Once upon a time at UC Santa Cruz, a group of renegade grad students started mixing physics with math and computers, determined to discover underlying patterns in the seeming-randomness of systems like the weather and roulette. Their research led to major insights in the emerging field of chaos theory, and eventually to the new discipline of complexity economics — which brings models from ecology and physics, cognitive science and biology together to improve our understanding of how value flows through networks, how people make decisions, and how new technologies evolve. As the human world weaves new global economic systems and sustainability looms ever-larger in importance, it is finally time to heed the warnings — and the promises — of this new paradigm of economics.
Welcome to Complexity, the official podcast of the Santa Fe Institute. I’m your host, Michael Garfield, and every other week we’ll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers developing new frameworks to explain the deepest mysteries of the universe.
This week on Complexity, we speak with SFI External Professor J. Doyne Farmer at INET Oxford, to tour his fifty years of pioneering work and current book-in-progress, The Complexity Economics Revolution. Topics include how ecology inspires novel forms of macroeconomics; how “bounded rationality” changes the narrative about rational self-interested economic actors; how leverage leads to greater instability; how new tools can help us predict emerging innovations and engineer a better banking system; the skewed incentives of science funding and venture capital; his take on cryptocurrencies; and more…
If you value our research and communication efforts, please rate and review us at Apple Podcasts, and/or consider making a donation at santafe.edu/podcastgive. You can find numerous other ways to engage with us at santafe.edu/engage. Thank you for listening!
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:
Twitter • YouTube • Facebook • Instagram • LinkedIn
Doyne Farmer’s Personal Website | SFI Page | INET Oxford Page | Google Scholar Page
Doyne Farmer and related talks on our YouTube channel
Complexity Economics from SFI Press
Related Complexity Podcast Episodes:
W. Brian Arthur on The History & Future of Complexity Economics
[Farmer’s PhD student] R. Maria del-Rio Chanona on Modeling Labor Markets
Matthew Jackson on The Science of Social Networks
Geoffrey West on Scaling and Superlinear Innovation
David Krakauer on Collapse & High-Beta Investment Strategies
This transcript was generated by podscribe.ai and still needs a human edit. If you'd like to help us edit podcast transcripts, please email michaelgarfield[at]santafe[dot]edu. Thanks and enjoy:
Doyne Farmer (0s):
Do you figure out what people are going to do when confronted with an economic decision, by assuming they have a utility function that tells them what they want and they know all the possible events that might happen in the future, and they know the probabilities of those events. And then they do a calculation to find the decision that maximizes their utility and therefore take that decision. There are circumstances where that's a reasonable thing to assume. Like if you're designing an auction where you've got specialists thinking really carefully about what's going on, where the world is really constrained, you may actually know the probabilities of the events and that you may know all the possible events and you may be able to solve that. But in the real world, if you're talking about climate change or inequality or financial crises, we don't have good models for what's happening.
Doyne Farmer (46s):
We may not even be able to anticipate all the things that could happen in that kind of world. The world we've evolved in people, reason differently. People use heuristics, people imitate their neighbors, or pick an example out at random, then just follow that. But the reason in a very bounded way, it's just a whole different way of modeling human behavior than the way economists currently do it.
Michael Garfield (1m 32s):
Once upon a time at UC Santa Cruz, a group of Renegade grad students started mixing physics with math and computers, determined to discover underlying patterns in the seeming randomness of systems like the weather and roulette. Their research led to major insights in the emerging field of chaos theory and eventually to the new discipline complexity economics, which brings models from ecology and physics, cognitive science, and biology together to improve our understanding of how value flows through networks, how people make decisions and how new technologies evolve as the human world weaves new global economic systems and sustainability looms ever larger in importance.
Michael Garfield (2m 16s):
It is finally time to heat the warnings and the promises of this new paradigm of economics. Welcome to complexity, the official podcast at the Santa Fe Institute. I'm your host, Michael Garfield, and every other week, we'll bring you with us for far ranging conversations with our worldwide network of rigorous researchers, developing new frameworks, to explain the deepest mysteries of the universe this week on complexity, we speak with SFI external professor J Duane farmer at Oxford to tour his 50 years pioneering work and current book in progress. The complexity economics revolution topics include how ecology inspires forms of macro Economics, how bounded rationality changes the narrative about rational self-interested economic actors, how leverage leads to greater instability, how new tools can help us predict emerging innovations and engineer, a better banking system, the skewed incentives of science funding, and venture capital.
Michael Garfield (3m 20s):
His take on cryptocurrencies and more. If you value our research and communication efforts, please rate and review email@example.com slash podcast. Give you can find numerous other ways to engage with us at Santa fe.edu/engage. Thank you for Jay Duane farmer. It's a pleasure to have you on complexity podcast. Well, it's a pleasure to be here in preparation for this. You sent me a draft galley of your upcoming book, the complexity economics revolution. And so, as I was just telling you a moment ago before we started recording, I think using the way that you have organized this book as a kind of spine for this conversation makes the most sense, which I think humorously drops us directly into one of the deepest questions we could ask on this show, which is a question about the nature of randomness and prediction.
Michael Garfield (4m 19s):
So if you're up for it, maybe the right place to start would be to talk about how your journey to complexity economics began at the university of California Santa Cruz.
Doyne Farmer (4m 31s):
Sure. Well, I guess, you know, one of my revelations, or let's say after cogitating on what randomness is for a long, long time is a randomness is really a subjective property. It depends on what you know, and you know, the first seminal event that caused me to realize that was my experience predicting relapse, where, you know, rillettes designed to be a random number generator, but if you know, information that most people don't like the position and velocity of a ball, when it's sprouts and the equation of motion, then you can make a roulette and do it more predictable system where the numbers aren't random anymore. And in a sense, that's a good metaphor for what science does.
Michael Garfield (5m 12s):
Science takes phenomenon. It seems fluky and arbitrary and difficult to understand and gives us insight into causal mechanisms that may explain what they do and make the events around us less random. Could you tell a little bit more about the project that actually like how you actually kind of came to this understanding because this is, this is one of these stories that for deep established fans of complex systems, I can imagine they already know this, but for some of the listeners like my buddy Mitch, you provided the music for this show and used to be a professional gambler, a card counter in Vegas. I think this is the kind of story that really situates the history of this science in some, some escapades that are worth.
