dead kings, and we each of us deserve nothing, not a mouthful of bread in hunger. Have we not eaten while another starved? Will you punish us for that? Will you reward us for the virtue of starving while others ate? No man earns punishment, no man earns reward. Free your mind of the idea of deserving, the idea of earning, and you will begin to be able to think.
There aren’t enough Calvinist communists in the real world; for that, we have to go to Tau Ceti.
Albert Hirschman
I mostly knew of Hirschman from Oliver Kim’s wonderful posts about the Nazi-fighting development economist. Early last month, I read his 1970 book Exit, Voice, and Loyalty, a monograph on economic and social feedback mechanisms. He sets up a comparison: economists prize exit as driving competition, and politics scholars model voice mechanisms through voting and protest. Yet the line between these two often blurs, and they are informative in different ways. This binary is really simple, but interesting things fall out of the models that Hirschman constructs. For example, Hirschman explores cases in which exit becomes easier (e.g., a new competing brand arrives in the shop), and voice becomes less prevalent. This is a good and dense book, with relatively little development economics in it.
One of my favorite papers from the Joural of Economic Literature is Hirschman’s “Rival Interpretations of Market Society: Civilizing, Destructive, or Feeble”. The cultural role of capitalism and markets is another of his obsessions in history of economic thought, more famous in his book The Passions and the Interests which I’m reading now. His JEL paper catalogues various theses: perhaps markets maketh manners, or perhaps markets undermine our social relations, or perhaps capitalism promotes civility, but only in democratic cultures. These theses predominated in waves, and Hirschman likes all of them. In this paper and the book, he tries to synthesize them.
For capitalism to be both self-reinforcing and self-undermining is not any more contradictory than for a business firm to have income and outgo at the same time!
He writes that, among all of his works, this idea “has the special quality of standing free and of evolving freely and independently.” All of this other books were “written in order to prove someone else to be – or to have been – wrong.” Hirschman still finds plenty of opportunities to call out historical economists for self-contradiction, but I get the sense that this is the work where he feels most a part of the great conversation of social scientists.
Links
I am very glad that baseball is back. Here’s John Rawls on why he and Harry Kalven loved baseball. Some of these are copeFor example, this written in 1981 looks more foolish today: ‘the rules of the game are in equilibrium: that is, from the start, the diamond was made just the right size, the pitcher’s mound just the right distance from home plate, etc., and this makes possible the marvelous plays, such as the double play. The physical layout of the game is perfectly adjusted to the human skills it is meant to display and to call into graceful exercise. Whereas, basketball, e.g., is constantly (or was then) adjusting its rules to get them in balance.’ Everyone loves the pitch clock and ABS and bigger bases. But it’s still a miracle that stolen bases are remarkably stable. 
but largely.
It’s not going to be an especially great season for the Giants, so here’s another of my favorite early season baseball essays.
There are those who learn after the first few times. They grow out of sports. And there are others who were born with the wisdom to know that nothing lasts. These are the truly tough among us, the ones who can live without illusion, or without even the hope of illusion. I am not that grown-up or up-to-date. I am a simpler creature, tied to more primitive patterns and cycles. I need to think something lasts forever, and it might as well be that state of being that is a game; it might as well be that, in a green field, in the sun.
Here’s a fun book review by John Psmith about John Boyd, the Air Force strategist and father of dogfighting. The comments are very informative.
I’m told that this medical case report has been making the rounds among cryonics people. It’s about an eight-year-old Pennsylvanian who fell in an iced-over pond, spending 2.5-3 hours underwater, and eventually made a full recovery. The worst part of EMS for me, by far, was working cardiac arrests which seemed hopeless. Bodies are more robust than I imagined, and miracles are possible.
This is a lovely story about life on a small Scottish island in the Atlantic, thousands of miles from anything.
The Standford Encyclopedia of Philosophy absolutely cooks Zeno on his fourth paradox:
The general verdict is that Zeno was hopelessly confused about relative velocities in this paradox.
Here’s a fun very short story: “What will one day destroy each of the 50 US States.”
The novelist Helen DeWitt was awarded a prize for $175k by Yale. This prize required her to travel, record a podcast, and record a video. She was incapable of doing this. Here’s her side of the story. I love her novels, though I’m struggling with her latest Your Name Here, and so I was initally very sympathetic. But imagine being a publicity guy at Yale and getting this email:
Tried to return to Starbucks today to reply and got lost 6 times wandering the streets of Amsterdam, which is typical when the voices coming into the head take up 90% of capacity. I know from experience that if I keep pushing to deal with practicalities while the mind is in this state the next stage is losing passport, cards, keys, finding myself locked out in the street with no way to contact anyone who could help. Important to back away before cracking up.
Or this one:
You might not understand. Last year I agreed to do a seminar on zoom for Brandeis and so booked a blow-out with a hairdresser on the day - deliberately not booking a cut bc much cd go wrong. The hairdresser suggested a little trim, did not have resolution to refuse, her “trim” of the bangs left me with a Mr. Spock cut ON THE DAY. So when you suggest a video, I think that I am not strong enough to control a hairdresser, and I don’t want to be filmed looking like Spock.
I felt bad for everyone involved, except for me, because I got to read this:
Long silence. Made what would later prove fatal mistake: spent hours play mah jong on my phone, under impression I had unlimited roaming data in EU.
And then, I didn’t even feel bad for DeWitt, because Tyler Cowen gave her $175k, no strings attached.
Dynomight in Asterisk magazine on why we’re taking the wrong painkillers. Acetominophen almost always beats ibuprofen. I’ve a big fan of Dynomight, especially his footnotes:
At one point while researching all this I had what I thought was a good idea: Why not sell acetaminophen in pills bundled together with NAC? The NAC would replenish glutathione stores in the liver, seemingly reducing the risk of overdose. Later on, I developed more humility and felt very stupid for fantasizing that such an obvious idea could be novel or useful. I think that this is indeed a bad idea because NAC itself has side effects, though I can’t find much formal discussion. In fact, I found a 2010 editorial called “Why Not Formulate an Acetaminophen Tablet Containing N-Acetylcysteine to Prevent Poisoning?” In another study, Nakhaee et al. (2021) actually tried giving NAC together with acetaminophen to rats and found that this seemed to make it better at reducing pain. So maybe this isn’t a completely stupid idea. That last paper also led me to discover that “rat hot plate test” is a standard phrase, and one that drives home what humanity’s dominion over nature means in practice.
Previously
…