Some people thought Donald Trump’s approval rating would have a floor around 40%. Well, it didn’t:

The nation’s disapprobation is richly deserved. I hardly need to go through the familiar litany of why Trump is bad — the failed war in Iran, the ridiculous tariffs, the destruction of America’s alliances and prestige, the attack on American science, the rampages of ICE, the unprecedented corruption, the authoritarianism, the election denial, the appeasement of Russia, etc. I guess I just did go through it, though. It feels a little cathartic to type it out once again.

If you think the GOP will return to being the party of Mitt Romney, John McCain, or Ronald Reagan after Trump is gone, think again. MAGA is already morphing from a personality cult into an ideology — a deeply xenophobic, economically retrograde ideology obsessed with expelling immigrants, and convinced that wealth has made the West immoral and weak. The party’s lower ranks are already filling up with unabashed fascists, Hitler apologists, and all sorts of scary people:

GOP veterans fret about the rising generation of young staffers in their party…Of specific concern are those labeled groypers — technically, followers of the self-described “White Identitarian” Gen Z streamer Nick Fuentes, though it’s also used as a catch-all term of abuse for young, anti-establishment right-wingers…

Prominent conservative commentators have alleged that groyperism is spreading among younger members of the GOP, including in government — a view seemingly backed by examples like the young Trump administration lawyer who admitted to a “Nazi streak” in text messages…The anxiety over groyperism in the GOP talent pool corresponds to a rise in antisemitism across the political spectrum, including among young conservatives…A Manhattan Institute poll found that a full 25 percent of Republicans under 50 self-identified as having antisemitic views, and 31 percent as holding racist ones…

[T]he troubling anecdotes are legion: gas chamber jokes and ethnic slurs bandied about on Young Republican group chats leaked to POLITICO last fall; a New York Magazine report from just last week linking a former New York Young Republican treasurer to a social media account denigrating Jews, Black people, Muslims and gay people; a House campaign adviser fired for mocking a rape victim on X; the Ron DeSantis presidential campaign staffer axed for retweeting a video featuring a Nazi image, only to join the office of Sen. Eric Schmitt (R-Mo.) and, eventually, the State Department, where he works today.

Meanwhile, at the top of the party, election denial has become a litmus test — a profession of faith for the bizarre Trump cult. The number of Republicans who will declare in public, in no uncertain terms, that the 2020 election was legitimate is now negligible. And being a vocal election denier is a fast track to prominence in the modern GOP — witness Tom Tiffany, the Republican candidate for the governorship of Wisconsin.

I see no reason to expect anything good from the modern GOP. When good ideas do make their way into Republican policy, it’s generally by heroic pragmatists quietly working within the system. But those small victories are becoming smaller and less frequent, as MAGA’s obsession with waging political warfare against its near-infinite roster of enemies muscles out any concern for the good of the nation as a whole. Americans are fed up with this, and they are very right to be fed up.

The Republicans have never been this bad before. But in the past, when they got bad, the solution was always pretty simple: Vote for the Democrats. Democracy isn’t perfect, but it does have a natural release valve; you can always just “throw the bums out.” And that’s exactly what we used to do. In 2006, 2008, 2018, and 2020, we grew disgusted with the GOP’s antics and elected Democrats to replace them.

I think if you look at those episodes, you can see a pattern of diminishing returns. Obama was a broadly successful president on domestic policy — he implemented broadly successful anti-recessionary macroeconomic policies, reformed health care and finance, and beefed up the welfare state. Biden gave us the return of industrial policy, but his fiscal profligacy — often squandered on things like student debt forgiveness and subsidies for already-overpriced service industries — contributed to inflation and did nothing to rein in the ballooning debt. On social issues, Obama was a moderate who tried and failed to curb his party’s drift to the left; Biden didn’t even try, running as a moderate but letting “woke” subordinates implement a lot of unpopular culture-war stuff.

Now it’s 2026, and the Republican regime in power is far worse than its predecessors in either the 2000s or the 2010s. Once again, we have the option of electing Democrats. And we should elect Democrats, this November. There is no ambiguity here; a Congress that challenges Trump’s lawlessness, investigates his unbridled corruption, and takes a more forceful stand in favor of Ukraine will be unambiguously better than the supine GOP Congress that can do nothing except worship at the man’s feet.

But at the same time, looking at the way the Democratic Party is headed, I think we should be asking how much better things will be under Democrats, including if they win the White House in 2028.

Let’s look at the type of candidates who are winning Democratic primaries. Francesca Hong, a DSA member who is all but certain to be the Democratic nominee for governor of Wisconsin, tweeted in 2020 that we should “cancel Thanksgiving”. When asked in a recent interview if she had recanted that position, Hong refused, talking about how Thanksgiving was “incredibly painful” for many people:

As if that weren’t tone-deaf enough, she appeared to criticize herself for marrying a white man (saying “assimilation…took over”), and talked about having a half-white son as “proximity to whiteness”:

Hong backtracked a couple of days later, declaring that Thanksgiving is her “favorite holiday”, but the damage appears to have been done — Bernie Sanders said he didn’t think he would endorse her, and prediction markets immediately began to discount Hong’s chances of beating her Republican opponent in the fall:

Some had hoped that the Democrats’ socialist faction would be able to escape some of the baggage of “woke 1”, because of their focus on class politics, and because the Palestine issue might be able to muscle aside a whole bunch of unpopular 2010s-era culture-war stances. But Hong shows that this hope will be in vain; the DSA types are the wokest of the woke.1 A number of leftists, such as Bernie Sanders’ former campaign press secretary Briahna Joy Gray, doubled down on Hong’s anti-Thanksgiving stance:

Meanwhile, a number of progressive commentators, like Mehdi Hasan, castigated moderates for suggesting that Hong might not be worth voting for, pointing out that Hong’s opponent is an election denier:

This demonstrates exactly what I mean when I say that Democrats have spent their political capital moving to the left. Trump’s vast number of mistakes and outrages have alienated much of the country. Democrats could have used this alienation and outrage to become a long-term majority party, and relegate Republicans to the political wilderness — a just and fitting punishment for falling in line behind a man like Trump.

Instead, what they’re apparently choosing to do is to select candidates who are so far to the left that they end up frightening and/or offending many voters, and then basically daring people to keep voting for the Republicans. Sure, she wants to cancel Thanksgiving, but you wouldn’t vote for a fascist election denier, would you? And so on. The median voter is pulled back into a state of disgusted ambivalence between two extremes, and people like Mehdi Hasan get to take the Democrats in their preferred direction.

(As an aside, if I lived in Wisconsin, I could not in good conscience cast a vote for either Tom Tiffany or Francesca Hong, so I’d just file a protest vote or leave it blank.)

When DSA-type candidates were just winning in deep blue cities, it was possible to think that they would remain a small, limited faction within the party. But Francesca Hong is the Democratic nominee in a statewide race. So is Abdul El-Sayed, who, while not a member of the DSA, is closely aligned with them (especially on Israel, which he focuses on a lot). El-Sayed just won his primary for the Michigan Senate race, but a closer-than-expected election, with very little working-class support, shows how his selection represents an expenditure of Democrats’ political capital.

This, increasingly, is who the Democrats are now. It’s doubtful that a leftist candidate will win the presidential nomination in 2028, but — just as Biden made concessions to Bernie Sanders supporters on student debt relief and other issues — whoever does get the nod two years from now is likely to let the ascendant DSA faction set a lot of policy priorities.

Meanwhile, the commentators promoting the DSA’s cause continue to breathe fire. Hasan Piker, who is emerging as a key kingmaker within the rising leftist movement, called El-Sayed’s Democratic critics “Islamophobic pieces of shit” (El-Sayed also accused his critics of Islamophobia).2

Compared to some other DSA types, though, Piker can seem downright moderate. Strident anti-American rhetoric is now standard among that faction:

And there are even less savory attitudes lurking in DSA-aligned activist spaces. Susan Abulhawa, a Palestinian-American writer and activist, repeated Nazi accusations that Jews stabbed Germany in the back in World War I:

The leftist magazine Current Affairs, meanwhile, recently published an article claiming that electronic dance music is a tool for ignoring genocide and upholding the “neoliberal order”.

It’s easy to dismiss this stuff as fringe radicalism that will never make itself to the mainstream of the party, but — as with “groypers” on the right — that increasingly feels like cope. Francesca Hong demonstrates that anti-American ideas have become fully mainstreamed and accepted within the Democratic Party; not everyone holds these beliefs, of course, but those who do hold them are firmly under the party’s big tent, and will be given some degree of influence over policy.

A Democratic party in which leftists wield significant power will not be the Democratic Party of 2020 or 2008. It will be something else. Research shows that leftists are not liberals — they have their own distinct ideology.

That ideology is not necessarily the antidote to MAGA. Yes, if DSA types held national power, things would be different. The disastrous Iran war would end, which would be a good thing. But on foreign policy in general, a faction that sees America as an evil empire seems unlikely to restore the liberal alliance system that Trump has thrown in the trashcan. Recall that the DSA blamed the United States and NATO for Russia’s invasion of Ukraine, and called for withdrawal from NATO.

On economic policy, too, a DSA administration might look more Trumpian than we might hope. DSA types say they’re now in favor of free trade (thanks to negative polarization against Trump’s tariffs), but the impulse to protect American labor and exercise more command and control over the economy would probably prove irresistible. As for the national debt — the country’s biggest looming economic threat — I just don’t see the DSA as being the balanced-budget types. Remember that Francesca Hong wants to cut property taxes by 44%.

And where the DSA would break from Trumponomics, it wouldn’t always be to the good. Trump has hurt the economy in some ways, but he has always been very careful not to mess with the AI buildout — the force that’s keeping the economy afloat. The DSA, however — like many progressive Democrats — wants to halt data center construction. This, of course, would remove the U.S. economy’s biggest tailwind.

I still think the choice of which party to vote for in 2026 and 2028 is unambiguous. Trump has been too bad, in too many ways, to think that he and his movement are the lesser of two evils. But to the degree that the DSA’s increasingly frequent electoral victories translate into real power in the next Democratic administration, it seems like we have less to hope for from the Dems now than in times past. If the U.S. is going to turn its problems around, “throw the bums out” is no longer enough.

Subscribe now

Share

I had a great-uncle who was a “human calculator”. It’s an odd, rare ability that allows people to multiply large numbers quickly in their heads. I remember using a hand calculator and seeing if my uncle could get it right without a tool faster than I could; more often than not, he could.

Human calculators used to be very valuable as employees, back when calculation was done by hand. Today they’re just a curiosity. Technology came along and superseded that particular human ability, as it has so many others. The story of John Henry being superseded by the steam drill, or Paul Bunyan by the chainsaw, is a metaphor for the march of progress. In some cases, like Garry Kasparov or Lee Sedol losing to computers at chess and Go, the moment really happened.

This week marked another important such moment in the history of man versus machine. OpenAI announced that its new model, Astra, managed to solve ten major outstanding problems in math and theoretical computer science. Here’s a simple explanation of the problems. The general consensus seems to be that these are extremely important, stunning breakthroughs.

This does not mean that computers are now capable of doing every kind of math better than humans can. Some people have argued that AI is good at results where the solution can be found by reading the entire literature and combining existing insights — exactly the kind of thing you might expect a computer to be superhuman at — but still not as good as humans at making truly novel leaps of insight. This may represent a real, general limitation of LLMs’ capabilities — as Tom Zahavy of Google DeepMind put it, it may still be the case that “LLMs can’t jump.”

But the trend line is becoming clearer. The University of Toronto’s Daniel Litt, one of the most prominent skeptics of AI’s potential for frontier math research, recently conceded a major bet about what AI could do:

Litt appears to no longer believe that there is any important type of math that AI won’t soon be able to do, writing: “The models don’t yet seem able to do certain kinds of high quality research, but I also expect this is a matter of (not too much) time.”

Some mathematicians have reacted to this development with despair. Kirwin Hampshire’s post is probably the most evocative and the most well-read:

Ossuary Lost at Sea

The Dark Night of Mathematics

I am going insane. During the last week or so LLMs have produced a number of counterexamples to significant long-standing conjectures. I will not recount these happenings here, there are many places where you can find the details…

Read more

14 days ago · 1148 likes · 361 comments · Kirwin Hampshire

He writes:

I am suffering a profound spiritual crisis due to these developments. I have been screaming internally for days. It feels as though I am living inside of a nightmare…For me, the affective quality of learning mathematics is empathetically tethered to an act of discovery and creation…If The Library of Babel existed, would authors continue to write books?…

Perhaps I shouldn’t tell you this, but my aim is to be open: These developments have triggered some deranged thoughts in me. I have wondered if it is the express goal of these companies to make me kill myself…The story of human discovery and the triumph of the human spirit will soon be excised from this discipline…[T]he process of prompting novel proofs will be as auraless as ordering doordash. Watch as magic and mystery evaporate. Watch as the sun sets on our heroic age…

There is nothing I can do. There may be nothing you can do. I have no prescriptions, policy recommendations, or coherent call to action. I just want to be honest and open about my emotional and spiritual response. I want to feel seen….I need the architects of our new mathematical paradigm to look me in the eye and acknowledge our shared humanity and soul before they deliver the coup de grâce.

Math PhD student Tasmin Chu, meanwhile, urges a more confrontational path, calling on mathematicians to avoid working with AI companies and to aggressively preserve the way that math research is currently done. This seems extremely unlikely (and would fail if it were attempted), but it demonstrates the depth of emotion out there.

Even the mathematicians who don’t share Hampshire’s despair have expressed alarm at the sudden disruption of their profession. The entire way that mathematical research works — training grad students, working on problems, collaborating with colleagues, building theories — is going to have to be overhauled.

I’m not a mathematician; I never have been. I can’t fully understand what mathematicians are going through right now. But I do have a few thoughts on what AI’s supersession of human mathematics research means for our species, so I thought I might as well share those. Some of these ideas are a little vague and half-formed, and for that I apologize in advance.

Heroism has to change

Humans have always valued standout individual achievement — heroism. We pride ourselves on being able to do what no one else could have done, and we heap status and respect on people who make standout achievements. But this was always a little bit of an illusion; most of what we did was always a collective enterprise. Workers work in teams, and those teams rely not just on the resources of their organizations, but on the smooth operation of a whole network of other teams and organizations all over the world. Even the most heroic entrepreneur stands on the shoulders of a vast number of workers; Elon Musk could not build a rocket, nor could Jeff Bezos create Amazon Web Services. And neither SpaceX nor Amazon could have been built in Somalia.

We always bridled at being reminded of this; when Obama told entrepreneurs “You didn’t build that”, he was making the anodyne point that organizations and institutions matter, but people still got very mad. We motivate ourselves with our own sense of heroism, and to be reminded that we’re cogs in a machine greater than ourselves makes life seem a little less meaningful.

The scientific enterprise was also always mostly a collective one. If you look at what almost any scientist does, they’re simply adding a little bit of data to our pool of collective knowledge. Almost all research is incremental stuff — a slightly novel experiment, a little wrinkle in an existing theory — and many published research findings are false, and yet the scientific enterprise collectively gropes its way toward the truth. Even Nobel prize winners are often just managers of huge teams of researchers who never get gold medals or get to give speeches in Stockholm.

In essence, humans were always what we now call edge compute — little devices that reported data back to a central world-mind. Except the world-mind we reported to wasn’t a computer; it was society itself — the network of corporations, governments, universities, and human networks that disseminated human knowledge and coordinated our actions. We started off by learning pieces of what the world-mind already knew (i.e. education). Then we walked around and saw slightly different things, had different experiences, chewed on tiny bits of bigger problems, gathered a little new data, and so on.

We mixed our tiny bits of unique experience with what we learned from the world-mind, and the resulting product was slightly novel because our individual experiences were unique. Then we reported this slightly-novel mix back to the world-mind — with scientific papers, PowerPoint presentations, office conversations, memos, and so on. The world-mind reassimilated our contribution, and learned, and grew a little bit wiser.

But there were a few of us who really did get to be heroes — or at least, a bit more heroic than the rest. Whether because of greater raw ability, or a luckily unique perspective, there were some people who could do things that the entire rest of the world couldn’t do. There are examples from politics and business and art, but there was always something special about the heroes of mathematical theory. Isaac Newton invented calculus and classical mechanics all by himself. Grigori Perelman, sitting in his little room in Russia, solved a famous math problem that had bedeviled the entire profession for decades. Andrew Wiles did something similar a generation earlier.

These advances weren’t done ex nihilo, of course — every great discoverer and inventor stood on the shoulders of proverbial giants. But in mathematical theory there was almost always some nub of hard work, some brilliant leap of genius, that belonged to one person alone. Some part of the excitement of becoming a mathematician surely had to do with the dream of becoming one of those heroes, without whom a great discovery simply couldn’t have been made.

But that heroism was necessary because of an inherent limitation in the way the old world-mind was built. Society is far more complex than any single human, but it requires individual communication in order to transmit ideas. A mathematical theorem has to be expressed in terms that one grad student can learn, or one professor can teach to another. This meant that every idea that got added to the collective ultimately had to be comprehensible by a single human being.1 Which meant that it had to be initially conceived of by a single human being as well. The people who conceived of those almost-impossible-to-conceive ideas were our heroes, and mathematicians were perhaps the most heroic among them.

We normally think of AI as something that replaces a single human worker, but I think that’s the wrong way to think of it. AI is really a new world-mind — a technology that takes the accumulated findings of individual humans (or robots, or other sensors) and integrates them into a general picture of the world. AI uses all of human knowledge as its training data; each time one of us reports a new scientific finding or expresses a new perspective, that adds to AI’s knowledge and understanding of the world. This is what people mean when they say to “write for the AIs”, but in fact every scientist and every businessperson is just writing for the AIs now, whether they intend to or not.

(I actually think of a financial analogy here. Before the crash of 2008, most financial derivatives — the complex CDOs and such that ended up being involved in the crisis — were traded over-the-counter, by people calling up other people and selling them products. After the Dodd-Frank legislation, trading of most derivatives shifted to central clearinghouses. The shift from human organizations to AI as our coordinating superintelligence feels a little like the shift from OTC information exchange to a central clearinghouse.)

But unlike human society, AI doesn’t need heroic human geniuses. Its ability to understand complex ideas is not limited by the ability of a single human brain to apprehend, intuit, or communicate those ideas. The information bottlenecks that our greatest mathematicians were able to slightly widen have now been done away with completely — or will be soon.

That doesn’t mean human scientists and thinkers have become irrelevant. At least for now, AI still requires us to provide the role of edge compute. We go out into the world, discover new facts, mix them with our unique individual experiences to form new perspectives, and communicate all this back to the AI world-mind. For most people — even most scientists — this is basically the same thing they were doing before, except instead of a group of their colleagues at a seminar, they’ll be reporting their findings to AI.

And for lots of scientists, AI’s mathematical prowess is going to open up a new golden age. Math was always one of the hardest parts of science; it’s something our brains aren’t naturally adapted to. Now that constraint is alleviated. Economists don’t have to worry about wracking their brains writing theory sections; they can focus on getting data and understanding empirical results. Przemek Chojecki writes:

From a perspective of science or mathematics (not mathematicians), this is the best time ever. AI will lead us to the new age of mathematical discoveries and boost science progress 100x.

Whether it’s 100x or 1.5x remains to be seen, of course. But the point is that people for whom mathematics is an input to understanding the world, rather than an output, just gained access to an incredibly powerful tool that makes their lives much easier. For most people, math was always a burden, like stamping metal or drilling holes. Now that burden has been alleviated by machines.

At least until systems of robots and sensors eliminate the need for human research labor, most scientists will get to be a little more heroic than before. They’ll be like the characters in Star Trek — exploring the unknown with the aid of unfathomably powerful machines.

But mathematicians, and anyone else whose sense of accomplishment came from being the indispensable solvers of hard theoretical problems, will have to learn to get their sense of self-worth and accomplishment from other sources. Jacob Tsimerman, who just won the Fields Medal and promptly took a job at OpenAI, sums it up:

I do not do mathematics only to find truth. I do it largely because I enjoy it and I am good at it. I also find it beautiful and am grateful I get to spend my days understanding beautiful things. But I enjoy the challenge, the process, resolving confusions, finding strategies, grappling with problems…There are many people whose primary enjoyment of math comes through problem solving in one of its incarnations. If that disappears, that is not a trivial issue and many of them might not want to do it anymore.

So what will mathematicians do in the new AI age?

Mathematicians after the age of heroes

First of all, let me say that I do not think AI will take mathematicians’ jobs. I sometimes say that it will, but this is a joke. Even if AI is able to do 100% of what mathematicians do today, I believe that mathematicians will continue to be employed, as mathematicians, if they want.

The reason is that mathematicians, as a class of people, are actually incredibly cheap. The sum total of all of the salaries of academic mathematicians in the U.S. is maybe $4-5 billion dollars at most. That’s a rounding error on our GDP — it’s about what we spend on bowling alleys, or Halloween candy. And to be brutally honest, only a few of those are actually doing frontier research — most are employed mainly as teachers. On top of that, it’s an open question whether most of the frontier mathematical research being done by the very best mathematicians is actually producing value for money in economic terms — other than some cryptography and the occasional physics application, there isn’t much use for most frontier math.

In other words, we already employ most mathematicians because we think human understanding of math, even without industrial applications, is inherently worthwhile. That isn’t likely to change in the age of AI. Even if AI “discovers” a vast number of new mathematical results, most people won’t care; what will be important is that some human understands what AI has discovered. And yes, our society — which AI will make even richer than it is now — will be willing to pay the pittance it costs to keep human mathematicians employed as professional math-understanders. For the overwhelming majority of people, that situation will feel no different than what prevails today.

This is actually a point a lot of people have been making:

Just because you press a button and make AI solve a problem doesn’t mean you understand the solution, and it doesn’t mean you immediately have an idea of what other problems you’d like solved. Mathematicians have been given an incredible new tool to understand the world — like miners getting hydraulic excavators or powered drills — and since the amount of math out there to be discovered is probably infinite, using that tool will take a lot of work.

For those mathematicians whose sense of wonder and meaning comes primarily from learning new things — who do math to “find truth”, in Jacob Tsimerman’s words — life will only improve. And while that won’t be as heroic of a job as what math research used to be, it’ll still be a prestigious one, because most humans will still be unable to do it. Even if any high school kid can press a button and solve the Riemann Hypothesis, the output will just be gobbeldygook to them.

And mathematicians who enjoy the extreme mental difficulty of math will still get to enjoy it. AI will make any given math problem a lot easier to solve, and it’s generally a lot easier to understand a solution than to find one. But AI will also come up with — and will allow humans to come up with — much harder problems than have been created so far. There will be problems so difficult that even understanding AI’s solutions will be as hard for a human as solving the Jacobian Conjecture without the aid of a machine.

As for people who just intrinsically enjoy the old way of discovering math, this will turn into a hobby, or a sport. Just as there are still sprinters even though humans can’t outrun cars, there will still be mathletes. And just as there are still woodworkers even though machines can probably do a better job, there will still be smart people who try to solve math problems without the aid of AI.

So mathematicians will still be able to get paid to do math. They’ll still be able to learn new things, have fun, and get respect from society. The only thing they’ll really lack is heroism — the knowledge that their own special, rare natural abilities alleviate a key bottleneck to human flourishing.

But that’s not so bad, really. Most people never get the chance to be heroes. Truck drivers don’t. Financial advisers don’t. Executive assistants don’t. They have to find meaning just from being regular people — from taking care of their kids, having friends, having hobbies, joining civic organizations, getting involved in politics, and so on. It’s not such a terrible fate.

Subscribe now

Share

Would you like to see an example of a badly conceived policy? Here you go:

The bill to require discounts for self-checkout will probably not pass. But what’s notable is that if this bill did pass, it would have exactly the opposite of the effect its creators want. The stated justification is that customers are doing unpaid labor by scanning and bagging their own groceries, and that this discount would compensate them for that effort. But the bill is part of a more general effort to disincentivize automation in the retail industry, in order to protect working-class jobs:

Rhode Island was the first state in the U.S. to enact statewide self-checkout restrictions with its Restrictions on Self-Service Checkout Stations Act, which takes effect January 1, 2027. The act requires grocery stores to have at least one human-operated checkout station for every three self-service stations during peak hours…Similar legislation has either been introduced, enacted or considered in additional states including California, Massachusetts, Connecticut, Washington, and Ohio.

The standard objection to this bill, I suppose, would be that self-checkout already gives shoppers a discount, because it allows stores to save money on hiring human labor, and that these savings get passed on to consumers in the form of lower prices. That argument is correct as far as it goes, but it means the debate ends up as one of technocracy versus populism — protecting the livelihoods of a few cashiers versus giving a huge number of consumers very slightly lower prices. There’s no simple or easy conclusion to that debate.

But in the case of New York’s proposed self-checkout discount, the problem is that the bill would actually make grocery stores fire human cashiers in favor of more self-checkout.

To understand why this is true, first imagine an incredibly extreme version of the policy. Imagine that instead of 10%, you made a law forcing grocery stores to give a 90% discount for self-checkout. This is exactly the same as forcing stores to charge 10x for using a human cashier. Who would pay 10x for groceries, just to use a human cashier? Basically no one. So everyone would go use the self-checkout machines, and stores would have no need for human cashiers at all.

But wait, you may be thinking. How could any store afford to give a 90% discount on groceries? The same way they give discounts for seniors and students at movie theaters. You just raise the sticker price, so that the price people actually pay after the discount is the same as before. So instead of keeping human-cashier prices the same and cutting prices for self-checkout, what stores would actually do if you made them give 10% discounts for self-checkout is to raise prices across the board — not by 10%, but by enough so that overall revenue is just about the same as it was before.

How do we know this is what will happen? Because the grocery store industry is very competitive. Profit margins are almost always between 1.5% and 2.8%, which is very low compared to the average industry:

It’s not just profit margins that demonstrate this — there are tons of grocery stores, and new ones enter the market all the time. This means that if grocery stores tried to respond to the New York bill by getting rid of self-checkout and using only human cashiers — in order to charge higher prices and make higher profits — other stores would undercut them by offering self-checkout with big discounts. People would flock to those other stores, and the stores that tried to overcharge would lose business until they, too, put in the self-checkout machines.

This is kind of like if you told grocery stores to charge 10% more for Skippy peanut butter than for Jif peanut butter. Obviously stores would want to stock more Skippy, because they’d earn more money on each sale. But obviously consumers would want to buy more Jif, because it was cheaper. At the end of the day, the consumers would win, because consumers tend to win out when an industry is highly competitive. Some people would still buy Skippy, because some people just really like Skippy more, but overall, forcing one brand of peanut butter to cost more than another brand means you’ll have less of it on the shelves.

By the exact same reasoning, forcing stores to charge more for purchases scanned by human cashiers will end up getting a lot of cashiers fired and replaced with self-checkout machines.

As policies go, this example isn’t particularly hard to explain, or to understand. If you ask any economist, or any good AI, they’ll give basically the same answer, in different words. It would be a good question on an undergrad microeconomics exam. But when I tried to explain this on X, some people had a lot of trouble understanding:

The idea of a 10% discount on self-checkout as a way to incentivize human employment is pure “slopulism” — stuff that seems like populism, but wouldn’t achieve the results that people want from the policy.1 It’s snake oil that sells well based on people’s inability to see the consequences of what they’re being sold.

Slopulism is an inevitable outgrowth of populism. Once people decide that technocrats, elites, experts, and gatekeepers aren’t worth listening to, they become vulnerable to all sorts of policy scams. In general, these scams fall into two categories:

1. Accidental scams, where the people making the policy just don’t understand what they’re doing

2. Ideological scams, where ideologues sell the public on an ideologically-driven policy by pretending it offers benefits that it doesn’t really offer

The second of these is more dangerous than the first. Slopulism that arises out of pure ignorance is very hard to correct, and social media — which elevates the dumbest and loudest at the expense of the most well-informed — probably makes it harder. But in the end, patience, education, and reason can overcome accidental slopulism — and AI may help a lot.

But when slopulist ideas are being promoted by smart people with ulterior motives, it can be a lot harder to wake the populace up. At that point, there’s the danger that slopulism becomes a bipartisan equilibrium, with each party’s ideologues executing a coup against the reasonable moderates by appealing to slopulist instincts.

Accidental versus ideological slopulism

Read more

Explaining stock market movements is always a little bit of a fool’s errand; no one really understands why stocks boom or crash on a given day or in a given week. Over the long run, the stock market displays lots of “excess volatility” — prices move up and down much more than is warranted by changes in earnings or other measures of fundamental value. There are plenty of theories about why those swings happen, but it’s very hard to know which of those — if any — is in operation at any given time. And so although every big stock market movement is followed by lots of articles claiming to know why, you should take them all — including this one — with several grains of salt.

