Strange spreadsheet concept

[Guenther Ossimitz]

Jim Hines wrote:

> Just a little point on instantaneous feedback. I think there is a (single?)
> class of examples in which simultaneity exists in the real world. If you
> push down on a table, doesnt the table instantaneously push up against you?
>

There are other examples for simultaneity. Consider the gravitational
force between two stars, rotating around a common center of gravity
in a twin-star constellation. Star A attracts star B and at the same time
star B attracts star A. Since the attraction changes the position of the other
object, this is a kind of instantaneous feedback loop.

Well, I know that Heisenbergs Unschaerferelation allows a slight
counter-argument against this claim of instantaneousness; we might
have to take into account that time in our physical world does not run
smoothly but in discrete steps of 10^-35 seconds or so.... So let me
conclude that I do not mean instantaneousness on a level of
10^-35 seconds, but on a level of continuous physical time and space.

--
Dr. Guenther OSSIMITZ
University of Klagenfurt
A-9020 Klagenfurt, Univ.str. 65 Austria/Europe
mail: ossimitz@bigfoot.com
home: http://go.just.to/ossimitz

["Daniel Poulson"]

I havent seen the package, but your criticism may be inappropriate for =
two reasons:

Firstly, circular flows do occur in the real world; consider refrigeration =
processes and the atmospheric water cycle.

Secondly, the correspondent refers to circular references, and not =
circular flows; circular references are necessary to describe any feedback =
system.

Regards,

Dan

Daniel Poulson
Vanguard Brand Management
+44 171 973 7655
dpoulson@vanguard-bm.com

[David Exelby]

I think there may be some misuse of the phrase "circular flow". I
expect that Khalid really meant circular referencing at an instant in time.
If so, then he is right to express the view that this does not occur in the
real world and we require the addition of the concept of time and delay to
describe feedback processes. Our stock flow representation of the world in
system dynamics forces us to think in this way (our software tools even give
us a warning too if we attempt circular referencing at an instant of time!).
A real system does not and indeed cannot respond instantly in a self
referencing manner (response times may be very short or very long but never
instant). To take the baby example below, there is positive feedback but it
is the level of adults that drives the flow of producing babies that
increases the level of babies and after a delay increases the level of
adults. I am not sure I would use the term circular flow (adults do not
turn into babies i.e. it is a one way street - an aging chain) but rather an
example of a positive
einforcing feedback/loop.

The brief description of the spreadsheet tool (I have not had the
time to check out the website) implies that it can be used to solve a
solution to a static problem that involves circular references but that any
of the variables can be expressed without reformulating the equations in the
cells that you would probably have to do normally. Its use in describing
dynamics may be of more limited use.

David Exelby
David Exelby <David.Exelby@hvr-csl.co.uk>

[Yaman Barlas]

After Khalids critical comment, there has been three responses that I
believe stem from differences of terminology. Khalid never said that ther=
e
are no closed loops! He said much more specifically that there can not be=
a
feedback loop in between two flows, without an intervening stcok variable.
His usage of the term "flow" is very specific; NOT in its general sense.
(So, you can find many loops involving "flows" only, if some of your
"flows" are actually "stocks").
And there is absolutely no problem in the example below, which depicts a
closed loop between Births (a flow) and Population (a stock). There WOULD
be a problem is one argued for a closed loop in directly between Birth an=
d
Death rates, without an intervening stock.

Yaman Barlas
From: Yaman Barlas <ybarlas@boun.edu.tr>

-------------------------------------------------------------------------=
--
Yaman Barlas, Ph.D.
Prof. of Industrial Engg
Bogazici University,
80815 Bebek, Istanbul
Turkey. Fax. +90-212-265 1800. Tel. 90-212-263 1540; ext.2073
http://www.ie.boun.edu.tr/faculty/barlas

System Dynamics Society (http://www.albany.edu/cpr/sds/ )
-------------------------------------------------------------------------=
----=9D=9D=9D

[dajoy@hoy.net (Ajoy Victor)]

I havent tested this but I wonder if these concepts can make better
simulation tools out of spreadsheets.


From: dajoy@hoy.net (Ajoy Victor)


The Spheresoft Modelers allows users to enter a value and multiple
formulas into each cell of Microsoft Excel, and it permits circular
references, making for spreadsheet models that are much more compact and
easier to use.

