Capitalisn’t: Why Human Progress Is Not Inevitable

Episode Transcript

Carl Frey: I'm not of the view that every technology out there is good, but unfortunately, many times it's hard to know beforehand. It's hard to know before somebody has invested in it and it's been adopted in the marketplace.

Bethany McLean: Welcome to the show, Carl Frey, author of the new book, How Progress Ends. The core message of your book is a deeply unsettling one. It's that technological progress isn't inevitable. It can stall, it can reverse, and often it does, not because we run out of ideas, but because our institutions stop letting those ideas transform society. I wanted to start with, especially today with concerns about AI, I think many people have become disillusioned with the idea that progress actually even is a good thing. Can we start out with you making the case for progress and why it is that we should be concerned about it ending?

Carl: Well, in the end of the day, our living standards, material standards, depend on it. Even politically, an economy that isn't growing over the long run eventually becomes a zero-sum game in which the only way I can improve my standard of living is at the expense of somebody else. That doesn't strike me as a very healthy situation. The sort of raw correlation between self-reported life satisfaction and income is around 0.8. It's a fairly strong relationship.

In the interest of improving the well-being and life satisfaction of the population, technological progress is an important part of that, not just because it raises our incomes, but it gives us access to things that people couldn't even dream of a century ago.

Luigi Zingales: Now let me try on the other side. Many economists and many venture capitalists are enamored with technology at the point that they want technology at the highest speed. I don't know if you read Marc Andreessen's technology manifesto. It's basically saying any day we delay technology, we're killing people, so we should go as fast as possible. Are you subscribing to this view or not?

Carl: I'm not sure if I'm subscribing to that view, but I think the pace of technological change is quite hard to control. I think there is a concern that the more barriers you put up to slow technological change, those barriers are not unlikely, I would say, to be captured by an incumbent that keeps those barriers in place for a long time. That said, I can certainly think of technologies I wish that we hadn't developed. The cotton gin being one example of that. I'm not of the view that every technology out there is good, but unfortunately, many times it's hard to know beforehand.

I think, frankly, many people criticize OpenAI releasing ChatGPT into the world. I can understand that many people do feel that it was premature, but I don't think we would have much of a debate over AI regulation to begin with if that hadn't been the case. I don't believe that we will be able to specify a set of rules and regulations ex ante that will deal with any potential harm that AI could cause down the line. I think, like technological progress, institutional change is also a matter of trial and error.

Bethany: Is your argument overall, despite the mistakes in the past that have come from perhaps moving forward too quickly and embracing new technologies too quickly, that the risk of not moving forward is higher than the risk of simply of moving forward?

Carl: I do think that the risk of not moving forward right now is pretty high. One of the arguments I make in the book is that it's not just Europe and Britain that has a growth problem at the moment. There are many concerns that artificial intelligence will threaten democracy through misinformation, some worry about existential threats. I also think it's a real risk that the technology, frankly, doesn't live up to the expectation. Maybe in part because institutions don't adjust. To me, it remains a puzzle that if we go back to the '90s and you take the web and the personal computer. These are astoundingly powerful technologies.

What do we get out of it? A decade streak of productivity growth mostly confined to the United States, a decline in research productivity by most estimates, a decline in breakthrough innovation, as far as I can tell. I feel like we have to ask ourselves why that is. I may not been able to fully answer that question in the book, and so I plead guilty to that, but at least I hope it does inform the debate around that a little bit.

Luigi: No, it certainly does. At the heart of your book, there is a core tension between decentralization that forces exploration and innovation, and centralized bureaucracy that accelerate exploitation and the scaling of technology. Can you give us a quick explanation of this tension and why it's so important that it's at the heart of your book?

Carl: In the early stages of the technology lifecycle, you don't know whether what you do is going to be a success or not. If you're an investor, you don't know what's going to catch on and what's not. You want to preferably invest in different technological trajectories. The Soviet Union was the most centralized economy we've seen so far. If you were an aircraft engineer in the Soviet Union and you wanted funding for your idea, you could go to the Red Army and ask for funding. If they declined, maybe you had two or three other options. If they declined, your idea would die with you.