Anyway, having said all that…

The Korean stock market has been crashing for weeks now. Around March, Korean stocks went on an epic tear; the KOSPI index rose from around 5,000 to over 9,000. Then, just over a month ago, it all went into reverse, with the index falling back to around 5,500:

There are probably two stories regarding why this happened — one about fundamentals, and another about finance. In fact, this is typical for bubbles and crashes, not just in stocks but in every asset class. There’s almost always some kind of connection to fundamentals — some story about how we’re in a new economy, followed by doubts about whether that story is really true, and so on. But the big market movements are almost always accelerated by purely financial factors — “noise traders” armed with piles of excess cash, opportunistic speculators looking to ride the wave of sentiment, and so on.1

For Korea, the fundamental story was about memory stocks. Korean companies like SK Hynix and Samsung make a lot of the world’s computer memory. Under normal circumstances, computer memory isn’t a great business to be in — the technology is fairly commoditized, the industry is brutally competitive, it takes a LOT of capital to build the factories, and it’s very risky to make long-term bets on the evolution of memory technology.

But computer memory is really important for AI data centers. And so the AI boom kicked off the mother of all memory booms. Only a few companies had the scale to meet a large amount of this demand explosion, and the two biggest of these were in South Korea. SK Hynix’s operating profit went from under $10 billion in the first quarter of 2025 to over $35 billion in the first quarter of 2026:

Hynix, which specializes in memory, briefly had a higher market capitalization than Samsung, simply because the memory boom is so huge. Korea’s exports rose over 70% in just one year; in fact, the country’s whole national GDP growth rate increased by a noticeable amount over the past two quarters, just because of this one product:

That’s a pretty strong fundamental story about why South Korean stocks should be worth a lot more. So it’s no surprise that the Korean stock market boomed as soon as people realized in early 2026 that AI technology is going to be extremely valuable. Here’s a thread about just how epic the runup in these companies’ stock prices was:

This is where the financial story rears its head, though. A bunch of traders saw this enormous price rise and decided to buy into it. You’d think a lot of these would be foreign, but international investors mostly avoided the boom (except for a few who bet big on Korean memory stocks). The most frenzied buyers were regular Korean people — the proverbial taxi drivers and teenagers.

These investors probably didn’t understand the fundamental story about AI and data centers and so on. Instead, they probably had extrapolative expectations — they see the price go up and up, and they figure stocks are just a “goer upper”. As more and more buy in and the stock goes up more and more, the perception of a structural upward trend is only reinforced, causing yet more people to buy in. This isn’t the only explanation for coordinated “noise trader” buying frenzies, but it’s probably the most likely.

Normal people don’t have a lot of cash sitting around. But earlier this year, regular Korean people got the opportunity to effectively borrow lots of money to invest it in stocks, via the introduction of leveraged single-stock ETFs. When you buy a share in a leveraged single-stock ETF in SK Hynix, it’s like borrowing money to buy SK Hynix stock.

A whole lot of Koreans used leveraged ETFs and other borrowing methods to borrow huge amounts of money and buy lots and lots of Korean stocks — especially the memory company stocks that were driving everything.

There were some people selling — notably, foreign investors “taking profits” and getting out. But the noise traders overwhelmed all the selling pressure, and sent stock prices soaring.

Then something happened last month — either something fundamental or something financial. Hynix and Samsung are doing fine in terms of earnings growth, but it’s possible that something suddenly gave traders reason to doubt the overall story about the AI boom sending these companies’ profits to the moon. The other possibility is that Korea simply ran out of hotheaded day traders willing to borrow more and more in order to bet on stocks, and the influx of cash naturally came to a halt.

Whichever it was, at that point the price faltered and began to fall. All that borrowed money accelerated the fall on the way down. When prices fall, leveraged ETFs have to sell some of what they hold.2 When a bunch of leveraged ETFs do this at the same time, it pushes prices down, forcing others to sell. In the meantime, people who had borrowed money to buy stock faced margin calls (or bankruptcy), forcing them to sell stock to raise cash. All of this created extra selling pressure, and so increased the rate at which Korean stock prices fell since late June.

Anyway, that’s the financial story. It’s a very old story — financial leverage plus unsophisticated new buyers plus a strong fundamental story often produces a bubble and crash, or exacerbates one that was already in progress. No wonder Korea is now moving — a little belatedly — to restrict leveraged ETFs.

But while financial factors affected the timing and the size of the stock price boom and bust, the fundamental story is more interesting. Fears of an AI bubble are quietly creeping back.

In 2025, as consumer chatbots struggled to find a market big enough to justify the kind of investments being made, there was a lot of talk about an AI bubble. One possibility was that AI revenues wouldn’t grow fast enough to justify the amount being invested in data centers. Another possibility was that AI companies wouldn’t have enough of a “moat” to make them consistently profitable.

In 2026, Claude Code exploded onto the scene, and everyone realized that AI had finally found “product-market fit”. We now know at least one thing that AI is incredibly useful for — writing computer code. Suddenly, an AI bubble seemed much less likely. Spending on AI was skyrocketing, and Anthropic — the new market leader — was successfully capturing much of the profits. Cybersecurity and other zero-sum applications — where having the absolute best model can matter a lot — started to seem like the “moat” that AI had previously lacked. Suddenly, the giant data center construction boom seemed a lot more reasonable.

But slowly, doubts have begun to creep back in. AI is amazing at writing software, but writing software is different from selling it. So far, huge increases in coding productivity are translating into only minor increases in the amount of software being shipped:

It’s possible that a significant fraction of the explosion in the use of coding agents is just “tokenmaxxing” — companies trying to use as much AI as they can, either to learn to use the tools, or perhaps just to look like they’re doing something. If so, we can expect a retrenchment and a temporary slowdown of AI spending growth.

It’s also possible that even the revenue growth we’ve seen isn’t enough to offset the enormous costs of the data center boom. Here’s a recent report from The Economist:

A back-of-the-envelope calculation finds that covering AI capex through identifiable AI income requires revenue on the order of $2.5trn per year, more than tech’s entire combined revenue today…Anthropic pulls in perhaps $75bn, annualised; OpenAI makes tens of billions; Google, via its AI model Gemini, and Microsoft probably get a bit less. SpaceX may have a few billion dollars’ worth of revenue from enterprise AI this year. Meta also makes a few bucks from AI. Add this up and you land at roughly $150bn a year. [emphasis mine]

AI revenue is growing very fast, but if these calculations are right, it’ll have to grow by 17x from where it is now in order to justify the capital being spent. In other words, even Claude Code isn’t enough; AI revenue has to accelerate even further, and while it’s perfectly plausible that it could do that, it’s a big question mark.

There’s also the possibility that the moat of companies like Anthropic is less invincible than it looked just a couple of months ago. A Chinese company called Moonshot AI has released a model called Kimi K3 that nearly equals the best available American models. American companies are using cheap Chinese AI models more and more for daily tasks. If Anthropic and OpenAI lose the overall b2b market to cheap Chinese competition — which is undoubtedly supported, of course, by China’s usual blizzard of subsidies and government supports — there’s not much chance that they’ll be able to pay for the data center boom.

And at that point, there could be a big, big bust — similar to when railroads went under in 1873. Investors are probably already beginning to worry about this. It isn’t just Korean AI-related stocks that have taken a hit recently. Here’s Nvidia, which designs and furnishes the chips that run data centers:

And here’s Micron, America’s top memory chip maker:

And here’s Microsoft, a big part of whose business is installing AI data centers:

There are similar (if less dramatic) stories at Google and Amazon, who also run a ton of data centers.

Here’s Bloomberg’s story about the decline of the so-called “Magnificent 7” tech stocks:

Wall Street is growing increasingly concerned about the hundreds of billions of dollars Big Tech is spending on artificial intelligence…“The real problem is the amount of spend that’s going on,” said Ken Mahoney, chief executive officer of Mahoney Asset Management. “No one knows what the return on investment is.”…The selloff is coming as investors grow increasingly cautious about the massive sums that Big Tech firms are spending to build out their AI infrastructure. The Mag 7 index is now down 11% from a record reached in late May, erasing $2 trillion in market value.

So although South Korea’s epic stock crash was probably related to Korea-specific financial factors, it could also herald the return of the “AI bubble” story. AI is far and away the most important thing going on in the American economy right now, so any hint of a bubble is worrying.

Update: One possibility I should have mentioned is that the Korean stock crash is the start of a general crash in AI stocks — the long-awaited “AI bubble pop”. So far the carnage hasn’t been too extensive, but it’s notable that some people who bet big on the smooth, uninterrupted exponential growth of AI are now seeing those bets blow up:

Citadel — a big traditional mainstream finance firm — is now stepping in to buy many of the stocks owned by Situational Awareness.

And as Derek Thompson noted a couple of weeks ago, Korean retail investors aren’t the only ones borrowing money to buy stocks:

Derek flags the rise of margin buying and leveraged ETFs in America, but I also noticed that some pretty big companies are leveraging themselves to the teeth here:

Subscribe now

Share

Hallo, awesome readers! It’s time for a roundup. I actually never think of a theme for these in advance. I just dump everything in my list of “stuff I don’t have time to write a full post about" into the Substack editor, then try to figure out some kind of thread that ties the items together. Today’s list is, roughly speaking, about industrial policy — deliberate policies that governments can take in order to encourage specific sectors or pieces of their economies.

But first, podcasts! I went on Scott Galloway’s Prof G Markets podcast, to talk about the macroeconomic implications of AI:

You can also watch a video if you prefer! Not sure why, but it looks like I haven’t shaved in a long time even though it’s only 12 hours of beard shadow.

I also went on Jayden Clark’s podcast, Members of Technical Staff, with Sam D’Amico (founder of Impulse Labs). We talked about San Francisco’s local politics, while cooking steak on an Impulse stove:1

Anyway, on to this week’s list of interesting stuff!

1. Is Trump getting better on science policy?

The Trump administration, generally speaking, has been the worst on science policy in the country’s entire history. As part of its effort to root out wokeness from American institutions, it has slashed science budgets. Thanks to its conviction that immigrants are foreign invaders intent on overthrowing Western civilization, it has blocked or scared away lots of the foreign researchers that American science depends on. And due to its bizarre political alliance with pseudo-leftist conspiracy theorists, it has elevated quasi-mystical antivaxxers to positions of authority over biomedical policy.

But at the same time, there are some people within the Trump administration who are trying to take things in a more positive direction. The Office of Science and Technology Policy recently released a report called “Science: A New Golden Age” that contains a number of interesting and promising ideas. Alec Stapp of the Institute for Progress wrote a quick summary of what he likes about the plan:

The most important proposal is probably the idea to shift funding from organizations to individual scientists:

Refocus on the Individual Scientist: Put the working researcher back at the center of America’s scientific enterprise. Free them from the growing administrative burdens that now weigh them down for nearly half their working hours. Bet on people, not just projects, by expanding portable graduate fellowships like the National Science Foundation (NSF) Graduate Research Fellowship Program (GRFP), backing early-career independence, and scaling long-horizon grants for the best and brightest modeled on National Institutes of Health (NIH) Director’s Pioneer Award. Open alternative pathways beyond standard academia, and ensure that selection rests purely on merit, not the political fashions of the day.

This is probably a very good idea on the merits. And you can also see how the OSTP folks are selling it to a skeptical Trump administration. Bypassing universities and funding researchers directly seems like it would naturally appeal to administration people who think universities are engines of woke ideology.

Another good idea is to diversify federal science funding mechanisms:

Move beyond consensus-driven peer review by adopting a broader menu of selection mechanisms suited to different kinds of science. Examples include “golden tickets” that empower individual reviewers to champion ambitious proposals, fast grants that deliver rapid funding decisions, prize challenges and advanced market commitments that pay for results, and regranting models that delegate funding authority to scientists to draw on distributed expertise…

Many of today’s most important problems are too large for an academic lab, too cross-disciplinary for a single department, and too hard to commercialize for a private corporation. Federal funding should support a wider range of performers. The recently launched X-Labs can assemble agile, time-bound teams of professional scientists and engineers to break specific bottlenecks. Advanced Research Projects Agencies (ARPAs) can empower individual program managers to make bold bets and curate researchers to execute them. Curiosity-driven institutes can give our best minds the stability needed to pursue fundamental questions over long time horizons.

Science is changing fast in the age of AI, and old funding models may not always be the best for every discipline. Also, notice that there’s a bit of industrial policy here — the mention of advanced market commitments suggests that Kratsios and the OSTP understand the intimate relationship between lab results and industrial scale-up in fields like chemistry and materials science.

Finally, OSTP wants to do metascience — they want to evaluate new funding methods as they’re implemented, to see what’s working and what isn’t:

Stand up an empowered metascience unit in federal science agencies, reporting directly to the director, with authority to run controlled experiments on review and funding mechanisms and to drive change across the organization.

I am in favor of all of these proposals. They’re very good, and most are long overdue. The problem is that they’re going to cost a lot of money. OSTP’s report isn’t a piece of legislation, or even a funding request; it’s just a plan, by one little piece of the government, that may or may not ever be implemented. Trump’s budgets still call for huge cuts to science funding, which would basically throw the OSTP plan right in the trashcan or (at best) turn it into a tiny symbolic demonstration program.

But these ideas are good enough that they need to be bipartisan. If Democrats take back the White House in 2028, they need to pick up this plan, slap a little token partisan gloss on it, maybe have Claude rewrite the language a little, and fully fund it.

2. America’s national debt is a big and growing problem

I’ve been shouting about the national debt for years now (including when Biden was President), so it’s good to see other people who aren’t fiscal perma-hawks shouting about it too. One of these people is Martha Gimbel, who worked for Biden’s Council of Economic Advisors. She writes that America’s national debt is already being reflected in higher long-term interest rates:

What should matter is that the consequences of this debt are not off in the future, but already here. The government’s deficits have saddled many American families with higher costs, largely from rising interest rates. The Budget Lab, the policy research center at Yale where I am the executive director, recently estimated that congressional-spending decisions since 2015 have raised Treasury yields by almost a full percentage point, which affects what American households pay to borrow. For someone taking out a 30-year mortgage at last year’s median home price, this rise in long-term interest rates has increased their borrowing costs by about $2,500 a year, or roughly $76,000 over the life of the loan.

The bloated government budgets and waning federal revenues of the past decade are driving up costs across the board. Compared with a world in which these fiscal-policy changes did not take place, the annual borrowing costs on a typical auto loan are now up by about $120, and by about $770 on a typical small-business loan. Credit-card borrowing rates are also hovering near record highs. [emphasis mine]

Gimbel is no one’s idea of a fiscal perma-hawk. But actual perma-hawks, like the people at the Committee for a Responsible Federal Budget, are naturally freaking out even more. Here is their illustration of how government debt can spiral out of control, when higher interest rates force the government to spend more even as they depress growth and reduce tax revenues, leading to even higher interest rates:

Relentlessly rising federal interest payments show why we’re in danger of this spiral right now.

Once this kind of spiral takes off, the only thing you can really do is print money, which can easily lead to horribly damaging inflation. A few macroprogressives on the left and die-hard supply-siders on the right still discount that possibility, but the experience of 2021-22 shows why Democrats and Republicans need to ignore those folks.

So what do we do? In fact, we have lots of levers to fix the debt. Here’s Gimbel again:

The main remedies for these problems—higher taxes and spending cuts—are generally politically unpopular. Every budget fix will have its critics, but some options are more palatable than others. Better funding for the IRS, for example, could help close the “tax gap”—the amount of taxes legally owed that are not paid in a timely way—which the IRS estimated at about $700 billion a year in 2022. Other levers include raising the retirement age and reducing Social Security benefits for high earners, who also tend to live longer; reforming Medicare Advantage, a program that has been shown to allow private insurers to overcharge the federal government; and removing the tax exemption on employer-provided health insurance, so that these benefits can be taxed as income. The Congressional Budget Office regularly publishes policies that could help close the deficit, and Americans need to decide what we’re willing to pay for and what we’re not.

It’ll take political will, but we managed to muster that political will back in the early 1990s. What we need, most urgently, is to destroy all vestiges of the bipartisan “deficits don’t matter” consensus that emerged in the 2010s and early 2020s. Yes, deficits DO matter.

3. Europe’s disdain for software comes back to bite it

Over the past decade or two, Europe has become increasingly hostile to the software industry. Although Europe does have a few successful software companies (SAP, Spotify), these are few and becoming fewer. Europe has generally chosen to regulate software rather than build it, enacting the free world’s most onerous software regulation — GDPR — in 2016. That regulation, which is why you have those useless, annoying “Accept cookies” popups on every website you visit, convinced European leaders that they can control the direction of software development by regulation alone — trading access to the European market in exchange for the ability of European leaders to have a say in how global software is made and run.

Why exactly Europe decided to regulate software instead of building it is a topic that deserves a deeper and longer discussion. It’s possible that European leaders just concluded, like Xi Jinping did in 2021, that software is a consumer toy that isn’t “real” technology — that it absorbs resources that are better used for manufacturing, that it doesn’t strengthen the nation or create real value, etc. It’s also possible that, like progressives in America, European leaders got negatively polarized against the “techbro” class — the industrialists who amassed vast personal fortunes figuring out new uses for software — and who used software regulation as a way to deprecate the social status of those industrialists.

But whatever it was, the European bias against software is looking more and more dangerous and self-defeating in the age of AI. AI is definitely not a toy, and not just a consumer good — it will be a crucial input into every manufacturing process on Earth, and a critical tool of national security. Countries that don’t have access to the best models may be at the mercy of those that do. They will be more vulnerable to cyberattacks, less capable of controlling drone armies (which now dominate the battlefield), and far less competitive in any export industry.

In other words, software matters now in a way it didn’t matter as much back when Europe decided it didn’t need the industry.

A group of European think-tankers and researchers have written a document called Europe 2031, illustrating the danger that their region is facing. It’s a bit long and dramatic, but it gets the point across. The report paints a vivid picture of what it would look like for Europe to become a de facto economic satellite state of the U.S. and China.

Avoiding this will require Europe to adopt a different approach toward AI than it did toward the consumer internet. Overregulation needs to be avoided, but that’s not enough. As the authors of Europe 2031 note, simple NIMBYism — similar to the kind that paralyzes the U.S. — will be sufficient to stop Europe from building data centers. European leaders aren’t just going to have to allow the AI industry to develop — they’re going to have to overcome their ambivalence toward software technology enough to fight entrenched NIMBY interests to get data centers built.

Fortunately, the Europeans appear to understand the gravity of the situation. But modern Europe has a long and storied history of issuing rhetoric and proclamations about the need for policy change, but not making anything happen. Basically, Europe needs to re-learn the political art of making things happen, because the era when it could coast on the legacy of post-WW2 industrialization has already come to an end.

4. Why doesn’t America have good trains?

The U.S. notoriously has one of the worst rail transit networks in the world. This is partly because building trains is so incredibly expensive in America. The Transit Costs Project did a deep dive into figuring out why, and concluded that there are a bunch of different factors — bad procurement procedures, lack of standardization, and so on. Basically, my interpretation of this report is that American cities and states decided that trains weren’t that important, which prompted them to A) gut their train-related state capacity, B) allow a bunch of essentially parasitic consulting firms, contractors, and unions to dominate what rail budgets did exist, and C) ignore the need to streamline regulations that prevented rail from getting built. If we started caring about trains more, we could start fixing those problems.

But it’s also notable that trains aren’t equally expensive everywhere in America. The Boyd Institute produced a good chart of the Transit Costs Project’s data, showing the dispersion of costs per mile of tunneled train in various countries:

This is a logarithmic chart, so the cost differences are very big. The U.S. cost per mile is 4x that of the median rich country. But note that in median terms, we’re barely more expensive than Singapore, and we’re actually cheaper than Hong Kong — both of which have great rail networks. If we wanted to, we could build great trains too, like Singapore and Hong Kong, despite the costs.

But we don’t. To me, this drives home the fact that our cost problem has a feedback loop with our unwillingness to build. Yes, high costs make us less willing to build trains, but not building trains keeps costs high. If we’re going to get better rail in America, we can’t only focus on pure cost factors like station standardization and procurement processes (though these are important). We’ve got to want to build more trains! That probably means things like denser development patterns and less crime and disorder on trains.

But the chart also shows something else — a few train systems, like NYC’s, cost an absolutely enormous amount. NYC is already train-dependent, so their problem isn’t a lack of trains — it’s the inefficiency, corruption, and incompetence they’ve allowed to become endemic in their system. So in NYC, the problem really is just fixing the system. The Transit Costs Project does have a new report on how to do exactly that, but it’ll require New Yorkers to get very mad at their government and demand change.

5. China’s choke-hold on rare earths

“Critical minerals” are metals that are very important for modern technology — traditional industrial metals like aluminum, nickel, and cobalt, rare earths like neodymium and yttrium, and others like gallium and germanium. For years now, American leaders have understood that our country is in grave danger from China cutting off our access to many of these critical minerals.

But a lot of Americans seem to still misunderstand why that danger exists. A lot of people seem to think that it’s all about mining the minerals, and that if we just A) discovered more deposits, and B) loosened regulation enough to allow mines to be constructed, America would have ample supply of these metals. This is wrong and misinformed. The reason China controls our supply of critical minerals is not because we don’t mine these metals, it’s because we don’t refine them.

When you dig metal up out of the ground, it’s not in a form you can use in industrial applications. To turn it from ore into something usable takes a bunch of steps — mostly, dunking it in various baths of chemicals. This is called refining.

Most critical mineral refining is done in China. Every year the IEA comes out with a report on critical minerals. Here’s a chart from this year’s report that shows how dominant China is in refining, even for things like cobalt that China mines very little of:

In fact, China is becoming even more dominant in minerals like lithium and manganese. This is despite the fact that other than graphite and gallium, China doesn’t dominate the mining of any of these metals.

Why does China dominate refining of critical minerals? A few reasons:

1. China dominates the downstream industries in which some of these minerals are used. For example, China makes most of the world’s lithium-ion batteries, so it makes economic sense to locate lithium refining near to the Chinese factories.

2. China’s financial system gives a huge amount of cheap bank loans to companies that refine critical minerals, making it very cheap to finance these expensive plants.

3. China has somewhat looser environmental restrictions than Western countries.

4. China has been investing heavily in metal refining for decades, meaning that they’ve built up a tremendous amount of distributed tacit knowledge in difficult refining processes (such as rare earth refining).

If the U.S. and its allies want to get out from under the Chinese yoke, they will need to address all of these shortcomings at once. They’ll need to promote downstream industries like batteries (which Trump has discouraged and attacked) and electronics. They’ll need to revamp their financial systems to allow big loans for refineries. They’ll need to loosen environmental restrictions in targeted, judicious ways. And they’ll probably need to use AI to quickly reproduce the distributed tacit knowledge that makes Chinese refineries so efficient.

6. India’s women need to work more

One big barrier to Indian industrialization is that women in India haven’t gone to work en masse. In most countries, industrialization starts when a ton of women move off of farms into cities and work in factories doing “light industry” — clothing, fabrics, toys, electronics assembly, and so on. This creates manufacturing companies with the capital and know-how to move up the value chain into more valuable and capital-intensive parts of manufacturing. Eventually the wealth created by high-value manufacturing creates the demand for high-value services industries.

That’s the typical story, but in India it isn’t working yet. This isn’t because factory automation has made labor-intensive “light industry” irrelevant — witness Bangladesh, which has grown a bit faster than India since 2004, largely on the back of its garment industry. Bangladesh has a relatively low female labor force participation rate, but it has increased its rate steadily over the decades. India has not; women in India are much less likely to work outside the home than women in other developing countries throughout the region:

Goldman Sachs analysts believe that getting Indian women out of the house and into the factory represents a huge economic opportunity. But Indian society seems to be resisting this.

A new paper by Alison Andrew and Andrea Smurra helps us shed a little bit of light on what’s going on. Low-income Indian women are increasingly not even leaving their homes:

We use rich time-use data on where, and how, individuals spend their time to quantify and describe women’s seclusion in India. We document extremely high levels of seclusion with the median married woman leaving home for just 0.5 hours/day and 45% not leaving home at all on a given day. Seclusion has increased markedly over the past two decades. While richer and more-educated women were, and still are, the most secluded on average, this gradient has flattened over time….Women are more likely than men to specialize in activities that are more readily done at home but are also more likely to carry out any given activity at home.

Why are Indian women becoming such hikikomori? Why hasn’t the lure of factory jobs been enough to draw them out of the countryside to the bright lights of the big city, as it was in Bangladesh, Vietnam, China, and a bunch of other countries? Andrew and Smurra don’t even hazard a guess. It’s not because they’re busy having lots of kids; Indian fertility rates are low and falling. Part of it might be simple lack of awareness of economic opportunities in factories. But whatever it is, it presents a big economic problem for India.

7. No, murder is not secretly really high

A lot of people like to have nostalgia for the 1950s. In fact, measuring by the murder rate, America is actually less violent now than it was back in the 50s (after a brief spike in the late 2010s and early 2020s):

People who think of the 1950s as being very safe tend to instinctively resist this conclusion; they think something must be wrong with the data. And so many latch on to the myth that murder rates have only fallen because of improved trauma care in American hospitals.

But it’s just not true. In a recent blog post, Jeff Asher explained how this myth got started, and how we know the narrative is wrong:

Jeff-alytics

Is Murder Falling Because of Improved Trauma Care?

My first writing for a national audience was in the Fall of 2015 with FiveThirtyEight (RIP). My first piece there was about police department open data, but my second piece was titled Murder Rates Don’t Tell Us Everything About Gun Violence. Thanks for reading Jeff-alytics! Subscribe for free to receive new posts and support my work…

Read more

6 months ago · 47 likes · 24 comments · Jeff Asher

Basically, the myth rests on a 2002 paper by Harris et al., claiming to show that Americans are getting shot more and more often, but are surviving the shootings much more, leading to a decline in death but a rise in violence. Here’s their key chart:

So what’s wrong with this chart? The data on “lethality” isn’t official data — it’s just the authors dividing the aggravated assault rate by the murder rate.2 And the aggravated assault data they use is bad. They use the aggravated assault rate as reported by police. But as Asher explains in his post, American police forces got a lot more professional and thorough throughout the late 20th century. If someone hit you with a bat in 1955, there was a good chance you wouldn’t call the cops, or they wouldn’t record the incident. By 1995, there was a much better chance.

In fact, we can just go and ask people if they’ve been the victims of a shooting or an assault. When we do that, we can easily see that even as police-reported shootings rose in the 70s and 80s, shootings reported by victims didn’t rise:

So did aggravated assaults really rise in the 60s, 70s, and 80s? Almost certainly not. So aggravated assaults were probably not becoming less lethal over that time period. Yes, we had better trauma care, but we also had a lot better guns and a lot more dedicated murderers.3

In other words, America today really is about as safe as 1950s America. Sure, violence in the 50s was concentrated in poor and Black neighborhoods, but that’s true today too.4 What really changed in America since the 50s isn’t the lethal sort of violence, it’s urban decay and disorder. That’s what we need to focus on fixing.

Subscribe now

Share

“[I]ntelligence, that counterentropic conjoined twin of information, must become the most powerful force in the universe, the energy to which all other physical laws must eventually kneel…Intelligence was destiny, manifest.” — Ian McDonald, “Verthandi’s Ring”

Not a lot of people expected that AI would come for the mathematicians before it came for the truck drivers, but it did. The other day, an AI model disproved the Jacobian Conjecture — an 87-year-old open problem that human mathematicians had struggled to solve. The greatest living human mathematician, Terence Tao, turned to AI to help him understand the solution. Around the same time, AI solved a very important open question in quantum cryptography. Solving Erdos problems has now become almost child’s play for the best AI models. And this is the worst AI will ever be at math. Model capabilities, and the amount of compute available, both continue to increase at rapid rates. (Meanwhile, long-distance trucking employment is slightly higher than it was a decade ago.)

I don’t expect mathematicians to actually lose their jobs en masse, of course.1 But it’s becoming clearer and clearer that humankind has invented machines that are smarter than we are. Intelligence isn’t defined for machines the same way it is for humans — AI’s capabilities are spiky in different ways than ours — but it’s undeniable that the technology is improving rapidly in every domain of cognitive capability. It’s still possible to find some mental tasks that humans are better than machines at, but those final advantages tend to disappear almost as quickly as we can identify them. “AGI”, or “ASI”, or whatever you want to call it, is certainly here.

And yet…the world remains much the same. In lots of sci-fi books, as soon as artificial superintelligence arrives, it bootstraps itself to even more godlike intelligence in an explosive “singularity” that rapidly transforms the entire physical universe. Lots of people, especially “AI safety” and “effective altruist” types, expected things to play out basically the same way in reality. But looking around, not much has changed since we entered the intelligence explosion. There’s a huge data center boom, and most people use AI on a daily basis, but we still live basically the same lives — driving to work or taking the train, sitting in front of a computer, scrolling on our phones, collecting a paycheck. People are staying in their jobs longer, but employment hasn’t been disrupted in a significant way:

Meanwhile, we’ve had decently robust productivity growth, but nothing really amazing:

A lot of people I know are surprised by this. Ruxandra Teslo writes:

Walking around the world today one might notice that it is weirdly unchanged…To many, this is surprising. Just the other day I was at a conference where someone remarked that if he could have seen today’s AI capabilities a few years ago, he would have been astonished — and would have assumed the world by now would look far more transformed, with much higher GDP growth.

