[From Bill Powers (2003.05.17.0806 MDT)]
Marc Abrams (2003.05.16.1728)--
>No it's not a requirement. But you still didn't answer my question. Why did
>you decide to speak for Rick? Did you think he was incapable of answering?
>I'm just curious.
I didn't decide to speak for Rick. I decided to comment.on your post.
>I said: The TEST is a set of actions ( words ) that occur between two
>people.
The only appropriate response to that would be "so what?" The more general
you make your description, the less it means. The test is certainly not
limited to words. You can do the test with a dog, or as I said, with a
thermostat. You can guess that a thermostat is controlling the temperature
in a room. So you open a window and see what happens to that temperature.
If it's cold outside, you'd predict the temperature to drop if there is no
control system acting. You can also try adding heat, as by building a fire
in the fireplace. The temperature should go up. But with the thermostat
there, the temperature will change a lot less than predicted, which tells
you that the temperature at the point of measurement may be under control.
By moving the thermometer around in the room, you can locate the place
where the temperature change is smallest, at which point the thermometer
will be next the sensing unit on the wall. Also, investigation will show
that if something acts to cool the sensor, the furnace will turn on, and if
it's already on and something acts to heat the sensor, the furnace will
turn off. By this time you will be pretty sure you have located a control
system, even though you haven't traced the wiring or found out how changes
in temperature are sensed or how the furnace is turned on and off.
You're focusing on a trivial aspect of the test that is sometimes seen --
communicating between people with words -- and missing what makes it work,
which includes applying disturbances to a variable you think is under
control, and seeing whether any ensuing action opposes the effects of the
disturbance on that variable. From your description of the test, nobody
could learn to do it.
You interpreted it to mean; The Test is [only and always ] a set of actions
(words) that occur between two people.
Yes, because that's what you said it was. You didn't say "except when it's
not." You might have added "sometimes" or "in some applications" or "except
when no words are used" or "except when one of the parties is not a
person." When scientists speak they have this annoying habit of adding
little phrases like that when it's important to say that they don't intend
to be including all cases. Of course that takes the drama out of sweeping
overstatements, but drama isn't the point when you're trying to say what
you mean to say.
Do you see what you did? Do you see why, what you did was not necessarily
what I intended? probably not.
Marc, I am not a mind reader and I don't care to guess my way through
poorly constructed communications. If you don't want your words to be taken
literally, say so. Of course then the point of what you're trying to get
across will get a lot weaker, but you can't have it both ways. Either try
to say what you mean, or admit that you're just blathering.
Just because you _can_ does not mean that it does, or is always done a
certain way.
The principle of the test remains exactly the same however it is done.
When I say the word _error_during the coin game, that is an
action.
Yes, but that word doesn't have to be used. It just saves time when the
experimenter has executed what he thinks of as a disturbance, but has
failed to change the variable that the subject is actually controlling.
When that happens, the subject makes no correction, but the experimenter
may take a while to realize that there isn't going to be one. Saying "No
error" cuts the wait short but it doesn't contribute anything else to the Test.
It is an action in that I am communicating an idea, some
information, that another individual is using to validate thier own
perceptions about what I am doing. It is an action just as surely as my
giving you the finger is an action. They communicate an idea. How the other
party percieves it is anyones guess, but it is _ONLY_ through the continuous
validation of action and thought that we can even hope to come close to
understanding what someone else might be controlling for.
You can wiggle and squirm all you please, but making your meaning more and
more general (so "action" means "anything that happens during behavior")
only makes your comment less meaningful. This isn't a court where lawyers
make a living by showing that their client meant one thing when he said
something else. The above paragraph doesn't come close to saying anything
about the Test for the Controlled Variable. It doesn't indicate in any way
that you understand how that test works (if you do understand it, you
haven't shown that you do). What you say could apply to any kind of
interaction between people -- it's the sort of comment people make when
they don't know what they're talking about but want to give the impression
that they do.
The key in your
statement is "COMMUNICATION". as far as I know mental telepathy is not an
option. _ANY_ communication, by any means, requires some form of
behavior/action. Verbal, being only one way. There are many ways to
communicate. In fact _That_ is the point. _COMMUNICATION_ requires
action/behavior. You cannot communicate without behaving/acting, in some
fashion. So behaving is central to communicating. It is only through
behaving/actions that we can interpret purpose or intent about another
individual. I will say it again and for the very last time. When you do the
Test or the Coin game, or the Tracking task. You are validating someones
"purpose" through communications ( actions/behaviors ) of some kind.
If that's the case for the defense, your client is in big trouble. Now
everything that happens between two people is "communication," and to heck
with what you call it when the other party isn't a person. All this shows
is how badly you want to be right.
