Meta Dynamics?

This forum contains all archives from the SD Mailing list (go to http://www.systemdynamics.org/forum/ for more information). This is here as a read-only resource, please post any SD related questions to the SD Discussion forum.
Locked
"John Gunkler"
Member
Posts: 31
Joined: Fri Mar 29, 2002 3:39 am

Meta Dynamics?

Post by "John Gunkler" »

Ill venture a partial reply. (Im not prepared to respond to all of the
"windup" to the "pitch," but Ill respond to the "pitch" which I take to be
your question about whether System Dynamics has "anything fundamental or
unique OF ITS OWN to say about the way the universe AS A WHOLE works....")

Yes, I think System Dynamics potentially does have something to say about
how the universe works, but our perspective is far from unique. A major
portion of what System Dynamics has to say comes from our feedback
perspective, and were not alone in that. George Richardsons brilliant
book, "Feedback Thought in Social Science and Systems Theory," does a
terrific job of explicating this.

Also, Im not sure how prepared we are to actually offer as theory the
notion that feedback rules the universe. We certainly have found many, many
examples where this seems to be true, but I know of little foundational work
that would turn our myriad observations (which may be colored by our
tendency to see feedback everywhere -- i.e., by our own biases and cognitive
limitations and errors) into what can be truly called a theory.

It may be, and I tend to think this is the case, that feedback loops are
truly a fundamental characteristic of dynamic behavior at any level of
complexity beyond the extremely simple -- which is to say, of any system
that is interesting enough to warrant study or to be of practical value.

Ill offer another candidate for taking part in the System Dynamics theory
of how the universe works: the extremely powerful claim that one needs only
two kinds of "things" in ones model of a dynamic system (i.e., stocks [or
accumulations] and flows [or rates.]) This is one of those nice, because
unexpected and simplifying, assumptions that good theories seem to consist
of.

Having said this much, I must also say that any of these speculative
candidates for theory must always be put in the context of what chunk of
"the universe" we claim to be able to explain and predict. Just as Newtons
theory of "force" does not intend to explain "the force of your argument" or
how one person can "force" another to do something, so our "theory" has its
limits as well. There are levels of prediction of the behavior of dynamic
systems for which System Dynamics is not well suited, and Im personally
struggling with how we build self-adaptation into our models -- i.e., how we
model what Jay Forrester calls "policy changes" without us (the modeler
builders) stepping in, like a deus ex machina, and changing a rate equation.
And if we dont have a nice way to model self-adapting systems, we wont
have much claim on a theory of how the universe works, in my opinion.
Others have their favorite shortcomings as well.

All of which may mean that what we have is a very useful tool for
understanding (quite a bit but by no means all about) dynamic systems, which
is based on some theories about how the universe works, but does not itself
contribute such a theory.

From: "John Gunkler" <
jgunkler@sprintmail.com>
"Michael Fletcher"
Junior Member
Posts: 2
Joined: Fri Mar 29, 2002 3:39 am

Meta Dynamics?

Post by "Michael Fletcher" »

Richardsons book is a very good look at SDs contribution to feedback
thought. SDs worldview in regards to causation is also interesting but not
discussed in detail anywhere that I know of. Dynamicists generally beat on
the point that correlation and causation are not the same, but maintain that
they can easily establish prove causal connections.

The worldview of System Dynamics is interesting in the fact that it takes a
rather firm view in favor of causation; that is, causation can be derived
and tested in models. (Structure causes behavior, and I can test best in a
computer model, therefore fulfilling the requirement for testing stated in
the Scientific Method.) The causal arrows in SD models are pretty powerful
statements, that Hume might have had some issues with. For the most part
arguments in causation have been the venue of Philosophers and
Statisticians. Statisticians have, until recently at least (if Pearl is to
be believed) have for the most part been keen to avoid causation arguments
and to stress the difference between correlation and causation.
Philosophers, including Hume and others, have always discussed causation
theory with gusto. One of the main assumptions of the Scientific Method is
that it can confirm causal links. Empirical observation or logic are
insufficient to prove causal links. (As Hume argued.) Others might see SDs
claims as a bit of a stretch. Can SD prove causation between anything? After
all, all we have is very simple "wrong" models, which are, at best,
imperfect reflections of the real world. What can we truly know or learn
from them? The models, as Forrester mentioned, may not be perfect, but they
beat out of the horribly fuzzy and inaccurate mental models which the models
hopefully help to update and correct.

In any event, this might be a good subject for a paper! Right now Im too
busy learning to write papers!

From: "Michael Fletcher" <
mefletcher@speakeasy.net>
=?iso-8859-1?Q?Jean-Jacques_Laub
Junior Member
Posts: 16
Joined: Fri Mar 29, 2002 3:39 am

Meta Dynamics?

Post by =?iso-8859-1?Q?Jean-Jacques_Laub »

Hi Mike and everybody.



As I have understood the question SD is getting interested in cosmogony.



I think that the SD being mainly the study of feedbacks, the question is:



Is any influence of something on something else belonging to a feedback that
after a time influences back the first thing?



Another way to express this is:



Are there circumstances where one can demonstrate that an influence exists
that

has a big chance not to belong to a feedback that influences back the first
thing?



In that case SD would not be the right tool.



I will illustrate this with an example.



Suppose that the world is finite and the time ahead is finite.

And suppose you fire a rocket out from the earth, and suppose that you
control the

trajectory so as to increase continually the distance of the rocket from
earth.

As the time passes you can get more and more convinced that it will not get
back on you head.

Now as the world is limited after a time, you will no more be able to
increase the probability

that it will come back unless the limit of time has come and everything is
being finished, there is no feedback loop. In a limited space, the distance
of the rocket cannot increase for ever.

So if the word is finite, depending on the time, you can never demonstrate
that there is no feedback.



If the word is finite and the time infinite the demonstration is even
easier.

At any time in the future, you will always be able to say, the rocket has
not yet fallen back,

but never mind, the probability is now stabilized, it is just a question of
(infinite) time to wait until it falls back.



Now if the world is infinite and the time finished. It is easy to
demonstrate that as the time passes, the distance is increasing and the
probability to fall back is decreasing and the probability of a feedback is
decreasing. Of course at the end of time, there will not have been any
feedback.



If the world is infinite and time is infinite, the answer is the same. The
distance will increase indefinitely, and the probability of feedback will
decrease indefinitely. So it will be easy to

Find influences that do not belong to any feedback.



Of course it is not a demonstration, but I feel that the problem of SD or of
feedback is governed by the finiteness of time and space.

Everything is or is not governed by feedback is depending of the finiteness
of space and time to my point.

Any controversial thoughts are expected with interest.

Regards.

JJ
From: =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <JEAN-JACQUES.LAUBLE@WANADOO.FR>
Locked