Re: Theories, models, and simplifications.

From: John Edser (edser_at_tpg.com.au)
Date: 01/25/05 

Date: Tue, 25 Jan 2005 00:23:04 -0500 (EST)

This reply will be two sections
due to its size.

> > > > JE:-
> > > > __________________________________________
> > > > What was just the name of the rational
> > > > (not 100% relative) theory from which
> > > > Hamilton's (100% relative) model was
> > > > oversimplified?
> > > >
> > > > This question has to answered.
> > > > ________________________________________

> > > JM:-
> > > Clearly, the answer you are looking for is "Hamilton's
> > > model is an (over)simplification of Darwin's theory (as
> > > Darwin's theory has been interpreted by Edser)".

> > JE:-
> > Any answer I may be "looking for"
> > is simply not relevant. I just wish
> > to be provided with YOUR answer.

> JM:-
> And I HAVE provided it. Repeatedly. Most recently
> in this post. Hamilton is not a simplification of
> any theory. It is a complication of Fisher's model.

JE:-
You seems to think that the word "simplification"
is never used in conjunction with Hamilton's
Rule. Do I have to list all the authors from
the establishment that state that
Hamilton's model is a simplification?
Felsenstein called it a "toy". Now why was that?
Fisher's model itself employed simplifications
as you well know. Unless Fisher's simplifications
are removed then any more complex form of Fisher
still remains a simplified model.

Felsenstein, after being pushed by McGinn
listed 11 simplifications for HW.
Fisher's model was built from HW. Hamilton's
model is built from both with an added over
simplification: the deletion of cmax from
the rule only allowing a headless model.

> JM:-
> John, if I were to ask you why you have stopped beating
> your wife, I don't think that it would be fair for me
> to take your response "I'm not married". as an evasion.

JE:-
The question I asked is NOT a "beating your wife"
question. Everybody here admits that Hamilton's Rule
is a "simplification" but nobody here will say
what it is simplified from. This provides proof
of chronic denial.

> JM:-
> If a question is based on a misconception, it is not an
> evasion to point out the misconception.

JE:-
In the sciences you cannot dictate what is and what
is not a "misconception". For Newtonian mechanics
is was a misconception to suggest that the maximum
velocity light in a vacuum is a constant. If you
allow such things to be dictated then science
ceases and witch burning begins.

> > Like Felsenstein, you keep evading
> > this basic issue. It remains hopelessly
> > irrational to keep on writing that Hamilton's
> > model was and remains, a simplification/
> > oversimplification but refuse to
> > provide just the name of whatever-it-
> > -is that it was supposed to be a
> > simplification/ oversimplification
> > from.

> JM:-
> It doesn't simplify ANY theory. It contains
> simplifying assumptions regarding the SITUATION.

JE:-
"Situation"? Where and what is this "situation"?
Is it just pure fiction in somebodies head or
does it have some testable basis in reality?

> > > JM:-
> > > However, I cannot give this answer, because I use the terms
> > > "theory" and "model" in a different way than you do. (Also,
> > > since Edser's interpretation of Darwin postdates Hamilton,
> > > it would be historically incorrect to say that Hamilton
> > > (over)simplified Edser. If anything, it is the other way
> > > around.)
> > > Both terms have various meanings, but I understand that we
> > > are using "theory" in the sense meant by Popper. My
> > > understanding is that, for Popper, a "theory" is a candidate
> > > law of nature; it is a hypothesis, frequently stated as
> > > a mathematical equation, which says that the results of
> > > certain physical measurement operations will (always) have
> > > a certain definite relationship, in a certain defined class
> > > of situations.

> > JE:-
> > Models can only use two types of Lego
> > building blocks: variables and constants.

> JM:-
> Well, there are a few more kinds of building blocks
> than those two.

JE:-
Please provide just one example.

> JM:-
> But I'll accept your oversimplified
> model of modelling for the sake of argument.

JE:-
The above is not acceptable to myself.
Please provide at least one of the
missing alternatives

> > JE:-
> > A simplification can remove a variable
> > by artificially setting it to zero.

