modeling systems

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
Bill Buchanan
Junior Member
Posts: 10
Joined: Fri Mar 29, 2002 3:39 am

modeling systems

Post by Bill Buchanan »

Re: modelling systems.

Periodically a position is expressed in some "systems" circles
that "systems cant be modeled". Every time I see it I
wonder, what could this position be thinking, and why does it persist in
the face of such overwhelming evidence to the contrary? What motivates
it? Does it offer any value, other than the obvious, that subsystems
can be modeled?

Of course whole "systems" can be modeled. Anything can be modeled, and
simulated, and even possibly synthesized into a "virtual" or
"semi-virtual" entity. Examples abound, but here are a few.
One need only look to the emerging "complexity sciences" (e.g., see the
Santa Fe Institute or New England Complexity Science web pages,
santafe.edu, and necsi.org, respectively) for increasingly sophisticated
and rapidly evolving examples.

Biology........modeling of.genes, chromosomes, cell structures, allowing
for.genetic engineering of plants and animals

Chemistry.........the periodic table, modeling of.all molecular
structures........allowing for synthesized proteins

Physics.......general physical model, allowing construction of new
elements, and building of.nanostructures

Organisms and populations...........sequencing of complete organisms
(about 25 to date), simulations of artificial life structures, and new
traffic structures for roadway and transportation systems

Business organization.........."virtual"/distributed organizations,
study and development of
new team communication/productivity structures, re-engineering of whole companies.

Cosmology.......Newtonian and Einsteinian mechanics and dynamics, Theory
of Everything, Grand Unified Theory, General Systems Theory,
.........generating new astrophysical and astrobiological breakthroughs
for instrumentation and exploration of space frontiers

These are just a few examples of the contemporary advance of modeling,
simulation, and synthesis of whole systems.........complete
computational analysis and synthesis......."no wholes barred". So, I
have a certain impatience with the position that "systems cant be
modeled"..........of course they can, and not only modeled but simulated
and even synthesized, depending on their physical scale relative to
human scale. The "big systems" thinkers never shirk(ed) at this
possibility, like Newton, Einstein, Darwin, Freud, Crick, Gell-Mann, and
many others, esp. those designing and building schemes like new towns,
cities, planes, cars, organisms...............and the list goes on.

Cheers.......Bill Buchanan
From: Bill Buchanan <
waba@scientist.com>
"George Backus"
Member
Posts: 33
Joined: Fri Mar 29, 2002 3:39 am

modeling systems

Post by "George Backus" »

This note seems to indicate a different view of the word "systems." It is
true these noted individuals did grab a system with large boundaries
(although Einstein only assumed one simple steady-state universe rather than
the multiple dynamic ones we now "know."). But the boundary is not the
system. ALL of these people wanted to look at one problem. Einstein was
only after the self-consistency of physical law (special relativity) and the
extension of it in gravity (general relativity.) Einstein did not worry
about galaxy formation (still a problem), or black hole formation (his
equations did allow such research that he tried to originally discredit).
Einstein hated most of quantum mechanical theory. (There is no "whole
system" without both.) Watson and Crick were only interested in the
geometry of the DNA molecule. They had no vision of polymerase reactions and
how to understand isolated genetic material. Newton was only interested in
the basic laws of force. These "ideas" were more like the, in vogue, SD
"molecules" than a system.

Darwin applied one mechanism to the changing biota. He did not understand
genes. He only vaguely understood its relation to economic and politics.
Again this would be a "molecule." Gell-Mann simply wanted to find the
pattern reflecting the primary forces of physics. The quantum mechanical
implications of theory required other researches and theoreticians. He
modeled the connections of several "SD molecules" to allow the making of
sub-systems (particles).

ALL the examples you will ever find of a successful "discoverer," will show
the search for the solution(s) of a small (WELL-DEFINED) problem set. Only
those aspects of the system needed to solve the problem are included in
their thinking or their theory -- and nothing else. They were spectacular
systems thinkers. They could determine what was important and only focus on
those dynamics.

The size or importance of the system is irrelevant. Any real system is
infinitely complex with infinite interactions and mechanisms (even it they
simple need to be repeated 10 godzillion times as in a crystal lattice
simulation). Infinity tends to be rather overwhelming to finite beings with
finite-lives. You are only interested in a few aspects of the system (those
relevant to the problem of interest), so why waste time modeling those
aspects not concerned with THE PROBLEM?

