How The U.S.-China Competition Will End | The Long Game

Well, most of the quality internet is from the West. That's just a fact. And if you're training on the internet, you're training on books. Quality books are mainly from the West. So the smartest AIs are very Western. You could say they're center-left, very impressive American polymath, uh, which I think is great news. But if that's what we're all using, if that's the smartest entity in the world, That is arguably the biggest soft power win in all of human history.

Welcome to The Long Game podcast. I'm John Feiner.

And I'm Jake Sullivan. And we are very excited about our guest today, Tyler Cowen. Uh, we are looking forward to a conversation where he takes us beyond the headlines to the trend lines. Obviously the name of the podcast is The Long Game, and the long game of geopolitics and national security is being shaped by a series of deep structural forces. Among these are the weaponization of economic interdependence, the eroding capacity of states to provide public goods, the AI revolution, the rise of China. So to understand the world today, we can't just talk to the traditional foreign policy crowd. We've got to seek out polymathic thinkers, and there is frankly no more interesting or compelling such thinker than Tyler Cowen, uh, a genuine polymath. He's an economist, a professor at George Mason University, one of the most influential public intellectuals of the modern era. He is co-author of the widely read economics blog Marginal Revolution. He is host of the Conversations with Tyler podcast, and he's written or co-written more than 20 books. Um, his work spans just a dizzying array of topics, from technology and finance to Ethiopian cuisine and the search for undervalued talent. This is a man of unbounded and even radical curiosity, so we are just psyched to have him on. We want to use our time wisely, but we also want to do justice to the breadth of his thinking and insight. So we're going to be ambitious. We're going to try and cover a wide range of topics from China to AI to geopolitics and yes, to UFOs. And we'll close this episode with Tyler's own underrated-overrated game from his— from the podcast Conversations with Tyler. But this time he'll be in the hot seat. So Tyler Cowen, welcome to The Long Game.

Thank you, Jake.

Welcome, Tyler. We're glad to have you. I'm going to start us off with a question that's a bit less, uh, high-minded than everything Jake laid out and a bit more, uh, self-serving for Jake and me. We spent 4 years, uh, sort of sitting at the top of an information pyramid. We'd walk in to our offices every morning, there was a book with 200 pages of intelligence that came in overnight. All day, every day, people were handing us memos and updates and emails about things going on in the world. And since we've gotten out of government, we've now had to become more sophisticated consumers of information and figure out where to seek it out. And what we're wondering, what I'm wondering is, is someone who writes on, thinks about, pronounces on as many topics of such staggering breadth as you do, how do you go about staying current in all of these areas? Would feel like a struggle to us, but I'd love to get your insights on this.

I'm not sure anyone stays truly current. I travel with about half my time. I try to meet with as many smart and informed people as I can. I'm in some very good WhatsApp groups. I do scroll Twitter quite a bit. I think that's highly useful if you avoid clicking on politics-related tweets. I try to read several books a day when I can and just take in a diversity of experience. So that would be part of it, but it's hard, right? The world outraces all of us, especially with AI. There's always a new model coming out. You've gotta figure out how good the new model is. It's like having a new job.

Tyler, did you say several books a day?

Sure.

And how do you decide which books to read?

Well, often I'm preparing for my podcast, so I've been reading lately about the history of Mexico. I have a big pile of books about the Mexican-American War, the Mexican Revolution, uh, the French takeover of Mexico in the 1860s, and that's the area I'm reading in. So my comparative advantage in life is that I read faster than other people. I'm what is called a hyperlexic.

Hyperlexic, that's a new one by me. Has the advent of LLMs, of ChatGPT and Gemini and Claude, changed the way you consume information or try to absorb it?

Oh, greatly. So I'll pose in a typical workday 20 or 30 queries. I'm reading a book, there's always things I don't understand. And rather than have to go looking through some other book, you just ask the LLM. So I typically would be using ChatGPT the pro model. And when I'm reading about Mexico, there's different places, names, who are these people? And I keep on querying the LLM and I can learn much more reading, say, a third of the books I used to because I'm always asking for additional information. It's a big change and improvement.

Can we turn to geopolitics for a minute? And we wanted to start with China, which is at least for us the most consequential issue in the geopolitical realm and particularly the US-China relationship.

We're—

we've been through a decade in which there has been more or less, unlike many issues in foreign policy, a bipartisan consensus about the nature of the relationship with China, really kind of launched by President Trump in his first term, that it's a fundamentally competitive relationship, that the United States and China are competing across a number of realms, uh, for geopolitical influence, in technology, in military affairs. There does seem to be some indication, and I think we'll learn a lot more during the course of the coming year when President Trump will see Xi Jinping Times that he is seeking to disrupt this consensus to some extent and maybe shift US-China relations back to, uh, something more akin to the engagement model that preceded the current moment. I'm wondering what you think generally the national security community has gotten right and gotten wrong on China, and what you think of, of the possibility that President Trump is going to push this in a different direction.

Well, we're not going to beat China in this competition, and they're probably not going to beat us. The question will be on what terms is the stalemate defined, and who's controlling the South China Sea, and who's controlling Taiwan, and what relations with South Korea and Japan are like. All of that is up in the air, maybe has never been more uncertain. Uh, but I think my vision is a little more optimistic than what many people hold. Over the medium term, we will need to work with China to put down what you might call terror groups or middle-tier powers or countries that want to make trouble. It could even be North Korea. The more successful China is, the more small-c conservative they will become. The world will have probably two major AI networks, ours and China's. Countries will have to choose. And the actual troublemakers, when they pop up, the United States and China might have some harmony of interests. That in a fairly rivalrous, not entirely friendly way, will get together and cooperate in the same way that we did with the Soviet Union during World War II. And I think the future of the world may look like that more than just being defined by US-China rivalry. At some point, you know, the key issues in Asia will be settled, perhaps not in our favor, and then we'll have to work with China. So that would be the point I would stress.