That's quite different to the US system where Bessemer Venture famously declined to invest in Google back in 1999, and they probably regret it today. It also illustrates that Google wasn't a safe bet at the time. AltaVista and Yahoo were dominating search, and so you need to invest to know if something will catch on. You can invest in a broader range of projects in a more decentralized economy. At some point, economies of scale do matter. Without that, Ford would not have been able to produce the T-Ford at a sufficiently low cost for it to become the people's vehicle. Both of these processes matter to technological development, but they are supported by quite different institutions, as I argue in the book.

Bethany: One of the fascinating things about your book is that you don't actually argue that either system is better. You argue that right now, both China and the US, and it's often posited as one wins and the other loses, but you argue that both actually might be headed for a period of stagnation. Can you explain that for our listeners?

Carl: Yes. If you look around the world today, productivity growth has been rather disappointing. After a brief uptake in the US between roughly '95 and 2004, productivity growth has been disappointing. It's been disappointing in Europe, and it's even been disappointing in China, not least since the Great Recession, whereby some estimates have even turned into negative territory. I think what-- US and China obviously has many differences, both in economic and political systems, but they do, I think, share a common feature in that more power is being centralized.

In China, I think after the opening in the late 1970s, it's fair to say that much of the growth was facilitated by the fact that provincial governors had a large degree of autonomy, and so they could experiment with different institutions. The party retained a monopoly over promotion within the party system, and it set growth targets. Essentially, governors would compete for positions inside the party on the basis of meeting growth targets. Since the 2010s, more power has been centralized in Beijing in particular.

Industrial policy, which used to be more local, has become increasingly centralized. Objectives are no longer just focused on economic growth, but other political priorities, such as technological self-sufficiency and national security. Because private firms are often not very keen on pursuing objectives that are not economic, China is once more dependent on state-owned enterprises, which also have more political power by having seats inside the party.

In the United States, as you know very well on this podcast, market concentration has been going up. Spending on lobbying has been going up since the 1990s. We see a range of anti-competitive practices, whether it's patent examiners at the USPTO granting low-quality patents and then moving into work for technology incumbents that they've granted these patents to. You see some outright killer acquisitions, whereby incumbents buy up promising firms just to shut them down, and other practices, too.

There's been a growing concentration across most industries, and that has gone hand in hand with less dynamism, less entry, even in the technology sector. I think that's part of the reason why growth has slowed.

Luigi: Certainly, I'm not going to quarrel with that. I'm curious because your book emphasize a lot innovation diffusion, and you claim that one of the reasons for the success of Silicon Valley versus Route 128 in Massachusetts was the fact that in California, the non-compete clauses are not enforced. Maybe I miss it because it's a big book, but I don't see you talk a lot about the increased enforcement of intellectual property rights in America and the war stage.

In a sense, TRIPS, I don't remember what the acronym is for, but it's Trade-Related Intellectual Property Rights Enforcement, was started in the mid-1980s and made it much more difficult, actually, for follow-up countries, but also for follow-up firms to use new ideas and new technologies coming along. Do you think that this played a role?

Carl: I do think that the pharmaceutical industry in particular has lobbied quite well in making sure that prices are kept high and generic alternatives are kept to minimum, and that has had an effect that it slowed down the diffusion of pharmaceutical products. I'm less sure when it comes to digital and ICT products. I think, in my mind at least, the problem there is more that we have these very complex technologies with a lot of intersecting patents where you really only know exactly what the patent covers after you ended up in court. Then you have large firms with large patent portfolios using intellectual property rights as barriers to entry. They use it as bargaining chip in cross-licensing negotiations.

James Bessen has estimated that outside pharma and chemicals, litigation costs are actually higher than the value firms derive from patents in production. It used to be one product, one patent, right? Now it's more one product, 1,000 patents. I think that complexity has made the patent system very litigious and has meant that barriers to entry have increased. I think that's been the bigger factor.

Bethany: When you think about how advances in the world today are being made through sheer scale and the development of manufacturing prowess, whether it's in semiconductors or in batteries, is there any evidence that the arc of the world is shifting away from exploration and toward exploitation? Does DeepSeek play into that at all?

Carl: I think most industries naturally shift away from exploration to exploitation as they mature. In the industries you mentioned, I think that's certainly true. I'm less sure about AI in particular though. I don't think we know yet whether the future of AI is large language models, whether it's small language models, whether it's some fusion between deep neural networks and symbolic AI, or if it's something else. We have this paradigm of scaling in AI, throwing more compute and more data at the problem and getting better performance as a result of that.

