Hello, all --
Just a note to say Mary and I arrived safely home on the 7th after an
overnight stay in Cleveland while we waited for the rain to stop in Houston
(last stop before Durango). The meeting with Jay Forrester, George
Richardson, and Bob Eberlein went very well indeed -- they went through
about six of the demos including Squaring the Circle and the latest one on
the Adaptive Illusion and seemed to grasp all of them. They were quite
impressed by the fit of the tracking model to data we took from George's
performance -- better than 0.5% RMS (of the peak mouse movements)
difference between model and real performance. In fact, at first they
looked at the plot of the results and asked where the data for the model
was. It was on top of the curve for the real performance with just a little
green peeking out from behind the red trace!
George wants to do more publishing of PCT stuff in System Dynamics venues.
We'll be talking about it.
As to the model of pseudo-adaptation, I have now done a final tweak on the
program to get the frequency-response data scaled correctly in decibels (20
* log of gain). Now for the first and third plots, the slope of the
apparent frequency response of the controller is negative and positive 20
db per decade, respectively. I think this completes the proof that the
apparent changes in the forward characteristics of the human controller can
be accounted for pretty well by a model with fixed characteristics,
requiring no adaptation. A screen shot is attached -- executable program or
Delphi source code available on request..
Best,
Bill P.
PS: Congratulations to Rick for the new publication!
(Attachment Hierarchy31.jpg is missing)