Avoiding walls; supply & demand; theory

[From Bill Powers (931016.1730) --

Avery Andrews (931016.1148) --

Try playing with the avoidance gain parameter. You can easily get

                * * *
              * *
             * W *
           * W *
  P* * W * G

Yet if you remove the wall, P will go straight to G. Also, if
there's just a single W, P will get much closer before starting
to turn. It's as though P understands that it can get around a
single W without turning so much, although of course it doesn't.


Hank Folson (931016) --

In your comments on the Williams Effect, you mention the Law of
Supply & Demand. I suspect that this is one of the non-laws
that Dag Forssell has been describing.

It's actually a generalization, but a venerable one. There are
cases in which it holds true, but of course what makes the
Williams Effect so interesting is that it shows a prominent
exception to the revered generalization, calling its status as a
"law" into serious question.

My example of non-compliance with this 'law' is the topical Too
Many Doctors Paradox: When there are more doctors than needed
in a community, medical costs go up .. My personal PCT
explanation is that doctors control for a reasonable income

I think you're perfectly right, with respect to most doctors. The
Law of Supply and Demand assumes that every supplier is trying to
maximize income, implying that prices will rise until demand
falls enough to create a stationary situation. So the model of
Economic Man you use makes a great deal of difference in how you
expect an economy to work.

However, the LSD can still occur when doctors are "satisficing."
If enough doctors come into the community, causing doubling,
tripling, etc. of prices to maintain constant income, at some
point the community will be unable or unwilling to pay more and
will start cutting back on its patronage of doctors in general.
Now when a new doctor enters the community, the other doctors
can't raise their prices without losing more patients than they
gain in income (it's cheaper to travel to another town, etc.).
All the doctors are then earning less than they want to earn, and
as far as an observer can tell they act as if they're maximizing
(who can tell where the goal-income is set?). The Law of Supply
and Demand really applies in situations where nobody is getting
as much as is wanted - a competitive scarcity economy. There are
no existing economic laws that could handle behavior in an
affluent society (except PCT).

This has generated a lot of traffic on the 'net, but I don't
understand the point or value of this experiment either for
linguistics or PCT.

As Bruce will no doubt explain, the "spin/sbin" pair is only one
of hundreds of examples in which parts of words are pronounced
differently depending on what the other parts are. This isn't
important to a child learning how to speak, but it is important
to a PCT modeler trying to devise a model that can understand and
produce speech -- and I mean actually do it, not just give people
some words to say that sound like explanations of the process. If
two sounds in different words are actually the same, why are they
heard as different? If two sounds are actually different, why are
they heard as the same? That's the challenge to the modeler.

I think PCT can explain spin-sbin: We are hierarchical control
systems, interpreting & using language (solely for controlling
purposes) through a hierarchy, processing the vibrations we
sense through many levels until the information makes sense.

That's a description of what a PCT explanation might look like.
It is not itself an explanation. A real explanation would lay out
the control systems that do the interpreting and the using, and
show by demonstration that they actually accomplish the necessary
operations and produce the same behavior we observe. Most people,
not being interested in real explanations, would be happy to
accept your description in lieu of a real explanation, and
probably wouldn't know the difference.

When I design a new bicycle part, I have a big error signal if
I can't apply a theory of stress and strain to design my new
light weight highly stressed part that must perform without
failure. However, this is a recent perversion of living
control systems.

I think of it as a recent advance, which started around the time
of Galileo. Before that time, the only way you could design a
structural member that wouldn't break was to try out endless
combinations of materials and designs and stack heavy loads on
they until they broke. You had no way to predict which designs,
given materials of certain strengths, would support the largest
load. So you could end up with satisfactory designs all right,
but in competition with the guy who had a theory of design you
would be left in the dust.

So I guess the question is: Is an interest in truly scientific
theories an aberration that will die out quickly when it
provides no evolutionary advantage, or does it represent a new
improved controlling technique which will prove to be in the
long run an evolutionary advantage for those who use hard
theories over those more primitive control systems that don't?

Do you see any major engineering projects other than drug
research that are still carried out in the way Thomas Edison did
things? Would you like to build a television set by starting with
a pile of parts and trying out different ways of connecting them
until you suddenly had a picture, or would you rather learn
electronic theory first?

I think truly scientific theories are here to stay, even if most
people can't see what they are good for.
Best to all,

Bill P.