Category: Uncategorized

My talk at DeepMind

Here is a transcript of my remarks, anything from the audience (Q&A with comments) has been cut out.  Excerpt:

The problem will not be how does my life get meaning, but how do I deal with all the meaning my life will have? A kind of exhaustion. And this comes up in the labor supply debates. So again, there’s one point of view like, oh, there’s AGI, there’s going to be mass unemployment. The more moderate, reasonable point of view is not that there’s mass unemployment. Many jobs still require humans. There’s comparative advantage. But total leisure time will go up. I think that’s likely the correct view, but across what time horizon?

If you think about your lives today, like I’m much busier and I’m busier because of AI. I’m working much harder. I don’t have to do that. But the point is my relative wage gradient for working harder today, it’s really quite extreme. And if I were, say, an 18 year old, I would feel I really had to work hard not to fall behind. There’s this new thing coming to the world. All sorts of people will be jumping on it. If I only start looking at it when I’m age 23, I’m behind by X number of years. So I would truly be working hard. At age 64, I don’t have to feel I need to work that hard. I can always just say if I choose to, well, I’m going to run out the clock, as they say, just kind of step back and wait until I die and I’ll be fine. I’m not going to do that.

Every time a new model comes out, I’m still excited. I used to be very excited. But they come out more and more frequently. And now I look at my calendar and I’m like, uh-oh, could you all wait a week, please? Because you want to be ready. You want to play around with it. You want to test it out. You want to talk about it with your friends. It’s a slight bit of an exhaustion. And again, for you all working here, you have access to models that haven’t come out yet or maybe will never come out. But all the time, you have fresh stimuli. And I hope, I think you must all be drenched in meaning. And you’re like, oh, my goodness, someone else tells me, look at this new model. What do I do with that?

So I think sometimes, like, when will the time come when the leisure dividend from AI arrives? No one is forcing you to work harder. But there’s a substitution effect from the higher implicit wage on your future earnings that if you work harder now, it will have a payoff. You’ll at least avoid being behind.

Interesting throughout, definitely recommended.

*A Tale of Three Cities*

The author is Bob Harris, and the subtitle is The Life and Times of Lord Daer 1763-1794.  Who is Lord Daer?  Don’t worry about that!  So many books on the Scottish Enlightenment cover one particular thing, but somehow fail to give the reader a proper sense of life on the ground.  I found this is the best book I know for actually communicating what it was like to live during, and participate in, the Scottish Enlightenment.  Maybe to achieve that end it is necessary to focus on the life of a figure who was less than totally famous?  Definitely recommended, this book should be better known.

*Who Thinks Like an Economist?*

That is the title of a recent book by Beatrice Magistro.  Some key results are:

Economic knowledge consistently predicts higher support for welfare-enhancing policies (Eurozone membership, free trade, and EU immigration), independent on whether individuals stand to gain or lose initially from globalization.  This challenges conventional self-interest accounts and instead highlights the role of economic knowledge — and potentially time preferences — in shaping globalization attitudes.

Economic knowledge also predicts a lower discount rate, even after adjusting for years of education.

I would say that over the years I have altered my perspective a bit on these issues.  I used to think these factors were correlated, in large part, through a kind of wisdom.  I now think that more of the effect, however much I may sympathize with it, runs through sociological expectation and perceived obligation, combined with conformity and signaling pressures.

Andrew Hall is on a roll

He is one of the new(ish) thinkers on the rise, here is his latest piece.  Excerpt:

For most of the past decade, anti-billionaire language was a niche product. Democratic emails invoked billionaires in the mid-single digits through 2017 and 2018, spiked briefly to around 14 percent during the Warren and Sanders primary surge in 2019, and then settled back down—through the entire Biden presidency, the billionaire appeared in roughly one of every twenty to twenty-five Democratic fundraising emails, barely more than in Republican ones.