Michael Garfield (5m 58s):
Doyne Farmer (5m 59s):
Well, the story began when Norman Packard and I, who had both been pursuing gambling as a way to make money as a summer job, me playing poker and Norman County blackjack, Norman had the realization together with his friend Jack piles. That roulette is a physical system. Therefore it should be predictable. And the key fact is that the crew K does not close the bets until a couple of seconds before the ball exits the track. So that gives you enough time. If you understand Newton's laws as applied to a roulette wheel after the crew PA releases the ball. So it's spinning around the wheel. You can measure the velocity of the ball and its position at a given point in time.
Doyne Farmer (6m 43s):
And we did it by putting switches in our shoes and clicking every time the ball passed a reference point. So the time it takes to go around is proportional to the losses or inversely proportional to velocity. Then we could, after solving the equations of motion for roulette ball, which is just a rolling ball on a circular track with wind resistance, we could predict where the ball was going to exit the wheel. The ball would then bounce around on the cups and the little diamonds for a while, but it turns out we were able to did a bunch of experiments to show that didn't totally randomize things. And over a saga that span quite a few years of my time in graduate school, we made 11 different trips to Nevada.
Doyne Farmer (7m 24s):
We spent a lot of time at the roulette wheel. We were able to beat roulette with about a 20% advantage as long as our equipment was running. But, you know, we were pushing the envelope of what was possible. We built what in hindsight, I mean, I had no idea at the time as a purse wearable digital computer, it wasn't just wearable. It was concealable. So it was small. And the initial version was stuffed under an armpit with a pack of 12 AA batteries under the other armpit, because back in those days, chips still use a lot of power Moore's law was not as far down the curve. This would be 1977 or 76 to 1980, roughly.
Doyne Farmer (8m 6s):
So we beat the house, but we had a lot of problems with equipment failures. And to be honest, we kind of burned out. We were also pretty nervous about getting our kneecaps broken, so we didn't have a lot of money. So we were nervous about making the stakes too big. You know, with hindsight, we should've just gone to Caesars and played with a hundred dollars chips, but we were a little too nervous about her kneecaps for that. And so it was a great learning lesson. We're the total of about 30 different people involved at one point or another. And we had a lot of fun, never really got rich as a result
Michael Garfield (8m 39s):
That brings us to the end of the seventies. And at that point you're at, UCFC starting to form the dynamical systems collective. And so, you know, talking about being at the edge of, you know, the frothy edge of, of innovation, you speak your book to the problem. As I think, you know, many people who are interested in studying chaos or complex systems, the problem of not having any PhD advisors that can actually advise you on early chaos theory. And so I'm curious to hear a little bit about the UCFC chaos cabal, the early sort of wild West days of this kind of research, and then how that ended up getting you to Los Alamos.
Doyne Farmer (9m 27s):
Well, I learned about the concept of chaos. It wasn't even called that then from my friend, Rob Shaw, who introduced us to the Lawrence equations on an analog computer and an oscilloscope screen, I was totally inspired as was my friend Norman. And we joined up with Jim Crutchfield who was an undergraduate at the time and decided that we wanted to understand that now Norman and I were really primed cause we've been banging our heads against relapse for several years. And so it clicked because we let an example, you know, that's what creates the randomness is the chaos in the trajectory of the ball.
Doyne Farmer (10m 7s):
We began by using the analog computer as if it were an experiment and resolve to come up with methods for, if there was a strange attractor inside of a turbulent flow, how would we show that was true. So we wrote a paper in 1980 that had a lot of influence and answering that question and we had a lot of fun, you know, the problem was there was just nobody at the university that was really equipped to be our advisor. There was no faculty members that were doing that when we started it. Some of them got interested in a bit later and, and did some things in the end, but they were very kind of generous in letting us pursue that. Despite the fact that, you know, we were just co-advising each other. And we were kind of lucky in being in the right place in the right time and doing that before anybody else did.
Doyne Farmer (10m 53s):
I mean, it did require a certain amount of bravery to do that. Given we were doing mathematics in a physics department and doing mathematics with computers with back in those days, wasn't really normally done that, set me up to go to Los Alamos because Los Alamos, you know, I happened to be reading a biography of J Robert Oppenheimer and I saw a poster for the Oppenheimer fellowships and I'm from New Mexico. So yeah, you're going back to New Mexico really appealed to me. And so I applied for a job at Los Alamos, despite my misgivings about the possibility of going for a weapons lab, thinking it was kind of a long shot. But when I went to interview, I was blown away by how exciting the intellectual environment was, how much people were thinking in a kind of free and open-ended manner.
Doyne Farmer (11m 37s):
There weren't any disciplinary boundaries in the theoretical division. If you did something interesting, that was good enough. It didn't have to fit in the pigeon hole. And in fact, I've made a list of my favorite and most influential dynamical systems papers. And four out of the top 10 were from Los Alamos. So it was a very natural place to go for me. And I spent 10 years there during the course of my time there at first, I was at the center for non linear studies. I was an Oppenheimer fellow. And then later I started the complex systems group there. And that's around when the Santa Fe Institute got going
Michael Garfield (12m 10s):
In talking about co-advising. And then also, you know, to draw the comparison to the kind of as interdisciplinary work that Los Alamos and later SFI has afforded people out at the edge of these kinds of questions. I'm reminded of the conversation I just had with James Evans about identifying unasked questions. There's a relationship in like the funding of science, the problem of not having an established advisor, this question of how do you direct attention to an unknown thing? How do you find the question? You, you're not asking the network of co-advisors or of interdisciplinary colleagues, maybe kind of like throwing a hail Mary pass here, but it seems to lead us into your work in studying economies as ecologies, where you have this evolutionary system that has no telos, but everyone is just adapting in this open-ended process to everyone else.
Michael Garfield (13m 10s):
And so far, as you know, David Krakauer has spoken about SFI as the mutant offspring of Los Alamos and people like Andreas Vagner have studied the rate of mutation in a sort of guided open-ended creative response to challenges. I'd love to hear you talk a little bit about the analogies between biological evolution to markets, and then how we, you know, we can actually start getting into some other research papers you sent me in which you're looking at the dynamics of marketing ecologies.
