[spam] Re: recent Chinese military history

[From Bill Powers (2006.08.11.1235 MDT)]

Tracy Harms (2006;08,10.09:55 Pacific) --

Although brief, the 1979 war between China and Vietnam
did have Chinese troops in serious combat.

Sino-Vietnamese War - Wikipedia

Thanks, Tracy. I had forgotten that war (though I'm embarrassed to say it). It's not quite clear whether the Chinese suffered an estimated 60,000 casualties or 60,000 dead. I assume it's casualties, with the ratio of wounded to dead being somewhere between 3 and 10 depending on the medical technology available. So that leaves us with our 45,000 to 50,000 dead in Vietnam, plus all the others since, in the running with those with whom we have been having wars. Whatever the exact numbers, it's clear that this proposed measure of conflictedness does not clearly distinguish totalitarian from free societies. Actually it's not a very good measure because in a war between a free and an equally strong totalitarian society, both sides will rack up large numbers of casualties.

This suggests another measure: the number of countries with which each country has been effectively at war during, say, the 20th Century. This is something like finding out who the class bully is by finding out how many different students each student has had a fight with. Such a measure avoids the question of justification, which is good because naturally every kid in a fight has a good reason for it. Perhaps the count should be weighted to include the relative populations of the countries in each war (the ratio of weights of the kids in each fight). For each other country, add up (My population) / (other country's population), then divide by the number of wars.

Here's a measure of internal conflictedness. When I was a kid, the rule was that one person in a family, usually the father, worked and brought home enough money to feed, clothe, house, and educate everyone in the family. Now I suspect that the median number of jobs per family is 2 or more. This means that to accomplish the same ends or at least a satisfactory life, twice as much work or more is required. This implies that work has become less efficient, which might prove to be because so much labor is used just to prevent other people's labor from being effective (think of recent advertising campaigns in the auto industry). Think of what happens when NASA puts out a Request For Proposal on some job like building a new spacecraft. Many companies each spend millions and thousands of man-hours in doing the research, testing, and writing needed to make a bid -- but only one of them gets the contract.

The general principle of competition (institutionalized conflict) implies that when N entities (people or companies) compete for the same indivisible resource or work, (N-1)/N of them must fail, so that the average effectiveness of a person or company in obtaining a resource or a job is reduced by a factor of N.

Here's another (I'm just tossing these in the pot in raw form). There is always a basic conflict between those who produce and sell goods or services and those who buy them. The buyers want to get the highest possible quality of product for the lowest possible price, whereas the producers and sellers want to provide the lowest possible quality (quality is expensive) at the highest possible price (return on investment). Similarly, there is a conflict between the people who produce goods and services directly, and those who control the means of production and hire workers to do the producing. The workers want to get the highest possible wages for the least possible amount of work and the producers want to pay the lowest possible wages for the largest possible amount of work.

In the above I omit the concept of "enough", which really needs to be considered: sane people to not want infinite quantities of anything, money or goods/services.

It's often claimed that competition drives improvements in the economy by forcing entrepreneurs to offer better goods and services than the other guy. Note that the effect on the "other guy" is not factored into the measurement of "improvement." But I have met few people outside of sports who would admit that they do their best work when in fear of losing their livelihoods. I think there is a lot of spurious assignment of causality in such arguments, as well as claims made under circumstances where it is impossible to compare approaches by going back and doing the same thing a different way. The economy works as it works in the presence of competition. So what is causing what? The only real trials of non-competitive economies of which I know were not "pure": that is, not only was competition eliminated, but at the same time intellectual repression, force, ideology, and brutality were introduced, or increased. So what caused the test to fail? You can't determine that by changing half a dozen major variables at the same time.

I suspect that competition, being a form of conflict, has the same effects that conflict always has: it wastes energy and reduces the ability to control. I don't know right off the bat how to test that suspicion.

Best,

Bill P.

[From Bill Powers (2006.08.14.0830 KMDT)]

Marc Abrams (2006.08.14.0637)–

But it seems that this is a one
way street here. That is, both Bill and Rick want, and expect economists
to read and understand B:CP but both seem to be unwilling to familiarize
themselves with just some of the most basic tenants of the
discipline.

Here is a basic tenet from the cited wiki page:
Austrian economists reject
observation as a tool applicable to economics, saying that while it is
appropriate in the natural sciences where factors can be isolated in
laboratory conditions, acting human beings are too complex for this
treatment. Instead one should isolate the logical processes of human
action - a discipline named “praxeology” by Alfred Espinas.
Ludwig von Mises is commonly miscredited as coining the term
‘praxeology.’

