the GREAT DEBATE

[Jim_Dundon_940220.18]

{From Jim Dundon 940209.1850

  >Martin Taylor 940207.1945
  >>Rick Marken 940207.0900

  >If I remember correctly, my first naive attempt to introduce
  >information theory to CSG-L was based on the information
  >stability of a closed loop, stability implied by the
  >fact of control.

  >>The behavior of variables in a closed loop is completely
  >>different than the behavior of the same varables when
  >>the loop is open. This is why open-loop based model
  >>model is irrelevent to understanding purposefull
  >>behavior.

    Rick, Martin,
    I beg your patience, the engineer may look at this somewhat
    differently, but the concept "open Loop" Is difficult
    to grasp.
    As I see transmission lines they are loop
    segments that have no purpose except to function in
    conjunction with purpose. They are useless except in use.
    They are given value by purposefull organisms.
    For all practical purposes they are non-existent unless
    in use. When in use they are part of another loop. The loop
    that Rick, Bill, Martin, Jim help to create
    when we sit down to
    the computer and use plastic, copper, paint, electricity,
    fuel, etc. in a warm room filled with furniture, The loop
    of political economy, of money flowing in one direction
    and consumable goods in the other. A little like neurons
    and energy flowing in opposite directions.

    In any event, a transmission line is a loop segment that
    has no purpose unless in use, and its value is value in
    exchange. It's purpose is to facilitate exchange of value
    Call the medium intelligence, information, whatever,
    it's function is to facilitate a certain kind of exchange
    of value, and as such is part of a loop, not a loop.
    It facilitates the influence of one variable upon
    another in the exchange of value. Sounds like a PCT axiom.

    Rick, Martin

    If the above view is accurate, can a transmission line
    have a counterpart in function in a living organism
    closed loop, that is as part of the loop. as a nerve
    maybe?

    Would axioms of PCT be usable in an economic loop?
    In view of the fact that you view human behavior as
    purposefull, would it be safe to assume that you
    consider the whole economic activity as purposefull
    or as an expression of purpose?

    loop?

    Rick, Bill L.
    thank you both for for your help. thirteen-fourteen hour
    days standing up will limit my participation mostly
    to weekends. I will look them over between now and then.

    Best, JIM D.

[_Martin_Taylor]

<Martin Taylor 940210 11:15>

Jim Dundon 940209.1850

When we talk about "open loop" vs "closed loop" we tend to be dealing
with what might better be called "control" versus "outflow actions,"
rather than with what concerns you. As we have used the terms, in
both systems there is a thing (complex) in the presumed-to-exist "real
world" for which some state can be determined (e.g. the position of
an object, the state of human rights in a country, the agreeableness of
the weather today...). We call that "thing" a "Complex Environmental
Variable" or CEV. In both systems there is a desired (reference) state
of the CEV, and its actual state can be affected by actions. From that
point, the "open loop" and "closed loop" systems differ.

In a "closed loop" system, actions affect the state of the CEV, which is
perceived. The perception is compared with the reference for that perception,
and the resulting "error" affects the actions on the CEV, hopefully in a
direction which reduces the error, bringing the CEV toward the desired state.
The overall effect is to maintain the perceived state of the CEV near the
reference value. This is NOT the same as maintaining the actual state of
the CEV near a reference value. The CEV exists in an unknowable "real world,"
that affects but is not the same as the set of perceptual signals available
for control.

In an "open loop" system, the reference state induces actions which "should"
bring the CEV to the desired state, but that state is not perceived, and
if the actions fail in their intent, the system will do nothing to inprove
the situation.

An open loop system cannot work if there is any uncertainty in the state
of the CEV consequent on the action. In the world in which we live, it is
almost always impossible to determine beforehand what the state of a CEV
will be as a consequence of an action, especially if the current state of
that CEV is unknown.

Most CEVs are subject to influences that the system can neither observe
nor predict, which in CSG-L discussions are called "disturbances."

