COMPLEXITY: Physics of Life

W. Brian Arthur (Part 1) on The History of Complexity Economics

Episode Notes

From its beginnings as a discipline nearly 150 years ago, economics rested on assumptions that don’t hold up when studied in the present day. The notion that our economic systems are in equilibrium, that they’re made of actors making simple rational and self-interested decisions with perfect knowledge of society— these ideas prove about as useful in the Information Age as Newton’s laws of motion are to quantum physicists. A novel paradigm for economics, borrowing insights from ecology and evolutionary biology, started to emerge at SFI in the late 1980s — one that treats our markets and technologies as systems out of balance, serving metabolic forces, made of agents with imperfect information and acting on fundamental uncertainty. This new complexity economics uses new tools and data sets to shed light on puzzles standard economics couldn’t answer — like why the economy grows, how sudden and cascading crashes happen, why some companies and cities lock in permanent competitive advantages, and how technology evolves. And complexity economics offers insights back to biology, providing a new lens through which to understand the vastly intricate exchanges on which human life depends.

This week’s guest is W. Brian Arthur, External Professor at the Santa Fe Institute, Fellow at the Center for Advanced Study in the Behavioral Sciences at Stanford, and Visiting Researcher at Xerox PARC.  In this first part of a two-episode conversation, we discuss the heady early days when complex systems science took on economics, and how biology provided a new paradigm for understanding our financial and technological systems.  Tune in next week for part two...

If you enjoy this podcast, please help us reach a wider audience by leaving a five-star review at Apple Podcasts, or by sharing the show on social media. Thank you for listening!

Visit our website for more information or to support our science and communication efforts.

Join our Facebook discussion group to meet like minds and talk about each episode.

Podcast Theme Music by Mitch Mignano.

Follow us on social media:
TwitterYouTubeFacebookInstagramLinkedIn

For more information:

Brian’s Website.

Brian’s Google Scholar page.

Where is technology taking the economy?” in McKinsey, 2017.

The Nature of Technology: What It Is and How It Evolves.

“Punctuated equilibria: the tempo and mode of evolution reconsidered” by Gould & Eldredge.

Episode Transcription

Michael: So Brian, it's a pleasure to meet you here amidst the complexity.

Brian: Thank you. Delighted to be here.

Michael: I think I'd like to take this conversation in three parts. One, kind of looking back at the history of the development of complexity economics and the argument for it that you've put forward in a book and in numerous papers. And then to dig into the actual mechanisms involved and to explore some of the ideas that you get into in The Nature of Technology.

Michael: And then you have a 2017 article for McKinsey that I thought was really fascinating in terms of looking forward into the shape that the new economics systems are taking. So if that sounds good to you…

Brian: It sounds great.

Michael: Awesome. So you were there at the beginning of the Santa Fe Institute's articulation of complexity economics. And I'm curious, what brought that together in the first place, and what got you involved and what you saw, you and the other people involved saw as the need that you were addressing at the time.

Brian: Right. In 1987, there was a famous meeting held here at the Santa Fe Institute in September, and it was decided that about 10 scientists and 10 economists would be brought together by Phil Anderson, who's a Nobel prize winning physicist, David Pines, very eminent physicist, and Kenneth Arrow, Nobel economist, and they brought 10 of us together. The science group included luminaries like John Holland, very famous these days, David Ruelle mathematician, and Stu Kauffman and others. And on the economics side you had Larry Summers who went on to be president of Harvard, Tom Sargent, who in the future would get a Nobel in economics, and others including myself.

This was financed by John Reed of Citibank. There's a general feeling with John Reed and others that economics somehow hadn't properly delivered. They were heavily involved, Citibank was heavily involved in South America, made a lot of sovereign loans to the tune of many billions of dollars. And the governments in some of those countries had reneged on the loans. John Reed thought there must be a better way to do economics because economists had assured him all of this would be okay.

So we met here in Santa Fe and there was great interest for 10 days in that meeting about looking at the economy quite vaguely, not as an equilibrium perfect machine humming over and needing a few bits of fine tuning, but rather as an evolving complex system. None of us quite knew what that meant, but we were trying to figure that one out.

Michael: So I remember, you spoke at the annual symposium this weekend and you mentioned that there was a period where all of you were kind of fishing around for the new framework, for the new lens through which to look at this.

