Economist and 2019-20 CASBS Fellow W. Brian Arthur chats with host John Markoff about the evolution of technology, the application of complexity theory in economics, and society’s struggle with distributive economics in an age of increasing productivity.
W. Brian Arthur’s homepage at the Santa Fe Institute
“Where is technology taking the economy? ” - W. Brian Arthur. McKinsey Oct. 2017
“Complexity and the Economy” W. Brian Arthur. Oxford University Press, 2014
Center for Advanced Study in the Behavioral Sciences
Narrator: From the Center for Advanced Study in the Behavioral Sciences at Stanford University, this is Human Centered. What could be more complex than the study of complexity? How about the mashup of complexity and economics? In this episode, renowned economist and complexity theorist W. Brian Arthur sits down with host John Markoff for a lively talk about his work exploring the interplay between technology and the economy. We'll hear them dive into Arthur's early work in complexity studies using simulated markets, his thoughts on hidden and emerging digital economies, the topic of automation, and the question of whether capitalism is flexible enough to survive these shifts.
John Markoff: When did you come to CASBS with the idea of exploring?
W. Brian Arhur: Quite a while ago, about 15 years ago or more, 20 years ago, I started looking at how technologies evolve or how the whole collective of technology evolves. Since day one. Machines, steam engines, computers, the whole works. And I had figured out that it wasn't by Charles Darwin's variation and selection. Of course, there's plenty of that happens, but new technologies come about by being combinations of previous ones. This might sound very strange, but actually when you make a speech or any long expression, say in English, you're combining words and phrases and ideas that are already there. So when people invent technologies, what they're doing is combining things that are already there, whether they're computer processors or circuitry. So I came to CASBS really to investigate what I call evolution by combination or combinatorial evolution.
John Markoff: Beyond— you're going to go— you're going beyond technology.
W. Brian Arhur: Yes, absolutely.
John Markoff: There have been CASBS scholars here in the past, actually one of the people invented a school of thought called the social construction of technology, and I was wondering if that necessarily is antithetical to the way you work at technology, or you could actually accommodate a sociological context for that integration idea.
W. Brian Arhur: Yeah, I've read some of that stuff over the years. Many technologies are socially constructed. You don't get social media without a lot of social interaction, for example. Women's bicycles came about because 100-something years ago women had long skirts and and so on. So, sure, there's plenty of social construction of technology, but it occurred to me, or seemed pretty clear to me, that nearly all technologies, the ones I'm interested in, were invented to serve other technologies. When radar came along, you're basically generating very high-frequency radio waves, bouncing that off some metal surface like a faraway aircraft. But you need to get very, very high-energy radio waves to be able to do that. The cavity magnetron and the klystron tube, these were invented for that purpose. Most technologies are invented with the idea of serving the purposes of some other technology.
Narrator: There was a great show, Connections. That was the whole premise, right? He would sort of trace—
W. Brian Arhur: That's right. My background— people get puzzled, they think, "The guy's an economist, why should he be interested in technology?" But my background is in engineering, and I wondered, "Where do technologies come from?" Most of the literature says, "And then something very inventive happens, something very creative happens, and all of a sudden you get the computer." the laser printer or whatever it is, the polymerase chain reaction, but it doesn't work that way.
John Markoff: Does place fit into your, to your model at all? You know, there are historically, you know, whether it's Venice or Silicon Valley, or there are places where there are centers of creativity.
W. Brian Arhur: Yes, absolutely. I think this is true in all human endeavor. Because very often you get concentrations of people who know something, who really understand, say, how to make violins in Cremona in Italy, places like that. Putting together technologies isn't a matter of being creative or inventive, it's a matter of solving problems. And I noticed when I first went to Xerox PARC that if you had a particular problem, how to etch a 3-micron whatever groove, you could walk down the corridor and ask somebody. It's really important to have people locally who understand how things work, and even more important to understand how things don't work. You don't want to be wasting your time. It's always good to have a concentration of people.
