It is generally recognised that Feedback is fundamentally crucial in SD.
However, what is an operational definition of "Feedback" in SD?
Both your qualitative and quantitative (cf. feedforward?)
statements are welcome.
Is Feedback mainly a thinking perspective or a concept/ construct?
Or it is merely another name (a subset) of information (flow).
IMO, the word Feedback has been already implied when mentioning
closed-loop in SD as it might have both Feedback and Feedforward
functional implications.
Comments?
Shaun TANG <stang@fcit.monash.edu.au>
Definition of Feedback
-
George Richardson
- Senior Member
- Posts: 68
- Joined: Fri Mar 29, 2002 3:39 am
Definition of Feedback
System dynamicists use the phrase "feedback loop" to stand for all manner
of circular causal loops. It is fair to say that system dynamicists
probably do not have a definition for the word "feedback" by itself. It
is the loop we label.
Historically, that usage is not universal. Von Bertalanffy, for example,
objected to using the term feedback in such an all-encompassing way and
preferred to reserve it for the deliberate "feeding back" of information
in a system, probably for control pruposes. So he objected to thinking
about the structure of, say, a predator-prey system as a feedback system,
because all the loops were naturally occurring -- none involved the
deliberate "feeding back" of information by actors or managers in the system.
Ashby was comfortable with a usage like ours, but he got confused by the
discrete "loop" x(t) = .5*y(t-1), y(t) = .5*x(t-1), which he concluded
was a goal-seeking positive feedback loop, and that was so bizarre to him
he concluded people should stop using the feedback concept.
The Macy meetings in the 1940s and 50s, which began the field of
cybernetics, adopted the definition that feedback was "the alteration of
input by output." That definition is much like the engineering one
sent to this list. It is unlike the way a system dynamicist thinks about
it because it focuses on ideas like input and output, which do not have
much meaning in an endogenous, circular causal system. Since were
dealing with a fundamentally endogenous perspective, we think about
"feedback loops" but not about "feedback" by itself.
Thus we dont think much about "feedforward" either. If it has a real
meaning in an endogenous circular causal system, it would appear as just
another link or set of links which add to the causal complexity of the
system. I could imagine that restructuring the beer game, for example,
so that the order stream at the customer end is feed to the other
positions in the game could be thought of as a kind of "feedforward" in a
new control system designed to eliminate the oscillations in the game.
But system dynamicists would probably refer to this as a restructuring of
the information system and would not really know whether the restructuring
was "forward" or "back."
One can find definitions of feedback loops and the feedback perspective
in various system dynamics writings, notably the early writings of
Forrester. Youll also find much discussion of this in Feedback
Thought. I became very interested during that research to see the number
of different ways the idea(s) was discussed. There are serious
differences in emphasis in the various definitions, which have serious
implications for practice, which apparently depend on the literature and
contacts one has access to.
...GPR
-----------------------------------------------------------------------------
George P. Richardson G.P.Richardson@Albany.edu
Rockefeller College of Public Affairs and Policy Phone: 518-442-3859
University at Albany - SUNY, Albany, NY 12222 Fax: 518-442-3398
-----------------------------------------------------------------------------
of circular causal loops. It is fair to say that system dynamicists
probably do not have a definition for the word "feedback" by itself. It
is the loop we label.
Historically, that usage is not universal. Von Bertalanffy, for example,
objected to using the term feedback in such an all-encompassing way and
preferred to reserve it for the deliberate "feeding back" of information
in a system, probably for control pruposes. So he objected to thinking
about the structure of, say, a predator-prey system as a feedback system,
because all the loops were naturally occurring -- none involved the
deliberate "feeding back" of information by actors or managers in the system.
Ashby was comfortable with a usage like ours, but he got confused by the
discrete "loop" x(t) = .5*y(t-1), y(t) = .5*x(t-1), which he concluded
was a goal-seeking positive feedback loop, and that was so bizarre to him
he concluded people should stop using the feedback concept.
The Macy meetings in the 1940s and 50s, which began the field of
cybernetics, adopted the definition that feedback was "the alteration of
input by output." That definition is much like the engineering one
sent to this list. It is unlike the way a system dynamicist thinks about
it because it focuses on ideas like input and output, which do not have
much meaning in an endogenous, circular causal system. Since were
dealing with a fundamentally endogenous perspective, we think about
"feedback loops" but not about "feedback" by itself.