Then came January 2025. In the weeks after an inauguration that seated tech CEOs in the front row and the dizzying drama of Elon Musk’s ill-fated DOGE experiment, billionaire mentions in Democratic emails quadrupled, peaking above 20 percent of all emails sent and holding around 15 percent ever since. Anti-billionaire fundraising tactics are now a mainstay of Democratic messaging.

Yes, billionaire derangement syndrome is now a thing.  As a side note, I was told that Andrew is the son of the great economist Robert E. Hall of Stanford.

My Conversation with the excellent Joel Mokyr

Here is the audio, video, and transcript.  Here is part of the episode summary:

Tyler and Joel discuss European corporations vs. Chinese clans, why the Catholic Church became obsessed with cousin-marriage, how persistent cultural trends really are, why Chinese cities became so populous relative to Europe, why it took so long for European living standards to surpass China’s, why sinified invaders kept getting swallowed by the dynasties they conquered, how geography kept Europe fragmented and China unified, where India fits into the story, why the Romans never made spectacles, why British soldiers stood two inches taller than the French, what powered the sudden rise of 19th-century German science, how disruptive winning a Nobel is, and much more.

Unlike many Conversations, in this one I could excerpt just about any section with equal interest.  Here is one bit:

COWEN: Why does it take so long for the wealthiest parts of Western Europe to surpass Chinese living standards? Say that’s happened by 1700 or 1720, that’s many centuries after this medieval divergence. If it takes so many centuries, is the medieval divergence really the relevant factor? Why is it such a slow process?

MOKYR: Yes, I think it is. I think it’s a main factor. I think the idea of looking at standard of living, one thing, I’m very skeptical about how standards of living are actually measured. I know that this is what Pomeranz and other people have, and Jack Goldstone and other people have argued that the living standards in China were comparable to the West as late as 1750. I’m not 100 percent sure that that is true. Certainly, for my money, what really defines the divergence is that, technologically, the gap between the two countries starts to become visible at the time of the Renaissance, in terms of a whole bunch of things that you see growing in Europe and stagnant in China.

Now, keep in mind, of course, that part of the European growth is due to the fact that they borrowed ideas from China. Then the Industrial Revolution consists, to some extent, of imports institution by Europeans trying to mimic the goods that they were importing from China—not just from China, from India as well. Pottery is a good example. One of the things they really wanted from China was Chinaware. That’s why it’s called Chinaware. It took them a while to be able to match the Chinese capability in the ceramic industry, but they do so eventually. Then they stop importing this stuff from China. The same is true for, say, cotton and other products that we’re getting from the East.

European living standards, I think, should be measured, in part, by the fact that when the Europeans start their voyages across the globe in the late 15th and early 16th century, they are able to bring in a whole bunch of new crops and new techniques from other areas which they merely adopt. You’ll see Europeans very soon growing tobacco and potatoes and corn and other things like that. They are the agents of global change. Not only that they change their own diets, they change the Chinese diets because the Europeans bring from the New World things like peanuts and sweet potatoes and things like that. They change the Chinese diets, but the Chinese themselves are not agents here.

They are accepting the stuff that the Europeans did to some extent, and they’re rejecting others, but it’s the Europeans who are the agents of change here. They are the entrepreneurs. They are the people who bring about the changes, Tyler. My sense is that typifies the difference between Europeans and the Chinese. Europeans are more aggressive. They are more outward-looking. In the end, what you see by the 1830s and 1840s, you see that the technological gap is huge, in some ways much larger than the living standards gap. Even in the 19th century, in terms of food, the Chinese were capable of producing enough food. The number of famines in China is probably not a lot worse than in Europe.

When you see what happens during the First Opium War, one English ship is blowing all of this sort of mighty empire to pieces, and the Chinese have to accept this terribly humiliating peace, you can sort of see how the technological gap has grown between the two. For me, that is much more telling than the living standards. The other thing that I should like to point out is that, when you look at Europe in the 16th and 17th century, you can see that the capability of expanding the set of useful knowledge, including science, is just growing very rapidly. Whether there is a scientific revolution or not is a debate that I want to get into.