Doyne Farmer (13m 40s):
Okay, well, so you spend a lot of stuff there. So first of all, I would say telos is an emergent property. And I think there's almost a paradoxical catch 22, understanding that because Telus is a very useful concept because if there is a key thing that needs to be done for something to survive and propagate a Telus will emerge to do that thing. And it plays a central role in nature in a sense of Darwin and Wallace. His key insight is that if you have a system where things are competing with each other and where they have limited capabilities, so they're, they're evolving by changing their programs or their genomes or whatever, their strategies, so that they can make them a little better and survive a little better than they were before, because they're being selected for that.
Doyne Farmer (14m 34s):
The ones that survive better are going to propagate more because of specialization, you get an ecology of competing specialists. They may also be cooperating at times, but overall there is a kind of competition for who's going to be there, understanding how they affect each other is key to understand the system dynamics. It's key to understanding the evolution markets in particular, and having experienced starting a, an investment fund and running a quantitative investment front for eight years. You know, it's very much that kind of world traders are all looking at the system, trying to figure out who's doing what and how they can fit into it and make a little more money as a result. And they're very specialized in what they do just because they have to be.
Doyne Farmer (15m 18s):
And so thinking of the financial system as an evolving ecology is to my mind, the centerpiece of the view we should take. And that view up until I wrote a paper on that was largely missing though. It was pre saged actually by some of the earlier agent-based models, like the Santa Fe stock market model. But the broader idea wasn't fully articulated
Michael Garfield (15m 38s):
Before that that's like one of the three pillars, right?
Doyne Farmer (15m 43s):
Well, there are three pillars of mainstream economics you're talking about from my book. Yeah. So the three pillars of mainstream economics that identified our utility maximization, a model of beliefs like rational expectations is the one that's most common, but people are pursuing other ones and equilibrium, which is what you put those two elements together defined. I'm arguing. The complexity economics is abandoning all three pillars. So it really is a revolution. And that utility maximization reaches back to the middle of the 19th century. And we're really saying, got to throw the whole thing out, or you can use that framework. We're not saying that mainstream economics, as it exists now is wrong, per se.
Doyne Farmer (16m 27s):
It just, there's a lot of things that doesn't apply to very well, like complicated problems. It's not a very good model of what happens in a complex world.
Michael Garfield (16m 37s):
So one of the main features that comes in to complexity economics is this notion of bounded, rationality speaking, somewhat to your points about randomness, that rationality is something that's sort of relative to the history of an agent in this system, the kind of choices that they're making based on heuristics and prior experience that they've had. Right? I mean, it seems like that's like the origin of all of these different investment strategies, trading strategies, bounded rationalities seems like, well, am I completely wrong in saying that it's not so much a complete eraser of a utility maximizing things so much as it is adding a relativistic dimension to a Newtonian simplicity, right?
Doyne Farmer (17m 28s):
Well, it's worth parsing out the pieces of the story. So, you know, bounded rationality, well, Hey, we're all bound to be rational, right? We're not really smart. We typically don't have access to all the data. And we don't have models that allow us to use the data fully, but it's, it's really a relative thing. You know, two kids playing tic-tac-toe when I was about nine or 10, I discovered the strategy whereby Oh, can always get a tie. And then my friends discovered it and game got boring and we quit playing it because we just always had a tie. So that's a good example where as kids, we were effectively rational, we understood the problem. We were solving well enough to find the equilibrium of us all playing that same strategy that made the game boring.
Doyne Farmer (18m 13s):
On the other hand chest that never happens no matter how smart you are, the game remains challenging. And open-ended because we are bound to be rational when we play chess. So it's a REL relative statement. So rationality, the model of beliefs is just one part of it. Then there's a question of, do you figure out what people are going to do when confronted with an economic decision, by assuming they have a utility function that tells them what they want, and they know that all the possible events that might happen in the future and they know the probabilities of those events, and then they do a calculation to find the decision that maximizes their utility and therefore take that decision. There are circumstances where that's a reasonable thing to assume.
Doyne Farmer (18m 56s):
Like if you're designing an auction where you've got specialists thinking really carefully about what's going on, where the world is really constrained, you may actually know the probabilities of the events and that you may know all the possible events and you may be able to solve that and find a good answer. But in the real world, if you're talking about climate change or inequality or financial crises, these are complicated things where we don't have good models for what's happening. We may not even be able to anticipate all the things that could happen in that kind of world. The world we've evolved in people, reason differently. People use heuristics. People do things like imitate their neighbors, or pick an example out at random and just follow that.
Doyne Farmer (19m 40s):
And they reason a little bit about what's going on, but they reason in a very bounded way, it's just a whole different way of modeling human behavior than the way economists currently do it. And that's the way we agent based modelers typically end up modeling the world.
Michael Garfield (19m 54s):
So it would seem that that brings us directly to this paper that you co-authored with Shoal and unkindness. Ooh, how market ecology explains market malfunction. Right? You're doing a really interesting thing in this paper using ecological models. Like if people know the lock, the Volterra equation, you know, predator, prey cycles, you're applying something like that to, you know, a population of noise, traders, value investors and trend followers. And I'd love to hear you talk a little bit about how you're rigorously extending this analogy into this space. And then what you found in the relationships between these three sort of species or genera of market strategies and, and what it means for macro economics.
Doyne Farmer (20m 43s):
With this idea that trading strategies are like species and they have a colleges and that they may interact with each other like lions and zebras and grass. In this case, the food source ultimately are inefficiencies in the market ways in which things are not perfect that allow traders to make money. And who's present in the market is going to influence what those inefficiencies are and what the available niches are. So the goal is to understand why markets, why do they malfunction? Why do they do things that seem imperfect and flawed? Like why is it that prices often seem to deviate from fundamental values?
Doyne Farmer (21m 28s):
Why is it that market it's often can get very volatile for reasons that seem to have nothing to do with the underlying fundamentals about what's going on in the market and nothing to do with outside news, the market just gets volatile because the market is volatile right now and because of its own internal decks. And so we show how in a world where you have three strategies, all of them are bound to be rational. They're none of them has access to complete information and none of them perfect model where you let the market evolve by having the strategies that accumulate profits, cumulate wealth, which then means they have more influence on how prices get set.