John Markoff: I wanted to ask you about one model for creativity. I think it comes from the Santa Fe Institute, you're affiliated with, maybe even you were involved in. I, I've, I associate this idea with Stuart Kauffman, who said creativity happens on the edge of chaos, is the way they describe this sort of thesis. And it worked very well for my sort of analysis of why did personal computing and the predecessor to the internet happen in, you know, around Stanford when they happened. It was a very chaotic time. It sort of fit their model. Yeah. I was wondering if you're familiar with that and if If you know more about it, then—
W. Brian Arhur: Well, I know plenty about the edge of chaos. I've lived in Santa Fe, or been at the institute there. But no, it's usually when there are times when it's necessary to solve a lot of technical problems, you tend to get a lot of answers. For example, if there's a really— horrible outbreak of Ebola, that's a major technical medical problem. And you will see a lot of inventive or innovative work on the way to trying to find a vaccine. If we go into a war, as we did in the Second World War, then you have to solve certain problems. How do you land high-speed aircraft on an aircraft carrier? How do you detect enemy bombers coming towards you? These are all problems, and they were solved technically. So you tend to get an outburst of technological invention or innovation at times when society faces particular problems.
John Markoff: I wanted to ask about the evolution of your own interests. This may be my interpretation, but you seem like you've reinvented yourself in a sense multiple times. I've just sort of watched your interests. I mean, can I describe it as a random walk following what you're interested in? I mean, you've been an engineer, you've been a mathematician, you've been an economist, you've been a biologist, you've been a demographer. Maybe I'm missing some. Who are you today? Did I say economist?
Narrator: Which one are we talking to right now?
W. Brian Arhur: Yeah, I have to admit that 5 or 10 years ago I worried about this. I thought, oh my God, you know, I'm getting older. I was looking back on my career, and about every 7 or 8 years I just didn't change the problems I was working on, I changed fields. I was a demographer, it's true. I was an expert. What I taught here at Stanford was childbearing in developing countries to human biology students. I started to get really worried that I was a dilettante, you know, kind of like a butterfly going from here to there to somewhere else. I thought, am I going to finish up a career and think, well, I've flitted around from one field to another and I haven't really gone anywhere deep? But a while back it occurred to me that all of these problems actually were instances of something that was the same. All the problems I've ever looked at had to do with systems changing over time, things evolving, how things form.
John Markoff: Is that a general theory of— I mean, is there not relativity, but is there a general theory here that you're working toward in a sense?
W. Brian Arhur: Not so much a general theory, just I would call it a general fascination. How does technology evolve? How do population age distributions change and form over time? These were various expertises I picked up, but I was always fascinated by things coming into being or things changing in structure, not just growing more of this and that. In economics, we have a tendency to say this goes up, this goes down, more quantities, lower prices, all this sort of stuff. Stuff. But I was fascinated by something different. How does the economy change in character over time? How does it change in structure? So I'm interested— if there is a common denominator, and I'm sure there is in my work, it's how do things structurally change or change in character over time. We now have a whole subject that studies this. It's called complexity. So I discovered quite late in my career that I was doing complexity studies. Of course, I'd heard the word and I'd made some associations. But quite early on in the Santa Fe Institute, it still hadn't really got started. In 1987, I was brought to Santa Fe and there was a whole institute studying these same sets of problems.