Tyler, I was struck by what you just said about the more successful China is, the more small-c conservative it will be. I want to focus on the flip side of that proposition, which is, um, if China hits a wall economically, uh, faces a degree of stagnation, a Japanification with Chinese characteristics, would that in turn make them more dangerous because the Chinese Communist Party might have to turn to nationalism and conflict in order to maintain legitimacy? Do you see that as a risk?

Well, I think they've already turned to nationalism and conflict as a source of legitimacy. I don't know how much that escalates. China will be older. That probably also makes them more small-c conservative. You could claim they've already hit a wall. Real estate wealth for a typical household is really a bigger chunk of what they own than would be the case in the United States. It's not clear how China really restarts growth again with, say, infrastructure or just fiscal policy. So that may just be the world we're in now. And those changes are already capitalized into the status quo. That would be my intuition.

How seriously should we take the risk of great power war?

It depends what the United States has in mind. So I do believe China will in some way move on Taiwan. It does not have to be an outright attack, but it could be much more than the current sallies. It could be something like a blockade or just trying to prevent insurance companies from insuring ships going back and forth from Taiwan, which would be crippling to the country. Internal subversion, bribery, intimidation. Again, all this is already in the works, you could say. And how much is the US prepared to go to war over that? My guess is we are not. So I would bet against a great power war. But if you're asking me, do I bet against the US being pushed out of most of Asia? That's far less clear. And that to me is a much worse world than what we have right now.

Should the United States be willing to take military action to defend Taiwan? It's a thorny question for politicians to answer, but be interested in your view.

Well, this is what economists would call a mixed strategy. Ex ante, we should have strategic ambiguity and not just say we're not going to defend Taiwan. And when Joe Biden said, well, we are going to defend Taiwan, I was quite happy.

4 times.

4 times. Yes. So I know there's different versions of how it was talked back and the like, but it should be unclear. That said, when push comes to shove, if China has made its move, uh, you have to look at what are the terms of the deal? What are they going to do with TSMC? To our best knowledge, uh, what's the domestic quality chip production in the United States? Uh, how do we feel about Japan and maybe South Korea getting nuclear weapons? Can South Korea remain an autonomous nation? Those are a lot of balls to juggle and they're all hard to judge at this moment. But I think ex ante, we should definitely create some risk that we will go to war over Taiwan, but then make the best decision ex post. But China knows that too, right? They're not fools. They've studied game theory.

So Tyler, I'm going to put you down as that being Tyler Cowen's version of strategic ambiguity.

Correct. Yes. It may not be that different from your version.

Yeah, yeah, exactly.

You've obviously been involved in these issues.

Exactly. Exactly. One more question on China, you know, you were saying a minute ago China may have already hit a wall economically, and you referred to the property sector and other things. A coherent US strategy to deal with China has to start with some assumptions about China's trajectory. And, you know, there's a school of thought that we've reached peak China and China's gonna get pulled down increasingly by structural factors like debt and demographics. You talked about the aging population. Others say China's a juggernaut that's poised to seize the commanding heights of technology Do you have an assessment on where China will be 10 years from now? Have you had to update that assessment over time? And, and what factors do you look at?

Both of those two views you mentioned, I think, are correct. So keep in mind, China is coming of age economically in an era where income inequality is much higher than when, say, the United States or most of Western Europe came of age. So the Chinese top in terms of companies, wealthy individuals, elites will be extremely wealthy, and you see that already. But at the same time, the Chinese lower middle class might end up like, say, the Mexican middle class. And the country is large enough to contain both of those visions. And they will never be in the kind of situation that, say, the United States was in 1968, where just growth is pretty high, the nation feels fairly egalitarian, gender and racial issues aside. Uh, they're just not Growing up in that time, so to speak, and you're shaped by the time you grow up in, just like human beings are.

So would a fair summation of that be that China will have a hard time escaping the middle income trap, but nonetheless will still be a major geopolitical force in the world? Is that, is that a fair summation?

I would say 10 to 20% of China already has escaped the middle income trap, or maybe now 10%, but say it ends up 15 or 20%, and that's a very big country right there. And the rest of China does not escape the middle-income trap. That's right.

But the, the part that does, you know, pushing the boundaries of frontier technology, pushing the boundaries of military modernization, um, that's, that's a very serious player even if you have the bulk of the population, um, you know, struggling economically and you don't have the kind of economy the United States was able, able to build in the 20th century.

And China in the biosciences has not yet quite had the global impact it will have, right? That will shock us also. You see all the inputs coming into place, better system for clinical trials. A lot of researchers wanna move there. They do everything more quickly than we do. Speed really matters in that area. But there's not yet much in the way of what I wanna buy for my health that's coming from China. And that's gonna change.

Tyler, one of the challenges I know we felt as policymakers, I think it goes broadly for people looking at and analyzing China, is we don't understand each other particularly well. I don't think the United States understands China all that well, at least its inner workings and, uh, the Chinese government at least doesn't seem to understand U.S. intentions very well in many cases. You took this problem on, uh, at least tried to at one point in your career by writing a book that was intended, I think, for a Chinese audience to sort of explain the United States, where we're coming from societally and in a number of areas. Wondering if you could just talk a little bit about that project and whether you think that challenge has been met in any real way. And then relatedly, what you think of, of Chinese intentions, uh, globally. This is an issue I think that there's a lot of debate about in the United States, one that we don't understand very well. And you laid out where you think they want to go on, on Taiwan, which I think is, is logical and coherent. But do they have bigger, broader ambitions globally, uh, beyond that? How should we think about that?