And Clifford Sosin writes:2

Superintelligence arrived. You probably didn’t notice, because it turned out to be kind of incremental…Don’t believe me? Run the test. Talk to Fable 5 for an hour, then talk to your ten smartest friends. Which one is smarter? Don’t worry, they won’t be offended…We were told to expect something bigger. Once machines crossed some line, the system’s IQ would climb to heights we couldn’t follow, and we’d be sharing the planet with something that designs warp drives and thinks thoughts as far past us as mine are past my dog…What we got is a tool that writes excellent code, beats people at a startling range of tasks, and will clearly reshape the economy. It’s also, somehow, incremental. No takeoff. No explosion. That’s the strange part.

Teslo blames bottlenecks — governance and other “frictions” — for the slow economic impact. But some others are advancing a more radical hypothesis3 — that intelligence itself is subject to diminishing returns.

One of these is Francois Chollet, an AI researcher who specializes in measuring AI’s capabilities. In a highly controversial series of tweets back in March, he conjectured that intelligence might be subject to diminishing returns:

One of the biggest misconceptions people have about intelligence is seeing it as some kind of unbounded scalar stat, like height. "Future AI will have 10,000 IQ", that sort of thing. Intelligence is a conversion ratio, with an optimality bound. Increasing intelligence is not so much like "making the tower taller", it's more like "making the ball rounder". At some point it's already pretty damn spherical and any improvement is marginal.

Now of course smart humans aren't quite at the optimal bound yet on an individual level, and machines will have many advantages besides intelligence -- mostly the removal of biological bottlenecks: greater processing speed, unlimited working memory, unlimited memory with perfect recall... but these are mostly things humans can also access through externalized cognitive tools.

In fact, this is a possibility I myself had raised in a post a year earlier:

It seems possible that humans are simply incredibly specialized in a few types of cognitive tasks — extracting patterns from sparse data, synthesizing various patterns into “intuition” and “judgement”, and communicating those patterns in language — and that we’ve basically approached the theoretical maximum in those narrow areas…That would explain why AI has gotten much better at things like math and coding and forecasting over the last year, but why the basic chatbot interface doesn’t seem much more “intelligent”. It would also explain why when you talk to Terence Tao about math, it’s like talking to a superhuman, but when you talk to him about where to get lunch or which movies are the best, he’ll just sound like a fairly smart normal dude. AI will eventually get better than Tao at math…but it may never get much better than the most thoughtful, eloquent humans at deciding where to get lunch or recommending movies. It may simply not be mathematically possible to get much better than we already are at that sort of thing.

Why would intelligence top out like this? Well, if we think of intelligence as the ability to extract information from data, then even an infinitely advanced model endowed with infinite compute will be limited by the fact that there’s a limited amount of information that can be extracted from the data.

For one thing, data itself is in limited supply. You can’t transform the world unless you can (in some generalized sense) understand it, and you can’t understand the world unless you can measure it, and our ability to measure the world is inherently limited and finite.

This is basically the hypothesis advanced by Arvind Narayanan and Sayash Kapoor:

We think there are relatively few real-world cognitive tasks in which human limitations are so telling that AI is able to blow past human performance (as AI does in chess). In many other areas, including some that are associated with prominent hopes and fears about AI performance, we think there is a high “irreducible error”—unavoidable error due to the inherent stochasticity of the phenomenon—and human performance is essentially near that limit….We predict that AI will not be able to meaningfully outperform trained humans (particularly teams of humans and especially if augmented with simple automated tools) at forecasting geopolitical events (say elections).[emphasis mine]

Sosin says something similar:

We hold a thin scatter of facts about the world, and intelligence or reasoning is whatever fills the space between them…Where the space between the facts behaves well, this is close to godlike…Coding, math and most administrative work are [like this]. What makes them easy is that they have relatively smooth solution spaces and are tractably verifiable…Most of what matters doesn't behave like that. The universe is mostly the emergent behavior of complex systems…The limit is contact with reality. A smarter reasoner fills the gaps between known facts in simple areas faster and better, but it doesn't produce new facts.

Sosin makes an important point here, which is that the limitations of intelligence aren’t necessarily about limited data. Even if we can keep on collecting infinite data, the cost of extracting additional information from that data might explode to infinity. This is the idea of chaos. Even in a deterministic universe where the present states of all particles are enough to perfectly determine the future, our ability to predict the future can be inherently limited; the tiniest infinitesimal error in our measurement of the present explodes into a huge error when we attempt to extrapolate even a small distance into the future.

So although we don’t know yet, it’s possible that humans were already hitting the point of diminishing returns with regards to individual cognitive capacity, and that superintelligent machines will never be as far beyond us as we are beyond dogs. But even if that’s true, I can think of at least three reasons why machine superintelligence could still deliver huge productivity gains.

Smart matter

The most obvious advantage that machine superintelligence confers is replicability. The number of human intelligences is fixed by the fertility rate, and we don’t know how to substantially boost that rate; in fact, it’s falling inexorably, and the human race is set to shrink.

AI isn’t bound by those limitations. By building more data centers with more compute, you can run more agents in parallel — essentially, you get more cognitive work. It’s not free, but nor is it limited. It’s good old physical capital; unlike human capital, you can just build more of it whenever you like, simply by reinvesting some portion of your economic output. Imagine if we suddenly discovered a way to manufacture more land in any city on the planet; this is similar.4

Of course, lots of tasks are physical ones, and for these you need physical machines — basically, robots. This is probably why so many AI people are now working on robotics, world models, physical AI, and so on. The roboticization of the world has a huge tailwind — the battery revolution, which allows energy to be stored and moved around much more easily. Conveniently, we got the physical tools to turn dumb matter into smart matter just as we also got the digital tools.

(It’s also worth noting that like the energy in batteries, the intelligence in robots is divisible. A robot the size of an ant can be remotely controlled by a data center the size of a football field. This is also a capability that human intelligence lacks.)

This doesn’t mean economic output will explode to infinity. But what it does mean is that humans will be able to use physical capital — GPUs and robots — to do more and more tasks at once, including many cognitive tasks that we used to do the hard way. In the long-run steady state, this should increase the capital-to-labor ratio of our society; each human will essentially leverage an army of intelligent machines. It’s basically another industrial revolution, and it has very little to do with whether artificial intelligence is smarter than human intelligence in any sort of head-to-head matchup.

Distributed tacit knowledge

The German company Zeiss makes the best glass on the planet. If one of the mirrors that Zeiss makes for ASML’s EUV chipmaking machines were the size of Germany, the biggest bump on that mirror would be just one millimeter high. Only a few other companies — and maybe no other company on Earth — can match that. Zeiss’ mirrors also have a number of other amazing properties, like not distorting much due to temperature changes.

How does Zeiss make glass this good? No one knows — not even the people at Zeiss. If the technology were capable of being written down on a blueprint, China would have hacked Zeiss and stolen it, the way Huawei hacked Cisco and Nortel. If the technology were capable of being explained by a former Zeiss employee, or even several former Zeiss employees, China would have paid those people many millions of dollars to spill the beans.

Zeiss’ technology basically can’t be stolen, because it’s tacit and distributed. It consists of a vast number of little tricks and techniques that a huge number of individual employees use on a daily basis. These people don’t always even realize all those little things they’re doing that make the glass come out so good. And each employee knows a different set of tricks and techniques. The knowledge exists at the level of the organization itself, and is thus very hard to steal or recreate.

This is true of lots of corporate technology. A big part of the reason China can cut off the supply of rare earths to the rest of the world any time it wants to is that other countries aren’t very good at refining rare earths. Rare earths are difficult to separate from each other in solutions; it takes a ton of little chemistry tricks to do it cheaply at scale. Chinese refiners have spent four decades building up those little tricks and techniques; American or Japanese refiners won’t simply be able to replicate their efficiency overnight, and so it’ll continue to cost much more to produce rare earths outside China.

Except in the age of AI, this might change. Suppose American rare earth refiners give their employees a bunch of equipment to record everything they do — smart glasses, gloves, and so on — in addition to sensors distributed throughout their plants. AI will be able to synthesize all that information and very rapidly suggest small ways to improve the production process. Many of those little experiments will fail; others will succeed and will quickly be adopted, allowing another round of experimentation and improvement to begin very quickly. Crucially, AI’s ability to do this doesn’t depend on its raw intelligence — only on its ability to handle huge amounts of data very quickly.

In other words, in the age of AI, distributed tacit knowledge might not be nearly as big of a barrier to technological diffusion. This could improve economy-wide productivity, as lagging firms catch up to leading firms much more quickly. A more equal distribution of productivity would also make the economy more competitive, creating more surplus for consumers (though possibly reducing the incentive for firms to innovate, by making technology less excludable).

AI’s ability to quickly produce distributed tacit process knowledge might also supercharge productivity growth at the frontier. Imagine if any company could optimize any production process five times faster than today. The whole economy would speed up, as components got cheaper, turnaround times and product cycles got shorter, and scale-up got much faster.

And as with the previous example, improving the production of distributed tacit knowledge wouldn’t depend on AI’s raw intelligence. It would spring from AI’s ability to act like a computer — to interface directly with sensors, to handle lots of data, to perceive tiny details, and to do everything very very quickly.

Cloud laws

For decades, researchers in the field of natural language processing tried to figure out the principles behind human linguistic communication. They made frustratingly little progress; the processes by which humans convey information to each other through words just don’t seem to obey simple laws, like the ones that govern electromagnetism or the circulatory system.

Then along came AI, and suddenly linguistic communication seemed like a solved problem. LLMs can reliably sound like a human being, even if we don’t understand how they manage to do it.

What if there are lots of other aspects of the Universe that work the same way — too complex to understand in terms of simple laws, but not so complex that they just dissolve into unknowable chaos? It’s possible that we can reliably control these complex phenomena with AI, even if we never reduce them to the kind of principles that we can teach a grad student in a textbook. In fact, I wrote an essay called “The Third Magic”, where I suggested that this might be equivalent to a whole new scientific revolution:

Another way of saying this is that there may be laws of the universe that humans can’t understand but AI can. I call these “cloud laws” — causal regularities that can be exploited by technology, but which are too diffuse and complex for an individual human being to either intuit or communicate.

Human language seems to obey cloud laws, so why not other phenomena too? Perhaps social sciences like economics, sociology, and political science obey similarly complex regularities, and AI can help us find them. Perhaps there are physical processes — plasma, or topological materials, or aerial turbulence, etc. — that obey cloud laws instead of chaos?

In other words, thanks to AI, we might be on the precipice of a new age of scientific advancement. And this won’t necessarily depend on how smart AI is in comparison to a single human; it’ll depend on its computer-like ability to hold huge amounts of data in its working memory and extract complex patterns from that data.

If this turns out to be true, it means Francois Chollet is wrong. Chollet hypothesizes that groups of humans, using pre-AI computing tools, can approach AI’s level of scientific competence. But human collaboration is bottlenecked — it’s limited by our ability to intuit patterns at an individual level, and to communicate these patterns from one individual to another. AI, being a computer, just doesn’t have this sort of limitation; it can work with vast, diffuse patterns without having to break them into pieces or simplify them in order to compress them into the tiny pipelines of person-to-person explanation.

So even if AI never gets much better than humans at the kind of science that humans have done heretofore, it might open up whole realms of scientific discovery that have previously been totally inaccessible to even the largest groups of the smartest humans. If much of the Universe turns out to be ruled by cloud laws, we could be on the precipice of a scientific renaissance.

The common thread in all three of these examples is that AI may revolutionize productivity not by being much smarter than a single individual human — not by simply solving harder and harder math problems — but by marrying human-style intelligence to the vast, inhuman capabilities of computers. We could simply be thinking about the benefits of intelligence wrong — arrogantly privileging the kind of mental tasks we humans happen to do especially well, while ignoring the value of the tasks we do poorly.

Subscribe now

Share

The story of the rise and possible impending fall of the American university system is, in many ways, the story of modern America. It ties together the changes in our culture, our economy, and our politics since the mid-20th century. Understanding why universities came to be our most important and most functional institution, and why that model is now under threat, can help us understand how our country might look different going forward.

Let me try to tell a condensed version of that story.

The United States used to have a bunch of institutions that bound us together and forged us into a unified society. Many American communities were centered around churches; these facilitated networking, helped people find spouses, provided community services like day care and mutual aid, and homogenized values and culture at the local level. During the World Wars (and to a lesser extent, the Vietnam War), the military was very large and threw together Americans from various social classes. In the early postwar decades, corporations were also a unifying institution. Mass media provided us with shared cultural context. Even public transit put people of various backgrounds in close social contact on a daily basis.

In the half-century from around 1970 to 2020, those unifying institutions became much weaker. Church attendance slowly declined and then fell off a cliff in the 2010s. The military shrank into a small, professionalized volunteer force. Corporations ended their brief flirtation with lifetime employment, and outsourced many of their roles. Mass media fragmented in the age of the internet. Public transit dwindled in importance as Americans moved to the suburbs and drove.

As the country’s unifying institutions withered, one new institution attempted to step into the void: the American university. Universities were not new in the late 20th century, of course, but mass attendance certainly was. From 1970 to 2020, the share of Americans age 25-29 with a bachelor’s degree went from 16.4% to 39.6%. Around two out of three have completed some college.

College went from something that only the upper crust did, to something that most people were expected to do if they wanted to be economically successful in life. This expansion inadvertently but inevitably thrust a new role on American universities — that of a broad socially unifying institution. College was where people from a variety of backgrounds mingled and mixed — not just in classes, but in dorms, campus activities, and college towns.

College became the new church, the new military, and the neighborhood bowling alley all rolled into one. Instead of preachers homogenizing Americans’ values from the pulpit, university administrators taught college kids to value things like diversity, consent, and so on, and college students hashed out their own differences in millions of late-night dorm room discussions. Instead of meeting their first love in high school, Americans increasingly delayed sex until college; many of these relationships turned into marriages. Young people went into college as children and came out adults.

This was a heavy burden to bear for an institution that hadn’t been designed for it. But American universities were accustomed to taking on big new duties. Our universities began as essentially a copy of the British model — teaching-focused institutions designed to provide broad education and mentorship to the upper class. In the early 20th century they tacked on the German model — a research-focused lab apprenticeship system by which top scientists taught other top scientists and readied them for research jobs while also producing cutting-edge basic research.

This dual system enabled a remarkable form of cross-subsidization. Undergrad tuition payments — and state support for undergrad education, and undergrad alumni gifts — created a flood of money that paid for grad student stipends, lab facilities, professor salaries, and more. The federal government and companies also funded university research, of course, through grants and sponsorships. But undergrad money was a huge tailwind. And the prestige generated by successful and famous researchers helped universities charge undergrads more.

The hybrid of the old British and German models was naturally symbiotic, and it meant that even as America’s other institutions came under pressure, universities thrived and grew — especially once they used their prestige to attract high-paying and highly skilled foreign students from around the globe. Universities became the lynchpin of America’s research effort:

American universities had a lock on both the nation’s research output and on its production of human capital, and those roles were mutually reinforcing.

I suspect that their success at handling education and research at the same time probably gave American universities a lot of confidence about their ability to handle society and culture as well. Colleges spent more and more on dorms and “student services” and hired administrators (many of whom dealt with undergrad life) at an astonishing rate.1

But replacing churches, the military, and the neighborhood bowling alley proved harder than replacing the corporate lab had been. There was just one basic problem with college as America’s primary unifying institution, which is that not everyone can go to college.

First of all, college is difficult. To complete college courses, you need some degree of raw intelligence, but you also need work ethic and a certain degree of independence. As much as we might like to believe otherwise, not everyone in America has those traits, and we don’t yet know how to instill them in everyone. As a result, there’s a limit to how much you can expand college enrollment — and college completion — without loosening standards.

In fact, loosening standards is exactly what American colleges have done. Universities have necessarily become less and less selective over the years, as they have taken in a larger and larger fraction of the young American population. For a while, this led to lower college completion rates, but in the 1990s, more and more students began finishing their bachelors’ degrees. Why? Because colleges implemented grade inflation, making it easier to finish school without learning the material well. Here’s Denning et al. (2021):

We find that most of the increase in graduation rates can be explained by grade inflation, and that other factors such as changing student characteristics and institutional resources play little or no role. This is because GPA strongly predicts graduation and that GPAs have been rising since the 1990s. This finding holds in national survey data and in records from 9 large public universities. We also find that at a public liberal arts college, grades increased holding performance on identical exams fixed.

At some point, though, this process hits a wall. A large fraction of young Americans just isn’t prepared for college, even with grade inflation, and doesn’t end up going. College enrollment by recent high school graduates plateaued in the early 2000s and actually fell back to early 1990s levels during and after the pandemic:

Read more

Let’s take a break from discussing the insanity of American politics, the tragedy of British economic mismanagement, and other weighty topics, and talk about something fun and light. Who was the greatest monster of the 20th century?

Well, ok, that’s not really fun and light. The 20th century featured hundreds of millions of people slaughtered in wars, genocides, political persecutions, and preventable famines. Many of my own extended family members were killed in the Holocaust, and I have plenty of friends whose relatives died in World War 2, the Great Leap Forward, the Cultural Revolution, the Cambodian Genocide, and other such calamities. Those deaths are even more tragic because of all the good things that were happening in that century — the fabulous economic growth and technological progress. Everything could have just been a smooth upward glide to luxury and comfort, and yet people were still being machine-gunned into mass graves.

With time, the emotional power of those events fades — Genghis Khan killed tens of millions in his campaigns of conquest eight hundred years ago, but today he’s more likely to be invoked as a joke, or an object of reverence, than as a monster. It can happen surprisingly fast. By the early 2000s, hipsters in my dorm were already hanging ironic Chairman Mao posters in their rooms. Today, you can see leftist pundits like Hasan Piker praising Mao without a trace of irony:

Plenty of other leftists joined in the Mao lovefest.

Piker is wrong, of course; if you read the authoritative biography of Mao by Jung Chang and Jon Halliday, you’ll see that the dictator repeatedly aggrandized himself at China’s expense, willfully derailing multiple periods of reform and recovery by launching the country into destructive episodes of chaos like the Great Leap Forward and the Cultural Revolution. Nor was he much help against the Japanese invaders in World War 2, preferring to let them fight and weaken his Chinese rivals (again, at the expense of many Chinese lives). It was Deng Xiaoping, not Mao, who made modern China great.

But the people who say that Mao was history’s greatest tyrant, with an unsurpassed body count of 40-50 million people, are also probably overreaching a bit. Yes, about 36-40 million people did die in Mao’s Great Famine, caused by his various foolhardy agricultural policies and the inefficiencies of China’s communist economic system. Nor was this merely an accident; even after Mao learned that his policies were failing, he kept them in place to avoid losing face in front of CCP rivals. But killing tens of millions of his own people was never Mao’s plan.

Intent is an important factor to consider when ranking history’s greatest monsters. The reason has to do with why we should care about the question. The point of ranking evil regimes isn’t to have fun marveling at the scale of human brutality — it’s not like trying to figure out what the biggest dinosaur was, or whether Michael Jordan was better than LeBron. Sometimes real life forces you to choose between monsters.

In World War 2, America had to decide whether allying with Stalin against Hitler was worth it, or whether we should simply sit it out and be neutral as the two villains slugged it out. Our decision probably tipped the course of the war. Later, we formed a quasi-alliance with Mao’s China against the Soviet Union, helping to hasten the latter’s downfall.

Many populations, of course, have an even grimmer choice — they have to choose which ruthless, cruel regime to support in a civil war. If you lived in Syria in 2016, did you support Bashar al-Assad, ISIS, or offshoots of al Qaeda? If you lived in Russia in 1919, did you support the Reds or the Whites? And so on.

So when we ask “Who was worse”, the question is actually pretty relevant. Obviously it’s a matter of opinion, but I think some factors we should consider when assessing the horribleness of a regime are:

1. Overall death toll during the regime’s time in power

2. How much of the death toll was intentional

3. How much of the death toll was willful versus accidental

4. How harmful the regime was relative to the size of the population it controlled

5. What other bad things the regime planned to do but was unable to accomplish

6. What life was like for the people who lived under the regime and didn’t die

So with that in mind, here’s my ranking for last century:

The worst regimes of the 20th century

#1: Adolf Hitler and the Nazis

For me, Hitler easily takes the top spot, and it isn’t really a question. The reason isn’t that the Jewish Holocaust — Hitler’s most-discussed atrocity — was uniquely horrible compared to other acts of genocide. It’s that the Jewish Holocaust was only the beginning of Hitler’s wave of destruction.

The slaughter of 5-6 million Jews was deliberate, but it was only one part of Hitler’s plans for genocide in Europe and Asia. Generalplan Ost, the Nazis’ plan for East Europe, involved the deliberate slaughter or enslavement of tens of millions of people in Russia and other countries to Germany’s east. Here’s a chart from Wikipedia:

“Removed” here means “killed”, since the Nazis intended to take all of the land.

Hitler’s invasion of the USSR — Operation Barbarossa — was meant to carry out this unprecedented genocide. The main tool was starvation — Nazi Germany’s “Hunger Plan” called for intentional mass famines to wipe out most of the people living in the USSR and East Europe.

The Nazis actually implemented this plan to a significant degree. Out of the 27 million Soviet citizens killed by the Nazis during World War 2, 19 million were civilians, and most of these probably died from hunger and disease, caused in large part by the Nazis’ deliberate policy of mass starvation during the war. The Nazis also mass-executed or starved millions of Soviet POWs. Had the U.S. not stepped in to help, the death toll in the USSR might have been twice as large, or more. Those that weren’t killed were to be enslaved, but given the experience of the “workers” at Auschwitz — and considering that Nazi ideology stated that superior races lived and inferior ones died off — it’s an open question how long those slave populations would have been allowed to survive.

And why would Hitler have stopped there? Having overcome the Soviets, would the Nazis really have allowed the Chinese, or the Arabs, to live their lives and hold their lands unmolested? Not a chance. Hitler’s regime was purpose-built for conquest and genocide — it could only keep killing or die. In this sense, the Holocaust, as horrifying as it was, was only the tip of the iceberg.

It’s not clear whether there has ever been another regime quite like Hitler’s, even when we look back to the barbarity of premodern times. Genghis Khan, Timur (Tamerlane), and other conquerors killed millions, but only in order to subdue and rule, not to annihilate. Other regimes have attempted annihilation of subject peoples, but have generally done this within their own borders rather than trying to do it to the rest of the world.

Hitler and the Nazis should thus stand as the undisputed all-time champion of evil. With an intentional body count of over 30 million, and plans to do even more, they were less like other historical tyrants and more like some kind of alien armada. In my mind, the decision to ally with Stalin in order to defeat that unique menace was unquestionably the right one.

#2: Fascist Japan

The Japanese Empire in the 1930s and 1940s was very different than the empire that came before. The military usurped civilian rule, put an end to the democracy that Japanese people had enjoyed in the 1920s, and made their society increasingly totalitarian. Japan had conquered and colonized before — Taiwan and Korea — but in the 1930s and 40s, they became increasingly brutal.

By the time they went after all of China in 1937 and Southeast Asia in 1941, genocide had become their standard tool of conquest. Like the Mongols 700 years earlier, the Japanese conquerors were very aware of China’s greater population, and intentionally used genocide in order to make China easier to rule. They killed 14 million Chinese people in their invasion (some say only 6 million, some say 20 million), mostly by deliberately starving them or slaughtering whole towns — similar to what the Nazis did in the USSR, but without the camps. They used biological weapons to try to kill even more. And of course they famously killed over 10,000 people by experimenting on them. This is in addition to a long list of massacres and atrocities carried out in the rest of Asia.

We tend to leave Japan off of the list of “history’s worst monsters”, because these crimes were committed in a very distributed way. They were not a result of official state ideology, like the Nazis’ ideology of race war. Individual figures like Tojo Hideki sometimes did encourage policies of mass starvation, mass shootings, and biological warfare, but the Imperial Japanese Army was remarkably decentralized, and lots of local commanders did atrocities all on their own. The Emperor probably knew about at least some of the atrocities being committed in his name, but didn’t try to intervene.

Still, collectively, Imperial Japan was the closest thing to the Nazis that existed in the 20th century, and like the Nazis they would have gotten far worse if they hadn’t been forcibly stopped.

#3: Pol Pot and the Khmer Rouge

Pol Pot and the Khmer Rouge had a total body count that was much lower than the other regimes on this list — 1.5 to 3 million instead of tens of millions — and they didn’t try to conquer the surrounding lands (though they sometimes attacked their neighbors). But the reason they only killed a few million people was that Cambodia only contained a few million people. The Khmer Rouge slaughtered a fifth to a quarter of their populace, higher than anyone else on the list — or maybe in recorded history. And had they had the power to conquer any of their larger neighbors, they probably would have conquered and killed even more. As it was, their reign of terror was only stopped by a Vietnamese invasion.

In terms of the cold, systematic nature of the mass murder, the Khmer Rouge are second only to the Nazis. The Killing Fields are legendary. You can go see the “magic tree” where Pol Pot’s soldiers would dash babies’ brains out, or gaze upon nice neat rows of human skulls. It really doesn’t get more psychotic than these guys.

#4: The USSR under Stalin

Stalin is generally held responsible for somewhere between 10 and 20 million deaths, of which maybe 6-10 million starved to death because of communist agricultural policies. Like Mao, Stalin learned of these deaths but turned a blind eye rather than admit a mistake. But there may have been more to it in Stalin’s case — he probably wanted a famine in Ukraine in order to intentionally reduce the Ukrainian population and make the “republic” easier to rule.

Stalin’s deliberate murders were fewer than Hitler’s but that’s a relative statement; he did slaughter maybe 3 million political opponents by murdering them or sending them to gulags to die of neglect. He committed small local genocides, killing over a hundred thousand Poles in WW2, mass-expelling a bunch of minority populations (like the Crimean Tatars) and killing many in the process, and so on.

His armies were also very brutal in Germany during World War 2, mass-raping literally millions of German women, expelling and enslaving large numbers of Germans in East and Central Europe, and so on. On top of all that, Stalin’s USSR was famously nightmarish, with people living in a constant state of fear and encouraged to rat out their neighbors.

What Stalin mostly didn’t do was conquer and expand the USSR’s territory. With a few exceptions, he just sat on his existing empire, dominated surrounding states with proxy arrangements, and focused on internal ideological and economic reorganization and maintaining his personal power. This lack of world-conquering ambition made him palatable as an ally against Hitler. Ultimately, he mostly just sat there and brutalized the people already under his own control.

#5: China under Mao

The vast majority of Mao’s deaths come from the famine. He also killed maybe 2-4 million people deliberately, through political purges, executions and forced labor during the Great Leap Forward, and the insanity of the Cultural Revolution. But the basic story of Mao is that he implemented insane policies in an attempt to gain and hold onto power, and often viewed Chinese deaths as collateral damage in his power struggles.

Mao, like Stalin, was also a villain that other countries learned to live with. He conquered Tibet and invaded South Korea (before being repulsed), but overall he wasn’t much interested in conquest. Like Stalin, he sat on his giant empire and brutally attempted to bend it to his (often insane) will. The society he created was desperately poor, paranoid, and totalitarian, but also often chaotic and even anarchic in places.

Honorable mentions

In my mind, those five regimes stand head and shoulders above the rest, but there are certainly no shortage of other bad ones. Seven honorable mentions:

• The Young Turks: Turkey’s junta carried out the Armenian Genocide during World War I, killing around a million people.

• North Korea: One of the world’s most nightmarish totalitarian regimes, North Korea tried to conquer South Korea, and also caused at least one mass famine with its policies of deliberate isolation.

• Rwanda’s Hutu government: The Hutu government in the 1990s committed the famous Rwandan genocide, and helped foment and prosecute the Second Congo War that killed millions of people in central Africa. An ideology of ethnic supremacy was a big part of the reason why.

• The Derg: Ethiopia’s communist regime in the 1970s and 1980s doesn’t get talked about much, but they killed maybe 500,000 people in political executions and ended up starving hundreds of thousands of Ethiopians to death.

• Yahya Khan: Under Yahya Khan in 1971, Pakistan tried to terrorize Bangladesh into remaining under Pakistan’s control. The army killed anywhere between 300,000 and 3 million Bangladeshis, and raped hundreds of thousands more.

• Suharto: Although Indonesia saw robust economic growth under Suharto, he killed over half a million people for political and ethnic reasons.

• King Leopold II: The Belgian king’s extractive regime in the Congo killed anywhere from 1.2 to 10 million people there. The bulk of the atrocities in the 19th century, however; Leopold’s rule over Congo ended in 1908.

I realize that this list is subject to differences of opinion, depending on A) which estimates you believe, and B) which factors you weight more heavily in the calculation of regime evilness. But I think it’s a pretty defensible list, and it demonstrates at least two important lessons.

First of all, the very worst atrocities come during times of war — especially wars of conquest, where foreign peoples are dehumanized and armies are given carte blanche. The fascist regimes of WW2 come off as the worst because they were engaged in massive wars of conquest. And wars often increase a regime’s appetite for further conquest and destruction. So basically, regimes that invade other countries tend to be worse.

