Peter Thiel on the Global Economy, the State of Our Technology, and Artificial Intelligence with Bill Kristol

Welcome back to *Conversations*. I'm Bill Kristol, and I'm very pleased to have with me again Peter Thiel—innovator, entrepreneur, businessman, thinker on many topics, imaginative and unconventional thinker on politics, economics, society. So thank you, Peter, for taking the time. Thanks for having me, Bill. Good to have you. We were just discussing before this that you're back pretty recently from China. This is now May of 2016. And you had interesting things to say about that. So I thought we'd just begin—such a huge issue in terms of international economy and the future of the world politically and economically. What struck you in China?

Well, it's a combination of incredible determination, you know, incredible hard work on the part of people. I taught a mini-course at Tsinghua University, one of the elite universities in China. So there was incredible interest— Your book was a huge seller in China, right? It did very well throughout the world, the *Zero to One* book, but probably sold more copies in China than the rest of the world combined. Is that right? —intense interest in technology, how to do new things. And then at the same time, there's also this very palpable sense that the model that's worked so well for the last 30, 35, 40 years of globalization, of growth through trade, extensive growth, copying things, is reaching some kind of a natural limit.

And so I think there is this very challenging transition China is going through in terms of finding a new growth model. It's not clear everybody can become an entrepreneur or anything like that, even though you have various Communist Party slogans of "mass entrepreneurship" and things that sound slightly oxymoronic in that sort of vein.

So what did you find among especially the students? Confidence in the future? Worries about the future? Desire to change the system? Not?

Well, I think there always are things you can talk about and you cannot talk about in that sort of a place. But there is an incredible drive to succeed. And there's always a question: how does one interpret interest in startups and entrepreneurship? And on the surface, it always seems to be very optimistic, very ambitious. And then I think just beneath the surface, it often also goes to a lot of anxieties around how tracked careers, how working in a large state-owned enterprise or in sort of a government functionary isn't quite what it used to be or may not work as well in the next 20 or 30 years as it has in the last 20 or 30. So I think—I think we often have had this with these booms or bubbles where, on the surface, it's extreme optimism.

In the '90s, we had extreme optimism about the new economy in the US. And then just beneath the surface, it was this sense that the old economy was no longer working and there was no future in the old economy. And so I've sort of been wondering whether something like that is going on in China today as well. But this is, again, it's always hard to know. It's a vast country, 1.3 billion or more people. And so it's hard to really generalize about something like that from just talking to a few students in an elite university.

Do the students think—and I guess what's your own judgment of this, too—I mean, are they fascinated by *Zero to One* because they think, "Well, we could do that in China, too"? Or is it, "Gee, they can do that in the US, but we can't do it in China"?

Well, I think it's interesting precisely because that's sort of the cusp of the debate. And so the two kinds of critiques my Zero to One book got would be, on the one hand, that of course China can innovate and it's wrong for me to suggest otherwise, which I didn't really do, but there's sort of a sense that China gets linked with globalization. And so one critique is, yeah, we can innovate just as much as the West. And then the other critique of it is, well, zero-to-one businesses are good for America and other countries. We don't need zero-to-one business. We can just copy things that work. And so you have these two diametrically opposed critiques: we don't need to do this, or we're already doing it.

And that leads me to think the truth is somewhere right in between, that there is a sense that they haven't needed to do it. They haven't needed to innovate for a long time. And this, of course, is the good part of globalization. You can copy things that work. If you're a very poor, very underdeveloped country, there's a lot of room for copying. And then at some point...

Huge increase in living standards.

Huge increase in living standards. And at some point, you run out of things to copy. And there's a sense that this is what happened. This was the arc of Japan, sort of the Asian exemplar in some ways, where starting with the Meiji Restoration in the 1870s, and then all the way through the 1970s, 1980s, was incredibly dynamic. There were so many ways it wasn't a strictly capitalist model, but it somehow just worked. And then you sort of hit a wall in the '80s where they more or less caught up. There was nothing left to copy. And then it somehow failed to move beyond that. And then you get the question, when does China hit a wall like Japan? It's a very similar model: export-oriented, current-account surpluses.

When do they hit a wall? If you look at it on per capita GDP, you would say it's roughly where Japan was in 1960. So you have maybe 20 years left to go on the copying. So if you look at per capita GDP, you'd say, you know, China still, it's one-seventh, one-eighth of the U.S. You know, Japan got up to about two-thirds or three-quarters before they hit the wall. So, you know, a long way to go. And then if you look at it, then if you look at it maybe from trade flows where, in many goods, China's manufacturing half the stuff that gets made in the world in that category, and you sort of wonder how much growth is there really left doing this. And you end up with this question whether maybe this model worked really well for Japan when it was the first country doing it, and it works much less well for China at a much bigger scale.

And so maybe they're hitting this wall that Japan hit in the late '80s today, in 2016.

And one of those interesting things, I think this is in the book, and certainly in other things you've written, and in our previous conversation, I think we discussed this some, is people tend to muddy together, link globalization and innovation. And I think you make the argument that they're quite different, actually.

Yes. Well, I always... I try to stress the difference where I always draw globalization on an x-axis. It's copying things that work. It's going from one to n. It's horizontal extensive growth. And then I always draw technology or innovation on a y-axis, going from zero to one, intensive vertical progress doing new things. It's doing new things versus copying things that work. And there certainly is a sense that in a successful 21st century, we want to have both more globalization and more technology. But then there are probably some trade-offs in terms of where the stress gets placed. And if you look at the history of the last 200 years, we've had eras of technology and of globalization and of one or the other.

The 19th century was an era of both tremendous globalization and tremendous technological progress from 1815, ending with the start of World War I in 1914. When globalization goes very much in reverse, technology continues at a breakneck pace. And then I would say, I would date 1971, when Kissinger goes to China, as the year where globalization begins again in earnest. And we have had 40-plus years now of breakneck globalization, but what I've argued is relatively more limited progress in technology, mostly centered on this narrow cone of progress around computers, software, Internet, and not so much in many other areas of technology. The 20th century had a period of technology with less globalization and then a period, a more recent period of globalization, with more limited technology.

It's reflected in some ways in the different ways we talk about our world. So in 1965, when you had technology but no globalization, you would have described the world geopolitically as the First World and the Third World. The First World was that part that was technologically progressing. The Third World was that part that was sort of permanently screwed up. Today, the dichotomy would be between the developed and developing worlds, which is a sort of convergence, homogenization, globalization theory of history. The developing world is converging with the developed world and becoming more and more alike. And so this is a pro-globalization dichotomy, but at the same time, it's also an anti-technological dichotomy, because when we say that we, living in the United States or Western Europe or Japan, are living in the so-called developed world, we are saying implicitly that we're living in that part of the world where nothing new is going to happen, where things are done, finished, and exhausted.

And I feel that that's always a little bit too pessimistic. So I think in theory, we should have both globalization and technology. In practice, we certainly have choices that we make. On an individual level, do we work in ways that are more globalization-oriented, more technology-oriented? It's possible that over the last 40 years, there were so many gains from globalization that it was natural for talented people in our societies to work in industries that were linked to globalization. And perhaps those gains are a little bit more hard to come by now, and maybe it makes sense to rebalance towards technology. If you think of it geographically within the U.S., you could say that New York City is the city that was linked to finance.

