[From Fred Nickols (2004.11.28.1016 EST)]
Somewhere along the way in one of the more recent CSGNet digests, Marc
Abrams wrote:
Fred Nichols has had some personal experience with using SD in a large
organization. What was your experience with it Fred?
That of an observer, client and customer. Several years ago I contacted
John Sterman and asked him who was the best SD modeler in the world. He
said it was Jack Homer. So, I got Jack to come in a do some SD modeling
work with us regarding the forecasting of test-taker volumes (I was with ETS
at the time). Jack did some really good work and our ability to forecast
test volumes was greatly improved. Jack was later brought back by others in
ETS to do some additional work and again all were quite pleased with the
results.
I don't consider myself any more knowledgeable about SD than I do PCT. I
try to know enough to recognize when a particular tool might prove useful
and to then get someone who does know what he/she is doing to confirm my
view and then pursue the actual task. Jack, for example, initially thought
that what I wanted to do could be done with a spreadsheet; later he agreed
that SD was needed.
I also had a young fellow working for me who had taken quite an interest in
SD and I had him work with Jack. Sadly, my confidence was misplaced because
I could never get this young fellow to put boundaries around his work. He
always insisted on doing what I called "modeling the entire universe so as
to look at one cookie crumb." I finally managed to foist him off on some
unsuspecting colleagues of mine and he became their problem. I'm afraid
they never forgave me for that one.
So, in sum, my experience with SD was mixed. On balance, I think SD is a
useful tool and I think that SD and PCT have some things in common and I
suspect they have some major points of conflict.
Regards,
Fred Nickols
nickols@att.net
From [Marc Abrams (2004.11.28.1318)]
Thanks for the response and critique, my comments follow;
In a message dated 11/28/2004 10:28:49 AM Eastern Standard Time, nickols@WORLDNET.ATT.NET writes:
···
[From Fred Nickols (2004.11.28.1016 EST)]
I also had a young fellow working for me who had taken quite an interest in
SD and I had him work with Jack. Sadly, my confidence was misplaced because
I could never get this young fellow to put boundaries around his work. He
always insisted on doing what I called “modeling the entire universe so as
to look at one cookie crumb.”
So, in sum, my experience with SD was mixed. On balance, I think SD is a
useful tool and I think that SD and PCT have some things in common and I
suspect they have some major points of conflict.
Fred, my question had more to do with what SD was successful at helping you with. My experience has been that it is a wonderful (usually anyway :-)) tool when you deal with homogenous collections of data. It has been less effective in dealing with individual items. Whether they be humans or widgets.
When certain modelers Homer and Levine being two prominent SD Modelers have done models that entail individuals they have use an output control model with the environment providing the reference.
THIS in my opinion, is the reason why SD models have had an extremely difficult time in translating the ‘collective’ behavior of an SD model into actionable ideas for an individual.
I think both PCT and SD can benefit each other. But that will not happen until someone who knows PCT can show how a PCT inspired model will help them implement their models more effectively then how they are currently doing it.
SD modelers already ‘know’ about PCT. What they don’t know is the significance the model has on an indivuduals behavior and for a large majority of the work currently done in SD it just doesn’t matter.
With the knowledge of PCT, I think this could open up a whole new avenue of exploration for SD modelers and it would provide PCT with a base of modelers capable of future exploration.
Am I the only one who sees this potential?
Marc
Yes, this is a big, common problem in modeling. Over enthusiasm is very easy when your imagination starts considering the possibilities. Rick’s attempt at a model of our ‘economy’ is in the same vein
[From Bruce Abbott (2004.11.28.1615 EST)]
My own experience with system dynamics (SD) is limited to some fussing
around with Vensim and reading a couple of Jay Forrester’s books (e.g.,
Industrial Dynamics, 1961). From that admittedly limited contact,
it seems to me that SD is just a set of tools for modeling dynamic
systems. You can use SD to implement a specific PCT model, for example.
SD is more general in the sense that one could create SD models that are
not consistent with PCT principles, such as the restriction that all
variables and signals be intended to represent potentially real physical
entities in the system being modeled. Put another way, PCT models are SD
models, but not all SD models are PCT models.
Bruce A.
From [Marc Abrams (2004.11.28.1700)]
Hey Bruce, thanks for weighing in here. 
In a message dated 11/28/2004 4:21:06 PM Eastern Standard Time, babbott@MFIRE.COM writes:
···
[From Bruce Abbott (2004.11.28.1615 EST)]
My own experience with system dynamics (SD) is limited to some fussing >around with Vensim and reading a couple of Jay Forrester’s books (e.g., >Industrial Dynamics, 1961).
Have you read Principles of Systems? Have you read George Richardson’s book Feedback Thought in Social Systems?
I ask for a very specific reason and will not elaborate until I know the answer to these questions
From that admittedly limited contact, it seems to me that SD is just a set of >tools for modeling dynamic systems.
You can use SD to implement a specific PCT model, for example. SD is more >general in the sense that one could create SD models that are not consistent >with PCT principles, such as the restriction that all variables and signals be >intended to represent potentially real physical entities in the system being >modeled. Put another way, PCT models are SD models, but not all SD models >are PCT models.
Well said Bruce
Marc
[From Bruce Abbott (2004.11.28.1900 EST)]
Marc
Abrams (2004.11.28.1700) –
Hey Bruce, thanks for weighing in here.
. . .
Have you read Principles of Systems? Have you read George
Richardson’s book Feedback Thought in Social
Systems?
I ask for a very
specific reason and will not elaborate until I know the answer to these
questions
I have read Principles of Systems but not the Richardson
book.
Bruce A.
From [Marc Abrams (2004.11.28.2351)]
In a message dated 11/28/2004 6:56:28 PM Eastern Standard Time, babbott@MFIRE.COM writes:
···
[From Bruce Abbott (2004.11.28.1900 EST)]
I have read Principles of Systems
Great.
but not the Richardson book.
You should, you would enjoy it immensely.
I asked for this reason. Richardson does a wonderful job of both providing a historical perspective AND the philosophical context for PCT among other work in the field. You would appreciate it. It is out of print, but I would lend it to you if you wanted to read it.
To summarize; Richardson found the roots to feedback thought in the social sciences had two distinct branch’s. The Cybernetics thread and the Servomechanism thread.
The major difference is the Cybernetic thread deals ONLY in negative feedback loops and the Servomechanisms deal in both negative and positive feedback.
Richardson gave a full chapter to PCT and to this day, loves it. He has been a bit disappointed that Bill decided to stick to the bottom up approach and not go for the bigger prize at the higher levels. (it’s NEVER too late Bill :-))
PCT was unique among all the theories presented by George. PCT is the only theory that resides in BOTH threads. Although cybernetic by philosophy, Bill’s modeling and a few other things (like control) place him in the Servomechanism thread as well.
SD is in the Servomechanism thread.
SD’ers think they are ‘doing’ PCT, usually using widgets instead of ‘perceptions’ for the quantities moving through the negative feedback loops. And in reality they are. Mathematically, there of course is no difference between either negative feedback loops. The only thing is what travels through the loops.
But you and I know this is NOT the case. A ‘perception’ is NOT simply a ‘signal’ it has cognitive significance and PCT is NOT simply a negative feedback loop, EVEN though the mathematics may be the same.
In PCT what enters a control system is determined by the RECIEVING input function NOT the environment. THIS is HUGE.
Marc
