[Swarm-Modelling] Re: [Swarm-Support] Repast vs. Swarm

[Steve Railsback]

Anju Dahiya wrote:
> Quoting "Marcus G. Daniels" <address@hidden>:


> By the way I need your answer to fill the just started MAS FAQs topic: Is agent
> based modeling a dodgy scientific activity?
> http://ecoinformatics.uvm.edu/projects/ibm/mas.html

(we're getting into stuff that belongs in Swarm Modeling)

I must comment because I'm deep into producing a book on how to do 
serious theoretical science with agent-based models. I think 
"innovative" is a better word that dodgy, but I know exactly what Marcus 
means. To some people (including one reviewer of a draft of our book), 
if you are not doing differential equations you are not doing science at 
all. But all important innovations have to deal with that kind of 
conservative response. I think agent-based modeling is rapidly becoming 
mainstream in some fields, especially economics and ecology.

But we are all here because we know that classical modeling approaches 
like differential equations are themselves dodgy for many systems. The 
problem is that we're still figuring out how to do science based on 
simulation. What is "theory", how do you "prove" stuff, how do you draw 
general conclusions? The answers to these questions are easier than 
people realize, but do require thinking about systems in different ways. 
(Especially: how do you model *individuals* so that *system* dynamics of 
interest are reproduced?)

One key to making ABS more acceptable as science is developing common, 
widely used modeling concepts, terms, frameworks, and tools. To me one 
of the coolest things about Swarm is that the project started off with 
the goal of meeting these needs; and has succeeded to some extent.

Steve

--
Lang, Railsback & Associates
250 California Ave.
Arcata CA 95521
707-822-0453; fax 822-1868
address@hidden