> JM:-
> Yes.

JE:-
Do you agree that any model that
simplifies a theory variable to zero
is a simplification of the theory
that created the variable?

Would you also agree that such a model
can represent a point of refutation for
the theory it was simplified from?

> > JE:-
> > It can also define that variable to
> > be any value where this defined
> > value may not even exist within nature.

> Conceivably. Though this would be a bad simplification
> except in the case of "limits". For example, assuming an
> infinite population as a limit might be a reasonable
> simplification even though there are no infinite populations
> in nature.

JE:-
That is not the issue here.
A model assumes a range for a variable that
remains outside of empirical evidence then
these values represent simplifications.
Do you agree or disagree?

> > JE:-
> > An oversimplification is the
> > removal of a constant. While
> > the removal of a variable
> > does not alter the logical form
> > the removal of a constant must.
> > An oversimplification entirely
> > changes any proposition.

> JM:-
> I get what you are saying. But I think that
> you are missing one posibility for a simplification
> that removes a constant in an quite innocuous
> way. The constant can simply cancel out (from
> both sides of an equation, for example).

JE:-
This process does NOT remove the constant from
the parent theory it just heuristically sets it
to zero. Zero is a measure of state and not the
absence of any state for the said constant.
Zero turns a constant a mathematical
phantom that is ready to haunt misused models
within the sciences, e.g. Hamilton's Rule. Of
course mathematics could not care a less because
it does seek to provide theories of nature.

Hamilton deleted cmax. This entirely changed
the mystery something from which the model was
over simplified. Do you still maintain that
Hamilton's headless model was beheaded from
just nothing at all, i.e. was Anne Bolin's
head always on a spike at the Tower Of London?

> JM:-
> doing
> so, though, I think that you are right that *something*
> has been lost.

JE:-
If you delete a constant via the process
of oversimplification you can reverse cause
and effect within the parent theory.
This means the model becomes a contradiction
to its own parent theory, e.g. Hamilton's Rule.
Here genes use bodies to maximise the replication
of genes. However I argue, the yet unnamed mystery
which was beheaded to make Hamilton's model always
maintained that bodies use genes to maximally
reproduce more bodies. Felsenstein will not even
consider cause and effect arguments let alone
any reversal of it.

> JM:-
> The simplified model will be incapable
> of addressing some of the questions that the original
> model was capable of addressing.

JE:-
If the simplification reduces a variable
to being outside of an empirical range it can
misrepresent whatever it is it was simplified
from. If zero degrees Kelvin
was ever documented within nature it would
refute physics as we know it. Yet, within a
model this can be reached. So, within such
a model what is now being represented by
zero degrees Kelvin?

> JM:-
> But it is not
> automatically WRONG simply because the constant has
> disappeared.

JE:-
If a constant is mathematically deleted
(heuristically set to zero) or remade into a
variable then the parent theory from
which the model was oversimplified becomes
a headless version of the same theory.
Such a model remains irrational until it
grows a new head. Hamilton's Rule remains
headless.

> JM:-
> An example. Newton's law of gravitation involves a
> constant G. But we make no use of this constant in
> deriving Kepler's laws. One can derive Kepler from
> Newton using an (over)simplified version of Newton
> which contains no constants. The (over)simplified
> law of gravitation would be expressed as a proportionality
> without a constant, rather than as an equation with a
> constant.

JE:-
The above were all built from Newtonian Mechanics.
Here time and mass are assumed constants. These
are never deleted from anything. Do we agree that
time passes within a Newtonian universe at the same
rate anywhere within the universe? EK refuses to
debate with me that mass remains constant within
Newtonian Mechanics. No mass is created or
destroyed within a Newtonian Universe. If
you agree (or disagree) I suggest you debate
such a basic with him. These constants are
always implied even if they are not referred
to directly. When you heuristically set a
constant to another number such as zero,
you zoom into an idea using a microscope.
You see more of absolutely less. However
what you see still remains connected to what
you do not see. Hamilton's Rule not only pretends
that what it sees is not so connected, it can
set up a logic to compete against it.