The Modeler represents a major advance in the power and flexibility of
spreadsheet technology. A standard spreadsheet limits the models you
can express to unidirectional flows from input cells (with one value) to
output cells (with one formula). Since the Spheresoft Modeler allows
you to enter a value plus multiple formulas into each spreadsheet cell,
with it you can create spreadsheets that specify all the relationships
between all the cells in your model -- end users enter values for the
cells they know and the Modeler calculates values for the cells left
blank. There is no more strict distinction between input and output
cells. This provides a more intuitive interface paradigm and it allows
users to work forward and backward in a problem very easily.

Special requirements: Excel 97/2000.
http://www.spheresoft.com/

[Khalid Saeed]

A good model must be identifiable in real world. Circular flows are
illogical as far as the real world is concerned. A software that allows
this is not appropriate for use in system dynamics modeling no matter
how compact a model it creates.

Khalid


_____________________________________
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
email: saeed@wpi.edu

SSPS Dept: http://www.wpi.edu/Academics/Depts/SSPS/
System Dynamics Society: http://www.albany.edu/cpr/sds/

[Jay Forrest]

>A good model must be identifiable in real world. Circular flows are
>illogical as far as the real world is concerned. A software that allows
>this is not appropriate for use in system dynamics modeling no matter
>how compact a model it creates.

Is this to imply that you believe there are no closed loops (circular
flows) and/or no conservation of mass (or on a reasonable basis elements)
in a closed or essentially closed system - such as the global carbon cycle?
Or of oxygen in a space station. I would propose that if one goes to the
highest level holistic view of systems many physical flows ultimately show
some characteristics of circularity.

Could it be that by focusing on lower level subsystems you are overlooking
higher level and ultimately more controlling loops?

I look forward to knowing more about what I am missing!

Thanks!
Jay Forrest



11606 Highgrove Drive
Houston, Texas 77077
Tel: 713-503-4726
Fax: 281-558-3228
E-mail: jay@jayforrest.com

[Ford Stone]

Khalid Saeed wrote:

> A good model must be identifiable in real world. Circular flows are
> illogical as far as the real world is concerned. A software that allows
> this is not appropriate for use in system dynamics modeling no matter
> how compact a model it creates.

Im sorry, but I dont quite get the nature of your objection. The more
people there are, the more babies arrive. The more babies there are, the
more people there are. Obviously, a bunch of other loops tie into this one
that keep it from ultimately accreting to infinity. But this is a circular
flow if ever there was one. In fact, the problem I have with the kind of
population model one creates in Stella, for example, has babies emerging
from a "cloud", a device I find extremely counterintuitive, when its
obviously a circular flow.
Can you elucidate?

Ford Stone
From: Ford Stone <fordston@caverns.com>

["George Backus"]

I think we are letting cross-paradigm semantics grossly misinterpret the
critical, key point Khalid is trying to make. All good SD models contain
causal circular flows. We call these feedback loops. These feedbacks loop
are NEVER defined in terms of simultaneity - remember good ol rates and
levels being required in loops? Rates and level break the simultaneity and
give real causality -- a time dependent flow phenomena -- a real
understanding of what is happening. If the problem (bounded system) being
addressed by the SD model had time constants that are much longer than some
loops being considered, then they are either converted to auxiliary
equations (we are only worried about the resulting condition) or they are
neglected as being irrelevant to the problem being addressed.

In spreadsheet, circular flow implies the simultaneous dependence of one
cells results on that of another that in turn references the original cell.
This is the defintionally the simultaneous solution of a "System of
Equations." The term "System of Equations" most often occurs in
econometrics or mathematical economics where it is assumed the world is
optimal and all information and action occur simultaneously. This view is
antithetical to SD. This view of the world may be mathematically insightful
and mathematically viable, but it contains no causal (time-dependent,
cause-and-effect) information. It is not SD.

Khalids note is simply pointing out the logical flaw (from a SD
perspective) of using such a approach and its absolute inconsistency with SD
causal thinking.

The definitions of a term within each paradigm has a different meaning in
another paradigm. I think people got badly caught not realizing this issue.
It may be a good lesson because our "client" audience often starts of in
another paradigm camp and if we start using our lingo without defining it,
the message we are delivering will be far different than what we (or they )
thought. It is like a foreign speaker misusing (has minimal knowledge of )
the language of the local country. The conversation (wording) sounds like it
should make sense but often doesnt.

G
From: "George Backus" <gbackus@boulder.earthnet.net>

[Khalid Saeed]

I have already responded to several private emails on this issue.

Indeed, feedback loop is the building block of systems and a model without
feedback would not create any interesting dynamics. A feedback loop must,
however, have at least one stock in its circular path that decouples flows,
creates integration and delays. In fact, as Jay Forrester pointed out in
his book Principles of Systems, stocks and flows must alternate in a
feedback loop. A feedback loop consisting only of flows does not exist and
if a software allows this, it would create models that are not identifiable
in real world.