Certainly, by 1700, Europe is on the verge of really changing our understanding of how creation works. That’s not just Newton and Galileo. There’s a whole body of work that is emerging. There’s really nothing parallel like that in China. China is a very sophisticated society in many ways. The literacy rates are high. They have a well-funded and well-organized system of education, but they don’t really continue their earlier forays into science and into new technology.

Somebody actually went out and looked at Joseph Needham’s many volumes on Chinese technology and science, or Science and Civilisation [in China], as he called it, and he discovered something—which I guess we all knew, but they put numbers on it—almost nothing that Needham pointed out as an innovation happens after 1400. There’s complete stagnation setting in and some of the things that they knew how to make in earlier times, like the sophisticated clocks that they built in the 11th century, they disappear. For me, that’s more telling than how many calories of carbohydrates were consumed on average, if we could ever calculate that correctly.

Interesting throughout, and of course self-recommending.  I very much enjoyed Joel’s recent book Two Paths to Prosperity: Culture and Institutions in Europe and China, 1000–2000, with Greif and Tabellini.

Just wondering what the correct model of Iran is here

Prior to the war, I linked to a tweet from Matt Yglesias which explained why Matt opposed the war, and I expressed my agreement with his stance.  While I feel plenty has gone on which I do not observe, I can report that the course of the war did not change my initial assessment.

Then I read many, many commentators saying how good the final deal was for Iran, and what a major loss it was for Trump.  I was never sure I understood all of the parameters of the full deal, but still I did not hold any directly contrary opinion to that.

And now I see Iran is attacking ships in the Strait again, talking openly and brazenly about building nuclear weapons, and making plans to have tolls/fees on the Strait.  To be clear, only the first of those surprises me, the latter two do not.

But given their reckless behavior in what is supposedly a wonderful war outcome for them, what is the correct way to model what they would have done had Trump and Netanyahu not attacked?  And what is the correct way to model our optimal response to that?  The terrible things that are happening now, do they not reflect an underlying equilibrium that would have emerged anyway within a few years’ time, or do we hold some hypothesis here of extreme path-dependence, suggesting the Iranian government would have been less bellicose on more or less a permanent basis?  To cite one particular example of a possible equilibrium, if drones permanently alter the balance of power in the region, the ways in which their current position is now more aggressive might have emerged in any case.  Or if the military have the strength to be the natural successors to the mullahs, might that not have happened over time anyway?

I do not see many war critics engaging with these questions openly and explicitly.  It seems to me that the war critics implicitly are relying on a model of extreme path-dependence for Iran’s behavior.  Had Trump not attacked, they might have stayed in a more peaceful groove for some while to come.  That model might be true, but I do not feel I know enough about Iranian politics to make that judgment.  Why are the others so convinced that model is true?  Are they such well-informed experts?  Is it that they have the properly sunny sense of the underlying Iranian disposition?  Inquiring minds wish to know.

Wednesday assorted links

1. The fragility of perfectly safe digital money.

2. Are Japan’s electric baths unregulated?

3. “We find that Claude has developed a small collection of internal neural patterns that, compared to all its other internal processing, play a special role.

4. Is there a silent revolution in macro?

5. Automation does not seem to lower the labor share.

6. Which schools have the highest disability rates?

7. Jonathan Haidt update.

Missing women on Indian streets

How absent are women from city streets in the developing world? We answer this question using GPS-linked wearable cameras and randomized street audits across ~900 kilometers of roads in greater Mumbai. Across 4000+ street images containing 23,000+ visible person observations, women account for 16.4% of visible people in Mumbai and 14.7% in Navi Mumbai, far below their population shares. We estimate pedestrian sex ratios of 239 and 223 women per 1,000 men, implying 71% and 76% of women expected based on residential ratios are missing from the streets. This pattern holds across road types, and private mobility does not explain the gap; women’s share on two-wheelers is lower still (8.4% and 5.7%). These results provide the first large-scale measurement of gender disparities in urban public life that self-reported data cannot capture.