I think we seem to be running into some diminishing return there. I think we're probably back at the paradigm where we need more research, more new ideas. I'm not so sure about AI, but I think in EVs and solar and some of these other technology fields where China is doing quite well, I think it's now more about exploitation and scaling than experimentation.

Bethany: Is it possible to identify a chief villain in all of this? The villain that blocks progress? If so, is it corporate power that becomes entrenched, is it the state that doesn't want to be challenged, or is the broader issue simply human nature? That if we can, we'll stop and preserve the world as it was rather than the world as it might be?

Carl: It differs a bit from case to case. Often, today's revolutionary becomes tomorrow's incumbent. I think there's certainly something to that. I think that's the reason why we see quite similar patterns across countries, because it is, to some degree, human nature. If you have something to defend, if you have a product in the marketplace, for example, and you fear that those revenues will be undermined by a new product, you're less likely to launch it. Kodak is a good example of that. It's probably also a reason that Google was hesitant early on to launch the kind of generative AI that OpenAI launched, for example.

Luigi: This is an interesting tension because in the book, you emphasize the role of the state in actually enforcing antitrust and enforcing diffusion. You refer to, of course, the famous consent decree of AT&T that spread the transistor. You refer to the fact that the computer revolution took place at the time where IBM stopped integrating the software with hardware as a result of antitrust. Also, the Microsoft trial made the internet easier.

Now, the challenge is that for the state to perform this function, the state should be independent of corporate power, which makes it very difficult, because today, in the United States, I don't see the state being interested in playing this role. Even under a Democratic administration where Lina Khan and Jonathan Kanter were working very actively, Congress was completely on the sideline and nothing was produced. A lot of the gains achieved, in my view, were temporary. How do you solve this tension?

Carl: I don't think there is an easy answer, unfortunately, and that's why the book is called How Progress Ends. I think if you go back historically, I think there are two ways, broadly speaking, out of it. One is geopolitical competition. In Europe, before the Industrial Revolution from the 12th century onwards, production in Europe was basically controlled by craft guilds. I think one of the reasons that the first Industrial Revolution happens first in Britain is that their political clout is undercut in Britain first, partly because Parliament begins to underwrite acts through the turnpikes that spurs market integration that entails overriding the rights of select landowners as well.

That competition means that guilds which control the craft in their city suddenly face competition from the outside. Then the state begins to clamp down on craft guilds more directly as well because it realizes that British wages are quite high, and so for Britain to compete globally in trade, it needs to mechanize production. Similarly, all across Europe, you see that Prussia's defeat to Napoleon at Jena in 1806 leads to guilds being abolished. Russia begins to liberalize a little bit after devastating losses in the Crimean War, and so on. The state can perform that function when faced with geopolitical pressure.

The other mechanism is by having a relatively mobilized public. I think we saw that in the United States during the Gilded Age where grassroots movements pushed for changes through the Pendleton Act, which led to meritocratic civil service, which then went on to institute new laws and regulations, such as the Sherman Antitrust Act. When the public is more mobilized on issues on rare occasions, they can serve as, I think, a healthy countervailing force to entrenched economic corporate power.

Luigi: Given this fragility of progress, if you were king for a day, what kind of institutional reform you will introduce in the United States or in China or in Europe to make sure that progress does not end?

Carl: If I was asked to be king for a day, I would politely decline. If I were able to give some advice, I think market integration in Europe is quite important, particularly when it comes to services. What's striking to me in Europe is that if you look at the post-war period, Europe managed to catch up quite effectively in manufacturing industries. In digital, Europe has not even managed to catch up yet. Push the technological frontier. I think part of that is that the single market for services is just much less harmonized. That makes it hard for firms to scale.

I think in China, get rid of the party system, is that good advice? I'm not sure if that sort of advice will have much of an effect in the United States. I think it's hard to know where to start these days. Is it immigration policies? Is it funding being cut to leading research institutions? Is it tariffs and the range of exemptions that mean that only incumbents that have political clout are able to navigate them and get those kind of exemptions? I wouldn't proclaim that there is one sort of a unifying solution to all those problems.

