Strong influences

[From Bruce Abbott (981019.1010 EST)]

Bill Powers (981019.0653 MDT) --

Bruce Abbott (981018)

writing to Rick:

So you are saying that a PCTer doesn't believe that the environment has any
strong influence on behavior. That is what "control" means in one sense of
the word -- exerts a strong influence on. Thus a PCTer doesn't believe
that, for example, disturbances have any strong influence on actions. Wow,
I learn something new here every day.

I'm sorry, but I don't think anyone would be satisfied with taking your
definition of control literally. Suppose I hand you a button and tell you
that pressing it will instantly cause your car to move 25 miles. That, I
think, would count as a "strong influence" on the car. This could be very
useful sometimes, when your destination is close to 25 miles away -- think
of the savings on gasoline.

But I don't think you would use the button. If you have any imagination,
you won't even use it once, but you would certainly never use it a second
time. The problem is that while the button gives you a strong influence on
the car, it doesn't let you pick the effect this influence will have. If
you're foolhardy enough to use the button while you're in the car, you have
an excellent chance of finding yourself 15 miles in the air or 15 miles
underground. You can have a strong -- very strong -- influence on the car,
but you can't control the outcome. And that's what control is all about:
preselecting the outcome and bringing it about. Apparently the makers of
dictionaries have failed to think through their definition sufficiently to
see what they are taking for granted.

If you had a control system (PCT definition) that worked that way you
wouldn't use _it_, either. It follows (according to _your_ logic) that the
PCT definition of control is absurd.

In truth, your example has more to say about the degree of control possible
in some cases (very limited) than about the correctness of the definition.
The definition says that X has the major influence on Y, so that Y is
essentially determined by X. This can be because (a) disturbances have been
minimized and the system calibrated so that a given setting of X produces a
given value of Y, within acceptible limits, or (b) because the system has
been stabilized against disturbances via an appropriate negative feedback
(PCT definition of "control").

Regards,

Bruce

[From Bill Powers (981019.1535 MDT)]

Bruce Abbott (981019.1010 EST)--

I'm sorry, but I don't think anyone would be satisfied with taking your
definition of control literally. ...

If you had a control system (PCT definition) that worked that way you
wouldn't use _it_, either.

Of course not -- it wouldn't be a control system. A control system not only
allows you to have an influence on something, it allows you to select in
advance the kind and direction of effect you intend to have, and then to
know what effect is actually occurring so you can make corrections.

Your definition and the dictionary's neglects to specify that the person
having the influence is able to verify that the desired influence has been
had. Of course when you include that, you're talking about a negative
feedback control system, not an open-loop system.

It follows (according to _your_ logic) that the
PCT definition of control is absurd.

No, it's the _other_ definition of control that's absurd: it's absurd to
think that anyone would be satisfied to have a strong and (supposedly)
reliable influence on something if the person were never allowed to observe
the effects of the actions. Even when it's not mentioned, observing the
effects is a tacit part of any concept of control, even so-called open-loop
control. If you can tell me what the effect of your action is, it's not
open-loop.

Best,

Bill P.

[From Bill Powers (981019.1605 MDT)]

I guess you'd just better publish your results, Bruce. I can't take any
more of this.

Best,

Bill P.