An Economist's View of Crisis, Human Nature, Rebellion, and the Future Role of Technology | Interviews

Wang Dingding, a professor at Peking University, once said: "Arthur's story is a permanent reminder to me that any mainstream — including the mainstream of economics — inevitably suppresses and excludes humanity's independent spirit and free thought."

On November 5, 1979, W. Brian Arthur picked up a ruler and a pencil, drew two lines in a black notebook, wrote "New Economics" and "Old Economics" at the head of the two columns, and listed 24 differences between the old economics and the new.

The old economics, for instance, treated economics as a "soft" physics built on 19th-century physical theory — equilibrium, stability, deterministic dynamics. The new economics treated it as an extremely complex science built on biological theory — structure, pattern, self-organization, life cycles.

In the nearly 40 years since, academics have cited Arthur's table many times, and it became the source of the new framework of economic thought he founded: complexity economics. Looking back at those notes this year, Arthur, now 72, feels he would not change a word. He still holds the views he held then.

"I don't think of it as the product of a burst of inspiration. It was more the product of being provoked, or disappointed. There was so much in standard economics at the time that I didn't agree with. I sat down and thought: here are all these old views — do I actually believe any of them? No! I said to myself," he told QDaily.

Arthur is a well-known American economist and a major founder of complexity science. He has won the Schumpeter Prize and the Lagrange Prize, among others, and is the author of Complexity and the Economy, The Nature of Technology, and Increasing Returns and Path Dependence in the Economy.

Kenneth J. Arrow, the Nobel laureate in economics, once said of him: "Brian Arthur changed the way we look at economic phenomena! The complexity of the economic system comes, on one hand, from the wildly varying expectations of different people, and on the other, from the law of increasing returns, which also determines how the economy will evolve."

Arthur was born in Belfast, Northern Ireland. In Ireland, he believes, a call to obey authority never worked. Every Irish hero was a revolutionary, and the highest form of heroism was to lead a revolution with no hope at all of winning, and then, the night before you were hanged, to give the greatest speech of your life from the dock. It was this rebellious streak in the Irish that, in a strange way, set him off on his academic career.

"Belfast Catholics have a marked tendency to despise intellectuals. So of course he was determined to become one," wrote Mitchell M. Waldrop, an editor at Nature, in his book Complexity: The Emerging Science at the Edge of Order and Chaos.

Arthur wanted to be a scientist from childhood. At school he threw himself from the start into engineering, mathematics, and physics. In 1966 he took first-class honors in electrical and mechanical engineering at Queen's University Belfast. He then went to Lancaster University in England and the University of Michigan in the United States, taking master's degrees in operations research (1967) and mathematics (1969), and after that to UC Berkeley, where he earned a master's in economics (1973) and a PhD in operations research (1973).

In the classrooms at Berkeley, Arthur recalls in Complexity, economics felt like a branch of pure mathematics. Neoclassical economics, celebrated as the foundational theory of the field, had reduced this rich and intricate world to a set of narrow, abstract laws that could be written out in a few pages. Every textbook was full of mathematical equations. The best young economists all seemed to be giving their academic lives to proving one theorem after another, without asking whether those theorems had anything to do with the real world.

The simplifications and unreality of neoclassical economics left Arthur more and more dissatisfied. By 1979, working at the International Institute for Applied Systems Analysis in Austria, he was reading widely in biology — above all the work of Jacques Monod and François Jacob, along with research on self-organizing systems by several groups in Brussels and Stuttgart. On the basis of these ideas, and of his own long thinking, Arthur came to see that the economics of the future would have to differ sharply from traditional economics and rest on different principles. The end product of that thinking was the table he wrote out on November 5, 1979.

In 1983, at 37, Arthur became the youngest economics professor Stanford had ever had. In his first years there he published a great many papers and served as department chair. He then spent several years trying to publish his work on increasing returns — and that is what drove him out of Stanford. Almost nobody in the academy at the time believed in increasing returns; they thought Arthur's ideas were heresy and would not publish them. Increasing returns is the tendency whereby a technology that gets out in front is carried further ahead by positive feedback, until its lead grows larger still. The QWERTY keyboard is a good example. In the end, of course, Arthur's ideas won acceptance.