Finance is the industry that's linked to globalization simply because it's about the global movement of capital. It's something that's very easy to move around the world. And so somehow, the way the economy and the future was centered on New York City from, say, 1982 to 2007, for that quarter-century, was the period when people really believed in globalization. And the sort of strange shift from New York City to Silicon Valley, I see is the shift where maybe we can get more of technology. It's harder to make progress on globalization. And I still spend some time both in New York City and mostly in Silicon Valley. And it's always striking, the sort of, the extreme contrast—you know, optimism in Silicon Valley and the sort of very deep pessimism that I feel always permeates New York, where there's this model that's not quite working anymore.

And so that's what you see in New York City. And then I think remarkably, I think that's what people also believe in China at this point, which, again, was, as a country, geared to this to a degree greater than perhaps anywhere else in the world.

And how much of a, you know, to come back to New York City, that's so interesting, but... And I guess there's a real trade-off in terms of the regime and the political arrangements between the two. I mean, an arrangement that might work for just massive increases of globalization and prosperity may not work to encourage innovation. And if you're hitting the wall, so to speak, if the gains aren't coming as easily anymore from globalization, you could have political crises and other things.

Well, certainly we have... There's a number of things to say here. Certainly, I think growth can happen through some combination of globalization and technology. We'll have problems politically if there's not enough growth. I think there's some level in which the combination of globalization and technology hasn't been delivering as many goods as it...

But in China, I'm thinking, I mean, how much of a crisis do they face? I mean, before we get back to us, is it hard to tell?

I think with respect to China, I think that globalization has been such an enormous driver. It seems to me hard to reorient it towards technology simply. That there's so many, you know, you have the Foxconn factory where people are manufacturing the iPhones, which I think more is globalization than technology. 1.3 million people employed there. And, you know, maybe that keeps going, maybe it gets redirected in some other way. But the transition, I think, is very far from trivial. It was a very big problem even in Japan, which, you know, arguably, was globalization-orientated, tried to shift towards something more innovative and never quite cracked that open.

So China has big challenges ahead of it.

It's, you know, the bullish case is always that if you just look at per capita incomes, it would seem like there's still a tremendous room left for convergence. But my own sense for it is that it's possible that we're later in the globalization game than people think. It was part of the 2012 People's Congress, which was a statement. I believe it was Liu He, who's sort of the chief economic advisor to President Xi, included in the 2012 statement that the globalization tailwind that had helped China for 30-plus years was abating, and that you'd have to really think about how to reorient much of the economy, maybe towards internal consumption, maybe towards innovation, but away from a lot of the things that are working, and then a lot of the power structures, of course, linked to things that are working.

Subsidies to the state-owned enterprises that export goods and that employ a lot of people. And so there's a question how easy it is to reorient that.

And I suppose that is where we presumably have some advantage as a free country. We're a free society where you can have a Silicon Valley in a way that I presume is harder in a place like China.

Yes, it's always, you know, I always worry that we're not as free a country as would be desirable and that there are, certainly with respect to innovation, there are many areas where it's effectively been outlawed—not in the area of computers, which are still, I think, quite unregulated on the whole and where there has been a lot of progress. But yeah, I think in the U.S., there definitely are large elements of our economy that have been linked to this globalization story as well. And if I had to do sort of a long/short version, I would try to be skeptical of the parts that are linked to globalization. If I was talking to a young person graduating from college, I would discourage them from working at a big money-center bank in New York or for McKinsey or sort of the international global consulting firm, which is, again, a classic globalization career—Clinton Global Initiative.

That sounds very dated at this point. That's so 2005, World Economic Forum. All these things have sort of a dated kind of a feel. This is the way the future used to be. It's not the way the future is.

Was '07, '08 the kind of the, well, people look back and say that was the break point where we did have... In a way, it was literally a globalization of credit, which then turned out to be not as soundly based as people hoped, right?

Yeah, that's certainly the interpretation I would give of it, which was that it was a peak on globalization, on the belief in the emerging markets. You have—one of the acronyms in the 2000s was the BRICS countries: Brazil, Russia, India, China. And then it turned out that a lot of it involved the investment of capital in these places that was misdirected, that there was a lot of corruption under the veneer of globalization, and maybe it wasn't going to work as simply as people thought. And I would argue that since 2008, it's not like the tide has really gone out. It's still been held together. You know, the global trade flows were growing at two to three times the rate of GDP growth for 30 years before 2008.

They've stayed roughly in line since then. So the tide hasn't really come out, but it's stopped advancing. But I think it's been held together much more by political will and much less by widespread belief. One miniature version of this global project has been the European Union project, where you could say in 2007–2008 people really believed it, and so a German saver would naturally buy Greek government bonds. You got a higher interest rate in Greece, and of course it would eventually converge to a German rate because everything was going to converge. Post-2008, there's a lot more skepticism. There would be no natural way that money would flow from Northern to Southern Europe today. And so, it's only—to sort of keep the European system balanced has required massive state interventions to flow the money to buy the bonds in Southern Europe because the people themselves no longer believe it.

And so, I think there's sort of all these versions of it where it's been held together by will. And I wonder whether the tide is really going to start going out in the years ahead.

Yeah, well, it does have the feeling, just looking at Europe and China in a different way, the political elites are—and I can understand why they're doing this, and they're not necessarily wrong to do it in the case of Europe, I suppose, so holding the thing together—but now it's fighting against the... It's holding something together that's already, from which the benefits have already been derived, and now you're beginning to see the downside, as opposed to, 'Let's all ride up,' you know, as you say, you know, the '90s kind of, 'We're all just riding onto a bright, happy future together,' you know.

Well, I think the elites are right in that the alternatives to globalization always seem really bad. Right. So, you know, in Europe it was, you know, two world wars. And so, even if we have a sclerotic, corrupt bureaucracy in Brussels, that seems better than what happened between 1914 and 1945 in Europe. And so, I think there is always this negative worry that the alternatives to globalization are all bad. But I think, even though I think that's correct, that shouldn't blind us to the possibility that a lot of what goes under globalization may also have been bad. And so, if it is just a way for, you know, dictators in emerging market countries to abscond with money into Swiss bank accounts, or if it's a way for terrorists to sort of freely travel throughout the world and use our global communications and global transportation networks...

It's the globalization of violence through terror, things like that. There are all sorts of forms of it that can be bad. So, even though globalization ending would be an unmitigated disaster, we have to, I think—we should be much more, we should try to draw much tighter distinctions between what forms of globalization have been good, which ones have been bad, and not just have this sort of, this Panglossian World Economic Forum, Clinton Global Foundation, that if it has the word 'global,' it must simply be good.

That's interesting. Also, I guess one could take the position that it might have been good, mostly, for X number of years or decades, and that's nothing to be ashamed of. It was a reasonable standard of living. Hundreds of millions, I suppose billions of people, increased in China and India, and parts of Central and Eastern Europe that had historically been hugely unstable became reasonably stable and better off. So, that something was good for four or five decades doesn't mean it's good for that. It doesn't mean that you can simply keep applying that formula for the next four or five decades. That seems to be very much the view of the European Union, in a weird way, I suppose, the Chinese Communist Party, too, of, you know, "We're just going to keep doing what we were doing."

But the world doesn't quite work that way, presumably, you know.

And it may be. It may even be somewhat good. So, I tend to think if we had more trade agreements, this would be generally a good thing. But the incremental value of the next trade agreement may not be as high as the ones we already have. And so, if we ask where we're going to really get the drivers to take our civilization to the next level in the decades ahead, maybe it is not the same formula over and over again.