Doyne Farmer (22m 11s):
Cause more wealth means you're making can bigger trades, which means you have more influence on how the price, this moves every day. So we just put some strategies in and let things go. And we see what happens. And we use some ideas from ecology to try and understand it. We were able to do things like compute. What's called a community matrix, which tells you whether the species are they competitive, meaning suppose you have species a and B or in this case, trading strategy, a and B if the wealth of trading strategy B goes up to the returns, the profits to trading strategy, they go up and vice versa. That would be what's called mutualism.
Doyne Farmer (22m 52s):
If the population of stray trading strategy B goes up and the returns of strategy, I go down and vice-versa, that's, what's called competition. And if it's as symmetric, so B goes up A's profits go up. But if A's wealth goes up, these profits go down, that's called predator prey, where a is preying on B. If you go back to the analogy of say lions and zebras, that's the way it's going to work for lions and zebras. Now we found several interesting things about our model ecology. We studied. One is that when we reached the equilibrium where the returns of all the strategies were the same, we actually saw that we had mutualistic interactions between all the strategies that is, if they're only one went up, the returns would go down.
Doyne Farmer (23m 41s):
But if anybody else's wealth, well-funded, the returns would go up that kind of us. But then we realized, well, that's actually maybe what you should expect at an equilibrium and an efficient place where the market's efficient in some sense. And that at the efficient place, all, all the returns to the strategies are the same. And if you deviate from that, then one of the strategies starts to have an advantage. Again, the way in which you deviate is you want the others to have more wealth than you have less wealth. It's like, no, the foxes do well when the rabbit population is high. So that was one of our insights. Another insight is that we're able to compute, what's called trophic levels for strategies that kind of tells you who eats whom and how do they lie?
Doyne Farmer (24m 24s):
Like with the lion zebra and grass ecology. Because if you assume that zebras eat only grass and lions eat only zebras, then the trophic levels are one for grass, two for zebras and three for lions, because they're trophic levels by definition, one higher than the thing you eat and the real world where there's more complicated diets, trophic levels can be more complicated. You can still compute them. And we're able to compute these in a financial ecology by looking at what happens when we knocked one of the trading strategies out to see how that changed the profits and saw that, you know, in the typical case, we had noise traders and at a trophic level, close to one and value investors at a trophic level or close to two trend followers that trophic level close to three, but that could change depending on the wealth of the strategies.
Doyne Farmer (25m 16s):
And in some cases, in fact, the trophic levels even cease to be, to find the key thing that we found is that if we want to understand why the market's malfunctioning, like why as volatility high, why are prices mispriced? Why are they straying from fundamental values? Then the wealth of each of those strategies in the ecology determine how mispriced the market is. And the system's own spontaneous dynamics can cause substantial excursions away from the equilibrium and substantial market malfunctions
Michael Garfield (25m 53s):
Elsewhere. In your book, you talk about leverage and how leverage causes the market to make its own news. This is an input into these imbalances. Could you go into a little bit more detail about this and you know, the, the ecology of leverage.
Doyne Farmer (26m 9s):
So there's really two, two kinds of effects that can cause the market to, as you said, make its own news. One is that the populations are the Wells. So different strategies changed through time and take it to places where, for example, you have de-stabilizing strategies like trend following that cause high volatility, or you can have leveraged through that and leverage as a kind of amplifier because under leverage means that investors are allowed to borrow money and buy assets with borrowed money, which might sound weird. But you know, most people end up doing that. When they buy a house, you put 20% down on your house, you just bought the house at a leverage of five to one, leverage, amplifies everything that is, if you have a leverage of five, your returns go up by a factor of five, but your volatility goes up by a factor of five.
Doyne Farmer (26m 59s):
But even more than that, leverage can force certain actions to be taken. If you have a leverage limit of say five and prices go down, then you're going to be forced to sell assets in order to keep your leverage at five, this kind of forest selling can de stabilize the market because it means that you can have a downward price glitch. And then if lots of participants, the domain tain, a leverage limit, they sell. So they're selling into falling markets. So they're amplifying the drops in the market, even creating instabilities, where they didn't exist before. And so you see this in several different models we've made where if you turn the leverage up, high enough things become unstable and you can get chaotic attractors and the dynamics, for example.
Doyne Farmer (27m 44s):
So you get spontaneous oscillations in the market. Like we have a model of what we call the Basel leverage cycle inspired partly by work of SFI, external faculty member, judge, and accomplice, and the Basel leverage cycle where, you know, we have an investment bank and we have a fundamental investor and the investment bank is borrowing money to use leverage, maintaining its leverage target, according to what Basel the Basel two agreement tells it to do and using historical volatility as an indicator of future volatility. If you turn the overall average volatility up high enough, you spontaneously get this chaotic attractor that consists of a 10 year slow rise in prices followed by an abrupt crash.
Doyne Farmer (28m 27s):
You know, the magnitude and amplitude vary because it's chaotic. But we see something that looks an awful lot like the great moderation and the great financial crisis just happening spontaneously from what was deemed to be prudent risk management by two representative investors, it makes its own volatility on a very long timescale on this case. So we have several different examples of models that do that, where the use of leverage creates these instabilities that can give rise to chaos or other noise amplifying phenomenon.
Michael Garfield (29m 1s):
I would like to draw a hypothetical link between what you just said and a conversation that I had with Jeffrey West last year on the show that has been keeping me awake at night ever since, which is his comments about the finite time singularity. And I just curious whether you think this is a valid bridge or not, he's got this insight that spills out of his work on scaling laws and the city as a social reactor. And this notion that, you know, as the city grows, that the interactions between people in the city scale greater than a linear rate. And so you end up in this accelerating crisis innovation that ultimately undermines itself because the crises start coming faster than the society can respond with innovations.
Michael Garfield (29m 50s):
There's a point at which you can no longer see the ball's coming at you fast enough to respond to them. And I've been thinking about this in terms of like, is there a way to escape this? And it sounds to me like what you're saying is no, because, and this is the part that I'm curious, whether you think this is valid or not. The externalities created by economic models represent a form of like off the books, leverage where, you know, if you're claiming, for example, that the ecosystem services provided by a Hector of Congolese rainforest are worth this then, because we know, like you said earlier that you can't make a complete model of the value you're buying in at a significant discount from the actual value of that rainforest.