Bethany: Continuing on that thread of the United States at the risk of being a US-focused person, how do you see what's happening now? It seems to me, and maybe I disagree with this formulation, that we're in this period of change in the relationship between corporate power and government power. With tariffs, with the government taking stakes in companies, with the government controlling more of what corporations do. Is there any historical analogy for where the United States is right now, or is this uncharted territory?

Carl: I think going back to the 2000s, many people predict that China would become more like the United States. Now it looks like the United States is becoming more like China. In terms of historical parallels, I think the closest I can think of would be the Gilded Age, but with the difference that during that period, you had this grassroots movement, you had eventually push for reform, you had the emergence of a meritocratic civil service. Now events are going in the opposite direction, right? So, the civil service is being completely handicapped in its ability to carry out its tasks by the Trump administration.

I think what is happening in the United States is that you're seeing more power being concentrated in the hands of the executive, but then without increasingly the civil service to actually be able to execute and implement on those policies. The key difference between China and Russia today is that in Russia, most of the political power is in the hands of one person, but he doesn't have the state capacity to implement many of his wishes. In China, meritocratic civil service means that the state is much more able to implement its policies. I think United States is probably starting to look more like Russia than China in that regard.

Luigi: That's very optimistic.

[laughter]

Luigi: Thinking about China, one of your argument, which I think historically has certainly been true, is that if you try to restrict circulation of ideas because you want to maintain a control of a party or an ideology over a country, you make it also more difficult to have the circulation of ideas that are necessary for progress to develop. During the period of the Soviet Union, it was difficult in Russia to have foreign magazines, whether these were political magazine or even scientific reviews. Clearly, this was an obstacle.

However, if we look at China today and how they master the censorship in the internet, they are able to let 99% of the ideas and things circulate, retaining and filtering the 1% that they don't want to circulate, which is certainly not the scientific component. Is it possible, and I see this as a danger, not as a wish, but is it possible that technology now has made possible the combination of a central ideological censoring power and a free intellectual environment for scientific innovation?

Carl: I think there's something to that. The disadvantage, in the academic sphere, of having a highly centralized political system is that you're less likely to attract talent. Ask your college how many of them would like to move to China and would like to be a researcher in China. That is not to say that you wouldn't have access to the same flow of information, and you would have access to the same publications in China, but I think if you look at Chinese scientific output, a lot of it comes from Chinese researchers who studied abroad, come back. A lot of it comes from international collaborations.

When you start to clamp down on those knowledge pipelines, I think it's quite likely that you will have a problem down the line. I don't think the key bottleneck is the access to information. It's more about the clamp down on international collaboration and the ability to attract scientific talent that are the key bottlenecks.

Bethany: Right now we're in a time, at least in the US, where we're tossing off the shackles of regulation in the AI race, because everything has become about beating China. That the US has to be able to beat China, therefore, we can't regulate because that will get in the way. With the historical parallels you've seen in your book, is it that clear cut that a huge government push to win and to beat China and toss off these shackles of regulation will lead to innovation, or is there a possibility that that does the opposite?

Carl: It's certainly a possibility that it does the opposite. It's quite striking to see at the moment how China is openly embracing open-source AI, how those models are gaining market share all around the world, and how that might actually undercut proprietary models, which may not be able to turn a profit because of that competition. It's not just a matter of national regulation. It seems that American tech industry has made one big bet, at the moment, on proprietary large language models. China, out of necessity and being behind, opted for open-weights. So far, that seems to have been a good bet.

As I mentioned earlier, we don't know really what the future of AI looks like. It may well be that we need new ideas, and the question is then which system is best placed to deliver that? Right now, it's far from clear to me. Yann LeCun recently left Meta. He's now setting up his startup in France. I don't think America can take for granted that it will lead in artificial intelligence just because it's saying that, "We're not going to regulate artificial intelligence." At the same time, I don't think that the EU AI Act and those European approaches to regulation have no bearing on innovation.

I think by most estimates, the GDPR has been harmful to smaller firms in particular. Larger firms have found it easier to offset compliance costs by capturing a larger share of the market. AI regulation could do the same.

Luigi: Your book was largely written before the election of President Trump, and you were pessimistic about the future of the United States even before the election of Donald Trump. To what extent, not just his election, but his behavior in the first year in office made things worse, and is there any sort of positive sign that we say maybe is not as bad as we think?

