--George McC 03:27, 11 January 2007 (PST)
Good stuff. I have some exposure to ABM but I am yet to use it seriously.
By its nature, ABM will generate 'surprising' results in the sense that you really have little idea what to expect when you set the model running. How do you reconcile this with the 'typical' expectation that a model will answer questions rather than possibly generating lots of new ones?
Note that I am asking this in the role of Devil's advocate - I personally don't have any problem with 'exploratory' models, but among the less informed population - including much of 'management' - the prevailing view is that you build a model to get an answer, not to experiment.