This is very Greek, of course. Arisotle declared that men have more
teeth than women, and apparently he never actually counted them because
the ancient Greeks considered experimentation and observation to be an
inferior way of arriving at Truth.
I have done some studying of economic texts, of a modest sort (more than
economists have studied PCT texts, for sure). I remain unconvinced that
the secrets of human interactions in the economic mode have been
successfully described or understood by anybody – Marx, Keynes, Mises,
Samuelson, or anyone else who claims the title of economist. There seems
to be a great deal of self-interest in economic theory; that is, theories
are proposed that seem designed specifically to justify whatever status
quo the author already supports, or to demolish what the author already
rejects. The spirit of exploration and discovery is hard to find. I
remember a letter from Galbraith to my father in which he admitted that
while TCP may have made a few good points, Galbraith simply loved the
capitalistic system and was not about to give it up because of what any
theory or data said. His conclusions, in other words, had already been
selected, so even if TCP was right about something, Galbraith was not
about to change his mind. I’m sure others felt the same way but he’s the
only one who said it.
One common thread I have found is what economists consider to be the
community whose interests guide economic processes. In the paragraph
following the one cited above, we find:
**Austrians view
entrepreneurship as the driving force in economic development, see
private property as essential to the efficient use of resources, and
often see government interference in market processes as
counterproductive.**This is exactly the point of view one would expect an
entrepreneur – but not a consumer – to defend. Judging from what
entrepreneurs complain about, the entrepreneur does not want to be
hampered by any precepts of honesty, fairness, compassion, or integrity:
he just wants to be left alone to do what he thinks will work to make
more money. This is called letting “market forces” determine
the outcome. Of course from the standpoint of the entrepreneur that is
only reasonable, but from the standpoint of the consumer who bought a
driveway installation last week only to find that it is crumbling away
this week, this lack of accountability is far from desirable – it
negates the very reason for having an economic system, from the
consumer’s point of view. It is the consumers who bring about the
regulations that entrepreneurs try to escape from, and consumers call for
regulations because all too often entrepreneurs come across as
sociopaths.
From the standpoint of PCT, the only basic driving force in any economy
is the set of all reference conditions that people seek. Mises admits
that some of these reference conditions may pertain to survival, or to
the manner in which one survives. In fact he sounds very PCTish in
places:
**When applied to the means chosen
for the attainment of ends, the terms rational and irrational imply
a judgment about the expediency and adequacy of the procedure employed.
The critic approves or
disapproves of the method from the point of view of whether or not it is
best suited to attain the end
in question. It is a fact that human reason is not infallible and that
man very often errs in selecting and
applying means. An action unsuited to the end sought falls short of
expectation. It is contrary to
purpose, but it is rational, i.e., the outcome of a reasonable–although
faulty–deliberation and an
attempt–although an ineffectual attempt–to attain a definite goal. The
doctors who a hundred years
ago employed certain methods for the treatment of cancer which our
contemporary doctors reject
were–from the point of view of present-day pathology–badly instructed
and therefore inefficient.
But they did not act irrationally; they did their best. It is probable
that in a hundred years more
doctors will have more efficient methods at hand for the treatment of
this disease. They will be more
efficient but not more rational than our physicians.**For Mises, “rationality” is a catchword for all
higher mental processes, what we in PCT would call higher-level
perceptions with their associated control systems. But the emphasis is on
logic, not principles or system concepts. If there are economists who
employ different principles or support different system concepts, to
Mises they are simply irrational, because they fail to use the logical
processes he uses. This self-centered view negates a lot of what is
PCTish in Mises’ writings, by eliminating perceptions higher than those
of logic. It may be that installing a driveway using cheap materials and
then moving on to the next city or state is a rational means of
maximizing return on investment, but that is only under a certain set of
principles (If you get away with it, it must be OK) or system concepts
(nature red in tooth and claw). Mises does not discuss these principles
or system concepts; he simply acts out the ones that happen to determine
the premises he uses in his own view of what is logical.
If Mises had been acquainted with HPCT, he would realize that what is a
purpose at one level is a means at the next level up. A particular set of
premises might determine the course of logical reasoning, but why choose
that set of premises instead of another? The answer lies in the
principles one is controlling for at a higher level. Different principles
lead to the choice of different premises. And why those principles? They
too, are means to an end, the system concepts that make up one’s picture
of the kind of self, social, or inellectual system one admires and tries
to help create.
Here, more or less, is Mises’ system concept of Economic Man:
**In every living being there works
an inexplicable and nonanalyzable Id. This Id is the impulsion of
all
impulses, the force that drives man into life and action, the original
and ineradicable craving for a
fuller and happier existence. It works as long as man lives and stops
only with the extinction of life.
Human reason serves this vital impulse. Reason’s biological function is
to preserve and to promote
life and to postpone its extinction as long as possible. Thinking and
acting are not contrary to nature;
they are, rather, the foremost features of man’s nature. The most
appropriate description of man as
differentiated from nonhuman beings is: a being purposively struggling
against the forces adverse to
his life.This PCTish paragraph, by the way, illustrates a peculiar sort
of expository style which appears often in economics and philosophy. It
consists of making a series of statements of fact without any
support or proof, as if there could be no doubt that they are true. Of
course every assertion above, and some sentences contain two or three of
them, could be false, in which case there would be no reason to read
further. After each assertion, one could be justified in asking “How
do you know that?” and not allowing the exposition to go on until
the question is answered, or at least until some plausible basis for the
assertion is indicated. How do you know there is an Id?
The primary effect of this style, it seems to me, is the early
elimination of dissidents, leaving only those who already believe these
assertions to accompany the author the rest of the way.
Mises considers “thinking and acting” to be “the foremost
features of man’s nature.” But he doesn’t say thinking and acting
about what. If one breaks thinking down into more detailed types
of thinking, as I have attempted to do with my proposed levels of
perception, it becomes clear that traditional logical entrepreneurial
thinking, “bottom-line” thinking, is not the only kind there
is, or even the most important kind. It’s not enough to explain, Donald
Trump-like, some despicable action by saying “That’s just good
business.” One must also explain what’s good about it, and in
explaining that, one can’t help but reveal the principles and system
concepts that motivate the specific strategies.
But I agree strongly with one of Mises’ tenets.
I
t is customary to find fault with
modern science because it abstains from expressing judgments of
value. Living and acting man, we are told, has no use for Wertfreiheit;
he needs to know what he
should aim at. If science does not answer this question, it is sterile.
However, the objection is
unfounded. Science does not value, but it provides acting man with all
the information he may need
with regard to his valuations. It keeps silence only when the question is
raised whether life itself is
worth living.**If science provides acting man (ideally) with all the
information he may need with regard to which acts will achieve the ends
he freely values, the Austrian rejection of observation ceases to make
sense. How can you determine what economic effects a given action has, or
is likely to have, without making observations? Pure reason is simply not
enough. To try to understand economic systems without using careful,
verified, and replicable observations, organized by a working and
verifiable model, is to construct a systematic delusion. Of course
economists think that all economic theories (but one) are systematic
delusions, which means that all economic theories are considered
by the great majority of economists to be systematic delusions.

The start of an economic model that I presented a few years ago was
nothing more than an attempt to look at the interacting effects of
actions in an economic system. The problem with all attempts to do this
that I had read about before, from Keynes to Samuelson, is that they
tried to derive the behavior of the system by looking at just a few parts
of it in isolation. The idea of supply and demand, for example, is
handled in a way that leaves reference conditions out of the picture;
there is no concept of “enough.” And other simultaneous
influences on buying and selling are ignored while talking about supply
and demand.