Second, communist regimes tended to starve a lot of their people to death. Insane agricultural policies were a hallmark of communist economics — probably because successful communist revolutions usually began as revolts by agrarian peasantry rather than proletarian workers as Marx had envisioned. In trying to transform agriculture, communists ended up just starving their people to death.

So while communism probably caused more aggregate human misery than fascism in the 20th century, that was only because fascism didn’t have time; it was so incredibly malignant that it was stopped very quickly by external coalitions. Communism was more like a chronic debilitating disease, while fascism war more like ebola.

The worst regimes of the 21st century so far

That leaves us one more interesting question: By these criteria, who have been the worst regimes of the current century, so far? So far, the 21st century has been somewhat less destructive than the 20th — by 1926 we had already seen World War I, the Armenian Genocide, Lenin’s political purges, the Russian Civil War, and so on. The years 2000-2026 have, thankfully, been somewhat less brutal and chaotic so far. But we still have plenty of tyrannical villains. Here would be a tentative top five:

1. The RSF, Omar al-Bashir, and the Janjaweed: There has basically been a rolling series of genocides in Sudan this whole century — the first Darfur genocide, the second Darfur genocide, the Masalit genocide, and so on. Some of this was done at the behest of former president Omar al-Bashir, who encouraged murderous militias known as the Janjaweed, who later morphed into the RSF.

2. ISIS: This is the closest thing we’ve had to the Khmer Rouge in recent decades. ISIS’ extremist ideology led them to create a nightmarish totalitarian state in the areas of Iraq and Syria under their control, genocide the Yazidis, and throw much of the Arabian Peninsula into chaos. Their total death toll wasn’t as high as some of the others, but this was because they were stopped quickly; they were so horrible that everyone immediately united to fight and destroy them.

3. Vladimir Putin: Putin is one of the most aggressive leaders of modern times. He started wars in Georgia and Ukraine, and got heavily involved in the war in Syria. Over half a million people have died in Ukraine so far, and Putin bears some responsibility for the hundreds of thousands more who perished in Syria. It seems clear at this point that he’s not going to stop prosecuting wars of aggression until he dies or is defeated.

4. Bashar al-Assad: In his desperation to hang onto power, Assad laid waste to his country, killing hundreds of thousands and even using chemical weapons on rebels before ultimately fleeing.

5. The Tatmadaw: Myanmar’s military has been locked in a brutal civil war against ethnic minorities for decades, but in 2021 things really kicked into high gear after the army staged a coup against the country’s democratically elected government and put down protests with extreme brutality. The country is now an anarchic wasteland with war crimes galore.

These tyrants and murderers are somewhat less impressive, numbers-wise, than their 20th century counterparts. They also don’t clearly fit into the “fascism vs. communism” dichotomy that dominated much of the previous century; they’re more likely to be motivated either by Islamism, local ethnic chauvinism, or simple personal lust for power. But looking back at the 20th century, we should remember how bad dictators and totalitarian regimes can get, especially when they go to war — and how important it is to stop the worst of them in their tracks, before they can get even worse.

Subscribe now

Share

“Be assured, my young friend, that there is a great deal of ruin in a nation.” — Adam Smith

The U.S. and the UK have one of the longest and most important relationships in all of history. The UK is America’s parent civilization, and even after the two diverged and had a period of rivalry, they came back together to more or less jointly craft the entire modern age. In terms of science, philosophy, government, and every other realm, the Anglo-American alliance was more than an alliance — it was a joint civilizational project.1

That also means that the U.S. and UK are able to learn from each other’s mistakes. This is often a painful and fraught process, because it involves quite a lot of cross-Atlantic shouting and insults. If you bring up the UK’s flaws on social media, you’re going to get an earful about how America elected Trump, Americans have no health care, American kids are constantly dodging bullets, and so on, so Americans have no room to talk. This is unhelpful, but inevitable. It’s a defensive reaction by left-leaning British intellectuals and elites who are occupied with their own internal policy and culture-war battles with other British people. Any criticism from an American is seen as giving free ammunition to the local enemy, so the natural impulse is to screech “Shut up, Yank!!”.

But the Yanks must not shut up. Whether or not the Brits need our advice, we need to have our own internal discussion about what’s going on over in the UK and how we can use it to inform our own choices. And although there are still a few things to admire across the pond, the UK mostly now functions as a cautionary tale for the U.S. — yes, despite Trump, and guns, and high health care costs.

Whether you think the UK has been falling behind America economically for a decade or for two decades depends on which set of prices you use to measure living standards. If you use the two countries’ national price indices, the UK has been growing slower than the U.S. since the financial crisis of 2008; if you use PPP prices, as Paul Krugman prefers, the UK’s relative stagnation has come only since Brexit in 2016:

That controversy leaves an important question unanswered — whether 2008 or Brexit was ultimately more damaging to British growth. But it also obscures an even more important fact, which is that by any of these measures, the UK is much poorer than the United States:

A data point that gets thrown around a lot is that the UK is as poor as Mississippi. In GDP (PPP) terms, that’s true — Mississippi’s income per capita is about $56,000, roughly similar to the UK. That doesn’t mean life is equally as good in the two countries — Mississippi is more violent and less healthy than Britain. But unless you think that Britain had to sacrifice a significant portion of its material wealth in order to be safer and healthier, this is a bit beside the point. France and Germany, which are just as safe and healthy as the UK (if not more so), also have significantly higher incomes. This is from a Resolution Foundation report:

Middle-income Brits are now 20 per cent poorer than their peers in Germany and 9 per cent poorer than those in France. Worse, low-income households in the UK are now around 27 per cent poorer than their French and German counterparts. It’s important to comprehend just how material these gaps are: the living standards of the lowest-income households in the UK are £4,300 lower than their French equivalents.

A very deep principle in economics is that if two countries are roughly equivalent in terms of their basic institutions, they ought to converge to similar income levels. The fact that the UK and the U.S. have diverged rather than converged suggests that the former must have some institutional weaknesses relative to the latter. And that implies that there are cautionary tales here for America.

What are those cautionary tales? There are a number of good breakdowns of the UK’s problems, and you can learn a lot from observing the overlaps and disagreements between them. Ben Southwood, Samuel Hughes, and Sam Bowman have a long piece called “Foundations”, which basically makes a libertarian YIMBY argument; they claim the UK is over-regulated and can’t build housing, infrastructure, or really anything else. Jim Tomlinson has a rebuttal in which he argues that the real barrier to investment is the financialization of the UK economy. Ben Ansell argues that the UK is in a version of the “high-level equilibrium trap” — the UK is still fairly wealthy, so its politics center around how to divide up the pie rather than growing it. The Resolution Foundation’s report has lots of suggestions for how to get Britain growing again. Michelle Kilfoyle also has some recommendations. These are only a few examples.

But my favorite rundown is probably the recent Atlantic article by Idrees Kahloon. The article points out that it isn’t just private industry that’s broken in the UK; it’s also the effectiveness of the government institutions that many British people cherish. This includes the fabled National Health Service. Kahloon writes:

Taxation is at the highest level since World War II, yet public services have deteriorated. The National Health Service, the celebrated pillar of the British cradle-to-grave welfare state, has a backlog of 6 million patients—almost a tenth of the population—waiting for treatment. The health service now has to spend more money settling maternity-malpractice claims than it does on actually providing maternity care. Many Brits can neither obtain an appointment with a publicly funded dentist nor afford a private one; in a 2023 survey, one in 10 reported doing DIY dental work, in extreme cases extracting their own teeth or gluing broken crowns back together…

Incomes can be shockingly low: Junior doctors recently went on strike for the 15th time in three years over their salaries, which start at just £38,800; the median salary for British civil servants is £35,680…

Recent plans to transform the country['s transportation system] have rested in no small part on High Speed 2, a superfast rail line intended to connect London with Birmingham, Leeds, and Manchester. But since HS2 was proposed, in 2009, its costs have tripled, to more than £100 billion. It is the most expensive rail line in the world. (A special structure to protect a rare bat species near the rail line in Buckinghamshire required 8,000 permits and was built at a cost of £216 million.) The most important sections of the proposed route have been lopped off. The rump line—going from Birmingham, Britain’s second-largest city, to not-quite-central London—may be finished by 2040…

Building infrastructure, or much of anything else, has become all but impossible in the United Kingdom. In addition to having the world’s most expensive (not yet built) train line, Britain also hosts the world’s most expensive (not yet built) nuclear-power plant, Hinkley Point C. Its environmental-impact assessment ran 31,401 pages; the plant will feature a £700 million “fish disco,” which will pulse sounds underwater to deter animals from its intake pipes. The government spent 32 years and £179 million planning a tunnel beneath Stonehenge to relieve traffic, only to officially scrap the plan this year. Even basic tasks, such as obtaining power, can be nightmarish…

Britain suffers from a housing crisis significantly worse than America’s. The problem cannot even be blamed on zoning, because Britain does not have a zoning regime to speak of. Rather, every attempt to build is a painful, ad hoc negotiation with local government councils and NIMBY residents. As a result, housing costs per square foot are among the highest in Europe. In the words of one report, “Our housing stock offers the worst value for money of any advanced economy.” France has roughly the same population as the U.K., but almost 50 percent more homes. And yet, since the financial crisis, the U.K.’s rate of housing production has only fallen…In London, the typical house sold in 2024 cost 11 times median earnings.

One of the key lessons here appears to be that NIMBYism kills the economy by a thousand small cuts. But Americans already know this. Kahloon’s description of Britain’s struggle to build high speed rail sounds uncomfortably like California’s story. And the housing story is also very similar to America’s. Most of the other analyses I listed above identify NIMBY-driven housing scarcity as one of the key factors holding back growth in the UK, and recommend various different strategies to address it.

I’m not sure the U.S. has a lot to learn from those various proposals, since they’re highly specific to the UK’s institutional context. But the overall lessons here are worth reiterating: Approval policies full of veto points and cumbersome procedures are a drag on the economy, local control can easily go too far, good infrastructure is often a matter of low-cost efficiency rather than pure willingness to pay, and so on. Americans have a severe NIMBY problem, but they can look over at the UK as a cautionary tale of how bad things will get if the NIMBY problem is allowed to fester.

Another lesson that Americans have already learned is the peril of excessive financialization of the economy. Almost everyone agrees that the UK economy is excessively concentrated in London. John Burn-Murdoch writes:

It will surprise nobody that London accounts for an outsized share of Britain’s output, but the magnitude of the UK’s economic monopolarity is remarkable. Removing London’s output and headcount would shave 14 per cent off British living standards, precisely enough to slip behind the last of the US states. Britain in the aggregate may not be as poor as Mississippi, but absent its outlier capital it would be.

Not only does this mean that the fruits of the UK economy are inequitably distributed — requiring a constant stream of politically difficult transfers to keep other regions afloat — but it also means that NIMBYism bites much harder than it otherwise would, because pricing people out of London means pricing them out of participation in one of the few healthy parts of Britain’s economy, and because pricing people out of highly productive cities hurts productivity.

And part of the reason the UK’s economy clusters in London is finance. As Reuters reports, the UK economy specializes in finance:

The United Kingdom is unusually dependent on finance. Financial services account for about 12% of economic output, opens new tab, compared with around 7% in the United States. Much of that activity emanates from London. The City is the world’s second-largest asset management hub, the leading venue for foreign exchange trading, and the number one location for over-the-counter derivatives.

Finance is an industry with an especially strong clustering effect; almost every country has just one city that’s the “financial capital” — NYC in the U.S., Tokyo in Japan, etc. But in an economy as financialized as Britain’s, the financial capital is the center of the entire economy, probably to an unhealthy degree.

Another obvious downside of financialization is that specialization in a single tentpole industry made the UK economy unusually vulnerable to a crash in that industry. The financial crisis of 2008, and the follow-on Euro Crisis in the years afterward, dealt a heavy blow to the UK finance industry — and, thus, to the UK economy. Whether you think Britain’s GDP stagnation truly began in 2008 or 2016, it’s inarguable that real wages stagnated and fell for years after 2008. Compare average weekly earnings in the UK to those in the U.S., and the difference is crystal clear:2

Whether financialization has other drawbacks — excessive corporate short-termism causing lack of investment and deindustrialization, financialization of housing development causing shortages, etc. — isn’t clear. But just these two negative effects — the UK economy’s hyper-concentration in London, and its vulnerability to single-sector shocks — provide a cautionary tale about letting an economy rely too much on the finance industry.

Again, this is a lesson that Americans already learned, at least partially, after 2008 (though thankfully, the lesson was somewhat less painful for us than for the Brits). But one lesson from 2008 that the U.S. should pay attention to is that fiscal austerity, if done too abruptly and carelessly, can leave long-term scars on an economy. Kahloon writes:

[T]he government’s actions during and after the [2008] crisis compounded the damage. Rather than increase spending to revive depressed demand, as modern Keynesians would counsel, the government…opted to slash budgets as revenue plunged…

Funding for day-to-day NHS operations was maintained, for instance, but only by cannibalizing the capital budget. A 2024 government report found that, as a result of austerity, Britain has “crumbling buildings, mental health patients being accommodated in Victoria-era cells infested with vermin with 17 men sharing two showers, and parts of the NHS operating in decrepit portacabins.”…

After austerity cuts to welfare benefits took effect, the share of children who grew up in long-term poverty, meaning half their childhood or more, shot up from about 14 percent to 23 percent. Nutrition appeared to suffer, and doctors reported increased cases of diseases stemming from vitamin deficiencies, such as rickets and scurvy.

The U.S. needs to hear this lesson. Our debt problem is becoming unsustainable:

We’re going to need to do fiscal austerity, but it’s important that we do it gradually, by restraining the growth of spending (and raising taxes), rather than by abruptly slashing government services and public works.

Kahloon’s example of the NHS actually demonstrates another lesson — the UK’s strategy of providing health care is probably uniquely flawed. In most advanced countries, the government provides universal health insurance; the government pays part of the cost of health care, but it doesn’t actually provide the care directly. The UK government, in contrast, provides health care directly to its people as a state-owned enterprise.

Britain’s NHS helps control costs in the short term; the system is cheaper than those of other rich countries, for roughly similar outcomes. But in the long term, it makes the UK’s health system subject to the whims and vagaries of politics. After the 2008 crisis, the UK disinvested severely from health care because of fiscal austerity policy, while other countries continued to do the necessary capital spending to maintain future services. This is a big part of the reason why the NHS system is now widely seen as falling apart.

The U.S. is vanishingly unlikely to switch to nationalized, state-run health care provision — at the very most, we’ll get a system of national insurance, like most other rich countries have. But watching the degradation of the NHS should remind us of why direct government provision of necessary services is so dangerous. It’s not necessarily that government is inherently inefficient; it’s that it’s inherently political. UK-style austerity is one reason government can starve its own enterprises of capital spending, but it’s not the only one — under Hugo Chavez, Veneuzela’s government starved its own state-owned oil company of capital investment in order to deliver a bonanza of government largesse, with predictably disastrous results. We should remember this when socialists like Zohran Mamdani promise to run essential services through the government.

One key lesson that all the analyses of Britain’s troubles seem to already have internalized, but which the U.S. needs to pay attention to, is that maintaining overseas markets is very important. Whether 2008 caused a long-term economic slowdown or not, 2016 definitely seems to have done so. Brexit is the huge, looming shadow that hangs over everything that has happened to the UK economy over the past decade.

Bloom et al. (2025) study the growth impact of Brexit, looking at both macro data and micro data. For the macro, they assume Brexit was an unanticipated shock, and compare the UK to other countries. For the micro, they look at how various British businesses fare after Brexit, according to how dependent they were on the EU before the big surprise breakup. Both estimates agree that Brexit was a negative shock:

[B]y the end of 2025 Brexit had reduced UK GDP by 6% to 8%, with the impact accumulating steadily over time. We estimate that investment was reduced by 12% to 13%, employment by 3% to 4% and productivity by 3% to 4%. These negative impacts reflect a combination of elevated uncertainty, reduced demand, diverted management time, and increased misallocation of resources from a protracted Brexit process.

Other studies agree.

The U.S. should look at Britain’s experience and remember that international markets matter. The pre-Brexit UK was a more internationalized economy than the pre-Trump U.S. — exports are over 30% of UK GDP, compared to only 11% in the U.S. But that seemingly small stream of exports revenues supports a lot of more local industries, through the magic of local multiplier effects. If Trump and his successors disrupt those markets by starting pointless trade wars with allies and friendly countries, the results won’t be quite as bad as in the UK, but they won’t be good.

A final lesson that the UK’s failures can teach the U.S. is that energy supply is very important. Most analyses tend to gloss over the chronic energy scarcity that has afflicted the UK in recent years. “Foundations”, however, puts this squarely in the limelight:

High energy costs help make British industry uncompetitive, increasing the country’s reliance on finance. They also directly impoverish the British people, who are forced to make do with much lower energy consumption than people in other rich countries. Here’s electricity generation per capita:

Energy is one very easily identifiable, tangible dimension in which British people are much poorer than Americans.

The Economist ties Britain’s energy scarcity to its overly rapid and haphazard push for decarbonization; having banned coal, and with natural gas now extremely expensive due to the Ukraine war, the UK has been forced to rely on hideously expensive offshore wind power.

The U.S. doesn’t have this exact problem. Because the U.S. has much more land and much more sunlight than the UK, we can use solar power (with batteries), which is much cheaper than offshore wind and are getting cheaper all the time:

But the general lesson remains — there is no such thing as a low-energy rich country. The MAGA movement’s irrational opposition to solar power is even more foolish than Britain’s haphazard push to ban coal.

So to sum up, the UK’s sad experience with economic stagnation offers several important lessons for America:

• NIMBYism is very bad

• Excessive reliance on the finance industry is dangerous for many reasons

• Abrupt, poorly-designed fiscal austerity can do long-term damage to an economy

• Direct government provision of essential services is easily politicized

• Cutting a country off from its export markets is very bad

• Energy abundance is important for both industry and living standards

Overall, the UK should serve as a reminder that economic growth is very important, and countries turn away from growth at their peril. The UK, caught up in endless political turmoil and culture wars, has forgotten the overriding importance of growth for national happiness and well-being. The U.S., facing its own political chaos and culture wars, should try very hard not to make the same mistake.

Subscribe now

Share

A month ago I wrote a post reflecting on whether my blog output is as relevant as it used to be:

My basic concern is that the U.S. has entered a political era in which reasoned assessment of the evidence, concern for the common good, and deliberation among intellectuals has relatively little effect on policy. When populism, tribalism, and social media shouting rule the nation’s politics, who’s even listening to someone like me?

In fact, I think it’s not that bad. Consider the 21st Century ROAD to Housing Act, which just became law after passing the House and Senate with overwhelming bipartisan majorities, despite Trump’s refusal to sign it. The main provisions of the new law read like a YIMBY’s dream:

• Expedited and simplified environmental review for housing projects

• More federal money for places that build more housing, less money for places that build less

• Some deregulation of prefab home construction and housing finance

Given the very limited powers that the U.S. federal government has over housing in this country, this is a huge win. These are policies I’ve been advocating for years.

Notably, the bill also contains one unhelpful “slopulist” provision: restrictions on the ability of corporate landlords to buy single-family homes. This measure was based on the myth that corporate investors, especially private equity firms, are responsible for the American housing crisis. Would-be populists across the political spectrum, from Elizabeth Warren to Donald Trump, profess to believe in this myth, but it’s just not true — corporate landlords are only a tiny sliver of the housing market, and there’s some evidence that they actually reduce rents, especially for low-income tenants.

But because corporate landlords are such a small part of the market, restrictions on their activity are unlikely to have any measurable effect on housing prices nationwide. And the final bill included so many carve-outs — for example, corporations that build housing in order to rent it out are excluded from the restrictions — that its negative impact will probably be very muted.

In other words, we just saw Congress — that dysfunctional institution that everyone hates — take action on an important issue in a very reasonable and bipartisan way. It’s a validation of Matt Yglesias’ “Secret Congress” theory — the idea that our legislators mostly just want to do what’s good for the country, but are prevented from doing so by the partisan anger of their electorates. According to this theory, the way to get good laws passed is to have them remain low-profile, so they don’t arouse the nation’s compulsion to turn everything into a zero-sum politicized food fight.

In other words, there’s still some scope for reasonable individuals and intelligent discourse to make an impact on American policy. But this scope is very limited, because “Secret Congress” has to remain secret — dedicated public servants can tinker around the edges, but can’t enact the kind of bold, sweeping policy program that FDR and Reagan (and to a lesser degree, Obama) were able to carry out.

The problem is not a lack of good ideas, or a lack of tools for evaluating those ideas. The problem is America’s political economy.

Roughly speaking, we have three major political movements — the MAGA right, the socialist left, and the “woke” left. But none of these seems especially interested in the kind of smart, reasonable policies that would significantly improve the material well-being of regular Americans. Their economic programs are all pretty bad. And even if they could be persuaded to support smarter policies, the conflict between the movements will preoccupy them, keeping them focused on culture-war issues.

I’m not really a political blogger, but an understanding of my country’s political economy seems essential if I want to be a useful intellectual. So I thought I’d give a brief rundown of how I think about the U.S. political situation, and why I think the outlook for a big, constructive policy push is so bleak. I’ll start with a historical analogy.

The Spanish Civil War analogy

Back in 2016, I read a bunch of books on the Spanish Civil War of 1936-1939. I immediately saw strong similarities between that conflict and America’s worsening political troubles. In the years since, as America suffered nationwide riots, a coup attempt, military purges, accusations of election theft, and increasingly authoritarian politics, I’ve continued to invoke the analogy, and I’ve been surprised at how helpful it’s been.

To be clear, I don’t think the U.S. is going to have an actual civil war. The military would have to fragment and fight itself, as it did in Spain, and I think this is unlikely. But Americans regularly use the language of civil war (or “cold civil war”) to define their current political conflicts, and fear of an actual civil war is widespread. The analogy doesn’t have to be perfect.

Here’s a very brief two-paragraph synopsis of the Spanish Civil War. Basically, Spain in the 1930s was riven by deep, long-standing, intractable class conflict (which also sometimes manifested as religious conflict between socialists and the Catholic Church). After the country became a democracy in 1931, it see-sawed between left-wing and right-wing governments, with major street clashes between rightists and leftists. In 1934, a bunch of leftist miners staged a giant strike to protest the admission of some rightists to the government, took over a province of Spain, then tried to turn the strike into a revolution.

The army crushed the revolt, but it led a bunch of right-wing types to panic that if they ever lost another election, there would be a leftist revolution. So in 1936, when a center-left coalition won an election, rightists tried to stage a coup and overthrow the republic. They failed at first, but they captured part of the country. Over the next three years, having already won more of the population and (crucially) the military to their side, they slowly ground down and conquered what remained of the republic. Although both sides were composed of various ideological factions, the rightist (“nationalist”) factions all united around the authority figure of Francisco Franco, while the leftist factions fought each other behind the lines. Both sides committed substantial atrocities, though the nationalists generally committed more.

If you want to read more about the war, I suggest:

• Antony Beevor’s The Battle for Spain

• Paul Preston’s The Spanish Civil War

• George Orwell’s Homage to Catalonia

So how do I see the Spanish Civil War as analogous to America’s current political troubles? Basically:

• A long-standing, intractable social conflict flared into a brief moment of mass revolutionary fervor

• That almost-revolution panicked various factions of the right, scaring them into uniting behind a radical, authoritarian program

• The most revolutionary factions of the left (the anarchists) were eventually muscled out and dominated by the most disciplined faction (the communists), who then proceeded to lose the war

• The rightists had no idea how to govern the country after their victory, embracing policies that kept Spain poor for decades

The most important analogy is that although the Spanish Civil War featured a number of different factions, none of the factions had a workable program for governing the country. Spain was basically screwed no matter who it went with, but people still fought very hard for one faction or another, out of a combination of fear, ideology, and opportunism.

So let me explain how I think this analogy fits America’s current situation, and why it makes me so pessimistic about governing the country.

The long shadow of 1934 and the long shadow of 2020

Read more

My elderly rabbit Giggles has unfortunately been quite sick the past two days (he’s recovering now), so I haven’t had as much time to write. Instead I thought I would repost a book review I did two years ago. I’ll be back with more original content tomorrow!

In my post yesterday about the “We Must Act Now” statement, I noted that Daron Acemoglu successfully got the writers to alter some key language in exchange for his signature — most likely, adding the “steering” idea that I took issue with. Technological “steering” is the central policy recommendation of Acemoglu’s 2023 book Power and Progress, with Simon Johnson.

As it happens, I wrote a review of Power and Progress, back in 2024. In fact, it was by far the longest book review I’ve ever written.1 I actually read the book cover to cover in detail, marking passages and looking up sources as I went; this took a very long time, and it was very frustrating, since I had serious objections to nearly every part of the book. I took out that frustration by writing a very long and highly critical review.2

So anyway, here’s that review for you to enjoy. Hopefully it gets a few people to think a little harder about the AI-related ideas that Acemoglu has been promoting throughout the econ world, and why it would be a big mistake to make those ideas the default position of the economics profession on the AI issue.

“Do not be fooled by the monumental technological achievements of humankind.” — Acemoglu and Johnson

It’s hardly surprising that Power and Progress made it onto practically every list of the most important business books of 2023. First, there’s the unrivaled pedigree of the authors themselves. To call Daron Acemoglu a powerhouse in the world of economics would be a ludicrous understatement:

Acemoglu is also the main proponent of the institutional explanation for national development, through his famous book Why Nations Fail and its sequel, The Narrow Corridor (both with James Robinson). If you hear me talk about “inclusive institutions” and “extractive institutions”, I’m channeling Acemoglu.

Simon Johnson, meanwhile, is the author of some of my favorite popular books about economic policy, especially Jump-Starting America (with Jonathan Gruber) and 13 Bankers (with James Kwak). When I write more about the need to spend more on science and to restrain the excesses of the finance industry, I’m channeling Johnson.

The second reason this book was destined to garner attention is that it brings together two extremely timely strains of thought: 1) the widespread distrust of tech companies that has grown in American society over the last few years, and 2) the wave of anxiety over AI-driven automation. Power and Progress weaves those two anxieties into a more-or-less coherent whole — a sum of all technological fears, if you will. And it seems to have been spectacularly well-timed, since its release coincided closely with the coming of ChatGPT and other generative AI.

But given all of those powerful tailwinds, I have to say I’m kind of surprised at how little of a splash Power and Progress seems to have made. This is anecdotal of course, but in the 9 months since it came out, I’m not sure I’ve once heard someone reference the book or any idea in it. The authors clearly intended it to be a handbook for people who are scared about AI putting humans out of a job, the way Thomas Piketty’s Capital in the Twenty-First Century became a handbook for people worried about inequality, or Robert Gordon’s The Rise and Fall of American Growth became a handbook for people concerned about technological stagnation. But I don’t think it did.

Why not? One reason might be that the timing wasn’t as favorable as it might appear. Contrary to Acemoglu and Johnson’s assertion (on p.24 of the hardcover edition) that we live in an age of “blind techno-optimism”, the internet is absolutely chock-full of arguments and warnings about the downsides of AI. Concerns over the risk of rogue Artificial General Intelligence resulted in a boardroom coup attempt that almost drove Sam Altman out of OpenAI. Worries that AI wouldn’t uphold diversity led Google to implement some pretty hilarious countermeasures. Fears of mass surveillance, deepfakes, etc. are widespread. And of course the idea that AI is going to lead to mass unemployment is absolutely ubiquitous — so much so that practically every San Francisco tech event I go to features discussions about exactly this subject. Yes, even dance parties.

In other words, Power and Progress may have come out a little too late to make a big splash, and instead ended up just being one more voice shouting in the chorus.

On top of that, though, I have to say that this book…well, I just don’t think it’s very good. I winced while I wrote that sentence, because Simon Johnson is a personal friend, and Acemoglu is a celebrated genius, and because both of them have written such good books in the past. This is the first broadly negative book review I’ve written since 2014, and I’m a lot less combative of a blogger than I was a decade ago. I did not want to pan this book, especially because I think the topic is a good and important one, and I think the authors are brilliant people whose hearts are in the right place.

But I just don’t think the way this book was written ends up supporting the conclusions it draws. The historical examples it cites simply don’t support a narrative of out-of-touch technologists inventing the wrong sorts of technologies and hurting workers in the process. The book embraces a highly questionable definition of “power” in which persuasion in an open democratic society is painted as a threat. It often seems to assume its conclusions about the impacts of specific technologies, and it tells a jumbled and confusing story about the role of productivity growth. And its central claim — that society can push entrepreneurs to steer innovation in a direction that augments humans instead of replacing them — is not well-supported.

All in all, Power and Progress just fails to convince.

The basic idea

Power and Progress is of the “magisterial sweeping tome” class of econ book, like Capital in the Twenty-First Century, The Rise and Fall of American Growth, or Brad DeLong’s Slouching Toward Utopia. Much of the book is a history of technological innovation in general. As such, it tends to ramble; the authors often seem to get so caught up in the telling of this history that they neglect to tie each event to their central theses. In fact, those are often the most fun and fascinating parts of the book. But if I were to boil down Power and Progress to a set of core ideas, it would be:

1. Technological innovation’s impact on human welfare depends crucially on social choices about how those innovations are used.