At the time, Kenneth Arrow was one of the few scholars who encouraged Arthur to think for himself and keep going.

One day in April 1987, Arthur was crossing the Stanford campus on his way to his office when a man on a bicycle circled him and stopped in front of him. It was Arrow. He told Arthur that a conference would be held in September, bringing a group of theoretical economists together with a group of natural scientists to trade ideas — the economists chosen by Arrow, the physicists by Philip Anderson, the Nobel laureate in physics. The conference would take place at the Santa Fe Institute, a small institute at the foot of the Rocky Mountains that had only just started up.

Arthur gladly accepted Arrow's invitation. Later, again on Arrow's recommendation, he received an invitation from the Santa Fe Institute's science board to come and lead a research program called "The Economy as an Evolving Complex System" — the institute's first. Santa Fe went on to become a world-renowned center for research in complexity science. Arthur sat on its science board for 18 years and its board of trustees for 10, led and took part in many projects, and became a major founder of complexity science.

"Arrow always championed Arthur's economics of increasing returns loudly. Another man deeply influenced by this idea of Arthur's was the economic historian Douglass North, who won the 1993 Nobel Prize in economics for his work on new institutional economics. When I was teaching at the University of Hong Kong, I leafed through North's new 1990 book, Institutions, Institutional Change and Economic Performance, at the HKU bookstore, and what struck me most was his use of Arthur's idea of increasing returns in the study of institutional change. Institutions in every country — and especially, I think, in a country like China, with its long and unbroken history — have an extremely strong path-dependent character. From then on, Arthur was one of the central figures in Western scholarship I followed. Arthur's story is a permanent reminder to me that any mainstream, including the mainstream of economics, inevitably suppresses and excludes humanity's independent spirit and free thought," Wang Dingding, professor of economics at Peking University, recalled in his essay "Path Dependence: Population, Economy, Technology."

In May 2018, Arthur published in China what may be the last book of his life: Complexity and the Economy. It gathers the major essays on complexity economics he has written over more than 30 years. "This book reflects the contribution I have made to the development of this new framework. Broadly, the core idea running through these essays comes to this: the economy need not be in equilibrium. In this system, agents are constantly changing their actions and strategies in response to the outcomes they jointly create. That is, in this system, agents keep creating an 'ecology' that they themselves must then adapt to," he writes in it.

In June 2018, we met the 72-year-old Arthur at a hotel in Beijing's Chaoyang district. He was warm, with flashes of Irish humor. He told us, for instance: "Someone once came up and asked me whether the Santa Fe Institute was a university. He'd never heard of it. I remember telling him that a hundred years ago, around 1890, Santa Fe had a choice between becoming home to a state university and home to a state prison — and chose the prison, on the grounds that this would be safer for the students."

Q = QDaily

Arthur = W. Brian Arthur

Economic crises seem to have got worse over the years, because economists always seem to imagine the system in equilibrium

Q: To help ordinary readers get at your thinking, let's start with the question of economic crisis, which is probably what people care about most. In your new book you write: "If we look back at the economic crises of the past 25 years, we find that all of them were, to a great extent, caused by a small number of well-placed 'big players' exploiting loopholes in the economic system, or by markets running out of control." This year is the 10th anniversary of the 2008 financial crisis. Looking back at it, what strikes you most?

Arthur: Ten years ago, in that crisis of 2008, I thought the financial system was very unstable. My thinking hasn't really changed since — not about that crisis, and not about the financial crises at other times either. A crisis starts with speculation by some group of people; they keep investing, and the whole financial system becomes dangerous. When some parts of the system start to collapse, other parts collapse with them. The whole thing comes down like a house of cards. That is my basic view of financial crises, and it hasn't changed much over time.

But about crises themselves I do have something to say. Data from the past 50 years show that air passenger safety, seismic ratings for buildings, medical safety, food safety — in China as in the West — have all become safer. Everything except the economy. Economic crises seem to have got worse over the years, like the one 10 years ago. The reason is that economists always seem to imagine the system in equilibrium, as though a building standing up were in its equilibrium state.