Khalid Saeed
_____________________________________
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
email: saeed@wpi.edu

SSPS Dept: http://www.wpi.edu/Academics/Depts/SSPS/
System Dynamics Society: http://www.albany.edu/cpr/sds/

["Jim Hines"]

>In fact, the problem I have with the kind of
>population model one creates in Stella, for example, has babies emerging
>from a "cloud", a device I find extremely counterintuitive, when its
>obviously a circular flow.

There are two flows in the population model - a conserved flow and a causal
flow.

The CAUSAL flow is circular and is represented by the connection from births
to people (this is both a physical flow and a causal connection wrapped into
one -- births "cause" people), and by the purely causal connection from
people to births.

The CONSERVED (or physical) flow is NOT circular in your model. This flow
carries the stuff of which people are made. Because this flow is not
circular in this particular model, SD requires us to identify a source of
the conserved flow as a cloud (meaning we dont really care where the stuff
comes from).

Of course, it would be perfectly possible to represent the conserved flow as
circular. Materialists, viewing the people-stuff as organic molecules,
would track the decomposition of people into stocks of chemicals (perhaps,
still measured in "equivalent people units") which would then be transformed
back into people. If you are of a certain spiritual bent, you might view
the people-stuff as souls in which case you could track the movement of
souls from the deceased into a stock of souls waiting to be born and then
back into people.

System dynamics also lets you represent NON-circular CAUSAL flows. A
non-circular causal flow would either start with a constant (or exogenous
variable), or would end in a stock or auxiliary that was not an input into
any other variable.

Regards,
Jim
MIT and LeapTec
From: "Jim Hines" <jhines@MIT.EDU>

[Bryan.A.James@gd-is.com]

>>
A real system does not and indeed cannot respond instantly in a self
referencing manner (response times may be very short or very long but never
instant).
<<
An earlier response made the point about how semantic differences in different
disciplines/paradigms can lead to serious misunderstandings or errors. The
above statement is another instance of this. It certainly appears to be true of
the world as viewed from a classical paradigm. But it increasingly does not
appear to be true from the quantum physics paradigm. An area of significant
interest at the moment is "quantum entanglement" ( see, for example, New
Scientist, 6 March 1999, pp. 25-28) in which instantaneous response does seem to
occur. Not to be trivial, but it shows the importance of being clear about
assumptions and boundaries of meaning.

Bryan James
General Dynamics Information Systems
8005 S. Chester St., Suite 325
Englewood, CO 80112
bryan.a.james@gd-is.com
303-649-7548

["Jim Hines"]

Just a little point on instantaneous feedback. I think there is a (single?)
class of examples in which simultaneity exists in the real world. If you
push down on a table, doesnt the table instantaneously push up against you?

Regards,
Jim Hines
MIT and LeapTec
Jim Hines JHines@mit.edu

["George Backus"]

Relative to Jim Hines comment about the force and counter-force on a table.
The logic is just the force law and it is described (even in conventional
physics) as an auxiliary equation -- no loop. If the force does work, there
is integration (momentum or distance or velocity depending on your
interest), and a feedback loop is formed. Physics and SD are never in
conflict -- both describe reality (as best we understand it.)

G
From: "George Backus" <gbackus@boulder.earthnet.net>

[kevin.a.agatstein@us.arthurander]

In the very interesting discussion of "instantaneous" feedback Jim Hiness in
SD2299 points to a negative feedback loop that apparently demonstrates
simultaneity. That example was pushing on a desk, which "immediately" responds
with an equal force (compression) to stabilize your hand. I would like to
challenge Prof. Hiness
example as it is my understanding pushing on the table creates pushes down on
the top atoms of the desk,
which of course translates that force the to the underlying atoms. The desk
responds with the compressive
force upwards as the atoms are being squeezed and electrostatic forces are
building up. Regardless of the
physics (of which my gross simplifications do not do it justice anyway), the
point is the time constant for this balancing
loop is very small, but not zero. We only perceive it as zero as we as humans
operate on such a large time
horizon compared to atomic movements. We should not be surprised by this
phenomena, as mountains, which if they posed intelligence, would laugh at us for
thinking that they are static and immobile, while they rise and fall over the
centuries.

Of course, static models for solid compression serve engineers extraordinarily
well. Unfortunately, static
compression may not provide SD with an example of simultaneity.

I encourage more discussion relating to this topic.

Warmest regards,
Kevin A. Agatstein
kevin.a.agatstein@us.arthurandersen.com