> > JE:-
> > Quite clearly a theory has to
> > incorporate all empirical variables
> > within defined ranges
> > but a model does not have to do so.
> > Therefore no simplified model can
> > even exist without its parent
> > theory. Any alteration of
> > variables beyond their empirical
> > range (a range documented to exist
> > within nature) constitutes a
> > simplification from a parent
> > theory.
> > Popper argues that the
> > parent theory must be refutable.

> JM:-
> And I agree with Popper. An irrefutable theory
> is worthless. If it prohibits nothing, then
> it tells us nothing.

JE:-
Exactly, it is just commonsense.
However unless you can separate
within a theory a non verification
from a refutation you may make a
really big epistemological error.

_____________________________________
Why can't you reason that
even if a model cannot be refuted
(note: all refutations are empirical)
BECAUSE it is a simplification the
whatever-it-is this model was simplified
from must be refutable or both remain
irrational?
______________________________________

> > JE:-
> > This must possible by attempting
> > to observe a prohibited
> > variable/variables within the
> > theory logic or more directly,
> > by attempting to observe a
> > prohibited constant.

> JE:-
> Agree.

JM:-

If Hamilton's Rule IS a
simplified model then:

What does the whatever-it-is
that Hamilton's Rule was oversimplified
from, prohibit? (please do not confuse
prohibition with non verification).

If Hamilton's Rule is NOT a
simplified model then:

What does Hamilton's 100% relative Rule
prohibit? (please do not confuse
prohibition with non verification).

You must supply an answer to at least
one of the above.

> > JE:-
> > Models are not refutable because
> > they are not an _empirical_ test
> > they are just a heuristic test.
> > All a heuristic test can provide
> > is a verification or non
> > verification of the parent
> > theory. Models that become
> > divorced from their parent
> > theory remain meaningless.

> JM:-
> I'm not sure why you are using the word "heuristic"
> here, or what you mean by it.

JE:-
Any value allocated to a variable
that remains outside of an empirical
range is just a heuristic, e.g. zero
degrees Kelvin.

> JM:-
> Are you saying that
> theories are either correct or incorrect,

JE:-
No, refutable but non refuted or
refutable but refuted. An invalid
theory cannot be refuted. Such
a theory can be logically valid
but is not rationally valid,
e.g. Hamilton's Rule employed
as a stand alone fitness accounting
device. Nobody can claim the
rule has not been so used.

> JM:-
> but models
> are evaluated by different criteria - they are either
> useful or unuseful? If so, then maybe I can see
> why you are using the word "heuristic". But then
> I don't see how a model provides "verification or
> non verification".

JE:-
If just a heuristic value is applied
to any variable then the model can only
verify or non verify changes in other
variables within that model when using
empirically correct values for
all constants. Such a model is very useful.
Because the allocated value to a variable
remains outside of its empirical range,
this model cannot refute the theory
it was simplified from but it can
provide a logic that presents a
valid point of refutation for that
theory, e.g. zero degrees Kelvin.
Unless zero degrees Kelvin becomes
empirical it only remains a heuristic.
Such a model cannot _in itself_,
refute anything. In Hamilton's Rule
unless a lineal gene fitness is
empirical it remains a heuristic
gene fitness value. Not one single
lineal genomic gene fitness has ever
been documented within nature.

> > > JM:-
> > > Examples of theories might be Newton's law of gravitation
> > > or the ideal gas law "PV=nRT". Examples of limitations
> > > on the class of situations might be the Einsteinian limitation
> > > on Newton's theory - that it only works if the velocities
> > > involved are much less than c. Or the limitation on the
> > > ideal gas law that it only works if the density is much
> > > lower than the density of a liquid.
> > > Theories may be fundamental or phenomenological. Newton's
> > > theory is fundamental. We don't (yet) try to explain it
> > > reductionistically as the consequence of some more fundamental
> > > law.