Narrator: Does complexity studies have predictive capability or do you— do you attempt to, or is it more descriptive of—
W. Brian Arhur: I think that's a really good question. I think so. Complexity is really about looking at how patterns form and patterns change. You put all these little elements together, cars, say, on a freeway, and the pattern they produce together is called traffic. Sure, you can predict that at certain densities you will get traffic jams happening. You could probably predict how often and so on. So yeah, if you're designing the system, there are— it's good to be able to see certain phenomena are going to happen. And this is used in physics, it's used in often in branches of mathematics that you can see the sorts of patterns you're going to get. It's still early days in economics, but we're starting to understand in markets how you get bubbles and crashes and how you get— let me just give you one example here. One puzzle in economics has been that in a stock market there are periods of extreme volatility and then periods of quiescence for no particular reason. You know, you can't blame anything, but we figured out in the Santa Fe Institute that you could model such markets using— we created a market in the computer and had these little artificially intelligent investors trade among themselves. We discovered that very often not that much would happen, things would just go along, but these little guys were able to learn. John Holland methods so they could get smarter. Occasionally, one of the investors would discover something that worked better. It worked better, they started to make a lot of money, they would buy in because they had more confidence that they could use this, and that would subtly change the market. It could be 2 or 3 of them discovered the same thing, but that changed the market and changed the mechanics of the market, so everybody else had to shift. And so what we were seeing— and this was early days, 30 years ago— we were seeing what you call little cascades or avalanches of change ripple through the system. It turns out things like that are not predictable. And this is more subtle. You can't predict when the next one will arise. You don't have enough knowledge to do that. But what you can predict is that that will happen. You get periods of volatility and periods of quiescence. And just like you can predict that traffic jams will happen, but exactly when you can't predict. So, this is general in complexity. There are many phenomena that you can predict. And again, I thought perhaps this was all somewhat artificial and cooked up in Santa Fe, but I began to read heliophysics. The physics of the sun. And in the physics of the sun, you can show that, you know, with nuclear fusion happening and with the effects of gravity and the effects of electromagnetic effects, you get plasma ejections and you get magnetic loops that could be the size of Jupiter. These things are huge. You can't predict when those are going to happen. It's a complex system. But you can predict that such things will happen and what their frequency might be and what their size might be. So this is general in a complex system.
John Markoff: Something happened right before the holiday here at CASBS that I wanted to ask you about because it was so striking to me. You got a call from the Swedish news agency saying that I guess the right way to interpret it, you were on the shortlist for a Nobel Prize. And it just was, for me, it was an excruciating couple of days. I don't know what it must have been like for you to have to go through that. I mean, it just seemed like, I mean, what was it like to experience that?
W. Brian Arhur: What had happened shortly before that was that there is some scientific institute had creates a list, I think every year, of people. They're called Citation Laureates. Whose really rigorously figured-out citations are equivalent to people who get Nobel Prizes. Somehow I'd been nominated.
John Markoff: You were one of the—
W. Brian Arhur: I was one of those. Then I don't know what happened in Sweden. I wouldn't even conjecture, but I started to get calls or emails asking where would I be at 3 AM and what is my phone number and would I mind being called in the middle of the night. I had a couple of attitudes to this which now I see as amusing. First of all, I was delighted that this happened. Then, of course, I thought it's highly unlikely. The reason it's unlikely is that the kinds of approaches I've brought to economics and ways of studying the economy are not yet central in economics. They may be in the future. So I thought, "Hmm, if I get a Nobel Prize, it's going to be a bit premature." So I I thought it wasn't super likely. Truth be told, I went to bed that night, fell asleep, but I kept my cell phone on and next to the bed. Woke up at whatever, 6:00 in the morning, and I think they'd already given the prize out. My overall comment was, For about a week and a half, it was really fun to feel I had a horse in the race. And I thought, I'm going to extract every little bit of pleasure out of watching this horse. But I wasn't expecting it to win.
Narrator: How would you describe to someone who doesn't know any better, what were you nominated for?
W. Brian Arhur: Well, I certainly know I've been nominated several times. I think the contribution would be to regard the economy as an evolving complex system. So, rather than see an economy as something like a structure with all the forces in balance, which is the standard way, seeing the economy as in equilibrium and then something happens, oil prices go up and that equilibrium gets shifted. My view of the economy, not surprisingly, was to see it more like an ecology. This comes in, that changes, something gets driven out, other things happen, and so the economy is always forming and reforming. So, along with the team in Santa Fe roughly 30 years ago, we figured out a way of looking at the economy as an evolving complex system, not an equilibrium, an economy that looked much more like an ecology. And that's catching on. But if I were to make bets, the question would be, am I going to die first or is the Nobel Prize going to arrive? I'd bet the first one.
John Markoff: I heard from Mike Katani that you were quite excited by a talk that Wendy Cho gave a couple of weeks ago, which I missed. And I'm just wondering if you could describe what she was focusing on and what sort of caught your attention?
