Let me expand a bit on my previous answer with some examples ...

Scott wrote:
> There seem to me to be two issues here (and certainly these are the
> issues that led me to start this thread):
> 
>   1. Can the approaches taken by the social simulation community
>      usefully inform software engineering?
>   2. Can the agent oriented, BDI based software engineering approach
>      usefully inform social simulation modelling.

Case 1.  Yes, if they look more towards the quick-and-dirty heuristics
that real social agents tend to develop to cope with the messy and
complex environments they inhabit; yes if they are looking for ideas for
ways of structuring societies in order to semi-regularise certain
aspects of behaviour ...

Case 2.  Yes, if the software approach is used to pin down those aspects
of social behaviour that is predictable so as to cleanly seperate it
from the emergent behaviour, yes if it is used merely as a software
architecture with which to program descriptively relevant stuff ...

BUT...

Case 1.  No, if they look to heavily idealised versions of human
rationality (unless they only want to work with toy problems); no if
they want algorithms that will work in all cases (i.e. with 100%
reliability) ... 

Case 2.  No, if we are looking for appropriate structures to produce
descriptively accurate models of social behaviour/cognition; no if we
are looking for agent design methodologies, ...

One interesting area where the two approaches would have an interest in
(but from opposite points of view) is to establish some of the the
conditions under which macro behaviour IS predicatable.

Regards.

--------------------------------------------------
Bruce Edmonds, 
Centre for Policy Modelling, 
Manchester Metropolitan University, Aytoun Bldg., 
Aytoun St., Manchester, M1 3GH. UK.
Tel: +44 161 247 6479  Fax: +44 161 247 6802
http://www.cpm.mmu.ac.uk/~bruce


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