Interesting People mailing list archives

Previous By Date Next
Previous By Thread Next

Re: The Economic Crisis and its Implications for The Science of Economics

From: David Farber <dave () farber net>
Date: Mon, 27 Apr 2009 18:34:26 -0400


Begin forwarded message:

From: Thomas Lord <lord () emf net>
Date: April 27, 2009 4:37:50 PM EDT
To: dave () farber net, Bob Frankston <Bob19-0501 () bobf frankston com>
Cc: ip <ip () v2 listbox com>, drallison () gmail com
Subject: Re: [IP] Re: The Economic Crisis and its Implications for The Science of Economics 

Bob,

I think that there's a simpler, more conventional
way to get at what you are getting at.  Perhaps
I'm misunderstanding you but I'll take a shot:

Instead of framing things in an evolutionary
context, a good way to frame the implications
of uncertainty about the future is in terms of
resilience and its market value as wealth.

We're used to thinking about the resilience properties
of systems with concepts like distributed decentralization,
redundancy, tolerance of components to a wide ranges of
operating conditions, graceful failure modes, and so forth.
These concepts aren't too foreign to just about anyone
from any background, even if the vocabularies aren't always
the same.  They're intuitive.

There's even a little cottage industry growing out
of the current war against terrorist extremism and
militant fundamentalism dedicated, in no small part,
to a kind of systems study of resilience.  It's motivated
by the thought that we have to out-compete our new
distributed, decentralized, highly resilient enemies
in those regards but our typical developed-nation infrastructures
are way behind (in everything from water and power to, yes,
the financial system).   I guess that fellow "Jon Robb"
is a fine public example of one of those thinkers.

Within that "resilience is fundamental" framework, why,
yes indeed theories of natural selection do emerge as
a handy analytic tool.   And, yes, indeed, they offer a
kind of retroactive explanation of why a heavy emphasis
on resilience is a good idea - it's quite the convincing
little hermeneutic there.

But the point is that "resilience" and the design
patterns that help create it are already commonly
recognizable.  It's latent in "common sense".   You
don't need to get people to buy into some radical
shift in world view - only to get them to change
valuations.

To a classical economist we might argue that the
markets as they stand deeply under-value resilience.
People don't pay a premium to obtain it - they'd rather
hedge in other ways, for the most part.  And that's
a bug and it's a bug you can put a dollar figure on.

Indeed, the current financial crisis is in no small
part a reflection of loss of confidence combined with
some actual failures of "falsely resilient" systems
like the CDO and CDS markets.   People bought CDSs,
for example, with the ostensible rationale that
they increased the resilience of a firm's books but,
clearly, they didn't think that through seriously enough
and got it badly wrong.   This is a teaching moment.

I think it was you but perhaps it was someone else
in this thread that called out for a greater emphasis
on "lots of small projects" in contrast to big,
monolithic, all-eggs-single-basket projects.  That's
really easy to understand in terms of common sense
knowledge about resilience.  Sure, there's an "evolutionary"
angle on it that adds depth but it is much simpler to
explain in terms of resilience.

Where I think the dismal scientists could pitch in
is with some creative thinking about how to
finance "lots of small projects".   Lots of small
projects implies fewer economies of scale, lower
and less reliable returns on each project, and a
significantly higher cost identifying, vetting,
and participating in investments.

There is a kind of paradox in that big spenders who
undertake "lots of small projects" are, historically,
likely to wind up with a portfolio that very often
under-performs larger markets - perhaps even
under-performs treasury rates.   The main "dividend"
to potential investors is simply a more robust
playing field - a functioning society in which to live
and aim for bigger gains.

And that kind of dividend is not to be underestimated
but there is a kind of prisoner's dilemma / tragedy of
the commons, there.  If I dedicate 80% of my capital
to that kind of project but you just ride on my back
and go only for the big windfalls, then eventually either
I'm screwed because you're so much richer that I'm powerless
or else we're both screwed because a black swan exposed
the consequences for society as a whole of your underinvestment
in resilience.