Michael Garfield (30m 38s):
And that in a weird way, like maybe the more we try to swallow the spider to catch the fly with better and better modeling, we're actually creating precisely the kind of instability that you're talking about here is am I right?
Doyne Farmer (30m 53s):
If you're suggesting that our models become so good that they become self-defeating? Is that what you're asking about? In other words, if I have a really good model of something, then I, if I see it going somewhere, I don't want it to go that I take an action that will invalidate the model. I guess
Michael Garfield (31m 8s):
The simplest question would be, are externalities a form of unaccounted leverage that is making our efforts to predict this sort of inherently destabilizing?
Doyne Farmer (31m 22s):
I don't think leverage is quite the right analogy, but let me maybe don't let me riff on that in a different direction. We live in a world of increasing complexity. Things are changing. They're evolving all the time. That's inherently hard to model because it's a moving target and we don't know where it's going. We don't know what the, where the bulls-eye of that target is. It's the world is changing rapidly at an accelerating pace. And meanwhile, our technology for modeling it is also increasing rapidly. And the question is, do we have a parody in that arms race? Are we able to model the world well enough to keep up with the accelerating complexity of the world? And I think the jury's out, you know, in a sense the whole point of my book is we damn well better at least try.
Doyne Farmer (32m 7s):
And the only, the only way we're going to get there is if we have modeling technologies that take full advantage of 21st century technology, mainstream economics is using modeling ideas from the 19th and 20th centuries that don't make proper use and can't fully incorporate 21st century technology. So if we're going to survive and thrive in the 21st century and beyond, we really have to start modeling the world and the complexity of the world so that we have some guidance to help us understand what the right decisions are. And we need to come up with ways of governance that help us actually use that information, making our decisions.
Doyne Farmer (32m 50s):
We haven't been doing a very good job.
Michael Garfield (32m 53s):
Yeah. This seems like the appropriate time to bring up the point you make in your book, that in part, because of your clandestine shoe computer, the state of Nevada subsequently passed a law against using a computer to predict the outcome of a game. It's like, if that were, this whole thing would be under some sort of weird oppressive authoritarian regime. If you can imagine, you know, trying to outlaw the use of computers to predict how do they even play the stock market in Nevada. Now it's curious, this seems like an interesting time to pivot as you bring up the ratcheting complexity of the 21st century to pivot into some of the statements that you make towards the end of your book about applying this kind of thinking to problems such as climate change, how far we have to go before complexity, economics functions in the way that meteorology functions, this is maybe biting off more than we can chew in one section here, but the relationship between using complexity economics to understand our response, our adaptation to climate change at the planet scale, and then how this relates to applying this for like microeconomic interventions,
Doyne Farmer (34m 9s):
I'll take a crack at it and we can see where it goes. So climate change provides a good example where I think complexity economics is desperately needed. That's obviously a complex problem. There's a lot of subtle effects going on. It's not a zero sum game. We need to look around and see what solid things we can grab onto in modeling climate change. And one of the solid things that my work has grabbed onto in the last 15 years is technological change because ironically, you might think technological change predicting technological progress would be impossible because it's innovation. Maybe by definition, it's unpredictable.
Doyne Farmer (34m 49s):
The details of the innovations are difficult or impossible to predict. But one of the things we observed just from collecting data is that once a technology gets on a trajectory, it tends to stay on it for a long time. Moore's law being the most famous example, but there are actually lots of things follow more as like laws, solar, foldable techs, which have been on an improvement trajectory of around 10% per year, since the 1950s, meaning the cost to generate a kilowatt of solar power to electricity goes down at about 10% per year. So that actually allows us to make predictions about technological progress and compare things. And if you look at fossil fuels, they've been roughly at the same cost for more than a century prices, Bob, up and down.
Doyne Farmer (35m 34s):
You know, it's not that different from what it was more than a century ago. Similarly for coal, we showed this in a paper with James McNerney who was a graduate student at SFI for several years. And so part of what we predict, which is actually good news about climate change is that because renewables have been, are coming down at a steady pace, whereas fossil fuels and nuclear power are not, we can expect to see actually cheaper energy prices in the future. And we should expect the economy of its own accord to move towards renewables, just because they're cheaper. And we can assign probabilities to those events by analyzing other technologies with a model of technological progress that we've tested against those other technologies.
Doyne Farmer (36m 16s):
So we ended up having a very different prediction about the costs and benefits of making the transition rapidly than others, such as William Nordhaus, who won a Nobel prize in economics a few years ago have made now more generally what we would really like to be able to do is model the economy and its complexity and glory, you know, at the level of individual firms and understand how they're linked to the financial system, how they're linked to the innovation system through the record of the patent record and so on how they're linked to the occupational labor landscape so that we can think about things like the green energy transition and give good answers to questions.
Doyne Farmer (36m 58s):
Like what happens if we implement the green new deal? Does economy go to hell as the right suggests or does it actually work pretty well? And I think by having models that really represent the economy, as it is track things down to a fairly fine-grained level of detail, we can provide good answers to those questions. It's not a cheaper, easy enterprise to do that, but I do think it's feasible with 21st century technology and a reasonable scale effort to make models that could actually give us fairly reliable answers to questions like that.
Michael Garfield (37m 32s):
So you have this one paper, a rapid green energy transition is cheaper than a fossil fuel future that you coauthored with Rupert way. And penny mili, you seem to be speaking directly to this paper.
Doyne Farmer (37m 43s):
Yes, that's right. One of
Michael Garfield (37m 45s):
The interesting findings from this particular study is not only that a rapid green energy transition looks like it's an economic benefit, but that a nuclear scenario is substantially more expensive than a rapid or a slow transition. And so it's kind of funny how you've come around here in some sense to undermining the Los Alamos thing here. But I'd love to hear what exactly is going on in that scenario. And why does this research show that nuclear is not an economically viable future?
Doyne Farmer (38m 21s):
Sure. We do everything in that paper based on a principle that you can, can accept or deny, which is that we just make extrapolations based on past history. Now, the paper that that paper is based on, you know, we, we assembled a of 50 different technologies and look at histories through time. And we showed that actually past is indicative of the future. If something has been coming down, it tends to keep coming down that hasn't been coming now, it tends to not come down ever, or it fluctuates around a bit, but doesn't do much. Nuclear is a good example. You know, nuclear power and solar power came into being at almost the same time in the late 1950s. And since then, solar photovoltaic energy has come down by a factor of about 5,000 nuclear costs, about three times, as much as it did then per kilowatt.