> > JE:-
> > A fundamental theory is entirely
> > based on just a guessed constant
> > so it is not just reductive.
> > Guesses are produced via
> > induction (imagination). Of course
> > guessed constants must constitute
> > a part of a logical structure.
> > Together they form a theory.
>
> I can see where you are coming from here, but I am not
> so sure that constants are as indispensible as you seem
> to think. Newton's law of gravitation involves a constant,
> but Newton, as far as I know, never attempted to guess the
> value of that constant. It wasn't measured for over
> a century (Cavendish was 19th century, I believe).

JE:-
In Hamilton's Rule you are looking at a maximal
over simplification: a situation where _every_ constant
term has been mathematically deleted within the rule
Also, the rule does not just implicitly refer to any other.
You cannot delete every constant
term from Newtonian Mechanics or any other theory
and make rational sense. For verification purposes
some constant terms can be ignored. This enables
you to zoom in on an idea using an epistemological
microscope. However "out of sight is not out of
mind", e.g. relatedness IBD is always r^e. Resetting
e to always be 1 does _not_ delete epistasis it
resets it to zero. For the process of refutation no
constant term can be so, deleted. All constants
must be set to their empirical measure.

> > > JM:-
> > > The ideal gas law is phenomenological. We can reduce
> > > it to more fundamental laws of mechanics using a "model".
> > > Maxwell first presented this model. Assume, he said, that
> > > the atoms of the gas behave like tiny billiard balls.
> > > Now, clearly, this is a good model only if the atoms actually
> > > do behave as billiard balls. Maxwell assumed, for purposes
> > > of his model, that they did. And, it turns out that they
> > > do behave somewhat like billiard balls if they are monatomic
> > > and if the density is not too high. In this case, the model
> > > sacrificed some of the range of the theory it was trying
> > > to explain - the theory applies to diatomic gasses too.
> > > But in limiting the density, it was mirroring the limitation
> > > of density in the "parent" theory. That is a success of the
> > > model.

> > JE:-
> > This model was not misused, i.e. it supplemented
> > and did not attempt to replace, the theory it was
> > simplified from. Models can be used or misused.
> > A properly used model acts like a microscope.
> > You see relatively more of absolutely less.
> > Of course this is extremely useful.
> > You are seeing absolutely less because
> > at its most extreme magnification (i.e.
> > maximum model simplification)
> > the model ends up seeing all
> > of just nothing at all. At no time is any
> > model proposed that does not include,
> > either implicitly or explicitly, at least one
> > constant term. Just a set of variables cannot
> > constitute a theory of anything even if
> > it can represent valid mathematics.

> JM:-
> But if I know that gravitation is proportional to the inverse
> square of the distance, then I can use Newton's insight
> even if I don't yet know anything about the constant. I can
> prove that orbits are elliptical, for example. And the theory,
> even without the constant, is refutable. If, for example,
> Mercury's orbit is found to not be exactly elliptical, then
> Newton is refuted.
> I am talking about valid physics here, not just valid math.

JE:-
Constants can be ignored for the process of limited
verifications using a "higher magnification" but they
cannot be deleted for any process of refutation.
To suggest that an epistemologically magnified
view produced by resetting constants to other
values including zero does not remain logically
connected to its 100$ non magnified view (all
constants are only set to their empirical values)
is just absurd.

> > > JM:-
> > > Furthermore, an improvement in Maxwell's model
> > > (complicating it) by allowing a finite diameter to the billiard
> > > balls yields, not the ideal gas law, but a slightly more
> > > complicated phenomenological law ("theory") called van der
> > > Wall's law.
> > > So that is the relationship that I see between theories and
> > > models.
> > > Models are not simplifications of theories. Models
> > > attempt to reductionistically explain theories.

> > JE:-
> > Yes, "models attempt to reductionistically
> > explain theories" but they do not and cannot
> > attempt to replace these theories

> JM:-
> Not sure about this. For example, the second law of thermodynamics
> can be reductionistically explained using billiard ball models
> of gasses. But these models suggest that the second law is actually
> not really a true law - it is only true on average - i.e. for a
> large enough collection of molecules so that the law of large
> numbers holds.

