Ventana software support forum

On Wed, 4 Nov 1998, eduardo giordanelli wrote:

> My question could be stated in the following terms:
> We use System Dynamics because is an easy, but not accurate, tool to use
> ? Or it could be an accurate tool when dealing with not learning systems
> and becomes not accurate only when confronted with learning
> environments?

Saying one cant point forecast with a system dynamics model doesnt seem
to me to be the same as saying the model cant be "accurate." Our
model-based predictions are contigent forecasts and they can be quite
accurate within the contingent world encapsulated by the model.

If people learn from a model and alter the contingencies accordingly, that
means the model wont predict the future well because people altered the
future as a result of the model-based insights. But those same people
would not then turn around and accuse the model of being inaccurate. If
they had enough confidence in the model to alter policies, I would think
people would say the model looked quite or very "accurate" to them.

Maybe we need a better (more accruate?) term here.

But the basic point is true, that forecasting in human systems is
confounded by the fact that the forecast could cause humans to change and
thus make the forcast false. Such a phenomenon was labeled a "suicidal
prophecy" by none other than the father of Venn diagrams, John Venn
(1888), and was seen as the counterpart to Mertons "self-fulfilling
prophecy" (1948).

Presumably, nonliving systems, or systems that cant make sense out of the
structure, dynamics, and insights of system dynamics models, arent
subject to the suicidal prophecy phenomenon, and thus can be predicted
"accurately," up to the point allowed by the vagueries of any prediction.

Whatever we make of the "accuracy" discussion, Im pretty sure we dont
use system dynamics because its "easy." How many really insightful system
dynamics models do we know of?

...George Richardson
From: George Richardson <gr383@cnsvax.albany.edu>