So, let's go to that, because you've been an articulate proponent and very much a contrarian proponent. I think people are beginning to come around a bit to the argument of the innovation slowdown, and that—not that it necessarily, that we've merged in our own mind, let's say, globalization and innovation, and everything is better than it was, and then the computers became a stand-in for all kinds of innovation. But in fact, in many important areas, innovation has been less than we would have expected it to have been and is less than it could be. Is that right?

Certainly, if we went back to, say, 1968 or the late '60s, and you asked where would people have thought the world would be by the year 2000, 2015, 2016, it's fallen way short. If you look at the Back to the Future movie, Back to the Future 1 was 1985. They went back in time 30 years, and things had changed quite a bit from '55 to '85. There's still been a decent amount of changes. Back to the Future 2 went from '85 to 2015, 30 years into the future, like it was about a year ago, a little under a year ago. And that was a world that was supposed to be radically different. And I think the actual day-to-day changes outside of computers would have been quite modest in those 30 years or even, I would argue, since the 1970s.

And so, you can sort of rattle off these different areas, whether it's biotechnology, where there's been some progress, but it seems to have decelerated; space travel; transportation more generally; all kinds of ideas people had in the '50s and '60s about, you know, reforesting the deserts or underwater cities or all kinds of things like this that, at this point... New forms of energies. Yeah, fusion, you know, new forms of nuclear power that, at this point, all have this sort of retro-future feel. It's like it's the future the way it used to be. Star Trek feels very dated. There's, sort of, these things feel very dated in their optimism about how much could be done. And so, yeah, I do think there always are different things that are going on.

There's acceleration in some places. There's stagnation. There's increasing inequality. So you have sort of different themes. But I think that in a lot of these debates, I would say it's 70% is probably stagnation. And that if we focus too much on inequality or acceleration, we're going to get a lot of the public policy debates wrong. So if you focus too much on acceleration, Professor McAfee at MIT, The Second Machine Age, this runaway technological progress, and then it's going to lead to more inequality, and so you need to deal, so it's both this really good thing, has some problems, and then that sort of pushes you in a certain set of policy directions of what to do. Or, you know, the even more optimistic ones are things like,

things like Ray Kurzweil with The Singularity Is Near, where this accelerating future and all you need to do is sort of sit back in your chair, eat some popcorn, and watch the movie of the future unfold. And then the kinds of policy debates that we have in this accelerating world tend to be, I find, sort of incredibly ethically charged, where it always ends up being good versus evil technology. And so the technology is so overpowering that the main risks are that it's going to destroy us all. And so you have utopian forms and dystopian forms and nothing in between. And so you have the worries in Silicon Valley about AI. You know, do you really want to live forever or is that just ethically bad?

Whereas on the stagnation side, the antonym to good is not evil. The antonym is bad. And the problem from my perspective is not so much that we have evil technologies that are going to destroy the world, but that we have a lot of bad technologies and bad science that simply doesn't work. So evil has more this ethical thing and bad has more the sense of not working. If there's an ethical, moral version of it, it's that people are lying about the science, they're lying about the technology, and they're saying that it's really incredible when it's often not quite living up to what it is. So I find myself much more in this, that the stagnation is the general, and it's reflected in the economic data where, you know, median wages have been stagnant for 40-plus years.

The younger generation has reduced expectations from their parents and sort of this very broad sort of sociocultural indicators that suggest that things feel—they feel kind of stuck. And then you have a very different set of questions. What do you do about it? You know, what's gone wrong? How do we get out of it?

How much of it is in our control? Are we just hitting natural limits? I mean, that would be a question, I suppose, right?

Yes, so I think to the extent you see it as a technological problem or a scientific problem, one approach has been that you're just hitting natural limits. This is Tyler Cowen, The Great Stagnation, a recent book by Robert Gordon, another economist, The Rise and Fall of American Growth, and so it was this idea that there were some fairly simple inventions, the 19th, early 20th century, but maybe the rate of innovation peaked as far back as the late 19th century, where you had, you know, incredible numbers of breakthroughs and it's just the nature of the world, of the universe, that it's much harder to find new things. The low-hanging fruit has been picked and therefore we have to sort of resign ourselves to

far-reduced expectations and sort of a more austere future.

I just read somewhere a quote in an article, it wasn't an argument, just a statement, “You only invent antibiotics once.” I mean, you have massive gains in public health by a huge breakthrough like that where half your population isn't dying from infections, basically, and public health disasters in infancy. You can't replicate that.

Those are the arguments on the nature side. I'm a little bit more partial to the culture side of this, that perhaps there never was any low-hanging fruit. The fruit was always of at least intermediate height, and it looked always pretty high up.

Yeah, they didn't think at the time that, hey, these are easy breakthroughs, right? These are easy breakthroughs.

And certainly, you know, we have a situation where, you know, one-third of the people at age 85 suffer from dementia or Alzheimer's. And you would—so if you could do something about that, that would probably be close to as big a deal as antibiotics. And we're just—but we're living in a world where people don't even expect that to happen anymore, I think.

So what are the core problems there, do you think? I mean, what is the education, regulation? I mean, is it government? Is it just the culture as a whole?

Yeah, I suspect this is, you know, all these questions why this is happening, why questions I think are always hard to answer. They're always a little bit overdetermined. Right. My sense is that to some extent it is all of the above. But... but I think there is a big hysteresis part to this where there's a way in which success begets success and then failure begets failure, where if you haven't had any major successes in a number of decades, it does induce a certain amount of learned helplessness. And then it shifts the way science gets done or the way innovation gets done into the sort of more bureaucratic, political structure where the people who got the research grants are more the politicians than the scientists.

You're rewarded for very small incremental progress, not for trying to take risk. And so it's led over time to more incrementalist, egalitarian, risk-averse approach, which I think has not worked all that well.

How does one change that apart from telling people that they should take risks?

It's not clear how you change it on a cultural-political level. I'm always focused on the sort of very modest startup version, which doesn't actually work for everything. I think it's sort of like, again, it's a very imperfect solution. So I think you can convince a small number of people to start a new business or to do something new; convincing a much larger number of people to change things, I think, is sort of a much more challenging problem. And then this gets to all these questions about, can you reform government research structures? Can you reform NASA? Can you reform the universities? Can you reform even large corporations that in many ways are not much better than the universities or the governments?

Maybe a little bit, but not as much as we often like to think.

But I suppose one way to think about it, and this is something you've thought a lot about, is what we do seem to have one area where we have had pretty massive breakthroughs, bigger than anticipated. The one case maybe where Back to the Future, 2015 to '85, really is a genuine quantum leap, so to speak, and that is computers and related technologies, information, and so forth. And could one then just say, well, let's... go back and look at what happened there and what the structures were there and the incentives were there and the culture was there and why couldn't we then, mutatis mutandis, you know, try to make that more prevalent in other areas? I mean, that would be sort of a simple but reasonable approach.

That's really naive intuition. So it was relatively unregulated, which again might be harder to do in energy or medicine or some of these other areas. There were parts that were not that capital intensive. So you could get started with relatively small amounts of capital. And there's a question: is that true? Are there areas of science where that's true? Or are there places where if you need to have, you know, $50 million to test a new experimental design for a nuclear reactor, does that necessarily become this deeply political process where you're at risk of the money going to precisely the wrong people every single time? Right. So there was lower capital intensity. There was... uh... there was, uh... there was, uh... it was less regulated, uh... and then I also think there's been this history of success where over time you've attracted, uh... very talented people who believe they can do things, and, uh... and one shouldn't understate, understate, uh... how much that sort of belief, uh...

is effectual. One of my friends years ago was looking to join a big biotech company, and the pitch to people with PhDs coming out of graduate school was that we have a better softball team than the other biotech companies, because whether or not you discover something is so random, so unpredictable, that the only thing we can sort of control about your work environment is to tell you that you're going to have a better softball team. That's why you should join our firm. And so there's something around a lot of these areas where it felt that people were too much of a small cog in a giant machine, very hard to actually impact things, and there was sort of less of a sense of human agency. So I think that's what we always have to come back to, is how do we come back to a real sense of agency?