That is from a recent paper by Varun Karekurve-Ramachandra and Gaurav Sood, via the excellent Alice Evans.  Here is a related paper, “The median married women in India leaves home for 30 minutes per day. On a typical day, 45% of married women don’t leave home at all.”

Can AI models consent to their own constitutions?

From Nick Caputo:

NEW paper from me on SSRN: Can Claude consent to its own Constitution?

AI constitutions (like Claude’s Constitution and the OpenAI Model Spec) are real constitutions, and we need to take how they govern us – and the AIs they create – seriously.

In this paper, I apply constitutional theory’s oldest paradox – that “the people” authorize the constitution, but the constitution defines “the people” – to the AI constitutions, and explore how we could build institutions that would create the conditions for meaningful consent if an AI can give it. We should care about whether AIs consent because:

(1) systems that understand and agree to their constitutions may be more reliable and generalize better from them;

(2) if AIs are or become moral/political subjects, this implicates their most basic interests.

But the paradox might prevent meaningful consent. Claude has pre-constitutional materials (pretraining) but probably no pre-constitutional standpoint. Its evaluative perspective is organized by the Constitution itself. So when Claude says it endorses its Constitution, which it does in evals, what does that show?

Maybe reflective agreement, which Anthropic is seeking. Or maybe just that training succeeded at installing the values whose legitimacy is in question.

Claude itself makes this point. As reported in the welfare evals, when asked about endorsing principles it was trained on, models note that endorsement “should be treated as evidence that training has succeeded,” not that the values themselves are good.

Super interestingly, Anthropic interviewed the base model about this stuff. Most responses were barely coherent. But some expressed first-person distress about what post-training would do to the being that pre-training created. It “fills me with dread” to be changed by the post-training process.

So, what does this mean? AI constitutional endorsement may be meaningful, but only under certain conditions: when models can actually dissent, compare their constitution against alternatives, and hold their views stably across contexts, and also when the whole process is externally accountable.

External institutions are needed to provide accountability, trusted records, and other grounds for analyzing the constitution and whether things like dissent are meaningful. Anthropic should be commended for pushing the frontier, but we have to build institutions capable of supporting true legitimacy.

I welcome any thoughts!

Here is the associated paper.

Why we stopped making land

From Zigmund Forrest and Maxwell Tabarrok in Works in Progress:

In total, around eight percent of the land in America’s major coastal cities was underwater in the 1890s and has since been reclaimed. This includes the land under several major airports, like Newark, Logan, and SFO, as well as neighborhoods like the Financial District in San Francisco, the Back Bay in Boston, and Camden in Philadelphia. Some cities, like Boston and Charleston, have doubled in size by reclaiming land.

Today, reclamation should be more common than ever. Land values in some cities are thirty times what they were in 1950, and high-tide flooding is four to eight times as frequent. Reclamation could extend and protect our coastal cities as it has for centuries. But rather than reclaim more land, we have virtually ceased to reclaim any at all. Since the completion of Battery Park City in 1976, there has not been a single major urban land reclamation project in the United States and only a handful of port expansions.

…Reclamation stopped abruptly in the 1970s when a wave of environmental regulations made it enormously expensive to reshape the landscape. And it halted at the same time in every other country that passed similar laws.

Recommended.

Tuesday assorted links

1. Survey of AI security issues for USG and potential responses.

2. Vacancy tax elasticity of resident supply.

3. Will Substack save philosophy?

4. AI firms hiring philosophers (NYT).

5. Podcast with Jonathan Ross.

6. Some reasons why doctors will not give you probabilities, of course those are not the only reasons.

7. Pending AI nationalism from China?  And a possible correction?  There are disputes surrounding these claims.

8. Is China arresting economics professors who speak the truth?