Thus we dont think much about "feedforward" either. If it has a real
meaning in an endogenous circular causal system, it would appear as just
another link or set of links which add to the causal complexity of the
system. I could imagine that restructuring the beer game, for example,
so that the order stream at the customer end is feed to the other
positions in the game could be thought of as a kind of "feedforward" in a
new control system designed to eliminate the oscillations in the game.
But system dynamicists would probably refer to this as a restructuring of
the information system and would not really know whether the restructuring
was "forward" or "back."
One can find definitions of feedback loops and the feedback perspective
in various system dynamics writings, notably the early writings of
Forrester. Youll also find much discussion of this in Feedback
Thought. I became very interested during that research to see the number
of different ways the idea(s) was discussed. There are serious
differences in emphasis in the various definitions, which have serious
implications for practice, which apparently depend on the literature and
contacts one has access to.
...GPR
-----------------------------------------------------------------------------
George P. Richardson G.P.Richardson@Albany.edu
Rockefeller College of Public Affairs and Policy Phone: 518-442-3859
University at Albany - SUNY, Albany, NY 12222 Fax: 518-442-3398
-----------------------------------------------------------------------------
Definition of Feedback
Benny Budiman in SD0222 offers a clear description of feedback and
feedforward in engineering. He concludes by saying:
"IMO, feedbacks in SD are actually causal links that include both
feedback and feedforward mode of "controlling" the system."
Benny is exactly right. The use of the term "feedback" in system
dynamics covers any process in which the state of a system influences
action, which affects the system, and thereby affects further action.
This obviously covers the engineering use of the term feedback. But, it
also covers "feedforward". In feedforward, information on the state of
the system is used to create an estimate or projection. The estimate or
projection is also part of the "state of the system" and guides action
which in term affects the system, estimates about it, and therefore
further action.
Jim Hines
jimhines@interserv.com
feedforward in engineering. He concludes by saying:
"IMO, feedbacks in SD are actually causal links that include both
feedback and feedforward mode of "controlling" the system."
Benny is exactly right. The use of the term "feedback" in system
dynamics covers any process in which the state of a system influences
action, which affects the system, and thereby affects further action.
This obviously covers the engineering use of the term feedback. But, it
also covers "feedforward". In feedforward, information on the state of
the system is used to create an estimate or projection. The estimate or
projection is also part of the "state of the system" and guides action
which in term affects the system, estimates about it, and therefore
further action.
Jim Hines
jimhines@interserv.com
-
saeed@ait.ac.th (Prof Khalid Sae
- Junior Member
- Posts: 17
- Joined: Fri Mar 29, 2002 3:39 am
Definition of Feedback
Re Definition of Feedback, may I refer the readers again to George
Richardsons book:
Feedback Thought in Social Science and Systems Theory.
Philadelphia: University of Pennsylvania Press (1991).
Khalid
Khalid Saeed
Infrastructure Planning & Management
School of Civil Engineering
ASIAN INSTITUTE OF TECHNOLOGY
G.P.O. Box 2754, Bangkok, THAILAND
phones: (66-2)524-5681, (66-2)524-5785; fax: (66-2)524-5776
email saeed@ait.ac.th
Richardsons book:
Feedback Thought in Social Science and Systems Theory.
Philadelphia: University of Pennsylvania Press (1991).
Khalid
Khalid Saeed
Infrastructure Planning & Management
School of Civil Engineering
ASIAN INSTITUTE OF TECHNOLOGY
G.P.O. Box 2754, Bangkok, THAILAND
phones: (66-2)524-5681, (66-2)524-5785; fax: (66-2)524-5776
email saeed@ait.ac.th
-
"William Steinhurst"
- Member
- Posts: 21
- Joined: Fri Mar 29, 2002 3:39 am
Definition of Feedback
ON Wed, 15 May 1996 11:58:04 -0400 (EDT)
George Richardson <gr383@cnsvax.albany.edu> said:
> One can find definitions of feedback loops and the feedback perspective
> in various system dynamics writings, notably the early writings of
> Forrester. Youll also find much discussion of this in Feedback
> Thought. I became very interested during that research to see the number
> of different ways the idea(s) was discussed. There are serious
> differences in emphasis in the various definitions, which have serious
> implications for practice, which apparently depend on the literature and
> contacts one has access to.