W. Brian Arhur: I very much like studies that combine a practical problem, a real problem, with some real ideas. These days I think very much in algorithmic or computational terms. When Wendy Chou gave the talk, it struck me the problem was extremely important that she was looking at. She was looking at gerrymandering in districts in the United States, and in some states it's quite extreme. So then the problem comes up: if you followed all the rules properly, or even guidelines properly, are there ways to redistrict where you could say things were operating normally and hadn't been gerrymandered. She used some very sophisticated statistical methods. She was basically looking at all the possible combinations you could make of districts. This turned into a huge problem that had to be solved by supercomputers if they were to meet the conditions. But the bottom line excited me. The bottom line, the outcome of her work, which I really did think was very good, was that politicians had been saying, "Hey, look, it might be gerrymandered, but you know, you show me something better." She was able to show something better with an awful lot of computation. The problem then came, how do you explain that to judges? How do you explain it to the Supreme Court?
John Markoff: The harder problem, right?
W. Brian Arhur: That's the harder problem. I don't think she cracked that one yet. But it seemed to me CASBS, where I am, I think one thing it does very well is have smart people looking at real-world problems but taking a fresh look at these problems theoretically. These are not at all the questions I would have thought, but keep going.
John Markoff: You wrote a piece a couple of years ago for McKinsey that your argument was sort of to redefine the way we look at Economies from a prism of production to a model based on distribution.
W. Brian Arhur: Yeah.
John Markoff: And I, I guess I want to start by asking, what was the backstory, um, behind that article? What brought you to that perspective?
W. Brian Arhur: There's an article in 2011 where I pointed out that there were now two economies, a virtual, or what I'd now call an autonomous economy. I call it the second economy. Which is digital, and then the physical economy. And we're seeing more and more of this. A driverless car isn't just a driverless car, it's an enormous amount of computation, invisible and digital and autonomous, meaning it's not really controlled by human beings. So what I'm seeing more and more of is not just jobs, but tasks in the economy being handed off into an autonomous digital economy. I think quite shortly, meaning in the next 5 to 10 years, we're going to have autonomous air traffic control. So, tasks are disappearing into the autonomous economy. I started to wonder what AI was really about, what artificial intelligence was really about. And sure, it seems to be all these miracle things, but I realized that what we have been able to do as human beings was to enable digital machines to do association. I began to realize that for human beings and the economy, this was a bit of a landmark. We'd arrived at a suddenly a new era. You know, it would be almost cliché to say we're not in Kansas anymore. What was new in the economy all of a sudden wasn't that we could have better agricultural production or better factories or have the internet. Suddenly, as human beings in the economy, we were able to make associations. In general, economists are worried as to whether this is creating a new scenario that will throw people out of work, and that's been debated for a long time, 15, 20 years, and in fact further back than that. And they fall into two camps: that we're probably going to be okay, humans always invent something new.
John Markoff: So this is what you refer to as the Keynes point. Yes, Keynes had this article in the '30s.
W. Brian Arhur: Thank you. Yeah, around about 1930, actually, John Maynard Keynes worried about the same problem. This is almost 100 years ago, and robots were new at the time. The whole idea of robots and automation was pretty new, and Keynes foresaw an economy around about this time, in the time of his grandchildren, where we could produce, or machines or robots, something could produce all we need. And he was pretty much right on the money about that. But the problem then would be what he called technological unemployment. There mightn't be enough jobs. And if there is such a thing happening, I believe there is, I also proposed in that article a couple of years ago that we've reached that point. I call it the Keynes point, the point where production ceases to be the main problem in society—getting more goods and services. Of course, that's always a problem, but the emphasis had shifted over to If there weren't enough jobs to go around, if there's pressure on jobs, how do you get access? So, plenty of fodder in the piggy trough, but how do the piggies get to the trough? And we've solved that one for the last few hundred years by having jobs. So, we have jobs, we're paid for the jobs. Now I think things are becoming dicier. There's huge swathes in Europe or in the US where jobs are tight or non-existent, and there's been plenty of suffering in that regard. And many people have been thrown into the gig economy where they don't have possibly an official job, but they can get into this or into that temporarily and hoping to keep their head above water. What I think is really happening is that we are moving into an economy where the chief issue is not producing more, the chief issue is how do you get access to what's already possible. And I began to realize that it had several implications. One of the