Doyne Farmer (39m 9s):
You know, you can grumble that's because they put all those safety restrictions on nuclear power plants, maybe true, but there was no other forest bringing it down, certainly not by a factor of 5,000. And you know, we've looked now at data from lots of different countries, France, South Korea, South Korea. They've come down a little bit like at 1% per year, but that's not 10% per year. It's just a matter of what history tells us about technological change. There are, we believe some good reasons for these differences, but they're not well understood at this point.
Michael Garfield (39m 41s):
Yeah, just a moment ago, you touched on talking about this real granular modeling of the economy, looking at the relationships between firms and, and it calls to mind a paper. I just saw coauthored by Ricardo Hausmann, where he's using directed networks to examine the relationship between simple and complex exports and looking at like a different nations and, and how the economic relationships between different nations steer, the evolution of technology by providing a map of how these different nations are able to build on what is already there.
Michael Garfield (40m 22s):
And then you can kind of predict where the next combinations of things are going to be both in terms of the technological sector, what they're going to look like, and then also where they're going to happen. And so this is, this seems like a very different application of these kinds of trophic network models, but it seems like a really exciting one and one that in spite of your historical aversion to weapons development, I just have to ask about these kinds of tools where you, where you sit on all of this, if, if it's not too presumptuous.
Doyne Farmer (40m 58s):
No. So I think that's a V. So the paper you mentioned by Naval Clery and Ricardo Hausmann, it's a great paper. One of the building blocks in a line of research that Ricard has been pursuing for a long time. And it's just showing once again, that if you think about the process of economic development from a complex systems point of view, you're led to models with a lot of predictive value in particular economic growth follows standard developmental patterns that as you say, depend on things like the capabilities that countries can bring to bear, to make the products that they make and what they need to add to make new. It
Michael Garfield (41m 34s):
Just sort of folds into the sort of larger question you've already addressed about the arms race between different people, modeling the world, and then leveraging those models in competition. And so on.
Doyne Farmer (41m 47s):
Maybe I don't quite understand your question. Ricardo makes a predictive model about development. If countries start to use that model, will that invalidate the model? I doubt it. I think, I think the model actually can survive. Having people use it and it will just lead to better development. It should allow countries to develop faster because they're doing their development. It actually gives insight into the right way for a country to develop in order to make rapid progress, bring wealth and prosperity to their people sooner, rather than later,
Michael Garfield (42m 20s):
I guess this sort of ties into one of the three points in the final chapter of your book. We've already addressed one, which is a transition to a sustainable economy, at least an economy built on sustainable energy and resources. Another one of the points is about creating a world of equal opportunity yet elsewhere. You and many other people working in complexity economics have pointed out that there are sort of natural structural inequalities that emerge from this. When I had Brian Arthur on the show, you know, he gave an argument from scaling laws for universal basic income saying that, you know, at some point the system becomes large enough that it actually has to pump resources into its own capillaries, similar to how most ants in a large colony are effectively doing nothing.
Michael Garfield (43m 12s):
And any given time, you have to actively incense this kind of opportunity. I wonder to what extent the proliferation and refinement of predictive models do not themselves just sort of become a tool whereby existing inequality can be magnified by people capable of acting on these models.
Doyne Farmer (43m 35s):
Let me maybe rephrase your question. If the people who get access to predictive models are the rich can't. They then use that to manipulate the world, to make themselves even richer and perpetuate inequality. And I think that's probably true, particularly if they use them to do things like manipulate public opinion and so on. As I think we see with things like Fox news. On the other hand, if we can show conclusively that decreasing inequality actually is good for the overall growth of the economy, decision-makers will be incentivized to act, to take policies into account that decrease inequality. And as you said, a moment ago, it's true that there are intrinsic forces that create inequality.
Doyne Farmer (44m 18s):
If we understand those forces, we can also understand the degree to which those forces act, depending on the policies that we put in place. We have a range of policy measures, you know, minimum wage, universal income, traditional welfare system, income tax, as we play with those levers, how does that change the landscape? How does that affect the economy? So you, you potentially can, can study all those options and just see which one gives you better outcomes and hope that democracy will push you in the direction of those outcomes.
Michael Garfield (44m 52s):
To the point of shaping incentive landscapes. This speaks to a question that we got internally from, from Kaitlyn McShea, which was about the project that we started talking about, the, the desire to try and make enough money to bounce out of the system fund your own research through, through your gaming of the system. Her question was about your thoughts on the problematic system of incentives for scientific research and how it takes a lot of effort to get funded such that you end up having to literally game the system as you did, in order to actually ask blue sky fundamental questions elsewhere.
Michael Garfield (45m 38s):
I've seen Murray, Gell-Mann speak to this about skewed incentives and society actively punishing people involved in a synthetic or translational function. This is an essential role in a network ecology, and yet it seems actively disincented no, it's true. So I'm, I'm curious how you see this problem and what if any solution you see for it?
Doyne Farmer (46m 4s):
It's certainly true that the disciplinary silos that we have created have created a strange system where it's relatively easy to fund research that is making incremental steps around the questions that each of the disciplinary silos are focused on. But there are chasms that exist between these silos and exploring what exists in those chasms is very hard to do because there is no institution for funding work that investigates those questions. And in fact, they're active disincentives because people inside the silos want to keep all the money and they want to maximize the money inside their silo, and they have allegiance to the silo and they have incentives to perpetuate themselves within it.
Doyne Farmer (46m 51s):
So they're big, big disincentives for more interdisciplinary work. And for, just for any work that doesn't fall inside. One of the silos, it's unfortunate, it's bad for society. And I'm just speaking in somebody who's always found the question I want to ask exist in those chasms, because those are the places where we don't know things and the best questions are about the things we know the least about. How can we change that? Well, SFI has made a step towards changing that just by existing people, recognizing the problem and funding it out, may mostly outside of the mainstream, the NSF for awhile. And maybe they're still doing this. I'm not getting it funding anymore. So I don't know that a program whereby a certain amount of money was allocated that required investigators from different disciplines.