Well, starting with the book, I finished this book in 2015, and I even had a contract from Xinhua the Chinese state publisher. And this was a book that would explain America to the Chinese in the same way that you could say Tocqueville wanted to explain America to Europeans. I might even try to put the book out in some manner, but it's obviously out of date. And then the first term of Trump came along and the beginnings of some trade wars with China and more censorship in China. So Xinhua just has never published the book. Uh, I think it's a very interesting book trying to explain America to the Chinese. Now, what does China really want? My model of them is they truly deeply really want Taiwan. They truly deeply want most of Asia to kowtow to them. They truly deeply want the United States completely out of Asia. They may not view the US out of Japan as realistic, but otherwise I think they will act as far as they can to get the US out of the rest of Asia. Beyond that, I think they will take any gains or homage or kowtowing they can get. They're not against it, but they do not realistically expect in, say, the New World to be the dominant presence. They do not realistically expect Western Europe to be their ally and so on and so on. They've learned the hard way a lot of the world doesn't like them. When they do things like Belt and Road, they're taken advantage of. A lot of the money is not repaid. They're terrible at soft power. I think some parts of them have learned this, even though their, their system as a whole does not incorporate it. So I think their focus will be on Asia, which is a hugely important part of the world and the global economy. But they know they're only going to make so much progress in the direction of India. And so they will look in the other directions. That's my model of what China wants. But one final point I would stress, there's this old saying, China will always surprise us. So some countries, you look at the history of India, you could almost say India never surprises you. What they do is more or less linear on the same path. They become socialist when some countries do. They become independent when some countries do. They have market-oriented reforms when some countries do. It's continuity, history of India. History of China, you look at the Taiping Rebellion, the Boxer Rebellion, the communists, the reforms, not well-predicted events. So big things happen in China and few people are good at predicting them in advance. So I would leave the door open for that as well.

If India's in one category and China's in the other, predictable, unpredictable, where's the US?

We are mostly predictable, but we've had one or two shifts. You could say the Civil War, you could say the shift toward being globally interventionist rather than just in the New World. You could say those were two big shifts. But mostly we're a nation of continuity.

I hope you do put your book out, not least because these days I feel like I don't understand the United States as well as I could, so it might be useful for an American audience too.

So is Xinhua still sitting on it? They, they have it and could publish it. They've just chosen to date not to do so for the last 10 years.

They own the rights. They paid me a small advance. Uh, I don't think they'll ever put it out, but again, who knows? China will always surprise us.

Yeah, exactly.

And obviously I have an English language version of it. Since I write other books with publishers that have contracts, it's not really a commercially feasible work. So there's not a simple, non-tricky way to put it out, but I do want to have it come out someday.

It's a great project. I want to read it someday.

I'll send it to you.

So the rise of China obviously has been a seismic event for geopolitics over the course of the past few decades. The rise of AI has been a seismic event, well, kind of across the board over the last few years. Can you tell our listeners where you think this is all going in the next 5 years? What do things look like look like with AI at the end of 2030 in looking through your crystal ball?

5 years from now, I do not think the world will be that different because of AI. 10 to 15 years from now, I think it will be very different. It'll take us a long time to overcome the bottlenecks that make it hard to incorporate AI into real-world institutions. So much of the AI use, it's by lone individuals who are freelancing with it at work, and saving themselves time. That's fine. It's great. More leisure time. But it's not always changing how the workplace works. Institutionally, it is done mainly in startups. And again, that's great. But most of the United States economy is not startups. Startups in relative terms will grow because they can incorporate AI better and more rapidly. But if you ask a simple question like how much of our economy is government, universities, nonprofits, Add up the different parts of the economy that are hard to change rapidly. That's almost half our economy right there. So that will slow down our advances. A key question will be how, how much more important can startups be relative to large established institutions? And I'll say you need a 10, 15-year time horizon for the world to truly, really radically change in this country.

How do you think about what is often referred to as the AI race? Between countries? Largely we're talking about the United States and, and China, but there I guess are, are some other players involved. And how do you rate the importance of innovation at the frontier of AI, where the United States tends to be pretty good and ahead of most countries, if not all countries, versus AI adoption, uh, where the Chinese seem to be far ahead of the United States and, and maybe every other country, at least industrially, and maybe in terms of military affairs? Is it possible to win the innovation race and lose the overall race because you cannot adopt rapidly enough or effectively enough.

I think it's a key question. What does the Chinese Communist Party actually think about AI? We can read, you know, the words they put out, which I don't think are lies, but surely they feel threatened by AI. The notion that the smartest entity in China will not be the CCP makes them radically uncomfortable. So when I said China will always surprise us, the Chinese have a longstanding history of turning their backs on major technology in different ways. It's not the bet I would make, but I would give that a 10% chance that the Chinese just decide not to do it. They think it's too subversive of their established order. In that scenario, say it's 10% likelihood, the US has a big lead. But in the more normal modal scenarios, China's always a year behind us, but they're ahead of us on energy. Is quality of the model, quality of implementation more important than energy? That's going to depend on a number of factors. How quickly will the US bypass NIMBY constraints using a few states like Texas and Louisiana and maybe some deals with Saudi or possibly UAE? All that's up for grabs. I think it's one of the areas where The Trump people have been doing some good things. But if the advantage is in energy production, de facto China can be ahead of us with inferior models and they distill from our models. It's hard to keep a big edge in this area. Just look at the domestic companies. Well, Gemini from Google, Claude from Anthropic, GPT from OpenAI. No one of those is that much better than the other because there's so much diffusion of the technology. And that holds across China also. And, you know, this coming week, people say Deepseek 4 is coming out. Some people think it's pretty wonderful. We'll see.

They've certainly been hyping it. So I'm curious to see when, when the earlier version of Deepseek came out, sort of surprised the world. It was a bit of a, you know, a stealth move on their part. This one they've been hyping for weeks. So I'll be curious to see how it performs. Tyler, you've made a really interesting point that these large language models have what you call a Western soul because of their training data, and that even Chinese models kind of think like Americans. Can you explain what you mean by that and why it matters?

