I'm here with Doyne Farmer,

who is a SFI external faculty,

and he's also a professor at
Oxford University in England.

Doyne, can you tell us a little bit about
what your position in Oxford is?

Well, I'm in the mathematics,

I'm a professor of mathematics,

and I'm also the director of
complexity economics

for the
Institute for New Economic Thinking

at the Oxford Modern School.

Okay, that's good!

So, tell me about how you got into the

field of complexity, what brought you here.

Well, complexity is something I was

thinking about when I was in college, and

you know, couple of friends of mine and I,

other physics majors, would go out and
sit on the lawn after physics class,

and instead of talking about physics,

we were really talking about complexity,

which didn't have a name at the time,

and we didn't think anybody could study

stuff like that, and

then when I was in graduate school,

studying physical cosmology,

I learned about chaos and
dynamical systems,

which seemed like the closest

thing you could get at that time,

so I dropped cosmology for that,

and then I discovered that there were

other people who were interested in

thinking about similar things, and I

ended up at Los Alamos, where one could

think about such stuff, ironically,

a weapons lab is a place for intellectual

freedom to cross disciplinary boundaries

and think about complex systems.

And then things evolved, I actually

got asked in 1988, to start a group,

in the theoretical division, the group
was basically for the people that didn't

belong in all the other groups because
they were doing funny things,

and we at the time thought about calling
it the Likened Intelligence Group,

but we decided that was too constraining,

and so, we decided complex systems

was a good, you know, general purpose

thing where we could do what we wanted to,

and we were very interested in emergent

phenomena, and so on.

And it also happened around when the

Santa Fe Institute got going, and
the name caught on.

Okay, so your background was in physics,

you did chaos, dynamics, and so on,

how did you get into economics?

Well, so that's another curious
coincidence,

one of the things I always viewed as my

day job was my ability to predict stuff.

And I got into that because of Roulette

when I was a graduate student in physics
at UC Santa Cruz,

I teamed up with Norman Packard,

and we used physics to beat the game of
Roulette.

And then chaos always seemed like,

well you should be able to take advantage
of chaos to make better predictions,

even though it's about not being able
to predict.

And so, in 1987, I wrote a paper with

"Sid" Sidorowich, on basically,

building nonlinear models to make

better predictions in the presence of
low-dimensional chaos.

And, as a result of that, we started
predicting lots of stuff,

like transition fluid flows, and ice ages,

and sun spots, so we had a collection of

things where we were making pretty good

predictions. And everytime I would give

a talk about this, some clown would say,

"Well have you tried applying this

to the stock market?"

I got tired of hearing that, and it happened that

I'd also been at
Los Alamos for almost 10 years,

and they were going to give me
a nut dish

to commemorate my 10 years of service,

and that freaked me out,
and then I decided I needed to leave,

so I quit and started a business

to predict the stock market.

So that got me into reading, into actually

trying to understand more about finance,

which has to do with economics,

and then after 8 years, I had a
prediction company,

again with Norman Packard.

Once I met my financial goals,

and I felt like, okay,

what am I going to do now?

So I decided to merge my background in

complex systems, with what I'd learned,

my domain knowledge about financial markets

and economics, and then I came here

for 12 years as a faculty,

and worked on that here.

And then went to Oxford to do Economics
And then I went to Oxford to continue

so, I'm actually doing the same thing for
more than a decade now.

Wow. So, I don't know if you remember,

but when you first got into the field

of economics, and started like,

looking into those models and so on,

is there anything that particularly

struck you as surprising, or weird, or...

Well, I think what struck me as surprising

or weird is just how differently

people from different disciplines,

think about the world, and that the

paradigms are different, the whole middle

constructs, that underlie, that are viewed

as the foundational way people view
the world, are different.

And really, this
Could you give an example?

Yeah, you know a complex systems person
would naturally start thinking about,

"Okay, let's break this thing up into its
pieces,

and let's do a computer simulation,

and let's look for emergent phenomena."

That's sort of the paradigm of
complex systems now, I would say,

or at least one way to view it.

In economics, it's, "Well let's think about,

let's try and understand what the beliefs

are of all the agents, let's try and

understand what their goals are,

their utility, their preferences, and then

we'll make a model in which we assume

that they optimize their preferences,

subject to their beliefs, and we'll assume

that there's a fixed point where that happens.

So that's the fundamental model
in economics.

It's very different than the one in
complex systems.

So the one in complex systems doesn't

assume a fixed point
Doesn't assume a fixed point, it assumes

that simulating stuff on a computer is
just fine,

if you can get some analytic results,
great,

Economics, if you want to publish in a
top journal,

you'd better get some analytic results,

at least if you're doing theory,

so it's just a different mindset,

and a different notion about

what questions you should ask,

how you should go about answering them,
what kind of answers are acceptable,

So what counts as true science
Yeah, what counts as good science

So you know, you said you've been doing
this field,

which I think a lot of people

refer to as "Complexity Economics" now

for a long time,
Yeah

Have you seen that shift that those views

I think it is shifting,

I think it is shifting.

Economics is changing,

economics is a field where the

scientific method is evolving in time,

and I've seen it evolve in my own lifetime,

in the seventies, the hot stuff

would have been models with perfect
rationality,

and by the nineties,

people were beginning to start to think about behavioral effects,

and take some psychology into account,

and even do experiments.