In China, in Europe, anywhere else, we have not understood the economy structurally. We haven't grasped that tiny cracks inside a building's structure can bring the whole building down. What we need is a shift in understanding.

Q: On the same point, you say in the book that "this new framework of complexity economics has policy implications" — that it helps with policies for reining in excesses, policies that use "nudges" to bring about good outcomes, policies that create conditions favorable to innovation. You give one detailed example in your essay "Even the Best Economic and Social Systems Can Be 'Gamed.'" Are there other concrete examples you could share?

Arthur: The concept of equilibrium exists in standard economics and in neoclassical economics both. It is like a spider's web, with various forces woven together into a state of balanced tension. Once you see that, you will see that the research path of complexity science assumes the economy is not in equilibrium — because there are always different incentives pushing people to make changes inside the system and to exploit those changes. In short, you don't assume the economy is in equilibrium by default, so the policy implications change accordingly.

You might want to ask: in that case, where should our economy go? Take electricity. If a country wants a lot of nuclear power, it will use more nuclear power. At first these systems are all balanced, and then the government tilts them. So once you recognize that the economy is inherently unstable, you can push it in the direction you want.

Which brings us to economic policy. You want to know which parts of the economy are unstable, which parts have room to grow, and how to help them grow. Standard economic theory studies only states of equilibrium. Complexity economics asks a different set of questions: what are the various parts of the economy like? How does the economy itself develop? What path of development should we take? Those are the questions complexity economics sets out to answer.

Q: You mentioned that complexity economics helps us understand and address the persistence of poverty. Why?

Arthur: One assumption of complexity economics is that you get positive feedback. Neoclassical economics assumes the economy runs entirely on negative feedback — on the tendency of small events to die away. If some variable increases, the returns it brings diminish. In complexity economics, positive and negative feedback exist together.

There are plenty of examples of this in Western economies. You can see how the children of the very rich all get a good education and already hold a great deal of wealth at the start of their lives. Donald Trump's father, for instance, gave him a lot of money as startup capital. So the rich get richer, while at the other end poor families struggle, and their children don't get a good education either, so they fall into a poverty trap. If one generation is poor, the next has no chance of escaping dire poverty. The rich get richer and the poor get poorer.

So what complexity economics has to answer — and I don't work on this myself, though colleagues of mine do — is how we understand this mechanism and intervene in it. But you can't say everyone's economic circumstances have to be identical. Society has plenty of powerful people who can make sure their friends and relatives and descendants live well too.

I'm not an expert on the Chinese economy, so I won't say much, and I won't comment much on China's situation. But I will say that a fair share of China's very rich also come from families whose parents held political power.

Q: Seen through complexity economics, what is the relationship between institutions and the economy?

Arthur: Neoclassical economics doesn't discuss institutions much, because you're mostly doing calculations with formulas, and it's hard to bury yourself in institutions that way. But in complexity economics we try out many different institutions. We might try setting up a capitalist market on a computer, or try different patent-law systems, and so on — and see what happens.

After the 1870s, economics went in two different directions

Q: In the preface to your book you say that many of the "modern" themes in this new framework of complexity economics fit closely with the ideas of great thinkers such as Schumpeter, Smith, Mill, Marx, and Keynes, and with the theories of many institutionalists and political economists. But you also say the formal link between complexity economics and those ideas hasn't been fully established — that there are only some threads of thought. Could you spell out what those threads are? What exactly connects complexity economics to that earlier economic thought?

Arthur: That's a good question. I have to tell a story about economics itself.

The study of the economy isn't actually that old. It began roughly in the eighteenth (and nineteenth) century, mainly with a few economists on the European continent at the time: Adam Smith, John Stuart Mill, Karl Marx, and others.

These economists cared about two things. First, what we in economics call allocation. How much is produced? How much consumed? How do households consume? How do countries export? How are prices formed? And so on. But besides that, they were also very interested in something else: the process by which things develop. Karl Marx, for example, was interested in how the economy takes shape within technology. Adam Smith held that as firms in an economy grow larger, their division of labor becomes finer and more specialized.

