Ventana software support forum

One thought keeps coming back when I think of feed-back/-forward and
"circular causal structures" (it says the right thing, but it doesnt yet
flow naturally off the tongue .

Feedback (and feedforward), as I learned about them as a former practicing
electrical engineer, both seem to be used often with rather simple single
input single output structures. I have discovered that mechanical
engineers sometimes deal in what they call modal analysis and MIMO
(multiple inputs multiple outputs) measurements, in which there may be tens
or hundreds of inputs and outputs to a system. For example, you can put a
747 on an array of shakers and inject specific vibrations into different
points on the structure. A mechanical engineer may want to place hundreds
of accelerometers on the airframe to determine the deflection (or velocity
or acceleration) of the airframe at all of those points. From that, the
response of the measured points to inputs at each of the shakers can be
calculated. The end result for them might be to create an animated wire
frame model of the airframe and to calculate the modes of vibration of the
airframe as a system.

While the analogy certainly isnt perfect, something about the number of
integrators, the number of inputs and outputs, and the fact that there is
no clear _one_ output seems related between modal analysis and SD. Not
being a modal analysis expert, I cant say what sort of language they use.

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

This conversation on feedback has been extremely useful to me!

During a visit to MIT and Albany last December, I talked with JWF and GPR
about some of the issues of applying SD to biology and medicine. In
looking through some of the D-files I found references to some of the
different viewpoints of engineers and socioeconomic workers re systems,
dynamics, feedback, etc.

The engineer is generally designing and building a new system. Specs can
be established, and specific testing devices (computer models, flight
simulators, wind tunnels) can be used to determine system behavior, improve
the design, etc.

In contrast, the socio-economist/manager is trying to understand the
behavior of complex systems which already exist, and within which we work
and live. Predictions can be made (or attempted), but controlled
experiments are next to impossible.

I noticed as I read these descriptions that biology did not fit easily into
either category.

As complex as biology appears to be, we may have some distinct
advantages. If I want to study thermoregulation, I can build an SD model,
run simulations, examine the rat, improve the model, examine another
-genetically identical- rat, hopefully under almost identical lab
conditions, and so on. This is not being done much (yet), but I believe
this perspective offers great promise.

I find the above discussion to be fascinating, because it highlights the
profoundly different ways that different disciplines might approach the use
of SD as a tool. For example, reading about policy decisions just does
not connect for biomedical scientists, even though the concept is right on.
So some social engineering (a la learning organizations?) needs to take
place here.


The thoughtful offerings on feedback from Sterman, Richardson, Shervais,
and Budiman provide a further excellent example of these cultural
differences. JS used the term feedback more generically. GPR provided a
nice contrast between the use of feedforward in behavioral decision
theory and engineering. BB then pointed out that cooling coffee is not
feedback at all, but rather an approach to equilibrium.

Whether this latter case would be described as feedback or not, the fact
remains that the temperature affects the rate of heat loss, and the rate of
heat loss affects the temperature. So there is certainly an interaction
between the two.

GPR said:

>Does the terminology matter? Yes, if the people you are trying to talk
>with have these terms but do not have our meanings. It would be bad if
>people with a "feedback/feedforward" lexicon were to decide that the loop
>in Newtons law of cooling is "neither feedback or feedforward," as Ed
>offered, since that is perilously close to saying there is not "a
>feedback loop" there.

This is absolutely critical for me! I am skating on thin ice when I
present an SD lecture to biomedical scientists who largely think that
modeling is for dilettantes. Given that there may well some engineers in
the audience, if I use the terms "incorrectly" (to them), I could very well
shoot my credibility all to hell.

On the other hand (if youre still with me), I REALLY like the term used by GPR:

"Circular causal structures" !

I think I could sell this term to my colleagues, regardless of their
background, without offending anyone. SD is -ideal- for studying the
behavior of CCSs. And yet it avoids any discipline-based pre-conceived
notions on whether this is feedback, feedforward, approach to equilibrium,
or compound interest. All of these can be defined as Circular Causal
Structures.

THEN within each discipline, based on the -problem- to be solved, and the
-relevant audience- (as GPR and others continually stress), the SDer can
tailor the terms "feedback" and "feedforward" as the situation suggests.

Does this make sense to John, George, Benny, Steve? And others?


ed gallaher
gallaher@teleport.com

"I for one" (as Floyd R. Turbo, AKA Johnny Carson used to say) think that
this has been a VERY productive thread!