In summary you can model problems that imply systems with big boundaries and
"global" impacts, but you only model a (very) small part of that system.
That part contains only the components that solve the problems and contains
only those components that you actually "believe" you can understand (your
hypothesis).

As an aside:

We havent yet gone to discussing "belief" systems modeling: We cant even
prove the hypotheses in our simply models. We simply select those components
based on our best understanding. We can call it postulates, axioms, or can
call it guided blind-belief. We use data but that just gives us confidence
in the use of the model, not in its real validity (Something about
"paradigms" vaguely come to mind here....) We cannot really model a "system"
because we can never truly understand any system. We can, at best,
understand our idealized (made up) view of it. Thus, the mechanical and
practical impossibilities of "whole systems modeling" still pales to the
philosophical limitations.

G


George Backus, President
Policy Assessment Corporation
14604 West 62nd Place
Arvada, Colorado, USA 80004-3621
Bus: +1-303-467-3566
Fax: +1-303-467-3576
George_Backus@ENERGY2020.com
Khalid Saeed
Senior Member
Posts: 79
Joined: Fri Mar 29, 2002 3:39 am

modeling systems

Post by Khalid Saeed »

Bill:

Interesting point. I think our loose use of terminology might have created
some confusion. Drawing from the GST jargon, the systems we model are
ABSTRACT systems built around specific problems or families of problems. We
try not to model CONCRETE systems representing organisms with a discernable
boundary all functions circumscribed by that boundary, since there is too
much detail in those not relevant to the problem being addressed. Drawing
an analogy with the practice of art, our models are created in the
impressionist tradition rather than in a realist tradition. The former
highlights a pattern, the later incorporates copious detail not necessarily
relevant to the main theme.

It seems your examples fall in the category of abstract systems and hence
are consistent with our practice.

Khalid
From: Khalid Saeed <saeed@WPI.EDU>


_____________________________________
Khalid Saeed
Professor and Department Head
Social Science and Policy Studies
W. P. I., 100 Institute Road
Worcester, MA 01609, USA

Ph: 508-831-5563; fax: 508-831-5896

http://www.wpi.edu/Academics/Depts/SSPS/
"Tom Forest"
Newbie
Posts: 1
Joined: Fri Mar 29, 2002 3:39 am

modeling systems

Post by "Tom Forest" »

I enjoyed Khalids comment that "our models are created in the
impressionist tradition rather than in a realist tradition." And George
Backus makes a good distinction between the scale and the comprehensiveness
of a model. There are in all of Mr. Buchanans examples of "Whole System"
models a finite number of state variables described. The periodic table
has only a handful, as do Newtonian and Einsteinian mechanics and dynamics,
and C-G-T-A sequencing of DNA was narrowly defined.

The distinction between a model, a prototype, and a complete recreation had
been blurred in this thread. In SD, we are interested in the dynamics
behavior of a narrowly focused set of state variables. They may be
disaggregated and replicated into discrete elements. For instance,
consider a model of Turkey. Start with population, capital stock, natural
resources, and the educational experience of the people. Disaggregate each
into an aged pipeline. Then create arrays such that each element
represents a square kilometer. Now I have millions of state variables
(necessary to fully describe the system as Ive defined it), describing a
large system. But it is not a prototype, much less a complete recreation.
My knowledge of the people is still vanishingly small. I know nothing
about the health of these people, their language, music, religion, or food
(aside: well, actually I do, and all I can say is - yum!).

As a modeler, I start with a reference mode of behavior for a small number
(half dozen or less, typically) of state variables. I build a model of
sufficient complexity to reproduce that behavior in a causally plausible
way. Then I experiment with the policies and parameters to see if I can
modify the reference mode in an interesting (better, worse, faster, slower,
less oscillatory, et al) way. I may add or subtract some state variables
along the way while building or experimenting, but I am never close to
using all the state variables that are theoretically available to me. Most
omitted variables act on the wrong time scale (too fast or slow to affect
my variables of interest) or have no plausible causal connection to my
chosen variables.

Tom Lum Forest
From: "Tom Forest" <
tforest@cordada.com>
Locked