Doyne Farmer (47m 39s):
You had to have at least two PIs, and they had to be from distinctly different disciplines in order to even have a shot at getting funded. That was a revolution because it gave a leg up to us, interdisciplinary people. I got several NSF projects funded as a result of that, that never would have gotten funded without that. One could hope that somewhere in the future, we have a real discipline of complex systems. Arizona state is trying to create such a discipline. Now, university of Chalmers, and you have to more in Sweden has been doing this for some time. And if that becomes more widespread, then we may start to have money allocated where you actually get credit for the fact that you've proposed a question.
Doyne Farmer (48m 21s):
That's not a standard question. That's really different than the questions other people want to answer, and that it doesn't land inside of a disciplinary silo. But until that happens, I think we're relying on little tweaks here and there, like the NSF program.
Michael Garfield (48m 37s):
It seems like a decent time to dip into some of the questions that our Twitter audience was eager to ask you. Fabi on dabble lender would like to know what are in your view, the three best and three worst ideas in economics. Maybe we've already spoken to some of this.
Doyne Farmer (48m 57s):
Gosh, that's, that's a question that requires more reflection than I can probably do on the fly. I can't say I'm going to give the definitive list of it's my definitive list of the three best in three words that would require some reflection, but traditional economics works pretty well for things like designing an auction. I really liked the model that you mentioned that Ricardo just made, because I assume when we talk about economics, we're talking about not just mainstream, but complexity, economics, you know, the whole picture that's emerged of the way financial markets really work from the complexity economics point of view was a major accomplishment. The three worst, the idea that we are all rational, selfish, utility maximizers has caused a lot of harm in economics by encouraging things like neo-liberalism that suggest that laissez-faire capitalism is just the right way to go and will create good outcomes for the world.
Doyne Farmer (49m 53s):
The idea that you can model climate change with a simple representative macro model with unrealistic mitigation and damage functions is another seriously bad idea. I don't know what my, what my third candidate would be.
Michael Garfield (50m 10s):
Okay. I mentioned earlier you have three points about the benefits of complexity economics in the final chapter of your book. And one of them we haven't touched on yet. And so it just seems like practicing due diligence to invite you to talk about your vision for the central bank of the future, how you see these practices in place as competitive evolutionary forces, drive them into the mainstream of the practice of economics and finance. And, you know, after this transition, this transformation in the way that we think about and practice money and value, what do you see that actually looking like what is going on inside a central bank in 2050?
Doyne Farmer (50m 57s):
So I think I envisioned a central bank that's dramatically more technologically sophisticated than current central banks, where banks collect data about the economy in real time, at a fine grain level and have models that simulate the economy at that level and allow them to think about the policies that are enacting and how they will affect the future. A good example. That's doable right now with things we know how to do is understanding systemic risk in the financial system goes back to the discussion we had about leverage and vision. This financial system is a multi-layer network where the nodes are the key financial institutions, big banks, and the interconnections are their interactions.
Doyne Farmer (51m 40s):
So money they've lent to each other common portfolio holdings being two most obvious examples. Every time one of those banks does something, it affects the other ones. If a bank sells an asset, the presses, the price of that asset, which affects every other bank that holds the same asset. Similarly, when banks default on a loan, for example, then you're going to get a cascading failure mode where one bank defaulting causes another bank to the fault as they all start trying to call in their loans and potentially not paying them back and these different channels of contagion interact with each other. We, and some others have papers where we've shown how one can really very effectively model these things.
Doyne Farmer (52m 21s):
You can really model the way the contagion cascades around the economy and much the way that you can model how the COVID epidemic will cascade. And not that those models will be perfect, but they will really allow central banks to think in a fairly quantitative manner about resilience and stability of the financial system. When new changes come in, like we've been doing this in the early part of the millennium and you see a new species like mortgage backed security trading coming in, then you can ask, is this destabilizing the financial ecology? Ultimately what I'm saying is central bank should be able to track the financial ecology, just the way a, you know, jungle ecologists might measure the abundance of species in the jungle and track that ecology and think about their interactions in a similar way.
Doyne Farmer (53m 12s):
Central banks can do that and we know how to do it. Now. It's just a question of throwing enough effort at it and, and making sure that banks are really empowered to collect all that data. In real time, they could have been presenting results to key decision makers in 2005 saying guys that financial systems headed for a bad place. I think it was quite apparent. And 2005 or six, certainly that, you know, leverage was out of control. You could see that if a little glitch in housing prices would cause the whole thing to collapse. And, and you could have seen that in the models. Hypothetically, in my book, I opened with a story about the fed. In 2006, they asked their best model what's going to happen.
Doyne Farmer (53m 53s):
If housing prices dropped by 20%, it said not much. So that model didn't any way capture the fine grade ecology of the financial system as it really exists.
Michael Garfield (54m 3s):
That's an interesting point about the future practice of economics, which in the same way that science fiction is always just talking about present day issues. I agree with this, but it's just worth pointing out the meadow, which is that it seems each of us have a stake in the future, looking like our strategy wins and that this is a strategy that looks very much like the kind of interdisciplinary work going on. And I know at Oxford and SFI and Los Alamos, where you talk about there being a teams within each bank that are specializing at different layers of granularity for large firms, small businesses, the entire financial system household level.
Michael Garfield (54m 47s):
And this looks a lot like the way that David Kenney and Tyler Millhouse have talked about the disciplines of science emerging out of different layers of fine graining. And so it's perhaps a Testament to the empirical validity that this is the natural response. You start to see a fractal structure in the way that we actually study these flows.
Doyne Farmer (55m 11s):
It's certainly true that almost by definition people, advocate for the things they believe in, I'm going to advocate for complexity economics. And I do that because I think it's the right thing to do. The only thing I can say there is, it's a bit different in that if I just decided to play ball with a mainstream, then I would probably be much better positioned for being part of it and receiving the accolades that it Accords to itself rather than being the outsider challenging them, which is not as profitable a role at least in the short term, but it's just what I believe in. So I've got to do it. And that's what I think is the right thing to do all your second point. It's certainly true that you know what the central bank of the future, I imagine people would be specialized.