> Most PCT experiments are a formal version of the Test. This is why, for
> example, we always check the data to see, when we can, what reference
level
> for the controlled variable the person actually set, as opposed to what
the
> instructions described.
The person will _assume_ the reference condition you have set, and try to
maintain it.
If the person understood it as you meant it. That isn't always true.
In Rick's "mind reading" program, the computer can tell what
position you are trying to maintain by the behavior ( mouse movements ) you
do. The illusion is that you think it's reading your mind and knowing your
purpose, when in fact it "knows" by your behavior.
Marc, that doesn't convince me that you have the least idea of how Rick's
mind-reading program works. OF COURSE it knows by your behavior, but that's
true of knowing anything about anything experimentally. It's a
wise-sounding sentence that says nothing. In fact, the program knows by
examining the relationship between disturbances being applied invisibly to
each object, and the person's actions. To be more exact, it continuously
calculates the correlation between the mouse positions in X and Y and the
positions of the objects in X and Y, and selects the object whose behavior
shows the lowest correlation with the person's actions. You could sit there
all day looking at the person's actions and you'd never figure out which
object is under control.
>We define the controlled variable in an unambiguous
> way, and measure to see exactly how accurately the action opposes the
> effects of the disturbance on this variable. The control error is part of
> the data we take. We don't just assume that we're dealing with a control
> system. We prove it every time, as nearly as we can. For a lot of people,
> that approach is just too fussy and slow. But I think it's how you keep a
> theory in contact with reality.
Fine, how do you move beyond the tracking task? Do you have any desire to
move beyond it, or are you simply interested in finding control in everyone
you test and be satisfied with that?
The "tracking task" is simply a control task in which a person controls a
visual relationship by moving a control stick. Demo 1 shows a number of
control tasks that do not entail tracking. It already goes "beyond the
tracking task." The point of doing these experiments is to establish what
kinds of variables people control and to measure the characteristics of
their control systems. As we learn more about the simpler systems, we
prepare for looking at higher levels of control. We need to see the lower
systems first so we can sort out what parts of higher-level control belong
to the higher level, and which are characteristics of the lower levels.
Simple example: if we see a lag of half a second in responding to a
disturbance of a higher-level system, how much lag is there in the
higher-level system? We can't answer that until we know the lags in the
lower systems that are used by the higher system.
You can't simply jump into a study of higher-level control systems without
knowing anything about the lower levels. Well, I shouldn't say "can't".
Actually, lots of people love to do just that. They can't wait. They want
to play a Litzt piano concerto before they have learned their scales. Lots
of luck.
>And when the experiment "fails" what was the person controlling for? We know
>they were controlling for something. Or are you saying a "failed" experiment
>shows no control.
It means we would interpret the results as though the person was
controlling what we said was to be controlled, when in fact he was
controlling something else. The result, of course, could be that he would
appear to be controlling the variable we intended, but very poorly. In
fact, he could be controlling a different variable very well.
Me:
> >The Test is a formal structured conversation, not unlike any other formal
> >structured scientific experiment, which is nothing more then a set of
> >behaviors/actions that help validate theories ( words and thoughts ) we
have
> >about things. What in this statement don't you undersatnd? What do you
> >disagree with?
Bill:
> I disagree with the whole thing.
You don't say what you disagree with me on.
I disagree with the following statements:
1. The Test is a formal structured conversation.
The test can be a conversation, but normally it is not.
2 not unlike any other formal structured scientific experiment, which is
nothing more then a set of behaviors/actions
A structured scientific experiment is a lot more than a set of
behavior/actions. Not "nothing more."
3. ... behaviors/actions that help validate theories ( words and thoughts ) we
have about things.
validity of. You are saying that the Test, and a scientific experiment,
help to confirm the validity of theories we have about things. That is a
layman's false impression of what scientific experiments (and the Test) are
or can do. What the Test does is check to see whether the proposed
controlled variable behaves when disturbed as we would expect if it were
under control by the organism being studied. If it does, we have not ruled
out the possibility that some other variable related to the first one is
the actual controlled variable, so we can say only that the results are
consistent with the hypothesis that the variable as defined is under
control. It is still possible that some other variable is being controlled
instead.
However, if any part of the Test is failed, we can say without doubt that
the variable we proposed to be under control is NOT under control. So the
Test very quickly weeds out wrong guesses. If we want positive
confirmation, we have to use other means -- making predictions about
behavior under varying circumstances, dissecting the system to trace out
its internal functions, looking for alternative controlled variables, and
so forth.
