Ventana software support forum

It was argued that SD is a specialised tool, like a hammer, and that one
would not gain by combining it with another specialised tool, say, a
screwdriver. Through my own professional experience I got a different view.
Usually there is a customer, such as a company or a government organisation,
who wants a solution and does not care much about the tools used to produce
it. In building terms they are not looking for a hammer, nor for a
specialist in using hammers. They are looking for the equivalent of a
handyman to make alterations to the THIS IS SPAM.

A real handyman has his trusted toolbox with hammer, screwdriver, power
drill, jigsaw, etc. He uses them all, as required, to do the job. In
modelling this is presently not the case. There are the equivalents of
hammer experts (SD), screwdriver experts (discrete simulation), etc. To
alter a THIS IS SPAM one would have to employ a team of such experts plus the
managers required to coordinate them. This is prohibitively expensive. If
the job is of a kind that a hammer or a screwdriver expert can handle it
alone, a contract may be given. If this is not the case, the customer may
solve his problems in another way. I assume for this reason most potential
modelling contracts are never awarded.

Often there are no suitable packages available. In addition there is also
the problem of cost. Many packets that might be useful for a certain
application are prohibitively expensive, especially if several of them would
have to be used in one project. It is then often cheaper to build a package
tailored to the job. I am quite convinced that my professional experience is
not unique. Here are a few examples.

* For my masters thesis I needed an SD package. Then (1977) we had only
DYNAMO on the university mainframe. The integration (linear) did not do the
job, so I developed my own package, using the Runge-Kutta method. I needed a
non-linear solver as well, so I developed one.
* When I simulated gas transport in divers during decompression, I used
another self-made simulation environment
* To model financial and environmental aspects of energy supply systems I
wrote yet another modelling environment. It focussed on process chains and
ignored feedback structures, but would they have been available with little
effort, I would have used them.

Imagine similar efforts of numerous other modellers and developers would not
be spent in re-building the same basic tools with slight modifications.
Imagine these efforts would be spent contributing to an Open Modelling
System (OMS). Once developed, components could be re-used with little
additional effort, and the development time saved could be spent on
understanding the system better. Alternatively, inventive minds could spend
that time building new types of components. It seems obvious that such a
system would soon exceed the capabilities of any commercially available
system today.

In order to make use of an existing structure, modellers would have to
strive to make their special requirements fit into the existing structure
and to minimise the need for additional components (which they have to
produce themselves). This would encourage the development of an integrated
modelling concept in an evolutionary manner.

To make use of the considerable intellectual capacity at educational
institutions, OMS would have to be free. Just like LINUX. Somebody would
have to start it off by defining the framework and the standards, and by
creating a useful kernel that makes it worthwhile for others to start
considering OMS for real work. As for the administration of such an effort,
much could be learned from the LINUX movement.

Some questions in this context are: Is the OMS suggestion sensible at all?
Would you use OMS? Can you see yourself contributing functional modules to
OMS? Can you imagine an agreement on the required standards? Who should
produce the kernel? Who might sponsor such an effort?

Roland Schiefer
Technical Consultant
schiefer@iafrica.com