[Hans Blom 930619]
(Rick Marken (930618.0900))
You are of course right: the individual control
laws together 'control' flockness, but in an emergent way
I think we are in violent AGREEMENT.
That only seems so, for the moment.
(Tom Bourbon (930618.0958))
My guess that you speak from the position of a designer, engineer, or
user of a system was confirmed when I came to your comments
Not only. I have been a control engineer for so long (25 years) that I
could not miss starting to question its basic assumptions. What I re-
discover every time is that modelling is the basis of control. Without
an approximately correct knowledge of the system to be controlled it
is impossible to achieve any control quality. In Bill Powers' approach
the 'knowledge' about the system to be controlled is very limited;
most of it is stored in the choice of the parameter that he calls the
'slowing factor'. Choosing an inappropriate slowing factor either
results in oscillations or in a system that is too slow to be useful.
Bill might protest and say that he needs no model in order to be able
to control, but his having to search for an appropriate slowing factor
indicates otherwise. Make it a factor of 10 or 100 larger or smaller
and chaos or stagnation results. The knowledge incorporated into
Bill's models is of unacceptably limited, so other control approaches
go further. Using adaptive control, for instance, you would also pro-
vide the system with a capability that allows it to discover the best
slowing factor. Such adaptive controllers will always ALMOST oscillate
(but not quite). The tremors in our limbs may indicate that something
like this goes on in humans as well.
But my basic discovery was that each model is an APPROXIMATION of
reality, geared toward some goal. A model incorporates the features
that you focus your attention on in that one particular application,
because you want to study only those, and it does not contain the
features that you deem unimportant in the study. Each model is there-
fore a human invention, a useful approximation, a tool to study an
otherwise too complex reality. It is simply not true that humans ARE
control systems; it is true, however, that some human qualities can be
MODELLED as control systems. The map is not the area.
could you describe the mechanism the "keeps
flockness within specified (but maybe varying) limits or almost so?"
I thought I did: the control laws of all individuals together create
the flock implicitly. There is no explicit control for flockness.
*You* use the heating system to control
your perception of a nice room temperature; the heating system does
not. I *think* I understand your position pretty well, or am I
wrong?
I think you do. *I* want to control for a nice room temperature. The
heating system can only control the temperature at its sensor. So what
do *I* do to control the controller? I put the sensor at a location
where what the heating system can do approximates as closely as
possible what *I* want done. There are two levels: the heating system
controls, and I control using the heating system's contrller. Could
that be a hierarchical control system?
From the perspective of the heating system, which knows nothing of the
room temperature except the temperature at its sensor, the temperature
at locations other than at its sensor is uncontrolled. From MY per-
spective, I control the temperature at the position where I sense it,
using the heating system. The perspective matters!
(Bill Powers (930618.1000 MDT))
Please be explicit in what discriminates a 'natural tendency'
from control (should you do so, I am tempted to say, I will
give you a counter->example :-).
Explicitly, it's power gain. A control system has an output
function that produces vastly more power output than its inputs
provide (drawing, ultimately, on external energy supplies).
The counterexample (see below for another one): You will agree that
your hifi amplifier is a control system. It amplifies the small power
at its input terminals into the larger power that the speakers need.
Here is the power gain that you stress as the important thing. The
feedback is of course needed because the amp's components are not
quite ideal, yet we want hifi quality. Now imagine an almost identical
situation, where it is the electrical power engineer's task to track
the current and voltage of those energy-rich 380.000 Volt overhead
power lines on his oscilloscope. What he needs is a tremendous power
LOSS. The components of his measurement system are non-ideal as well,
and therefore he too needs feedback. Where is the power gain now?
You can, no doubt, think of many more examples in which energy or
power must be scaled DOWN rather than up. This example stresses what
*I* consider the intent of control: to improve otherwise inadequate
results, with or without power gain.
You don't arbitrarily perturb a real control
system without spitting on your hands first and bracing your
feet. When I say that control systems resist disturbance, I mean
that they resist it energetically and with as much effort as
needed (up to their limits of output).
This and the next section of your reply make your position much
clearer for me. What you talk about is servomechanisms, and you also
seem to exclude servomechanisms that have no power gain such as
micromanipulators that atomic physicists or surgeons might use to
operate on a scale much finer than that of the human hand.
Now servomechanism theory is, indeed, part of control theory, but
control theory is much wider. Control theory also studies adaptive
control, time-optimal control, and robust control, for instance. In my
previous contributions, I have referred to these areas of control
theory as well, not knowing that you wanted to limit things to servo-
mechanism theory only. That may be the basis of our mutual misunder-
standing about what constitutes control.
Isn't it remarkable how robust flockness is in the face of
disturbance?
I don't think you know what "robust" is.
I was talking in terms of robust control systems. Robust control is a
new branch of control theory, that you may not be familiar with. It
has greatly gained in importance during the last ten years or so. The
goal of robust control is to design systems that continue to control
well in an environment which may change greatly or with sensors and/or
actuators whose characteristics may vary randomly or systematically,
as in aging of catalysts or decay of magnetic properties and so on.
For example, it is one of the limitations of servomechanisms that they
break down when their sensors become very noisy. There are solutions
for that problem using other approaches.
If it were not for the fact that I remain upright while I walk,
the air along my path six feet above the ground would not be
displaced. The movement of the air follows lawfully from the
effects of my remaining upright while I walk, but it is not a
controlled variable. It is an uncontrolled, although lawful,
outcome of my control processes. A side-effect, as far as I am
concerned.
Again a matter of perspective. If it is my goal to walk, you are
right. If it is my goal to displace air (through walking), you are
not.
(Rick Marken (930618.1500))
So Hans' and my idea that "flockness" is under control (a "collective
controlled variable" by my terminology -- ie. a variable that is
controlled by the collective but not by any individual) is incorrect
unless there is a control system present that is perceiving and
acting to influence this variable.
We are NOT in violent AGREEMENT now. See how difficult these things
are? One moment you see control, the next moment it's gone. That is
the best demonstration you, an expert on control, could give me that
the concept of control is slippery.
Greetings,
Hans