Ventana software support forum

Jay wrote:

>It appears to me that the "boom" in popularity since Senges book has
>resulted in a lot of different descriptions of the field - some of which are
>somewhat in contradiction. As the field becomes "muddy" from confusion

Yes. I see an increasing number of people who claim that causal loop
diagrams are all one needs to worry with. While I accept that the
archetypes are useful, I also accept the notion that, in general, we arent
good at estimating the temporal dynamics of complex systems.

>The contraction I see happening here is that the debate is helping a lot of
>us lurkers (like me) who are relatively self-trained hone our understaning
>of the field and to move toward a "core" understanding. That contraction

To that point, thanks to whomever posted the "Desert Island Dynamics"
pointer. Im planning on catching up on some of the study I havent done
yet.

Finally, it seems that the discussion on discrete vs. continuous has
sometimes mixed that dichotomy with feedback vs. open loop. It seems that
the software Ive seen usually joins continuous and feedback OR discrete
and open loop. I conjecture that there are times when feedback would be
nice to have in the discrete case. For example, Ive done some modeling
work (discrete) to understand and improve the way we design printed circuit
boards. The quantities are small enough so that its hard to think of
"smoothing" it all into a continuous process. However, there are feedback
effects we ignored, because the software we used didnt express those
notions. I could have coded it using Fishwicks multi-modeling approach
using SimPack, but graphical tools are nice when communicating the results
with others.

(Yes, I did hear the comment from someone far more experienced than I that
the feedback/discrete case has occurred rarely in his experience; Im just
thinking out loud, as it were.)

Regards,

Bill

--
Bill Harris Hewlett-Packard Co.
R&D Productivity Department Lake Stevens Division
domain: billh@lsid.hp.com M/S 330
phone: (206) 335-2200 8600 Soper Hill Road
fax: (206) 335-2828 Everett, WA 98205-1298

>On Tue, 23 Apr 1996, Ed Gallaher wrote (among other things):
>
>[...]
>> I did see the articles in Science. However, these papers described
>> differential equations and not SD models. This is not a trivial
>> difference!
>[...]


In a thoughtful reply George Richardson commented:

>In fact, a system dynamics model *is* a model of the form dX/dt = f(X,p),
>where X is a vector of n states (levels), f is a (usually nonlinear)
>n-dimensional function, and p is a vector of parameters. System dynamics
>models are systems of differential (or, if you prefer, integral)
>equations. John Sterman is right to point out to us system dynamicists
>excellent work in Science in using using such models to address
>significant biological/medical problems, particular when the authors
>translate some of their structural insights into feedback terms. Such
>work is system dynamics.

This topic raises many issues, and I will expand on these in another message.

But first, George, the definitions and semantics contained in this
paragraph are disturbing and confusing to me.

Id prefer to discuss this privately, and then summarize the result to the
group.

I am quite serious about this. We all agree that modeling is useful
because it clarifies and explicates our mental models. My mental model of
"what is system dynamics" does not fit the paragraph above. And perhaps
our collective "marketing" of SD is fuzzy as a result.

Premise:

There are a number of real-world PROBLEMS that involve continuous processes
over time. Many of these involve feedback loops. These processes and
problems existed long before calculus, diff eq, or SD, and are independent
of our ability to understand them or solve them. They just "are".

Various techniques have been developed to study such problems.

If we describe the above problems as "systems", then I suppose any study of
them could be defined as "system dynamics". If so, we need another clear,
unamgigous name for whatever it is that Forrester "invented/developed".

Calculus (Newton and Leibniz), diff eq, control systems engineering, etc.
provide us with a set of tools to study and interpret these problems.
These techniques evolved over several hundred years. The concept of a
limit, and thus calculus itself, was invented largely because it was not
feasible to add up all the little bits and pieces.

Forrester and his colleagues proposed and developed a unified method of
Stock-and-flow diagrams and computer simulation (DYNAMO, STELLA, Powersim,
Vensim, etc). Computers can now add up the bits and pieces for us.