I don't know how to fix it other than, by god, hacking
the ethical norms of rich people.   If a few believe in
resilience-oriented investing they should do everything they
can to not invest with and not cooperate with and even
throw up obstructions to other rich people who neglect
such things.  If a critical mass is reached then the lazy,
selfish, irresponsible opportunists won't find enough trading
partners to make it worth their while.

(Hey, you flush?  I think it would be interesting to
examine the needs of my (relative to the area) low-rent 'hood
and bootstrap a bunch of small projects in relation :-)

Anyway, "resilience" is the concept I think you'll find
people more generally able to intuitively grasp and the
evolutionary analytics are just esoterica on top of that.

-t






On Mon, 2009-04-27 at 04:10 -0400, David Farber wrote:



Begin forwarded message:

From: "Bob Frankston" <Bob19-0501 () bobf frankston com>
Date: April 26, 2009 10:32:39 PM EDT
To: "'Dennis Allison'" <drallison () gmail com>, "'Eric Weinstein'"
<erw.phd () gmail com>
Cc: <dave () farber net>, "'ip'" <ip () v2 listbox com>
Subject: RE: [IP] The Economic Crisis and its Implications for The
Science of Economics


I realize this response is a bit long (yet frustratingly short) so
I’ll start out with an overview so you don’t need to read it all and
then I’ve got the notes I made as I read some of the documents that
Dennis cited.

Dennis, thanks for the references – they gave me a useful framework
for my responses. The essential issue is in how we frame the problem
statement. I agree we can learn from “Limits to Growth” and the many
other examples. I want to be very clear – I’m not saying that these
approaches are totally useless. It’s just that we need to understand
the limits rather than simply trying to refine them.

My assumption is that the future is inherently unpredictable and that
we have a complex soup of possibilities at any one time. The challenge
is to figure out why things seem to work so well despite the chaos.
Even more surprising is the degree of progress and even hypergrowth
according to measures we like. I look at system (and markets) and try
to understand how and why each works as it does and in what ways it
fails.

We need to be careful as we tend to find patterns and the definition
of “works” can beg the question. There are many ways in which things
are seen to work and there are many that we view as not working – just
by how we interpret what we observer. My central thesis is that some
market configurations “find” order amidst this chaos because of their
ability to capture what works and survive what doesn’t and I call this
the degree we can make the sharp distinctions and capture nuance a
digital property. The found order is really an interpretation of what
is observed. This gets self-referential which is why implementation is
an important anchor. In fact, writing this response is the kind of
forcing issue I need to organize my ideas against a particular
context.

As a programmer and in business I’ve learned that implementing is
applied modeling and you learn a lot in the reality of failures as
well as from the freedom and need to challenge the givens. To use my
roulette example, I’m able to focus on taking advantage of the number
we happen to have whereas economists seemed obliged to predict what
number will come up.

Obviously there’s a lot more to it than that and I don’t want to be
dismissive of economists (tempting as that may be) because there is a
real role and need for modeling. My concern is the degree of hubris in
confusing the models with reality. There’s also a tendency to try to
make things work rather than accepting that things generally do work.
Though we can use our understanding to deal with dysfunctions relative
to the results we want.

I do tend to jump too quickly to implementation (one reason I never
finished my doctorate). This is why I am so interested in the things
we call telecom and the Internet. My approach is a result of applying
these basic principles to real systems. Telecom is one of the rare
cases of high value problem with a simple solution. We don’t need no
stinkin’ economic models – OK we do need models but ones that aren’t
stuck on false “givens”. In this case the models themselves perpetuate
the problem as we cling to them instead of checking back against
reality.

The models in “Limits to Growth” are limited by not being able to
incorporate the unanticipated. But the approach does demonstrate the
complex ways in which systems are coupled. Another more recent lesson
is in Leonard Mlodinow’s The Drunkard’s Walk about the history of
probability and statistics. My take-away from that book is that analog
systems can’t help but turn to mush because they don’t preserve the
sharp distinctions (which I call digital) necessary to preserve sharp
results. As an aside, we get learning curves by taking the important
digital distinctions and losing them a static stew.