Carl: To the extent that I pay attention to current events, I struggle to find any positive signs at the moment. I have to tell you. We mentioned earlier growing markets concentration across industries. You could say, at least previously, that, well, you have still competition from other players in other countries, but now with trade barriers on the rise, that is no longer true either. In addition to this, you have this system of exemptions, which is easier for politically connected firms to navigate. That is probably going to have some impact on the kind of managers that are being hired as well.

We now have a lot of tech CEOs in Silicon Valley. If the future of these businesses is politics, well, you're going to get a different set of managers running those companies, too. It's going to be managers that are skilled at politics. It's probably going to have some spillovers there. I've been quite astonished by the degree that scientific funding is being cut. I haven't been surprised on immigration, but it's not working in favor of American innovation. I've been quite struck at how the rhetoric on cutting the administrative states. The actions there is actually diminishing American state capacity. The capacity to regulate, the capacity to build stuff.

If you wanted to, for example, get rid of the Internal Revenue Service, you would introduce a flat tax and you could get rid of the Internal Revenue Service. If you cut the Internal Revenue Service, then without reforming the tax code, you're just cutting the ability to collect taxes. If you want to get rid of the administrative apparatus, you start by cutting regulation, but that is not what's happening in the first instance either. I'm puzzled on many fronts. If I would say something optimistic, is that the United States have seen challenges before. It's a resilient economy. It's been quite institutionally resilient through multiple crises, but I don't think that it's a lot of cause for celebration, frankly.

Bethany: Well, I'm not sure this podcast qualifies as uplifting, but it was incredibly interesting, so thank you so much for your time and for coming on.

Luigi: Yes, indeed. Thank you very much, Carl.

Carl: Thank you for having me. I really appreciate it. Particularly nice to be on a podcast you listen to yourself.

Luigi: [laughs] I think that the book is definitely very rich and very broad. Some of the facts are reminiscent of his previous book that was more focused on the fact of how basically incumbent elites don't want innovation. One of the stories that I think he repeats in both books is that one Roman emperor put to death a inventor that was trying to save labor because he said, "If I implement this technology, I'm going to have a lot of people unemployed." I think that one of the important point we need to remember is that we give for granted today, a culture of growth, to use Mokyr-- this year's Nobel Prize words.

The culture of growth is actually a product of Western civilization starting in the 17th century. We discussed this a bit with Beckert, because at the beginning, if the world is zero-sum, there is not much of culture of growth because what I make more, I subtract from you. I think that at some point, the Western nation first started to steal so much from the rest of the world, that it grew very much. After that, the scientific revolution really increased the size of the pie and made all of a sudden possible to be much better off. We give for granted that capitalism and scientific evolution and progress go hand in hand, but this is not obvious.

We have seen countries slow down. We have seen my own country of origin not growing productivity for 25 years in a row. I think that it is possible to stagnate. It's not that easy to restart progress when you are in that equilibrium.

Bethany: Yes. He was obviously a defender of the idea of progress, but what do you make of the argument that we should be putting conventional notions of progress aside? Because humanity has a lot, at least particularly people at the top end have a lot. Can our world tolerate continued growth if it means continued seizure of the world's natural resources? Do we all have to learn to live? Do we all have to stop prioritizing growth in the way that we have in the past? Or maybe a better question is, do you equate growth and progress? Can you have progress without growth? Can you have one without the other?

Luigi: I would say you can have both without consuming more resources. You can consume less resources. I think progress, it could be to produce the same quantity with much less input. Also, progress is the creation of intangibles that are not necessarily finite, they don't consume resources. I think that think about new philosophy or new religions or new poetry. I think that they don't necessarily consume resources, but they do benefit humankind.

Bethany: In the traditional economist's view, don't we equate progress and growth? In other words, is there a society that economists believe has made progress that has not shown growth? Do we need new measurements?

Luigi: I think that new measurement will definitely help. Even with traditional measurement, we value a lot of intangibles. If you go to the theater in New York, you pay a lot of money to go to the theater. That is GDP, but has not produced any tangible thing. It's not a machine, but it is growth, in some sense.