Of course there is a relationship between supply and demand, and another
between demand and supply, but it is not the only relationship that
exists between buying and selling. The economic system is full of closed
loops of causation, and there is no way to understand it or predict its
behavior correctly without understanding the laws governing interacting
closed loops. The only people I have seen who manage to do this
correctly, using models, are those who study “agent-based”
economics (Charlotte Bruun is one of them). And even they try to use
short-cuts based on abstractions like supply and demand without
understanding how they emerge from the more basic model.

My model was certainly not complete, but it could have been made a lot
more complete. Unfortunately, the person who could have helped make it
more complete couldn’t bring himself to do it.

It has been extremely
frustrating for me in trying to talk about economics in this forum as it
applies to PCT as it was for Bill Williams.

Bill Williams, who was a close friend of Mary’s and mine for over 20
years, suffered a radical change in his last few years, which may perhaps
be understandable now. He was never hesitant to apply his acid wit to
ideas he disagreed with, including mine, and for most of our acquaintance
he could take reasonable criticisms of his ideas as well as dish them
out. Toward the end, however, all that changed. He became offended at the
idea of people without degrees in economics proposing unorthodox (or any
other) views (though he had little good to say about orthodoxy). His
ideas of modeling changed from his practical approach toward the Giffen
Paradox to something more impressionistic and much less organized. When I
did not accept his new approach, I very quickly turned into his
arch-enemy. I was angry and upset about that at the time, but now, in the
light of subsquent events, I am only sad.

My suggestion is for them to
read a bit from the Austrian School and Ludwig Von Mises. Again, as you
note, it is extremely important that they stay away from that
“macro” mush

The quotations from Mises above were taken from some exerpts I made in
2001. Unfortunately I didn’t write down the title of the book, but it’s
probably one of the main ones you get to through his web page. As to the
Austrian school, I see no reason to accept their pessimistic views about
scientific observation in economics – it sounds more like an excuse than
a view, to me. As to their claim that human behavior is just too complex
to study scientifically, all I can say is that it may be that they
don’t know how to do it, but that does not mean that nobody can do it. I
think I’ve made a pretty good start on it, and others have, too. The
Austrian rejection of science sounds like sour grapes to me. As I’ve said
before, to me, science just means studying things honestly and
systematically and trying not to let your own desires and beliefs distort
the results. To reject science is to embrace ignorance, prejudice, and
superstition. Can they really be proposing that we do that?

As to that “macro mush,” there is no way to avoid it. Even the
law of supply and demand, or any other regularity that is seen in
economic interactions, is macro mush. A truly micro account of economics
would be as meaningless as a half-tone photograph viewed under a
microscope. Any time you’re talking about more than one person, more than
one producer, you’re in the realm of macroeconomics. It looks like mush
only when you don’t know how to deal with complex systems, but there is
no excuse for that any more.

To the believer in molecular interactions, thermodynamics is macro mush.
But thermodynamics is a lot more powerful as a way of predicting the
behavior of large assemblages of molecules than is any analysis of
molecular dynamics. Macroeconomics is far more powerful than
microeconomics in revealing the consequences of system-wide policies or
properties, and it’s the only way to unravel the effects of multiple
closed loops acting at the same time. I think that a lot of objections to
macroeconomics arise from the fact that a macro analysis leads to
conclusions that are contrary to popular beliefs and superstitions, such
as – well, I’m not here to throw gasoline on the fire. Suffice it to say
that there is no way to get a correct picture of economic interactions by
looking at the whole system one entity at a time. But it is very easy to
develop wrong ideas that way.

I think there is a much bigger
problem that both Rick and Bill face and that is the complete
revision of their current mind sets about what economics actually is and
what it describes.s revision will create utter havoc with their current
world views and I’m not sure either one of them is up to
that.

Thus spake Zarathustra. My view is that modern economics is a total mess
and needs a complete revision of world-views, which are mostly so
transparently self-serving as to be laughable, not just wrong.

I’m afraid both will fight tooth
and nail against what I would think could be the only reasonable outcome
given their beliefs about perceptual control.

It’s really time to stop alluding indirectly to things and start saying
what you mean. You refuse to describe what you refer to as your research
because you anticipate criticism and stubborn refusal on the part of
others to admit the truths you have discovered. Anyone who expects
scientific work to be accepted sight unseen as valid just doesn’t
understand how the world works. If I had taken that attitude, my name
would not be associated with PCT (though someone else’s might be). If you
continue that way, your work will be ignored, and why not?

Just what is that “only reasonable outcome?” If it were truly
that, wouldn’t everyone already agree with it? Or is your definition of
being unreasonable the refusal to agree with you? Why not just
throw it out on the table for discussion?

Bill P.

[From Bill Powers (2006.08.14.1410 MDT)]

Richard Kennaway (2006.08.14.1059 BST) –

Bill, with all respect to you
and your late father, have you ever read any real economics? Not
the macro-scale theological stuff that has been the subject of flame-wars
here in the past, and in which no individual people seem to play any
role, but the economists who study people as purposive entities who do
whatever it takes, given the actions of the other purposive entities
around them, to achieve their purposes.

My dad started reading economics when he retired in 1965. He published
his book in 1996, 31 years later. He probably read as much economics as
the average PhD in that field does, at least the major works as opposed
you journal articles. He did a LOT more exploring of the Statistical
Abstracts than most economists take the time to do.

But of course in some ways he believed much as traditional economists
did. In particular, when I suggested that what people wanted by way of
goods and services probably should be included in the concept of
“demand” he said that individual preferences have nothing to do
with economics.

The idea that economic properties emerge from the properties of
individuals has been very much a minority view. Mises certainly upheld
that view, and it seems that the Austrian ecomomists do, too. I don’t
think they represent a majority.

More at David Friedman’s web
site (for it is he). Many of his writings, including two whole
books, are online at

http://www.daviddfriedman.com/Academic/Academic.html.
(I have no
reason to think that he has any particular familiarity with control
theory of any variety; or, for that matter, that he does
not.)

I tried to get him interested a few years ago but he never answered my
posts. He probably gets a lot of nut mail, and it’s hard to say what I
have to say, cold, without sounding like a nut.

As for competition, your
characterisation above only applies to one-level control systems which
have no ability to change their references. Conflict won by the
stronger is the only possible interaction they can have.

I think the usual reaction to a conflict is to try to win it, which of
course makes it worse. Just having higher levels is no help if they are
not organized appropriately.

Businesses have learned some lessons – I’m thinking of the gas wars, in
which oil companies (or their gas-station operators) discovered that
price competition can be ruinous. Now they emphasize brand loyalty and
supposed superiority of their additives and services, or their
convenience stores or car-washes. In Durango, the price of regular ranges
from $3.06 per gallon to $3.36, with Sammy’s Peerless Tyres gas the
cheapest and most crowded – but the others are still comfortably in
business. You never see an ad saying “Buy our gasoline – it’s
cheaper!” That’s probably the higher-level adjustment you’re
thinking of – it’s a way to avoid direct competition based on
price.

Last year in Colorado, there were (if I remember right) 53,000 business
failures. About 3.5 million people live in Colorado, as I recall. I’d
sure that competition, or failure to meet it, had something to do with
some of the failures. But what most companies do is shift their
advertising to avoid direct competition – try to offer something the
other guy doesn’t have, find a niche as free of conflict as possible.
Make customers try to decide between apples and oranges, not 10 cent
apples and 11-cent apples.

Hierarchical systems (assuming
that the model outlined in B:CP is more or less the right account of what
people are) exhibit enormous flexibility in maintaining their topmost
references, because they will vary their lower level references to
achieve them. I suspect that competition between such systems will
– well, someone model it and make a prediction.

My prediction is that reorganization will tend to eliminate conflict
(i.e., competition). If you list all the ways there are to eliminate
competition, you will probably be able to find examples of each way in
real life (for example, assassination, bribery, circulating rumors and
lies, turning out out a clearly superior product, giving the false
impression
of turning out a superior product, price-cutting, adding
variety, etc.).
A question in return: have you seen or played with my Econ005 program
that simulates a simple economic system? Not very much theology about it,
although it is a macroeconomic model.
You mention “the economists who study people as purposive entities
who do whatever it takes, given the actions of the other purposive
entities around them, to achieve their purposes.” (That sounds a bit
like The Mob, doesn’t it?) You might classify some economists that way,
but aren’t you overgeneralizing a bit? I doubt that they put it exactly
that way. If they truly saw human nature in that way, wouldn’t they be
including such things as crime and terrorism in their studies? I know
that Friedman does, but isn’t he sort of unusual?
Anyway, a model that can apply to people in the way you describe
is a macroeconomic model. It does not deal with Individual
A, then with Individual B, and so on. You can’t get any kind of economics
out of that. You have to have interacting individuals, and even then not
just two of them. You have to include examples of different ways of
interacting in different economic roles, and you have to allow for a
range of degrees of interaction over many people (if you want realism).
That’s macroeconomics. I don’t see what’s so “theological”
about that. Maybe you’re thinking of some particular example of
macroeconomics. Take a look at my model and tell me what you think is
theological about it.

Best,

Bill P.

···

[Martin Taylor 2006.08.14.16.25]

[From Bill Powers (2006.08.14.1320 MDT)] to

Richard Kennaway (2006.08.14.1257 BST) --

Your counterarguments have all been to the effect of "But freedom is a better way." Have I ever said it isn't? Of course it is; less conflict is always better than more conflict if you know how to achieve that.

I think that's an unproven derivation (if not an axiom). We all have our axioms, and some of this discussion (along with Bill's lovely posting on the Austrian school of economics [From Bill Powers (2006.08.14.0830 KMDT)], makes that difference obvious.

My own basic axiom is that "better" for some set of control systems means "less error" in the aggregate.

Now, that axiom leads to a whole lot of questions about how should "error" be aggregated across control systems. Is it better for a few systems to have huge error in order that many could have small error, or is it better to add up "error" linearly in some mathematically dimensionless way? That question leads to the next question, of how to provide a non-dimensional measure of error, because intrinsically, the error has the dimension of the perceptual and reference signals in its control system. They might in one case be a chemical concentration, in another case a neural firing rate, and perhaps yet other things in other control systems. And when one has that dimensional question resolved, one has a scaling problem. is an error of "2" in a control system with a high gain more or less important than an error of "20" in a system with low gain? Is it "better" that a control system be induced to go to zero gain because it's no use butting your head against a brick wall? Would the error in such a control system count for nothing?

If there's any validity to PCT at all, this issue is one that has affected all of evolution, and if we could assess how biological systems balance the errors involved in the inevitable internal conflicts, we might get some sort of handle on how to mathematize the notion of "better" and "worse" when looking at the effects of dfferent economic systems on large numbers of people. Failing that, I fear we are doomed to arguments about "freedom" being better than ... than what, exactly? I believe that a lot of East Germans think that life was much better under the Commissars than subsequently. They could live mderately comfortably and securely then, which isn't true for a lot of them now. Perhaps they had more freedom before the Wall came down?

Which "freedom" are we talking about? The freedom to express opinions without fear of the Secret Police, the freedom to be able to go to the grocery and buy food, or the freedom to screw your business associates and customers without fear of the government? It's nice to have all of these, but are they even compatible? When faced with choices, some people have had the opportunity to ask themselves which is better. Different societies have operated using different axioms about which is better, and some people have done very well in all of those societies.

Look, for example, at Cuba, which is held up as an example of a country with little freedom. But Cuba happens to be the country in the world that has the very best public health relative to its per capita income (CIA World Factbook -- the US is one of the worst, apart from the central Asian "xistans" from the Soviet Union), and many Cuban people seem to be pretty well educated and tolerably happy. Isn't the fact of being in good health one influence on your abiity to control? Isn't that a kind of freedom that is important? Maybe Cubans aren't so lacking in freedom after all, if you define it in ways beyond being able to choose the faces in the government and the policies they follow. I wouldn't like to judge, however, which kind of axioms about "better" intrinsically leads to overall "better" conditions (ability of all control systems in all people to control well), or indeed whether conflict reduction always is better than more conflict.

If you take it as axiomatic that less conflict is better than more conflict, you then have to define "better" independently of that axiom. If you don't, you are just appealing to what you used to call a "dormitive principle." Likewise, Richard has to define "better" independently of "freedom" -- and he ought to define "freedom", too, because a simple naive definition that allows everyone to do whatever they please soon runs into the degrees of freedom limitation (no pun intended). More constraint can lead to more freedom, by -- in this case -- reducing conflict.

Do I have more freedom when I am coerced into driving on the right (for Richard, the left) side of the road and I know that everyone else is similarly coerced, or when I have to judge at every moment which side of the road to use, depending on where the cars coming at me happen to be at that moment? I personally think more freedom goes hand in hand with effective regulation and appropriate government interference, but that's just my intuitive (non-mathematical, non-experimental) assessment of the consequence of MY axiom, that "better" means "overall, more ability for more people to control more perceptions."

But that is seldom the choice that people face. They start out in a terrible mess, and someone steps in and makes it a little less terrible. There may be many ways to make it less terrible that permit more freedom and even better results, but that is completely irrelevant to the question of whether an improvement resulted. You (and others) seem to think that by opening the door even a crack to see that bad methods can have good results I am flinging it wide and letting the enemy in.

The mathematics is not complicated. It's true that 20 is greater than 10, but does that mean that 11 is not greater than 10? You seem to be saying it does.

Oh, I think the mathematics is _very_ complicated! And you can't even start on it until you define your terms, and in my language, your axioms.

Martin

[From Bill Powers (2006.08.14.1650 MDT)]

Martin Taylor 2006.08.14.16.25 --

Now, that axiom leads to a whole lot of questions about how should "error" be aggregated across control systems. Is it better for a few systems to have huge error in order that many could have small error, or is it better to add up "error" linearly in some mathematically dimensionless way?

I think you answered this question some time ago in a different context. The "objective" meaning of error must be defined by the reorganizing system -- that is, the set of all intrinsic reference signals. All the others derive from that, don't they? And the relative importance of intrinsic errors must relate in some way we could (in principle) determine from their contributions to "fitness," a word I don't really believe in but something like that.

That question leads to the next question, of how to provide a non-dimensional measure of error, because intrinsically, the error has the dimension of the perceptual and reference signals in its control system.

For purposes of determining tightness of control, error can be expressed as a fraction of the reference signal. That gets rid of the units. We could also speak of the relative "importance" of errors in terms of the amount of effort that is produced per unit of error, or alternatively (and inversely) the amount or error that is required to cause the largest possible amount of output effort. Instead of just looking at the error we could look at (error)*(output gain).

They might in one case be a chemical concentration, in another case a neural firing rate, and perhaps yet other things in other control systems. And when one has that dimensional question resolved, one has a scaling problem. is an error of "2" in a control system with a high gain more or less important than an error of "20" in a system with low gain? Is it "better" that a control system be induced to go to zero gain because it's no use butting your head against a brick wall? Would the error in such a control system count for nothing?

Again, I think this is answered by referring everything to intrinsic error.

Which "freedom" are we talking about? The freedom to express opinions without fear of the Secret Police, the freedom to be able to go to the grocery and buy food, or the freedom to screw your business associates and customers without fear of the government? It's nice to have all of these, but are they even compatible? When faced with choices, some people have had the opportunity to ask themselves which is better. Different societies have operated using different axioms about which is better, and some people have done very well in all of those societies.

Yes, and there's also the matter of levels of freedom. I am free to turn the steering wheel of a car in any way I please -- unless I also want to keep the car on the road in a crosswind. By choosing some goals, we limit our own freedom to set others.

If you take it as axiomatic that less conflict is better than more conflict, you then have to define "better" independently of that axiom.

Less conflict --> better control --> smaller intrinsic error ....?

The mathematics is not complicated. It's true that 20 is greater than 10, but does that mean that 11 is not greater than 10? You seem to be saying it does.

Oh, I think the mathematics is _very_ complicated! And you can't even start on it until you define your terms, and in my language, your axioms.

Shucks, and I thought you couldn't complicate it. You make it look easy.

Best,

Bill P.

[Martin Taylor 2006/08/14/22.58]

[From Bill Powers (2006.08.14.1650 MDT)]

Martin Taylor 2006.08.14.16.25 --

If you take it as axiomatic that less conflict is better than more conflict, you then have to define "better" independently of that axiom.

Less conflict --> better control --> smaller intrinsic error ....?

Actually, it was this set of assumptions that led me to make my post -- or rather, it was the unanalyzed acceptance of them.

Conflict leads to explosion, for sure, in a linear system. But real control systems aren't linear. They may have thresholds, and they certainly have saturation levels. If you believe in the hierarchy as usually described, you CANNOT have linear perceptual funtions in too much of it, and certainly not in successive connected levels. (The argument is that successive linear levels are simple rotations of the incoming data, which buys nothing).

If the systems are non-linear, which they must be in many of the perceptual input functions, probably are in the error functions, and must be in the output functions, then increased conflict may not lead to worse control; at least that is if conflict is defined in terms to the degrees of freedom bottleneck. The degrees of freedom bottleneck exists when more degrees of freedom are implied by the reference signals for the involved control systems than are available through the control loops.

If the error functions have thresholds, the effective error can be zero in each of two conflicted control systems. The perceptual signals may deviate from the reference signals, but a sub-threshold deviation results in no control action. There's conflict, as one would see if both systems were disturbed to a supra-threshold level, and because of the lack of degrees of freedom through the two control loops, that conflict is always there. But it doesn't necessarily affect the ability to control.

Degrees of freedom is quite a subtle concept. It should always include the temporal variable. A waveform has so many df per second, depending on its bandwidth (the Nyquist criterion). Those have to be taken into account, which is why two systems can control two variables through one controller by time-multiplexing, if each controls sufficiently slowly against sufficiently slow disturbances. Again, less conflict need not mean better control.

Backing off, I will accept your modification of my axiom so that it refers to intrinsic error, though that's (to me) a more nebulous concept, and less quantifiable, than the control error. I think, though, that the same dimensional an scaleability arguments apply, arguments I think your commens did not adequately address. Nevertheless, it is presumably the aggregation of intrinsic error that has influenced the course of evolution, rather than the aggregation of control error, so I accept the change.

That question leads to the next question, of how to provide a non-dimensional measure of error, because intrinsically, the error has the dimension of the perceptual and reference signals in its control system.

For purposes of determining tightness of control, error can be expressed as a fraction of the reference signal. That gets rid of the units.

That really doesn't help, does it? Suppose the reference signal is at a zero-crossing of its waveform, but the perception is non-zero. Clearly that doesn't resolve the scale problem, and neither would an appeal to the historical range of the reference waveform. I'm not sure whether it really avoids the units issue, though I have a conceptual problem in properly defining that issue. In my earlier message, I cast it in terms of whether the mechanism was chemical concentration of neural firing rate, but that's mechanism, which really isn't the issue.

In my mental quandary, it was an apples and oranges question: how does an error in my perception of injustice aggregate with an error in my perception of the colour of the paint on my wall? That's comletely external to any of the control systems concerned, none of whaich know what they are controlling other than a magnitude, but somehow it seems to matter. It's still a conceptual quagmire for me. Maybe it will become less muddy after a night's sleep.

We could also speak of the relative "importance" of errors in terms of the amount of effort that is produced per unit of error, or alternatively (and inversely) the amount or error that is required to cause the largest possible amount of output effort. Instead of just looking at the error we could look at (error)*(output gain).

Yes, I vaguely conceived of something like this, and it would allow comparison and aggregation of different kinds of perceptual control, like injustice and paint colour -- how much effort would you put (or are you putting) into conrolling those variables. But a problem arises here, too. To some extent the total effort is likely to be more constant than the effort put into individual controls, and to depend on your calorific intake and your metabolism. A starving person, or an unhealthy one, may have large errors (both control and intrinsic) that they would very much like to reduce, but against which they are powerless (in the mechanical sense) to act. When aggregating, do those errors count for less than the errors being effectively reduced by a rich, healthy person? If that's your model for aggregation, you probably will arrive at the Libertarian social system as being pretty good.

It comes down again to your axioms -- what you take to be true without argument or analysis.

I think you answered this question some time ago in a different context. The "objective" meaning of error must be defined by the reorganizing system -- that is, the set of all intrinsic reference signals. All the others derive from that, don't they?

Having got back to the start of your message, I now ask what are the intrinsic reference signals that are the basis of the reorganizing system for the set of control systems that together form a society of millions of people? Don't we get back to the individuals again, and aren't we back to square one in asking what it means to have a "better" society? It comes down to what we take as axiomatic. Richard takes as axiomatic that any government interference is worse than none. I take as axiomatic that more ability of more people to control more variables is better than less, and work from there to try to figure out what kind of society maximizes that multi-dimensional criterion. You seem to take as axiomatic that minimizing conflict is best.

It's conceivable that all three sets of axioms could result in the same optimal structure for society, but it's also conceivable that all three would lead to different concepts of "best". Apart from that, we are left with personal prejudice based on our upbringing and experience. And there we have no sound basis for argument -- at least not a scientific argument.

Martin

Re: recent Chinese military
history
[Martin Taylor 2006.08.15.15.34]

[From Bill Powers (2006.08.15.0720
MDT)]

This is really it for today – only 14 days until the big move.

Martin Taylor 2006/08/14/22.58 00

Conflict leads to explosion, for sure, in
a linear system. But real control systems aren’t linear. They may have
thresholds, and they certainly have saturation levels. If you believe
in the hierarchy as usually described, you CANNOT have linear
perceptual funtions in too much of it, and certainly not in successive
connected levels. (The argument is that successive linear levels are
simple rotations of the incoming data, which buys nothing).

I think that “linearity” applies only to a few of the lowest
levels – even object recognition probably is inherently nonlinear.
And its probably not true of any level when you get right down
to it. Look at the psychophysical data.

My point there is tha it can’t be true at ANY level in a
perceptron-like structure, which is the structure of the perceptual
side of the “classic” HPCT hierarchy.

Aside, for those who aren’t following this because they don’t
know the meaning of “non-linear”: A function is linear if it
has the form

y = a0 + a1x1 + a2x2 …

and nonlinear if any of the “xn” elements has a power
or some nonlinear function applied to it.

z = ax^2 + by^2 is the equation for an ellipse (a family of
elipses, actually). It is non-linear.

Often when we talk about the hierarchy, and especially in many
simple models, all the perceptual functions are linear, meaning that
if the inputs from level n are pn1, pn2, … the output perception
P(n+1)1 = a0 + a1pn1 + a2pn2 …

If there are K perceptual signals at level n+1, based on K inputs
from level n, all such a set of linear functions accomplishes is a
rotation and scaling of the input space. The control that can be
performed is identical at level n and at level n+1. The system has no
need for level n+1 to exist.

If, on the other hand, the perceptual functions at level n+1 are
nonlinear (say P(n+1)1 = a0 + a1log(pn1) + a2log(pn2) … then what is
controlled at level n+1 is different from what is controlled at level
n. In the example, level n+1 would be controlling something like the
product of the inputs from level n. The effect of a given change in
pn1 would be affected by the value of pn2, which is not the case in a
linear system.

Some operators other than addition and multiplication by a
constant are also linear. Integration and differentiation are linear
operators. If a control loop consists of a linear perceptual function
as described above, has a linear error fnction e = r - p, and an
integrating output function, and the effectthrough the environment is
also linear, it’s a linear control system. If any of those functions
are nonlinear (such as

e = 0 if |r-p| < epsilon

e = r-p otherwise

then the control system is nonlinear.

I’m arguing that the hierarchy of perceptual functions MUST be
nonlinear if it is structured as classically described in HPCT. I’m
also saying that individual control loops might (and I think often do)
avoid the loss-of-control problems of conflict by certain
nonlinearities. Colloquially, this is called “don’t sweat the
samll stuff”.

The degrees of freedom bottleneck exists
when more degrees of freedom are implied by the reference signals for
the involved control systems than are available through the control
loops.

Did that sentence actually say what you meant? Reference signals, as I
conceive of them, are one-dimensional, and so are perceptual signals.
Maybe what you were referring to were the input functions, which
receive multiple signals from systems below, thus “implying”
more degrees of freedom.

No, I was referring to the number of controlled perceptions,
which equals the number of reference signals.

If the error functions have thresholds,
the effective error can be zero in each of two conflicted control
systems. The perceptual signals may deviate from the reference
signals, but a sub-threshold deviation results in no control action.
There’s conflict, as one would see if both systems were disturbed to a
supra-threshold level, and because of the lack of degrees of freedom
through the two control loops, that conflict is always there. But it
doesn’t necessarily affect the ability to control.

I think this is making the idea of conflict too confusing. The problem
here is that when there are multiple real control systems operating in
a real common environment, there are always some residual
interactions. Should we say that all behavior, therefore, is
conflicted? I don’t think so.

As I try to use the term, conflict implies severe loss of
control.

OK. That’s a functionally different definition of conflict from
what I have always used, and from what has usually been taken as
conflict in discussions on CSGnet. Loss of control is a consequence of
conflict, but it isn’t itself conflict. That’s true of conflicted
linear systems, but not necessarily of conflicted non-linear systems
according to what I thought was the conventional definition of
conflict:

“Conflict occurs when N unidimesional control systems have
available among them fewer than N degrees of freedom though which to
exercise control.”

That definition “implies severe loss of control” under
many circumstances, not only with linear systems. However, as in my
example, some non-linear systems can be conflicted with only mild loss
of control (implied, for example, by the threshold in the error
function).

In my mental quandary, it was an apples
and oranges question: how does an error in my perception of injustice
aggregate with an error in my perception of the colour of the paint on
my wall? That’s comletely external to any of the control systems
concerned, none of whaich know what they are controlling other than a
magnitude, but somehow it seems to matter. It’s still a conceptual
quagmire for me. Maybe it will become less muddy after a night’s
sleep.

You said it yourself: “aggregation” is completely external
to the control system. It’s in the analyst, not in the system. I see
no reason to assume that there is any kind if aggregation of error in
the system as a whole. The errors simply exist, and there is a huge
number of states of error that are completely equivalent by any
aggregate measure you can think of.

All of that is true, but doesn’t address the point that it is the
analyst who is trying to assign an ordering of “better” to
“worse” among social structures that have different amounts
and kinds of regulatory constraintson individual abilities to
control.

It comes down again to your axioms –
what you take to be true without argument or analysis.

Right, and that is how we can tell whether we’re talking about the
system itself or the analyst. When we’re talking about the system,
there are data and functions and such things. When we’re talkiing
about the analyst, we’re worried about truth, axioms, proofs, and so
on.

Yep. The analyst has to do the analysis of the system. Let’s
hypothesize an omniscient analyst, who can perfectly determine exactly
how the system will behave. The problem still exists as to whether one
system is better than another, and that, I believe, has to depend on
the analyst’s axioms.

I don’t like optional axioms. They ought
to be forced on us, meaning that we can’t argue against them because
there is nothing to suggest they are false.

That assumption (axiom?) held back mathematics for centuries
after Euclid. Euclid’s axioms had that aura of “being forced on
us”, until Reimann and Lobachewsky showed the world that they
weren’t and that self-consistent mathematics was both possible and
useful by choosing different ones.

When I try to justify my axiom
about minimizing conflict, I do it partly by trying to define conflict
in such a way that we can identify it unequivocally. I think I’ve done
that by tying it directly to loss of control and putting it on a
continuum of interactions among systems.

I do, from time to time, come back to the continuum of
interaction among systems. One of the subtle issues of degrees of
freedom is that they aren’t always integer-valued, which happens when
waveforms don’t have purely rectangular spectral shapes. That makes
the definition above for conflict in terms of degrees of freedom shade
into the question of interaction strength, which is assocated with the
ease with which multiple control systems can control, rathr than their
ultimate abiity to control. But I don’t think here and now is the
place to pursue that.

… And finally, losing accuracy of
replication is not axiomatically bad, either: what’s bad is ceasing to
exist. And of course that matters only to the system in question, and
when it ceases to exist, that doesn’t matter, either. Poof! No more
axioms.

True, but PCT is the study of LIVING organisms, for which Proof
is preferred to Poof.

Martin

Re: recent Chinese military
history
[Martin Taylor 2006.08.15.19.45]

From [Marc
Abrams (2006.08.15.1222)]

But
does this notion of degrees of fredom actually apply to living
entities and control systems.

Yes.

Martin,
what warrants do you have that lead you to believe that this applies
equally to living and non-living entities.

Only that I believe living things
to exist in a physical world and to be subject to the normal laws of
physics. I know it’s just an item of faith, but it’s a faith to
which I adhere.

We had this discussion a couple of
years ago. I’m afraid I still don’t believe living things work by some
magic inaccessible to normal science. If you have evidence to the
contrary, my faith might be shaken, but in the absence of a workable
description of the specific violations of physics inherent in
living things, I’m afraid it wouldn’t be stirred.

Martin

[From Richard Kennaway (2006.08.16.2346 BST)]

[From Bill Powers (2006.08.14.1320 MDT)]
Come now, Richard, you know that my view is not that simple. What do you mean by "work"? I think it's perfectly obvious that sometimes when some people are told what to do, in fact forced to do it, the result leaves the majority better off than they were before. This has nothing to do with whether the use of force is better than some other approach. Of course it's not. But that's not what I've been talking about.

It might be obvious with an example. The general hypothetical doesn't really get anywhere. Sure, making people (not merely telling them) to do stuff can get things done, primarily when the people made to do what other people want them to do are children, the mentally deficient, in prison, or in the army. Making everyone in an entire society do their supposed duty (except, of course, for the ones giving the orders) invariably fails. I point again at the great dictatorships of the 20th century. What did they achieve for their people? Here is what North Korea is currently achieving:

"North Korea, one of the world's most centrally planned and isolated economies, faces desperate economic conditions. Industrial capital stock is nearly beyond repair as a result of years of underinvestment and shortages of spare parts. Industrial and power output have declined in parallel. Despite an increased harvest in 2005 because of more stable weather conditions, fertilizer assistance from South Korea, and an extraordinary mobilization of the population to help with agricultural production, the nation has suffered its 11th year of food shortages because of on-going systemic problems, including a lack of arable land, collective farming practices, and chronic shortages of tractors and fuel."