Brian: Yeah. Well a year later after that incredible meetings in Santa Fe, the Institute here decided they would form a program funded by Citibank on the economies involved in a complex system. I was brought here from Stanford to run the program. And in due course we hired some people. We got really good people here, first rate people because of Arrow and Anderson being able to lure them in. So we decided we were going to look at the economies involved in complex system, but we were sitting around the kitchen in the old Santa Fe Institute on Canyon Road and we didn't quite know what that would involve. We thought vaguely some people thought we should study chaos theory. This was 30 years ago now, 1988. Some people thought we should study high tech and network effects, which I'd studied a lot about. And other people in other ideas. And we didn't know quite what to do.

I should add at this point that standard economics looks at the world, as I said, as being quite mechanistic and sees a lot of forces as being in balance or in equilibrium. And in particular if you look at producers, firms and consumers and prices, they see the world as adjusting so the prices that appear in the economy are enough that no consumer, given what they're earning, can do any better with their budget, and no firm can do any better with its means of production. The whole economy's a bit like the spider's web where every little thread is keeping the other threads in equilibrium.

That was the point of view in economics that had come about in the 1870s, 1880s. Some of it borrowed from the physics of that day. And all through the 1900s that was the dominant way of looking at the economy. And it's been extremely successful. It's fashionable to complain or deride economists, but actually economists have been able to understand things like the macroeconomy pretty well, central banking, monopolies, how to set up auctions and markets that were properly trading agreements. These are all extremely well understood in economics, thanks to the theory that was built upon the standard approach.

What we wanted to do was not certain, we sat around and debated for a time and there were a couple of things happened at that stage. Stuart Kauffman, who's a theoretical biologist, was sitting there and he said to me, he said to our group, he says "There's something I don't understand. Why do you guys in economics do everything at equilibrium?" And I just said, "Well, because we do Stuart, so shut up." "No, no, why do you?" And he persisted. And finally I and a couple of the others who were trained economists said, "It's a good question, Stuart. Let's think about that and take it in." None of us could see how to do that. But I thought that was worth kind of noting on the blackboard as a direction that we could take.

In the meantime, because I wasn't sure I was directing the group, I wasn't quite sure what our thing would be for the year or two years hence. So I called Arrow and Anderson and I said, "Hey look we're happy to do something here. Do you want to give us a bit of guidance?" They were godfathering the program. So they called John Reed in New York and the word came back from Reed, "Do anything you want providing it is not conventional and providing it set the foundations of the field of economics."

And Arrow phoned me. There were no emails, and Arrow phoned me in Santa Fe with that piece of advice. I was floored. I didn't know what to say. I couldn't believe it. Later, a bit later, like a day or two later, I thought, "Oh my God, this is like 1520 in Wittenberg or somewhere, that the Vatican has phoned up the Augustinians and Martin Luther and said, "Hey you guys, I hear you want to redo theology, go ahead. But providing it's not conventional and totally to the fundamental field." It was like getting instructions from God, "Do anything you want."

And because Santa Fe Institute was isolated and it has no departments and no economics department, we could do this as a kind of skunkworks, a little private group without anybody complaining or wagging fingers, "You can't do that. You can't do that." So that's when we started to look at the economy as an evolving complex system.

Michael: So there's two dimensions that you discuss in “Complexity and the Economy” and I'll be sure to share the essay from that on complexity economics in the show notes here. You talk about endogenously generated non-equilibrium and how there's two factors that really feature here. One being fundamental uncertainty and the other being technological innovation. That the addition of these two factors are a major piece of this new way of seeing. I'd love to hear you dive into that a little.

Brian: Sure. So there we were in the old convent probably half a dozen of us in economics, maybe an equal number probability theorists and mathematicians and one theoretical biologist, etc. And we decided we'd start to look at the economy as a non-equilibrium system, but that brought its own difficulties.

First of all, there was a very strong feeling in economics that if the economy wasn't in equilibrium and be a bit like a ship teetering over, but it would soon right itself and be back in equilibrium. So really non-equilibrium was rare and it was an exception and maybe wasn't worth looking into, etc. Where would non-equilibrium come from in the first place?

And I began to realize that if you're looking at economics the standard way, we tend to assume that all the little players in the puzzle we're looking at in any part of the economy, they're all identical by assumption, and they all have a well defined economic problem they're trying to solve, and they're all perfectly rational, which means they can do an infinite amount of logic and mathematics on their own case and solve that and they'll solve it in such a way that we can call a solution the outcome of what they're all doing together.