Bethany: Unless you show up by yourself and get a single seat at the last minute, and then you don't pay as much. That's a new trick I've just discovered. It's very exciting. Anyway, [laughs] put that aside. One thing I was thinking about, actually, is that after our conversation we had with Dani Rodrik, who pointed out the immense progress that China made on green energy due to the structure of its society. Do you think-- does that challenge Carl Frey's view of the world? In that can you argue-- in other words, does whole-scale exploitation become exploration at some point, given that so many of the advances in green energy are being made by China?

I don't know. I'm struggling to reconcile those two views. One, Rodrik's view that China is the one that has made most of the breakthroughs on the green energy front because, basically, of this massive state-driven move, with Carl Frey's view that what China excels at is exploitation rather than exploration.

Luigi: I think we should bring a China expert to the show because it's very difficult to really understand what's going on in China. Actually, most people have their ideological view, either one way or another. Either they say that China is only able to copy, it's not able to do anything, et cetera. I think that that view is false. For example, now the self-riding car, I'm told, I'm not an expert, I'm told that the Chinese model is much more efficient than the one that Elon Musk invented. In part of it is because they started with a different technology, but they had enormous return to scale in that technology, so now it's much cheaper.

Bethany: Yes. I think there are examples like that that suggest that Rodrik might be-- I'm sorry, that Carl Frey might be underestimating China.

Luigi: I think so. I think that he's underestimating China. Now, I think the question that I was trying to get with him, which to me, the fundamental question, is historically, there is no doubt that scientific innovation and intellectual freedom went hand in hand. Yes, we had a little spurt of technology innovation, the Soviet Union at a particular moment, mostly military, but by and large, was very difficult. Part of it is because if you censor, like in the old Soviet Union, was censoring all material written in English, it was very difficult to have a bureaucracy able to sort out whether the material was a political pamphlet or a scientific journal. Okay?

Today is a piece of cake. With AI, you can automatically sort so that everything that is-- if you want more technological oriented, can filter through, and everything that is vaguely politically oriented can be blocked. In China, they even learn how to block satire. They know that Xi Jinping is equated to Winnie the Pooh, so anything that has anything to do with Winnie the Pooh is censored in China, no question asked. This ability to sort out, unfortunately, because I would like to believe that the only way to achieve progress is to be free, might separate the two. If they separate the two, then many of the things he's saying might not apply.

Bethany: Yes. I think that might get at the fundamental issue we both have. That his book might be a wonderful history of the world as it was, but may not be a predictor of the world as it is going forward.

Luigi: Yes, I think you summarize it perfectly.

Bethany: I love Frey's framework, but I think that we don't even understand. People don't even understand how progress comes about, or how ideas come about. It's amorphous. Let me just give you an example. In the run-up to the pandemic, everyone said it was utterly critical for people to have spaces where accidental congregation could take place because human interaction was the key to progress and to ideas becoming reality. When the pandemic hit, all the companies that had been previously talking about how important their workplaces were suddenly said, "It doesn't matter. We can continue to do all of the work we ever did without people getting together."

Now we're back to that, we need to get people together thing again. I think what that suggests to me is that no one really knows. Some of this is a matter of belief. I like the idea that you need people to get together in order to create progress, that you need those happenstance moments, because I like humanity and I like the idea of people interacting with other people, I don't like the idea of all of us being separate little atoms, but I admit that I'm biased. I bring my bias to that argument. I'm not sure anybody actually knows. What do you think, Luigi?

Luigi: Actually, Carl Frey has this beautiful example of innovation going down when you shut down the bars during the prohibitionist era, because people don't congregate and share information. I think that there is definitely that element of congregation. There is a reason why, in every university, they tend to subsidize coffee machines or places like this, because then people congregate near the coffee machine, and then you get to know your colleagues, and you interact, and sometimes even good research spurs.

Carl Frey, to his credit, has this other beautiful example of Karikó, the woman who won the Nobel Prize, that met the other guy she was doing research with at a photocopy machine. In line for the photocopy machine. I think that these congregations are important in breaking the silos in which we live.

Bethany: It's an absolutely beautiful idea, because when you're out at happy hour having fun with your friends, you are actually enabling creativity and progress. There you go.

Luigi: That's your point. That you're always working, right? Even at the happy hour, you're working. [laughs]

Bethany: Exactly. You're being productive, even when you think you're just having fun. There.