(I am going to state that this latter technique can be conducted without
higher level math background. We can discuss this in another dialog if
there is any disagreement on this point.)

MY PROBLEM: I am looking for a clear, unambiguous NAME for the SPECIFIC
PROCESS which uses stock-and-flow diagrams and closely-linked computer
simulations.

The name which describes the PROCESS must be distinct from name given to
the *set of problems* to which this procedure is applied.

(The railroads defined themselves as being in the railroad business. This
might be considered semantic trivia, until the airlines came along and
pointed out that this was the "transportation" business. The -solution-
(horse vs railroad vs airplane) cannot be equated with the -problem-
(transportation).)

So, I am looking for a clear, unambiguous name.

A fuzzy overlap between systems thinking and system dynamics is NOT OK.

This process does not include hexagons.
This process does not include causal-loop diagrams.

(I cast no aspersions at these latter techniques; they each serve a very
real purpose. But not *this* purpose.)


Should we call this invention/technique "System Dynamics"?

Or "Stock-and-flow Modeling"?
Or "Forrester Modeling"
Or what?

I would suggest "System Dynamics", and I would guess you would too.

But this works if, and ONLY if this term is taken to mean the
invention/technique of stock-and-flow diagrams and computer simulation.

Not if it is meant to describe the *set of problems* for which it is used.
Not if it is confused with Systems Thinking.
Not if it includes causal-loop diagrams.

I am not seeking to discuss the merits of these various techniques or
approaches. Just an unambiguous NAME.

If you still feel that your paragraph above is precisely stated, then can
you provide me with a name that we can agree upon to describe
"stock-and-flow diagrams and the collection of software including
DYNAMO...... Vensim", but NOT including diff eq and other numerical
analysis solutions?

This is not meant to be just argumentative, or a power struggle.

I would like to proceed with a discussion of "whatever it is" vs
"traditional math", but we cant even begin this discussion without a
resolution of the definitions.

Thanks,

Ed
gallaher@teleport.com

P.S. In discussing this several months ago with another colleague (by
phone), he implied that STELLA (i.e. the graphic interface which makes the
process user-friendly, and eliminates the potential diagram-coding
discrepancies of DYNAMO) was a HUGE step forward. But he felt that what
Forrester "invented" was just a way of diagramming and conducting
traditional numerical analysis, i.e. a technical footnote in the calculus
thread.

He (with a very strong math background, by the way), would probably agree
with your paragraph above. We never got this semantics question
resolved. He knew he was right, and knew I was wrong; needles to say
this did not shed too much light on the issue. Writing has advantages over
the phone of course; it allows some contemplation. In addition, I observe
that your style provides more light, less heat than his does, so Im
optimistic that I (we) can achieve some clarity here.

This might shed some further light on where I am coming from on this.

irt: gallaher@teleport.com (Ed Gallaher), Thu, Apr 25, 1996 7:44 PM EST

This is simply marvelous! I wonder what preculuded us from getting here for
so long. Ive read this message several times, now lets see if I have an
appropriate understanding.

The Universe IS! There are different techniques for describing and
understanding how it works. For us to communicate on some meaningful level we
must come to some acceptable agreement as to what labels applied to our
techniques mean, otherwise we just confuse each other, and are often apt to
debate unsubstantiated meaninglessness.

If two different techniques both prove equally adequate for describing the
operation of a system then, althought they remain different techniques, they
are essentially equivalent. I rather relate this to the "universal Turing
machine" being the simplest definition of a computer, and as all computers
can represent the univeral Turing manine they are thus equivalent (more or
less).

I would also relate the discussion to:

All grass is green. This is grass. Therefore it is green.

Which should not be confused with the fallacy:

All grass is green. This is green. Therefore it is grass.

Regarding part of Eds premise:

"There are a number of real-world PROBLEMS that involve continuous processes
over time. Many of these involve feedback loops."

Is there a real-world problem that doesnt involve feedback loops? Even if
one existed, as soon as it was considered, would that create a feedback loop?