> I will amend my statement to
>read;" The Test is a formal structured [ way of communicating ], not unlike
>any other formal structured scientific experiment, which is a set of
>actions/behaviors that help validate theories ( words and thoughts ) we have
>about things. Do you still disagree with this?
Yes, for the same reasons listed above, and in addition because now you are
characterizing the Test as a "way of communicating." To make that true you
have to define communication as any interaction whatsoever between any two
systems (even when they're not speaking or writing or even capable of
speech or writing), to which I will not agree. Why destroy the meaning of a
perfectly useful word?
> You said that Rick's and my attempts to communicate control theory to
> others (one of our main occupations for the past 20 or 30 years, or more)
> have been a total failure.
You have again added words and meaning to what I have said. The word "total"
is yours, not mine. The theory has not been unsuccessful, your ability to
communicate it to others has been. You can continue to believe what you
like.
What a masterful way you have of denying and un-denying at the same time.
You deny saying our efforts to communicate control therory to others were
total failures, and then say that while the theory has not been
unsuccessful, my ability to communicate it to others has been
[unsuccessful]. Since the sentence you dispute was about my ability to
communicate the theory, not about the success of the theory, all you have
done is change "total failure" to "failure." So does that take care of the
problem? Now you are merely saying that my efforts, and Rick's, to
communicate control theory to others over the past three or four decades
has been a failure. What am I supposed to say to that? How about "Oh, what
a relief, I thought you said it was a _total_ failure."
> Nonsense. I laid out some of the background courses that would make PCT
> easy to understand. Look over the list again; it included physics and
> control engineering.
Are these prerequisites? If so, how many on this list can say they have
them? When I speak to someone about PCT do I need to ask for credentials
first?
If they were prerequisites for students of behavior, those students would
find it easy to understand control theory (which is what I said). Some on
this list have studied such things, and I'm sure they would agree that this
knowledge helped them understand control theory. Others who have not taken
such courses can still understand control theory, but they have to pick up
the necessary knowledge a different way. Those who don't even try to
acquire the knowledge are kidding themselves.
You don't have to ask for credentials. If you're not being understood, this
will become evident fairly quickly.
>I can't imagine why you think physicists and trained
> control engineers can't grasp the theory.
I didn't. Your Buddy, Rick did, here;
What does that have to do with what I said? And anyway, your reading of
what Rick said was sloppy. He said that he was helped by learning computer
modeling, but he said that others with more training in sciences and
engineering "haven't adopted it." How did you get from that to a statement
that these others didn't _understand_ it?
And why are you so resentful of the fact that Rick might be my buddy? Is
this sibling rivalry?
[From Rick Marken (2003.05.15.1400)]
"...But I adopted it! True, I was lucky enough to learn computer modeling
in graduate
school. But many others have had even better training in modeling and
engineering
than I had (like my graduate advisor, who was trained as a physicist) and
they
haven't adopted it. ..."
See Bill. This is what I mean by showing the data. That is why I said I did
not say this. You imagined it. How about that? A perception from memory.
What you have done is mangle the data. Can't you tell the difference
between "adopt" and "understand?"
>They can, they have. I think you simply misread what I said.
No, I read Rick perfectly. You did the wrong "reading".
OK. You read Rick perfectly. That's what you mean by "perfect?" I'll keep
that in mind.
Here's a bit from earlier, to close with.
>This is funny. What don't I understand? So far you keep on saying that I
>don't understand this and I don't understand that. Are you saying these
>things to be helpful? If so, how do you think these statements are helpful?
>I find no value in them. Maybe because they are not intended to be helpful.
Finally you've got it. I am fed up with this manic you've been on, which
has had you blundering around trying to manage CSGnet and everyone on it
and spouting uninformed nonsense. You're a very bright person and energetic
enough for three people, but this does not make you into a superman or a
hypergenius. You have to learn to walk before you can fly, and all the
bluster and aggression in the world isn't going to change that. Just
yelling about how right you are doesn't make you right. You're trying to go
for the gold ring before you know where it is or what it looks like. I
understand why, but you don't need excuses. They won't help. When you're
working on an even keel you put out some great stuff. But it's time for you
to get over the idea that you're going to rescue PCT from its mismanagement
and turn it into the new Wonder Science. It ain't going to happen, and
you're not the one to make it happen. Any more than I am, or Rick is.
The problem is that PCT falls too strangely on the ears of conventional
scientists. On the other hand, there are thousands who understand it
reasonably well and support it -- but who hold back because it's not at all
clear that they could make a living from it.
We're having a little impasse here, and there's no predicting what will
finally clear it, or when. It could happen next year, it could be 50 years
from now.
Next Best,
Bill P.
···
To :"validate" means "verify" or "substantiate" -- to confirm the