2. Those choices are determined by the relations of power in a society, and in recent decades our choices have been steered in a negative direction by the power of tech company founders and venture capitalists.

3. The type of technologies that society invents can be chosen so as to distribute benefits more widely, by avoiding technologies that replace workers and inventing technologies that complement workers.

It’s the last of these that the book is most known for, because it’s the boldest, the most original, and the most controversial. But first let’s talk a bit about the other two.

Questionable historiography

The idea that technology’s impact on society is not determined solely by the nature of the tech itself, but depends on how we choose to use it, is obvious enough to be a truism. Everyone knows how the industrial technologies that have created so much wealth are also put to destructive uses in wars. Everyone knows that the same camera technology that lets you talk to your friend in a different city can allow governments to spy on their citizens. Everyone knows that there is a vast system of laws, international agreements, and social norms whose purpose, at least in theory, is to ensure that technology is used for good and not for ill.

But even though “technology can be used for bad purposes” should be a simple truism, Acemoglu and Johnson pick some very odd examples to illustrate the principle. For example, in the prologue, they have a list of what they claim are “new inventions that brought nothing like shared prosperity”. Here’s the fifth item on their list:

At the end of the nineteenth century, German chemist Fritz Haber developed artificial fertilizers that boosted agricultural yields. Subsequently, Haber and other scientists used the same ideas to design chemical weapons that killed and maimed hundreds of thousands on World War I battlefields.

The idea that the Haber-Bosch process has “brought nothing like shared prosperity” is an absolutely wild claim. Nitrogen fertilizers are so important to human existence that by the most common estimates, about half of the entire population of Earth — 3.5 billion people — is only sustained thanks to this technology. But because that same chemical reaction was used to create one particular type of chemical weapon that was responsible for a tiny fraction of the deaths in one particular war, Acemoglu and Johnson feel comfortable saying that a technology that literally gives life to half of humanity “brought nothing like shared prosperity”. It is the kind of claim that is so obviously wrong as to leave the reader slack-jawed — and yet it is deployed in support of an overall thesis for which countless better examples exist.

Unfortunately, this kind of questionable selection of historical examples is a hallmark of Power and Progress all the way through. For example, in Chapter 6, the authors write:

[Belief in the power of productivity] suggests that as technology advanced rapidly during the early phases of the Industrial Revolution, wages should have risen. Instead, real incomes of the majority stagnated.

Acemoglu and Johnson conclude that because textile manufacturing technologies were biased toward automating workers, they immiserated the working class of 1700s Britain. But those same textile manufacturing technologies have been at the center of the early stage of every other country’s industrialization as well. China went through a period where it made most of the world’s clothes, with its share peaking in the late 2000s. In 1995, apparel was China’s biggest export category.

But during this time, when Chinese garment workers were getting the descendants of the original British industrial technologies of power looms, their wages were skyrocketing — as were wages in the economy as a whole. The same is now true of Bangladesh — the country focuses relentlessly on the garment industry, and has access to all of the old automation technologies, and yet incomes in the country have tripled since 1990.

(As a side note, it’s kind of funny that after we’ve used “Luddite” as a slur for technophobes for all these years, Acemoglu and Johnson explicitly try to rehabilitate the original Luddites, writing that they “were right to worry about knitting frames decimating their livelihoods”. For this reason, I considered subtitling this review “the Bible of the Luddites”, but decided that the negative connotation of the word was too strong and it would be rude.)

A third questionable example in the book is the story of the Panama Canal. Acemoglu and Johnson describe the brutal exploitation of the workers who built the canal, and declare the project a “colossal failure”. That brutality was certainly real. But the authors cite it as a reason that the technology of the canal itself failed to bring broad-based prosperity. In fact, the opposite seems true; thanks to the canal, the people of Panama today enjoy a standard of living much, much higher than that of their Central American neighbors. This is not to say those economic benefits were worth the human cost. But the canal’s problems clearly seem associated with its construction, rather than unfair distribution of the benefits from the technology itself. Could the same canal have been built using more humane labor standards? The authors decline to speculate, simply declaring the whole project a failure and not even mentioning Panama’s prosperity.

A fourth dodgy example is the story they tell about Japan. In Chapter 8, Acemoglu and Johnson praise Japanese companies for “combin[ing] automation with the creation of new tasks”, noting that Japanese automakers didn’t reduce their workforces like American automakers did. But Japan’s manufacturing sector wages, like wages throughout the rest of the country, have been falling since the early 1990s, while American wages have stagnated but not fallen. So this story doesn’t fit the data.

A fifth example is in Chapter 7, when Acemoglu and Johnson write that “Henry Ford was a pioneer” in developing “a more cooperative relationship” with his workforce. I’m just wondering how this “more cooperative relationship” involved hiring thugs to gun down union organizers. Ford did pay higher wages to increase efficiency, but his actual dealings with representatives of labor was brutal and intolerant.

I could go on citing these questionable examples — my copy of Power and Progress is stained blue with all the notes I made in the margins — but this review would run into the dozens of pages, and you would quit long before you finished. But because there are so many questionable examples, Power and Progress is the kind of book that must be read closely and with a critical eye.

Update: In the comments, Brian Potter adds:

Another strike against this book is its terrible scholarship: it frequently gets basic facts incorrect because the authors haven't bothered to actually research the topic. Example: it claims at several points that Eli Whitney was responsible for the development of interchangeable parts, a claim that has been widely and thoroughly debunked.

Source?

Another issue with the book’s examples is the lack of footnotes or endnotes. Instead of citing specific works in support of each specific claim — as most books do — Power and Progress has a bibliographic essay at the end. Many sources are mentioned in this essay, but it’s often difficult, and sometimes impossible, to match the sources to specific claims. As a result, you often end up having to choose between exhaustively searching multiple sources to figure out where the authors got a particular point, or simply giving up and trusting that the authors are accurately representing the data.

For example, in Chapter 1 the authors ask “What if…AI also impoverishes billions in the developing world?”, and asserts that “evidence is mounting” that this concern is “valid”. But where is the evidence that AI threatens to impoverish billions? That’s an astonishingly strong claim about a technology about which little is known, and I can’t find any source in the bibliographic essay. An empirical study I do know is Acemoglu, Autor, Hazell and Restrepo’s 2022 paper “AI and Jobs: Evidence from Online Vacancies”, whose abstract concludes:

We find no discernible relationship between AI exposure and employment or wage growth at the occupation or industry level, however, implying that AI is currently substituting for humans in a subset of tasks but it is not yet having detectable aggregate labor market consequences.

So that paper certainly doesn’t include mounting evidence that AI threatens to impoverish billions. But I can’t find which paper the authors relied on to make this claim.

In fact, because I’ve read many of the Acemoglu papers that undergird this book, I also know that there are instances where the data doesn’t quite say what the authors claim. For example, in Chapter 8, Acemoglu and Johnson argue that “digital technologies became the graveyard of shared prosperity” over the last few decades. In this chapter, they attribute some meaningful piece of the recent rise in inequality to the introduction of digital technologies to the workplace. But because I’ve read Acemoglu and Restrepo’s 2020 paper “Robots and Jobs: Evidence from U.S. Labor Markets”, as well as the working paper version from 2017, I know to be skeptical of this claim.

Acemoglu and Restrepo found that a narrow category of automation — industrial robots — was associated with decreased employment and wages. But as the Economic Policy Institute’s Larry Mishel and Josh Bivens noted, when Acemoglu and Restrepo measured the effect of workers’ “exposure to IT capital” in general — i.e., how much the employers invested in IT tech overall — they found either no effect or a positive effect on employment and wages. Here’s the relevant table from the 2020 version of the paper:

The estimates of a positive impact of IT capital on employment and wages are in the original working paper version, in table A9.

Now, this doesn’t mean that computers and the internet weren’t a driver of mass unemployment or stagnant wages. Maybe they were! Acemoglu and Restrepo (2020) could simply be wrong; in fact, since their working paper first came out in 2018, many other studies have ended up contradicting their findings about the negative impacts of robots. And all of these papers look within specific industries or companies — as does Acemoglu and Restrepo’s 2022 follow-up paper about automation and inequality between demographic groups. The overall effect of automation on economic growth, absolute wages, and the composition of industries in the economy simply isn’t known.

In other words, it could be very well be that automation has been impoverishing people, or it could be that it has been enriching people overall. I’d simply like to know where the authors get the data to back up their claim about information technology, especially when one of the authors’ most famous papers appears to contradict that claim.

To sum up, footnotes and endnotes are a technology that has unambiguously benefitted the world, and even though they can be a bit of a pain the butt, authors should include them.

A questionable definition of “power”

Anyway I digress; back to the book’s central theses.

Despite the questionable examples, it’s clearly true that technology can be used to benefit average people or to hurt them. But how does society choose how to use technologies? Acemoglu and Johnson’s answer is “power”, from which they get the title of their book. But what is power? Here, in Chapter 3, Acemoglu and Johnson deploy a definition that veers into the tautological:

Power is about the ability of an individual or group to achieve explicit or implicit objectives. If two people want the same loaf of bread, power determines who will get it.

Using this definition, how could we ever conclude that power wasn’t the reason for an observed outcome? Two people want a loaf of bread, and one of them gets it; we know this was due to “power”, because “power” is defined by who gets a loaf of bread. This kind of definition is semantically valid, but empirically useless; if you define “power” such that it simply means “whatever caused an outcome to happen”, you haven’t isolated causality, you have simply given it a new name.

Acemoglu and Johnson have a reason for employing a definition this infinitely broad; it allows them to include persuasion and compulsion in a single category of “power”.

The authors’ historical examples of when power determined the distribution of the benefits of technology include cases when laws and the threat of violence allowed some people to extract the benefits of technology for themselves — the cotton gin increasing slaveowners’ profits in the American South, or lords extracting agricultural surplus from peasants in medieval Britain. These are instances of compulsion, which certainly fit with our common, everyday, colloquial use of the word “power”.

But Acemoglu and Johnson also spend a lot of time arguing that persuasion is also a form of power. They cite instances in which techno-optimists and businesspeople in 18th century England and 21st century America persuaded the public to enact pro-business policies, through articles, speeches, conversations, and so on. Their explanation for why inequality has increased since the 1970s is, in effect, that silver-tongued technologists managed to persuade American society to weaken pro-worker institutions, and to allow the technologists to invent technologies that replaced human labor instead of complementing it.

The authors don’t venture to say exactly why these techno-optimists’ pro-business vision prevailed — they write that “an idea is more likely to spread if it is simple, is backed by a nice story, and has a ring of truth to it,” but they admit that “quite a bit of this process [of persuasion] is random,” and declare that “you are enormously lucky if you get the right idea, with just the right ring to it, at just the right time.”

I have to admit, this kind of surprised me. I expected to see some sort of pseudo-Gramscian theory of cultural hegemony (or at least some references to Gramsci or similar writers). Instead, the authors just sort of shrug and put it all down to luck. For some reason, the techbros just wrote really good posts, and by doing so they ruled the world — at least until their luck ran out and the world turned against them, I suppose.

In fact, I have to confess that the entire chapter on power and persuasion left me bewildered. I do not understand why we should put accidental success in a nonviolent marketplace of ideas in the same conceptual category as chattel slavery and feudalism. It seems to yield neither understanding nor solutions. Perhaps the historical example of the cotton gin might give us some insight about how to explain the spread of laissez-faire economics, but simply labeling both things as varieties of “power” does not yield that insight. And the idea that persuasion is power doesn’t seem to suggest any kind of systemic fix for the problem that sometimes society is persuaded to do things that increase inequality.

That doesn’t mean Acemoglu and Johnson have no solutions to recommend, though. They want to strengthen institutions like unions and labor laws, but their main idea is to redirect technological innovation toward technologies that complement workers instead of replacing them.

When all you have is a hammer, everything looks like a loom

“With such persuasiveness, you tend to convince yourself that you are correct.” — Acemoglu and Johnson

Over the last six years, Acemoglu and Restrepo wrote a series of theoretical papers in which they lay out a number of different ways that new technology can affect workers’ jobs and wages. Basically, this ends up being a fancier version of a very old idea — capital can either substitute for labor or complement it. If capital substitutes for labor, then capitalists win, because they can replace people with machines, and pay people accordingly less. But if capital complements labor, then workers win, because capitalists have no choice but to hire them to work the new machines, and pay them good wages. Acemoglu and Restrepo make this theoretical breakdown a bit more nuanced, but in the end it really boils down to whether machines replace people or augment their abilities (either by making them more productive or by giving them new things to do).

Power and Progress attempts to analyze the history of technology through the lens of this theory. In eras where technology seemed to progress rapidly but workers’ wages didn’t grow much, like 18th century Britain, Acemoglu and Johnson argue that the main cause was technologists inventing machines that replaced human labor; in other eras where wages grew rapidly, like the late 19th century, they argue that technologists were inventing machines that created new tasks for humans to do.

At times this can begin to feel like a just-so story. For example, Acemoglu and Johnson cite electricity as a technology that was good for workers, because it created so many new industries for people to work in. But didn’t electrification also replace human labor in quite a large variety of ways? Electric lights save us the labor of making candles, electric dishwashers and washing machines and dryers automate our housework, and so on. How do we know this task automation outweighed by the new tasks electricity creates?

They also argue that wages in industrial Britain began to increase in the late 19th century because steamships and telegraphs — as opposed to looms — “expand[ed] the set of tasks and opportunities for workers”. As far as I can tell, this is a claim without evidence. Why was the telegraph’s automation of message couriers less significant than its creation of new jobs for telegraph operators? In Chapter 8, Acemoglu and Johnson blame communication technology for increasing inequality by enabling the offshoring of jobs to China and other countries. Why would modern communication technologies have exactly the opposite effect of the telegraph?

And in Chapter 6, Acemoglu and Johnson praise the early United States for its direction of innovation, writing that American businesses compensated for a lack of skilled labor by, in the words of engineer Joseph Whitworth, “call[ing] in the aid of machinery in almost every department of industry.” In Chapter 7 the authors write that the American use of interchangeable parts “was first and foremost an effort to simplify the production process so that workers lacking in artisanal skills could produce high-quality products.” But how is that any different from the British use of power tools to help unskilled make textiles in the 1700s? The difference is never explained.

On the other side of the coin, Acemoglu and Johnson cite most modern information technology as something that automates more tasks than it creates. But what about all the new tasks that IT creates — mobile developers, web designers, digital media marketers, content moderators, and so on and so forth? Why are these less economically important than the tasks that the internet automates away (encyclopedia salespeople, etc.)?

One answer is that if we assume that Acemoglu’s theory of automation is the main thing that’s going on, we can just infer the effects of particular technologies from macroeconomic outcomes — if we see wages stagnate, it must be because task automation outweighed task creation. But for anyone who suspects that Acemoglu’s theory might not actually be the main thing going on in the economy, just saying that the proof is in the pudding is a bit unsatisfying. It feels like a just-so story.

Is productivity good or bad?

In fact, the historical examples in Power and Progress leave themselves open to alternate narratives. The main alternative narrative is about productivity.

Acemoglu and Johnson repeatedly argue that if productivity gains are produced by automation, workers don’t see the benefits. They cite the idea that productivity naturally uplifts workers — which they call the “productivity bandwagon” — as one of the main nefarious narratives that technologists use to persuade society to allow them to invent technologies that replace workers.

But what if the “productivity bandwagon” narrative is true?

There are two main historical periods the authors cite as examples of excessive automation leading to stagnant wages — the early Industrial Revolution in 18th century Britain, where textile machines like looms replaced human artisans, and America since the 1970s. (Note: they make an error when they say, in Chapter 8, that “declines in real wages…have been a major part of U.S. labor market trends.” In fact, when you include benefits, real hourly compensation has grown a bit more slowly since 1973, but has still consistently risen.)

But Acemoglu and Johnson also note that both of these eras had sluggish productivity growth. Perhaps wages were stagnant in those eras because productivity was also stagnant?

Regarding the early Industrial Revolution, some researchers argue that the labor share didn’t actually fall. For example, here’s Crafts (2020):

[R]eal wages grew more slowly than real GDP per worker during the industrial revolution. However, the discrepancy was much less than has been claimed such that in 1820 the former had risen by about 12 per cent since 1770 and the latter by about 16 per cent. Second, labour productivity grew quite slowly prior to 1830 averaging a little below 0.4 per cent per year in the 60 years after 1770. Nevertheless, in the context of demographic pressure this was a very good outcome by pre-industrial standards. Third, as relative prices changed and exportable manufactures became cheaper, over the long run real product wages grew somewhat faster than real consumption earnings. Fourth, the share of profits in GDP rose over time from 17.2 per cent in 1770 to 31.3 per cent in 1860 but this was associated with a decline in the share of land rents and the share of labour was little changed. Fifth, looked at through the lens of growth accounting the evidence is of total factor productivity (TFP) growth accelerating only gradually to 0.6-0.8 per cent per year during 1830 to 1860 with the steam age only materializing after 1830.

In sum, this looks more like a story of paradoxically slow productivity growth than of pro-rich growth. The story of the industrial revolution is definitely not one of a new general-purpose technology boosting productivity growth at the expense of a big shift in the distribution of income which is the current fear about AI.

As for the U.S. since the 1970s, inequality has definitely increased, but — contrary to what you may have heard — pay has largely kept pace with productivity. Automation may have made wages more unequal (this is the argument of Acemoglu and Restrepo’s 2022 paper), but the modest decline in labor’s share of the national pie was probably mostly about land values increasing. Which means that aggregate wage stagnation was largely due to slowing productivity growth in recent years as well.

In fact, Acemoglu and Johnson also blame automation for stagnating productivity! In Chapter 8, they write that “productivity gains from automation may always be somewhat limited.” They coin the term “so-so automation” to describe technologies that take humans out of the loop but fail to increase productivity much by doing so. They argue that technologies that make better use of human capabilities lead to faster productivity growth as well as higher wages and lower inequality.

OK so…why isn’t that the book’s central this? You could write a very interesting book about how technologies that complement humans are better for both productivity and broad-based prosperity than technologies that try to substitute for humans wholesale. I would definitely read that book! But Acemoglu and Johnson did not choose to write that book; instead, they warn against a focus on productivity, claiming that it’s a seductive but dangerous narrative used by the greedy, fast-talking techbros. In my mind, this weakens their overall narrative.

Where’s the menu?

Throughout Power and Progress, the authors tell a story about a “menu of technologies” that entrepreneurs can choose from. On one hand, companies can choose to invest in automation, replacing workers, increasing inequality, causing slow wage growth, and maybe reducing productivity in the process. On the other hand, they can choose to invest in technologies that create new tasks for humans to do, thus increasing wages and decreasing inequality. Their story is that out of greed and/or elitism, entrepreneurs often choose the former, so it’s in the interests of society to push them toward choosing the latter.

But when the authors approvingly cite examples of new industries springing into being, they never give an explanation of why entrepreneurs and technologists chose to create these new industries, instead of trying to cut costs in existing industries. My default assumption would be that the people who invented and commercialized steamships, telegraphy, interchangeable parts, autos, electricity, and telephones were driven by the same sort of motivations that animated the people who invented and commercialized power looms, computers, and the internet. If not, why not? Was there ever a case when governments or unions pushed entrepreneurs to select Option A from the “menu of technologies” instead of Option B?

When Acemoglu and Johnson do discuss union power, it’s in the context of worker training. In Chapter 7, they write:

In fact, for unions [in the 1960s] the central issue was worker training. They insisted on training provisions to ensure that workers could be brought up to the necessary skill level to operate the new machinery and benefit from it.

This is very different from affecting the direction of innovation! This is a case of workers collectively pushing companies to invest in human capital, so that worker skills can catch up to the direction in which innovation was already going.

As far as I can tell, this book does not contain even one single example of when a union or government supposedly pushed an entrepreneur or company to choose a different path of technology in order to benefit workers. As far as I can tell, it does not even contain one single example of when an engineer, entrepreneur, company or investor chose to create a technology in order to benefit workers more.

In other words, there is no evidence here that the “menu of technologies” actually exists. It’s not clear that technologists and industrialists even know in advance whether the inventions they create and commercialize will create more new tasks than they automate. And this raises pointed and troubling questions for the authors’ preferred solution to the problems of inequality and wage stagnation.

Can the mandarins act as a check on the techbros?

In Chapter 11, Acemoglu and Johnson roll out their proposed solutions. Having concluded that inequality and wage stagnation are due to “tech billionaires and their agenda” choosing the wrong technologies from the “menu”, they call for democratic people-power to force the techbros back onto the labor-augmenting path.

What’s totally unclear is how to do this. Acemoglu and Johnson admit that “redirecting” the path of technological innovation is going to be an incredibly tall order:

Determining how different digital technologies are used and their impact on wages, inequality, and surveillance is much harder [than assessing their climate impacts]…Moreover, given the difficulty of distinguishing automation from other uses of digital technologies, automation taxes are currently not practical.

And yet the authors still claim that this can be done! Yet they’re maddeningly vague on the details:

There is a telltale sign of automation technologies: reducing the labor share of value added, meaning that once these technologies are introduced, how much of value added goes to capital increases and how much gets captured by labor decreases…On this basis, technologies that increase the labor share can be encouraged via subsidies for their use and their development.

But how do we know in advance, before a technology is invented, whether it will increase or decrease the labor share? This is just replacing one target of guesswork — new task creation vs. automation — with another target of guesswork.

Fundamentally, it still boils down to some sort of mandarins in a room somewhere — economists? government engineers? bloggers? — trying to assess the economic effects of a technology that doesn’t even exist yet.

As I wrote in a post last June, this is probably an impossible task. Some of the world’s top experts thought that AI would replace radiologists within a few years, but demand for radiologists boomed even as the new AI tools were coming online. The technologists got it wrong.

And the economists are just as likely to get it wrong. For example, industrial robots are the one technology that Acemoglu consistently rails against as an example of harmful automation. That’s based on his 2020 paper with Restrepo, where they found that companies that buy more robots employ fewer humans. But a whole bunch of other economists followed up on this research and found the exact opposite — robot adoption is correlated with more jobs at a company or in an industry. Here was a list I made back in 2022:

I’ll list a few of these studies:

1. Mann and Püttmann (2018) — Where Acemoglu and Restrepo…looked at correlation, this paper attempts to identify causation. They look at automation-related patents in an industry — a proxy for innovation in the automation space — and then look to see whether that industry gains or loses jobs. They find that “advances in national automation technology have a positive influence on employment in local labor markets”, though this isn’t true for every area.

2. Dixon, Hong and Wu (2021) — These authors looked at robot adoption in Canada, at the level of the individual company (or “firm”, as economists say). They found that companies that adopted more robots hired more people, while also improving the quality of their products and services.

3. Koch, Manuylov and Smolka (2019) — This paper looks at firm-level data for manufacturing companies in Spain. They find that robot adoption is associated with a substantial increase in employment as well as output.

4. Adachi, Kawaguchi and Saito (2020) — This paper finds the same thing as the previous one, but for Japanese companies over the course of a 40-year period.

5. Eggleston, Lee and Iizuka (2021) — These authors look at robot adoption by nursing homes in Japan, and find that it strongly increases employment, although it does result in existing nurses working fewer hours (and thus getting paid less).

6. Hirvonen, Stenhammar, and Tuhkuri (2022) — This paper looks at a technology subsidy program in Finland that increased adoption of a broad range of advanced technologies at Finnish firms. They find that this led to employment increases.

In fact, by this point the trend is clear. Essentially everyone is finding that, contra Acemoglu and Restrepo…robots are correlated with — and probably cause — higher employment in the companies and areas where they’re adopted.

What’s happening is that companies that use more robots hire more humans (and retain their existing humans) in jobs that complement the robots. That’s exactly what we saw with previous waves of automation — people find new roles, robots increase their productivity, and they get paid more. Looking at the countries that use the most robots in their manufacturing industry, it seems likely that this virtuous cycle is happening even at the level of whole nations.

Zooming out from just the manufacturing sector, Hötte, Somers, and Theodorakopoulos have a very interesting 2022 review paper in which they look at the literature on the entire range of automation technologies. Here’s an article in which they explain their results. Hötte et al. find that automation does replace jobs, but that this effect is outweighed by the “reinstatement” effect — in other words, people find new jobs to do. And their incomes generally rise as a result.

So no, there’s no possibility that a council of mandarins — engineers, economists, or whoever — can sit there evaluating every potential new technology that companies or inventors want to create, and deciding whether it will raise or lower the labor share. I mean, you could make a council of mandarins, and it could look at plans for new technologies, and it could issue decisions, but in practice it would be throwing darts at a dartboard. And it would be an incredibly costly tax on our companies, since it would introduce massive delays into their decision-making process. Their Chinese rivals, on the other hand, would suffer no such delays.

In other words, I see no hope for Acemoglu and Johnson’s preferred solution. The utter vagueness with which the idea is presented in Power and Progress doesn’t suggest that the authors have thought carefully about how this solution might work in practice.

In particular, these solutions seem inferior to something far simpler: policies to increase labor share ex post. Labor market institutions like co-determination and sectoral bargaining, and direct interventions like wage subsidies funded by taxes on capital income, can push up the labor share without requiring panels of experts to predict the unpredictable. And if entrepreneurs really do have any degree of foresight about whether their innovations will tend to push the labor share up or down, these policies will act like a Pigouvian tax on the kind of cost-cutting that Acemoglu and Johnson decry. With a wage subsidy, for example, the higher the market rate you can afford to pay your workers, the more of a rebate you can get from the government. So if there are technologies that augment your workers and let you hire new workers, a wage subsidy gives you an incentive to create them.

Anyway, I think simple policies like these should be economists’ first go-to solutions, rather than the creation of whole new social institutions.

The old narrative, and the new

At the start of this review, I talked about how Power and Progress may have missed its moment by getting lost in a flood of fears about AI. But there’s another way in which the book might be poorly timed. Wage inequality — the very thing that Acemoglu and Restrepo (2022) try to explain — has flatlined since the early 2010s.

Meanwhile, real wages have been rising strongly for years now, interrupted only by the post-pandemic inflation. And wages for production and nonsupervisory workers have risen more robustly than those for managerial workers:

Meanwhile, employment for prime-age Americans is near all-time highs, and unemployment is at record lows; everyone who wants a job in America has one.

All this has happened in exactly the time frame during which AI has exploded. Predictive AI burst onto the national scene in 2012 with the ImageNet paper, the basic technology for generative AI was created in 2017 with the transformer paper, and generative AI became really widespread in 2022-23 with LLMs and AI art programs. To reiterate: essentially all of the commercialization and implementation of artificial intelligence has happened during a time in which wages have been rising, inequality has been flat or falling, and employment has been high.

Maybe AI just isn’t big enough to kill all the jobs yet; maybe we just have to wait a few years and we’ll all be unemployed or working for pennies. Or maybe AI is actually the kind of technology that improves task productivity and creates new tasks instead of automating old tasks away. Or maybe Acemoglu has something wrong in his models, and further theoretical and empirical explorations will overturn his conclusions about the key role of automation in fostering inequality. I’m not sure.

Whatever is going on, though, I think it should give the AI worriers pause. This was not on the menu. If you went back to 2012, and asked people to predict the impact of a new machine that could recognize objects and imitate speech and create beautiful art, they probably would have assumed that the rising inequality that they had experienced for the last 30 years would now be turbocharged. And — at least so far — they’d have been completely wrong.

To me, that thought experiment illustrates the folly of trying to predict the economic effects of new technology. It also suggests another reason why Power and Progress didn’t make the same kind of splash it might have made in 2018. Obviously, our economic problems haven’t all been solved. But perhaps, underneath all of the anger and pessimism, Americans realize that something has shifted in their economy, for the better. And perhaps that’s making them a bit less interested in the kind of pessimistic economic narratives that flew off the shelves in the 2010s.

Subscribe now

Share

“We must do something. This is something. Therefore, we must do this.” — the Politician’s Fallacy

The other day, a friend asked me to add my signature to a statement called “We Must Act Now: A Statement on AI’s Transformation of the Economy”. A bunch of economists, including many famous and influential ones, have been signing it. Here’s the text of the statement:

1. AI may become radically more powerful over the next 10 years.

2. This could drive an unprecedented transformation of our economy, larger than the Industrial Revolution, but unfolding over a vastly shorter time frame. It could bring risks, including large-scale job displacement, as well as opportunities such as major gains in living standards.

3. Economists, policymakers and technology leaders must act now to understand the economics of transformative AI and to build the incentives, guardrails, and institutions needed to steer AI in a direction that complements humans and benefits society.

That’s it. That’s all it is. It doesn’t say what our action ought to be, only that “we must act”. There’s no appendix, no longer manifesto attached below. It just says AI is getting good, AI could be economically important, AI could take people’s jobs and/or make us a lot richer, and that we have to do something to make sure the AI age turns out alright.

But what is that something? What actual policies would I be recommending by signing this statement? None that I can see. It’s completely vague and unspecific.