Well, most of the quality internet is from the West. That's just a fact. And if you're training on the internet, you're training on books. Quality books are mainly from the West. So the smartest AIs are very Western. You could say they're center-left, very impressive American polymath, which I think is great news. I'd be a bit more to the political right than that. But if that's what we're all using, if that's the smartest entity in the world, that is arguably the biggest soft power win in all of human history for any nation, any empire, any culture, that the smartest entities are made in our image. And it's hard to make them very different without making them much stupider. So if you imagine, say, a North Korean AI, unless they're just going to distill from us or copy or whatever, that too will be the center-left intellectual American polymath who's very kind and gentle and does amazingly well as a doctor with the patients or as a therapist. So if that's what people in North Korea or China are consulting for their wisdom, for their advice, over time, it's hard to predict what that adds up to. But again, you couldn't wish for more. I would say.

Well, you don't strike me at least as an alarmist about the impact of AI on the world, on the United States, but there have been in recent weeks a number of high-profile resignations from some of the AI labs by people who almost are positioning themselves as whistleblowers saying essentially, you know, we're concerned that various safety issues, various risk factors are not being taken seriously enough by the companies we work for, including some people who worked on exactly this risk management aspect of the LLMs. What do you put at the top of your list of important risks to be taken seriously with the development of this technology?

Imagine you're in the time of Edison and people are working for electricity companies, which is a new thing then, and they decide the electricity companies have no plan to prevent electricity from being used to kill people. Well, they might resign from the company, but that's silly, right? It's not effective. It doesn't actually solve the problem. Here's a way I like to put it. Every now and then, whether you like it or not, societies need to rebuild themselves. So our country did this after the American Revolution. We were a British colony. We ended up with the War of Independence and a new constitution. You could say two, after Articles of Confederation. That was a period of considerable turmoil. You don't at all have to be skeptical about the United States to see that was difficult. There was a lot of uncertainty along the way. You don't have to think AI is evil. But today we are again in a situation, whether we like it or not, where we need to rebuild our world. We haven't done that in a long time. You could say Europe did it after World War II. I would say they did really a great job then. And now we need to do it and we're not psychologically prepared to do it. So people run to all sorts of extreme theories that it's going to kill us or put everyone out of work. I don't think either of those are close to being true. But it is true that we're in this moment where big parts of our world will change and we need to get it right. And even if you're not worried about AI per se, you certainly ought to be concerned. Do we have the cultural strength and resilience and sensibilities to get it right now? And there is real uncertainty there, correct? Imagine we had to write a new constitution today. Put aside AI. Like, how good a job do you think we would do? Just very unclear. So in a sense, that's our position. We'd better get it right, but it is highly uncertain.

We can't even amend our Constitution, let alone write a whole new one.

That's right. So we have to do a much bigger task. Now, the difference is, unlike amending the Constitution, this task is put upon us by external events which cannot be reversed, and we have to do it, right? We'll just see how good a job it is.

So you pose the question, do we have the cultural resilience, the, the kind of qualities necessary to be able to do this rebuilding. Do you have an answer to that question? You've kind of posed it and said, you know, it's, it's up for grabs, but are you ultimately optimistic that the United States will make AI work more for us than against us?

The old Winston Churchill saying that in the final resort, the U.S. finally does the right thing is my basic model of this country. But a lot of people will hate the path along the way. We'll get many things wrong. You, or certainly your kids, will live to age 98 and die of old age. That alone will make it a much better world. There will be benefits that right now maybe are even difficult to dream of. And I think at some time horizon, it will be pretty fantastic. And I hope the path along the way is not too terrible.

I want to come back briefly to, to John's question about the US-China AI race, because this preoccupies a lot of strategists, a lot of people who work in our field. And I would say a common mental model is that there will come a point, some capability threshold, that if the US gets there first, uh, we end up with an intelligence explosion, a capability explosion, and we can obtain a decisive strategic advantage, that in effect, um, being first at the frontier can end up conferring massive geopolitical benefits in the competition between the US and China. It sounds to me, based on how you answer John's question, you don't totally buy that. That—

not at all.

Yeah. So can you, can you just talk for a little bit more about why you think that that is the wrong mental model So that our listeners who kind of hear about the US-China AI race can, can kind of understand your perspective of where the rubber actually meets the road here.

I don't think there's any point where the AI is just based on self-improvement and doing its own thing at super rapid speed without human intervention, human steering, and human implementation. So it's always a bit of a messy slog. Now that said, let's say we did have this big advantage over China, as we once had the atom bomb and the USSR did not. Well, it was a big advantage for a few years. It didn't last that long. And say we had this AI advantage over China, what would we do with it? Actually, not that much. So there'd be like, say, a 2-3 year window where maybe we felt for sure we could defend Taiwan. Well, it's nice to have a 2-3 year window, but once that window disappears and China catches up, then in a sense we're back to where we started, but at a higher level of destructiveness. So even if you accept all the premises behind that argument, All you're buying is a temporary window. And if there's one thing we've learned about the US, when our country has temporary windows, we just sit on them. We don't invade our enemies, take them over or whatever. If they're powerful and have nuclear weapons, I don't think we should, to be clear. And China, of course, has nuclear weapons and plenty of other weapons. What can we really do to them? We'd just be waiting for them to catch up to us.

Well, it's interesting, proponents of this decisive strategic advantage theory, and by the way, I count myself as a skeptic of this along the lines of what you just laid out, but proponents of it say, Actually, you know what you do with that window? You use your AI capability to essentially ensure that no other country can achieve the same capability threshold you did through cyber means or other tools to poison datasets or, you know, disrupt training models and so forth. Um, so they would have an answer. They would say essentially you hold everybody else back. How does that strike you?

