limit cycles

[Avery Andrews 930510.1047 (Eastern Oz Time)]
(Bill Powers (930509.0830 MDT))

>And in the second place, a feedback control system that shows
>limit cycles is simply badly designed. This harks back to one of
>the earliest objections to control theory, which is that
>artificial control systems "hunt," while real behavior doesn't
>show any hunting. God, how these stupid ideas get lodged in
>people's heads!

My understanding (possibly wrong) is that a perfect feedback system
would be an l.c.o. whose limit is a point (like a capacitor discharging,
it never quites gets there). In real life, at least some of the
feedback systems I'm made out of aren't very well-designed -- if
I hold up my finger in front of my face and try to keep the tip aligned
with some object, it won't stay still, but jiggles around in a small
region.

I do agree with you that the dynamics people are out in left field in
being (a) descriptive/taxonomic rather than explanatory (b) uninterested
in mechanism (& therefore unlikely to notice the subtle phenomena that
are needed to tell which of the plausible mechanisms is the actual one
(c) unequipped with at least one of the ideas that is required to
figure out how these things work, the test for controlled variables.
(Let's not forget that the claim that PCT is necessary for the
understanding of living systems does *not* imply that it is sufficient.)

More on the other stuff later, with any luck.

Avery.Andrews@anu.edu.au