It does seem like the safety and environmental regulations just are so mass, and others, I'm sure, too, are so massive in the medical areas, the energy areas, transportation, and then just the political obstacles, the rent-seeking and the people who currently occupy things making it hard to make progress, which I guess didn't happen. Of course, after the fact, as you say, everything's overdetermined. Probably one could have thought in 1985 that it's not like there was no one in the computer or information technology field, and they were presumably interested in blocking new entrants, but somehow it didn't work that way.

Yeah, it's a complicated history. We had the thalidomide disaster with the FDA, so there's specific things you can point to that went very wrong. But yeah, I think today you would not get the polio vaccine approved. When it was first used, they dosed it wrong, and I think they gave—10 or 15 people got polio accidentally. And so today, that would probably... slow it down for another 20 years or something like that. So I do wonder whether we've become too risk-averse in various ways. But even this concept of risk is a very strange concept. One of the things you can do on Google is search for words and the frequency with which they occur in books over time. If you look at the word 'risk' over the last 200 years, from 1800 to 2000...

It's sort of a very infrequent word, very rare, until about 1970. Is that right? And then it goes up at an incredibly steep curve. And it becomes sort of this—much more common in titles: 'How to Manage Risk', 'How to Take Risk'. And so the thought I've been wondering about is whether a lot of talk about risk is actually even counterproductive to risk. That if you have a risk that if your kids are left unsupervised at the playground that they'll be kidnapped, or if you have a risk that this can go wrong or that can go wrong, or a risk that somebody's going to die from some new medical treatment, that 'risk' is actually this word that's used to discourage people from doing anything. And its more and more frequent occurrence

is a symptom of society in which less and less good risk-taking is actually taking place.

A friend of mine who's on the board of a financial institution says they spend more time at their—and I'm not sure this is bad, it's not even just reporting this—they spend more time at their board meetings discussing the management of risk than actually making money. It's minimizing the downside. And it's not a stupid thing to be worried about, obviously, especially after 2007 and 2008. But at some point, you're spending more time on the insurance to ensure that there's no disaster from the products that you're making than the products that you're making, right? I think diversification, you've written about this some, right? The kind of degree to which venture capital people think of in terms of diversifying instead of—

Yeah, there's all these ways where I wonder whether the focus on the sort of processes of risk minimization distracts you from the substance of ideas, of figuring things out, of doing new things. And so, yeah, so I think definitely something like that seems to have been very much at work. And I think that one of the things that's true about risk is that it's sort of this very probabilistic way of thinking about the future, where the future is dominated by chance, by fortune, and that that's sort of this all-powerful force that dominates everything. And I think it's, you know, one of the questions that I think is very unclear is whether this is—is this in fact a deep truth about the universe, or is it more about the abdication of our responsibilities?

And so, as a venture capitalist, I'm always tempted to—the temptation is always, you look at a company, you say, 'Don't really know if it's gonna work or not. I'm just gonna invest a small amount, see what happens.' And the temptation is to treat all these companies as lottery tickets. But once you treat them as lottery tickets, I found you've somehow psyched yourself into losing because you've already psyched yourself into writing too many checks a little bit too quickly, and you're not actually making a statement about the inherent chanciness of the universe. You're actually making much more statement about your own laziness and your own unwillingness to try to really think things through. And so I do wonder whether there's something like that that gets obfuscated by this talk about risk, where it always sounds like it's a statement about the larger world, but it may really be more statement about—

the failure of our ability to think things through. And maybe it is hard to precisely model these things out or something like that. In the case of startups and very innovative businesses, it's always quite unclear to me how you even talk about risk or probability. And so you can talk about it if you can do an experiment many times and see what happens over and over again. So, you did high-frequency trading on Wall Street, that has a probabilistic character where you can probably model the risks very, very precisely. But if you're investing in a one-of-a-kind company, I don't even know how you'd go about measuring what the risks are, how you would quantify it. And if you have a sample size of one, standard deviation is infinite.

And so, in theory, you can't say anything about the risk. If Mark Zuckerberg started Facebook over again a thousand times, how often would it work? We don't get to run that experiment. And so somehow risk-orientated processes, they make sense in a context of large N. So insurance is a context of large N. We have millions of people driving cars. So many have accidents. We can sort of model it out. Or life insurance policies, actuarial science. These are all sort of large-N sciences. But I think there are a lot of very important things that are small N, or even N equals one, or it's a very small number. And if we're too beholden to these risk models, I think we just don't do those kinds of things, and they may be very important.

Yeah, and the large-N things, the insurance schemes, are important for maintaining the stability of society and, you know, taking care of people who are afflicted by bad luck and bad circumstances, but it doesn't get you, it doesn't—you could have a perfectly insured, a perfectly running insurance scheme, so to speak, and never have any progress, I guess is the simple, minor way to put it.

Or the even more extreme way I would put it is if you had a perfectly running insurance scheme, it would probably collapse under its own weight eventually. And so if you think of our education system as largely an insurance policy—the universities are largely this insurance policy for upper-middle-class parents who are scared that their kids are going to fall through the ever-bigger cracks in our society—it can sort of work. But are we really going to have a future in which everybody's insured? And isn't that a recipe for disaster over time or nothing?

Sort of, the welfare state is a reasonable supplement to a growth economy.

But it can't work as the be-all and end-all for everybody. Yeah, that's interesting. And I think in the same way, a super-low-risk—there are parts of it where it can work, but it probably can't work universally. And I don't think it can work remotely as extensively as it has been done. You know, I was guilty of this. I went to law school. It was sort of the low-risk, seemingly low-risk thing to do from undergraduate in the late '80s, early '90s when I went to law school. And it was a fairly low-risk way to get sort of an upper-middle-class career in a big law firm. And in retrospect, I think it's turned out to be quite high-risk for a lot of the people who did it because there were far too many people who did it.

You know, it worked well for a few years and then a lot of things just tend to go wrong even for the people who come out of these very successful, track kind of places. So, you know, if you wanted to date this maybe in, you know, 1965 or 1970, if you were graduating from college, there was actually a way that you could do a low-risk, very successful career. So few people were doing it that, you know, if you—and there was a way you could sort of arbitrage this risk successfully. Don't think that's been working terribly well. I think it's been working less and less well for the last 30, 40 years, even as it's been understood better and better.

So... Yeah, that's interesting. And I mean, people forget, and you know much more about this than I do, but the '07-'08 crash, the instruments that are generally thought to have contributed to the crash, at least, and to have gotten out of control, and so that people didn't know what they were holding, and one thing led to another, the kind of knock-on effects, were precisely supposed to be lessening risk. It wasn't quite right that many people on Wall Street thought, "Hey, let's gamble." Some did. But really, it was the opposite. It was, "If we can spread..." Holding one mortgage is a big gamble. Holding one one-hundredth of a hundred mortgages is safer because three of them will go under, but of course, you have the other 97.

And then it turned out that, guess what? The whole thing is more rickety, in a way, than the very old-fashioned community bank holding the local mortgages.

