World model - Chaotic?

[forda@mail.wsu.edu (Andy Ford)]

>Hello, everyone.
>
> This is Andy Ford at Washington State University. Im replying to
>one of the questions about the world modeling efforts. The question was
>whether the world model was ever run under chaotic conditions, and whether
>we learned anything useful from the analysis.
I am not aware of such an analysis, but I did read about something
similar. I read that someone decided to test the world model by
initializing the model at the END of the planning horizon. They then
changed the sign of time and allowed the model to run backwards. (I dont
belive they were using DYNAMO, the original language of the 1970s world
models; I think they had translated the model into FORTRAN or some other
language that would permit them to make time run backwards).
You are probably wondering why in the world anyone would want to
run the model backward? Their goal was to learn if the model would
correctly retrace its steps backward through time. If the model did not
arrive at the 1900 starting conditions, the investigators were ready to
conclude that the model was "unreliable."
As I remember the story, they conducted this test and found that their
model did not correctly retrace the path backward to the year 1990. They
concluded that the model did not "pass this test" and published the
ressults.
I strikes me that changing the sign of time is like changing all the
negative controlling feedback loops in the model into positive feedback
loops. They would then act in an unstable manner. The effect is to create
a highly unstable system that will have trouble retracing its backward path
in the face of small numerical errors in the simulation. The investigators
had created a system that was highly sensitive to initial conditions.
The backward test exercise is similar to a bicycle rider wishing to
test a bicycle by riding it backward. If you were to sit on a well
designed bicycle and try to ride it backward, it would not provide you with
the control needed to maintain a stable backward path in the face of small
disturbances from pebbles on the road. But does that mean that the bicycle
is "unreliable" for its intended purpose (presumably to ride forward)?
The "backward test" teaches us NOTHING about the usefulness of the
world modeling effort. But it does teach us that people from different
backgrounds can approach a topic from a point of view that is quite foreign
and mystifying to those of us who have studied and adopted a System
Dynamics point of view.

The publication of LIMITS TO GROWTH provoked an interesting and
heated debate during the 1970s. It seems that everyone had a comment:

Anthony Lewis: "The most fascinating and the most disturbing book I have
read in a
very long time"
Robert Townsend: "if this doesnt blow everybodys mind who can read without
moving his lips, then the earth is kaput"
Henry Wallich: "a piece of irresponsible nonsens"
Rasell, Roberts and Ross: "an empty and misleading work"

and to sum it all up--

Robert Socolow: "the amount of commentary -- pro, con and orthogonal--has
exceeded the original text by a factor of several hundred.
Like LOVE STORY and GREENING OF AMERICA, it is simply not
permissible to have nothing to stay about THE LIMITS TO
GROWTH.

These and other views of the LIMITS debate are summarized in an interesting
and fair book by Martin Greenberger entitled MODELS IN THE POLICY PROCESS
published by the Russel Sage Foundation in 1976.






----------------------------------------------------------
Andy Ford
Program in Environmental Science and Regional Planning
Washington State University
Pullman, WA 99164-4430
(509) 335-7846
FordA@mail.wsu.edu
----------------------------------------------------------