In a control system, disturbances alter the perception of the CEV whether
the reference state changes or not, but the closed-loop structure of the
control system maintains the CEV state more or less stable despite these
disturbances. The words "more or less" are important, because there are
many factors that make perfect stability impossible to achieve except over
infinite time. In an open loop system, they simply make the state of the
CEV differ from its "intended" value. There is no "control" in an open loop
system.

In our discussions, the kind of loop involved with "money flowing in one
direction and consumable goods flowing in the other" is not considered, since
a control loop requires there to be a substantial negative gain somewhere
in the loop (amplifying the difference between the reference state
and the perceptual state in the usual diagram). The world has many
situations involving loops in which A affects B affects C affects A, but
these are not usually discussed in CSG-L unless there is a substantial
negative gain in one of the links.

   Would axioms of PCT be usable in an economic loop?
   In view of the fact that you view human behavior as
   purposefull, would it be safe to assume that you
   consider the whole economic activity as purposefull
   or as an expression of purpose?

Yes. Bill Powers has several times tried to start a discussion thread
going on the application of PCT to economics.

Martin

[Marken_Richard_S1]

[From Rick Marken (940210.1330)]

Martin Taylor (940210 11:15) to Jim Dundon 940209.1850:

When we talk about "open loop" vs "closed loop" we tend to be dealing
with what might better be called "control" versus "outflow actions,"
...etc

We may have a "great debate" going about whether there is information in
perception but we certainly have no debate about the difference between
open and closed loop systems.

This was really an EXCELLENT post, Martin. Your description of the
distinction between these kinds of systems was wonderfully clear
and lucid. I'm sure a great sigh of relief will go up all around the
net as I admit that I can say no more about open and closed loop systems
than you did - and I certainly could not have said it better.

Best

Rick

[Bill_Leach1]

<[Bill Leach 940210.18:50EST(EDT)]

Jim Dundon 940209.1850

A transmission line is sometimes thought of as a loop because the
individual electrons (so the theory goes) actually travel through the
conductors and whatever is in series (in one path at a time) until they
are back at their starting point. Indeed an individual electron may
traverse the "loop" many, many times in a comparatively short period of
time.

As far as control theory goes (that is control theory with respect to
"industrial control equipment" as well as PCT), the term "loop" can apply
to about anything from the most simple (input, reference, comparator,
transfer function, and output) controller to the entire system.

In closed loop control system (PCT or otherwise) the "input" (Perception
for PCT) is compared to the "reference" by the "comparator." The
"difference signal" (that is the difference between the input and the
reference) is "fed" to the transfer function which generates an output.
Under "normal" control conditions, the output should then "act upon"
something which will bring the "input" (or Perception) into alignment
with the reference.

This is the essence of closed loop control; within limits, the system
adjusts its output to make the perception match the desired value
(reference). The concept is true for something as "simple" as blinking
to as complex as climbing a mountain.

As I see it, one of the great difficulties present in PCT research
(currently anyway) is that you can not "watch" a single control loop.
Even the simple tasks such as the "tracking" experiment probably involve millions of individual control loops acting in "concert". You can think
of these many control loops in terms of levels in a hierarchy.

The entire structure can be viewed as a control loop and the various
levels can be viewed as interdependent control loops. The "higher" level
control loop's output then is the reference for the lower level and
conversely the lower level control loop appears as a part of the higher
level's "transfer function" when multiple levels are viewed together.

The term "open loop" or more correctly "open loop control" is actually a
poor choice of terms. In reality an open loop control is not a loop at
all but the term comes about because it describes a control method that
is radically different from "loop control" (or closed loop control).

"Open loop control by design" assumes that the output WILL cause the
desired action "without checking." Such "control" is sometimes valid for
industrial control purposes.

In human behaviour, I suppose that such a control state might be proper
term when "reorganization" is occurring, or something physical is wrong
with the person such that the Perceptual variable is absent.

-bill
                          Government is evil...
                  Less evil than anarchy but still, if
                  left to itself, fundamentally evil!!