Information, Thanks Dennis

CSGnet Archive CSG1994
1

[From Rick Marken (940228.1400)]

Kent McClelland (940228) --

I haven't had the time to pay close attention to Rick and Martin's long-
running debate about information, but my impression has been of two people
talking past each other.

Bill P. and Tom B. have been part of the debate as well; my impression
is of three people making a simple point about control (that the controlled
perceptual signal is the net result of many influences, including those of
the control system itself) and one person trying to make believe that
it doesn't matter.

Part of the problem may be that Rick and Martin haven't found a mutually
satisfactory definition of "information" that makes sense in PCT terms.

We tried the definition game a year ago. It didn't help.

even PCT faithful must admit that the environment has "an effect" on
perceptions.

A major part of the discussion was based on the idea that p = o + d.
Both o and d are environmental variables. So, yes, the "PCT faithful"
admit that the environment has an effect on perception; it's basic to
the PCT model.

As I see it, this definition of information allows both Martin and Rick to be
right simultaneously. As long as a perception is in control at a given
reference level, disturbances are uninformative, and Rick is right. When the
system shifts reference levels to maintain control, the "information" in the
disturbance has now been embodied in a new state of functioning of the
control system, and Martin has a point (though it may not be the point he was
trying to make).

Disturbances are uninformative, not because they are cancelled out by
outputs (which may be lower level reference variations when we are
dealing with systems at a higher level) but, rather, because their
influence on perception is always combined with the influence
of the system's own outputs. For controlled perceptions, it is always
true that p = o + d so the system has no way of knowing, at any instant,
the extent to which the value of p is a result of o or d.

Trust me, Kent, information theory is simply irrelevant (at best) to control
theory -- not because we want it to be but becuase it IS. I keep discussing
the relationship between IT and PCT because it is an example of what I
discussed this morning. IT advocates believe that the ideas of PCT were
anticipated by or are contained in IT. This is a problem because it
diverts smart, competent people, who could actually help with PCT science,
into worthless pursuits (such as trying to determine how the uncertainty of
the disturbance influences control). It's just kind of depressing -- maybe
not DSM-III depressing but sad nevertheless.

Dennis Delprato (940228) --

"Originality is favored by freedom from cultural restraint, from profes-
sional specialization and narrow and prescribed training. It is for
this reason that so many major scientific advances have been made by
amateurs--interlopers Darlington calls them.

What a wonderful, heartening post, Dennis. Thank you!

Best

Rick

2

<Martin Taylor 940228 19:15>

Rick Marken (940228.1400)

Kent McClelland (940228)

I haven't had the time to pay close attention to Rick and Martin's long-
running debate about information, but my impression has been of two people
talking past each other.

You aren't the only one with that impression. I also find it so. This
response of Rick's is another in the long line that proves he never reads
what I write. I have HOPED that I have read what he writes, but one can
never know for sure.

Bill P. and Tom B. have been part of the debate as well; my impression
is of three people making a simple point about control (that the controlled
perceptual signal is the net result of many influences, including those of
the control system itself) and one person trying to make believe that
it doesn't matter.

I don't know who that one person is. It isn't me, and never has been. The
foregoing statement has been a given, agreed basis, ever since information
theory first was mentioned in this discussion group. Everything I have
ever contributed with respect to information theory takes this for granted.

We tried the definition game a year ago. It didn't help.

No. It didn't. The definition of information, in Information Theory terms
is quite straightforward. Rick won't use it when he discusses information.
He always substitutes something else that makes nonsense out of any rational
statement involving information. With his definitions, all he says is
true. But they are not definitions related to the science of information.

Disturbances are uninformative, not because they are cancelled out by
outputs (which may be lower level reference variations when we are
dealing with systems at a higher level) but, rather, because their
influence on perception is always combined with the influence
of the system's own outputs.

This is the core nonsense statement that sits up like a stainless steel
wall between Rick and an understanding of the role of information theory.
It is a simple non-sequitur. The "because" is completely irrelevant to
the thing it is supposed to explain. We started last year with trying to
point out what nonsense this is, and here we still are. Sad.

IT advocates believe that the ideas of PCT were anticipated by or are
contained in IT.

What IT advocates? I have seen none on CSG-L to make claims anywhere like
that. Ever. That statement is close to libellous, if you are applying it
to me. Even Bill P. in Durango went so far as to say that I was almost the
only one who did NOT say that his theory encompassed, foreshadowed, or
otherwise took in PCT, but instead acknowledged that his own theory could
be seen as a consequence of PCT.

However, if you are saying that the ideas of control theory are contained
in arithmetic or algebra, there's some truth in it.

Trust me, Kent, information theory is simply irrelevant (at best) to control
theory -- not because we want it to be but becuase it IS.

Don't trust him, Kent. He knows not whereof he speaks.

By the way, your own ideas may turn out to be useful. They aren't what I
have been talking about, but they may link up when I manage somehow to
climb or cut through the stainless steel wall. I'm keeping your message
handy for later reference.

Martin