We call it a solution of that outcome. Doesn't ask for further changes. But that assumes that other people are like you, that you know what problem you're in, that you can do all this mysterious logic that might have taken an economist six months to figure out and the people in the little model you're working on can do this instantaneously, etc. And we began to realize — we’re not the first to realize this, but we began to realize that any realistic economic problem where you've said four or five other players, like some startup in Silicon Valley, you don't quite know who the other players are going to be at the outset.

You don't know what the game's going to be. Suppose you're starting up and you're in the field of autonomous trucking. You're going to have driverless trucks and convoys going across the country and you might be Google or you might be Volvo or someone financing this.

You don't know how the technology is going to work. You don't know who the other players will be. You don't know what the regulations are going to be or the legalities or the insurance background or how other people on the road are going to accept this.

So you're in a whole cloud, not just of not quite knowing probabilistically. You completely don't know. There's a large area you don't know and it's a fundamental unknowing. Economists call that fundamental uncertainty. And so you kind of assume an equilibrium in a case like that because people are trying to make sense of the situation and may or may not be in equilibrium at the outset.

And there's another big source of uncertainty. You don't know what technology is going to come down the pike in the next year or two or in the next five years, whatever horizon you're in. Fields are very rapidly changing. Commercial banking's changing at the moment, or retail banking because we now have blockchain, and we have Bitcoin, and we've all sorts of other things, artificial intelligence coming in. These things are coming in rapidly and they're disrupting and changing what might otherwise be a business that's relatively at equilibrium.

So non-equilibrium we began to realize was standard. It's the standard. Occasionally the sea goes flat, the waves vanish and there's plain sailing. But we began to realize that that was the exception rather than the rule.

Michael: Something I really appreciated in this is, I think we were talking about this right before we started recording, that this is symbolic of a larger trend in the sciences. A movement away from this sort of idealized deductive rationality and into a sort of more humble bounded rationality or inductive reasoning. And you made a great point in your talk the other night that this isn't just how we make these kind of like business decisions. This is how people decide to get married. This this is the reality of our mundane daily existence.

Brian: Yeah. We got to the point where we realized that non-equilibrium wasn't just an exception. That was the norm. It's like saying the oceans normally have waves and they're not in a perfect calm condition. Non-equilibrium was the norm. But the problem is how do you do economics when there's no equilibrium expected and the questions you're looking at are ill-defined? If you don't know what people are going to do, the problem you're trying to look at isn't well-defined. And that means rational behavior in that problem is not well defined. You can't have a logical solution to a problem that isn't well-defined.

And so we had to sort of throw overboard all the standard assumptions in economics that there are logical, rational agents, that problems were well-defined, that everything was in equilibrium. What could we or should we put in its place?

It was there that I began to realize that we had something we could put in its place and for me this was a real breakthrough. As part of our group, we had a really extraordinary scientist called John Holland who had worked for many years at the University of Michigan. As it turns out, John I believe was the first PhD in the United States in computer science. He'd been trained at MIT. And John's expertise was looking at how ill-defined systems could be such that some computer program like playing checkers or something could learn in such a system how to make sense of it and how to move ahead and how to improve.

It began to be clear to me that also that in our standard world ill-defined problems were the norm. They weren't the exception either. We get married and it's quite ill-defined who on Earth we're moving in with and how well that's going to operate. We have children. We don't know what that's about until we start to do that. We take up jobs. We're not quite sure how that will operate. Yet we do all these things and we do it happily and routinely.

What John was pointing out, John Holland was pointing out to us was that you could model human beings not as rational beings, but as people who could start to make sense in the problem they're sitting in maybe by analogy, maybe by having hypotheses that they'd brought in from the past. I think it might be a bit like this. I was in Japan before. Now I'm in Korea. Maybe it operates similarly, maybe not, and they would learn.

They would have hypothesis and have their own takes on how things were operating and they might change their hypothesis or the model they were using. They could be quite complicated models in their mind or they could be rather simple ones. They could come from behavioral economics. If we wanted to look for how people really reason. They could come from knowledge of the past. In the past I've faced similar problems. I'm going to try and go at it this way.

So John Holland was showing us how we could make mathematical models, rigorous models where the little players in any given situation didn't quite know what on Earth they were doing but were mutually learning together. As they learned, they acted. As they acted, the situation would change. As the situation changed, they would relearn and update their ideas of what they were doing. Maybe that would converge to an equilibrium. Maybe it wouldn't.

