The main point of our paper is not to label all agent-based models as
individualist and therefore 'wrong' -- although, I admit, we may come over
that way... the important thing is to be aware of individualism as a theory
(or set of theories, biases at least) about the social world, to which ABM
approaches are particularly prone.
There is a case to be made for models which produce social entities as
emergent from purely indiviudal interactions. These are 'experiments' in
'generative social science' as advocated by Epstein and Axtell in the
Sugarscape. But care is needed as to the status of such results. After
all just what is a model consisting of individuals only a model of? Where
is this society which is not a society, just individuals? Even coming down
from the trees (as it were) proto-humans were organised into tribal groups,
with social structures. Therefore, there may be little empirical interest
per se in a model which produces social structures from 'nothing'.
[Note on 'emergence': I'm yet to be convinced that most emergence reported
in our models isn't (i) almost entirely in the eye of the modeller, and
(ii) only a product of our limited analytical ability when dealing with
non-linear systems. Castelfranchi is persuasive on the first point when he
argues: "A notion of emergence... relative to an observer... is not
enough. We need an emerging structure playing some causal role in the
system... self-organizing emergent structures. Emergent structures should
reproduce, maintain, stabilize themselves..." (page 179, 'Modelling social
action for AI agents', Artificial Intelligence, 103, 157-182, 1998). On
the second point, see Joe Faith's paper 'Why gliders don't exist:
anti-reductionism and emergence' in ALife VI.]
In the long run, it may be more productive, in line with repeated prisoners
dilemma etc, to think about which types of social structure are
evolutionarily stable and maintained over time, and which are not. And
also about which are stable when they're the only game in town
(hunter-gatherers in pre-colonial Australia, perhaps) and which survive
when different social structures meet (Victorian capitalism in
post-colonial Australia, perhaps). Such work will require social structure
to be explicitly represented.
When it comes to building models for policy makers, corporate planning etc,
of contemporary societies, rather than abstract experiments, then it seems
impossible to avoid representing social structures explicitly -- or do we
really believe that there is no such thing as society?!
David
-----------------------------------------------------------------
DAVID O'SULLIVAN
RESEARCHER
CENTRE FOR ADVANCED SPATIAL ANALYSIS
UNIVERSITY COLLEGE LONDON, 1-19 TORRINGTON PLACE, LONDON WC1E 6BT
[p] +44(0)20 7679 1812
[f] +44(0)20 7813 2843
[e] [log in to unmask]
[w] http://www.casa.ucl.ac.uk/~david/
-----------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|