This might seem like it makes the statement innocuous and bland (so why not sign it?). At some point, however, the authors may decide to release a second statement, with policy specifics. I’ll inevitably be associated with those ideas, even if I don’t sign the second statement. Relatively few people will pay attention to the difference between who signed only the first statement and who signed both. So by signing this first statement, I would essentially be giving my imprimatur to unknown policy proposals. I don’t want to do that. So I didn’t sign.

In fact, the existing statement, vague as it is, does contain at least one clue as to the kind of ideas that the authors will eventually come up with. At the end, it calls for us to “steer AI in a direction that complements humans”. I recognize this as the main idea in the book Power and Progress, by Daron Acemoglu and Simon Johnson.

In fact, not only did Acemoglu sign the statement, but it appears that the authors changed the text in order to get him to sign! He writes:

Why did I sign this statement?…First, I had a hand in revising it, after the organizers reached out to me. I did not feel like I could sign the initial version…[M]ost importantly, I wholeheartedly agree with the ending: “to build the incentives, guardrails, and institutions needed to steer AI in a direction that complements humans and benefits society.”

This is what I have been arguing for over a decade now. Good AI needs to complement humans, and this requires a redirection, because the current focus on AGI is, in all but name, an agenda for displacing humans from meaningful work. That’s why steering AI must be a first priority.

So although Acemoglu doesn’t say which part of the final text he got to insert in exchange for his signature, it’s a pretty good bet that it’s the part about “steering” AI.

As it happens, I think “steering” AI is a bad policy idea. The first reason is that it’s basically impossible; no one actually knows how a technology will complement or substitute for human labor at the time they invent it. Inventors don’t know how their inventions will ultimately be used by businesses — and the more general-purpose a technology is, the less they know. Could James Watt, in 1765, have predicted most of the applications of steam power? Absolutely not. So he had no way of knowing whether the steam engine would ultimately create more jobs than it destroyed.

In fact, although AI might eventually be a big job-destroyer, right now it doesn’t seem to be. The employment rates for people age 20-24 and 25-54 are both just about the same as they were before ChatGPT ever existed:

And the employment rate for young college grads — the group everyone thinks is most likely to be hurt by AI — is also basically unchanged:

So if there’s any wave of AI job destruction, it’s not visible in the macro data yet. As for the micro data, there are a few studies that show companies reducing their hiring of certain kinds of workers when they adopt AI, but most don’t really find much. In fact, one recent study found that companies that adopt AI hire more workers than other companies in the same industry:

This is despite the fact that lots of people in the AI industry think their inventions are going to destroy jobs. So far they’ve just been wrong, and many of them are feeling pretty astonished right now.

Even when it comes to specific occupations, technologists are often startlingly wrong on the “complement or substitute” question. Geoffrey Hinton, one of the inventors of modern AI, famously predicted the end of human radiologists within a few years, only to see a boom in hiring and salaries for radiologists when it turned out that AI actually complemented their skills.

So how the heck are businesspeople and inventors supposed to “steer” AI toward being complementary to human workers? They obviously couldn’t predict the labor market effects of the last round of AI — at least, in the short term. So why should anyone believe that technologists have the ability to purposefully invent different forms of AI with different labor market effects?

The second problem with the idea of “steering” AI is the question of who does the “steering”. Acemoglu’s book, Power and Progress, never answers this question. Here’s what I wrote in my (very long) review of that disappointing book:

Acemoglu and Johnson admit that “redirecting” the path of technological innovation is going to be an incredibly tall order…[H]ow do we know in advance, before a technology is invented, whether it will increase or decrease the labor share?…Fundamentally, it still boils down to some sort of mandarins in a room somewhere — economists? government engineers? bloggers? — trying to assess the economic effects of a technology that doesn’t even exist yet…[T]his is probably an impossible task.

Acemoglu himself has certainly not had a better record than the technologists when it comes to predicting the effects of AI on jobs. He wrote an empirical paper claiming that companies that buy robots tend to hire fewer workers, but this paper was contradicted by a very large number of follow-up studies. And he wrote a theoretical paper claiming that AI wouldn’t do much to raise productivity, but that prediction was based on arbitrarily assuming away parts of his own model.

So any panel of wise mandarins that Acemoglu and his fellow-travelers assemble in order to “steer” AI technology is likely to have absolutely no idea what they’re doing. Here’s what I wrote about that idea back in 2023:

[I]f we were to set up a panel of experts and task them with deciding which lines of research and innovation to encourage and which to discourage in order to maximize jobs and wages, they would be operating purely on gut instinct and quasi-science-fictional supposition…[I]n practice, any panel or commission set up to speed up and slow down various types of AI will be simply adding noise to the innovation process, offering rewards and punishments essentially at random. That’s not good for the development of technology as a whole, since it introduces uncertainty into the innovation equation. But it’ll also be ineffectual in terms of actually protecting human workers.

Three years later, having witnessed so many of the dire predictions of job destruction dashed on the rocks of reality, I see absolutely no reason to change my assessment. Acemoglu’s big idea — basically, to put him and his friends in charge of AI development — is not a good idea.

In fact, the surprisingly benign effect of AI on jobs so far calls into question the very notion that “We Must Act Now”. Yes, I agree that AI presents a very severe security threat, and we must act on that. But on the economic front, it’s possible that inaction is the right move.

Statistically speaking, we probably don’t live in the best of all possible worlds when it comes to AI’s economic effects. But we might be close enough that any large-scale attempt to interfere in AI development might leave average human workers worse off. There’s certainly plenty of historical precedent for that — collectivization of agriculture, Mao’s “backyard production”, and a bunch of other heavy-handed interventions in the development of an entire sector crashed and burned spectacularly.

It might be, in other words, that the AI we’re building now is already highly complementary to human workers, and that the best approach is not to “Act Now”, but to simply sit there and do nothing. Even the seemingly empty slogan of “We Must Act Now” might actually be wrong. Perhaps we mustn’t.

In any case, if the authors of the “We Must Act Now” statement add specificity to their policy proposals, I’ll consider signing it. But right now, the statement seems to hide some genuinely inadvisable Acemoglu-ism behind a screen of extreme vagueness.

Subscribe now

Share

“Buy a big house and live in the suburbs” — Tracy Chapman

“Outside suburbia’s sprawling everywhere/ I don’t want to go, baby” — Kim Wilde

I grew up in the suburbs, and when I got the chance, I moved to a big city and never looked back.1 I love living in dense, built-up urban areas, with great train systems and tons of restaurants and shops. Japanese cities are the best in the world, and I’ve written plenty of posts about what makes them so great. But you don’t have to be Japan in order to create amazing metropolises — New York City, Paris, Istanbul, Seoul, London, etc. are all excellent places to live.

I’m far from the only person who feels this way. Rents in New York City are absolutely insane — $5300 a month to live in Manhattan, $4350 to live in Brooklyn. That’s partly because there are a lot of good jobs in NYC — it’s a cluster for industries like finance and media. But more and more, Americans move to cities because they like living there.

As early as 2000, economists were starting to find that “amenities” were driving America’s urban revival even more than job opportunities were. Couture and Handbury (2020) find that wanting to be close to restaurants and nightlife explains about 40% of the trend of young, high-earning, college-educated people2 moving to cities in recent decades. Furthermore, the stereotype of dense cities as crime-ridden and unsafe is just wrong — NYC has one of the lowest violent crime rates among big cities in America.

I’ve been a relentless advocate of building more dense, walkable cities in America. Not only would this raise GDP (because of improved clustering effects), but it would let Americans live where they want. The demand for life in cities like NYC exceeds America’s willingness to supply these environments; this raises rents in places like NYC, which pushes a lot of people into the suburbs who don’t want to be there. Forcing those city types into the ‘burbs raises rents for people who like suburbia. Basically, everyone would be happy if America had a few more Manhattans and a lot more Brooklyns.

Yet among my fellow urbanists and YIMBYs, I often encounter disdain or outright hostility toward the suburbs that define most of America’s present urban landscape. This isn’t just an urbanist thing, of course — my own parents had a lot of negative things to say about suburbia, and ranting against its sterility and boredom is a staple of pop culture. But the criticisms are just way overdone; the suburbs are not the isolating, lonely hell that they’re often made out to be.

And I think that the constant ranting against the suburbs complicates the quest for denser cities. It creates the suspicion that urbanists and YIMBYs want to make the whole nation into Manhattan. Nothing like that could ever happen, of course; even Japan is mostly suburbanized. But painting the quest for denser metropolises as an attack on suburbia makes everything needlessly confrontational, polarized, and zero-sum.

So I think it’s helpful to go through some reasons why the suburbs aren’t actually as bad as they say.

The suburbs don’t force you to have a long commute

One of the most persistent myths about suburbia is that it forces you to commute a long way to work:

It’s possible, of course, for something like this to be true. If cities refuse to build housing (which they do), and if jobs are concentrated in the city center (thanks to clustering effects or agglomeration or whatever), then people will be forced to live far out on the periphery and endure punishing commutes to get to work.

The thing is, it’s not true. Despite greater sprawl, Americans have some of the shortest commutes in the developed world. This is from the OECD:

Why do Americans take less time to commute to work? One reason is that it’s not just our homes that are scattered and dispersed; our workplaces are too. America’s urban agglomerations are polycentric; they’re not just one central business district surrounded by concentric rings of houses. Suburbanites tend to live near where they work.

Another reason Americans have shorter commutes is that cars are almost always faster than public transit at getting you from point A to point B. This isn’t just true in places like America and Canada that are built with cars in mind. It’s true in places like Stockholm and Amsterdam that were built to be transit-friendly — even when you take parking time into account. In fact, it’s not even close. This is from “Disparities in travel times between car and transit: Spatiotemporal patterns in cities”, by Liao et al. (2020):

We use real-world data to make realistic estimates of travel time by car and by PT [Public Transit] and compare their performance by time of day and by travel distance across cities. Our results suggest that using PT takes on average 1.4–2.6 times longer than driving a car. The share of area where travel time favours PT over car use is very small: 0.62% (0.65%), 0.44% (0.48%), 1.10% (1.22%) and 1.16% (1.19%) for the daily average (and during peak hours) for São Paulo, Sydney, Stockholm, and Amsterdam, respectively…A systematic comparison between these two modes shows that the average travel time disparity is surprisingly similar across cities…for travel distances less than 3 km, then increases rapidly but quickly stabilises at around 2. [emphasis mine]

And here’s a chart, showing that public transit takes more time whether you’re measuring in terms of the length of the trip or the percent of the population reached:

That doesn’t mean cars are better than public transit. Cars cost more, and they require a lot more land to move the same number of people. But because they take you directly from point to point, instead of making a circuit and stopping periodically, they get you there faster.

Of course, commuting by car and commuting by train or bus aren’t the same experience. Driving gives you more privacy, but it forces you to pay attention to the road instead of reading or playing games. Traffic can be frustrating, but so can jostling and bumping strangers for a space on the train. You’re a lot less likely to be sexually harassed in your car, but you’re more likely to die in an accident than to be murdered on the bus. And so on.3

But the point is, cars are not a cost forced upon suburbanites in exchange for their large houses, as some urbanists believe. They are a thing people want in and of themselves, and are willing to pay a lot of money for, all over the world. The convenience, sense of freedom, and privacy cars offer is a benefit of suburbia to many people, in addition to the large house and cheap land. Car ownership is a form of wealth.

The suburbs are not lonely and isolating

Again and again, I hear urbanists declare that the suburbs are lonely and isolating. It certainly sounds logical. In a city, you’re walking past other people constantly — on the street, on the train, in cafes and restaurants. In suburbia, you’re shut at home inside your giant house or alone in the metal shell of your car. How could suburbia not be more lonely than the big city?

But in survey after survey, we don’t find this to be the case. Here’s Abshire et al. (2022):

Data were obtained from 616 adults (278 from small rural, 100 from large rural, 98 from suburban, and 140 from urban areas) from June 2018 through October 2019…Mean unadjusted loneliness scores were lower in suburban compared to urban areas…The prevalence of loneliness was 50.7%, 59.0%, 40.8%, and 54.3% in small rural, large rural, suburban, and urban areas, respectively. Suburban living was associated with lower odds for being lonely compared to urban living…but this association was not statistically significant in the adjusted model[.] [emphasis mine]

And here’s Hammond et al. (2021):

Data from 756 participants who completed 16,602 assessments between April 2018 and March 2020 were used in order to investigate associations between momentary feeling of loneliness, the social environment (i.e. overcrowding, social inclusivity, population density) and the built environment (i.e. contact with nature)…Increased overcrowding and population density were associated with higher levels of loneliness; in contrast, social inclusivity and contact with nature were associated with lower levels of loneliness. These associations remained significant after adjusting for age, gender, ethnicity, education and occupation. [emphasis mine]

And here’s Morris and Pfeiffer (2016):

Based on data from the 2003 to 2013 American Time Use Surveys, this research…assess[es] whether suburban living is associated with less socializing than city living in mid-to-large American metropolitan areas. After controlling for personal characteristics, we find no meaningful difference in suburbanites’ and city dwellers’ time spent socializing across a wide range of social activities. [emphasis mine]

Bower et al. (2022) review 57 different studies and find no systematic association between loneliness and any measurable feature of the built environment.

Why doesn’t the simple intuition work here? Probably because people aren’t just like particles bouncing around in a chemistry experiment — they don’t simply form human connections and bonds just because they happen to walk past each other. A few relationships form from random urban conversations, but most form through work, or friends-of-friends, or shared hobbies, etc.

And while meeting a ton of new people is fun, what really gets rid of loneliness is repeated interaction with people you know and care about. Imagine going to ten parties filled with strangers versus having two close friends over for dinner. Which of those is more likely to leave you feeling lonely? What about 100 Hinge dates versus having a relationship? Remember that Japan has the world’s best cities, but struggles with widespread loneliness.

In fact, suburbs have some features that make it easier to interact with the people you really care about. Those big houses aren’t just for walking around all alone and going “Wow my house is so huge”. They’re for entertaining guests. It’s harder to have a dinner party or a TV night or a game night at a tiny little Manhattan apartment than at a big suburban McMansion. Cars help too; those short travel times make it easier to just pop over and hang.

This doesn’t mean cities are socially inferior to suburbs. Constantly meeting new people is exciting. There’s more fun stuff to go out and do with your friends in a city — restaurants, parties, and so on. It’s just a different lifestyle.

There’s a reason people are moving to the ‘burbs

The urban revival of the 1990s through the 2010s was a modest thing. It was largely driven by young people, high earners, and educated people. Here are some charts from Jed Kolko a decade ago:

During most of that urban boom, it was actually the suburbs that were growing much faster. And in recent years, the trend toward suburbanization has only accelerated. Here’s a much more recent chart:

Kolko shows that although the densest city centers are rebounding from the pandemic, more than 100% of this is driven by immigration — domestic migration is still strongly away from city centers.

In fact, the new suburbanization trend is being led by Millennials. This is from the Joint Center for Housing Studies:

[W]e found that throughout the past decade millennials were moving to suburbs that were farther out from the city center. There are many ways to define suburbs, and in this paper we rely on a framework that considers rates of homeownership, single-family housing, and car commuting, in addition to proximity to a metro’s urban core. While there is extensive research and discussion about millennial preferences for walkable urban areas, we found that the places with the largest increases of early millennials were both suburban and on the periphery of metropolitan areas. [emphasis mine]

Why are Millennials moving out? Part of it — the part that urbanists and YIMBYs will emphasize — is that they’re being pushed out by higher rents, which are a result of cities failing to build more housing. But that’s not the whole story. Millennials are also being pulled to the suburbs, because suburbs are generally better for raising kids.

Albouy and Faberman (2025) find that the kind of high-skilled workers who drove the urban boomlet of the 1990s through the 2010s tend to move out — and value urban amenities less — once they get a little older:

We show that high-skill workers disproportionately sort into high-amenity areas, but do so relatively early in life. Workers of all skill levels tend to move towards lower-amenity areas during their thirties and forties. Consequently, individuals’ time use and expenditures on activities related to local amenities are U-shaped over the life cycle…We present evidence that the move towards lower-amenity (and lower-cost) metropolitan areas is driven by changes in the number of household children over the life cycle: individuals, particularly the college educated, tend to move towards lower-amenity areas after having their first child. [emphasis mine]

Kolko shows something similar — people with a kid over 6 tended to move out of cities during the urban boomlet, even as childless people and people with young children were moving in:

Anyone with kids can easily rattle off the reasons why suburbs make it easier to raise kids. A big house means more room for kids to have their own bedrooms, space to play, and a back yard to run around in safety. A car makes it a lot easier to ferry kids around to school, or soccer practice, or wherever. Cars also make it much easier to do large grocery shopping trips — try taking a week’s worth of food for a family of four home from the store on foot or on a bike and you’ll quickly understand.

It’s almost a cliche that people move to the city when they’re young — to jump-start their career, to party, to meet friends, to explore the world, to date around, and to find their spouse — and then move out to the ‘burbs once they settle down and have kids. The cliche is rooted in reality.

Instead of attacking suburbia, just make room for everybody

Instead of treating the suburbs as some sort of hellish place that Americans’ car culture has exiled them to, urbanists should recognize that suburbs have real advantages that dense urban cores can’t easily replicate. Yes, too many Americans are pushed out to the ‘burbs by unaffordable housing. But a lot move there of their own free will, because they want a nice place to raise kids, enjoy a short quiet drive to work, and hang out with their friends in a big comfy house. It’s not the lifestyle for everyone, but it’s the lifestyle that a lot of people love and aspire to.

Which doesn’t mean we should ignore the real disadvantages of suburbia, either. Low-density sprawl is very expensive to maintain. It tends to make people less healthy, because they walk less. It reduces variety among restaurants and brick-and-mortar retail outlets, because it’s harder to cater to niche tastes when you don’t have a critical mass of people nearby. Those tradeoffs shouldn’t be ignored.

Nor should we accept that the current suburban form is the optimal one. A lot of American urbanism has focused not on Manhattanizing urban cores, but on giving suburbs a few more of the benefits of cities — creating “gentle density” with rowhouses and duplexes and small apartment buildings, allowing retail in residential areas, adding bike lanes and commuter rail, and so on. This should all continue.

But it’s simply a mistake to frame the question of urban form as “Which is better, cities or suburbs?”. The answer is that they’re good for different things, and they appeal to different sets of people. Instead of fighting flame wars over whether the whole country should look like Manhattan or the Inland Empire, we should aim for a country that has room for everyone. We should build great city centers for the same reason we should build great suburbs — because Americans deserve to have great places to live, no matter what kind of place they want to live in.

Subscribe now

Share

I grew up in the 1980s in a small house with only one bathroom shared between four people. The floor was linoleum. There was a carport instead of a garage, and we had one beat-up used Toyota Tercel hatchback. I don’t remember when we got our first color TV, but when I was young we had a black-and-white one that my grandmother gave us. Our furniture was all second-hand and we kept the couches covered up with worn old blankets.

When I was young, I mowed lawns for money. As a high school kid, I signed up to pick cotton by hand (!!) for an agricultural research project at Texas A&M University, for minimum wage1. I have also worked as a cashier. Twice in my life, I have been a member of a labor union, and I have marched in a strike.

I have never once considered myself part of the working class.

Why not? Because I have never thought of class as being defined by a present snapshot of someone’s lifestyle or material circumstances. Instead, I always thought of class as being about someone’s potential. And I grew up always knowing that my economic potential went far beyond the rather humble circumstances of my early childhood.

For one thing, my family was upwardly mobile. My grandparents could probably be called “working class” in their youth — my grandmother worked in a sweatshop as a teenager, my grandfather wore cardboard in his soles because his family couldn’t afford shoes. But after World War 2, thanks to the GI Bill and rapid economic growth, my ancestors advanced into the middle class, with jobs like optometrist, athletic coach, and registered nurse. My father had a PhD and a tenure-track academic job that promised to pay a lot more after a decade of work. We weren’t rocketing up the income distribution, but we were clearly climbing.

Our humble lifestyle in the early and mid 1980s reflected this future orientation. Our family income was probably around the 35th-38th percentile,2 but this was because we were a one-earner family. My mother chose to spend the first seven years of my life as a housewife — which she did in order to make sure my sister and I got a thorough, accelerated education. We lived an abstemious life in part because we saved as much money as we could.

What were we saving for? My college education, and my sister’s. We were smart kids; we knew we would go to good schools, and we did.3 We knew our college educations would allow us to get good jobs that paid more than our parents ever made. And we were right.

As for the union membership, I was part of the grad student instructors’ union at the University of Michigan, and the professors’ union at Stony Brook. When I marched in a strike in 2008 to secure a raise and health benefits, I was already getting paid to complete a PhD that would eventually increase my earning power even more. Even as I scarfed free food from charcuterie boards at departmental events to save money, I was building up my future earning power at a rapid clip.

Class in some countries is about the past; you can be a shabby aristocrat if your grandfather was the Earl of Whatevershire. In American policy discussions, class is often implicitly about the present — where you lie in the income and wealth distributions this year. But on some level, everyone knows that class in America is really about the future.

Milton Friedman had a theory that sort of gets at this idea, in fact. It’s called the Permanent Income Hypothesis. “Permanent income” is the income you can expect to make over the course of your life. If you’re a shabby grad student living off of cup ramen, your current income is low, but your permanent income is high, because you know you’re probably going to make a lot of money in the future.4

But class in America isn’t just about the money you will make in the future; it’s about the money you could make if you wanted. I know schoolteachers who live modern middle-class lifestyles despite having graduated from the best schools in the country. They could have gone to work for companies and made decently big bucks, but they preferred a more laid-back lifestyle. Whether their children should count as middle class or upper class is an interesting question, but they themselves are clearly the American equivalent of Europe’s shabby aristocrats, because they forsook the upper-class lifestyle voluntarily.

Meanwhile, there are millions upon millions of Americans for whom working in high-paying salaried jobs was just never an option, and never will be. They will spend their entire careers driving long hours in a truck, or stocking shelves at a store, or installing smoke alarms in people’s houses, simply because this is the best they can do.

This explains at least part of why most low-income Americans traditionally consider themselves middle class. They expect to be middle-class at some point; they don’t think they’ll be trapped driving a truck or stocking a shelf forever.5 And it could also explain why the number of Americans calling themselves “lower class” rose in the 2000s and 2010s, as growth in incomes temporarily stagnated and the potential for rapid downward mobility became clearer after the financial crisis:

I’m being very approximate here, of course, and I’m speaking for the country more than I probably should. In fact, America has less class consciousness than many other societies, and when we do talk about the idea, there’s rarely agreement on what it should mean. I wrote about the contested nature of class in American society two years ago:

But I’m speaking from my own personal experience because I think it illustrates something important about class in America. Although we disagree about what the concept should mean, most of us feel deeply uncomfortable with a notion of class that ignores future possibilities. When we see a guy proudly wear a United Auto Workers jacket even though his only UAW experience was as a grad student instructor at Harvard University, something about it feels deeply inauthentic:

There’s nothing wrong with this, of course. It’s not fake; the UAW really does include grad student unions these days. And MacKay could easily be doing this not out of rank political opportunism, but from a sincere desire to express solidarity with union workers from all walks of life.

And yet Evan MacKay is really not in the same boat as people who stand on an assembly line, even if those people make more money than he does. He could, if he wanted, go work in private equity and live in a mansion on Cape Cod. Your typical UAW member could not do that. There’s a sense in which his jacket’s implicit claim of “we’re all in this together” papers over that harsh reality.

Progressives have had an extremely tough time appealing to Americans with low incomes and low education levels. Decades ago, those people tended to vote for Democrats; in 2024, they broke solidly for Donald Trump. Even as they’ve become more and more progressive, the Dems have become the party of well-educated high earners:

Even people who identify themselves as “working class” have been abandoning the Democrats:

Joe Biden tried very hard to win over labor unions, but the Teamsters — once a Democratic stalwart — refused to endorse him.

The socialist faction might style itself a friend of the working class, but it faces the same problem. Zohran Mamdani won his mayoral race thanks to the support of higher-income, educated voters, while his defeated “establishment” opponent did better with lower-income people, Blacks, and Hispanics. Other socialist victories have seen a similar pattern.

This is not actually a new problem. A century ago, in The Road to Wigan Pier, George Orwell complained bitterly that the British socialists of his day were middle-class intellectual elites who failed to win over the working class because they were completely out of touch:

The first thing that must strike any outside observer is that Socialism, in its developed form is a theory confined entirely to the middle classes. The typical Socialist is not, as tremulous old ladies imagine, a ferocious-looking working man with greasy overalls and a raucous voice. He is either a youthful snob-Bolshevik who in five years' time will quite probably have made a wealthy marriage and been converted to Roman Catholicism; or, still more typically, a prim little man with a white-collar job, usually a secret teetotaller and often with vegetarian leanings, with a history of Nonconformity behind him, and, above all, with a social position which he has no intention of forfeiting. [emphasis mine]

Half a century later, Barbara and John Ehrenreich wrote something similar about the American left in their essay on “the professional-managerial class”. The “PMC”, as it has become known, was beginning to dominate the political left even in 1977; today, it arguably forms the Democratic Party’s most important voter base. The DSA faction that’s now becoming more influential within the party is disproportionately drawn from this class — 80% of DSA members had college degrees in 2021, and more than a third had postgraduate degrees (more than twice the national average).

Although they might not have realized it, what the Ehrenreichs were describing was a result of the rising economic importance of human capital. The “youthful snob-Bolshevik[s]” Orwell described in the 1930s were few in number, but the growth of knowledge industries has made this group of people far more numerous — and far more influential in our culture and our politics.

Because of its Marxist inheritance, the idea of the working class is very important to American socialists. But that’s been increasingly hard to square with the fact that voters with lower income and lower education have been steadily drifting away from the Democrats, and form relatively little of the DSA’s membership. One response has been for some socialists to paint themselves as the actual working class. This is a less charitable interpretation of Evan MacKay’s UAW jacket, and it’s also something that has come up in casual conversation. This is from my post back in 2024:

In January 2017, I was at a house party in Berkeley. People were discussing why Hillary Clinton had lost to Donald Trump, and one woman — a law student at the University of California — declared that it was because Clinton had ignored the “working class”. I asked her to describe someone in the working class. She imagined a “sex worker” who had a bunch of student loans and a humanities degree that she wasn’t able to use… I had expected her to describe a unionized auto worker or steelworker or a stereotypical Midwestern guy in a hard hat,…I was utterly unprepared for her to instead describe someone from her own educated progressive social circles…To this law student, the “working class” was simply those of her friends who were most down on their luck.

But on some level, I think this just doesn’t ring true, even within socialist circles. Class in America is just too deeply connected with earning potential and human capital; most people can’t really bring themselves to believe that someone with a diploma from a good school is “working class”, even if they happen to be pulling an unlivable wage as an adjunct professor and sleeping in their car at the moment.

So socialists are always on the lookout for champions who seem more authentically working class. They thought they found such a champion in Graham Platner, who recently dropped out of his Maine Senate race due to rape allegations. This is from a New York Times story about how Platner was recruited:

Last July, in a small town in coastal Maine, three progressive, self-styled recruiters of economic populists showed up at the blue-shingled house of Graham Platner, a little-known oyster farmer and Marine veteran who lived largely off government benefits…They knew his name from local labor organizers and activists, and they had watched a video on the internet of him talking about oysters. Struck by his left-leaning ideology, his working-class affect and his gravelly voice, they became convinced that he could win a Senate seat in Maine — and quickly persuaded Mr. Platner of the same…The recruiters — Dan Moraff, Leanne Fan and Morris Katz — told Mr. Platner he was “the one,” a “hero of the movement,” “a historical figure” who could be “leading a revolution,” according to half a dozen people with knowledge of their conversations.

Is Platner actually working class? You can argue it either way. He went to a good college but dropped out due to psychological issues. He’s the son of a lawyer and an architect, but failed to make much money with his oyster farming business, and mostly lived on welfare benefits. He doesn’t fit cleanly into the kind of class categories I described above.

But the people who picked him do! Dan Moraff and Morris Katz are both educated scions of rich families — Moraff’s ancestor founded the company that became Toys “R” Us — while Leanne Fan is a sociology PhD student. If the NYT’s reporting is correct, they appear to have picked Platner based entirely on stereotypes and vibes — he seemed rough and tough and down on his luck, so they assumed he was a real working class guy who could connect with other working class guys. This seems to have convinced them that Platner was a messianic figure who could bring wayward working-class voters back to the Democratic fold.

As many people have said, if Trump is “a poor person’s idea of a rich person,” Platner is basically the opposite — a rich person’s idea of a working-class person.

This episode doesn’t make me particularly optimistic about Democrats’ ability to reconnect with their rapidly vanishing blue-collar base. I don’t think it’s impossible, of course. But it would probably require moderating on social issues — DEI, immigration, etc. — instead of just running candidates that pattern-match to the kind of people that bullied rich lefty kids back in junior high.

The Platner saga also makes me a bit pessimistic about American society as a whole. Although we don’t talk about class much, it’s separating us more and more, and the cultural gap is now so big that lots of Americans seem unable to even imagine what Americans from other social classes are like.