>
Its amazing how much light a little history can shed on a semantic
debate!
I am often startled and disappointed that so many
disciplines fail to teach the historical evolution of their theories
and practices. In high school and college in the dark ages (late
50s, early 60s), I learned physics and astronomy through the history
route: Copernicus suggested this, Galileo that, then Newton
synthesized it this way, etc. *Everything* made sense because
everything was in context. Then in grad school, the engineering
coursework did not make those connections, and every step was a
struggle to make my own connections. (Not a fruitless activity, but
more painful and less efficient than it could have been.)
William Steinhurst
wsteinhu@psd.state.vt.us
George Richardson <gr383@cnsvax.albany.edu> said:
> One can find definitions of feedback loops and the feedback perspective
> in various system dynamics writings, notably the early writings of
> Forrester. Youll also find much discussion of this in Feedback
> Thought. I became very interested during that research to see the number
> of different ways the idea(s) was discussed. There are serious
> differences in emphasis in the various definitions, which have serious
> implications for practice, which apparently depend on the literature and
> contacts one has access to.
>
Its amazing how much light a little history can shed on a semantic
debate!
I am often startled and disappointed that so many
disciplines fail to teach the historical evolution of their theories
and practices. In high school and college in the dark ages (late
50s, early 60s), I learned physics and astronomy through the history
route: Copernicus suggested this, Galileo that, then Newton
synthesized it this way, etc. *Everything* made sense because
everything was in context. Then in grad school, the engineering
coursework did not make those connections, and every step was a
struggle to make my own connections. (Not a fruitless activity, but
more painful and less efficient than it could have been.)
William Steinhurst
wsteinhu@psd.state.vt.us
-
bbens@MIT.EDU
- Junior Member
- Posts: 10
- Joined: Fri Mar 29, 2002 3:39 am
Definition of Feedback
Here is a control engineers perspective:
o Feedback is used when one is comparing "outputs" of a system to their
desired values. The comparison is used as a basis to generate a
corrective action that will drive the system to its target state. The
terminology is due to the use of "information" available at any given
time in regulating or controlling states of the system in a later time.
For this very reason also, systems with huge time delay inherent in it
(e.g., thermal systems with huge diffusion time such as in manufacturing)
are more difficult to control than those with short time delay such as
electric motors.
o Feedforward is used when one is compensating for a known (estimated)
side effect of a certain physical system. For example, this is done
in, say, trajectory control in which one can estimate the effect of
friction that needs to be compensated for. One estimates this effect
and apply a corrective action to the system to mitigate the usually
negative effects of friction in this particular example. Of course one
can also wait until the effect shows up as outputs and apply corrective
action in feedback mode--but any significant delay in the system always
spells trouble so it is better to compensate early than late.
IMO, feedbacks in SD are actually causal links that include both feedback
and feedforward mode of "controlling" the system. So, it is more of a
concept/construct rather than a thinking perspective. The name was chosen
because of the way people use information to control a system.
Regards,
-Benny Budiman-
bbens@mit.edu
o Feedback is used when one is comparing "outputs" of a system to their
desired values. The comparison is used as a basis to generate a
corrective action that will drive the system to its target state. The
terminology is due to the use of "information" available at any given
time in regulating or controlling states of the system in a later time.
For this very reason also, systems with huge time delay inherent in it
(e.g., thermal systems with huge diffusion time such as in manufacturing)
are more difficult to control than those with short time delay such as
electric motors.
o Feedforward is used when one is compensating for a known (estimated)
side effect of a certain physical system. For example, this is done
in, say, trajectory control in which one can estimate the effect of
friction that needs to be compensated for. One estimates this effect
and apply a corrective action to the system to mitigate the usually
negative effects of friction in this particular example. Of course one
can also wait until the effect shows up as outputs and apply corrective
action in feedback mode--but any significant delay in the system always
spells trouble so it is better to compensate early than late.
IMO, feedbacks in SD are actually causal links that include both feedback
and feedforward mode of "controlling" the system. So, it is more of a
concept/construct rather than a thinking perspective. The name was chosen
because of the way people use information to control a system.
Regards,
-Benny Budiman-
bbens@mit.edu