implications— I'll mention a couple of the implications— but it means that everything I see now, an awful lot of political proposals are much more than before are being assessed on the basis of jobs. A couple of weeks ago we had this USMCA, the new version of US-Mexico-Canada trade relations, and I noticed that one of the main things that had been changed was it brings back automobile production jobs to the US. So, 20 years ago we were interested in efficiency, so we basically said, "Okay, let's have our cars produced in Mexico and our computers produced in China, and we'll benefit from that. That's efficient. Production gets done where it's cheapest." Now we're saying something different, and I can see the beginnings of this already. We're not worried about cheap production because we're not that badly off. We're interested in making sure everybody has access to what's already produced. So we're bringing back automobile jobs into the US. And in fact, I think this is politics that is running all through Europe, all developed countries, and the US, of course. And the rather strange politics we have at the moment are a symptom of this, not a cause of that. Similarly, there's been a huge movement since the 1980s saying that what really counted in the economy was efficiency. So if it's more efficient to get your clothes made or shoes made in China That's where we're going to have them made. Now I think things have changed and that sort of— the whole school of saying what really counts is economic growth and so-called efficiency will not be able to get traction as easily as it did. 30, 40 years ago. Anything that gets traction now is going to be how do human beings get access to the economy? So, if you like, the big issue in the economy, I think, has shifted from asking how do we get more goods and services produced in our economy to saying we're doing fine for goods and services, how do we get access to those? Whether it's healthcare or whether it's automobile production, the question is how do we get access?
John Markoff: Let me go in two directions from that. One is, if I could ask you to put your demographer's hat on for a second. My view of the Keynes point question was reframed when I spoke with Danny Kahneman at one point. I was clearly in your camp and I was arguing that automation was going to come to China. And it would lead to disruption because of jobs. He stopped me, he said, "You don't get it." He said, "In China, they'll be lucky if the robots come just in time." And I said, "Well, what do you mean?" And then he pointed to the falling birth rate in China. And then when I looked at that, I realized it was an important factor that not only in some cases will they not have enough workers, but they have this problem of the aging population and how do you care for the aging problem. And the robots might not come in time to care for certainly our generation. But maybe the next generation. But I mean, for you, how does demography factor into the question of—
W. Brian Arhur: Oh, I think you're right on target there. I think it was last week Russia announced that it needed more births, more children. They need to get population up. But I think that's an old-fashioned point of view. I think what's happening is that as jobs fall away, populations all over the developing world are aging out of the labor markets. They're living longer, and there are going to be— let me back up here. It may be a piece of really good luck that as jobs fall away, the number of young people seeking jobs here and there, whether it's in Italy or in Russia or in China, is lessening, so fewer young people coming along, but fewer jobs available for them. And with a bit of luck and a bit of good timing, that could be very good. It could be that just as fewer people are coming into the labor market as humans, non-humans are there to fill in. That would be a wonderful scenario if it works out that way, but it very rarely does, right? I've got my fingers crossed. I wouldn't bank on it, but I do think that overall, if you take a longer point of view, that in 100 years or so, if we're still doing well, there will be fewer people, or fewer births being born, but we won't have to be looking for jobs for them.
John Markoff: Yeah. So on the other side of the equation, you point at this distribution problem, and —There's lots of evidence that in the developing world there is this increasing problem of the concentration of wealth, and that's a distribution problem, basically. Do you see a path to solving that?
W. Brian Arhur: —Well, I do, but whether there are many paths to solve that, whether they're politically feasible, I'm not so sure. I think what happened to a very large degree was that since the 1970s and '80s Certainly into the '90s, manufacturing economy and the service economy were getting more and more efficient, and jobs were being done by machines. It turned out that labor, blue-collar workers and many white-collar workers, were not as needed as they were before. And therefore they didn't have political power, union power, as much as they did before. And then in the 1990s, globalization came along. So we started to get satellite communications and fiber optics, and that meant for the first time we could have supply chains in China, or in principle anywhere, in Myanmar or somewhere like that. Like that. And that meant that jobs were getting shipped off. Now something different's happened. Suppose we bring all those jobs back, which politicians are suggesting, if we could wave a wand and bring them back. I think they'll be shipped off into a different country, and that is the autonomous digital country.