Source: CIA World Fact Book, https://www.cia.gov/cia/publications/factbook/

It is not enough to say that since Stalin and the rest got into power and were not removed by their own people, that most people must have preferred what they got to what they had. They may have thought they would before the fact, just as the dictators may have thought they would give it to them. But it didn't happen. It never happens. Otherwise, Stalin, Mao, and Pol Pot would not have had to make their countries prisons.

Your counterarguments have all been to the effect of "But freedom is a better way."

That is the conclusion of the argument. Better not in a moral sense (I'd say it is, but that's not a subject admitting of constructive argument, and you agree with it anyway), but better in the sense that more people get more of what they want. One of the things they want, of course, is more freedom, but the freer societies are also more materially prosperous. Which countries in the world support shops all across the land offering eighty kinds of bread to the general public, all of it good? Which countries have the latest mobile phones and laptops? Which countries censor the Internet? Which countries shoot citizens who try to escape, and which countries have a problem keeping out the people who would love to go there?

If one sees a signpost saying "London" planted on the coast at Dover and pointing out into the English Channel, one does not have to make a great cartographic survey to justify wrenching it out of the ground.

So, if you want a scholarly treatment of the relationship between freedom and well-being, that has probably been done somewhere, and at any rate I'm not in a position to undertake it. But supposing the testament of the daily news over either of our lifetimes insufficient to form even a first impression (which I think it is not), here is one indicator.

From the CIA World Fact Book again, net migration (positive = inwards) per 1000 population:

USA 3.18
UK 2.18
China -0.39
North Korea 0
South Korea 0.42

The video I linked to had other examples.

-- Richard

···

--
Richard Kennaway, jrk@cmp.uea.ac.uk, Richard Kennaway
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.

[From Richard Kennaway (2006.08.16.2347 BST)]

[From Bill Powers (2006.08.14.1410 MDT)]
I tried to get him interested a few years ago but he never answered my posts. He probably gets a lot of nut mail, and it's hard to say what I have to say, cold, without sounding like a nut.

Presumably you also read him? What did you think of his ideas then?

Businesses have learned some lessons -- I'm thinking of the gas wars, in which oil companies (or their gas-station operators) discovered that price competition can be ruinous. Now they emphasize brand loyalty and supposed superiority of their additives and services, or their convenience stores or car-washes. In Durango, the price of regular ranges from $3.06 per gallon to $3.36, with Sammy's Peerless Tyres gas the cheapest and most crowded -- but the others are still comfortably in business. You never see an ad saying "Buy our gasoline -- it's cheaper!" That's probably the higher-level adjustment you're thinking of -- it's a way to avoid direct competition based on price.

Well, there you are. Everyone freely adjusting their business plans to find a place in the market. That is the real "competition".

Hierarchical systems (assuming that the model outlined in B:CP is more or less the right account of what people are) exhibit enormous flexibility in maintaining their topmost references, because they will vary their lower level references to achieve them. I suspect that competition between such systems will -- well, someone model it and make a prediction.

My prediction is that reorganization will tend to eliminate conflict (i.e., competition).

That is my expectation too. The free system works! When people find that something they are doing does not work, they try something else. People being what they are, some may pigheadedly refuse to do so: they fail. "Competition" is not what makes them fail. Their failure to think outside "competing" with everyone who is selling exactly what they are selling makes them fail.

This is known in business circles. The first rule of starting a new business is "Don't compete with success", otherwise expressed as "Have a Unique Selling Point". That USP is the first thing a venture capitalist will want to know about you.

A question in return: have you seen or played with my Econ005 program that simulates a simple economic system? Not very much theology about it, although it is a macroeconomic model.

I passed over all the economic stuff that went around last time, but I've dug out the attachments and I've now run the program. I have to say that the controllers control for rather strange things. Why do they have references for stock and cash levels? I don't think I do.

You mention "the economists who study people as purposive entities who do whatever it takes, given the actions of the other purposive entities around them, to achieve their purposes." (That sounds a bit like The Mob, doesn't it?)

I think it sounds like people with purposes. The Mob are only distinguished by the means they are willing and able to use.

You might classify some economists that way, but aren't you overgeneralizing a bit? I doubt that they put it exactly that way. If they truly saw human nature in that way, wouldn't they be including such things as crime and terrorism in their studies? I know that Friedman does, but isn't he sort of unusual?

Some others do. But yes, he is unusual. So to understand David Friedman, there is no substitute for reading him specifically, rather than anyone else his name may be mentioned with.

For exaple, as a corrective to the idea that he, and I, are raving fundamentalists to whom the slightest whiff of Gummint is the touch of Satan, he writes on problems of libertarianism and his answers in chapters 41 and 42 of his book, "The Machinery of Freedom". Those chapters and some others are on the web at http://www.daviddfriedman.com/Libertarian/Machinery_of_Freedom

Anyway, a model that can apply to people in the way you describe is a macroeconomic model. It does not deal with Individual A, then with Individual B, and so on. You can't get any kind of economics out of that. You have to have interacting individuals, and even then not just two of them. You have to include examples of different ways of interacting in different economic roles, and you have to allow for a range of degrees of interaction over many people (if you want realism). That's macroeconomics. I don't see what's so "theological" about that. Maybe you're thinking of some particular example of macroeconomics.

By the "theological" stuff I mean forms of macroeconomics that make no reference at all to any individuals. Not having followed the previous economic discussions on CSGNET, I can't say if any of it fits that description. Statistics aren't a problem, they're only a problem when the mass entities are imagined to be themselves control systems with purposes of their own, when the only control systems and purposes are in individual people.

I interpret the controllers in Econ005 as being averages or sums of large numbers of controllers, rather than literally representing a real aggregate producer etc. that has reference levels independent from what any of the mere cells in its body might want.

The model does leave out variation between people, which can have major effects on the behaviour of a system.

-- Richard

···

--
Richard Kennaway, jrk@cmp.uea.ac.uk, Richard Kennaway
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.