I was recently at an MIT talk about climate models (appropriately part
of the Darwin presentations) and the same problems frustrate attempts
to make definitive climate predictions. That doesn’t mean we should
ignore them. You’re right – just because the models can’t take into
account innovation doesn’t mean that we can always count on
innovation, especially with a system with as long a baseline as
planetary climate.

I argue that while models can be useful we can’t have a deterministic
model which is why I’m advocating a different approach.

Going through the references I’m looking through Kauffman’s Scientific
American piece and I see attempts to create models and find patterns
but I don’t see an evolutionary approach (though does take one in his
book). The article looks at Creative Destruction statistically without
a deeper sense of capturing success and surviving failure. It’s
treated as a single system that goes into phases. In fact he writes
about “the statistics of Schumpeterian gales of creative destruction”.

As a relevant aside, the SA page had references to the power grid
which made me think of the “smart grid” and conservation. When
preparing dinner last night I noticed the amount of plastic packaging
for food and thought about all the effort to get people to bring their
reusable bags but nothing about bringing in reusing the plastic
containers (recycling isn’t the same thing). It’s as if we are going
through the motions of conservation but without a larger view of
systems. It’s another reminder of evolution at work – you can champion
the bag issue but it’s far more difficult to find someone with purview
at a systems level.

Again, we could argue for better modeling but how can we have models
that take into account all these issues? Instead we need to think
about refactoring systems. We’ve done pretty well relying on random
innovation to address issues and we get solutions but if we were able
to be more explicit we might do better than rely on the whims of
markets – as long as we’re modest in our expectations and are willing
to accept unexpected consequences.

OK, continuing to comment as I write and looking
at http://www.edge.org/3rd_culture/brown08/brown08_index.html. Here
too I find the economy treated as a single system though Taleb’s black
swans are a way of characterizing the discontinuities that are so
important. I agree with Eric’s comment about our tendency to add
epicycles to models. Alas, there is no fixed reference point like the
Sun that can center a model.

I do think about why Wikipedia and EBay work – at least within an
operational definition of work. Sure, it’s useful to assume that
people are basically “good” just like we can assume that it’s usually
safe to cross the street but in reality those systems have mechanisms
to self-correct and it’s that kind of “trust but verify (or
work-around)” that allows us to take advantage of the “good” while
surviving the “bad”. I use quotes because I want to emphasize that I’m
using those words in an operational sense rather than moral sense.

I see Mike Shermer picking up on the evolutionary theme head-on. He’s
right about the futility of managing the economy – my challenge is
trying to make actionable suggestions including but not limited to
benign neglect. This is why I look at policies like decoupling
markets. In health care for example we can make great advances by
simply capturing information without having to pile on a coding agenda
or a monetizing agenda.


I like Romer’s point about the danger of depending on complexity
science (or economists?) to prevent forest fires. But that example is
a very good one because we’re now learning that preventing forest
fires is the wrong goal as those fires are part of a larger system.

I’m now going
through http://physicsworld.com/cws/article/print/38468 and it’s far
more interesting and here too I agree with Eric’s comments on
risk-averse funding for science. I see this first hand as my son is
being taught how to write wimpy grant proposals in the post-Proxmire
world. More to the point the article does emphasize the importance of
embracing the unpredictable.

http://www.versaggi.net/ecommerce/articles/romer-econideas.htm makes a
lot of good points but one comment stand out

"The economic problem," Romer says, "is really about configuring all
our institutions so that we search efficiently through this space of
possible ideas, finding better and better ones."

What is missing is the what Seymour Papert calls the powerful idea or
an acceptance of the unpredictability of the black swans. You can’t
“search” for an ear – what you do is start relying on the vibrations
picked up by a gill bone and then you discover you have an ear. You
can’t search for an ear until you have the concept of an ear. There’s
also the question of the measure of “better” – not just the lack of
metrics but the inherent ambiguities.

There’s also a mention of antitrust but I’d rather think of it in
terms of decoupling rather than control of existing markets according
to necessarily backward looking measures. Hmm – now it gets personal
with the Lotus copyright suit … I could tell you stories but let’s not
get diverted.

