[From Bill Powers (980620.0716 MDT)]
Bruce Nevin (980619.1624 EDT)--
Me:
I don't understand why you think that the victim is NOT in the model of
"coercion," whatever that is.
You:
Because so far we have seen only a model of a coercer.
I don't understand. Isn't the coercee also a control system? Do we need to
draw the models out completely every time, as if there were some
alternative? I think it's sufficient to assume that the coercer and the
coercee are both control systems, and to say that the coercer is controlling
the coercee's (nominal) output quantity (which is a controlled variable at a
lower level). Nothing new to HPCT is being introduced. The same diagram
would apply to interaction, conflict, and coercion -- the only differences
are the values of certain parameters and signals. No new relationships are
involved.
Rick has shown
output o' with no control loop, and (separately) a plot of a reference
signal r2, but these do not constitute a model of a victim.
Just assume the usual model of the victim. Since the coercer is assumed to
have greater strength and perhaps loop gain than the coercee, it makes no
difference (to the state of the controlled variable) what the coercee's
internal organization is. Nothing the coercee does is going to have any
material effect on the shared controlled variable (for the coercee, "former"
controlled variable) -- only the coercer controls it. The victim might, of
course, be experiencing very large error signals and might be producing very
large efforts, and the only way to prevent this is for the victim to set the
reference signal to whatever value the controlled variable has. But the
coercer need know nothing of that. The coercer has an iron grip on the wrist
of the coercee, and the hand is going to be placed wherever the coercer
wants it to be. No observer could tell where the coercee wants the hand,
because only the coercer's reference level determines that position. The
coercee's muscular efforts, because they're ineffective, are not easily
observable.
This is all just straight deduction from the HPCT model. It is completely
testable, and I can't think of any reason for all the argument except
perhaps that a lot of people on this net are suffering childhood amnesia, or
grew up in Paradise.
The victim has always been in the model --
_our_ model. Perhaps you missed the reference to what model is being talked
about when we say that the coercer does not take the coercee's goals into
account. That means that the coercer does not need a model of the coercee
(though he may have one, probably erroneous);
I'm not concerned with putting the victim inside your model of the coercer.
Coercion, like conflict, is a relationship between control systems. It is
not an attribute of the coercer.
No, it is an attribute of an observer.
I agree that it entails a relationship, a particular and I hope well-define
one, between control systems. One does not "coerce" a pencil. But neither
does one ask the pencil what it wants to write. In order not to coerce
another person when you are the stronger of the two, you must ask what the
other's desires are, concerning variables you want to control in a way that
might affect the other person as an overwhelming physical force. The same
relationship between control systems can be coercive or not coercive,
depending only on the potential coercer's willingness to take the potential
coercee's goals into account, modifying his own actions or goals or both so
as not to force the other party into unwilling acquiescence.
The teacher, who is unquestionably the more powerful, says to the student,
"Look, if we go on like this we're just going to get into a big fight, and I
will win it because I have the whole school system behind me. What do you
say we break it off here? You can go to the RTC and figure out how to handle
your problem, and I can go on with my job, and we'll still be friends when
you come back. Would you agree to that?"
This is pretty non-coercive, in that there's at least some sort of
persuasion being tried before the hammer falls. It could be totally
non-coercive, if the teacher chose to avoid even the opportunity to apply
force. My point is that with the two control systems having the same set of
properties (one strong, one weak) at the level of the conflict, we could end
up with or without coercion taking place. It all depends on what the
stronger person decides to do about the conflict: win it, or resolve it. The
coercer just wants to win it.
we, of course, who understand
the consequences of ignoring the coercee's desires, DO need to propose a
model including both partipants in the interaction, although we do not need
to know the coercee's goals to explain his behavior (which is totally
determined by the coercer).
That depends upon how far up the hierarchy the coercion extends. And *that*
begins to get at something interesting.
While I don't think that coercion at one level necessarily results in
coercion at any higher level, even if it does the coercer still doesn't need
to know anything about what's happening inside the coercee. The coercer only
wants to control the coercee's behavior, and has sufficient strength to do
so. And, of course, does so, which is what creates the coercion. If the
potential coercer refrains from using his superior strength, he is not coercing.
I know we don't
know how to determine what the victim would be controlling if she could (we
can't determine the value of r').
Of course we can -- just remove the coercer and see what the coercee does.
Let the coercive teacher leave the classroom and listen to the noise level,
watch the spitballs fly. Of course what you observe is subject to
interpretive problems, but that's nothing new, or insurmountable.
I respectfully disagree. Kids released from coercion behave differently
from kids free from coercion.
My point exactly. If no coercion was going on, the kids will show the same
social behavior with or without the teacher present. If coercion was going
on, the kids are likely to get pretty raucous when the teacher leaves,
because their goals for their behavior are different from the teacher's
goals but have been unsatisfied -- a big error exists. Coercive teachers
typically don't dare to leave their classes unsupervised.