JE:-
Models are used to predict a more limited
range of verifications from a parent theory
so they can be observed more exactly.
That is the only reason they are employed.
If they attempt to replace whatever it is they
are simplified from then you are left with the
microscope on high power to get on with reality.
It is difficult find the handle on your tea cup
under a 1000X microscope. You require a lower
magnification (less simplifications) to make
such a verification. To make a refutation you
must remove the idea from under the
epistemological microscope, entirely_.

> JM:-
> Based on this modeling insight, Tim Tyler, among others, claims
> that the second law is not a law. He may be right. Or, perhaps
> the problem isn't that the law fails when the number of molecules
> is small, instead the problem is that the model's definition of
> entropy fails when the number of molecules is small. Either
> position might be defended. But Tyler and many other people
> think that the theory has been refuted by the model.

JE:-
It remains absurd to claim any simplified
model can refute its own parent theory.
It can provide a heuristic condition
for such a refutation but _not_ the
refutation itself.
 
The false assumption here appears to be
the epistemologically magnified view of
the parent theory the model represents
is not connected to the parent theory
because all of the parent theory is
not now visible within the magnified
view the model represents. If you close
your eyes the world does not go away.

>snip<

> > > JM:-
> > > Models may
> > > be simplified or complicated from their predecessor models.
> > > They may introduce limitations on the situations that they
> > > attempt to cover, and those limitations may be more severe
> > > or less severe than the comparable limitations in the "parent"
> > > theory. But they don't really simplify the theory. They
> > > try to explain it.

> > JE:-
> > I agree that a simplified model tries to
> > explain a parent theory. What is the
> > parent theory that Hamilton's model
> > tries to explain?

> JM:-
> I answered this below. But, to repeat, Fisher's model tried
> to explain Darwin's theory. Hamilton's model complicates
> Fisher's model, in an attempt to provide a better model of
> Darwin's theory.

JE;-
Now we are getting somewhere. What
exactly constitutes a refutable
Darwinian fitness?

> JM:-
> Hamilton's model does not contradict
> Darwin (who certainly knew about parental care and altruism
> in the social insects).

JE:-
Yes it does for the simple reason
Hamilton's model is a headless
version of Darwin's theory.
A body without a head (Hamilton's Rule)
and a head without a body (Darwinian
total fitness without a theory structure)
are contradictions.

> JM:-
> It does contradict some simplified
> models of Darwin and some misinterpretations of Darwin. For
> example, it contradicts the naive group-fitness interpretation
> of Wynne-Edwards.

JE:-
Jim, do you realise that Felsenstein can
derive Hamilton's Rule from classical
group selection. Ask him yourself. Now
what does that tell you?

> JM:-
> It also contradicts the ultra-Darwinist
> mis-interpretation which suggests that altruism can never
> be advantageous, unless directed to direct descendents.

JE:-
Darwinism prohibits any _selected_ reduction
of total Darwinian fitness but Hamilton's
Rule requires it.

> JM:-
> However, Hamilton's model is NOT an attempt to model Edser's
> theory, therefore the fact that the "constant" cmax is
> missing from the rule should be of absolutely no concern to
> anyone.

JE:-
Utter nonsense. You said you
agree with Popper. Therefore,
what constitutes a REFUTABLE
DARWINIAN FITNESS?

> > > JM:-
> > > I have now twice used the phrase '"parent" theory of a model',
> > > each time with scare-quotes around "parent". To explain this,
> > > let me note that I think that a better use of the term "parent"
> > > is to describe the relationship between two theories (for example,
> > > Newton's theory of gravitation and Einstein's or between the
> > > ideal gas law and van der Wall's law). One can also use "parent"
> > > to describe the comparable relationship between two models.
> > > The infinitesimal billiard ball model is the parent of the finite
> > > billiard ball model. Fisher's pop gen model was the parent of
> > > Hamilton's, in a sense I will describe below.