The mention of free trade reminds of me Bad Samaritans which makes an
interesting case about the dangers of globalization in stifling new
discoveries by suffocating them in a forest of big business. While I
don’t agree with all that Ha-Joon Chang says he does make an important
point. The coupling of the world’s financial systems into one big
eco-niche is a glaring example of how the process can go awry. That
will lead to the topic of the concept of money as a measure but I
don’t want to get too side-tracked as interesting as the topic is.
Along these lines there’s David Landes’ The Wealth and Poverty of
Nations which he observes that economies that tolerate chaos do best.
http://seedmagazine.com/stateofscience/sos_fundamental_money_p1.html is also interesting but too much to comment on – I do think that science can be considered to be another business/market model. This isn’t to put down science but to generalize the concept of business models and market configuration in terms of motivations and interactions. If we didn’t romanticize the purity of science we might have a better way to counter Proxmire-phobia. We’d not only fund more risky research but we’d appreciate the externalities in education. 

http://www.nytimes.com/2009/03/10/science/10quant.html -- well I’ve
already written about my memories of IDC when Black-Scholes numbers
were introduced in 1973. Too bad people naively treated them as hard
numbers.

Which brings us to Reinventing the Sacred: A New View of Science… I
bought it on the Kindle because of the accidental property of
immediate delivery – the Kindle itself is another story.
First, http://letterfromhere.blogspot.com/2006/12/bohr-leads-berra-but-yogi-closing-gap.html -- I appreciate Kauffman quoting Bohr whereas I glommed onto Yogi Berra. Perhaps his choice is right but it also may indicate he’s been hanging around with physicists or I’m too attracted to the social story. 

His use of “preadaption” is similar to my use of “powerful idea”
though I am wary of the determinism implied by pre-adaption. More
interesting is his reference to the “learning curve”.
In http://frankston.com/?Name=BeyondComputing I point out that the
learning curve emerges by ignoring the details. If you peek underneath
you’ll notice a self-clocking process consisting of disparate steps
whose result only exist to the observer. You can make some predictions
such as I did in anticipating multi-core processors but that was
pretty obvious to all. But in terms of the learning curve it wasn’t a
linear speed up since those processors are faster only if you accept
different measures of speed as opposed to linear performance.

My bigger issue is with the emphasis on fancy (or fanciful) models,
especially about the future (J). What seems to be missing is an
appreciation for DIY (Do It Yourself). Maybe we need to wait for a
generation or two of programmers who are used to building new things
and then using those things as building blocks and repeating the
process – as in the class instructions “Lather, Rinse, Repeat” found
on every bottle of shampoo. We’re also used to finding solutions that
are solved by the tools we have and working around problems or
satisfizing and we’re very well practiced in the art of illusion.

Without the physical limits of waiting till the next crop comes in I
can see ideas iterating and combining and morphing rapidly. The
question is not why do we get an explosion of possibilities but why do
they form a seeming coherent whole. The emphasis is on “seeming”. I
argue this coherence and equilibrium is a misinterpretation and,
perhaps, a very hold theme from the days when things changed slowly
and events seemed more cyclic (seasonal) rather than progressing.

One example of trying to use an opportunity framing is in thinking
about high school dropouts and education. But why do we focus on high
school graduation as the end-goal? That’s a merely a means. We do pile
a lot of policies on top of the (moral?) idea that we owe students an
education until they are 16 and then they are thrown into the sea to
sink or swim. We then lock this in more deeply with testing against
arbitrary metrics and then patch it with GEDs. Shouldn’t we be
thinking about life-skills and view people as part of a larger
economic or social systems and that this is not about giving people a
free ride? It’s about enlightened self-interest and having a literate
educated population rather than one that’s trained for a future that
was supposed to have happened.

Stepping back thinking about today’s financial debacle … well, that’s
another topic …


From: Dennis Allison [mailto:drallison () gmail com]
Sent: Saturday, April 25, 2009 23:12
To: Bob Frankston; Eric Weinstein
Cc: dave () farber net; ip
Subject: Re: [IP] The Economic Crisis and its Implications for The
Science of Economics

I  listed some pointers to reference materials in the abstract for
Eric Weinstein's EE380 presentation.  You can find the abstract
at http://ee380.stanford.edu and the references page
at http://ee380.stanford.edu/Abstracts/090429-references.html.  It is
not intended to be comprehensive and can certainly be improved.  I
think I would start with the EDGE 269 article and comments:  CAN
SCIENCE HELP SOLVE THE ECONOMIC CRISIS?   I think you'd also find
Stuart Kauffman's Reinventing the Sacred to be interesting reading,
particularly Chapter 11 which deals with economics.

