PCT Officer Nabs Two Suspects

CSGnet Archive CSG1994
1

[From Rick Marken (940401.1159 -- last chance to be an April Fool)]

While prowling the information highway this evening, officer Rick (The
Enforcer) Marken ran across two suspects in "the case of the cause of
behavior". Martin Taylor (940401 18:00), looking tired but still putting up a
great deal of resistance, was overheard saying the following:

The result that was expected was that the correlation is positive and that it
is not large. The data look to me to be as expected. Still.The derivatives
are certainly NOT independent. Is that clear?

Taylor was cited for violation some elementary rules of inferential
statistics. The correlation presented was a sample statistic. The null
hypothesis is the the actual correlation between derivatives is 0.0. The
observed correlation, if large enough, allows us to reject that hypothesis
with a known probability that we are making a type I error (rejecting the
null lypothesis by mistake; convicting an innocent correlation). Since
Marken did not have actual r value or statistical tables available, he was
unable to determine wether the r was significant. But, even if it IS
significant, there is still a chance (Type I error probability) that the
derivatives are not "certainly NOT independent" (PCT officers are specially
trained to use double negative; do NOT try this at home).

Taylor was also booked on suspicion of using the ridiculously small
correlation between derivatives as evidence that the observed relationship
between disturbance and output in a control system depends on perception.
Given Taylor's previous record, there was reason to believe that the
purported "non-independence" of the derivatives was going to be used as
evidence that perception is the cause of control system output -- a violation
of the second law of PCT which says that output depends only on the
disturbance, not on perception at all.

While apprehending the suspect, Marken was caught off guard by Dag
Forssell (940331 1600) (actually, Marken had given Forssell a warning in the
afternoon which apparently went unheaded). Forssell was seen scrawling
the following near the Al Gore memorial information highway overpass:

When we are unhappy with the results of the performance of another, it is
best to ignore the action/behavior--the by-product or symptom--and ask
instead about the wants and perceptions, which are the causes. (As
suggested by exhibit 13).

"Perceptions are causes? I don't think so", said Marken. Forssell was booked
on suspicion of supplying arguments to the "you can't control me becuase I
can perceive what I want" gang -- a dashing band of PCTers who believe that
perception causes behavior, but that it can't control you because YOU create
your own perceptions. "Got a problem with a perception?" ask members of
this gang, "like a manager who's trying to take credit for all your work? Not
to worry. Just look at it in another way; just perceive the manager as the
kind of person you admire -- a tough sonovabitch who is trying his best to
live the American Dream."

"The arrest was a real surprise", said Marken, controlling several perceptual
variables simultaneously, "because you don't usually find the "derivatives
are not independent" and the "we create the perceptions that cause our
behavior" people in the same neighborhood -- let alone at the same time.
"But it does make sense", Marken nodded, doing his patented "controlling
without awareness" schtick. "They speak a different language, sure; but
they're both sayin' the same thing; perception causes behavior. I guess it's
just tough to get comfortable with the idea that it ain't so; perception
doesn't cause behavior in a negative feedback control loop. I know its hard
notion to get, but if Gary Cziko can get it, ANYONE can."

Happy day, fellow fools

Rick

2

<[Bill Leach 940402.10:15 EST(EDT)]

[Rick Marken (940401.1159 -- last chance to be an April Fool)]

Oh Rick... you had me in stitches. Particularly with your completely
unexpected "Al Gore memorial information highway overpass".

Seriously though (note that it is the second of April now)...

I don't particularly want to clutter up and already busy net, so I just
settle for cluttering you own personal mail box, I mean like cops are
public servants right? :slight_smile: (as is obvious, I changed my mind)

I think that I understand enough to know that perception does not cause
behaviour... simply stated, there are only three things that can cause
any change in output and those are a change in disturbance that results
in a change in a controlled perception, a change in one of more reference
signals, or re-organization (which in itself only causes a change when it
alters the reference and/or the transfer functions).

Amicus Brief:

Behaviour is not controlled... perceptions are controlled.

Thus, Officer Marken charges the defendent with making this statement (in
effect):

"Got a problem with a perception?" ask members of this gang, "like a
manager who's trying to take credit for all your work? Not to worry.
Just look at it in another way; just perceive the manager as the kind of
person you admire -- a tough sonovabitch who is trying his best to live
the American Dream."

The inconsistancy in this charge is present in the last sentence. To
wit: The verb "choose to" is missing. The correct sentence fragment is:

"...; just choose to perceive the manager ..."

"Choosing to perceive" anything is the act of setting a reference to
control a perception (in this case reseting the referenece for the
control of the perception of the "manager").

Therefore, your honor, I submit that the defendent is innocent of the
charges levied against him.

It is easy to understand how "disinformation" or missrepresentation of
facts could occur when one has been in the vicinity of an Al Gore
"anything" but nonetheless, this court is obligated to the truth and
nothing but the truth.

In that vein, the phrase "the by-product or symptom--and ask instead
about the wants and perceptions, which are the causes." deserves some
critical analysis.

"Our manager" in the case presented by the defendent is attempting to run
a business (expressed in the "common language"). The defendent, being
believer in the true belief is aware that this idea specifically means
that the manager IS controlling perceptions to references that are
established from other perceptions concerning the nature of business, how
businesses are operated, HOW PEOPLE FUNCTION, and how a manager should
interact with people to achieve the "want" of "run a business".

Further evidence can be offered if it pleases the court, that the
defendent is aware that few people (particularly managers and Vice
Presidents) have even the foggiest notion what a control system is much
less that they themselves might actually be one.