One of the things I learned from John Holland was I remember saying to him one night over a beer I said, "Don't grandmasters know a lot more about chess, and so they understand much more about the game than before?" But I said, "It's still an equilibrium. They're just better players then there were a century ago because there's been more experience with chess."

John said no. He said it's not just that there's been more experience with chess. The game has shifted. So the level of play has shifted. It's got much better. There's a frontier where the game is changing and people at the master level are learning and pushing out that frontier all the time so that if someone from a hundred years ago showed up, they wouldn't be able to compete anywhere near as well as they could a hundred years ago.

This shocked me. It meant that these problems, the problems themselves might be not that well-defined. It's not well-defined how I should move in chess because I don't know you very well. I don't know how you think very well. I'm trying to find out as I go along. I'm studying your past games maybe. But it's not that well known. But what John was pointing out was that the whole situation shifts as more is learned.

Michael: Yeah. So there's that “bringing a knife to a gunfight” kind of quality.

Brian: Exactly!

Michael: This is where it gets very interesting for me, because this dimension of having to anticipate or learn from the behaviors of other agents in the system. And you talk about how this means that the economy is a collective computation.

Brian: That's right.

Michael: And so there's a really interesting example that you bring up in this piece about Kristian Lindgren and the evolutionary game theory and running iterated simulations of the prisoner's dilemma. And I'd love to hear you go into that because I think that that's a really interesting place to leap from that and into technological systems as evolutionary ecological systems.

Brian: Okay. One of the examples that really interested me was by a physicist from Sweden called Kristian Lindgren. It was a little bit after the time I was talking about with John Holland. And Lindgren was looking at what we thought of in those days as a prisoner's dilemma tournament. You don't need to know much about prisoner's dilemma. Think of it as a hundred different strategies for playing this game called prisoner's dilemma are competing. And they're competing against each other for the first hundred moves and then somebody wins more than another, and so they've won that round.

What Lindgren had set up on his computer was the idea that strategies that won consistently could reproduce themselves, and so if one strategy was particularly good when playing randomly other strategies, including itself, it would reproduce and do pretty well, and others strategies that didn't do well would drop out. We'd have a sort of trap door and they'd be thrown out.

The interesting thing about what Kristian Lindgren did that fascinated me was he didn't just set this up as an automatic game automata or algorithms on the computer strategies can play. That had been done I think before. The kicker in his world was that the strategies every so often could mutate and get deeper. They could remember more moves back. And so you might be playing with strategies that just remembered one move back and then suddenly a strategy could discover how to play two moves back. And that would be obviously an advantage because they'd have more knowledge of what their opponent was doing and so on.

So suddenly these strategies began to mutate and deepen and have deeper memories of how to play that game as it was ongoing. Remember more moves back. And some of those strategies obviously started to take over.

What interested me in Lindgren's model, he ran this thing for, I don't know, maybe 60,000 tournaments or 60,000 gos. Must've taken in 1991 probably three days or two weeks on the computer. But Lindgren ran this thing and there'd be periods where there was clearly a best strategy and the other strategies started to disappear. But you'd see some of those other strategies staying in the game because they were also necessary. If the superior strategy just played itself, it didn't do that well. It needed some strategies almost as fodder. The wolves need some sheep to eat, otherwise they won't be wolves. So there's a balance there.

And then other times deeper strategies would be discovered. One or two of those might take off for a while and beat everything in sight. And then there'd be other periods where there's a gigantic free for all and you'd see for maybe a lengthy time of 1,000 or 50,000 tournaments, it would be like a situation where there was a lot of random strategies being created but nothing quite dominated. It looked quite chaotic. And then maybe an even deeper strategy would take over.

I remember looking at Lindgren's results and thinking this is what I would call paleontology or paleontological economics, meaning that there are long periods, perhaps like before the KT boundary, where you'd get all the dinosaurs and then suddenly something would happen. A smarter strategy would be discovered, might've been mammals or something, whatever. And the whole game would change and you'd be in a new eon where that would last for a while. But there was no equilibrium in this sort of game. Things just kept getting discovered and discovered.

And also it seemed to me that there were strategies that depended on other strategies. Some strategies were almost parasitic. They could stay a bit like the flu virus as long as there were hosts for them. Other strategies competed, other strategies collaborated, but none of this was built in. This all emerged from what Lindgren was showing.