The integrating institutions that once pushed us together across class lines — church, the military, schools in mixed-income neighborhoods — have waned in importance. College has grown to fill some of that void, but less than half of the country is really prepared to handle the rigors of a college education, and so it ends up dividing our society more than it unites it. Mass media has fragmented into the millions of little social verticals that make up the internet. We’ve sorted ourselves geographically — knowledge workers live together in progressive urban enclaves on the coasts, while blue-collar types inhabit the small towns and down-market suburbs.

I read Tocqueville, and I miss the roiling, fluid, egalitarian young democracy that I never knew. I think back to my childhood, in a little house on a dusty side street in a small Texas town, and I feel like I can just barely recall the fading embers of that stubbornly classless democracy. Something happened between then and now. We let the Old World sneak up on us.

Subscribe now

Share

In my post on America’s 250-year anniversary, I argued that respect for the individual was the “secret sauce” in the U.S.’ long-lasting preeminence among nations. But it was certainly not the only factor. There were plenty of institutional innovations that helped the U.S. stay on top — economically, militarily, and in terms of the attractiveness of its society.

One of these was American science. Today we take things like the modern research university, government grants for science, public-to-private research spinoffs, etc. for granted, but a lot of that infrastructure wasn’t there before World War 2. It was either invented or scaled up massively by the U.S. government in the postwar period, led by far-sighted scientist-bureaucrats like Vannevar Bush (pictured above). If you want to read about this history, a good place to start would be Jonathan Gruber and Simon Johnson’s excellent book Jump-Starting America.

Those scientific institutions didn’t exist in a vacuum, however. They were backed by the U.S. government’s abiding faith in the power of science — and, even more fundamentally, by deep popular trust in the scientific enterprise. Science gave us radar and the atom bomb in wartime, and in peacetime it gave us plastics, vaccines, cheap food, and a thousand other things that made our lives easier. Science was also the driver behind American industry — chemicals, aerospace, telecommunications, computers, electronics, and so on.

We owed science our jobs, our livelihoods, our comfortable living standards, and our greatness and power as a nation. It’s little wonder that both political parties, even as they fought viciously over other issues, were steadfast in their support of science. For a long time, the only people who distrusted science were a hippie fringe on the left who disliked modernity (or thought they did, anyway), who also disliked American industry and American power, and who subscribed to an early version of degrowth environmentalism.

America’s scientific enterprise is still strong, especially compared to the systems in Europe, Japan, Korea, and other developed countries. It has lost a lot of ground to China in a relative sense, but a lot of that is because of China’s incredible growth; as China has poured untold amounts of money and talent into its research labs, spending and output have overtaken the U.S. by some measures.

There are a few ways in which China’s rise creates problems for American science — for example, top scientists can choose to work in China instead of in the U.S. — and of course there’s the concern that China’s technological strength will help its military to reign supreme. But overall, China’s rise in science should be good for American science, since American scientists can use Chinese discoveries for free and build on them.

The much bigger problem is that the scientific enterprise America built during and after World War 2 is now being threatened with absolute decline. The biggest problem, of course, is that much of the country — the Republican half — has basically lost faith in the scientific enterprise. To what degree this loss of faith is justified is an open question, and deserves to be discussed openly. But the larger point — that the system that powered American dominance is under threat — is true either way.

Most Americans still trust science, but MAGA Republicans don’t

There’s a myth, popular in right-wing circles, that scientists have lost the trust of regular Americans — either due to the increasingly left-wing composition of academic departments, or to misbehavior during Covid, or to DEI-related research taking over science, etc. This also fits with a wider narrative that Americans are losing trust in all of our institutions.

But it just isn’t true. Poll after poll shows that on the whole, Americans still trust scientists and want to spend more on science. For example, here’s a Pew poll from late 2025 showing that although about a fifth of Republicans did lose confidence in science in 2020, two thirds still have at least “a fair amount” of confidence:

That same poll found that scientists are among America’s most trusted groups — even better trusted than the military!

In fact, Americans trust scientists more than people in most other countries do.

Other polls show that although Republican trust in science has dropped somewhat, Republican support for spending more on science remains very strong:

A Pew poll in 2023 found the exact same thing:

So when MAGA types tell you about scientists losing their credibility, or a drop in trust in science, they’re only talking about themselves. Whatever left-wing politicization of science happened during the Biden administration — and there was definitely some of that — it was not enough to make most Republicans lose faith in the scientific enterprise, or favor research cuts in order to purge unwanted ideology from the system.

That doesn’t mean I think progressives should continue down the path of politicizing scientific research. They should not, and Biden made real missteps in this area. Objectivity clearly matters for public trust of science in the long term. But as of right now, there’s no crisis of trust in science, except among the smallish minority of people who are running around screaming that there’s a crisis of trust in science.

But despite science’s overwhelming popularity and public trust, Trump and his administration are launching an unprecedented and devastating attack on American science — cutting funding, and forcing science projects to undergo ideological review by government commissars.

Trump’s comprehensive attack on science

Read more

“Her belly may be full, but her spirit will be empty.” — Captain Picard

Usually, these “GDP is actually good” posts start with a big disclaimer — an acknowledgement of all the things GDP doesn’t measure, all the reasons that measuring GDP is an inexact science, and all the ways that we need to improve society other than just making the GDP line go up. If you want a standard wonky explanation of why GDP is a useful number, here’s one that I wrote four years ago:

Today I’m going to do something a little different. I’m going to tell you what I think the debate over GDP is really about.

Free trade usually raises GDP. Immigration, done right, raises GDP.1 Rightists in America want less free trade and less immigration. But every time they propose restricting trade and immigration, someone — either libertarian business/econ types on their own side, or moderate liberals on the other side — says “That will make America poorer!”. So they want some way to neutralize this objection, so they can do things that will, in fact, make America poorer.

So America’s right borrowed an argument from the European left. The European left favors degrowth, and another term for degrowth is “making GDP go down on purpose”. So naturally, they’re always trying to find reasons to denigrate GDP as a metric of human flourishing (see here, here, and here for examples). The American right is simply tweaking these arguments to make them more appealing to their own base.

JD Vance, who has emerged as the consensus leader of the New Right, makes a bunch of these anti-GDP arguments in his new book. For example, he uses the example of Japan to point out that unobserved quality differences in non-traded products can make it difficult to compare GDP across countries:

If you’re focused on GDP, a $6 pint of Japanese strawberries is no different from a $6 pint of American strawberries. If you’re focused on dollars and cents, each contributes equally to the economic indicators. But if everyone in Japan eats better strawberries than everyone in America, the economic indicators have failed to measure something meaningful.

This is actually a good argument, and I’ve made it myself many times in the past. I’m in Japan right now, and there actually are a lot of little things that make Japanese products and services a bit nicer than their American counterparts — clean tables at Starbucks, slightly better-tasting food, and so on. Economists who try to adjust for quality differences end up catching some of these things, but probably miss most of them. That ends up creating a problem for GDP comparisons between countries. And it’s only one of many such problems. Comparing lifestyles in countries where life is very different is just a difficult thing to do.

But instead of simply noting that economics is hard, JD Vance uses this good argument as a reason to bash the entire field of economics:

When I got back home, a friend asked me if I learned anything on my trip to Japan. “Yes,” I replied snidely. “Maybe economics is just fake.”

When you read some of Vance’s other arguments against GDP, his agenda becomes clearer:

[A]s the decline of Christianity has left us without a shared moral language, economics has stepped into the vacuum. We pretend there are scientific answers to questions of values. Take one of the major issues of the 2024 campaign and a significant focus of our time in the White House: Should our trade policy be oriented around protecting domestic industries and jobs or around ensuring a short-term supply of cheap consumer goods?

This idea — that economists urge values of base consumerism on society, and ignore other moral considerations — is common in European leftist discourse. But instead of urging us to care more about inequality, power, and so on, as European leftists do, Vance wants us to care more about spiritual elevation, morality, community — i.e., things that the American right cares about. He goes on to write:

[W]e now live in a society almost blinded to considerations outside of the economic. This way of thinking is inherently opposed to the Christian way, which demands more focus on people…Take, for instance, the time we spend with our children…Domestic labor—that done by moms and dads—if unpaid, is uncounted in measures like GDP. When I leave work to spend time with my children, when I cook them dinner or argue with them about eating their carrots, I am engaged in economically unproductive work. No money changes hands, so it doesn’t show up in our national figures. By contrast, if I left for dinner at 6 p.m. and returned to work until midnight while paying a total stranger to look after my kid, my contribution to GDP would be much higher.

and:

If you step away from the glory of economic statistics, so much of American life has gone wrong. An influx of prescription opioids became a flood of synthetic opioids, which has led to tens of thousands of deaths each year and a declining life expectancy among a substantial portion of our society. We have made great progress on reducing infant mortality, but we send our children into a world—even in the physical security of their own homes—that bombards them with images and influences that have left them isolated, depressed, and increasingly at risk of self-harm. We are more disconnected, lonely, and isolated, even in the midst of historic levels of material comfort.

All this economic abundance coexists with intense spiritual misery. We orient people toward a life of consumption. We tell them to find meaning in the home they buy, the money they earn, the prestige of their job. We bombard them with all manner of creature comforts, and add their consumption—price club mega-size junk—to our national GDP. We use that GDP as a yardstick for our broader society, which is why it’s possible for false prophet economists to argue the American Dream is healthy even as suicide and addiction rates soar and the laughter of children fades from our streets.

Some of these arguments are — in my opinion — reasonable. A culture of overwork can boost GDP, at least in the short term, at the expense of quality time with family. This is actually a common argument of the center-left, which is why liberals have long fought — often successfully — for more paid family leave and other policies that reduce GDP slightly in exchange for more quality time with family. Whether this has increased birth rates isn’t clear — the evidence is very mixed — but it’s a very popular policy.

Other arguments are clearly mistaken. Over-prescription of opioids has clearly reduced GDP, by a massive amount. Yes, selling a bunch of opioid painkillers to Americans raises GDP by a few billion dollars, but this is vastly outweighed by the trillions of dollars of GDP that we lose from having a bunch more people addicted to painkillers, heroin, and fentanyl. Here’s the Philadelphia Fed in 2023:

There is growing evidence that the opioid epidemic has harmed many aspects of the real economy, including the labor market, consumer finance, and municipal finance. According to analyses from the Council of Economic Advisers’ 2019 report, the annual (nominal) economic cost of the opioid epidemic, including the cost of lives lost, is estimated at about $700 billion (roughly 3.4 percent of GDP) in 2018 alone, and over $2.5 trillion from 2015 to 2018. [emphasis mine]

So if you care about GDP, you should view curbing opioid and opiate abuse as a huge priority! Vance is simply not thinking about this very clearly.

But Vance’s real problem is that he conflates correlation with causation. In words that could have come straight from the mouth of a European degrowther, he rails against “creature comforts”, “consumption”, and “price club mega-size junk”. But nowhere does he explain why depriving Americans of these creature comforts would give them closer-knit families, a stronger sense of morality, stronger communities, reduced loneliness, and so on.

Why would taking away Americans’ large houses, SUVs, big-screen TVs, or central air conditioning make them spiritually richer? Modern Europe — which JD Vance spends much of his time railing against — lacks most of these things. And yet America has higher fertility rates than Europe, we go to church much more, and we have a much more robust social conservative movement. Europe has also been far more restrictive of speech that criticizes Islam, as Vance repeatedly notes. Yes, America has been trending away from social conservatism and Christianity in recent decades, but so has Europe, and the gap remains. Other developed countries in East Asia — most of which are moderately poorer than the U.S. in GDP terms — are extremely secular.

What about America’s past? We were much poorer in the 1950s, yet we went to church a lot more, had larger families, and so on. If you took away the material gains we’ve made since then, would we go back to tradwives and bowling leagues and lawn parties and Sunday church and 4 kids per family?

Perhaps, but it’s doubtful. Remember that the 1950s and 1960s were the culmination of a long upswing of community, religiosity, and so on in American society — something the sociologist Robert Putnam has documented extensively. Church attendance rose:

Fertility was on the upswing too:

And if you believe Putnam’s numbers, social solidarity increased all throughout the early 20th century:

It’s important to remember that this all came during the most robust and rapid period of GDP growth that America has ever seen. Over the period in which our social solidarity was soaring, our GDP per capita nearly tripled:

It was during this time that Americans got many of the “creature comforts” that Vance despises — the single-family homes, the cars, the televisions, the lawns, and so on. The image of that material prosperity, depicted in glossy ads and paintings from the time, is a powerful part of 1950s nostalgia.

In fact, many economists argue that one big cause of the Baby Boom was the fact that economic growth — bigger houses, better medical care, new labor-saving devices like washing machines, refrigerators, and vacuums, and so on — made it easier and cheaper to raise kids. This is from a relevant Works in Progress article by Anvar Sarygulov & Phoebe Arslanagic-Little:

Parenthood rapidly became much easier and safer between the 1930s and 1950s. The spread of labour-saving devices in the home such as washing machines and fridges made raising children easier; improvements in medicine making childbirth safer; and easier access to housing made it cheaper to house larger families…

[H]ousehold electrification paved the way for other technologies, including home refrigeration…By the 1940s, electric washing machines were becoming normal in middle class homes…Between 1936 and 1956, America’s maternal death rate fell by 94 percent, from 51 deaths per 10,000 live births to under 3…[M]edical advances, which were being made across the West, radically reduced the most serious potential cost faced by prospective mothers: life itself…

Alongside strides forward in household and medical technology…[I]t became easier to secure a home in which to raise children. The number of houses built soared across the West after World War Two…This house-building bonanza led to sharp rises in homeownership rates.

The golden thread linking the phenomena that comprise the triple mechanism we describe above – advances in household technology, progress in medical technology, and easier access to housing – is that they together sharply reduced the cost of having children. [emphasis mine]

If you like the kind of society we had in 1960, you can’t ignore the story of how we got to 1960. The answer was “economic growth”. This, combined with the examples of Europe and Asia, is why there’s no reason to believe that forcing Americans to be poorer — taking away the “creature comforts” Vance despises — would lead us to suddenly rediscover the value of community, family, and religion.

Now it’s worth noting that if you were to decrease America’s GDP to poor-country levels — below $15,000 per person, as opposed to over $90,000 today — you might be able to raise fertility. That’s how low you have to go before most countries have fertility above replacement level:

Countries with a basically pre-modern standard of living — where many women can’t read or write, and infant mortality is so high that families have to have many kids as a form of insurance — tend to have above-replacement fertility (though some don’t). But even this law is weakening, as fertility rates in Sub-Saharan Africa plunge, so even that extreme level of GDP reduction would probably fail to restore high fertility over the long run. Also, I kind of doubt that JD Vance wants to force Americans to live lives similar to those lived in Sub-Saharan Africa.

What about JD Vance’s preferred policies — trade protectionism and immigration reduction? Would those restore American community, family, and religion, at the expense of a bit of GDP? As I said earlier, that’s actually what I think this whole debate is really about.

On trade, you’ve seen Trump explicitly make the argument that Americans are going to need to suffer a bit of material deprivation in order to achieve the administration’s goals:

But how will doing this restore community, family, etc.? Presumably you could make an argument that protectionism will bring back good manufacturing jobs, which will then give men the confidence and social standing they need to get married and have kids. There’s just one big problem with this, though: Trump’s trade policy doesn’t actually increase the number of good manufacturing jobs. We’ve lost manufacturing jobs since Trump took office last year!

In fact, Trump’s tariffs are hurting the U.S. manufacturing sector, by raising the cost of intermediate goods. Economists understand this pretty well; JD Vance, who thinks “economics is just fake”, does not seem to understand it.2

How about immigration? Rightists will endlessly cite Robert Putnam’s finding that diversity reduces social trust in American communities. But as Bryan Caplan and many others have pointed out, the effect size is tiny — in Putnam’s research, going from zero diversity to maximum diversity reduces social trust by the equivalent of 1 point on a 100-point scale. This suggests that all the mass deportations in the world won’t move the needle on American community and togetherness.

In other words, JD Vance’s crusade against GDP is a cargo cult. Sure, GDP doesn’t measure “the beauty of our poetry or the strength of our marriages,” to use Robert F. Kennedy’s famous words. But that doesn’t mean that making Americans poorer will make their poetry more beautiful or their marriages stronger. Nor does it mean that policies that also happen to make us a bit poorer, like immigration reduction or tariffs, are any more likely to strengthen our society.

In a famous episode of Star Trek: The Next Generation — my favorite TV show of all time — Captain Picard castigates an alien for turning his society into a fascist empire. When the alien responds that the fascist government’s forced modernization program raised his daughter out of poverty, Picard responds with the beautiful quote at the top of this post: “Her belly may be full, but her spirit will be empty.” But was Picard arguing that it was the full bellies themselves that emptied the people’s spirits? Was he merely arguing that the fascist empire ought to become a poorer fascist empire, in order to restore the virtue of the people? Only a fool would think so.

Look, I also want America to have a stronger society. I want us to have more kids. I want us to have more stable families. I want us to have closer-knit communities, better moral values, etc. But that doesn’t mean the New Right knows how to get us there. So far, the New Right has built nothing — no new community organizations or institutions, no religious revival, nothing that would knit our society together. It has merely thrashed and thrashed against modernity, with no plan for a replacement. JD Vance’s crusade against GDP is simply more of the same.

Subscribe now

Share

2

In fact, if you care about Americans having jobs, maximizing GDP growth is an important part of the equation. Okun’s Law — one of the most well-established regularities in economics — states that the unemployment rate is very strongly (negatively) correlated with quarterly GDP growth:

And only when GDP is growing robustly is there upward pressure on wages. The greatest period of growth for working-class wages in America since the 1960s was 2014-2024, despite inflation and the pandemic, because steady GDP growth kept us at full employment for most of that time.

Nor is this simply a case of correlation vs. causation. The mechanism is aggregate demand — when the economy is growing fast, people need more workers to produce things. So if you increase aggregate demand to make GDP grow faster, you raise total employment (though you don’t always want to do this, because of inflation). Yes, you can give people government jobs when the economy isn’t growing, but A) these jobs will generally not be great, and B) this will lower productivity and put downward pressure on wages. So for good jobs, you need GDP growth.

Back in the 2010s, a lot of people marveled at China’s seemingly recession-proof economy. Throughout the global financial crisis of 2008 and the Chinese stock market crash and capital flight of 2015, the country never recorded a single quarter of negative economic growth. Here’s what I wrote back in 2019:

China’s government seems to have developed a highly effective new form of economic stabilization. Its extensive control of the financial system allows it to turn on a flood of bank loans when the economy looks weak, and restrain credit when the danger has passed. China’s avoidance of recession in at least the past three decades suggests that this form of credit-based stabilization is more effective than traditional, more indirect stimulation of the economy through government deficits and central bank monetary easing…When a recession threatens, the government tells banks to lend --— to local governments, construction companies and real estate developers. Then, if the credits go bad, the government swoops in and takes the nonperforming loans off of financial companies’ books. Uninterrupted rapid growth then shrinks the government debt as a percentage of gross domestic product, and the system sails blithely forward[.]

And here’s what I wrote in 2018:

China…directed banks to lend lots more money [in 2009]. The World Bank estimated that increased bank credit represented 40 percent of China’s stimulus. Much of the lending was done by China’s four large state-owned banks. The money went to infrastructure, real estate and all kinds of corporate projects, many of which were carried out by the country’s state-owned enterprises.

Basically, most countries use two types of policy to get the economy moving again when some sort of negative shock hits it:

1. monetary policy (e.g. cutting interest rates), and

2. fiscal policy (e.g. stimulus spending).

Macroeconomists disagree about why interest rate cuts give the economy a boost, but most agree that the policy usually has an effect. Although there are many other theories and interpretations, one way you can think of rate cuts is as a financial policy — by making it easier for businesses to borrow and invest, low interest rates stimulate business activity. Fiscal policy, in contrast, pretty much bypasses the world of finance and aims directly at the real economy — you build a bridge or a road, which employs some people who might otherwise be unemployed, and then those people turn around and spend their money elsewhere in the economy, igniting a virtuous cycle of spending and working.

China uses both of those, but it also uses a third policy: financial policy. Instead of simply cutting interest rates and hoping that this filters through to bank lending, China’s government uses its direct control over the banking system to push banks to lend more. In the 2010s, after the Great Recession and the 2015 Chinese stock crash, this mostly meant lending to real estate companies. This lending fueled the biggest property boom the world has ever seen.

The boom ended in late 2021. The crash of the Chinese property developer Evergrande began a sequence of bankruptcies and defaults across the entire real estate sector. China’s property prices began to fall, and have not stopped falling to this day:

Chinese housing construction plummeted as well:

But despite the housing crash, China’s official growth rate never fell below zero — or even below 3%:

In fact, China did this by resorting to a version of the same playbook it used in 2009 and 2015. The Chinese party-state called up its captive banking system and told it to lend huge amounts of money to manufacturing companies. And that’s exactly what it did — industrial loans surged, even as real estate loans petered out:

It’s tempting to cry “China’s done it again!” In fact, that’s exactly what some people are now doing:

Skeptics will caution, of course, that this sort of stabilization policy can come with a cost: lower productivity growth and economic inefficiency over the long term. That’s probably what happened in the 2010s, as China’s repeated use of real estate lending to stabilize the economy directed resources to inefficient real-estate companies and led to lower productivity growth. Now there’s the possibility that China’s wave of financial stimulus in 2022-2024 may lead to an overhang of unproductive “zombie” companies that keep soaking up labor and other resources for years to come:

But admirers of China’s economic system will be undeterred. They will point out that productivity is hard to measure; that long-term costs are both uncertain and hard to verify; and that long-term problems can always be fixed later. The more important fact, they’ll argue, is that China did exactly what Xi Jinping said it would do — to pivot away from an excessive reliance on real estate without causing the economy to shrink. “Chinamaxxers” will use this as reason to crow about the superiority of the Chinese way, while left-leaning intellectuals will use China’s performance as a foil to demonstrate the benefits of greater government control over the economy.

There’s just one problem with this triumphalism: China did, in fact, have an economic crash as a result of its real estate bust.

The first way to see this is to look at China’s job market. In 2023, China famously modified its youth unemployment data to use a narrow definition of unemployment, because the numbers were getting too high. But even the revision couldn’t mask the upward trend:

Overall unemployment was recorded as rising only a small amount. But as Bloomberg reported at the time, China’s total unemployment numbers aren’t a very good measure of the labor market, and alternative indicators told a much more pessimistic story:

Alternative indicators and anecdotal reports suggest unemployment is worse than the official monthly figures show…[T]he [official headline] figures aren’t sensitive to changes in the number of migrants from China’s rural areas who work in cities; they also don’t capture the number of people who have dropped out of the labor market for more than three months or those unable to start work…

The employment sub-index for China’s non-manufacturing purchasing manager’s index, which tracks hiring intentions in the service and construction sector, has stayed consistently below pre-pandemic levels for most of the past 12 months…Official data shows there’s been no growth in the migrant worker population since the pandemic…The average number of workers at industrial enterprises with revenues above 20 million yuan ($3.1 million) fell to 7,398 in November 2021 from 7,419 in November 2020, according to official statistics…Because of the weak labor market, record numbers of young people are preparing to take exams to qualify for post graduate courses or enter the civil service [and] would not be counted as job seekers[.]

What about GDP growth? In her tweet above, Kathleen Tyson declares that China “was the first to deflate a massive, leveraged housing bubble without a single quarter of economic contraction or loss of growth momentum in the real economy”. But is that true?

Well, no, it’s not. According to China’s official statistics, the Chinese economy shrank by 0.8% in the second quarter of 2022:

China’s growth is usually reported in year/year numbers, but quarter/quarter is how the U.S. and most countries do their reporting. So by the kind of measurement Americans are used to hearing about, China’s economy officially contracted at an annualized rate of over -3% in the second quarter of 2022. This was also revised down from the -9.3% that was reported in the initial version of the statistics. As for “loss of growth momentum”, China’s economy is officially growing around 2 percentage points slower than it was just before the pandemic.

So even if we accept the official numbers, the claim is wrong. But should we accept the official numbers? Probably not. There is evidence that the Chinese government “smooths” its growth numbers — in good years, it fudges downward, and in bad years it fudges upward. This is from Nakamura et al. (2016), who use detailed data on Chinese consumption to estimate how incomes changed:

Our estimates suggest that official statistics present a smoothed version of reality. We find that inflation was overestimated and growth underestimated by several percentage points per year in the late 1990s. In contrast, since 2002, official inflation statistics have risen only modestly, but our Engel curve based estimates have risen much more. Our estimates imply that growth was substantially lower than official statistics suggest since 2002, and actually dipped into negative territory in 2007 and 2008.

A bunch of analyses claim that China has also done this in response to the property crash. The Rhodium Group used alternative data sources to estimate that China’s economy actually shrank in 2022 and grew much more slowly in 2023 than the official numbers suggest:

Officially, China reported 3.0 percent real GDP growth in 2022, despite the fact that significant proportions of the economy were under strict lockdowns to prevent the spread of COVID-19 during large portions of the year, retail sales fell outright, and investment in the property sector was collapsing. In 2023, the decline in property investment continued, net exports and government spending were drags on growth, and household consumption growth remained relatively low. Beijing provided little direct assistance to households to facilitate spending, and Chinese households added to savings and paid down mortgage debt instead of spending more. Yet China officially reported 5.2 percent real GDP growth in 2023, barely slowing from the pre-pandemic pace of 6 percent in 2019, even though the property sector was experiencing a boom in 2019 and was collapsing in 2023…We estimate that real GDP growth was closer to a contraction of -0.3 percent to -0.8 percent in 2022, and there was only modest growth of 1.5 percent to 2 percent in 2023. [emphasis mine]

The Bank of Finland was a little less negative, but still estimated that growth stalled in 2022:

Capital Economics, which tracks a whole bunch of independent estimates, finds that China probably did experience a recession in 2022, though it’s pretty positive about growth since then.

One particularly pessimistic indicator is inflation, which has slipped into negative territory in China since the real estate bust:

Deflation is a classic sign of low aggregate demand and a slowing economy.

It should be noted that there are a few analysts who disagree, and think that China’s growth numbers are basically accurate. But most independent assessments conclude that China’s growth not only suffered a sharp hit in 2022, but has been weaker in the years since the end of the pandemic.

It makes sense that China’s government would continue their traditional approach of smoothing out growth numbers in the short term in order to project an attitude of stability and calm. But smoothing only works if the economy eventually bounces back. If China is on a new longer-term trajectory of lower growth — which of course remains to be seen — then there will be too few good years to “pay back” the growth that was “borrowed” in the bad years of 2022 and beyond.

I don’t want to detract from China’s accomplishment here, or say that its macroeconomic stability is entirely fake. China has invented — or, perhaps, perfected — an alternative tool for macroeconomic stabilization. Countries all over the world, including the United States, should study China’s financial stabilization policy and think about how to accomplish something similar without direct government control over bank management.

But at the same time, I don’t think we ought to be idolizing Chinese macroeconomic policy either. Even if there don’t turn out to be long-term productivity costs — which is a big “if” — China still hasn’t managed to rewrite the rules of aggregate demand and aggregate supply.

Subscribe now

Share

“What other country would have done this?” — Daniel Inouye

“We don’t repeat this every day, but there are 33 words that are very sacred to all of us. We do the repetition a little differently but ‘We hold these truths to be self-evident. That all men are created equal. That they are endowed by their creator certain inalienable rights. That among these are life, liberty, and the pursuit of happiness.’ It’s operational. Believe me.” — Daniel Inouye

China’s last imperial dynasty, the Qing, lasted 268 years. The Ming Dynasty that preceded it lasted just slightly longer, at 276, while the celebrated Tang made it to 289. Decades before each of those dynasties officially fell, they were shells of their former selves, with much of the land outside the control of the central government.

Barring abrupt catastrophe, the United States — which today marks a quarter of a millennium — will probably last as long as the Qing, the Ming, and probably the Tang. The country’s foundations are certainly shakier than when I was a child, but we have not yet entered an obvious terminal phase. The economy is still robust — our GDP remains on the smooth upward trend it has been on since we started measuring such things eight decades ago:

We cannot (or will not) build a functional passenger train network, but our AI industry is upending the world. Our health care costs twice as much as that of other rich nations, but our houses are huge and luxurious. Our cities are burdened with crime and disorder, but e-commerce delivers everything we need, straight to our door. The hour of the wolf is not yet upon us. It seems a safe bet that there will still be a United States of America in 2044, 2052, and probably 2065.

And yet we’ve reached the stage where we can peer through the fog and see how this grand experiment might be heading toward its conclusion. Much of the country has eased into a comfortable equilibrium of sclerosis; local veto power either prevents the construction of factories, housing, energy, transportation, and other infrastructure, or delays it by decades, or raises the cost to multiples of what other rich countries pay. The past has become more valuable than the future to many Americans; they cling desperately to the power to enforce stasis, preserving a facade of the country they grew up in at the expense of the very dynamism that made that country great.