Thus, the defendent in trying to help this rather large group of people
to understand why what they are trying to accomplish is not working is
faced with the monumental problems presented by such facts as:

  The managers know that something is wrong but absolutely do not believe
  that there is anything fundamentally wrong in the basis for their
  decisions (they think they have the wrong quality assurance plan or
  some such irrelevant thing).

  The managers and the people that they are trying to manage do not
  understand how and why they function.

  Natural language is imprecise under the best of conditions but in this
  case the defendent is almost guaranteed to be missunderstood.

  He is dealing with people that are already suspicious that
  "behaviouralists" know less about interpersonal relationships than they
  themselves do (and this perception is clearly not without merit)

  The people that he is trying to influence are "bottom line" people and
  he must address issues in a manner that relates directly to their
  existing focus.

Now I think that it is clear that the defendent knows that when trying to
teach the application of PCT principles many difficulties arrise.

For example in an attempt to resolve a problem a manage might ask a
worker: "What are you trying to accomplish?" The manager might then
observe (perceive) a behaviour (resulting from the complex processing
based upon wants that what the worker's perception of the question
triggered -- naturally resulting in the setting of references for
control) such as: "I am trying to achieve the goals of the company of
course --- sir"

In such an exchange, the manager, if she is PCT savy, will recognize that
the worker has some significant perceptions that are apt to result in a
long meeting and that the only "want reveiled" indicates that the worker
either never thinks (a distinct possibility -- remember the overpass) or
the worker has little faith in the "system" (again, not without merit).

So, Dag's percision might be a little sloppy in his tacking on the phrase
"..., which are the causes" but it is obvious from his work that he
ultimately strives for examination of wants. Controlling perceptions is
the way to "get there"... in fact controlling perceptions is the only
thing that we can do.

     //////////////////////////////////////////
     / /
     / -bill /
     / bleach@bix.com 71330.2621@cis.com /
     / ARS KB7LX@KB7LX.ampr.org 44.74.1.74 /
     / /
     //////////////////////////////////////////

3

<Martin Taylor 940402 13:50>

Rick Marken (940401.1159 -- last chance to be an April Fool)

Chance well taken. Lovely posting. Police Officer nabbed for false arrest
and speeding to conclusions.

Taylor was cited for violation some elementary rules of inferential
statistics. The correlation presented was a sample statistic. The null
hypothesis is the the actual correlation between derivatives is 0.0. The
observed correlation, if large enough, allows us to reject that hypothesis
with a known probability that we are making a type I error (rejecting the
null lypothesis by mistake; convicting an innocent correlation). Since
Marken did not have actual r value or statistical tables available, he was
unable to determine wether the r was significant. But, even if it IS
significant, there is still a chance (Type I error probability) that the
derivatives are not "certainly NOT independent" (PCT officers are specially
trained to use double negative; do NOT try this at home).

Ah, well. Let us consider this response also to be April Fools, since
any use of significance statistics, ever, has to be foolish.

I do happen to have statistical tables at hand. As measured, there are
3000 samples (or maybe it is 3600--Bill P. didn't say which set of data
he used, but it is usually 60 seconds of 60 samples per second) and the
stated correlation between derivatives of disturbance and of perceptual
signal (delayed) is 0.268. If the samples were all independent, the
probable error of this correlation would be 0.01044. The correlation
would be roughly 25 standard errors different from zero, somewhat
unlikely to have happened by chance.

However, this overstates the case. The samples are not independent, since
the bandwidth of the control system (and of the disturbance waveform)
is less than 30 Hz (as that of the control system must be, given a transport
lag of at least 130 msec in the tracking model fits on which the calculation
was based). So let's give our Court Fool a better chance to be right, and
take a sampling rate guaranteed to provide independent samples: 4 per
second, giving 240 samples for the test of significance. Under these
assumptions, the obtained correlation is only about 7 standard errors
from zero, giving a MUCH better probability that the observed correlation
was due to chance. At that level, the probability goes WAY up, to one
in 800 billion.

(To bring the probability that the correlation observed was due to chance
down to reasonable levels (say 1 in 100), the number of independent
samples would have to be no more than about 30, or one per 2 seconds.)

So Officer Marken is right on a technicality. It is not CERTAIN that
the derivatives are NOT independent. The probability is one in 800 billion
that they are independent, and that the correlation is spurious. One in
800 billion is NOT certainty. However, the chances are rather better
that the correlation is real than that it isn't.

Of course other factors are much more important than the computed
statistical probabilities, when you get to these levels. It is much
more probable, for example, that Rick does not exist, or that if he
does, he will be struck down by a thunderbolt from heaven this fiscal
year. Now THAT would be a significant result, though like so many
significant results, it would be much to be deplored.

I think that for the purposes of continuing discussion, I will continue
to say that the derivatives of the disturbance and of the perceptual
signal are NOT independent (though I may sometimes leave off the "certainly"
in deference to the .000000000012 probability that they are independent).

Given Taylor's previous record, there was reason to believe that the
purported "non-independence" of the derivatives was going to be used as
evidence that perception is the cause of control system output -- a violation
of the second law of PCT which says that output depends only on the
disturbance, not on perception at all.

Objection, your Honour. This is a leading comment. And uses the term
"the cause," a notion that I am unlikely to use in any serious discussion.
Except to further the Cause of PCT. However, the cited "second law" of
PCT will be appealed, conflicting as it does with the first law: "It's
all perception."

Happy New (Fiscal) Year.

Martin