I got George Richardsons permission to post his May 15 comments on feed
back loops to the CYBCOM list - the cybernetics equivalent to this list.
An interesting exchange developed between Jixuan Hu and Martin Taylor.
Here is Jixuans original posting in response to Georges comments. I
have his permission to post this to the system dynamics list with a
request to keep him updated. I will now try to summarize the various
positions and share them with both lists.
Hope this proves to be interest.
------- Forwarded message follows -------
Re the discussion about feedback loops, the following is what I present
to my students at George Washington University, CalState, and Beijing
University:
Principle of Circular Causality (The Law of System Inertia)
A complex system will keep its initial behavior (stable, growth,
oscillation of chaos) unchanged as long as the structure of the
dominating circular cause-effect loop within the system is kept unchanged.
There are two ways by which any cirular cause-effect loop is formed. The
first is "measurement," (i.e., a sensor is involved) in which a
measurement of the "effect" is linked back to influence the "cause." Thus
the word root "back" is used. The second is "predictive predication,"
(i.e., a forecasting agent is involved) in which a prediction about the
"effect" is generating influence on the "cause" before the "effect" can
be measured. Thus the word root "foward" is used. Therefore, "feedback"
and "feedforward" are two kinds of cause-effect loops.
There are also two kinds of outcomes for both feedback and feedforward
loops. The first is "stabilizing," or "goal-seeking." The second is
"growing," or "exponential departuring." By tradition, the first kind of
outcome is described as "negative," because the circularity is diminishing
the difference between the future state and the current state, and the
second kind of outcome is decribed as "postive," because the circularity
is amplifying the difference between the future state and the current
state. (This use of terminology confuses a lot of students, so Peter Senge
replaced them with "balancing" and "growing." I emphasize the amplifying
or diminishing the difference between future and current states - and
students found it very helpful.)
Therefore, combining two ways of forming and two kinds of outcomes, we
have altogether four types of circular cause-effect loops:
Positive Feedback (PfB):
A measurement about the "effect" of a process is influencing the "cause"
of that process in a way that the "effect" is growing exponentially.
Negative Feedback (NfB)
A measurement about the "effect" of a process is influencing the "cause"
of that process in a way that the "effect" is stabalized at a particular
point.
Positive Feedforward (FfF)
A prediction (anticipation) about the "effect" of a process is influencing
the "cause" of that process in a way that the predicted future state can
never be reached. (Ah-ha, "self-defeating prophecies" called in other
disciplines.)
Negative Feedforward (NfF)
A prediction (anticipation) about the "effect" of a process is
influencing the "cause" of that process in a way that the predicted
future state will be reached. ("Self-fulfilling prophecies.")
These four types of cicularities can be put into a digram with a
unitifying frame of reference. An intereting point is that for the
feedback side (measurement related), an observer (of that measurement)
can be revealed, and for the feedfoward side, a forecasting agent -
functioning at a level higher than the observer - can be revealed. This
leads to the discusion of first order versus second order, type I science
versus type II sciences, etc., etc.
Hope this is helpful.
~~~~~~~~~~~~~~~~~~~~~~~
Jixuan (Jason) Hu, Ph.D.
e-mail: jixuanhu@gwis2.circ.gwu.edu (o):301-422-6919 (f):301-422-2618
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Anthony Gill phone: +44 (0)1295 812262
Phrontis Limited
Beacon House fax: +44 (0)1295 812511
Horn Hill Road
Adderbury email: t2@phrontis.demon.co.uk
Banbury
OXON. OX17 3EU
U.K.