[From Bill Powers (931223.1200 MST)]
Martin Taylor, Rick Marken, Tom Bourbon (931222) --
I see that I didn't need to chime in on the conversation about
pull-only control. Rick and Tom are right. Even if the pull-only
systems are arranged at odd angles, this does not make the
controlled variable ambiguous. It is not output that is
controlled.
Each pull-only system controls a single scalar variable within
its possible range of variation. A higher system, controlling
some other perception, then partitions its output so that the
output effects (via reference signals for all relevant pull-only
systems) on its own perceptual variable are negative-feedback
effects. There is no difficulty in defining the controlled
variable. The only ambiguity is in how the output is partitioned
among the pull-only systems -- and there, the boundary condition
is that the total effect on the controlled variable must be in
the right direction.
ยทยทยท
---------------------------------------------------------------
Best,
Bill P.