[From Bill Powers (940905.0830 MDT)]
[I sent this once but the process aborted before I could type in the
subject line. Hope this is not a duplicate]
Bruce Buchanan (940904.2125) -
As a relative newcomer to this field I realize that I have a lot to
learn, and I much appreciate the feedback and clarification of possible
mistunderstandings.
I'm always amazed to see how people who select themselves for CSGnet
react to criticism, correction, new information, etc.. The level of
defensiveness is weirdly low. The desire to learn is unnaturally high.
If the whole world were made of people like this, wouldn't it be a
lovely planet?
Actually, as I wrote to you about controlling versus controlled variable
I was in considerable doubt about offering this explanation, because the
risk of telling you something you already know was rather high. But I
figure that even if my explanations are superfluous for you, there's
always somebody else who might see something new in them.
I further understand that the ControllING Variables are
developed/evolve to provide guidance as reference signals, i.e. to
modulate the appropriateness of (perhaps more than one) lower level
ControllED Variable.
Not sure what you mean by modulating the appropriateness of a controlled
variable. As I understand them, reference signals specify the level at
which a variable is to be controlled.
However it is also my understanding that the perceptual signals are
aspects which we somehow abstract from the larger "real" situation
which, admittedly unknowable to us in itself (Ding an Sich),
nevertheless affects us.
Basically perceptual signals are the world as we experience it. When we
act on the world (or our relationship to it) we alter the physical
world, and hence the perceptual signals that represent it.
Another consideration is that part of the problem of the hierarchical
control system is also to adjust the higher level reference signals...
Reference signals aren't adjusted once and for all. When you raise an
arm from one position to another, you're adjusting the reference signal
against which position signals are compared, adjusting it smoothly from
one value (arm at side) to another value (arm horizontal). It isn't
obvious that control is going on until disturbances occur. Then you see
that at every point the arm will resist a push, even if the push is in
the direction it's going. A higher system controls its own variable (get
a drink of water) by varying the reference signals for lower systems
(hold the arm in this position, now this position, now this position
...). You can see that picking one fixed reference signal for arm
position wouldn't be much use to the system that needs to vary arm
position in order to get a drink of water.
···
----------------------------------------------------------------------
Bill Leach (940904.11:17 EDT)--
Your beliefs sound OK to me, most of them. Aren't you glad? That means
you can go on holding them, right?
I even agree with this:
I am one of those that believes that in any compromise between "True
and False", True loses.
If you measure truth on a binary scale, there are only two values: True
and False. So anything that isn't true must be false. The only problem
is that the truth of most statements can't be determined accurately
enough, on this number scale, to make them anything but False. Even
something that's 99.9% true is False.
If you present a continuous signal to a logic gate, what comes out of it
will be only 1 (true) or 0 (false). This means that logic is not very
good for continuous control of anything. There are certain cases in
which it is appropriate: you can go only up or down at a constant speed
in an elevator, so the control of direction can be carried out
logically: push this button or that button. But you can't drive a car
that way.
--------------------------
I am even willing to admit that there are people that could possibly
direct my "own affairs" in a much "better" manner than I can or do
myself, I do not believe that a person or group of people can direct
the affairs of a much larger group of people in general in a manner
that will result in overall "better" performance than is obtained when
the larger group attends to its own affairs.
It's doubtful whether anyone but you can know what all your reference
signals are at all levels. Better and worse, for you, are judged in
comparison with your own reference signals. Nobody else is likely to
guess right as to what will suit you exactly.
It occurs to me that all this political theorizing is futile unless
something is done to implement one system concept or another. But as
long as people have different system concepts -- especially concepts
with basic features in them that are never ever over-my-dead-body-not
going to change -- the only way any of them is going to take over is
through force. This pretty much leaves the field to those who say
"Force? Sure, why not, nuke 'em!"
The basic problem people have in getting along with each other is not, I
think, a problem with achieving agreement on a common system concept.
It's learning how to live together when system concepts differ. What
this means is that people can't afford to be tight, accurate control
systems at the system concept level. There has to be some looseness,
some play allowed. There's a limit to the practical amount of loop gain.
Maybe what we need is not an official philosophy of democracy or
federalism or republicanism or communism or objectivism, but a Ballpark
Philosophy. If the other guy's philosophy (system concept) is in the
same ballpark as yours, wish him well and let it go. You can work out
the details when they come up in real life. Isn't this more or less how
our government actually works? It looks very sloppy, but maybe it needs
to be sloppy in order to work at all. When you get two fanatics trying
to interact, even the slightest difference between their perceptions and
their goals will be intolerable to both of them; they will freeze each
other instantly into a conflict and nothing will get done at all:
gridlock. To get things going again you have to crank down the loop gain
on both sides, so that each side can say "Well, you're not offering
exactly what I want but I guess I can live with that."