John Markoff: Yeah, that's, that's interesting. Yeah, so they won't come back. Well, they'll They'll be where it's appropriate for production, but which could be anywhere in the world.
W. Brian Arhur: Yeah, to make that more concrete, we can take a lot of automobile manufacturing jobs back from Mexico, put them in the US, but maybe they'll just be done by digital machinery then. But my hat is off really to Keynes. Keynes had, I think, a very accurate take on what we're going through right now. There's a strange thing— this is a bit like climate change— that people sometimes say to me, "Hey, you know, we're in an economy in the US where unemployment is very low. There tend to be an awful lot of jobs. So what on earth are you talking about, that there'd be a shortage of jobs?" I don't know. I think what's happening is that not so much the jobs are disappearing as that whole industries are being redefined. So it may be that the healthcare industry gets redefined into many of us having automatic monitors, and that's monitored, say, by algorithms, and we visit certain labs. I mean, this is a sci-fi scenario, but I think it's more than possible within 50 or more years. And you You could say, well, where are all the jobs? It's not quite that. That's not the right question. The question is how are we going to organize ourselves that human beings are still necessary.
John Markoff: A final point based on your McKinsey article and this shift from production to distribution. Do you think capitalism as an economic system is flexible enough to survive the shift you're describing?
W. Brian Arhur: Yeah, I want to speak as an economist here, not as a— private person. But yes, I think that capitalism, if you— that has to do with ownership of productive means, all this sort of thing— is about as good a system as we've managed to come up with over a few hundred years, and certainly where it's presence, such as South Korea. You can see the results vis-à-vis North Korea, West Germany versus East Germany. So that system works. What doesn't work is having such a system provide for people where people feel it's equitable. But that does exist in Scandinavian countries. What I would like to see is a sense of a sense of community arising and a sense of social responsibility. So it's not just, what do I get versus what do they get? It's more a sense that, well, can't we together create a system where we look after— collectively we look after our children in kindergartens, where collectively we have some form of healthcare? I shared an office with Cormac McCarthy for many years at Santa Fe Institute. Institute. We discussed this quite a bit, but there was always a theme to Cormack's work, and that was that in the time when he was young and growing up, the 1930s and '40s, there was a very strong sense of moral direction in the U.S. As Cormack would put it, we knew who we were. And I would love to see or study how that could be brought back. That's not an academic question. It's not even quite a cultural question. It's a 'what are we here for' question. And I think it's more important, it's the biggest question whether you're in China or whether you're in America or in Africa, what are we here for? We've lost the main religions to quite a degree, certainly in Europe, but what are we putting in its place? You know, 120 years ago in England you would have said, well, it's empire, isn't it? You know, the time of Kipling and people like that. So I'm not suggesting we put silly things in place. But if we are agnostic as to what society is all about or what we are about as people, then I think we just keep getting into these— well, you know, I can get more income than you can get, or I can elbow you out of the way, or whatever. So I would say a sense of moral purpose. So I think that I would like to see a system, at least in the West, where we're saying it's not just what's in it for me, which came in heavily in the 1980s but was already there, but what's in it for us and how can we help each other and help the planet. I don't think that's an overstretch. There are countries that are run much more along those lines. So I think capitalism, capitalism, yes. But shared responsibilities and a shared sense of community would make this pleasanter and more effective for everybody.
John Markoff: Well, thank you for taking the time with us.
W. Brian Arhur: And thank you indeed for having me.
Narrator: That was W. Brian Arthur in conversation with John Markoff. Be sure to check out the notes for this episode. We've got links to Brian's homepage at the Santa Fe Institute and some of the pieces he's discussed in this episode. Human-Centered is a show from the Center for Advanced Study in the Behavioral Sciences at Stanford University. To learn more about the center and the rich work of current and former fellows, head over to our website, casbs.stanford.edu, and check us out at Twitter @CASBSStanford. Until next time, thanks for listening.