What I want to say is that up to the 1870s, economists were focused on the question of "equilibrium" — a very neoclassical concern. They also cared about how the economy develops, how institutions are built, how technology changes, how it bears on politics, and so on. But these are two different kinds of question. The ones tied to equilibrium were measurable; in the 1870s, economists used linear algebra and calculus to handle questions of "equilibrium." About 100 years later, in the 1970s, neoclassical theory was finally accepted.

The other questions — how do social classes form? How does the economy affect political power? What can we do about poverty? Where do institutions and laws come from? — were not measurable, not mathematizable. Their answers were not then thought to belong to economics at all, but to sociology, political science, or history.

The picture is clear. After the 1870s, economics went in two different directions. What economics used to care about became the concern of sociology, jurisprudence, political science, and Marxist theory instead.

I was very excited at the time. We found that complexity economics could do more, that it could answer questions about processes of evolution. Then we went back to the work of our predecessors from the mid-to-late 19th century — the economists I just mentioned. And we could use computers, and mathematical methods more advanced than theirs, to take on the questions they had raised. Now we can simulate an economy inside a computer and watch how institutions and outcomes form.

So my friends and I felt we could suddenly sit down and say to Adam Smith or Karl Marx: do you know what we're doing now? And they'd say, oh, of course, we know all about it. It is as if we had trekked over the mountains and, on the far side, run straight into this crowd of predecessors. In the work I do now, this feeling of running into people who came before me, unintentionally and even delightedly, is a thrill. Quite unexpectedly, we can talk with people from the past in something close to a shared language. But our assumptions are more precise and our methods more advanced, and in computer-based models we are connected again to the economists who came before us.

Computer modeling and fieldwork aren't that far apart; theory always has to come back to the real world

Q: Since we're on methods — when you study economic problems, the method you mostly use is to take a core mechanism and build a computer model of it. Why this approach? What are its strengths and weaknesses?

Arthur: In the 1880s, a great many Western economists felt we had a perfect neoclassical theory of the economy. I was trained in that theory too at the start. I got my economics training at UC Berkeley, studying under a number of Nobel laureate economists — Gerard Debreu, Daniel L. McFadden, George A. Akerlof. They were all excellent teachers.

Later I found that to make this system (neoclassical economics) run, a great deal has to be simplified. Picture a standard economic model as a stack of paper on which you write equations; to keep the equations neat you make a series of simple assumptions — everyone is the same, all producers and firms are the same, everyone is rational, everyone faces diminishing marginal returns — and then you can solve the thing.

By around the 1980s, people were growing more and more dissatisfied with this way of working. Back in the seventies I said to my professor: "Do you really believe all this?" My professor said: "Well, I don't know either, but what can you do?"

Back then, in the 1980s, what could we do? Before that, our research tools in economics were basically linear algebra and calculus, plus maybe a bit of topology. Earlier still, we had charts and geometry. After that we got a new tool: the computer. Though the computers of that time weren't like today's. They were called "desktop workstations" and looked like boxes. We had computers by 1982 and 1983, and then in 1984 came Apple's Macintosh.

Once we all had computers, we could build models on them without having to make simple assumptions. You could create 10,000 different firms inside a computer, each with its own set of consumers. You didn't have to assume that diminishing marginal returns meant negative feedback either; we could have all kinds of feedback. Which is to say, you could build any model you liked. A computer lets you reflect the world more realistically.

What we did at Santa Fe amounted to calculation as well, but we weren't calculating on paper with linear algebra or calculus. We built a system inside the computer in which the participants in an economy meet, trade, form expectations, and so on. We modeled the participants too — each computer program stood for one participant, and they could buy and sell at random, but they could also learn to get smarter. Once we did that, we found we were no longer stuck with those simplified assumptions.

Back to your question about computer models. If you build a model on a computer today, it may mean nothing at all. There's a saying in computing: "garbage in, garbage out" (that is, if you feed a computer system wrong or meaningless data, the computer will naturally produce wrong, meaningless results). So it is entirely possible to build a very stupid, very unrealistic model, and to turn out a lot of bad economics that way.

What we try to do is be very careful about our assumptions. We don't regard what we're doing as simulation — at least I insist on that myself. We are not simulating the real-world economy; these are experiments run in a laboratory. If the economy worked this way, what result would you get? Then you can fine-tune your assumptions. People aren't that smart (they aren't all rational agents) — so what will they do?

