Steam Engine - RKC

March 20, 1992

[From Robert K. Clark (491-2499)]

Chris Malcolm and the Watt Steam Engine Governor.
Sorry to be slow to respond. My time for SCG is very limited as I am
responsible for several on-going activities. My CSG files are still
not well organized. There are many interesting topics, but I can't
follow them all!

I'm not sure what you are suggesting with this. Are you suggesting
that it is not "possible to trace all the signals/events through the
system??" Or that it may be difficult to do in some cases? This can
be particularly hard to do if digital computers are involved in the
absence of the system documents. With analog systems, or other
hardware systems, it may be easier.

What I had in mind was the tracing of each step in the system in
cause & effect terms. This is what sometimes results in tracing the
signals ("events") around and around the loop, with the reference
signal ("set-point") fixed. And then tracing the systems, likewise,
for the effects of changing the set-point both with and without
having the out-put "signal" fixed. Tracing the events through the
hardware may be awkward -- this is one advantage of the Control
System terminology. It has broad general applicability and is
convenient in many situations.

Of course much of the interest in Control System Theory of Behavior
is concerned with application to living beings, mainly, people. And
this quickly moves to tracing neural networks. Here it is not
uncommon to find people going "around and around the loop."

Perhaps my following response is unnecessary, but here it is:

The Watt Steam Engine Governor is indeed a negative feedback system
of brass and steel, with an adjustable set-point. Note I said "may
be possible to trace all signals/events through the system." These
"signals/events" need not involve conversion to alternative physical
form. After all, the concept of "signals" is just that, a "concept."

In the Steam Engine situation, there are several ways to describe the
"signals." For example, the distance of the spinning balls from the
axis can be regarded as the "feedback signal." This distance is then
converted by levers into a force applied to a movable valve,
countered by an externally adjustable tension on a spring or other
opposing force. As the valve opens (or closes) the engine operates
to change the speed of the assembly, particularly the rotating ball
system. Thus each step in the operation of the system is of a
"directly cause and effect" or "stimulus response" nature, while the
over-all operation serves as a control system. Note, however, that,
without some entity to adjust the "set-point," the operation of the
Steam Engine as a control system is incomplete.

There are many non-living systems in which the physical expression of
one or another signal may be different from what is expected. For
example, many thermostatic systems detect temperature by means of the
shape of a bi-metallic strip. In such a case, the temperature signal
can be regarded as the position of the end of the strip. Which may,
in turn, make or break an electrical contact -- the comparator --
producing an output signal to some part of the output function.

The physical expression of some feedback systems may be difficult to
untangle, but if the conditions defining such systems are met, the
over-all operation is clear.

You speak of "forecasting." Indeed, you are right. In some
situations a great deal can be accomplished without detailed
information. Meteorology, for instance. There are many examples of
this in physics and elsewhere. What I don't like about Determinism
(hence "Behaviorism" carried to the extreme) is the implication that
one is powerless to change ANYTHING! Of course this is extreme, and
usually handled by limiting the application of the deterministic
philosophy. I say I "don't like Determinism" because that is
essentially my orientation. I have been unable to find any way to
demonstrate either determinism (complete, unlimited) or free will
(limited, of course). So I "takes my choice."

I reacted to the quotes from Bill's remarks about BEHAVIORISM, with
which I agree, because it seems to me that a more philosophical
viewpoint is more effective. This also begins to show my approach,
primarily that of an experimentalist who needs a working theory --
that works -- in order to interpret and communicate his experimental

Regards, Bob Clark


Subject: Steam Engine - RKC