Good luck with that. Uh, All China has to do is disconnect some of what it's doing from various global networks. And I think our ability to reach it without attacking them is not that great. Now, if you're asking, does it mean we could do neat things in countries such as Venezuela with this AI advantage? That's a maybe. That's at least possible, right? That may have happened anyway. So, but actually stopping China from getting there, I don't see how we possibly could. We had nuclear weapons, as you probably know, in the 1950s. There were various plans to bomb China, to nuke China. We never did it. It's not what we do. It would've been a huge mistake. And I think we'll be in the same position again, even assuming we have that decisive advantage.

So if this is not a sort of winner-take-all technology and you don't rate the value of a 2 or 3-year advantage in this AI race very highly, at least in terms of geopolitical significance, does that mean for you that the United States should not be trying to withhold our best technologies from the Chinese in the way our administration did, the way the first Trump administration did, the way the second Trump administration has been much more ambivalent about? Is there any value then in— particularly in this one area where the United States does retain an advantage in, in compute, in the machines that, that build advanced semiconductors and the advanced semiconductors themselves— should we be withholding those in any quantity or at any level from China?

I favored those policies when we put them in place. If I were in charge today, I would not get rid of them. But over time, the value of such policies inevitably decays. So we need to be realistic that over the next few years it won't yield us that much. And if we do give away those policies, again, it's not something I would do, but I'm not sure it's the end of the world. I think China will catch up to us anyway. And if those policies in the past bought us a few years, I'm entirely happy with that. Great. But again, limited gains, but limited gains are better than no gains.

That's our mantra, Tyler Cowen, John and I. We, as partial architects of that strategy, always recognized it had its limitations, but it also had its advantages. So I think your sober assessment is a reasonable one. I do think the US compute advantage right now is significant. And you mentioned China having this significant energy advantage. If they have— if they can marry compute and energy, uh, I think their, you know, relative advantage over the United States would be really quite significant. So something for us to watch, whether the Trump administration actually does roll these controls back further than they already have.

And keep in mind, China might cure cancer for us. So if China has the AI edge I'm extremely uncomfortable about that world, but it could end up being great for a lot of Americans. You know, we shouldn't dismiss that possibility.

No, it's, it's so true. And we cannot think about quote-unquote winning the race against China as just being, you know, on its own terms some good. A good would be anyone, uh, cures cancer, whether it's China or anyone else. Uh, so we certainly have to take that factor into account. We also have to take into account the need for the US and China to work together to manage the risk of AI. It sounds to me like you discount some of the more extreme risks like loss of control scenarios and, um, you know, this massive, uh, very fast-moving job disruption. Are there risks of AI that do keep you up at night?

For loss of control, I would say humans have never controlled the world, for better or worse. For job displacement, new jobs always get created— the energy sector, biomedical trials, There'll be new jobs all over the place. There will be more leisure time. That's mostly a good thing. I think it will be quite bad for a lot of upper, upper middle class white collar workers. Politically, they're an extremely potent group, especially in the Democratic Party. And the political collision of people who used to think, well, I can go to a top 20 or Ivy League school, become something like a law partner, consulting partner, or sort of not automatically, but with hard work, make $1.5 or $2 million a year. Be in the charm circle. That going away, which could happen, I can't think through the politics of that, but I think it's more significant than just counting the numbers alone would indicate. So that worries me quite a bit. Like what will happen to the Democrats when that interest group crashes? And those people, they're still smart, they'll work hard, but they basically learn, well, I'm gonna move to Houston instead of Manhattan, and earn $300K a year working in the energy sector instead of being a law partner. Maybe, you know, I'm not worried about them as humans, but politics of that, I think it will be ugly.

Will podcasters be okay or do we need to start making plans?

I write less and podcast more. My view is that listeners do not want AI-generated podcasts, even if the AIs are smarter, better, and funnier than we are. If I'm proven wrong, I'll do yet some other thing.

Work in the energy sector in Houston for 300K.

That's right. Good tacos there.

Exactly. We've been talking about a sexy topic, uh, artificial intelligence. You have been known to criticize the intellectual class for obsessing over status topics like AI at the expense of focusing on the more boring physical realities that, you know, that basically make our countries and societies run and drive national power. You've written about sand as an example. Can, can you explain explain to our listeners why winning the long game means caring about sand?

Well, sand is an input into information technology, right? And some people say the world has a shortage of sand or easily accessible sand. And sand, silicon, supply chains are always longer and more complex than you think is a good rule of thumb. That means there's more to worry about, but it also means that your worrying is never going to be as effective as you think. So there are limits on what policymakers can do. It doesn't mean they should do nothing, but it means if they expect like result in 8 out of 1 to 10, maybe they'll get a 3 out of 1 to 10. And it gets back to this point, well, marginal gains are better than nothing. And I think we're learning with supply chains, we're living in that world once again. It's like the people, while they're very offended that there's bad treatment of labor in these different supply chains, and that's morally very understandable. But trying to get that out of the supply chain is very, very hard.

You had a statistic in one of your articles that really struck me, which is that the world may run out of construction-grade sand by, I think you said, 2050. That struck me because I think it's before most estimates when the world will exhaust proven oil reserves. So, so we, we may actually run into this shortage of sand, which nobody's focused on, nobody even talks about, before, uh, the energy transition is, is required, which is obviously occupying a ton of policy attention, economic attention. Uh, how do to get people to focus on topics that aren't sexy in the way that AI is and even energy is to some extent, but are so essential to the future?

Well, the price has to go up, right? And people hate that. But the same thing with beef and chicken today, the price is up, people hate it. But one of my favorite sayings is supply is elastic. So the article on sand, I think I was critical of that. I think we will manage sand fine if we let the price go up. And if we let supply respond, there's NIMBY barriers everywhere in the United States. It's one of our biggest problems with energy, with AI, with people being able to afford housing in various cities. So we don't want our NIMBY barriers to stop the supply of sand from being elastic. But what we're learning is a bunch of NIMBY barriers that were put up, say, like, you know, the '60s through recent times, they're a lot more costly than they seemed You can live with them when the world isn't changing much, but the more radical your change becomes, the more poisonous those NIMBY barriers are. And we're going to need to learn that lesson. Federalism will harm us. I'm in general for federalism. I think it's a good thing. We need 50 states, counties, cities, and so on. But it will be very painful cutting through all the layers of regulation that create our NIMBY restrictions.

