[Martin Taylor 990507 17:35]
The word "coercion" has been used to describe two entirely different kinds
of situation. There is precedent for each usage, but if the word is to
be part of the technical vocabulary of PCT, I think it should be restricted
to only one of the usages.
Usages:
(1) Dominated conflict: Two control systems attempt to influence the same
environmental variable, and are therefore in conflict. One is stronger
and can "overwhelm" the other. In standard electrical engineering
terminology, the environmental variable is "coerced" to the value
desired by the stronger. (Note that the other control system is _not_
coerced. Only the value of the environmental variable is.)
(2) Moral 'suasion: One control system (in the hierarchy of the coercer,
ER) is able to coerce (in the engineering sense) the value of an
environmental variable V the perception of which another control system
(in the hierarchy of the coercee, EE) is controlling. ER wants EE to act
in a certain way W that to EE is ordinarily unrelated to V. ER is
controlling a perception of W, and is using the ability to coerce V
to alter EE's action W. This cannot work unless EE understands (program
level?) that asserting the correct value of W will allow control of V.
If EE understands the linkage, or has reorganized to use action W as
part of the control of the perception of V, then ER's ability to coerce
V allows ER to control the perception of W (to get EE to act the way ER
wants).
Although in the engineering sense, ER is coercing only the value of V,
nevertheless, because ER is controlling W using EE's control hierarchy,
it makes sense to say in this case that ER is coercing EE to produce
action W.
If "coerce" is to be used as a technical term in the sense "person A is
coercing person B," I think it must be restricted to the "Moral 'suasion"
situation. Technically, the words "dominated conflict" seem adequately
to describe the condition in which one control system can bring more power
to bear on a conflicted variable than can the other.
What follows is a more extended technical discussion of the engineering
situation. Skip it if you want.
Martin
ยทยทยท
------------------------------
Coercion and Control
The word "coerce" is used in engineering, or at least in electrical
engineering, in a sense different from the dictionary definition of
getting "compliance" by force of threat. This engineering sense is
technical, and it may be where Bill and Rick are getting their usage from.
In the following, I will talk only about electrical circuits, not living
control systems, unless I specify otherwise.
To "coerce" a value of a signal is to ensure that no matter what other
influences on the signal there may be, the signal will take on the coerced
value. This can be accomplished by disconnecting any other possible sources
of influence.
If other sources of influence on the coerced signal cannot be disconnected,
the situation is one of "control"--the maintenance of something at a
desired value despite the influence of disturbing variables. How well can
the value be held at its reference value? The answer is given by the
control equation: error = disturbance/(1+Gain). The greater the Gain, the
more precise the control (assuming infinitely small loop delay).
Great Gain implies that the output value can get very large. This implies
the availability of "overwhelming force." If "overwhelming force" is not
available, then when the disturbance becomes sufficiently large, the error
will increase without the control system being able to correct it.
If the disturbance is due to some kind of random external event, then the
control system will usually be able to coerce the signal to something like
its reference value. But what happens when the disturbance source is
actually the output of a control system trying to coerce the same signal to
a different value? This is a classic conflict, and leads to escalation,
each control system increasing its output until one reaches the limit of
its resources. For the other system, this means that the disturbance is no
longer increasing, and the error can be brought arbitrarily close to zero
by exerting more force. The signal value can be coerced even in the
presence of weaker control systems attempting to coerce it to a different
value.
In this sense of "coerce," a rock can be coerced into a specified position
by a control system, a child's hand can be coerced into tracing a specified
letter-form whether or not the child tries to coerce the hand into
following a different form, but a child cannot be coerced into making the
hand trace a specified letter form. To coerce the _actions_ of the child is
a quite different phenomenon.