[From Bill Powers (921213.0700)]
John Gabriel, Bill Cunningham, Tom Baines, Martin Taylor, et. al.
At the risk of alienating friends, I have to make some further
comments on John Gabriel's post of 921211 11:12CST.
My first impression was that we are seeing hints of a great and
complex work by a group of people who have vast mathematical
expertise and deep experience with the world. After a night's
sleep, I awoke with the different impression. If this work is so
fundamental and important, why don't I understand it? And why,
beneath this work, does there seen to be such an unthinking
acceptance of the premises of military philosophers? Would a
scientific advisor to a torturer find it just as easy to become
immersed in the technical problems, and to ignore the underlying
repulsiveness of the whole undertaking?
I'm perfectly aware of my limitations and my ignorance, but the
fact is that when, as a teenager, I read Einstein's explanation
of relativity, I may not have known what a tensor was, but I
understood the idea perfectly well. When I read Norbert Wiener's
account of control processes in living organisms, I may not have
seen the relevance of stationary time series or followed his
arguments about Newtonian and Bergsonian Time, but I understood
how a control process worked and what it had to do with behavior.
When people describe a phenomenon of nature and offer a clean and
simple explanation of it, I usually understand what they are
getting at. So, I think, do most people.
When I read your post, John, I could not figure out what you and
the others were getting at. Beneath my surface reaction, I was
wondering "Why are they making it so complicated? What are all
these theorems supposed to be about? What makes them think that
any of this has anything to do with reality? Are these people
really of such a different order of human intelligence that they
can see simplicity and order in such seemingly vague and abstract
conjectures?"
Frankly, I don't think that the answer to the last question is
"yes" for ANYBODY who works in the field of human behavior. What
I see in morning's light is mathematics leading people around by
the nose, while they struggle to find real applications that will
give the resulting abstractions some shred of meaning. I see a
search for understanding of the mysteries of human behavior
misdirected, so it becomes a search for mathematical truths in an
imaginary universe.
What bothers me most is that this project of yours seems to
accept that the solution to human problems like warfare is to
figure out better ways to defeat the enemy. This is a total
misapprehension of the nature of the enemy and of the problem. It
bespeaks an acceptance of warfare as a fact of nature, something
we simply have to live or die with and about which we can do
nothing at all except to try to come out the winners. To applyPCT in this way is
like using Galileo's telescope to hammer a
nail.
In PCT there is the potential for understanding human conflict
and its resolution. That kind of understanding, not the endless
improvement of physical and organizational might, is what will
finally make the inthinkable undoable.
I wish you all would do the simple tracking experiments in Demo 1
and Demo 2, and ponder the outcome of the experiments at the end
of Demo 2. Here we have a model that will explain a simple motor
behavior carried out by a human being for one minute. It explains
it by predictively simulating the detailed movements by which a
person counteracts a random disturbance. If you fit the model
carefully, you can then predict the person's movements for any
new pattern of random disturbances, for a whole minute, with an
accuracy of about 3 percent of the peak-to-peak excursion of the
control handle. As Tom Bourbon has showed, you can predict with
this sort of accuracy how the person will behave a year later;
you can get the same predictively accuracy for 100 people picked
at random. Rick Marken has done the similar experiments in two
dimensions and with more complex controlled variables. If you
didn't look very carefully at the traces of real and simulated
handle movements, you wouldn't know which one represented the
human being's behavior and which one represented the model's
behavior.
Think about it. We can predict human behavior with this accuracy
for a WHOLE MINUTE. Tom Bourbon has shown that this can be done
with two people interacting in the presence of unpredictable
disturbances; two models will interact in the same way, with the
same accuracy, for ONE WHOLE MINUTE. Rick Marken has shown that
you can predict motions in two dimensions, with two-dimensional
disturbances and with a coupling between the dimensions, for ONE
WHOLE MINUTE.
There has never been this kind of ability to predict any human
behavior whatsoever in the whole previous history of the life
sciences -- not even the simplest of behaviors. Most
psychologists, cyberneticists, physiologists, and so on who have
seen this demonstration or who have reviewed articles about some
version of it have completely missed the point: they have never
seen anything remotely resembling this kind of accuracy of
prediction of behavior; as a consequence, they miss seeing it
altogether. They don't understand what the demonstration is
showing. You can say to them, ten minutes later, "Control theory
has the potential of explaining behavior with the precision of a
physical theory," and they will immediately launch into
counterarguments about variability and population statistics and
individual differences and environmental influences and such
like, proving conclusively that they didn't understand what was
in front of their eyes.
One whole minute, with one person or two persons, in one
dimension or two dimensions. That's the best anyone in the worldcan do right
now. That's where the true science of behavior
stands. The next step is not to solve the problems of military
organization or the causes of crime or the behavioral effects of
chemical concoctions or the complexities of adaptive control in a
network of 1E10 neurons with 1E14 connections. It is to try for
two minutes, three people, three dimensions, five different kinds
of controlled variables, always demanding that the prediction be
precise and work for every individual. This is how a science of
human behavior will arrive on this planet for the first time.
The vast complexities of human interactions will still be there
when this science of behavior has become able to handle them.
They will also still be unsolved by any of the present
approaches, even those that are trying to leapfrog all of the
hard and detailed work that is actually needed to bring PCT to
maturity. I think that any researcher who is trying to get lucky
with applying PCT to vast systems is wasting time. We are still
crouched beside Galileo, watching balls roll down an inclined
plane and timing them with our pulses. Nobody can imagine what
future will follow from these first primitive experiments. Of
course that future will never arrive if everyone insists on
trying to make a laser with the materials and knowledge available
to Galileo.
ยทยทยท
------------------------------------------------------------
Best,
Bill P.