So complexity economics is basically something you do with a computer's help. Through the computer you can make different assumptions, more realistic ones, instead of assumptions like "everyone is a rational agent." People aren't that smart, but they can learn.

Q: Set against this computer-based method, how do you see the value of fieldwork for the questions you care about?

Arthur: I'm not much of a fieldworker myself, but the approach has influenced me deeply. I once worked at Citibank's foreign exchange trading floor in Hong Kong, and that was rather like fieldwork.

The world neoclassical economics describes seems far removed from the world outside; it seems to have a world of its own, an imagined one. What we can do is observe the real world more. I might not call it fieldwork, but at least we can get closer to it bit by bit. We can watch how an international foreign-exchange floor works, and then put those observations to use when we build models. So our work isn't fieldwork in the strict sense either — it isn't that we only do theory on computers. It's all mixed together. Computer modeling and fieldwork aren't that far apart; theory always has to come back to the real world.

Of 20th-century economists, the two greatest were Schumpeter and Keynes

Q: In The Nature of Technology you discuss innovation and bring up Joseph Schumpeter's "creative destruction." You've also won the Schumpeter Prize. I'm curious how you see the connections and the differences between your thinking on innovation and Schumpeter's.

Arthur: When I was writing The Nature of Technology, two questions interested me especially. First, where does technology come from? The second is a bit harder: is there a theory of the evolution of technology?

When I started working on technology, I came quite naturally upon what Schumpeter had done. He had already thought through the very questions I was raising. It was like setting out on an expedition, picking your campsite, and finding that someone had been there 50 years before. I found that fascinating, and I became interested in Schumpeter.

Where does technology come from? You might say it comes from inventions like the steam engine or radar. In my book I concentrated on where these new things come from. Schumpeter pointed out that the arrival of a new invention — electricity and electrical machinery in the 1880s to 1920s, say — was like a great wind blowing through, changing every industry and the whole economy. People in the cities were swept up as if by a hurricane. It's a beautiful image.

What I could find was the finer detail. What I wrote about in my book wasn't only new technology but how existing technologies inside each industry give birth to new ones, and how new technologies combine with old. So I was very glad that Schumpeter had done so much related work before me. To me it meant I was on the right road. I trusted him.

Of 20th-century economists, the two greatest were Schumpeter and Keynes. They knew each other, but they didn't like each other — that's another story. Schumpeter thought he was the best economist in the world (I think so too), but if you had asked anyone in the 1940s, they would have said Keynes. Schumpeter couldn't accept that.

Q: This year the Santa Fe Institute turns 34. Looking back over these decades of the institute and of the field of complexity, what stands out to you?

Arthur: At Santa Fe, everyone thought the work they were doing was risky, hard to believe. I agree it's an unusual place.

In the early days, what mostly happened was that we'd hear so-and-so was working on such-and-such, and we'd have to pull them in. So around 1988 we really were a brand-new institution just finding its feet. I remember going to academic conferences with a badge that said Santa Fe Institute, and nobody knew what that was.

What people don't realize is that the Santa Fe Institute was built from nothing. My view of why it succeeded is that we insisted on bringing in a lot of outstanding people. We kept trying to recruit people who were doing creative research that no one appreciated. Once they were in, we'd give them a team, or they could bring their own people. So at Santa Fe we had the money and we had the people, and we built the institution around those people.

I was talking recently with some people who fund research in Singapore, and they asked me how I judge whether a piece of research is good, and how we choose research topics at Santa Fe — since that's my job. I said, I don't pick anything. I don't work out or analyze research topics. What I pick is people. If you want to do really outstanding science, you need to find the right people.

I'm not the only one doing the recruiting either. We have plenty of Nobel laureates. I'm not well-known enough to draw certain people in, so when I need someone I go to my Nobel laureate colleagues and ask them to do the asking: can you give so-and-so a call? Just tell them Santa Fe needs them.

I was taught from childhood not to trust authority

Q: I meant to ask some personal questions, but I found that Mitchell Waldrop had already written a very thorough book, Complexity: The Emerging Science at the Edge of Order and Chaos. Have you read it? Is there anything you'd add or want to say about it?