You are not alone in seeing the currently naive trust in a market
economy as faith-based rather than rational.  Neo-classical economics
seems to be an elegant solution to a problem quite removed from the
real world.  Much of the work of the Wall Street "quants" has been
directed at extracting money from the existing system and not at
understanding how the financial system works and considering the long
term impacts of particular strategies.  And, it is the latter which is
badly needed.

I think there are different levels of modeling which make sense.  I
don't think we have a descriptive model of the current financial
system and certainly do not have one which is detailed and complex
enough to deal with resource issues, global warming, and the like.  I
think the first step in understanding the financial crisis would be to
build a model of what's out there and what the dependencies are.

The complex systems folks argue that a bottom-up model using very
low-level agent-based elements is likely to be a better approach than
a top-down one.  This sort of system tends to evolve towards something
similar to what is observed in the wild.  Beinhocker'sThe Origin of
Wealth describes some experiments in this area, albeit with highly
simplified systems and limited behavioral rule sets.

I think you dismiss the 70's classic Limits to Growth study a bit too
quickly.  You might want to take a look at Graham Turner's 2008 paper.
"A Comparison of 'Limits to Growth' with 30 Years of
Reality" ( http://www.csiro.au/files/plje.pdf )  On the one hand,
theLimits to Growth results are based on a highly simplified dynamic
model of global resources and the global economy and did not take into
account (nor could it take into account) such factors as innovation.
On the other hand, Limits to Growth did attempt to look realistically
at the problems of limited resources and their impact on the society.
The detractors resorted to comments like "a piece of irresponsible
nonsense" (Henry Wallach, Newsweek) and hang their hats on arguments
like "human beings have always found a way around problems and we will
continue to do it" but did not present many credible arguments to
refute the basic qualitative result.  I think we, as a society,
learned a lot from Limits to Growth even though the model was only
qualitative and not predictive.

Just how to build a reasonable model of the economy is an open
question.  Given the emergent, self-organizing, and evolutionary
nature of the system(s) in question, it is not even clear that a
deterministic model can be built.  Many people think that we should
look more to biology than to physics for inspiration in learning how
to model complex systems.



On Sat, Apr 25, 2009 at 6:43 PM, Bob Frankston
<Bob19-0501 () bobf frankston com> wrote:
Rather than studying ethical behavior in a moral contexts we should be
less naïve about the role of people in these systems. Perhaps
behavioral economics is more on-topic in that sense. I see native
trust in markets as a form of intelligent design. I see prediction (as
in portfolio evaluation) as more art than science – it’s like playing
poker to see what price you can get.



I would appreciate any pointers to background material for these
presentations. I’m particularly interested in any attempt to shift
from predictive models in a Malthusian mold (claiming to know the
future by projecting from the present) to a framework building on what
I consider the essence of evolutionary processes – the ability to
capture success and survive failure without having to be predictive
beyond very modest contexts. (Too bad we use phrases like “survival of
the fittest” which confuses a mechanism with the dynamic that drives
evolution).



The reason why I put so much emphasis on evolution is that we have a
system that survives (and thrives) despite the essential unknowabilty
of future interactions and the absence of a single goal or optima.
Thus I consider efforts to develop sophisticated quantitative models
and tools to be the wrong framing. The primary “given” should be that
the future is unknowable – if we try too hard to predict the future
we’ll find ourselves in the Malthusian trap that Forester’s System
Dynamics did with the Club of Rome reports in the 70’s.



I bought a book, The Limits of Market Organization, hoping that it
would move beyond uniformism and look at the characteristics of
particular markets but instead I found the chapters on The Internet
and Telecom hopeless mired in false premises in treating telecom as a
railroad and the Internet as just another telecom service. It’s a
classic example of trying too hard to predict the future while naively
misinterpreting the present – a present that is inherently ambiguous.