Regarding Eds stated "MY PROBLEM," my personal perception which has
developed over time is that that "System Dynamics" is the technique (process)
which uses stock-and-flow diagrams and closely-linked computer simulations to
understand real world problems. I supposed this was why I felt comfortable
writing a paper titled, "System Dynamics: An Operational Guide to the
Universe," presumptuous as it may have been.

As for the comments regarding Systems Thinking, CLDs, etc. I consider them
each to be techniques which provide more and more explicit representations of
what I seek to understand. I seem to repeatedly progress from systematic
thinking, to systems thinking, to CLDs, to System Dynamics (or DEVS -
Discrete Event Simulations, which I perceive as a technique different from
System Dyanmics). As I progress through each technique the understanding (or
the fallacy of my understanding) seems to become more explicit with the more
explicit nature of the technique.

Gene Bellinger
CrbnBlu@aol.com

irt: gallaher@teleport.com (Ed Gallaher), Thu, Apr 25, 1996 7:44 PM EST

This is simply marvelous! I wonder what preculuded us from getting here for
so long. Ive read this message several times, now lets see if I have an
appropriate understanding.

The Universe IS! There are different techniques for describing and
understanding how it works. For us to communicate on some meaningful level we
must come to some acceptable agreement as to what labels applied to our
techniques mean, otherwise we just confuse each other, and are often apt to
debate unsubstantiated meaninglessness.

If two different techniques both prove equally adequate for describing the
operation of a system then, althought they remain different techniques, they
are essentially equivalent. I rather relate this to the "universal Turing
machine" being the simplest definition of a computer, and as all computers
can represent the univeral Turing manine they are thus equivalent (more or
less).

I would also relate the discussion to:

All grass is green. This is grass. Therefore it is green.

Which should not be confused with the fallacy:

All grass is green. This is green. Therefore it is grass.

Regarding part of Eds premise:

"There are a number of real-world PROBLEMS that involve continuous processes
over time. Many of these involve feedback loops."

Is there a real-world problem that doesnt involve feedback loops? Even if
one existed, as soon as it was considered, would that create a feedback loop?

Regarding Eds stated "MY PROBLEM," my personal perception which has
developed over time is that that "System Dynamics" is the technique (process)
which uses stock-and-flow diagrams and closely-linked computer simulations to
understand real world problems. I supposed this was why I felt comfortable
writing a paper titled, "System Dynamics: An Operational Guide to the
Universe," presumptuous as it may have been.

As for the comments regarding Systems Thinking, CLDs, etc. I consider them
each to be techniques which provide more and more explicit representations of
what I seek to understand. I seem to repeatedly progress from systematic
thinking, to systems thinking, to CLDs, to System Dynamics (or DEVS -
Discrete Event Simulations, which I perceive as a technique different from
System Dyanmics). As I progress through each technique the understanding (or
the fallacy of my understanding) seems to become more explicit with the more
explicit nature of the technique.

Gene Bellinger
CrbnBlu@aol.com

To: Ed Gallaher, regarding your message #58

Ed,

I think restricting the term "system dynamics" in the way you suggest will cause
a great deal of confusion. As an example, consider causal loop diagrams: Many
people (including me) consider causal loop diagrams as part of system dynamics:
CLDs appear in many of the texts on system dynamics; they are taught as part of
most courses on system dynamics; Vensims underlying graphic metaphore is CLD
(Althgouh it is possible to make these CLDS **look** like stock and flow
diagrams; "flow-pipes", like all of Vensims arrows, actually indicate causal
connections).

I think we should keep "system dynamics" a term that allows growth of the field.
Causal loop diagrams were added later (though not much later) than Industrial
Dynamics; but it seems that CLDs are now part of the fields vocabulary. I
think we should welcome this sort of gradual development.

If we want to refer to something that is part of the field, I think were just
going to have to name it as best we can. Your use of the term "stock-and-flow
diagram", for example, seemed pretty serviceable to me.

Jim Hines
jimhines@interserv.com