You can always spot a fanatic. When he sees someone compromising, he
berates him for being wishy-washy.
All of this teases me once again with the notion that there's a level
above system concepts. How could we even talk about system concepts if
there weren't? There just has to be a level from which we can look over
all the system concepts there are and turn up the gain on this one and
turn it down on that one, or eliminate another one, or pick one that's
appropriate for now. Sometimes I'm all for Me First, and other times I'm
quite humble and self-effacing. It all depends -- on what?
There were times during the cold war when I looked at reports on how we
treated third-world countries and how the communists treated them and
thought, "Hey, those communists are doing a lot better than we are for
that country -- maybe they should give it a try for a while." Of course
I wouldn't have wanted it here, or in a lot of other places (including
places where it existed), but there were places where the commies just
had a more rational and humane notion of how to help people take a step
up. How could I have thought that (wrongly or rightly, that isn't the
point) unless I was in a position from which I could compare system
concepts?
Well, I get no further this time than before. I can live a point of view
that seems to be higher than system concepts, but I can't see it.
-----------------------
The expanded concept was that cities would be most directly influenced
by their citizens. Since people are at least nominally free to move,
the cities would be affected both by election and the results of their
decisions. Errors could be corrected relatively quickly and would have
limited damage.
I'm thinking of Mayor Daley's Chicago (father or son). My attitude is
that most systems of government have advantages, but most also have
disadvantages that show up only when they're put into practice. Anything
is worth trying for a while, provided you have a way of getting out of
it again. But that's the catch, isn't it? If any proposed system doesn't
sincerely provide a way for dumping it, one that could actually work,
watch out.
The American System as you describe it sounds (and often is) wonderful,
but lots of people complain that it isn't really being put into practice
as our forefathers envisioned it. To me, this shows that the Founders
didn't really figure it all out -- namely, they didn't have any idea of
how to get people to live up to it. Somehow the system they set up was
subject to drift; in some way or another, this group or that couldn't
live with it, and in coping with such people, other people started doing
things not according to the charter, until we arrived at our present
state. The Constitution is, after all, words, and words are ambiguous.
Does a well-regulated militia mean a militia that is subject to strict
regulation? Or does that little clause have nothing to do with the
following one, about the right to bear arms? Depends on who you ask.
---------------------------
I suspect that they were not fully "alert" to the idea that government
might not be the only source of such "power problems" though they
clearly knew of the effects of Britain's Merchantile but may have felt
that it was an error in governmental structure that allowed such
organizations to wield excessive power and influence (and I personally
am not so sure that they were wrong even yet).
What do you think about the decision to give corporations the status of
human beings?
--------------------------
We need then, I think, compelling evidence that a proposal to force or
restrict the freedom to control for an individual is actually in that
individual's best interest.
Is that really what laws are about? Some are, to be sure, but most laws
seem to be concerned with protecting one individual or a group against
another. Naturally the person who is prevented from wronging another
screeches bloody murder about the infringement on his freedom, but once
we're convinced that a person did something against the common good, we
hear these protests with little sympathy.
When people protest against regulations, I want to know what it is,
exactly, that they want to be free to do. If what they want is basically
the freedom to screw me, they won't get it from me. And I'll be happy to
live under the same regulation.
--------------------------
I am very cynical of "grand schemes". Communism is such a "grand
scheme" and certain (to me anyway) demonstrates it effectiveness well.
I am particularly skeptical of "elitists" plans and especially in areas
of human behaviour, I think that you should be also.
I am.
You already KNOW that the experts in human affairs "sciences" are
operating from incorrect understanding of human behaviour. That you
should want to concentrate the influence of such people over others is
almost amazing to me.
Oh-oh, did I do that? When, where, and how? Wait a minute, you were
replying to Rick. Oh, well, that's different. Yeah, what do you mean by
concentrating the power of human affairs sciences, Rick? Who told you
you could do that? We want you to dilute it IMMEDIATELY.
That's what I get for horning in on someone else's conversation.
-----------------------------------------------------------------------
Martin Taylor (940904.1720) --
Horning in again --
You can control for being at a certain place at a certain time AND for
doing all your travelling on your own two feet.
Yes, but you can't simultaneously control for those things and also for
the path you will take to the certain place, or the direction in which
you will travel at any moment on your own two feet, or how fast you move
at any time. This is not a matter of degrees of freedom, it's a matter
of the unpredictability of the environment. If you pick a path to a goal
and control for staying on it, you will run into people, buildings that
weren't there before, and holes in the ground. If you control for moving
at a certain speed, you will arrive either late or early, because the
path you will take is unpredictable. The means must remain variable if
the end is to be reliably reached.
