Generic Structure

[CrbnBlu@aol.com]

After reading the generic structure article by Lane and Smart in the latest
issue of System Dynamics Review I went back and tinkered with some of the
archetypes I initially created in ithink.

I have found that the connection between an Archetypal Structure and its
implementation can be so loose as to be almost nonexistent and relatively
worthless.

I managed to take the "Limits to Success" Archetype and implement it with a
Stock & Flow structure which looked pretty much like the "Limits to Success"
CLD, and make it act like a "Fixes that Fail," an "Escalation," a "Success
to the Successful," a "Balancing Loop," a "Reinforcing Loop," or a "Shifting
the Burden." Each of these reference actions were produced with no structure
changes, just the equations in the diagram were changed.

As such, a picture may be worth a thousand words, and all the words may be
inappropriate. I guess the lesson I get from this is that Ill just be much
more cautious as to what I think I conclude from the visual structures I
perceive!

be well,
Gene Bellinger
CrbnBlu@aol.com

[Stephen Wehrenberg]

Gene,

At the risk of Bob wondering why we are having a discussion like this on
the list (a reasonable question) ...

I read the below, and your paper on translating CLDs to SF diagrams
...as usual, I wont try to answer your questions but will try instead
to add to the confusion amd your frustration.

> Each of these reference actions were produced with no structure
> changes, just the equations in the diagram were changed.

Well, of course ... the equations are the only things that REALLY
matter, yes? The very first time I ran the Peoples Express simulation
(and IThinks new RGP Financial Services simulation) I was frustrated
beyond belief ... in fact, I stopped participating on the PE simulation
because everyone involved was treating it simply as a game to beat ...
with no notion of the underlying structure of the model that was the
simulation. Similarly, I tried the SIM-WHATEVER stuff exactly once ...
since I couldnt get under the hood, it seemed like GUESSING to me
(unless one wanted to do an awful lot of controlled experiments with
individual variables). At least the RGP Financial simulation PRESENTS
the underlying structure, but (to my frustration) not when I wanted to
be able to see it.

So whats my point? First, a model, no matter how carefully crafted and
artfully presented, is really only UNDERSTOOD by the folks who developed
it. And maybe not them, either! Its only with a thorough
understanding of the underlying structure, REPRESENTED by the graphical
model (whether CLD or SF), that the model makes sense. Its the
"adjusted mental maps" that represent understanding (and learning), not
the physical representation of the modeling package ... thats just a
summary.

Second, I am beginning to think that modeling for any reason other than
solving a problem is a waste of time ... but I stand ready to be
corrected on that generalization!

Yours in co-frustration,

Steve
--
Stephen B. Wehrenberg, Ph.D.
Chief, Forecasts and Systems, US Coast Guard;
Administrative Sciences Program, The George Washington University;
wstephen@erols.com

[bbens@MIT.EDU (B. Budiman)]

A

IMHO generic structure is no more than the manifestation of conservation
principles: conservation of matter and resources, and the bookeeping of
information flow. The structure becomes generic as the flows of matter,
energy and information follow certain rules, e.g., paths of least
resistance.

Many members of this list are familiar with the notion of SD modeling
with "molecules." Jim Hines and Bob Eberlein are two of the leaders in
the conception of this modeling tool. Molecules, if I recall correctly,
are a collection of SD stock-and-flow structures that appear numerous
times in almost any modeling effort. So, one doesnt need to "guess" the
structure of, say, a project management model. Instead, one just looks
it up in a collection of stock-and-flow structures called molecules. It
is very useful to get an approximate structure and its behavior.

However, people say, "Its difficult to please everyone all the time."
This means that the molecules, while sufficient to model a behavior
quickly, may not encapsulate all possible structures. So, it works better,
for me anyways, to keep track of the flows of matter, resources, and
information in the system than it is to use the molecules. This worldview,
if you will, shows a strong connection between the causal-loop and
stock-and-flow diagrams. BTW, doesnt a stock-and-flow diagram represent
an accounting system of such flows?

Yet, another universality of system dynamics (science & engineering,
socioeconomic, you-name-it systems)!

-B-
bbens@mit.edu