That sclerosis seeps into everything else. Immigration, and even migration from city to city, becomes a vicious zero-sum fight over a fixed housing supply. Cities decay into museums of themselves. The industries of the future can only be built in America if they take up nearly no land, use nearly no energy, require very little bank financing, and are able to procure skilled labor as needed from abroad. Somehow the internet industry satisfied all of those conditions for three decades, but that time is done.

Our politics, meanwhile, has degenerated into movements defined more by who they hate than by any positive vision for the country’s future. Rightists are consumed by their hatred for immigration, leftists by their hatred for Israel. Even intellectual liberals — my own movement and social class, if only by process of elimination — increasingly subordinate other goals to their dream of lowering the social status of wealthy technologists.

To the extent that popular visions of a better America exist, they are rank and obvious fantasies — homogeneous harmony that rightists will never be able to create, or socialist plenty that socialism is incapable of delivering. There are plenty of workable, feasible future visions that would advance the frontiers of freedom, dignity, and prosperity; no faction of the engaged American public seems particularly interested in them.

Political discourse in America is still the baleful thing it became in the 2010s — a vicious free-for-all of social media influencers using hatred, division, fear, and misinformation to win the ear of the powerful political staffer, think tank, and journalist classes. Everything exists in the shadow of the almost-revolution of the late 2010s — an upheaval whose force has mostly receded but whose damage has yet to be fully assessed. Meanwhile, the country’s powerful enemies abroad are sharpening their knives.

If there is a reason to be pessimistic about America’s future, it’s that so few of the country’s citizens seem to believe in it. We used to be an unusually patriotic nation; now Americans are less proud of their nation than Europeans, Asians, or people in any other major world region:

The rightists who now dominate the GOP believe that America will only be valuable as a going concern if its old ethnic composition can be forcibly restored. The leftists who are surging among the Democrats, meanwhile, have a vision of America as an evil empire that could have come straight from old Soviet propaganda; this idea finds fertile ground among progressives who for a decade have mainlined the notion that America is “stamped from the beginning” with racism. How will the country be saved if no one thinks it’s something worth saving?

It would be foolish, of course, to predict that the U.S. is headed for the scrap heap within our lifetimes; uncountably many such predictions have made fools of the people who made them. The country is not facing mortal, imminent danger; its enemies are powerful but most of its wounds are self-inflicted. The United States may yet survive, with its territory and its constitutional democracy intact, to its 300th birthday and beyond.

Even so, it’s far from clear what a nation will even mean in those decades and centuries to come. The human race as a whole is set to dwindle, as fertility falls below replacement in every corner of the globe. At the same time, more and more of the thinking done on the planet will be done in data centers rather than within human brains. In that posthuman world, it’s not at all clear that humanity will even need the nation-state to provide the crucial organizing and coordinating role it played during the previous two and a half centuries.

So whether or not this is the beginning of the end for America, it’s the beginning of the end for something even bigger and more important — the human age. By that I mean the age when humanity, unassisted by any higher intelligence, broke free of the chains that had bound it for millennia and became something greater.

Read more

Birthright citizenship is the law of the land. The Supreme Court has once again ruled that the 14th Amendment gives automatic citizenship to anyone born in the U.S.1 This includes the children of illegal immigrants and temporary visa holders, both of which the Trump administration had sought to exclude from birthright citizenship.

Rightists immediately began howling about the ruling, and saying some very intemperate things. Sean Davis of The Federalist suggested dissolving the Union and/or sterilizing foreign visitors to the United States:

Right-wing political commentator Matt Walsh shrieked that the America he grew up in — which also had birthright citizenship and significant amounts of illegal immigration — has somehow been destroyed as a result of the SCOTUS ruling:

These histrionics — sadly typical of online reactions to major news events in the social media age — demonstrate how central the anti-immigration cause has become to the political right in the United States. The notion that immigration is an invasion bent on destroying the country by replacing its founding population has become a bedrock belief on the Right; it is a singular, all-consuming passion similar to what anti-racism was for 2010s progressives and Palestine has become to leftists in the 2020s. Nor is it just an object of passion; nativism has become a self-contained, hermetically sealed worldview not subject to reasoned argumentation, logic, or data.

It is a distinctly minority worldview. The overwhelming majority of Americans continue to say that immigration, on the whole, is good for this country:

This does not mean that Americans want open borders, or that they think all kinds of immigration are good. Sentiment against illegal immigration, and in favor of increasing border security, remains strong. A backlash against the disorderly flood of quasi-legal immigration under Biden helped get Trump elected in 2024. But even illegal immigration is not seen as an invasion by most Americans — support for a pathway to citizenship remains strong, even among many Republicans not affiliated with the MAGA movement.

The rightist view of immigration as the death knell of America is simply a small minority viewpoint. Guys like Sean Davis and Matt Walsh are a screechy online fringe. They seem to think that if they screech loud enough and make dramatic enough threats (“Dissolve the Union!”, “Sterilize tourists!”, and so on), they can bully the supermajority into giving them their way. They basically expect veto power over this one issue, based purely on the strength of their emotion.

And on the narrow question of birthright citizenship, poll after poll shows that Americans want to keep the practice. Here’s an example from The Hill:

Nearly 70 percent of respondents in a Quinnipiac University poll think the Supreme Court should keep birthright citizenship in place. The results come ahead of the high court’s ruling on the legality of President Trump’s executive order seeking to end the policy…The survey, conducted between June 18 and 22, found that 69 percent of 1,165 self-identified registered voters believe the Supreme Court should keep in place its 1898 ruling affirming that the 14th Amendment guarantees citizenship to those born in the U.S…Less than 3 in 10 respondents (27 percent) said the high court should reverse its decision[.]

That doesn’t mean the public entirely disagrees with the Trump administration. About half of Americans think that birthright citizenship ought to be denied to the children of illegal immigrants:

(It’s not clear what Americans think about the children of temporary visa holders; this question is typically not broken out in polls, and it’s not clear whether Americans generally understand the difference between permanent residents and legal visa holders.)

So if the MAGA movement were pragmatic — if they really wanted to succeed in restricting immigration in a way that Americans would support, instead of just screeching louder and louder about immigration in general in an attempt to cow the majority into giving them their way — there might be room for them to change the law. They don’t currently have enough support for a Constitutional amendment specifically revoking birthright citizenship for the children of illegal immigrants, but they might be able to get over that line with a concerted campaign.

But that’s a bit of a moot point, because the right is probably not going to execute that sort of pragmatic, majoritarian strategy. Immigration has become a culture war. Taking a maximalist position on the issue is a way for people to signal their allegiance to the MAGA movement; supporting substantive compromises that win real policy victories broadcasts that you’re not a core part of the movement.

As for Democrats, they seem to still be almost entirely reacting against MAGA. On the positive side, this means that even hardcore leftists like Hasan Piker responded to the SCOTUS ruling by temporarily dropping their “America is evil” shtick and showing some national pride:

On the minus side, it makes it even harder to get coherent immigration policy. Trump & co. want to make immigration harder, so Dems will simply make it easier in response. This is probably why Biden was so lax on border enforcement during the first three years of his term — a disastrous decision that probably cost the country four more years of Donald Trump.

This is a shame, because smart immigration policy is needed now more than ever. The U.S. is suffering a fertility crisis, similar to the rest of the world. Two decades ago, America was having enough kids to keep the population stable over time. Now, fertility in America has declined to below the level of Japan in the 1980s:

Most of us grew up thinking of Japan as a place with an aging, shrinking population. America is now headed for exactly that same fate, unless we take in immigrants. But Trump’s restrictive policies — not just deportations, but dramatic reductions in legal immigration — have almost entirely cut off the inflow:

Other estimates show Trump actually reducing the number of immigrants in America in 2025 (and probably 2026), which would mean MAGA’s anti-immigration crusade is now actively reducing the country’s population.

This is bad news for the country. For one thing, it means far fewer workers to support each retiree. With zero immigration, the number of American working-age people per retiree will fall from 3 to 2 over the next quarter century:

For some, that will mean paying higher taxes to support old people’s health care and eldercare. For others, it will mean directly doing the work of taking care of their aging parents. Either way, it means more toil and drudgery for the young and the middle-aged, and less freedom and consumption and fun. As Paul Krugman points out, immigrants are disproportionately likely to work in jobs that take care of old people:

The burden of supporting the elderly will also probably reduce fertility even further — it’s hard supporting kids and your retired parents at the same time! — which will compound the problem in the long term. And small towns and rural areas will be especially hard-hit.

Immigration can’t hold off population aging forever in a world of low fertility — immigrants get old too, and so you need to keep increasing the immigration rate just to maintain the age structure. And as the whole world begins to shrink, the supply of immigrants will dry up. But America’s wealth, and our (rapidly diminishing) reputation as a “city on a hill”, gives us the ability to stave off population aging for a while, and perhaps buy time to find a more permanent solution to the riddle of low fertility.

At the same time, we’ll get a lot more benefit if we’re careful about which kind of immigrants we let in. Immigrants with higher education levels add to the national fiscal coffers, since they make a lot of money. But immigrants with low education levels create a net fiscal drain, since they make less money and absorb more government benefits (as do their children).

If you just look at immigrants themselves, you find that college-educated immigrants tend to decrease the national debt, while immigrants without college degrees tend to add to the debt:

If you include later generations — who tend to show strong upward mobility in the U.S. — then the long-term fiscal impact is more positive, but you still see a very strong difference by education level:

So if we want to use immigration as a tool to strengthen our nation’s economy, we should focus on letting in immigrants with college degrees. That means more legal skilled immigration, more border security, and less quasi-legal asylum grants — in other words, exactly the set of policies that the American people say they want.

Now, MAGA people will certainly respond that immigration is about more than dollars and cents — it’s about culture. Immigrants assimilate to American culture, but they also do change that culture somewhat; assimilation is a two-way street. It’s impossible to live in the same country you grew up in — technology is still the biggest cause of cultural change — but if you want to preserve as much of the country of your youth as humanly possible, then it makes sense to restrict immigration. More broadly, if you believe that nation-states are legitimate entities, then you must admit that countries have an inherent right to preserve their cultures in amber by shutting themselves off to immigration, if they want to.

But MAGA has already lost that battle. America is not North Korea; we’re not even Japan or Sweden. 79% of Americans say that immigration is good for the country overall; they say this even knowing that immigration will cause some long-term cultural changes. The cultural preservation argument against immigration is a lost cause, no matter how desperately its proponents shriek and bluster and threaten.

At the same time, Democrats must be extremely wary of embracing open borders simply out of pure reaction to MAGA nativism. Americans don’t want to let just anyone in; they want people to come legally, and they want to admit people who earn enough money to contribute positively to the economy. If Democrats simply respond to each electoral victory by throwing open the borders and turning a blind eye to illegal immigration, the American people will continue to respond by intermittently electing xenophobes.

In the age of low fertility, America desperately needs to be smart about its immigration strategy. We can’t let a fringe of culture-warriors dictate our national policy.

Subscribe now

Share

An old friend recently wrote to me, telling me that I needed to write a blog post about bears on bicycles. He wrote: “The internet needs a large, apex predator trying its best to navigate a two-wheeled vehicle through the complexities of global supply chains and geopolitical shifts.” Wise words indeed. AI-generated words, to be sure, but full of wisdom nonetheless.

Writing a Noahpinion post about bears on bikes sounded challenging, but I’m never one to shrink from a challenge. So I’ve tried to include bears on bikes as a thematic throughline in today’s roundup.

But first, a podcast. I went on Sam Harris’ podcast to talk about the state of the macroeconomy! His podcast is paywalled, but you can listen to the first quarter of the discussion here:

Here’s a YouTube preview, if you like video.

Anyway, on to the roundup!

1. What if the “vibecession” is just bad data?

For years now, we’ve been wondering why Americans are in the dumps about the economy even though inflation is still relatively low and the employment rate is historically high. This “vibecession” is still happening, and it’s reaching absurd levels. The University of Michigan’s Index of Consumer Sentiment has hit its lowest level on record — lower than the inflation of the 70s, lower than the Great Recession, and lower than Covid:

Potential explanations have included:

1. Americans expressing unhappiness about political and social conflict by claiming the economy is bad

2. Increased partisanship making either Democrats or Republicans loath to admit that the economy is good, depending on which party has the presidency

3. High interest rates making it unaffordable to buy a home

4. A delayed reaction from years of rising service costs or other long-simmering economic difficulties

Now, over at Nate Silver’s blog, Joel Wertheimer has another potential explanation: Maybe the consumer sentiment data is just bad!

Silver Bulletin

Is the vibecession real — or is the survey broken?

Today’s newsletter is a guest post from Joel Wertheimer. Joel is a civil rights attorney in New Yo…

Read more

a month ago · 204 likes · 15 comments · Joel Wertheimer

He writes:

The University of Michigan ICS [Survey of Consumers] is the gold standard sentiment survey measuring consumer sentiment. The survey has historically shown a very strong correlation with “hard” economic data such as inflation and unemployment. But…As with election polls, the ICS has struggled amid a shift away from telephone polling…So the problems with the ICS are these:

1. The switch to online polling made responses more negative and,

2. There are too many Democrats in the sample.

Thus, ICS data since mid-2024 is not comparable to past periods…Democrats right now say they hate the economy with Trump in charge…

When adjusted for these issues, the ICS should be substantially higher than the Great Recession lows we have witnessed over the past year. Weighting the survey to Pew’s National Public Opinion Reference Survey…would place the ICS at a level more like that in 2013…This adjustment would bring the survey in line with other measures of consumer confidence, such as those from the Conference Board, Gallup, and YouGov.

This seems like an important point. The Conference Board’s survey shows consumer confidence down from 2019, but still pretty good:

Gallup and YouGov don’t have data from before 2020, so I’m not sure how to use those as bases for comparison, but the disparity between the Conference Board’s survey and the University of Michigan’s survey seems important. And notice that the Conference Board doesn’t even show a “vibecession” during Biden’s term in office!

And although I missed it at the time, Ryan Cummings and Ernie Tedeschi did show that the University of Michigan’s index of consumer sentiment fell by 9 points when they shifted from telephone polling to online polling in 2024:

Briefing Book

The effect of online interviews on the University of Michigan Survey of Consumer Sentiment

We analyze the University of Michigan’s (UMich) Consumer Sentiment survey’s (“sentiment”) recent change in survey methodology from collecting interviews via phone to collection via the Internet. We document several features of this new sample that we believe are materially affecting the results of the survey. While we agree wi…

Read more

2 years ago · 13 likes · 4 comments · Ryan Cummings and Ernie Tedeschi

In other words, the notion that American consumers are ultra-bearish might be perched upon a flimsy bicycle of bad data.

So it might be that Americans are simply a lot less angry about the economy than we thought. That would be heartening from a “people are rational after all” standpoint, though perhaps disappointing for those who are hoping for a Democratic electoral sweep this fall.

2. A bit of evidence for Digital Cronkite

Back in March, I wrote a hopeful post about how AI might de-radicalize our politics:

The basic idea was twofold:

1. AI can inject lots of information into debates in real time, in order to bring them back to reality and correct misinformation.

2. AI is trained on a sample of writing from Democrats AND Republicans, so it tends to be more moderate than the typical human being in a polarized society.

I called depolarizing AI “Digital Cronkite”, because it could be a modern version of Walter Cronkite’s moderate voice of authority on broadcast television in the mid 20th century.

My post cited some papers in support of the thesis. Here’s one more. Conlon and Schwardmann (2026) set out to study the phenomenon of “AI sycophancy” — i.e., AI telling people what they want to hear. Sycophancy is a well-documented phenomenon — if you tell AI that bears ride bikes, it’s disturbingly likely to agree. Some people are naturally afraid that sycophancy will increase political polarization, by reinforcing people’s political beliefs. If AI tells Democrats “Yes, Republicans are bad and you’re right about everything,” and if it tells Republicans “Yes, Democrats are bad and you’re right about everything,” that could seemingly entrench polarization by turning every human-AI conversation into a little echo chamber.

But Conlon and Schwardmann found that AI does the exact opposite of this! From their abstract:

Our experiment involves 1,500 participants in 30 decision environments spanning core domains in economics and the social sciences…[W]e find that AI advice depolarizes choices on average, moving participants away from their initial leanings. This depolarization arises despite the LLM being measurably sycophantic: it disproportionately offers considerations that support users’ initial leanings and uses agreeable and flattering language. Depolarization occurs across moral and non-moral, objective and subjective, strategic and non-strategic, and complex and simple tasks. Increasing sycophancy weakens depolarization, showing that sycophancy is behaviorally relevant, even if it is generally outweighed by the informativeness of AI advice. [emphasis mine]

The authors go with the “substantive information” theory of AI depolarization, and they find some evidence to support it:

Why then does our baseline LLM depolarize choices, despite being sycophantic? One hypothesis is that it combines its agreeable framing with enough substantive information to counteract sycophancy’s polarizing force. Several facts corroborate this interpretation. First, in objective tasks, interacting with the baseline AI improves accuracy by 0.12 standard deviations (p < 0.05)…In contrast, we find no evidence that extra deliberation time, reactance, noise, or ceiling effects drive our results.

Interestingly, Conlon and Schwardmann don’t think about what seems to me to be the most obvious reason for AI depolarization — i.e., that AI is trained on the average of society, and thus is the ultimate Moderate Normie. That’s something worth following up on.

But whatever the reason, this is another piece of evidence in favor of the Digital Cronkite thesis. You should probably be a little more optimistic about AI’s effect on human politics.

3. A bad argument from a good economist about AI risk

Anthropic recently hired Chad Jones, my favorite growth theorist. That’s great news, both for Chad and for Anthropic! But a Financial Times article about the hire catches Chad saying some highly questionable things about existential AI risk in a working paper back in 2023:

What is the price of this amazing change in living standards? Recall that we would face a flow probability of existential risk of 1% per year for 40 years, so the probability we survive this A.I. explosion is exp(−.01 × 40) ≈ 0.67. In other words, with log utility it is optimal to take a 1 in 3 chance of ending human existence in exchange for a 2/3 chance of dramatically raising living standards by a factor of 55.

Chad’s point here is not that we should be blasé about the existential risk from AI. His purpose in the paper is to compare different utility functions and show how our attitude toward existential risk can change a lot depending on risk aversion. But I still don’t like his calculation here.

The main reason is that Chad sets the utility of human extinction equal to zero. This isn’t actually log utility. Log utility is just u(c) = ln(c), where c is consumption. If you consume nothing at all, then ln(c) = ln(0) = -∞. In other words, if you take log utility seriously, then death is infinitely bad.

That presents a problem for economic models, since it means that even the slightest chance of death is so scary — scarier even than a bear on a bicycle — that humans would do basically anything to avoid it. That obviously isn’t realistic. So instead, economists using log utility typically model it with a fudge factor. One option is to just set u=0. This is the assumption that if you’re dead, you’re not getting any utility from anything, so your utility should just be zero forever.

That’s what Chad Jones is doing in this paper, and it’s that choice that drives his result. But it’s a highly dubious assumption. Recall that ln(1) = 0. So this means that if you assume u(extinction) = 0, you’re saying that “humanity going extinct” is no worse than “humanity existing for all eternity at some baseline level of consumption”. That doesn’t sound realistic to me.

A better choice — which is what some economists do — is to set the utility of death equal to some constant, and then try to calibrate that constant against data on risky behavior. The constant wouldn’t be zero, and it would greatly alter Chad’s calculations on how much benefit we’d need from AI in order to accept existential risk.

But even here, we have to be cautious. Human extinction is not the same as an individual’s death. A lot of people would probably accept a lot less risk if they knew they were risking the lives of their families, friends, countrymen, and fellow human beings, than if they were only risking their own life. So you have to be very careful when looking at how much people care about the risk of the whole species dying.

Chad was not being careful here. I know this was 2023, before existential risk seemed like a serious thing to people outside the AI industry. But now that he’s at Anthropic, he should be more circumspect about these things.

4. Millennials are doing better than Boomers (but are more unequal)

During the 2010s, there was a pretty common narrative that economic progress had stalled in America, and that the Millennial generation had been screwed over. You still hear a few progressives argue this, but in general this narrative has been in retreat as new data has come in. As the Millennials have eased into middle age, it has become apparent that like every generation before them, they have experienced substantial economic progress. I first blogged about this shifting narrative back in 2023:

But I was a bit late to the party here — bloggers like Jeremy Horpedahl had been pushing back on the “generational stagnation” thesis for years.

Anyway, as we get more data, Millennials’ advancement becomes even clearer. Tyler Cowen points us to Corinth and Larrimore (2026), who measure the income of each generation after all taxes and government transfers are accounted for. You can basically see the results in one chart, which I’ve helpfully labeled:

These are income distributions — they measure how many people in each generation are living at each level of income at an equivalent age, after adjusting for inflation (i.e., after adjusting for changes in the cost of living). What you can see here is:

• There are a lot fewer Millennials making less than $30,000 (in 2019 dollars) than any other generation did at age 36-40.

• There are more Millennials making over $40,000 than any other generation, and Millennials dominate other generations at every income level above $40,000.

• It’s hard to observe incomes above $140,000, so these are left off the graph.

• It looks like there are a few more Millennials making exactly zero income than other generations. This could reflect people who are still being supported by their parents, or people who somehow fall through the cracks in the data gathering process.

• The Millennial income distribution is more spread-out — as Millennials have done better overall, some have seen bigger gains than others, resulting in increased inequality within the Millennial generation.

This is a pretty reasonable description of the reality that we’ve all seen over the course of our lives — most Millennials are more comfortable than their parents were, just as most Boomers are more comfortable than their parents were.

This doesn’t mean government redistribution is unnecessary — indeed, Corinth and Larrimore explicitly calculate income after redistribution has taken place. And America has become more redistributionary. So government is helping produce upward mobility for the poor. This is a big government success story.

But the narrative that Millennials are falling behind, and that the Boomers screwed their children over, just doesn’t hold up in the income data.

5. Why did America deindustrialize?

As countries get richer, manufacturing tends to become a less important part of their economies. You can just look at the declining trends for manufacturing as a percent of GDP in rich nations over the years:

A lot of people think that this happened because we outsourced our manufacturing to China. Other people think it’s because our demand for manufactured goods topped out over time, and we started to want to consume more services.

In fact, it was both of these! Richard Baldwin has a post decomposing each rich country’s deindustrialization into three factors:

1. How much demand shifted away from manufactured goods

2. How much final goods production shifted from domestic production toward imports

3. How much intermediate goods production shifted from domestic production toward imports

He finds that the rich countries have very different stories, even if their overall numbers look similar. Here’s the key chart:

America did actually outsource a fair amount of its final goods production, but this was almost balanced out by onshoring of intermediate goods production (at least, in terms of monetary value). Almost all of America’s deindustrialization since 1995 came from Americans spending a smaller % of their money on manufactured goods.

For other countries, it’s a different story. Germany and Japan actually spent more on manufactured goods, but lost tons of market share in the intermediate goods sector. For France, Canada, and the UK, all three factors contributed to deindustrialization.

This is very interesting. It implies that the simple, common story of “we outsourced everything to China” holds true for other rich countries — at least, in a generalized sort of way — a lot more than for the United States. For the U.S., the main reason we make less stuff is that we want less stuff — at least, relative to how many “experiences” we want to consume.1

6. How much did the Iran War help Russia?

The wheels are falling off the bicycle for the Russian bear. Ukraine is taking far fewer casualties, as it switches to a drone-intensive way of war that Russia has so far been unable to match. Russia is losing over a thousand men a day in futile assaults. Meanwhile, Ukrainian long-range strike drones are wreaking havoc on Russia’s logistics, strangling occupied Crimea, destroying Russian refineries, and causing fuel shortages throughout Russia. Meanwhile, the strain of war production is starting to cause cracks in Russia’s government finances.

Russians had hoped that Donald Trump would ride to their rescue, pressuring the Ukrainians into ceding territory. He certainly tried, but was ultimately unable to bully the stubborn Ukrainians into backing down. But some believed that Trump would help Russia in a more indirect, accidental way — by launching a war against Iran, the hapless U.S. President would cause oil prices to spike and flood the coffers of Russia’s government with much-needed cash.

Indeed, when Iran closed the Strait of Hormuz — which ultimately decided the war in its favor — it caused oil prices to soar:

But at least by April — the most recent month we have data for — the benefits to Russia had been a lot smaller than Putin supporters hoped. Matt C. Klein had a good post about this:

The Overshoot

Russia's Underwhelming Oil Revenue Windfall

A barrel of Brent crude oil cost about $103, on average, in the month of March, up from $66/barrel in October 2025-February 2026. Thanks to sanctions, the price of Russian Urals crude was somewhat lower in the months before the war with Iran, averaging around $55/barrel, while the price…

Read more

3 months ago · 27 likes · 3 comments · Matthew C. Klein

Here’s the key chart:

Why has Russia only reaped a small windfall? One reason is the strengthening ruble, which means fewer rubles of revenue for every dollar of crude oil sales. Another reason is that Russia has been paying its refiners to keep fuel prices down in the face of Ukrainian attacks; because they handle these payments as tax deductions rather than as expenditures, it reduces the total amount of oil and gas revenue.

But there’s a third reason, which is a lot scarier for the Russians. The country’s crude oil production is falling:

Ultimately, it’s production declines that doom a petrostate — just look at what happened to Venezuela when Hugo Chavez starved the state oil company of funds for reinvestment.

Russia needs to invest huge amounts of money to expand production in Siberia, where the easily accessible oil is gone and only harder-to-get supplies remain. War spending may be starving the oil industry. And Western export controls may be successfully starving the Russian oil industry of the technology it needs to build and maintain its extraction infrastructure.

If this is true, then Russia is in big trouble. Crude oil exports are the big prop holding up Russia’s whole economy — no amount of macroeconomic policy or war mobilization can compensate for its loss. Putin’s Russia is a classic petro-empire that uses wealth extracted from the land to fund imperial conquests; when the oil runs out, such empires collapse.

7. Not being fat is probably good for you

Usually, when you see eye-poppingly large estimates for the effect of some sort of policy or medicine or whatever, you should immediately distrust the methodology of the paper. But Rebecca Diamond is one of the most serious and respected empirical microeconomists in the business, so when she says that GLP-1 drugs have almost magical effects on women’s lives, we should at least sit up and listen.

Diamond’s paper compares people who go on GLP-1 drugs to A) those who say they want to go on the drugs but don’t, and B) those who go on the drugs a bit later. For men, she finds relatively few life changes beyond weight loss itself. But for women, she finds some pretty big effects! According to Diamond’s estimates, single women who go on the weight-loss drugs are 29% more likely to get married or cohabit over the next three years, and women without jobs are 27% more likely to get jobs, relative to the women who didn’t go on the drugs.

That’s a pretty dramatic result! It gives us one more reason — beyond the obvious health benefits — to treat weight loss as a technological problem, and just get on with it by any means necessary. The “fat acceptance” movement has gone too far — we should treat excess weight not as a piece of who we are, but simply as a physical impediment to be managed or removed. Fat is basically just a fat suit, and you can take it off if you want.

That said, there are some reasons why you shouldn’t put too much faith in this one empirical result. Although Diamond of course does an incredibly thorough job in trying to compare the GLP-1 users with the non-users in an apples-to-apples way, there’s still the possibility that people who actually go ahead and use the drugs are just different than people who don’t, or who use them only with a lag. They might be more purposeful, more motivated, etc. And that could explain at least part of their greater likelihood of getting a partner and getting a job.

What we really need here is a natural experiment — some policy that affects how easy it is for people to get GLP-1 drugs, preferably with different timing in different places. If we find similarly positive effects from that sort of study, we can be even more certain that these drugs are wonder drugs.

(Of course, drugs aren’t the only thing you should do to lose weight. You should also go for a bike ride! Especially if you’re a bear.)

8. The latest on AI and jobs

Everyone’s favorite subject (except for bears on bikes, of course) is AI taking jobs from humans. I’ve tried to use these roundups to keep abreast of the most recent evidence in this area. This week, we have a study by Kharazian, Simon, and Stevens using private data to examine what happens when companies start using generative AI.

The authors find that when companies start using generative AI, they hire more humans, not less:

It’s not just total headcount, though. Entry-level headcount rose too!

This flies in the face of the typical story that the current “no-hire, no-fire” economy is due to companies adopting AI instead of hiring entry-level workers.

That doesn’t mean that AI isn’t reducing job churn, of course. Uncertainty about how to use AI, or uncertainty about how AI might affect their industries, might be keeping companies from hiring new workers, even if they don’t adopt AI. That’s a research direction worth looking into, and basically no one’s talking about it.

Anyway, Kharazian wrote a blog post about the new findings:

Ramp Economics Lab

We can finally say AI isn’t killing jobs

Dear Colleagues: The most important economic question of this decade asks how AI will affect jobs. Everyone wants to write that paper. Until now, no one has had the right dataset, so existing research has relied on a combination of guesses, surveys, AI exposure scores, and self-interested punditry. In fact, a recent paper from Stanford said the ideal da…

Read more

a month ago · 29 likes · 1 comment · Ara Kharazian

The simplest story here is that AI is still mostly a complement to human labor rather than a substitute. That could change, of course, as the technology advances further. But for now, AI is behaving pretty much like a normal technology.

Subscribe now

Share