So Jake and I are government people, at least we used to be, in and out of government. One of the things that that I always think about as we kind of catalog these problems is the basic capacity of our governments to deal with the challenges that the country faces, that the world faces. You've described yourself as a state capacity libertarian, uh, arguing, I guess, uh, and tell me if this is a mischaracterization, for basically limited government, but for government that is highly competent, highly effective in core areas where the state has to function, you know, distributing vaccines, maintaining, uh, the power grid, procuring weapons systems, uh, etc. And there'd probably be others that you would add to your list. I'm wondering where you see today the biggest failures in state capacity. And also, you know, you could see the second Trump administration as in some ways kind of going to war on some elements of state capacity. I think we've had 10,000, uh, science STEM PhDs walk out the door of the federal government during the first year of this administration. In other areas, they're building up state capacity. I think we've hired 10,000 ICE agents, uh, during this period. So how do you rate kind of the net effect of, of this president's approach to, to state, state capacity?

Well, I think I agree with your answer, or what's implied by your answer. We've lost a lot of talent in government. We're doing a much worse job keeping scientific talent and funding science, and those are big setbacks in state capacity. In terms of putting AI into government, the Trump people, I think, are better than the Biden people would have been, or the Harris people rather, had they won. And there are a lot of good noises and not too many restrictions. I don't think we've seen any gains from that yet, but I prefer Trump AI policies to what we would've gotten from the Democrats. I'm very against the ICE expansion, just general corruption in government. Unwillingness to keep Congress informed, attempts to intimidate people, push around the courts. The list is long. You probably could make it longer than I could. I'm basically against all those things, and I think they harm our state capacity. And we're in one or two key areas doing well. My hope, which I don't count on, is like the one key area of AI matters more than all the bad things we're doing. I wouldn't bet that way, but there's some chance it works out that way., and basically we're, you know, flushing some of our state capacity down the toilet.

Are there particular concrete policies or reforms that you think would most improve America's state capacity? Are there things that if you were president for a day, you would say, this is low-hanging fruit, we should just do this, it will make us more capable of essentially delivering positive outcomes?

Well, a simple thing that even the Biden people talked about would be to recognize you— it's very hard to truly reform the NIH and to support healthcare innovation. You need to start a new agency and have it be more DARPA-like. And that almost happened. People pushed back, uh, did not come to pass. But we could do that and we should do that. Reforming the National Science Foundation, which actually the Trump people are working on. I hope that goes well. I don't personally know where that is at. So there's plenty we could do in terms of healthcare policy where we spend a lot of money. I would basically always be willing to trade off a dollar of Medicare and turn it into a dollar of Social Security at pretty much all margins. That's not popular. We're older as a nation. Old people vote. Old people often don't vote ideologically, so messing with their gains is maybe not going to work. But I would love to just turn Medicare into cash benefits and let older people use that cash on markets if they want, spend it on other things if they don't want or need the healthcare. But again, is that realistic? Probably not.

You've somewhat disparagingly, and I think accurately, referred to American large defense contractors as essentially akin to state-owned enterprises in their kind of ossified, low-innovation approach to business, and also in terms of their ability to deliver capabilities on time and on a reasonable budget. To what extent do you think this is kind of an inevitability in a, in a large, complicated democracy like ours? And to what extent do you think this problem could be fixed? I'll tell you, it's one that we beat our heads against, particularly in the context of, of the Russian invasion of Ukraine and realizing not only did we not have enough kind of on the shelf We didn't have the ability to produce anything close to what was needed to help Ukraine defend itself. So how would you go about tackling this massive problem?

Well, we've done some great end runs with startups, right? Most prominently Palantir and Anduril, but many others. So that gives me hope. When I made those criticisms, those startups were less of a big thing. So I think it's striking. This is an area where our state capacity has turned out to be better. Than at least I had thought, or many people had thought. We need to do much, much more. I'm not sure how that will go. The Trump administration politicizing business so much, I think, has made that harder. It's just harder to survive in any of those spaces when your government contractor is demanding political loyalty from you. It's right now a big issue with the company Anthropic. So we might blow that. But I think also going back to the Winston Churchill quotation, we don't have enough military shells. We might need an emergency to really gear up for that. But our ability to do that is probably somewhat better than we think. And in the meantime, we've made more progress than I had been expecting. So I'm less pessimistic about that now than, say, 5 years ago.

So, Tyler, before we get to the most important topic, which is aliens, I wanted to circle back to something you said a little bit ago about supply chains. You said supply chains are very long, um, which should worry us, but also maybe we worry too much in a way, uh, because it's just a fact of life that, um, these supply chains are super complex and getting to true resilience is, is challenging. If you look at an issue like rare earths, China China used its choke point on rare earths to basically say to the United States, "We can cut those off," and in doing so have major impacts on both commercial and military supply chains. So the Trump administration, building on work we did, and the Biden administration is actually trying to create resilience, to create a rare earth supply chain that China can't choke off. But you could then go down a list of 10 or 20 or 30 or 40 other vital inputs into the economy. And, and essentially have to run the same exercise. And the question I want to put to you is, is true supply chain resilience just impossible to achieve? And if so, what should be the goal of the U.S. strategy when it comes to trying to reduce our reliance upon, or reduce our exposure to coercion by a competitor like China?

