Epistemologic question

["Antoni Oliva"]

I have been working with SD models for the last three years, including the
introductory course at MIT. So one or the basic things I am supposed to know
is that you need a concrete question or problem to address, formulated in a
very concise way, to start building the model. Nevertheless, and maybe
because of the kind of systems I am involved in, I always try to tackle the
system I am dealing with as a whole, and I always fail; then I focuse on
formulating the right question in the right way and start all over again,
then it normally works.
My interpretation is that the problem acts as an "attractor" (excuse me for
this terminology, I do not mean to use it in the chaos theory sense),
ordering all the relevant variables and relations, and rejecting those
irelevant. In some way the model is created by the problem.
Is that right?
What are the epistemologic consequences of this way of doing?
Best,

Antoni Oliva
From: "Antoni Oliva" <aoliva@22sistema.com>
http://www.22sistema.com
Barcelona

["David Castro"]

Antoni Oliva wrote:

"In some way the model is created by the problem.
Is that right?
What are the epistemologic consequences of this way of doing?"

Very interesting question to me. I help community leaders learn systems
dynamics in Pennsylvania, USA. The way I have understood this advice about
modeling (start with the problem) is that it helps you focus on what you are
trying to create. These human needs/wants/visions help you focus relevant
systems dynamics. It recognizes that systems dynamics is a pragmatic tool
for making small, contextual changes in the world. It is not a theory of
everything. Trying to create overarching models that you can then use to
find your particular issue (like picking up the Atlas of the world to find
an address in a particular city) doesnt work well because 1) the maps
change depending on what you care about 2) the world changes at a rate that
is faster than our modeling capabilities. (So by the time you locate your
address in the Atlas of the world, the street doesnt exist anymore). Thus
systems analysis is always local, always contextual, and always dependent
upon and framed by the goals and intentions of those doing the analysis.
Saying that the problem creates the model is similar to saying that the
modeler creates the model -- the model cannot be divorced from the specific
context/situation/desires of the modeler. To me the epistemological
consequence is that models cannot be really be right or wrong, they can only
be more or less helpful to the modeler in working to achieve the modelers
objectives. This can be liberating because you stop worrying about building
the perfect model and rather ask the question whether you are gaining
insights through the modeling process that help you move toward your goals.

David Castro,
Institute for Leadership Education
castro@i-lead.org
www.i-lead.org

[=?iso-8859-1?Q?Andr=E9_Reichel?=]

Antoni wrote:

In some way the model is created by the problem. Is that right? What are
the epistemologic consequences of this way of doing?

Knowledge is always, an maybe first of all, problem solving knowledge. The
process of gaining knowledge (learning) is an explanatory process, aimed at
solving a problem. Therefore, the model is an answer generator for a
predefined question aka the problem.
>From a constructivist point of view this means further problems. The
addressed problem is created in our mind, reflecting on our mental models
and behaviour patterns. In other words: in front of the historical
knowledge that has been created in the past, arised from past problem
solving patterns.
Now, the new problem will be defined in a way that fits this historical
knowledge and solved with your usual patterns (localised learning).
Unless the problem is of a somewhat awkward kind, you will go along with
it, without learning a great deal but further strengthening your perceived
knowledge base and world views.
This then would mean, that real learning (in a double-loop manner), only
occurs when we face an awkward problem and are forced to reshape our
problem solving patterns or even the mental canvas of our understanding of
reality.
To conclude: science only progresses when we fail to solve problems the
usual way. The problem itself makes us think, if we ask the right questions.
Without a challenging problem we cannot change our knowledge base. Human
beings are much too lazy animals, after all.

Viele Grüße / many regards
André Reichel
< reichel@sofo.uni-stuttgart.de >

["Michael E. Fletcher"]

This is probably more a question of particular problem solving style
than a question of epistemology. My guess would be that you use the
preliminary effort as a method to help you correctly boundary the
question (and thus define it more clearly). Using iteration to improve
the question portion of the process certainly fits with the normal
process of iteration used in the overall modeling process. Many failures
in problem solving are actually failures in problem statement; that is
people rush off to solve before they actually have defined the problem.
They they wonder why their "solutions" never solve the problem. The
answer is simple - they didnt really know what the problem was.

It seems that the iterative process you are using, perhaps
unconsciously, gets you from A to B, and to a workable model, even if
the road is a bit longer than you might like. Looking at the "bigger
system" first and then narrowing the scope actually reminds me of what
Russell Ackoff once recommended as a synthetic (vice analytic approach)
to problem solving; that is: understand how the "larger system" works
first, and then confirming how the system under study functions within
that larger system. (with apologies to Dr. Ackoff for a poor paraphrasing.)

As far as the epistemology of the whole thing, well my (minor) in
philosophy was 20 years ago.... but perhaps there is a little Karl
Popper is at work here. You falsified your first hypothesis and had to
develop a subsequent working hypothesis. Everything works out in the
end, and the "how" is not entirely a mystery.

Mike Fletcher
mefletcher@speakeasy.net

["Dr. Vassilios Karavezyris"]

Hallo Antoni,
this is a very interesting topic.

What are the systems you are involved in?

I think that you (or anyone else) can not really "deal with the system as
a whole". At least not by using or relying on thinking alone.
If we deal with -i.e. understand (or explain)- the system as a whole we
possibly need not to model it. (Unless we believe this is a better way to
communicate knowledge.)

Now, talking about problems is somehow different than talking about
systems. (I am not sure what you mean exactly when you refer to problems.)
At any rate, we may approach to this by examinig problem spaces (initial
states, goal states, barriers.) I feel that when the initial and the
desired states are clearly formulated then there is at least one solution
which you may find (out). In this sense, making the right question means
that you know where you want to get to. Then, with more or less effort and
time you have good chances to find your answers. If you do not know what
you want to reach, or even why, (this may easily happen when you deal with
the system as a whole) then you will fail in finding any solutions.

We may agree in this: The model is created by the problem solver.

Regards
Vassilis Karavezyris
From: "Dr. Vassilios Karavezyris" <vka@mms.tu-berlin.de>

--
TU Berlin, FG MMS, Sekr J2-2
Jebensstr. 1, D-10623 Berlin
Tel: (4930) 314 79525
Fax: (4930) 314 72581
E-mail: vka@mms.tu-berlin.de

[Alexander.Zock@t-online.de (Alex]

Hi,
I do like your question, as it points to one of the key issues of system
dynamics from my point of view. This key point can be summarized as: "You
never model a system as a whole, you always model your mental representation
of a subjectively perceived problem!". To state it a bit more provocative,
there is nothing like a system as a "whole", there are only subjective
representations and models of the so called external world. This might be a
bit too psychologic, but from my point of view it touches the heart of SD.
SD in that respect is a great tool to help people to formalize a discussion
about their subjective representations of a problem. It is not a tool to
build a formal model of a system (meant in an ontologic way).

Given the above understanding of the strength of SD it is not only a good
advice to try to focus on your understanding/ a groups understanding of a
problem, but it is also a good idea to focus on the process of group problem
solution/model building. You might want to check out Jac Vennix book, Goup
model building, which is a great treatise about this view on SD.

Regards

Alexander
From: Alexander.Zock@t-online.de (Alexander Zock)