Arthur: Well, I remember Waldrop wrote that book in 1990. It isn't a new book — it's twenty-seven or twenty-eight years old — but it's still the best book in the field of complexity economics, because it explains abstract ideas so clearly. The author also decided to build the book around people: Chris Langton, John Holland, Stuart Kauffman, and others. The reason I run through the whole book is probably that he needed a figure to hang everything on, a kind of coat rack that could string all the others together. I met everyone else in the book in that order, one after another. So the author decided to make me the "territory" where many of these people crossed paths. The book is very accurate; he had interview records for every story in it. As far as I know it's all well written and thorough. He's an excellent journalist. He has his own way of telling a story that makes everything hold together, our research included.

Q: I think it's a very good book too. The first chapter, for instance, is called "The Irish Idea of a Hero," and it says it was precisely the Irish streak of rebellion that shaped your academic career. I'm curious whether you agree. Do you think of yourself as a rebel? How much has that trait shaped you?

Arthur: I don't quite agree about the "hero" part, but I agree about the "rebellion" part. And there's a legitimate reason for the rebellion. If you know Irish history, you know we were ruled by Britain for 800 years. I grew up in much the same setting as other Irish people, and we don't by nature trust the people who govern us. I was taught from childhood not to trust authority. My parents never taught me this outright. I picked it up from the whole surrounding culture. If someone tells you this is the right way to do it, most of the time what he says is wrong. I also wouldn't exaggerate this influence, as if to say that because of it, ah, I feel an urge to smash the whole system, or something like that.

By the time I met neoclassical economics I was fairly old. I'd already studied mathematics and engineering for years, so I was quite old when I started on economics. You might say twenty-four or twenty-five isn't old, but if you send someone into the army early, by 17 they're a fine soldier. By twenty-four or twenty-five, a person has reached the age of thinking for himself. Some things get harder by then. So when I was learning economics at that age, I did a lot of thinking for myself, and asked myself: is this right?

But I don't really think of myself as a very rebellious person. I just think that, having grown up in a society like that, I don't trust those in power. And as I see it, neoclassical economics sacrificed a great deal of realism to achieve mathematical elegance. So I thought it was wrong. I'd rather have a messier world, because that's the more realistic world. I understood then that a supposedly elegant view of the world was a wrong one.

I found I understood the economy better through reading history than I ever had through mathematics

Q: Since complexity economics is closely bound up with history, I'm curious what your view of history is. Are there historians you particularly like?

Arthur: I've read a great many history books. My early training was as an engineer, and everything was mathematized — everything reduced to the level of mathematics. I studied that for a long time. But over roughly the past 20 or 25 years, from about my fifties on, I found that I knew an enormous amount of economic theory, mathematical economic theory, mountains of it, and that these theories weren't insightful enough. They were dealing with a perfect machine in which everything reaches equilibrium. They didn't tell us how the machine changes, or how it comes to appear in front of us.

That's when I began reading history — a great deal of the history of science, of engineering, of technology, and so on. At the start I hadn't set out to read those particular kinds of history, but in the end that's mostly what I read. Lately I've been reading a book about the Battle of Agincourt in 1415, the Battle of Waterloo in 1814, and the Battle of the Somme in 1916: The Face of Battle.

So I found myself reading history quite naturally. After 10 or 15 years of it, I found I understood the economy better through reading history than I ever had through mathematics. I gave a lecture in Dalian yesterday, and I said there that if you want to understand the importance of artificial intelligence, you have to understand the importance of printing in Europe at the time. Going from written books to printed ones changed the medium through which information could be had. In the early 15th century, information sat mainly in books; you could buy a book, think for yourself, and then discuss it with people who knew about it. Intelligence, too, comes from outside.

What I mean is that real intelligence tells you how to do things. That may be a good thing, and it may not. But I find that historians grasp this better, because they can see the resemblances across history.

That made me think that the best economists — the best economists of the 20th century — were in fact all mathematicians. They all won Nobel Prizes and they were all formidable. But to my mind the two truly great ones were Keynes and Schumpeter. Keynes needed a lot of mathematics, but Schumpeter didn't understand mathematics. He was a historian.