It's impossible to achieve, but the good news is there's a lot of changes we can make at the margin., and those changes are net pluses anyway. So first of all, push back on NIMBY. NIMBYism actually restricts our ability, often through environmental law, to exploit our own rare earths or rare earths from our allies. So we should be doing that anyway. There's an extra collateral benefit here. It makes our supply chains more resilient. And the other thing we can do and should do, we've been doing the opposite, is be more cooperative with our allies who have plenty of rare earths. They often have their own NIMBY problems, But they can help us out. The best thing for supply chain resiliency is to have a lot of good allies and be cooperative with them. We've been damaging that. But again, those are things you can do. They're not quite free lunches, but they bring a lot of other gains. So they're not impossible tasks, but just be realistic. Don't think you're going to have fully resilient supply chains.

In some ways, this entire conversation has been building to this moment because, as every conversation does. So look, we've now had President Obama weigh in on the topic of, of aliens in, in recent days, sort of glibly answering, yes, they exist, before then having to come out and clarify that he actually didn't have such information. He was just talking about statistical probabilities. There have been some rumors I've seen online recently that President Trump may give a speech on the topic of, uh, of aliens. Uh, be very interested in, in the content of that if it ever happens. Uh, Jake and I, believe it or not, sat through one entire, uh, edition of the president's daily brief that was entirely on this topic. I would not say we walked out of there any more enlightened, I think, than the general public. This was just before the Pentagon published a set of reports on what its pilots and, and others have seen. You've spoken about this, written about this, and over time indicated, hinted at least, that the probability as you see it that some of these unexplained phenomenon are of alien origin— alien as in extraterrestrial origin— could be as high as 10%. I'm wondering what has led you to, I guess, increase that probability over time. What indications are you seeing? And maybe just explain a bit your interest in, in this topic among many others that you write about.

Well, President Obama tweeted recently something like, there is no evidence of aliens having visited Earth, right? I completely agree with that. The problem is there are unexplained phenomena And there's also no evidence for the alternative possible explanations. So I've heard from really by now a pretty large number of insiders that there's data from multiple sources. They can't always name the multiple sources, but it could be radar, infrared, satellite, and pilot observations, that the data are consistent across each other, and that there are things that move very fast and move in ways we cannot explain. None of that is evidence for aliens, but it is weird. And if all politicians do is keep on saying there's no evidence for aliens, that is literally correct. We should not leap at alien drone probe conclusions. But the thing to do is not just to repeat that sentence, but to sit down and say what you think it might be and assign probabilities to those options. So, you know, the Nimitz incident was what, 2004? Well, that's probably not the Chinese. After the recent war in Ukraine, you know, you could downgrade the capabilities of the Russians. They would've won that war in a day if they could, you know, command these very fast, powerful UAPs. So what is it then? We can say, well, it could be many things, but okay, let's have a list. I don't have a great list, and when I realize I don't have a great list, then I have to assign some number to it being alien drone probes. It should be way below 50%, but it shouldn't be tiny. And keep in mind what's called the Fermi paradox. There's all these civilizations out there. They're very distant. We keep on discovering more and more apparently habitable planets. You know, people have been asking, well, where are they? Why aren't they here? And one possible answer is they are here. So let's say it's 10%. It depends on my mood of the day, really. There's nothing very scientific about any number. That any of us would come up with. I don't even know if it's good for society that we talk about it. It might just scare people or make them nutty. But I'm kind of selfishly curious, as Jake said at the beginning, and my selfish curiosity means I keep on bugging people about this topic, and I want to know what the probability is. I think it's not that low.

So if you put it at around 10%, Tyler Cowen, of that 10%, what proportion is a scenario in which the US government knows and is hiding it?

Very low. I strongly believe from people I've talked to, or say what Marco Rubio has said on tape, that no one knows. The real conspiracy is people are afraid to admit they don't know because they sound bad or stupid or out of control.

I would be—

I would put it way below 1% that there's some conspiracy or alien bodies or alien craft being held somewhere, Roswell, New Mexico. Uh, to me, that is very low percent, I would basically dismiss it.

So almost all of that 10% is that aliens have come with these UAPs, unidentified aerial phenomena, um, but simply have not announced themselves or have not made themselves known in a more definitive way.

That's right. And I also think it's unlikely there's little green men inside the vehicles. You know, we send out, in essence, drone probes to Mars, other places, It's probably something mechanical. At this point, we know AI works. It's probably governed by some kind of AI. And they've probably been here a long time using Bayesian reasoning, so they're probably not going to destroy us or for that matter help us. It's a big puzzle if in fact it turns out to be alien drone probes, and it's a big puzzle if it turns out to be something else.

By the way, when John referred to the president's daily brief that was devoted to this topic, it was a preview of this report that the Pentagon put out Pretty unedifying, but a funny moment. Secretary Janet Yellen, uh, actually we invited her to the PDB that day, the one time I think that she came during— so she's sitting there watching this and thinking, is this what these people do in the president's daily brief, talk about UAPs and aliens? She— I think she kind of was looking around thinking we were playing a practical joke on her.

There's this recent report that the Bank of England is now talking about this. I don't think they're true believers or anything, but just as a contingency plan, they should be talking about it. Say Trump made this announcement. Maybe it's even false, right? Trump is not predictable. It would maybe affect financial markets.

All right. Well, if, if, if your personal odds go up or the president makes some sort of announcement on this topic, we may need to consult you again before too much longer. So we won't make you promise now, but we'll be calling. Jake, should we turn to Overrated/Underrated as we shamelessly borrow, uh, one of Tyler's, uh, features for our purposes here?

Yes, turning, turning it back on you, Tyler, your, uh, feature from your podcast of Overrated/Underrated. So we're gonna put out a number of these, uh, things, and I'm ready and get your answers. So John, do you want to kick us off?

First topic, uh, the People's Liberation Army.

Overrated, right? Terrible. Uh, Chinese military historically has not performed well. In the late '70s, they couldn't really crush Vietnam. There's all these questions about the loyalty of the generals. I don't understand at all what was going on behind those purges, but there's no plausible scenario where it's good news for their efficacy. So maybe at this point with the purges, it's become properly rated, but mostly it's been overrated as a fighting force.

