Dear Jim,
I have been following your excellent pieces of advice to Martin, and I would
like to ask you two things:
1. Would you be interested in organizing and publishing this advice in the
brand-new modelling site "Environmental Dynamics Vitrine"
(
www.utad.pt/ed/vitrine)? They would fit perfectly under the "Theory"
section.
2. Would you be interested in joining the Editorial Board of the EDV? That
would be an honor to all participants, given your experience and "sense of
community" (i.e. sharing)!
Best regards,
Tasso
---
Anastássios Perdicoúlis
www.utad.pt/~tasso
Jim Hines wrote:
> Martin,
>
> Its good that you are using the "standard method" for each of your
> sectors. But, Im concerned for you about developing the model by
> sectors.
>
> Dividing a model into sectors is a common strategy in SD, but not one
> that always (usually?) works. The problem is that when the modeler
> hooks the sectors together, he will be completing tens, hundreds,
> thousands, even tens of thousands of loops (depending on the size and
> "density" of the model). This means that the modeler can be confronted
> with new, puzzling patterns of behavior at precisely the time the model
> is most difficult to analyze (i.e. when its as big as its going to get).
> The danger is multiplied when different people have developed the
> different sectors (which is usually the whole idea behind dividing it
> into sectors in the first place).
>
> The MIT National model suffered from this approach, I think: A number
> of dissertations (most of which I listed in the prior email) were done
> as new sectors, developed separately from the larger model. The sectors
> themselves were fine pieces of work, but incorporating them into the
> larger model was difficult, as has been understanding the dynamics of
> the National Model, though this is not only due to the early sectoral
> approach -- the National Model really is a complicated model.
>
> People sometimes mistakenly think the sectoral approach will work
> because the approach is lauded in other fields. For example, the
> sectoral approach is used by software developers. But in such fields
> the approach works only because interactions between sectors can be
> minimized (by good developers). Unfortunately in system dynamics, we
> dont want to minimize those interactions -- the interactions are what
> were interested in (otherwise why put the sectors together at all?).
> So, what works in software development and maybe other fields wont
> necessarily work in SD model development.
>
> A better approach, I think, is to apply the "standard method" to the
> real subject of interest -- The economy -- rather than to the
> individual sectors.
>
> The 5000 dollar question remains: How do you make use of a more than one
> person in creating a model, if you DONT rely on a sectoral approach. I
> know of one SD consulting firm that (at one time) could put three people
> on a model -- but in this case only one of the people did the modeling.
> One of the other two people would take charge of the data and the other
> person would take the role of project manager (plus relationship manager
> and coach).
>
> Id be very interested in hearing on this list experiences or
> speculations on how more than one person can participate in modeling.
> (Note: Im mostly interested in approaches that dont rely on sectors,
> but defenders of sectors are welcome to get their licks in (as if I
> could stop you!)).
>
> Jim Hines
>
jhines@mit.edu