-----------------------------------------------------------------------
Bill Leach (940904.1802) --
Ah, my very own post this time.
RE: 250E6 people not being able to take care of 250E6 people:
Perhaps that just means that no society so far has wanted to do it,
...
I don't think that is the case but then if you change the wording to
something along the line that no one has figured out how to do it yet,
I would agree.
I think I'll stick with my wording. If everybody who is insured under
the present system would agree to pay 15% more in health insurance (at
$3000 per family or $750 per person anually, that's $107 per capita),
everybody could have health insurance. People could do that if they
wanted to. They don't want to. I'm sure they have lots of very good
reasons, but the fact is they don't want to, and they could if they
wanted to. What could stop them, but wanting something else more?
The status of the poor, I think, has always been determined by the
balance between people's desire to take all they can for themselves,
and their level of concern for the welfare (or at least survival) of
others.
This is a pretty pessimistic view of humanity.
Not at all. I was just stating the conflict that actually exists. You
can't give away what you have and still have it. I stated that there is
a wide spectrum of solutions to this conflict, some amounting to giving
everything away without thought to the consequences, and some amounting
to giving nothing away except under duress. And everything between.
I fully believe that there are "enough" people that do want to TAKE all
they can for themselves but I don't believe that view is necessarily
the operative systems concept for even the majority of people.
The average view is evidently such that there aren't enough people
willing to lower their own standard of living to raise that of others to
keep the number of poor people in the country at any lower level than
the level that exists.
To give of what one has is to lose the use of it for oneself.
Yes, that is a valid view and even "appropriate" at times but it
certainly is not the only view point.
"At times?" Do you know of a way of giving away what you have and still
retaining the use of it? I thought I was just describing a fact of
nature. If I have $100 and give away $50, it seems to me that I now have
$50 to use instead of $100. If I have 1000 beans and give away 500 of
them, it seems to me that I have only 500 beans left to eat. Is my
arithmetic wrong?
Can one really put a dollar amount on a relationship? the "value" of a
friendship? There are some things that do not equate to other things.
I think there is enough friendship around for 250,000,000 people, but
that isn't usually what poor people complain about. Those jolly fat
mommas one finds in poor sections of Italy have plenty of friendship.
What they lack is enough money to buy anything to eat but pasta.
I agree that a control system that does not (or can not) reduce error
is in trouble. OTOH, a control system with NO error will never create
anything. I believe the "pop psychology" expression goes something
like: "Is stress bad?" "Yes... and no!"
...
It does not do any good to tell her that all her problems are someone
else's fault and someone else has to do all the "fixing". She or
anyone else must recognize that just living requires one's own effort
regardless of the amount of help available.
Let's get this discussion onto the level of PCT.
In normal operation, the perceptual signal of a control system is just
enough different from the reference signal to explain the level of
output effort that is occurring, the same effort that is maintaining the
perceptual signal. It is normal, therefore, for error to exist. However,
it is not normal for VERY LARGE errors to exist; few control systems can
operate normally with very large errors, as Martin Taylor has
documented.
The difference between the amount of error that causes normal actions
and the amount that causes maximum action is normally not large compared
with the setting of the reference signal. This is particularly true of
variables that are "important" to the organism -- that is, for which
loop gain is high. A rat being fed its daily ration by tube will not
press a bar at all to get more food. But if the daily ration is reduced
enough to make the rat lose 1% of its body weight, the rat will begin
pressing a bar to get food as fast as it can (in bouts) until its body
weight returns to normal. Body weight is evidently a very tightly
controlled variable for a rat, very important to it.
The loss of 1% of its body weight is therefore a sufficient incentive to
cause a normal rat to exert maximum effort. What would you conclude,
then, if you saw a rat with a body weight that was 10% below normal for
that animal, but the rat was not pressing the bar any more rapidly than
normal, or was pressing it even less rapidly? You would probably
conclude that there is something wrong with the rat.
According to one theory of behavior, what is wrong is that the rat lacks
sufficient incentive to work any harder. The solution is to increase the
incentive, which in this case means taking away some of the food that
the rat is getting so as to make the error larger. Doing this may
actually result in more efforts by the rat. The observer sees that the
rat is now working harder, and concludes that the added incentive
worked. However, since that increased effort is based on the increased
error, the added effort will not quite restore the body weight to the
original level (if it did, the output would drop to its former rate), so
the rat will be getting a little less food than before and weigh a
little less than before. Work out the equations; that's the result.
