About practical use of Model Analysis

[Jean-Jacques Lauble]

Hi

I am interested by practical model analysis. Most methods are loops centered, some are parameter centred and one is client oriented (reality check).
Does anyone use practically loops oriented methods? It seems that past a certain number of loops, all of them are totally impossible to apply. These methods look too terribly academic.
I personnaly use parameter sensibility analysis which relies on my knowledge of the model and reality and reality checks which seem to be extremely rarely used. I never saw a model with reality checks included publicly available.
Any suggestion?
Regards.
Jean-Jacques Laublé Strasbourg France

[Jim Duggan]

Hi Jean Jacques,

I think many people are also interested in leveraging the benefits from Formal Model analysis.

This is a topic that goes to the core of System Dynamics, as it focuses on ways to formally demonstarte how feedback structure influences system behaviour.

It also is a challenging topic in many ways, as the mathematics of feedback analysis can be difficult to conquer (i.e. Eigenvalue/Eigenvector approaches), as they are not as intuitive as "bathtub dynamics" and graphical integration.

I wonder what are the "barriers to use" of these approaches?

If a tool was available that was user-friendly and intuitive, might that start a reinforcing "market growth" process for the tools of Formal Analysis?

regards,
Jim.

[Jean-Jacques Lauble]

Hi Jim
To my opinion, the first problem with model analysis whether formal or not (what does really formal means ?) is: is there a market for analysis alone? It looks to me that people spend so much time building and verifying models that they have no more time to analyze them.
The fact that analyzing a model takes much more time than building it, reinforces that situation.
The same situation occurs with business people meeting. A report of the meeting is rarely produced and the next meeting, nobody has read the report.
One of the solutions is to build models that can be easily analyzed, or even that are analyzed automatically. This is why I like reality checks that are built in the model and verify the coherence of its results with some aspects of reality. It alleviates the burden of model analysis.
The problem with reality checks is that it needs practice and doubles the time to build a model if done seriously. It too necessitates a strict top down model building, starting from the simple (no details) to the more elaborate, reality checks being added at each step as new material is added to the model. But there is nothing new to that. More quality needs more work.
I share with you the belief that one of the prime SD’s problems is building models easy to use for the client.
Best regards.
Jean-Jacques Laublé