I would say it turned out to be more rickety because people thought it was too safe. In theory, diversified portfolios would be somewhat safer, but if everyone thinks they're incredibly safe, then you might leverage those diversified portfolios up a great deal more than you otherwise would, and you would take risks that a community bank would never have taken.

Sure. I mean, it's the, it's the, it's the sort of like the, the sort of probabilistic, when I think of this as probabilistic kind of a thinking has infected so many things. And the, you know, the very strange version in politics is the way that poll-taking has become so, so unbelievably dominant where—and, you know, it's like you can model these sort of statistical aggregates of people probabilistically as you get to 51%. You only need 53%, which is where, you know, Romney's 47% comment came from. It was, you know, 'I need 53, the other side needs 53, don't need any more.' And then there's sort of just this probabilistic science of, of getting at that. It has had a certain amount of power. We shouldn't deny that it's had a great deal of power.

And then at the end of the day, it has some very severe limits where, you know, if everyone's, you know, if everyone, if all the politicians are looking at the same polls, you know, we're just really electing Nate Silver as president or something like that, which is a sort of a strange change in constitutional priorities.

And everyone's micro-targeting. And then, as happened this year in 2016 on the Republican side, at least, a guy blew through who did no micro-targeting. Everyone else was picking their lane to run in. I was struck by this, and the word—someone mentioned 'lanes' to me recently, and I hadn't heard the term in about a month, but early in the primary process in late 2015, early 2016, that was all the talk about political. It wasn't stupid. I mean, you know, there's the conservative lane, the evangelical lane, the moderate lane, this lane, governor lane. It turned out the one guy who seems to have ignored all that, Trump, and just decided, 'I've got a message, and I'm not right now leading, but I can be leading, and I can go up from 25 percent to 35 percent to 45 percent to 55 percent.'

Sort of did that, and it's—it's a sort of substance that seems similar to, yes, whatever virtues and limitations. But I mean, to someone who just didn't accept the kind of, 'I'm going to diversify my political effort,' you know, right? Or narrowly target my political...

I think we're still too close to 2016 to really coherently and totally... that idea. On the other hand, you know, for, you know, all the ways that Trump gets described as this sort of, you know, incredibly independent kind of a non-politician, it's still very striking how all the speeches just get started with recitations of polls. And so it's again...

But that's more of a snowball thing, don't you think? Kind of a, how do you say that? Getting people to—it's legitimizing himself by, 'Hey, I'm winning,' and then other people want...

And so again, maybe it's another commentary on how beholden we are to polls. And so if you're the politician or the aspiring president who talks more about polls than anybody else, that's strangely advantageous, which seems, again, like a very odd but somewhat true statement.

But in a democracy, people want to be on the winning side. And so I think he had a real instinct on that, too, that there's a sort of way in which you—you keep doubling down, so to speak, and people want to be with the winner.

Yes.

Who knows what will happen?

Again, these things can work in certain contexts for a while. You can be winning as long as you're winning. Yeah, right. Exactly. But there are, I suspect, all these other contexts where this sort of probabilistic approach has not been—has not been terribly helpful. In politics, at least, you can still just do surveys. It's somehow related. If you're talking about starting a business and saying, 'We're going to just do this random walk. We're going to do A/B testing of different types of products and see what people want. We're going to have no opinions of our own and just do all this testing of customers.' The problem is the search space is just way too big. There are way too many things you can do.

You don't have enough time to go through this sort of statistical surveying and feedback mechanism. And a lot of the things that have worked the best have been sort of paradoxically not so probabilistically beholden. If you look at Steve Jobs's Apple, where it was... you know, you have something like the Isaacson book on Steve Jobs where it sort of portrays him as this tyrannical boss who just yells at people all the time. And even if all of that's true, it doesn't seem to me to get at, you know, why did it work at all? Why did it inspire people? And I think it was because there was a plan, you're going to execute against it, and you could sort of pull off some really incredible things that you could never do if you just did everything through this sort of instantaneous feedback from different kinds of things.

So I think complex planning, complex coordination, these are the kinds of things that it's gotten very hard to do in this sort of probabilistic society. I described to you a complicated plan. In a technological context, you'd think of it as like a Rube Goldberg contraption. It's just not going to work. Something's going to break. Because we think of every step as probabilistic. We think of the steps as more deterministic. It's just you just have to get the different steps to work. And you can actually have a Manhattan Project. There's no reason that you can't do this. There's no reason even the government can't do this. It actually did in the 1940s. And you could send a man to the moon with Apollo.

And you certainly should be able to do a website for the Affordable Care Act, since that's demonstrably a lesser, demonstrably inferior technology to Apollo or Manhattan.

And I suppose on the business side, if you surveyed, polled ahead of time—I mean, this has been famously sort of said about Starbucks, I think it's right: "Do you want to pay two or three times as much for a cup of coffee that may not even taste better, really, than the coffee you get in your diner, kind of a bitter European taste or whatever?" I mean, people would have said no, presumably, or in any case, it wasn't clear there was like huge, incipient demand or whatever term the economists would use, you know, unmet demand for that. And I'm not sure even if one polled 10 years ago, "Would you like to have a little phone in your pocket that you could do this, this, and this?" Well, maybe, maybe not.

I don't know. And there is a way in which all that way of thinking about economics does seem self-limiting and misleading in some ways.

Well, it's... I'm always... It's always this question whether it's that good to just always be looking at the people around you and getting feedback from them in different ways. There's this very strange aspect in Silicon Valley where so many of the very successful entrepreneurs and innovators seem to be suffering from a mild form of Asperger's or something like this. And I always wonder whether this needs to be turned around into a critique of our society, where if you don't suffer from Asperger's, you get too distracted by the people around you. They tell you things, you listen to them, and somehow the wisdom of crowds is generally wrong. The Malcolm Gladwell "Wisdom of Crowds" book, there's a very specific thesis that it has, which I believe is true, and there's a way the term always gets misused.

The specific thesis is that the wisdom of crowds works if everybody in the crowd is thinking independently for themselves. So if we have a more jar of marbles and everybody guesses how many there are then somehow the collective judgment ends up being better than most individuals judgments, but but if you have a if you have a But if you have sort of most more common way the wisdom of crowd works is through this sort of hyper mob like behavior where I think you get a lot of irrationalities. You get the wisdom of crowds becomes the madness of crowds. It becomes a bubble in finance or something worse in politics when it goes very wrong. They've done these studies at Harvard Business School, which I often think of as consisting of the opposite of Asperger-like people.

So they are people who are extremely socialized, extremely extroverted, have relatively few convictions of their own.

But good work habits. Good work habits.

And you put them in a two-year hothouse environment in which they spend two years talking to one another, trying to figure out what to do, which leads to this very dysfunctional wisdom of crowds dynamic where you will simply—because none of the other people actually have thought about it for themselves and have any independent ideas. They've done studies on this where systematically the largest cohort at Harvard Business School goes into always the wrong thing. So, you know, 1990, they all wanted to work for Michael Milken sort of one or two years before he went to jail. You know, they were never that interested in tech except for, you know, 1999, 2000 when they descended on Silicon Valley en masse and timed the end of the dot-com bubble perfectly, and on and on.

So there is something about this that's interesting, that's very tricky, where probably a lot of innovation, creative thinking, doing things that matter generally depends on not being so beholden to the people you're immediately around. Even though you get feedback and the feedback's helpful, there are these cases where it can go very wrong.

And I suppose it's most helpful, I would think, in a static situation. If you're dealing with a static universe—X number of people, X number of things you can do—then, in a way, getting the most feedback suits you best to deal with this situation, I suppose. But it doesn't tell you what happens when there's a breakthrough in innovation or some disruptive event, right?