The model of the victim is far more
interesting than that of the coercer, and one reason is changes in what the
victim controls, by going up to a level that is not coerced (other means
are available), by trying random alternatives, by controlling being
spaceman Spiff on an alien planet (ah, Calvin! Apologies to those who don't
see U.S. comic strips).... (You're also talking here about social
interactions among the kids, which are not modelled. Put a single student
with a single coercive teacher. When the teacher walks out does that single
student start yelling and throwing spitballs? That suggests that social
interactions can be involved in victims changing what they are
controlling.) Remove the coercer and you are not likely to see control
resume as though it had never been thwarted. But that is probably what the
model now predicts--though we can't tell because the present model only
says that whatever the victim is controlling is irrelevant.
TO THE COERCER, TO THE COERCER, TO THE COERCER. You're getting me in the
mood for some nice satisfying coercion. What the victim is controlling (or
would like to control) is very relevant from the victim's standpoint, and
from the standpoint of a modeler who wants to know what error signals are
present because of the coercion, and for a lot of other reasons such as
those you mention above. But what the victim is controlling is irrelevant TO
THE COERCER, which is why the coercer just goes ahead and controls the
victim's behavior without regard for what the victim wants. If the coercer
were aware of what the victim wants, and cared enough (positively) about
what the victim wants to modify his demands or actions, there would be no
coercion even though the coercer is stronger and more capable. The
conditions in the relationship that permit coercion to take place do not
determine that coercion _will_ take place. Whether it does take place
depends entirely on higher-level processes in the coercer.
You haven't mentioned the most common way in which coercees deal with
coercers. They say "Yes, sir, no sir, right away sir, I'm sorry sir, you're
perfectly right, sir," giving the coercer the behavior that the coercer
wants to see, while completely dissociating their higher-level goals from
that behavior. They give meaningless or even malicious compliance, while
managing to accomplish their own higher-level goals by other means or at
other times. "I'm sorry sir, you said to turn the heat lower so I turned it
off to make sure it was low enough. Shall I turn it on again, sir?"
I don't see any unsurmountable
methodological problems here.
Not insurmountable, probably, but not simple either. See above.
That's just a cantakerous remark. You haven't mentioned anything very
complicated, and anyway, does "not insurmountable" suggest "simple" to you?
To me, it suggests "difficult, but achievable."
There's a "logical fact" that I think is being overlooked in this
discussion.
The logic is impeccable. The premisses are slippery.
It would be helpful if you would say what premises you mean, and how they
are slippery.
If the coercer is controlling the state of environment variable C then C
(or rather the state A of C) is an aspect of the coercer's behavior. The
victim is unable to control, so it cannot be the victim's behavior.
Oh, I see how they are slippery: you're greasing them.
Suppose C is the position of the victim's hand, and the coercer is
controlling it by holding the wrist. The coercer is saying "Stop scratching
your face," so we have at least a reasonable indication that the victim has
been using the position of the hand for something. So if we see the hand
doing something, it is now the coercer, not the coercee, that is doing it.
Isn't that more or less what I've been saying?
Therefore the coercer is not controlling the victim's behavior.
Ooooh, slick. Of course if the coercer is now in control of the hand, the
victim is no longer controlling the hand's position, so the coercer is not
controlling the victim's behavior! You have just proven that coercion never
happens! Wow!
What you're overlooking is that the victim (unless he has given up) is
probably still _trying_ to behave, but the coercer is so much stronger that
the trying is ineffective. The mother dragging the child along by one arm is
meeting with resistance from the child, but the resistance is not strong
enough to keep the child's body from moving as the mother wants it to move.
It's clear from the angle of the child's body and knowledge of physics that
the child is still _trying_ to control the same behavior that the mother is
controlling. But the mother is simply overcoming the disturbance.
The coercer
is preventing the victim from controlling, so the coercer is actually
preventing the state of C from being an aspect of the victim's behavior. As
you have said for many years, it is not possible for one control system to
control the behavior of another. To do so, it would have to reach inside
and set one or more reference values in the victim.
Did you ever think of becoming a lawyer? Lawyers quite often distort
opposing arguments by misquoting them or leaving something out. In this case
you have quoted only part of what I said: It is not possible for one control
system to control the behavior of another without the application of
carefully selected disturbances, or of course overwhelming physical force --
i.e., coercion. And a few other methods, such a misrepresenting facts and
lying. Since coercion's what we're talking about, I should think you might
have quoted the whole statement (it has appeared in several forms and I'm
not quoting it exactly).
....
A PCT claim is that it is not possible for one control system to control
the behavior of another. Have you changed your views, and decided that not
the case after all?
It is not the case, and I have not changed my views (on that matter). Time
to reread B:CP.
Best,
Bill P.
a great fathers day weekend with the