> > JE:-
> > You seem to be presenting a model lineage
> > where each model is simplified from another
> > without end.

> JM:-
> Do you mean, "without beginning"?

JE:-
Both. Without beginning and without end.

> JM:-
> No, a model doesn't have to
> have a parent model in my sense of "parent".

JE:-
In that case it is a fully fledged theory
of nature. Since you agree with Popper then
it must be refutable. If Hamilton's model
is not a simplification of anything
(which is a hilarious suggestion) then
please provide the so far entirely missing
_empirical_ refutation of the rule.

>snip<

> > > JM:-
> > > But the relationship between a model and its "parent" theory is
> > > different; so different that I think that the word "parent" is
> > > inappropriate. The relationship between model and theory is
> > > one of (Aristotelian) final cause. The model is created for the
> > > purpose of explaining the theory. The relationship between
> > > model and parent model, or between theory and parent theory,
> > > is more like formal cause. I think that use of the "parent"
> > > metaphor is more appropriate for formal causality. So, talk
> > > about the "parent" theory of a model makes me uncomfortable.
> > > OK, enough philosophy! Lets talk about biology, or at least
> > > theoretical biology. Hamilton's rule CAN be called a theory.
> > > (I realize that I may have called it a model in the past.)
> >
> > JE:-
> > Hamilton's model is an OVER simplification
> > of SOMETHING because it does not refer any
> > of its variables to just one single constant term.
> > Such oversimplified models normally replace
> > one constant with another so they
> > now constitute a contesting theory. Hamilton
> > never replaced one constant with another
> > he just deleted it. His model remains
> > entirely irrational because of this
> > fact. A headless model remains mathematically
> > valid but rationally meaningless because
> > the variables, which are all that is
> > now left, can vary without limit.

> JE:-
> Could you explain to me what you think the constant is
> in Darwin's theory, and could you point me to the section
> of "Origin" in which he discusses it?

JE:-
Darwin implies one that is all.
It can de exactly defined: the total
number of fertile forms reproduced
into one population by each parent
HOW MANY MORE TIMES MUST I REPEAT
IT? This maximand fitness can be
tested to refutation. NO OTHER
FITNESS MEASURE CAN BE. What
does that tell you?

> > > JM:-
> > > It formulates a proposed relationship between (somewhat)
> > > measurable quantities. Hamilton (1964) also provided a model
> > > as a reductionistic explanation of his rule. This model
> > > was a *complication* of its parent - Fisher's model.
> > > Fisher's
> > > model included the assumption that a gene could only influence
> > > its own frequency by its effects on the fitness of the
> > > organism in which it happens to reside.

> > JE:-
> > The fitness of Fisher's MODEL organism was
> > just the simple sum of each supposed additive
> > gene fitness within one genome

> JM:-
> I think you are confusing fitnesses with selection coefficients
> here.

JE:-
Do you deny or confirm
that "Fisher's MODEL organism was
just the simple sum of each supposed
additive gene fitness within one
genome" ?

> > JE:-
> > because Fisher
> > stated that all genetic epistasis, even though it
> > can be inherited, remained non heritable
> > and non selectable.

> JM:-
> Fisher didn't just make this as an assumption for simplicity.
> He gave an argument justifying his claim of non heritability
> and non selectability. Do you understand this argument?
> Can you point out a flaw? Do you have a counter-argument?

JE:-
It is entirely flawed as a theory of
nature both rationally and empirically.
As a model it is valid but remains utterly
misused. I repeat: not one single linear
genomic gene fitness has ever been
documented within nature ERGO: all
empirical genomic gene fitnesses
are NON linear. What does that tell
you?

> > JE:-
> > ______________________________________________
> > However, not a single
> > lineal gene fitness has ever been documented
> > within nature.
> > _______________________________________________
> > This means Fisher's simplification
> > of fitness was an _oversimplification_ , i.e.
> > required the deletion of some fitness total
> > where no other total was supplied to replace it.