Let me use another comparison. A doctor — an obstetrician, an oncologist, a cancer specialist — can be produced by medical training. But if a cancer doctor has 35 years of practice behind him, and in those years has come to know his patients, talked with them honestly, understood their circumstances and watched what befell them, that doctor holds a much richer inner knowledge. In the West we have gradually lost our respect for that side of medicine, the understanding of patients' experience. It's the same in other lines of work; engineers and architects also need to build up a good understanding of experience.

Back to economists: if you're an economist, I've never once regretted knowing as much economic history as I do.

Q: Reading your book, I noticed you quote a number of Eastern philosophers, such as Cheng Yi and Cheng Hao of China. I looked into it afterwards and found you know something about Daoism as well. So I'm curious whether Confucian, Daoist, and other Eastern thought influenced how you built the theory of complexity economics.

Arthur: My direct answer is yes. But it's hard to say how. I'm not sure how these things influenced me, just as I don't know how growing up in Ireland influenced me. These influences are all very subtle.

What I can say is that around 1986, 32 years ago, I became very interested in Daoist thought. I was in Hong Kong then, and I had a teacher. I studied with him one-on-one for several years. I also read on my own, working through many books on Asian philosophy, and found them very stimulating. I read Laozi and Zhuangzi, and a great deal of Neo-Confucian thought too — Cheng Yi and Cheng Hao of the Song dynasty, and others. I don't know whether I really understood them. (Laughs.)

I hold ancient Chinese philosophy in respect, so I'm afraid of getting it wrong. If I had to sum it up more clearly, as I understand it, Daoism tends to see the elements of the world as combining into patterns, and those elements in turn respond to the patterns they have formed; the patterns keep changing, and new things emerge from them. That is the essence of complexity science.

Take complexity science itself. If you study it, think of a flock of birds — there might be many thousands of birds in it, an enormous number. The flock in flight is like a cloud, a cloud that dives down out of the sky and changes shape constantly. But every bird is one member of the flock, and a change in the birds can change the shape of the whole flock. There's an interaction between individual behavior and the form of the whole. The point is that there is no fixed state, no equilibrium. The point is that individuals adjust their behavior as circumstances change.

China has a concept of harmony, of being in accord. Western harmony means everything sitting in a delicate balance. Chinese harmony is dynamic — like riding a bicycle: stop, and the harmony is gone. So you're always adjusting. That is also the logic by which complexity science works. It is how we look at the economy too: as a system full of change, whose elements keep adapting and adjusting as things change.

The big question is how we can hold on to our own humanity, our humanistic spirit

Q: In "Even the Best Economic and Social Systems Can Be 'Gamed'" you write: "In the future, computer simulation ought to be able to fold in history well — to find, for instance, the ladders that answer the high walls, and present these to us... I think it may take several more decades of effort to make full use of this kind of practical machine intelligence." When that happens, do you think there will still be a difference between humans and machines? Why?

Arthur: Well, at that point there won't be a difference between people and machines. I wrote that essay in 2010. At the time I was working at IBM's Almaden research lab, and they were very interested in artificial intelligence. That year IBM trained its own computer to answer questions on the quiz show Jeopardy! The show has been very popular in the United States, with a 50-year history, and its questions are hard — only very smart people can answer them. IBM put a lot of effort into training a computer to answer those hard questions. We were very curious then about whether a computer could answer them the way a person could.

Let me give an example. In 2003, under President Bush, the American army attacked Iraq. I won't say anything too harsh about it, but the campaign wasn't well planned. Before long, within a few months, an insurgency began. Many groups inside Iraq started to resist, and the American army ran into trouble. When the army first went in, thoughtful people warned that care was needed, because after Britain's earlier invasion of Iraq there had been an insurgency too. But nobody listened to that warning at all.

So in that essay I made a guess: if you want to find where trouble might break out in a system, perhaps you could use a computer, one holding a lot of historical experience of similar trouble. A question would pop up on the computer: given Britain's earlier invasion of Iraq, Russia's invasion of Afghanistan (and the insurgencies that followed those invasions), what will happen if the United States invades Iraq now? Have you thought about that?

