Leigh Tesfatsion wrote:
> Please recall that ACE is simply defined as follows:
>
> ACE is the computational study of economies modelled as evolving
> systems of autonomous interacting agents.
>
> How anyone can infer from this general definition that "ACE
> ....respecifies issues and environments to render them amenable to
> analysis with a narrow range of techniques" (Moss, last email) is
> totally beyond me.
The point of this discussion is not an abstract definition of computational
economics but the extent to which models in practice correspond to reality --
whether their formal properties preclude such correspondence and, if not,
whether they are or can be validated as accurate descriptions of observed
behaviour, social institutions and structures.
So what is the practice? Leigh invites us to consider her introductions to
the special issues of Computational Economics and the Journal of Economic
Dynamics and Control as an accurate description of what ACE is and what
constitutes ACE. Leigh's claim that ACE is, in effect, a broad church seems
to me to be well supported by her introduction. But from those
introductions, I was able to identify only four papers out of 18 that address
observed social or individual processes: Kirman and Vriend on the Marseilles
fish market, Rouchier et al on the social organisation of the North Cameroon,
Epstein on norm development and Edmonds on human subjects in an auction
experiment. Perhaps Leigh would tell us how many of the other papers
developed agent representations specifically to capture observed social and
cognitive structures.
If she would include the Duffy paper which "describes a set of experiments
with real and artificial agents that were both conducted within similar
versions of the Kiyotaki-Wright model," perhaps she would also tell us how
these results were or would be validated. I ask because the Kiyotaki-Wright
paper (Journal of Political Economy, 1989) represents a world where
there is a continuum of infinitely lived agents.... (p. 930) Trade
always entails a one-for-one swap of inventories.... Each
individual chooses a trading strategy to maximize his expected
discounted utility....(p. 931)
How would this (or the simplified Duffy version of it) relate to any actual
situation that we could observe? The issue here, of course, is whether it is
acceptable in the ACE community to "(re)specify issues and environments to
render them amenable to analysis with a narrow range of techniques." (I did
not intend to suggest that this quote defines ACE -- certainly not as
practiced by Leigh Tesfatsion. The question remains as to whether it is
acceptable in ACE to specify environments that are amenable to pre-selected
analytical/modelling techniques without regard for their descriptive
accuracy.)
> In my mind, true scholarship is looking for the wheat in other
> people's work and building on it, not separating out the chaff and
> making grand pronouncements that the chaff represents fundamental
> error and failure of vision.
On this basis, astronomers would be "looking for the wheat" in astrology,
chemists in alchemy, physicists in the phlogiston theory of combustion,
biologists in the humunculus theory of reproduction.
> I hope in the computational social
> science community we can keep our eye on the prize, the exploration of
> the potential usefulness of computational frameworks for the study of
> important social science issues
I agree completely: start with the issues and then reason about the value of
a computational or other approach and, if a computational approach is
indicated, reason about an appropriate modelling technique and appropriate
representations of the actors, etc.
Would Leigh also agree that we should validate our choices of what is
appropriate by observing the target system -- both in relation to available
statistical data and qualitatively described observation by domain experts
including stakeholders?
> from a comprehensive inclusive point
> of view.
Would such a point of view involve the acceptance of mumpsimus?
regards,
Scott
--
Professor Scott Moss
Director
Centre for Policy Modelling
Manchester Metropolitan University
Aytoun Building
Manchester M1 3GH
UNITED KINGDOM
telephone: +44 (0)161 247 3886
fax: +44 (0)161 247 6802
http://www.cpm.mmu.ac.uk/~scott
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|