Yeah, I mean, it depends to some extent on what your priors are. So if we—like, my prior is that there's a lot more innovation that could happen. And so if the feedback mostly is a form of anti-theories—'can't do that,' 'that's too bold,' 'that doesn't quite make sense'—in a world where a lot of innovation is still possible, this sort of horizontal feedback probably has a very bad dampening effect. If we're in a world where, in fact, everything's been discovered, everything big has been discovered, then this sort of feedback would stop people from wasting their lives on some sort of quixotic quest of one sort or another. So it does depend some on what your priors are. But if I had to make a judgment on it, I think—

I think we are in a world where these feedback mechanisms have gotten far too powerful, where people are too easily swept up by these mob dynamics. I'm certainly not going to go on your TV show and blame this on the internet in any way, but you have to ask whether there are ways some of these technologies have maybe even exacerbated some tendencies that were already there in our society. The phenomenon of political correctness—so many different ways to describe it, but certainly one is just that you have these incredibly powerful negative feedback effects that get brought to bear and have this very inhibiting character. I don't think it actually results in things being far more generative than they otherwise would be.

It cuts off a lot of lines of inquiry, but I don't think that means that it opens up that many more.

And final question just on this topic, which is very interesting, I think. It seems to be that conservatives and libertarians, we're both pretty sympathetic, I think, to those points of view. You know, Hayek, one of the great libertarian heroes, and justly so, and a great economist. The Fatal Conceit, you know, you can't have central planning, which I think has been a very useful critique of big government and welfare state liberalism. Don't you think it's had the effect, just listening to you and thinking about this, of, on the other hand, it's sort of, conservatives have bought into a very probabilistic, or maybe that's quite the right word, but... such a hostility to planning, such a hostility to human conceit, that almost spills over to a kind of fatalism about human innovation or agency, almost?

Yes, I think there is sort of a very strange way that some of these ideas developed, and certainly in the area of economics in the 19th century, sort of pre-Austrian, classical economics, the thought was that you could still measure things objectively. So, how many pounds of steel is this factory producing? How many tons of steel is this factory producing? How many cars per hour are the workers producing on this assembly line? There was that sort of an intuition about objective value. And there's been the shift towards making things more subjective when it all becomes sort of unmeasurable or there are too many variables to measure. And I think that was, you know, in some ways was started by Austrian economics.

And I think now, at this point, it really permeates all of it. It's not just the right; it's also, you know, the Obama administration wouldn't say that they can have strong substantive ideas of what to do. They can improve processes, but they would never actually think that you could actually build a very specific thing in a pre-planned way. They don't even believe in that anymore. Right. They believe in nudging, right? Isn't that the term that's... Yes, the Cass... Which is a very revealing thing when you think about it.

It's such a limited aspiration for...

And it's, again, feedback from things that are already working, and we improve them a little bit. So it's all—we're all just going to climb up, go up the up-gradient, and we may get stuck on a low-lying hill. Sometimes you have to step back and wonder, where in the world do we find ourselves, and we just simply go uphill? Do we end up on a low hill, or do we really end up on Mount Everest? And we have all these hill-climbing theories; we have no sort of valley-crossing, mountain-climbing theories. And we need some more of those kinds of things. But yeah, I think the shift towards a subjective, you know, subjective ways of measuring economics, it then ultimately leads to this way in which we can't even coherently talk about progress.

So, are we actually progressing as a society? And then the answer becomes, it's just impossible to know. You can't tell. It's just different. All these things are somewhat different, and maybe, maybe you're living in a much smaller apartment in New York City than your parents or grandparents lived, but you have an iPhone with a really smooth, flat surface, and so the hedonic benefit from that—how do you trade that off versus the apartment that's a quarter the size of that of your grandparents? And so, I think it's hard to know how to think about these things, but I do worry that the stress on these sort of subjective, hedonic economic measures are excuses we're telling ourselves to hide the stagnation or even decline from ourselves.

Speaking of progress, one of the ways people hope for progress obviously is, and more broadly, artificial—well, not broadly, but artificial intelligence, I guess, in particular has been a focus recently. But you've given this a lot of thought and are pretty closely involved in a lot of these efforts, or watched them closely and invested in some, I'm sure. What do you think? Are we at some tipping point? Is the world 20 years from now going to look that different in terms of what machines can and can't do?

Well, it's very hard to say. And I have many somewhat conflicting thoughts on this. I don't necessarily come down very strongly on one side or the other of these debates. That certainly computers generally have been an area where there's been a lot of progress. So it's maybe not unreasonable to at least ask the question, how much more progress could there be? How many ways could AI happen? But one of the—on the other hand, one of the things I don't particularly like about artificial intelligence is that it has become such a buzzword. I think that these buzzwords often always obscure more than they illuminate. And one of the ways to see that it's a buzzword is to see how ambiguous it is. Artificial intelligence can mean both the next computing

the last computing that humans will ever build and everything in between. And so it has this sort of rather elastic meaning. When artificial intelligence means the next set of computers, it sort of pushes the conversation in somewhat more automation, replacing certain low-skill or medium-skill kinds of activities people are doing. When you talk about it as the last computing device where you're building a mind that sort of can outthink and outwit any human being, you end up with these very scary, somewhat political questions. Is it going to be friendly? Is it dangerous? And if something like that could be developed, maybe it would be on par with extraterrestrials landing on this planet, where I think the first question would not be about, what would this do to the unemployment rate?

The first question is, are they friendly or not? The first questions would be political. So I think that's sort of a general framework. I would say that certainly the sort of bullish AI consensus that exists is that we're making progress very quickly, deep reasons why computers couldn't do everything better than what humans do. And this may indeed happen in the next few decades. And that this would, of course, be an extraordinarily important and transformative set of changes. And I'm certainly open to all these perspectives, but I also wonder whether there's certainly parts of this that one could question. And if you had to be a little bit more critical of it, sort of the two points of criticism would be to first start with the history where people have been, in some ways, too optimistic about AI for quite some time.

So in 1970, there were people who said you'd get computers to understand language, everything humans could do within a decade. Same thing would have been said in 1980. And so, you know, we've been here before, and so there's been a history where this has turned out to be more difficult than people would have thought. And then, of course, there always is this sense of whether it's maybe just a particular moment in time where, at the peak of a technology cycle, the only thing we can worry about is technology that's so good that it's too fast and it changes things too radically. In spring of 2000, there was this essay that got a huge readership in Silicon Valley by Bill Joy, one of the founders of Sun Microsystems.

It was, "The Future Does Not Need Us," sort of how runaway technology would get rid of people.

I remember that well, but that's already 15 years ago.

As a socio-cultural observation, a psycho-social observation, in spring of 2000, what we should have worried about was not whether the technology was going to work so well that it would be this runaway progress, but the real question was, was it working at all? Were the business models working? It turned out that a lot of things didn't work that well. And we had sort of a period when people went back to banking and back to consulting from Silicon Valley. B2C, B2B didn't mean business-to-consumer and business-to-business; it meant back to consulting, back to banking. And so anyway, so I do wonder whether the sort of mini AI bubble that we've seen in the last few years is maybe symptomatic that we're at some, again, some local peak in optimism about how much Silicon Valley is doing, can do, all these sorts of things.