Doyne Farmer (55m 55s):
You need teams of people that interact with each other, where different members of the team have very different kinds of expertise. Because if you want to understand households, you have to really gather data on households and firms is a different story. And we need specialists in both of those. And we need models that can really fully incorporate the richly textured information and differences between those parts of the economy and ways of checking to see whether the models that we make for those parts are really accurately representing what the economy does. It's reminiscent actually heard Simon had his theory of the firm and arguing why you get hierarchies and firms. You get hierarchies because of information flows and the need for local information processing.
Doyne Farmer (56m 39s):
Versus since we aren't rational, not everybody can do everything perfectly. So you have to specialize domains of competence. So we do need disciplinary things. We do need specialization at the same time, we need more generalists to cut across and cross-fertilize things. So we need both of those at once. The system, as it's set up has tended to favor the specialists over the generalist. And as part of what SFI has been about changing is to facilitate those interdisciplinary dialogues by people who strive to be generalists.
Michael Garfield (57m 10s):
So to that point of encouraging the diversity of a given ecosystem, I guess maybe the last question I have for you is completely off the rails as far as any of the work that I've seen of yours. But I know that this is a, an area of research for SFI and of modest interest. I would say, how do you understand what's going on in the space of cryptocurrencies and its relationship to the incumbent economy as it is today? Because, you know, obviously it doesn't seem like you have any doubt that central banks will persist and you know, your work as we've already discussed in this episode makes the case that people that are just following other people's trades as seems to be largely the case in the cryptocurrency markets is for sure contributing to its enormous volatility.
Michael Garfield (58m 4s):
It's not just this small market cap of these markets. I mean, it seems like perhaps a small amount of insanity is actually good for the system in the sense that small quantity of poison has medicine. But I'm just curious how you, how you see the relationship between the, in some cases, stifling and counterfactual, sanity of the incumbent system and this sort of creative insanity of this next generation of like decentralized finance and so on
Doyne Farmer (58m 35s):
Cryptocurrencies are fascinating. I don't think it's completely clear where that's going so far. Most of the cryptocurrencies don't do a very good job of regulating supply and demand. And so, hence they've been hugely volatile and they're not backed by a bigger entity. I mean, one of the nice things about a dollar is it is backed by the earning power of the U S taxpayer. So the whole economy sits behind it, which makes you think it's not going to just disappear or evaporate overnight. Bitcoin is a fascinating example of a self-organized phenomena that plays into some peculiar mixture of illegal activity and the quirks of human psychology. You, it has led to some very interesting technological improvements like blockchains, which have uses that go far beyond cryptocurrencies and potentially could be used by governments in lots of other kinds of applications where you want to certify the authenticity of something in a sort of distributed holographic way.
Doyne Farmer (59m 34s):
So going to be very interesting to see where that goes. It's not obvious to me, whether in 20 years we'll look back on, Oh, this is a cryptocurrency or a, was it that funny, or will cryptocurrencies really come to play a major role in global finance? I'm sort of betting on the former, but I think they may also evolve. So cryptocurrencies are either picked up by governments and deployed, I mean, blockchain and some of the other technologies that are used in cryptocurrencies will be picked up. So they're digital currencies for nation States or other entities, but we'll see
Michael Garfield (1h 0m 13s):
Possibly a prerequisite technology for the kind of granular economic surveillance required of this hypothetical future central bank.
Doyne Farmer (1h 0m 22s):
Well, I'm not sure. I mean, you know, the thing is right now, central banks could do a lot of things. Most of them don't do, some of them are starting to do it, but like, you know, in Chile every time somebody swipes a zero Stripe, it gets sent to the central bank and they know who at least one of the counterparties work, it says some customer bought something in the store or, you know, two companies exchange an invoice or receipt that also gets sent to the central bank. So they're able to really track the flows in the economy in a much more detailed and real-time manner than we do. And it's just a fairly trivial change in the value added tax laws and you know, a little bit of extra technology, other countries like in the UK, they leave out key bits of information.
Doyne Farmer (1h 1m 8s):
So you can't do that. They have a value added tax and are keeping records, but they just don't keep a sufficient record. So I think there are a lot of areas where even without blockchains or anything fancy like that central banks could be tracking a lot of stuff that they don't track at this point in time.
Michael Garfield (1h 1m 21s):
Well, this has been fascinating. Thank you, Dwayne. Just as a parting shot, I guess, what of importance do you feel that we we've left on the answered? What stone have we left unturned that you feel like is worth people thinking about in this area?
Doyne Farmer (1h 1m 39s):
I don't know. I have been sufficiently occupied trying to answer these questions that I have. I'm just reflecting back there's anything in particular we haven't discussed.
Michael Garfield (1h 1m 49s):
What's the most fruitful area of unanswered inquiry for you right now, looking forward into, you know, future research projects?
Doyne Farmer (1h 1m 58s):
Well, I think at this point, actually, I'm at a point where I'm bringing a lot of different things to fruition rather than starting new new directions. And I think we've covered most of those. Excellent. I mean, I could comment on something else. I don't know if this is relevant for SFI talk, but I have decided that the only real way to demonstrate complexity economics is the right way to go in a definitive way is to actually do it commercially. That's the only way I think to get funding on the scale that's needed. It should be able to show that if done well, the complexity economic methods can be more effective at predicting things. So that's my next big thing.
Doyne Farmer (1h 2m 39s):
I started a company called macrocosm and we're in the process of refining a business plan and, and we'll be in the process of raising money probably within six months. That's exciting and new direction. Awesome. Well, everybody, the complexity economics revolution should be out by when it'll be out a year from when I, my publisher says, gives me the green light that it's really done. I hope that will happen within about three months. I've been eternally optimistic and made consistently bad forecasts about that question. Well, in the meantime, folks, you can find some of his work in SFI, press complexity, economics, volume books have been written about.
Doyne Farmer (1h 3m 23s):
You will link to a lot of this stuff, as well as some of the talks that you've given at SFI in the show notes. Thank you Darren, for taking the time. It's been a pleasure.
Michael Garfield (1h 3m 34s):
Thank you for listening. Complexities produced by the Santa Fe Institute, a nonprofit hub for complex systems science located in the high desert of New Mexico. For more information, including transcripts research links and educational resources, or to support our science and communication efforts. Visit Santa fe.edu/podcast.