And the bottom line in all of this was I realized that if you look at economics this way, this was a very well-defined game. It was well known in the literature. He was holding a tournament for strategies and watching how strategies changed and mutated over time. But the whole thing was not like any machine, it was biological, it was paleontological, it's for all the world reminded me of species competing, a new species being discovered, so to speak.

And for me, why Lindgren study so much was economics suddenly became biological. In fact, I've begun to realize that complexity economics, if you want to have a quick one liner, is standard economics views the economy essentially as a machine with all parts in balance. Complexity economics views the economy as an ecology with strategies or forecasts or actions competing to see which will dominate. But then maybe new things being discovered all the time. And so for me, suddenly economics became much more like modern ecology.

Michael: And it's funny looking at this over history, my background is in paleontology and you don't learn about dinosaurs without learning that there was this... Until very, very recently, at least in the West, we went through a similar kind of a paradigm shift where it was believed that all forms were created as they are. Everything was static, that things could shift. But ultimately there was a sort of equilibrium of species and their roles. And then right before the discovery of the dinosaur, you get this catastrophe in the paradigm of like realizing that things have gone extinct.

Brian: Yeah.

Michael: And then it wasn't long after that, that we developed a theory of the origin of species. And so in a lot of ways it seems as though economics has been trailing biology and ecology in this movement only because perhaps the systems are so much more abstract or so much more complex that it's just taken us longer to see them.

Brian: Yeah, I think that economics was born actually, essentially as an intellectual pursuit around about the time of the enlightenment in the 1700s. And the enlightenment tended to view the world as being very highly ordered. You didn't always know the mechanisms, just like before Newton, you didn't quite know why the planets moved in elliptical orbits. But if you thought about it, you might get the theory just like Newton did and things were highly ordered, things were largely in equilibrium. All of that apparatus got brought into economics in the 1700s. Now we're realizing that it's not that simple and the economy is certainly always changing and never, probably, at rest.

I don't say there's something terribly wrong with the equilibrium way of thinking. My view is that it works very well indeed for a number of problems and it's given us a huge amount of insight. But then I would say, "Okay, what about the rest of the economy?" Economic development, for example, is not by definition is not an equilibrium system. Structural change, that's not an equilibrium. Technological change, never an equilibrium, etc. So there are major parts that the theory couldn't look at if it insisted on equilibrium. And that's what I find is happening now.

There quite a while ago, Los Alamos, maybe during the war, mathematicians were arguing over linear systems with linear types of mathematics and nonlinear systems, which were somewhat unfamiliar. And it was Stan Ulam who pointed out very famously. Somebody said, "Well, we know an awful lot about linear systems, that's huge. These nonlinear systems, these are exceptions and they're not well studied and we don't know much about them. And with a bit of luck, that'll be a rather small set of things to think about." And Ulam said, "No." He said, "It's a bit like saying there's a theory of elephants and now we're going to have a theory of non elephants." So non-equilibrium is a theory of non-elephants and equilibrium, at least as far as I'm concerned, is very necessary. I wouldn't stop teaching it. It's central to economics, but it is a fairly constricted system of assisting things. Can only be studied at equilibrium.

Again, people in complexity are fond of pointing out that you can tell awful lot about butterflies if you chloroform them or whatever you need to do and nail them to a board. That's fine. We know all about their arms and legs and parts and wings, but if you really want to understand how they work, it's better to study them when they're not dead, when they're flying around. And you can study how they operate. So it's a widening of economics that I'm pointing out, it's not an either/or situation. As economists, and not just in Santa Fe, but in several other places…we’re starting to think, okay, we've had 150 years of equilibrium economics. What would it be like to have non equilibrium economics?

And I remember early on like 1999, 10 years after we started at Santa Fe, I was publishing a paper in Science on the economy and complexity. And I was asked by Science to give the approach a name and they insisted and I said, okay. This was on a telephone call. I said, "Okay, let's call it complexity economics." And I realized vaguely at the time that anything I said might lock in. And it did. Now, I don't know, it's debatable, but I sort of think maybe I should have called this non-equilibrium economics. And I wouldn't change now because there's quite a cohort of people who call this complexity economics and it fits well with complexity. So that's okay.

Michael: There's something that comes, looking at the 60,000 iterations of the prisoner's dilemma ecosystem in this. When you compare it to a paleozoological study, this looks exactly like around... Is it around the same time, maybe just a couple of years before that Gould and Eldredge published on punctuated equilibrium? You make this…

Brian: Evolution by jerks and creeps.