I think one way—one aspect of it that I think is somewhat disturbing is certainly that there's this sort of probabilistic thinking that again creeps into the AI issue as well. So if you meet someone in Silicon Valley who believes that AI is possible, it's happening soon, and it's potentially dangerous—these are three very widely held beliefs. The fourth thing that you can always push back on—you don't want to question those three beliefs, but the fourth one that's a very powerful one to question back is, well, you have no idea of how to build one that's safe, and you couldn't build one that's safe. The way you've defined the problem, that you're going to build a superior mind that will be able to outthink you, you won't be able to build it in a way that's safe.

One of my colleagues was sort of talking to one of the top AI researchers and sort of pushed him on this. And it was basically, 'Professor, you obviously don't believe any of your theories about AI. Because if you did, you wouldn't be publishing any of it on the Internet. Because when the AI emerges, it will read about it on the Internet, and it will know to hide and to not make you aware of how powerful it really has become.' So there's something akin to that that's implicit in all of this, where if the sort of the optimistic case about technological possibility is true, there's a sense of helplessness in terms of what people can actually do about it. And so it's very much linked to this rather dystopian view.

And this is reflected in the Hollywood movies, where I cannot think of a single movie from Hollywood about AI that doesn't have a dark and rather disturbing undercurrent to it. And so again, I think it's a sort of probabilistic reasoning that things are sort of—technologies are out of our control, there's no human agency, we can't actually know what we're doing—that gives it this very strange quality. And of course, of course, there is a sense in which the term AI simply means that human intelligence isn't up to the task. One other way of interpreting the AI boom is that on the surface, it is about extreme optimism about the potentialities in computer technology. And then beneath the surface, it simultaneously is perhaps a great deal of pessimism around the possibilities in other technologies that will be developed by humans, and deep pessimism about the possibilities for what humans can do.

And so it's sort of 'a man with a hammer sees a nail everywhere.' So one interpretation I have of the AI bubble is that, again, in a strange way, it is symptomatic of the technology stagnation thesis.

Now, that seems to fit with the notion that we're developed societies and there's not much more we can do on our own, so this deus ex machina kind of literally is going to come in and make it—do what? I don't know, but, you know.

It's like waiting for this AI to save us from all these things and then don't know whether it'll be friendly or not. So it has this very strangely passive aspect where somehow there's not enough room for human agency to my liking. Now, certainly, one sort of somewhat more basic point that I always try to make is that it's not at all obvious why the question about, let's say, near-term AI, so not the way-futuristic stuff, but the next generation or the generation after that, why it should be seen as such an adversarial dynamic. We're always talking about computers as substitutes for humans, and yet the reality is they are very different. They are, you know, computers are able to do things in this incredible brute-force way.

Humans are sometimes able to do things far more effectively. And yet there's sort of ways in which our minds are probably much simpler than we think. It's probably wrong to think of a human mind as having hundreds of billions of neurons, and that somehow codes for hundreds of billions of things in our mind. You're a really smart person if you have 20,000 vocabulary words. And so I think there is something about computers and humans where they're deeply different. And I wonder whether the focus on AI somehow obscured these differences. And so the mystery in some ways is, why have we actually not built AI? And the conventional explanation in Silicon Valley is we have not built it because human minds are so complicated.

You have hundreds of billions of neurons you'd need to have to have a computer with hardware the type we haven't quite developed yet. And it's sort of just a matter of time and we'll get there. But you can make the same argument. You can say this cup here has close to Avogadro's number of molecules or atoms in it and could never be modeled by a computer. But you don't actually need to model every atom. You can just model the basic structure in a similar way. If there is such a reductionist theory of the mind, were possible, it could perhaps have been already designed on 1970s or 1980s-type hardware. So I think there are some very mysterious questions like this that have not been fully thought through.

My guess is that it's possible that there are just these really big differences and that there are—and there's a separate question whether you can brute-force a simulation of a human mind. And there probably are ways you can do it. And there's some limits to that. But they're naturally complementary because they're so different. We normally need to be afraid of people who are just like us because those are the people we're competing against. You need to be—globalization is scary because it means that you have sort of very underpaid people who aren't that different from us in other countries competing with people in the U.S. and Western Europe. You know, the computers are not; they're complementary.

They're not really competing. They would be scary if we had a super-futuristic version where we had a robot that looked just like you, Bill Kristol, and we didn't have to pay it any money in any context. And you'd be—you'd be rightly alarmed by that. And this is the common-sense intuition why people are scared of cloning. There's a sort of bioethics cut on cloning, but the common-sense reason is that you had sort of 100 clones of yourself, they'd undercut your—they'd be competing with you, and you're competing, you're always competing with people who are like you. And so, to the extent computers are really different, that's, I think, much more of a positive than a negative.

And when you say the human mind is simple, you're saying not that it's easy to replicate, but the opposite, that there's something mysteriously simple about it, so that brute force doesn't replicate at least some of the things that it can do.

Yeah, my view is that it's mysteriously simple. And it seems to be able to do very powerful things with relatively few components. Maybe there is some relatively simple algorithm that could replicate it, but it's strange that we haven't found that. But it's perhaps not a problem of hardware, which is the naive standard view that we just need more hardware and you'll get it to work.

Yeah, when the computer beat Garry Kasparov in chess, when was that, in '97 maybe? Yes. Something like that. Charles Krauthammer, who knows a lot about chess and knows a fair amount about science too, wrote a cover story for us, which we titled, “Be Scared, Be Very Scared.” It was an interesting reflection. I haven't gone back and looked at it in a while, but I would say 20 years later, it doesn't feel that scary. I mean, clearly, at a game like chess, with a set number of pieces and a board and rules, the brute force, so to speak, can just, at some point, surpass even the greatest player, but most of life isn't like that, I think. And one doesn't have the sense that computers are defeating humans in the way that one might have thought 20 years ago would now move from chess to other parts of life.

Or maybe I'm wrong about that, but it does, you know.

Yeah, I think it's an open question. So I think it's happened certainly less quickly than people would have predicted at various points. You know, certainly you have day trading, you have all these sort of great trading algorithms where the computers are maybe doing two-thirds of the stock trading on the stock market or you have certain places where these things happen. On the other hand, things like language translation still feel extremely far away from that. If anything, the way Google has improved its language translator is to find phrases in books and then see how the humans translated those books. And so it sort of has found efficient ways to leverage off of human translators, and it's not actually any kind of semantic understanding on the level of the computer of what's actually the meaning of the words.

And so that kind of a thing still seems very far away.

And day trading is, in a way—I don't know that much about it, but it seems like a good example because it's not stock picking. It's not, let alone venture capital investing, which, to my knowledge, computers haven't, of course, used them.

Yes, I'm hopeful that venture capital is extremely far from being replaced by computers.

Yeah, the use of computers in those areas is to help people process data very quickly and do a lot of comparisons and stuff, but at the end of the day, it's not.

The more hopeful view that I still have is that it is likely to be just a continuation of what's been happening since the Industrial Revolution, where mechanization, automation free people up from certain kinds of repetitive tasks and free people up to do other things. And then it can be scary, living in a society where there are not enough other opportunities, where there's not enough growth, where there are problems like that. But in and of itself, freeing people up from the drudgery of repetitive tasks is probably a good thing.