It may be unfair to characterize economists as being too willing to
accept premises so they could then write papers. This is why I’m
interested in counter-examples.



I’ve been arguing that we need an economic/financial system that
creates opportunities for many small efforts rather than large
all-or-nothing projects. Or, to use evolutionary framing – we need
many ecological niches as units of failure. A corporation can be such
a unit if but we try too darn hard to keep them alive instead of
treating failure as a learning experience.



The question is whether the concept of “economics” is the right
framing and whether we should leave evolution to the biologists. We do
have “disciplines” (an ironic term in this case) such financial
management which does calculations within a framework and should treat
valuations as tentative and context-dependent. Unfortunately the naïve
(as well as willful) hubris leads the practitioners to act with
unjustified certainty. (Robert Laughlin’s A Different Universe tells
about the limits of piling models upon models).





From: David Farber [mailto:dave () farber net]
Sent: Saturday, April 25, 2009 18:59
To: ip
Subject: [IP] The Economic Crisis and its Implications for The Science
of Economics




Prior to any Economic Manhattan project maybe a study of Ethical
behavior in the financial area would be more profitable djf





Begin forwarded message:




From: Dennis Allison <drallison () gmail com>


Date: April 25, 2009 3:22:35 PM EDT


To: Dave Farber <dave () farber net>


Cc: allison () stanford edu


Subject: The Economic Crisis and its Implications for The Science of
Economics





For IP if you think it worthy:

Dave,

I'd like to call the attention of IP readers to an interdisciplinary
conference on The Economic Crisis and its Implications for The Science
of Economics being held in Toronto May 1-4 at the Perimeter Institute
and to a talk being given in the EE Computer Systems Colloquium at
Stanford this Wednesday, April 29th.  Both are related to the current
financial crisis and what we might do, as scientists and engineers, to
help resolve the situation.

CONFERENCE:

For more information on the conference.

http://www.perimeterinstitute.ca/en/Events/The_Economic_Crisis_and_Implications_for_Science/The_Economic_Crisis_and_its_Implications_for_The_Science_of_Economics/

Registration is required (seating is limited) but there is no charge
for the event.

From the Conference Abstract:

Concerns over the current financial situation are giving rise to a
need to evaluate the very mathematics that underpins economics as a
predictive and descriptive science. A growing desire to examine
economics through the lens of diverse scientific methodologies -
including physics and complex systems - is making way to a meeting of
leading economists and theorists of finance together with physicists,
mathematicians, biologists and computer scientists in an effort to
evaluate current theories of markets and identify key issues that can
motivate new directions for research. Perimeter Institute was
suggested to be the gathering point and conference organizers plan to
foster a very careful, dispassionate discussion, in an atmosphere
governed by the modesty and open mindedness that characterizes the
scientific community.


COLLOQUIUM:

One of the conference organizers, Eric Weinstein, will be speaking in
the Stanford EE Computer Systems Colloquium
(http://ee380.stanford.edu) Wednesday 4:15-5:30PM Pacific.  He will
argue for the need to have a broadly based, interdisciplinary
"Economic Manhattan Project", a global effort to develop sophisticated
quantitative models and tools that could be used to guide us out of
the financial/economic current crisis.

Eric's lecture will be given before a live audience in the Gates
Computer Science Building (room B1) on the Stanford Campus; this is a
public lecture and anyone is welcome to attend.  The lecture will be
available over the Internet by webcast in real time (questions via
Twitter) and will be available for on-demand viewing from the
Colloquium website about an hour or two following completion of the
talk.  In time, Colloquium lectures are available on YouTube and
iTunes and other distribution channels.  See the website for details.

Dennis Allison
Computer Systems Laboratory
Stanford University






______________________________________________________________________
Archives https://www.listbox.com/images/feed-icon-10x10.jpg


https://www.listbox.com/images/listbox-logo-small.jpg









______________________________________________________________________
Archives







-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com
Previous By Date Next
Previous By Thread Next

Current thread: