[From Rick Marken (930807.1100)]
Bill Powers (930806.1945)--
It seems to me that the French derivation of "counter-roll"
doesn't quite capture the meaning we now use. Which set of
figures is the reference set?
Yes, I saw that problem -- especially in my bank statement
example. But I think that the original function of a "counter-
roll" (as best as I can make it out from the OED description) was
very much like a reference signal. The bookeeper was apparently
given a "counter-roll" (just as an ECS is given a reference signal)
and told to make the accounts (the perceptual input) match it.
So I think "counter-roll" captures part of what we mean by control;
the entries in the "counter-roll" are equivalent to the value of the
reference signal sent to an ECS. Whether these are "good" entries or
the "best" entries that could be be used as a reference standard was
apparently not a matter to be decided by the bookkeeper (just as an
ECS does not decide whether the value of the reference signal sent to it
is "good" or "appropriate"); both bookeeper and ECS are supposed to make a
perceptual variable match the reference input -- no questions asked.
Allan Randall (930806.1520) --
Just to make sure I understand, is the following a proper
paraphrase?:
Disturbance(1) (Taylor & Randall): that part of the CEV
fluctuation that is not caused by the control system itself,
and thus is due to an external cause.
Disturbance(2) (Powers & Marken): the external cause of that
part of the CEV fluctuation that is not caused by the control
system itself.
Yes. And I agree with everything Bill Powers said in his reply
and then some. It is normally impossible to pick d(1) out of the
continuous variations in CEV. As Bill said, about the only time
you can detect d(1) is when a sudden disturbance (d(2) sense)
pushes the CEV (actually, p) from a steady state match with r.
d(1) is only evident for a brief instant, before the system starts
responding to the effect of the d(2) disturbance.
As far as detecting the d(2) disturbance, Bill gave the simplest
reason why it cannot be done; variations in the CEV are typically the
result of multiple influences -- disturbances AND system outputs: Y =
a + b + c. Given variations in Y (the CEV) it is impossible to
determine a, b and c variations. I am claiming that given the CEV (Y)
you cannot even determine the NET effect of the d(2) disturbances
That is, if CEV = a + b + c + o, where a, b and c are three
disturbances (d(2) sense) and o is the output of the system,then a+b+c
is the net d(2) disturbance. This is the disturbance to which o is
opposed. The fact that the net d(2) disturbance and output are
connected to the CEV by possibly non-linear and time varying functions
(disturbance function and feedback function, respectively) there is
no way to determine that net d(2) disturbance (a+b+c) from the CEV
even if you are given o.
In general, variations in the CEV cannot be used as the basis for
computing variations in output, o, that keep the CEV (actually the
perception thereof) in the reference state specified by the reference
signal. Output variations control perceptions, not vice versa.
Best
Rick