REPLY Adoption Dynamics (SD6155)

[System Dynamics Mailing List]

Posted by  Jean-Jacques Laublé <jean-jacques.lauble at wanadoo.fr>

Hi Colin

Building a model without any data is ate the same time easier technically
(no cumbersome calibration) and much more difficult conceptually without the
guidance of experiences.

To my opinion there are fundamentally two kinds of simulation: the ones with
experiences of the past with data and the other without data where you are
left with your imagination and your dreams too. These two kinds of simulation 
must probably be conducted in a very different way.

I have looked in the SD literature if somebody has tried to study the
subject but did not find any reference. I would be glad to have one if 
somebody knows something about the subject.

Aldo Zagonel once separated SD simulations into 5 kinds, the third one being
the classical SD with past data. He must have probably studied the question.
Kim Warren in his book is more oriented with simulation of the second
kind, without data that is to my opinion more adapted to the business
environment where it is more difficult to use past data due to the more 
changing environment.

The second kind of simulation without past data is difficult because you can
build as many very different sort of models that are not better than the others. It is
necessary to consider a lot of different scenarios. But it is too somewhat more interesting because
you are more free to invent new solutions that can too be better applied having not
too deal with the weight of the past.

As to the objective of studying the adoption dynamic it is not an easy one
that has many parameters to be considered. The objective to my opinion
should not to give any solution that could be pretentious but to help thinking about the problem.

After the discussion about the adoption of SD by American authorities which
is a process of adoption too, I have tried to make a model of the growth of SD a bit like
the one you can find in the Vensim examples but I stopped because it was too
difficult.

I tried with something like the Bass diffusion model.
It is very simple but unfortunately relying on parameters very difficult to
evaluate and at the end you finish with something not credible.
I have changed my objective and started a model that tries to understand the
policy of development of a new SD consultant.
Policy meaning for example what problems to accept.
Faced with a client's problem, the question is:
Will I accept the problem knowing that I know that the added value of my
work may be weak?
Will I spent time looking for customers with more added value?
What are the consequences of this choice on the dynamic of my development?
This kind of model is much more easy than a general model of the growth of
the field although less complete. But it can help understand some of the difficulties
of the field.
I think that one of the difficulty of the field is the lack of consultants which is mainly due to the
difficulty of the field.

It is the case in France where you do not only find any consultants but
nobody who ever heard about SD.

I will be glad to help you in your endeavour if you are interested.
Regards.

Jean-Jacques Laublé
Strasbourg France
Posted by  Jean-Jacques Laublé <jean-jacques.lauble at wanadoo.fr>
posting date  Sat, 6 Jan 2007 19:21:16 +0100