Early experiments back in the sixties,

but that was very low-brow stuff at the time.

Those things that broken through,

so economics is changing,
it's becoming much more empirical,

which I think is a really good thing,

and there's a little, some cracks in the

door towards openness, toward things like

agent-based modelling,

though so far, I would say that

it's almost impossible to get a paper

based on an agent-based model,

in a top economics journal. Still,
Still.

What effect did you think the

financial crisis of 2008 have on the field
of economics?

So it's had a big effect on the field of
economics, since 2008.

It shook the profession up.

You know, Duncan Foley, another SFI

external faculty, and I, wrote a piece

in Nature in 2009, where we were being

a bit snoddy, I'd have to say.

But we said that the existing set of

theoretical models were
not even good enough to be wrong.

And what we meant was that,

if we look at the dynamic stochastic
general equilibrium models

that several banks were using to try
and forecast

what was going on through the crisis,

those models didn't have the banking

system in, the possibility of default was
not there,

all of the machinery that the

crisis was about was not even modelled,

you weren't even testing the theory

to see if it was right or not.

Now since then, things have really changed.

People are struggling to try and create

models that have these elements in them.

I'm still critical of what's being done,

because the models continue to be,

in my opinion, too simple structurally,

to capture the key elements of the economy

that need to be captured.

Networks weren't part of the modelling.

They aren't actually in the production
of models, though

though they have been studied more theoretically

by economists. And there are lots of other

structural things that are not present

in the models, and that I think are going

to be very difficult to put into

the current kind of models.

And I think the fundamental assumption of
equilibrium,

that's built into existing economic models,

is simply inappropriate.

You have to let go of that

to accurately capture the collective

oscillations that happen in the economy

and the endogenous movements that
I think drive things

like the business cycle.

And so there has to be a key change
that has to happen

in the fundamental way models are put
together,

in my view, before they can
start to really go beyond

what's being done now.

So what scientific questions really
grabs you these days?

What are you focusing on?

Well, I'm really focusing on two things.

One is trying to make agent-based models

that do give us more accurate
representation to the economy,

and trying to do that
in a very quantitative way.

We're in the process of putting together
an open-source project.

We're still trying to finalize an aim for.

I'm sort of leaning towards an economic
simulation platform,

but my colleagues and I are still debating this,

so we're going to have an open-source project,

where anybody can put their agent-based models,

anybody can contribute to building new models,

we're going to try and build the
right set of lego blocks

to put together economic agent-based
models.

And we're trying to do this in a very

structured way with a lot of focus on

calibration and validation of the models.

The other thing I'm really
interested in these days is

the evolution of technology,

and in particular in creating a evolutionary
model of technological change.

Evolutionary in the sense
of Darwinian evolution?

In the sense of descent
with variation selection,

which I think is the fundamental driver

that's shaping the way technology is,

the broad principles, I think,

are shared with biological evolution,
in the sense of Darwin,

the details are all different.

So, I think it's a wide, open field

where a lot remains to be done.

You know, technology is

an ever larger part of our world,

that I think is just crying out for more
fundamental understanding,

both for practical reasons,

because we're facing something
like climate change,

well, technology made the problem,

technology is our only
way out of the problem.

And the right technological bets

are going to make a big difference.

And so I'd argue that nuclear power

for example, is not the right
technological bet,

okay, not just because of all those
safety considerations and so on,

but because there's no history
for progress in nuclear power.

Cost of nuclear power
now is the same as it was

in the 1960s, or higher actually.

Whereas other technologies like

solar energy have been dropping quite fast,

and there are underlying reasons for that

that we need to understand better,

there's very interesting regularities

in how these changes happen,

all have policy relevance.

So that's one side, and the other side is,

I think it's fascinating to,

for the unity of knowledge to try and
put it together with Darwinian evolution,

and try and understand,
other similarities and differences.

So, going back to your focus on prediction,
Yeah

do you think it's possible to develop a
predictive model of technological evolution?

Well, we have a paper in research policy,

called how predictable is technological
progress,

and we show in there that just simple
extrapolation models

provide predictions whose accuracy can be
quantified,

that actually are good enough

to be quite useful for making policy
judgements.

Like, for instance, if you want to ask the
question,

if you say, "Well solar energy has been
improving over the last 40 years,

what's the likelihood it will continue
to improve at a similar rate,

over the next 20 years?"

At which point, it should be quite a bit
cheaper than coal,

in terms of electricity price.

And, so we can put a quantitative answer
on that question,

and know that the odds are that it will be
more expensive in 20 years than it is now

are a few percent.

The odds for nuclear power,

from the same analysis,

are about 50 percent.

And so, it's just using statistics.

I think you could also,

I think you could do better, ultimately, by
having a proper theory

where you really understand why things work.

The prediction I just mentioned is made
by just analyzing lots of technologies

and looking at their patterns and improvement

So it's more like looking at data to build
statistical models,

rather than having a real model
of the mechanisms.

That's exactly what it is.

But I think one can go beyond that,

to make better models,

and it will also give,

by giving you a more fundamental
understanding of what's happening,

provide deeper insight in lots of other ways.

Alright, well, that's really interesting.

Thanks a lot, appreciate it.
You're welcome.

Thank you.
Thank you.