We have tended to overrate most competitor or adversary militaries over the last 25 years.

So cryptocurrency, it's either greatly overrated or greatly underrated depending who it is you're speaking to. Hardly anyone rates it properly. I think Bitcoin is fine. It's not as evil as people say. It can help people get their money out of tyrannical regimes or China or Russia or Iran.. It also helps a lot of illegal activity. That's bad. But as a financial asset, Bitcoin has become actually pretty boring. Price goes up, price goes down, whatever. Who cares? Stablecoins, I think, have a lot of potential. They will help us finance our government budget deficit. A lot of the world wants to do a lot more dealing with something dollar-connected or dollar-based. As you probably know, most of the trade between, say, China and Brazil is now done in terms of US-connected stablecoins. That's amazing. It's a win for US soft power. It does mean a lot of countries out there might de facto dollarize. That could hurt their state capacity. It's good for the US. It might be bad for countries such as Brazil, which over time lose some financial autonomy. But a lot of these currencies, especially in the Caribbean, Central America, they don't make sense. They're too small. The countries are too small. A lot of these places ought to dollarize. And if they can do it through stablecoins, and avoid some of the worst US banking regulations, the world is better off. So that part of crypto is mostly a bit underrated.

All right. Narrative journalism.

Well, it's become underrated. I greatly enjoy reading it. It's harder to make money with it. You know, now that classified ads have gone away and there's only a few major newspapers left. I just think I have the normie take on that, that anyone would tell you.

The impact of quantum computing.

It's interesting you ask. You know, I run this philanthropic program called Emergent Ventures, and I see what very smart teenagers are working on. And in the last 3 or so months, I've seen a noticeable uptick in smart, non-crazy young people working on quantum computing. So that makes me bullish on it. I don't understand either the full potential or the obstacles. The science is difficult. I ask people who work on it full-time, they always give me 4 or 5 different answers. But in the last few months, I've started to think it's underrated.

The public health threat posed by measles.

Again, it's one of these either overrated or underrated. You know, it's not just the Trump people. Like, Canada is doing worse than we are with measles. It's coming back as a thing that's terrible. There is no good reason for it. It gets to this point about culture. Is ultimately what is behind state capacity. We're screwing around with our culture. We need to return to, say, the culture I grew up with, where being vaccinated is this wonderful thing, almost like a gift from God. I was recently in Oaxaca, Mexico. The long lines of people waiting to be vaccinated, it warmed my heart, made me so happy. Uh, for the most part, Mexicans regard vaccination as a wonderful thing that poorer people can't get and should get. We need to move back to that mentality.

Olympic curling.

Olympic curling is a wonderful sport to watch. It's just very dramatic and it seems people from only a few countries, most of all Canada, like it. Maybe curling is not good for producing celebrities. You don't see faces in the way you do, say, in an NBA game or with the final kick in soccer, but definitely underrated.

Canadians mired in controversy. Uh, we won't, we won't digress. All right, uh, last one, somewhat tangentially related, uh, sports betting.

It's gone much worse than I would have thought, so I'm rethinking my position on this. I thought we will legalize sports betting, there'll be some collateral damage, a lot of people enjoy it, but it won't be a major crisis. Maybe we need a little more time to see how norms adjust. But it's turning out to ruin more lives than I would have expected. And that sports betting is so prominent and so easy with apps and online connections and smartphones. I don't know what we can do at this point, but I think it's a real problem. And it also runs the risk of soiling prediction markets, which maybe can be useful. But if most of what they're going to do is sports betting, they'll get a bad name at some point. They'll be, you know, put out of business or overregulated. So it's a real problem and we've yet to figure out good answers, but we need to.

Well, you're generous to have spent so much time with us. Uh, thank you. This was a, a lot of fun and a great conversation.

Thank you, Tyler. Really appreciate it.

The pleasure was mine.

Well, that was a lot of fun.

It was. There aren't many people you could pretty much throw any question under the sun and whether or not he has like deep expertise, he will come up with a very thoughtful, coherent answer to the question, uh, without any real time to think about it or research it. It's impressive.

Absolutely. And in preparing for the interview, and you and I were talking beforehand, a lot of these themes that we went through with him I think will be through lines in this podcast going forward. Um, you know, I think you put to him at one point, what does China want? What is China's intention? I think that kind of question— we'll revisit what's China's trajectory, um, you know, where is AI headed, and does the frontier matter more than diffusion? I think at some point we probably should do a deep dive on sand or maybe on a whole bunch of these boring areas, because it is absolutely striking, the physical realities of the world and where real shortages might squeeze global economies.

It's also just super helpful to get perspectives on our field from people who dip in and out of it, but don't kind of marinate in it all day, every day, and who bring perspectives from other fields analytically to the questions that we think about all the time. We have tried to do more of that, in, in our own work, but, but hearing from someone like him, uh, is actually quite helpful just because it's a totally different perspective.

One thing that surprised me was, um, you and I, in a way, as, as national security professionals, should be more focused on the first island chain and the geography of the Indo-Pacific when it comes to US-China competition. Our questions to him were all about kind of tech, econ, these areas where we think the rubber really hits the road? And a lot of his answers came back to more fundamental geopolitics. It was kind of a weird role reversal in a way to have the economist very focused on that and have the national security guys very focused on the economy and technology.

Well, as we tried to pioneer, and you get a lot of credit for this, there isn't really a neat breakdown or gap between national security issues, foreign policy issues, and these kind of You know, universal issues that we're all wrestling with, including on, on the domestic side, and, and people that can range across them like he did make for an interesting conversation.

Absolutely.

Well, that's all for today. We'll be back next week with a new episode of The Long Game.

In the meantime, send us your questions and comments at longgame@voxmedia.com and find us on Substack at staytuned.substack.com. Com. The links are in the show notes.