Michael: Yeah, jerks and creeps. Maybe it was a little too on the nose. You make a point later in this article that a lot of this has to do with, you're looking at this system at different scales and that adding the dynamical component, adding time gives you different rates of change and that system can appear to be an equilibrium certain times, can be in a kind of a complex flow at other times and then at other times becomes so difficult for the agents involved to predict it, that it's essentially chaotic.

Brian: Yes, that's correct.

Michael: And to just sit with the trinity there of these three systems and how they appear to be sort of attractors. I guess maybe that's like too recursive or meta, but that the system is moved from one phase into another phase and then back as endogenously generated complexity or external disturbances occur. But then at the same time… So there's that sense in which you find points of stability or change. But then over the course of that history, you're seeing the strategies of those individual players getting longer and longer memories. More and more capacity to model their environment. And there's this very interesting through-line to, what does a non-equilibrium economics tell us about the emergence of biological complexity?

Brian: Yeah.

Michael: And you talk about this quite a bit in your drawing of the analogy between biology and technology in your book.

Brian: Yeah.

Michael: So I'd love to use this as the opportunity where we explore, how you see technology itself as an evolutionary system.

Brian: Sure.

Michael: Yeah. And so just like for starters…where do you see technology start? How old is technology here? Because I mean, the more I read your book, the more it seemed like the differences were rather kind of arbitrary between biology and technology. We look at this as a multi-scale thing and like you get into extended phenotypes and you yourself in the nature of technology, extend the definition of technology to our ideas. That we're using to operate on the world. So I mean...

Brian: Well, let me dive in here.

Michael: Yeah. Maybe you should provide some context.

Brian: Yes, of course. What fascinated me. I'm trained as an engineer, I should say. And I've been fascinated just about all my life with technology. Funny enough, technology even more than science or mathematics, and I realized that with human beings, technology goes way back. You could argue that fire is a technology. It's something we developed or used for our own human purposes, say for cooking or keeping wild animals away or whatever we used it for. Even before that you could say humans’ linguistic abilities, human speech. Quite far later, the ability to write. These are all technologies. They're all means to human purposes and so you could go arbitrarily far back with human beings back to at least three million years and find uses of quite primitive technologies.

People looking at this quite vaguely for roughly since Darwin's book. Let me say a word or two about this. Darwin's book came out in 1859. It was read a couple of years later by Samuel Butler who had left England to get away from an overbearing father and actually Butler's grandfather had been the head of the school and had taught the young Darwin. I think in Salisbury. Butler was in New Zealand. Darwin's book arrived and it fired up Butler's imagination to ask could there be a theory of evolution for technology? He was thinking in terms of steam engines. This was the around 1861 and he started to ask that question. That question was never satisfactorily answered. Could there be a theory of evolution for technology? Could we see a line of descent, not just say for sailing ships from rather primitive dugout canoes, but for all of technologies. So that there might be a tree of speciation or of ancestry for all of technology, the way Darwin had demonstrated this for biological species.

People had to go at this on and off, but largely things didn't quite work out. They assumed that evolution meant the same as Darwin's mechanism. So you could say, well, if you vary railway locomotives, you will get newer locomotives. If you vary helicopters, you'll get more modern designs and helicopters descend from previous helicopters and things like that, but it wasn't very satisfactory because there were new technologies coming along like radar or whatever. Like the polymerase chain reaction in molecular biology. These things come along and radar didn't come out of variations of radio circuits. You could vary radio circuits til you’re blue in the face, you'll never get radar. Radar is a different principle.

Similarly, the jet engine didn't come out of variations of air piston engines in the late 1920s. It's a different principle. So when I was thinking about technology, was there a theory of evolution for technology? It was pretty clear the Darwin's mechanism didn't work, but I began to realize that novel technologies, radically new ones like the jet engine or radar, come about usually from some human problem. How are we going to detect enemy aircraft, metal clad aircraft that are strong and carry a heavy payload of bombs coming from the continent to England if there's a future war?

And so I began gotten to realize that novel technologies come along as combinations of previous technologies. Radar is a combination of something that can generate high-frequency radio waves. Something that can detect high frequency waves and possibly screens or cathode-ray tubes to show you a picture of what's happening. There's a lot of parts to radar, and those parts, somebody realizes, or maybe some group of people over several years or maybe over several months, realizes that there might be a principle, maybe we can bounce high frequency waves off an aircraft 30 miles away and maybe we can detect the faint echo of those radio waves. That was known as a phenomenon as far back as the early 1900s.

