Pyramids; Succinctness; Stella

[From Bill Powers (920903.1600)]

Thomas Baines (920903) --

You might consider looking at my _Living Control Systems_, the chapter
called An Outline of Control Theory. In it are defined 11 levels of
perception ranging from intensities to system concepts. These are, of
course, also levels of control. If the ordering is right, the lower
levels operate the fastest; also, it is necessary to control variables
of level n-1 in order to control variables of level n. I don't know
how well these properties actually hold up in every instance, but you
might find some suggestions in this hierarchical arrangement.

Use the term "model" carefully around here. We generally think of
models as things you can set up as a program, turn on, and run -- at
least in principle.


Greg Williams (920903) --

That's the trouble with succinctness. Somebody immediately comes up
with a counterexample -- in the worst case, in your own words.

"Controlling a person" means controlling some variable associated with
that person. This is possible without direct use of force only if the
variable is an output variable. Otherwise [Tucker] "... the term
'control' cannot be used to describe any activity that goes on between
"self-regulating systems"."

I was thinking of an activity "between" self-regulating systems,
meaning one in which they are interacting. If I'm controlling your
outputs, we aren't interacting; I'm just diddling around with side-
effects of your control actions and you're ignoring me.
Gary Cziko (920903.1340) --

I've seen Stella in operation, and it would be a fine tool for
simulating control systems on a Mac. I've heard that it's available
for PCs too. But my impression is that it's VERY expensive. If Rick
could get a copy I think he would freak out: it's really quite pretty.
You just draw block diagrams using standard components like amplifiers
and integrators, specify the parameters, connect the blocks by drawing
lines, and run! You can get plots of various variables, read in data
from files, and so on.

Greg and Pat Williams are going to be working on a general simulation
program that will probably work similarly. The only problem: it will
be written for the PC world. I presume it will be written in C,
however, so a C-competent Mac user might be able to port it to the Mac
world, in which case we'd all be able to run the same simulations.
That would truly be a great step forward.

Best to all,

Bill P.