9 March 2000

Dear SimSoc Participants:

>What Chris is saying here is that economics students are convinced to
>accept the assumption that agents are immortal by first getting them to
>accept that agents are omniscient.  This doesn't strike me as an
>improvement.

This sweeping characterization of current economics is at least 15
years out of date.  This type of modelling did reign supreme in
economics during the hey-day of the rational expectations revolution
1972-1986 but not now.  Indeed, much of Scott Moss's discussion
strikes me as backward looking carping about the past and I cannot
understand why he persists in it.  It seems to me to be a waste of
time not only because it is misleading (as a characterization of
current economics) but because it closes off communication lines
rather than opening them up.

There is a lot of exciting stuff going in economics these days.  I've
been a strong critic of conventional economic modelling (omniscience
assumptions, infinitely lived representative agents, perfect
rationality...) for twenty five years and I can attest to you all that
things really are changing in economics if one is open-minded enough
to see it.

I hope that the participants on this list are independent minded
enough to check out for themselves what is going in current economic
research.  The best place to see this for yourselves is in the
_Journal of Economic Perpectives_, which includes clearly written
nontechnical survey articles of all kinds of new research directions.
Some of these directions you will think are interesting and some you
won't -- is this any different from how things are in any other social
science?

I will also say again that Scott Moss's attempt to lump in agent-based
computational economics with his characterization of conventional
economics as making assumptions of omniscience and immortal agents is
so strange I have to wonder what is really going on.  Object-oriented
programming (as well as other recent developments in computer science)
permits us to get away from this kind of modelling, and people in ACE
are increasingly exploiting this powerful tool to do just that.

I again urge people to take a look for themselves at, for example, the
introductions to the special issues on agent-based computational
economics (ACE) forthcoming in _Computational Economics_ and _Journal
of Economic Dynamics and Control_ -- hardly fringe economic journals.
These introductions can be found linked to the ACE web site at

    http://www.econ.iastate.edu/tesfatsi/ace.htm

and are also linked at the ACE surveys site at

   http://www.econ.iastate.edu/tesfatsi/surveys.htm

Moss himself "approved" of four of the 18 articles in these special
issues, even though in his last email he seems again to dismiss ACE
researchers out of hand as doing nothing of practical interest.  But
despite Scott's apparent belief in the empirical irrelevance of the
other 14 articles, these articles include studies of the England and
Wales electricity market using actual market protocols proposed under
restructuring, a study of the dynamics of movie release box office
revenues using ideas about information cascades, a study of an
automated market negotiation protocol currently being considered for
actual e-commerce, and a study of a restart rule for Internet access
tested on the Internet itself.  Moss approves of the study of Edmonds
testing learning behavior of artificial against real agents in the
stylized context of a simple one-sided bidding process but apparently
disapproves of the studies by Duffy, Chen/Yeh, Linn, and others who
test learning rules in simplified market contexts and who also use
learning rules derived from observation of real agents in experimental
or real-world market settings.  Could it be that the latter work is
dismissed simply because it is done by economists?

Scott Moss will reply I am sure with withering derogatory statements
about all the papers in the special ACE issues that were not conducted
by the four researchers he deigned to favor with his approval.  I will
not respond.  

Instead, I express again the hope that participants in the SimSoc
list, if you have any interest in establishing good communication
lines with economists striving like you to understand complicated
socioeconomic processes, will look beyond self-appointed spokespersons
and communicate directly with those economists on the frontlines who
are attempting to cross conventional disciplinary lines and model
market behavior from a more comprehensive social viewpoint.  I am sure
they would welcome constructive comments and ideas from you regarding
their work -- I know I would!

Best wishes to you all,

Leigh

   Leigh Tesfatsion                Department of Economics
   Tel: (515) 294-0138             Iowa State University
   FAX: (515) 294-0221             Ames, IA 50011-1070 
   [log in to unmask]            http://www.econ.iastate.edu/tesfatsi/





%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%