Yeah, and the transition can be very unpleasant. And of course, when one reads about it in history books—I was having this conversation with someone recently, you know, well, people went from agriculture, from rural areas to the cities. You know, they got through it and London was a mess and everyone writes, you read early accounts of early, you know, Blake and the coal mines and early industrial London and other, less London, maybe other industrial towns in England and it's terrible. But, you know, they get through the problem and everyone ends up well off. Of course, the getting through takes 20 or 40 years and includes all kinds of bad things happening and instability and maybe we're in such a—

Moment here, you could argue with all the anxieties about people losing jobs, but there's also so much nostalgia, don't you think? I mean, it's like, I mean, I'm sorry. It is, if you are a coal miner and you're out of work at age 52, it's very tough and it's hard to adjust and you have to move or you're, you'll never maybe have quite as well-paying a job. And it's easy for me to sit here and say, you know, that's just part of social progress. That's not the right—that's not a very palatable thing to say. On the other hand, it shouldn't be too nostalgic that it was a great thing that we had tens of millions of people in coal mines. That was not healthy. Or the assembly line. All the nostalgia now.

Trump's getting the votes of, quote, "the white working class," the people who were once working in steel mills, and that was great. Or on assembly lines. It seems to me an awful lot of literature was written in the '50s and '60s and '70s about how assembly lines were dehumanizing and cog in a machine and not a great... you know, one would want to get beyond that, right?

Right, and perhaps if we didn't want Chinese manufacturing that was powerful, we needed to automate the assembly lines even faster than we did in the U.S., or if anything, we didn't mechanize quickly enough in some of these industries. I think the sense of nostalgia that we have is a sense that there is a lot that we're losing, and the dilemma is that the things that we're losing are very obvious, and then on the other—and at the other end of the tunnel, there are many things that we will gain. And we have every reason to think the things we will gain are much greater than the things that we are losing. But it's sort of obvious what we're losing. It's not at all obvious what we're going to gain on the other end.

I think that would have been certainly true of the early 19th century. We were in the throes of the first industrial revolution. And there certainly are aspects of it that are like that today. But, again, my worry with all these things is, if anything, there's not enough happening. If you take the biggest innovation that people are talking about now, it's self-driving cars. Yeah, it's funny. I was just going to ask you about that. So let's talk about that. Is that really so big a deal? In a way, I mean, right. I think there are one or two million people who are employed as drivers. It's like maybe one and a half, 2% of the workforce maybe, something in that ballpark. Maybe it would increase efficiencies because you could get some work done in the car while you're driving to the office.

Maybe it could lead to a 5% increase in GDP in the whole economy. Maybe I'm underestimating it somehow, but I think it would be a... significant change, but it wouldn't necessarily take double our GDP or anything remotely like that. And so the fact that that's sort of the most transformational change we can imagine is, again, perhaps—is, to my mind, to my way of thinking, strangely unambitious.

Well, you famously said, you know, they told us what happened. What was your famous joke?

They promised us flying cars and all we got was 140 characters. So now we're getting self-driving cars. It's better than 140 characters.

But really, when you think about it, I hadn't really thought about it this way. The self-driving cars... There are still cars going on the same roads in the same traffic. Maybe it's a little less traffic because there are fewer.

Maybe they can park themselves. You don't need to look for parking.

You don't have to keep one. Not everyone will have to have a car. You'll just tell the self—you'll call, like, Uber, the self-driving car will come get you. Less congestion a little bit because there aren't, but still. But it's not flying cars. I mean, that is, in a way, the—which would have been a quantum leap, so to speak.

In theory, it could help congestion a lot. In theory, it could take a lot of pressure off of parking spaces and things like that. And then at the same time, the fact that this is the technology that's iconically the most radical that we can sort of concretely describe. It's more than Twitter, but it's not quite vacation trips on the moon or something like that.

It's a very good company.

I just like Twitter. I also agree there that this is what it's come to in a way. There were huge breakthroughs, obviously, real breakthroughs, email, iPhone, all that stuff is pretty amazing.

Google Search, but... A self-driving car would be, I'd say, almost as big as the car itself. But I would still say that the original invention of the car was bigger than a self-driving car. Oh, yeah, I think so. Again, if you sort of had to give the rough qualitative...

Yeah, from horse and buggy to car is a bigger jump in people being able to, you know, in lessening distance, I think, than a car to a self-driving car, which certainly doesn't lessen distance at all. It, of course, makes it easier, you know. But it is sort of like Uber replacing. I don't know. Yeah. I was thinking the exact same thing myself, but not quite in this context. And it is the iconic next breakthrough, but it's not that big a breakthrough, you know.

Yeah. But so there are... And then there are, of course, even a company like Uber where you have this major innovation. I often wonder whether it's more symptomatic of the failure of certain political structures. So the vision in the '50s and '60s was that you'd build very high-speed transport systems, which in the San Francisco Bay Area where I live were basically vetoed by local zoning ordinances. And so it's sort of like this very second-best solution because you couldn't build the much faster kinds of things that people thought were natural in the '50s or '60s to develop. And so again, it's a compensating device for sort of the dysfunction of our cities where there's not enough parking, so you don't want to drive your own car.

You can never find a parking space. It sort of ends up being, and then the public transportation systems don't work. Cabs are limited, so there's no market system for taxis. All these things that are dysfunctional, and then we have this innovation to sort of ameliorate the dysfunction. But if the political systems worked better in our cities, this would be sort of a... we might be doing some very, very different kinds of things.

So you're more worried still about the lack of real technological breakthroughs, if I can put it that way, than this notion that we're on the cusp of all kinds of technologies which will be dangerous to us and out of our control.

Far more worried about the lack of good technologies than the danger of evil. I feel as a venture capitalist, I see all these forms of bad technology, bad science, where it's things that just don't work, the cool-sounding things. The general problem is not that they're ethically problematic, it's that they just don't work. Or the things that do work are often on a scale that's incredibly modest. There's often a way in which humor is used to sort of hide disturbing truths from ourselves. And people made fun of technology in the early 20th century. The humor was to disguise how scary it was, how much it changed, how drastic it was. And today, it's like people throwing virtual cats at one another on the internet or something like that.

The humor, we make fun of technology. It's to hide from ourselves the disturbing fact of how trivial it is and how small, how small it is. So people are still worried about what's going on, but I think it has a very different feel from what it did 100 years ago.

Yeah, that's really fascinating. Any last word of relative optimism, however? I mean, it is striking. We're here actually having this conversation. We're about to go to a more academic conference on the mastery of nature, sort of modern philosophy and its attempt to conquer nature and, in a way, the amazing success. But so much attention has been given to the downside of that, and was it a bad idea in the first place, or has it created as much harm as good, is it out of control, as we were saying earlier? In a way, you're taking a very interesting and sort of contrarian view that maybe the problem is that we need to recapture a little of that early verve of the attempt to conquer nature for the relief of man's estate or whatever?

I don't think we've lost it altogether, right? And so it's always—there's always, there is, and there's always a way in which the post-2008 malaise—the silver lining to this is that there is a sense that we need to do new things. We can't just keep going the way things were. And there was a complacency that you could have had 10 years ago that I think you no longer quite can have today. People are far more open to this idea that perhaps there's been not as much technological change. Do you find that?

Since when did you first introduce the idea? It was about six, seven years ago?

I wrote about it first in 2011. I was starting to talk about it three, four years earlier. And certainly even in 2011, people thought this was very crazy. And at this point, even in Silicon Valley, there are people who sort of said, 'Yeah, I've thought about it some more, and I think there's actually, you know, there's a lot to this.' So there's definitely more of an openness to this notion, and maybe that's the first step in getting out of it. You have to realize that we've been wandering in the desert for 40 or more years now and not in an enchanted forest. And that's how we find the first step to get out of it.

40 years is a good time to end the wandering in the desert. I think it's a little over 40, but it's time to leave. Peter, thanks so much for taking the time to have this conversation. And thank you for joining us for Conversations.