What wasn't known is, could we use this to detect enemy aircraft? And so people began to realize all we need to do is generate high frequency waves, point it in the direction of the aircraft, switch it off for an instant, which is not easy to do, so any echo comeback is not drowned out by the outgoing signal, detect the new signal, process it and etc. So I began to realize that all new technologies are combinations of technologies that had gone before and when I realized that I started to look in the literature and I found that wasn't a new thing to say. That people in the 1930s had said novel technologies are combinations of prior art. In fact, this goes back to somebody in the 1880s a guy called Thurston who had written about steam engines and pointed out that all parts of steam engines were known and used in previous technologies and that new technologies were combinations.

So when I thought of that, and I wasn't quite thinking of a book then, but I was thinking you could have the theory of evolution for technology that rested on the idea, not a variation but of combination. And then there'd be lots of combinations either put together mentally or put together physically or both. And many of them would be quite useless, but some of them would be for repeating use. And those combinations, like a jet engine or a radar device, could become new building blocks for yet further technologies.

So in that sense, if you take all of technology together, the whole collection of technologies, millions of different technologies, some of which are not used anymore, many of which are, they're all available for further combination, for further technologies. It turns out this isn't a terribly deep insight but nor was Darwin's, but it was crucial. It's a bit like saying technology works a bit like a very strange Lego set. You make combinations and if they happen to repeat and be extremely useful, you heat those combinations, fuse them together and throw them back in the Lego set for yet further combination.

And this was very unfamiliar to say the least. Although, people had thought of something like this before and what I did in that book, The Nature of Technology was work out the details of how such combination works. I had to figure out, where the new species, where the novel technologies come from, not just vaguely but in particular. And how did they diffuse through population, how do they in turn create something that we call the economy? You could say that no ecology exists. You might have 25 species in the Mojave Desert in some small local place. No ecology exists, but it's convenient to say they're part of a collective interacting set of things and we can call that ecology.

Turns out that an economy is a bit similar, that if you have technologies and they're helping us get what we want to live with — it could be transportation, it could be food production — those technologies together produce something we call an economy. So rather than thinking the economy was producing technologies, it seemed to me just as useful to think that technologies were producing the economy and therefore the economy keeps changing. Because the technologies keep changing.

I want to comment a little bit on this. I put this book out on this theory of evolution and by the way, it took me 12 years to write. And actually Darwin was a sort of beacon for me. Darwin's ideas came together very slowly between roughly 1832 and 1839. He'd been on the Beagle, on that voyage and he had done an awful lot of observation. It was clear that the Earth was old and that speciation had happened, but he couldn't figure out how. He borrowed the key idea from economics that was borrowed from Malthus, that if species were struggling to survive, then ones that were slightly better adapted to a changed environment would leave more offspring, would survive better.

And so if you had variations that were better adapted, they would survive and persist until they could no longer interbreed with the original population. You'd have a new species. When I was looking at technology, I decided that I would study about 20 technologies and maybe a dozen of those extremely well. So I studied professional books on jet engines and I talked to the chief engineer of the Boeing 747…and I mean I talked to him, asked him about certain things. I studied the evolution of computation of steam engines, railway trains of packet switching. Of penicillin, a means to human purpose…of certain technologies molecular biology, like the polymerase chain reaction.

And I read several books and I read biographies and I read lab notes on the laser printer, which was invented where I work at PARC. The ethernet. So it was a thrilling time. I did a lot of that study here in Santa Fe at St John's College library and I think I acknowledged them in the book, but a lot of that was done here at the Santa Fe Institute. And sitting for three hours every morning reading about technology. The strategy was simple. I didn't start with the theory of technology. I started with vague ideas, maybe combination was important and I read and read and read a bit like Darwin running around the Galapagos collecting beetles or looking at iguanas. And I kept reading until I started to see common patterns and I began to see that every technology had come into being as for some human purpose and as a combination of what had gone on before and what was used before, and then it joined the Lego set and things could go from there.

The interesting thing is that means the economy is open-ended. There's no finish to technology. There's more and more of it and I do think it's moving rather faster because we have more means devoted to that purpose. Doesn't mean people are thinking faster, and doesn't just mean there's more to combine with. It means there's more resources going into that. Not just DARPA or government, not just Silicon Valley, but all over the world. Huge amount of work going into it.