Another theory is that this rat is simply displaying a variation within
the normal range of rat behavior. It was born lazy. There is no point in
helping the rat species propagate laziness, so the best thing to do is
let nature take its course and eliminate this rat. The remainder of the
breed will be stronger for it. Too bad, but that's life.
A third theory is that this rat, for some unknown reason, has lost the
ability to get food for itself, and in its weakened condition it can't
produce the efforts required to restore itself to health. So what must
be done is to supplement the rat's diet with food that doesn't require a
bar-press to get it -- even to the extent of force-feeding the rat if
necessary. This, of course, has the effect of reducing or even
eliminating the error signal that is producing the bar-pressing, so the
bar-pressing gets slower and slower, and eventually stops. The person
who began by supplying part of the food ends up supplying all of it, and
the rat ceases to work for it at all, preferring nosing around the cage,
exercising in the running-wheel, scrambling after mates, and smoking a
little dope when there's some good music on. The helping person then
finds that it is impossible to withdraw the help, because the rat has
forgotten how to press the bar, its muscles have lost bulk, and it is
unable to relearn in time to survive.
There's a fourth theory called PCT. This theory would lead one to notice
the (apparently) extraordinarily low loop gain in this rat with respect
to controlling body weight. It is allowing its body weight to drop 10%
below what is normal for that kind of rat, whereas normal rats will work
hard if the drop is only 1%.
The first thing to do is to see exactly what is wrong with this control
system. By applying small disturbances (raising and lowering the food
intake by supplementing or withdrawing food), we measure the loop gain.
If the loop gain is normal, we know that something is wrong with either
the perceptual side or the reference signal. We might find that the
rat's perception of body weight is 10% too sensitive, so when its actual
weight is 10% below normal, the rat perceives normal body weight. Or,
the reference signal might be set 10% too low, in which case we would
have to try to see what is wrong with higher systems, if we could figure
out what they are. We might conclude that there is nothing wrong with
this rat; it just wants to weigh less than other rats do. There is
nothing to fix.
If the loop gain is only about 10, that would explain how a 10% error
could be allowed in this control system. In a normal rat the loop gain
is closer to 100. This says that there is something drastically wrong in
the comparison process, or in the output function that converts error to
action. We might be led to look lower in the hierarchy to see if there
is something that prevents the rat from producing as rapid a bar-
pressing rate as required for a normal loop gain of 100. We might find a
lesion in the spine that makes the response of the first-order control
systems inadequate; we might find an injury that makes all efforts
painful. We might find disease.
We would find _something_, because a loop gain of only 10 is entirely
abnormal for this kind of rat. And we wouldn't waste our time lecturing
the rat on its moral shortcomings, taking away what it needs and making
its situation worse, or force-feeding it into a state of torpor.
Presumably, we would care about solving the rat's problem enough to keep
working on its problem rather than just letting it die. What we want, in
the end, is a rat with a loop gain reasonably close to 100, so we can
turn it loose and forget about it.
Doing all this requires knowing something about how the rat works, and
taking the time to figure out what is not working right inside it. This
means we have to interact with the individual rat, looking for its
individual problem and trying to find a real solution if one exists. We
do this only because we have some interest in seeing a normally-
functioning rat instead of a malfunctioning one. This isn't do-
gooderism; we may want the rat restored to normal condition so we can
use it next week in an experiment that will kill it. All that's required
is that we have some reason to want to find out what is actually wrong,
and fix it.
Of course the last thing we would do in such a situation is to go around
moaning and weeping about the defective rat that we're stuck with, or
punishing the rat for not being a better rat, or becoming slaves to the
rat by trying to deal directly with the symptoms. PCT is a practical
science. The practical way to deal with a malfunctioning system is to
figure out what is malfunctioning and repair it if you can.
This is more or less how I view social problems. There's no point
berating people for not being more skilled at running social systems
than they are. There's no point putting patches over the patches.
There's no point in elaborate descriptions of the way people would act
if they were functioning perfectly in a perfect world. All that will
actually work is to get to understand what makes people tick, figure out
what is wrong, and try to fix it. I don't see any profit in getting
fanatical about any particular social system, either pro or con. The
particular ways we do things now are just not very important; they're
all just gropings in the dark. A century or two in the future nothing we
are doing now politically will be of any importance; it will just seem
quaint. Unless, of course, what we do is figure out the basics of human
nature and come up with a few facts so solid that they have a chance of
surviving for a couple of centuries. Our Founding Fathers camed up with
a view observations about human nature that have stood the test of time.
But a lot of what they did was just quaint. Why on earth did they think
that writing out a whole Constitution would lead everyone to interpret
it the same way?
-----------------------------------------------------------------------
Best to all
Bill P.