So my question was whether there could be an artificial-intelligence computer that holds a lot of basic knowledge and can warn you before a crisis happens. I put the question in the essay because IBM was so interested in artificial intelligence at the time. It was only a guess, and normally I wouldn't put that sort of thing in an essay. But it was interesting all the same. I thought my colleagues at IBM would enjoy thinking about it.

Q: What have you been thinking about lately?

Arthur: Lately I've been thinking about artificial intelligence — where it's headed, how it will work, that sort of question. I've been interested in AI for a long time, since 30 years ago. But despite that long interest, I'm not an expert in it.

Artificial intelligence is very fashionable now, but it isn't a technology that will turn every industry upside down. On that point Schumpeter was right. What's happening now is subtler. AI technology has many parts, like the bricks in a Lego set, and it seeps into every industry. Take transport: the control-system design in driverless cars, image recognition, and so on.

I think it is creating an economy of a kind. As I said, we now have many routes to intelligence outside ourselves. If you read a book, you gain a route into another person's mind, a route to information from outside. Now you only have to sit in a car and you have a great deal of outside intelligence available — information technology, radar, lidar, image processing. That's intelligence enough to help you drive. In the Middle Ages we didn't have so many routes to information. We could only talk with people in the village, or till our own fields. Europeans didn't know China existed, and the Chinese didn't know Europe existed. And of course nobody knew Jupiter existed, or knew anything about the bright points of light in the sky.

So once people got hold of outside information, the world became different, became modern. People began exchanging information, and science came out of that. One person's ideas could travel around the world; ideas led to technology, and to everything else. Five or six hundred years later, the world became what we now call the "modern world." And now we have this outside intelligence, and I don't know what change it will bring to the world. Our homes are full of the services technology brings us; we have all sorts of machines and cars, and these interconnected networks of services will make decisions for us.

That will be wonderful, but there is another side to it, and I don't know what other effects it will have on human beings.

Q: Are you optimistic about it?

Arthur: No, my feelings are mixed. If I could live another 50 years — which of course isn't possible — but if I could live longer, I imagine artificial intelligence will combine with things like genetic technology, and they'll be able to tell me about my ancestors. That would be interesting. We'll develop air traffic management in which machines act on their own and no longer need humans, and so on. I've heard these pictures of the future more or less everywhere else, and in China too.

I do have worries. Economists generally worry about the unemployment artificial intelligence may cause, but what I worry about is slightly different — not worry, really; concern. I'm more concerned about the meaning of human life. Why are we alive? What is the pleasure of life? Is it just to be more comfortable?

Robots will come, and they'll serve us. That's a good thing. But we aren't alive in this world to be comfortable, to take in this information, to sit there watching television. We need challenges. I once wrote in a book that we really do need to face challenges. Difficulty and challenge give us meaning. This isn't to say we should throw away comfort. Oh heavens, I served five years in the army, and it's days like those in the army that really make you a person.

I'm not recommending hardship, but we are meant to grow stronger through hard times. If you never face any challenge, you become weak. When you've been through enough difficulty, you find that something inside you has grown too. In The Nature of Technology I mentioned the story of Star Wars: a small band of humans and the "Death Star," that product of high technology, and how these people have to fight it to save the last of humanity in a world ruled by technology.

So the big question is how we can hold on to our own humanity, our humanistic spirit. Over these years, in California, in Singapore, in China and elsewhere, I've seen people begin, without quite realizing it, to answer technology with a redoubled emphasis on humanity. People say, ah, we want engineers to start reading novels, we want to care more about human beings. And I can tell you that 50 years ago we were training tough, resilient human beings to think like machines. Almost the exact opposite of now.

Artificial intelligence is rising now, but attention to humanity is rising alongside it. It might be attention to a sense of humor, or to courage, luck, challenge and hardship, or to caring for others rather than living for oneself. What form it will take, I don't know either; it will be a process lasting decades, maybe centuries. But I'm not pessimistic. I don't think humanity will finish itself off.

(Wang Sui's translation work also contributed to this piece — with thanks.)

Appendix: the "Old vs. New Economics" table Arthur wrote out in 1979

Cover image: a still from the film Rogue One: A Star Wars Story, via Douban.