> JM:-
> Clearly you understand neither the argument nor
> its conclusion if you believe that the boxed statement
> is evidence against Fisher.

JE:-
ALL EMPIRICAL GENE FITNESSES
ARE NON LINEAL (EPISTATIC).
Fisher deleted all _empirical_
gene fitnesses within his model
when he deleted all heritable
epistasis. You can refute what I
wrote if you can provide just one
_empirical_ lineal gene fitness.
Please provide it.

 
> > JE:-
> > Fisher's fitness model remained headless.
> > Consequently, so did Hamilton's and all others
> > in this line of decent.

JE:-
This is the line of decent
for almost all Neo Darwinian
model misuse.

> > > JM:-
> > > Hamilton removed that
> > > assumption, observing that the gene could also influence its
> > > frequency by the effects of its containing organism's social
> > > behavior on another organism's fitness.
> > > He analyzed the effects
> > > of removing Fisher's assumption (and replacing it with some
> > > different assumptions) and derived the rule from the model.

> > JE:-
> > However, this influence on the "effects of its
> > containing organism's social behaviour"
> > remained entirely non epistatic within just
> > a headless model.

> JM:-
> John, I'm still trying to understand your thinking on epistasis.
> Do you believe that my grandson (r=0.25) inherits more of my
> "epistatic information" than my half-brother (r=0.25) "inherits"?
> Do you believe that my son (r=0.5) inherits more of my "epistatic
> information" than my full brother (r=0.5) "inherits"?

JE:-
I have no idea. I am simply suggesting
that you cannot argue they are the same,
i.e. constitute a variable e within r^e
that can be validly deleted by fixing it
to one within Hamilton's Rule so it can
become entirely forgotten.

> JM:-
> You seem to accept that "inheriting" (in that strange "up the tree,
> down the tree" direction that IBD employs) is a useful concept
> for individual genes. So why is it different for "epistatic
> information"? Why does either direction work for individual
> genes, but only the downward direction for epistasis.

JE:-
Non epistatic fitness information alone allows
independent gene fitnesses. Non epistatic gene
fitness only allows the opposite, dependent
gene fitnesses. What are they dependent on?
The Darwinian fertile organism level of selection.
ONLY epistatic gene FITNESSES are EMPIRICAL.
What does that tell you?

>snip<
 
> > > JM:-
> > > Now, what about Darwin? Is Darwin's Natural Selection a "theory"
> > > in Popper's sense? Popper himself certainly didn't think so.
> > > I think I have to agree with Popper on this one. Natural selection
> > > is a tautology.

> > JE:-
> > Popper, like Hamilton, only "understood"
> > Darwin's theory through Herbert Spencer's
> > lens: the tautological jingle "the survival
> > of the fittest". This jingle utterly
> > misrepresents Darwin's theory. Since I have
> > been posting why for over 4 years need I repeat
> > the argument?

> JE:-
> Darwin himself didn't think that Spencer's slogan
> was all that bad. But that is not the point you
> are making here.

JE:-
Yes. However Darwin stuck to his
original term "natural selection" because
it accurately describes what he was
talking about.

> JM:-
> I am aware that you have submitted a description of an experiment
> which you claim gives empirical content to Darwin's theory.
> Since your experiment provides (you claim) a point of refutation
> for Darwin's theory, therefore you are quite correct in claiming
> that Darwin's theory (as you interpret it) is not a tautology.

JE:-
All of science boils down to refutable empirical
things and not model _non_ refutable empirical
things. Model have an enormously important
role to perform. The problem is they can
and are, hopelessly misused e.g. Hamilton's
Rule.

> > > JM:-
> > >(That doesn't mean it is useless, it simply
> > > means that it is not a fundamental law of nature.)

> > JE:-
> > All tautologies proffered as a theory
> > are less than useless, i.e. they are
> > harmful e.g. Hitler's use of
> > Spencer's jingle to justify ww2.

> JM:-
> I make it a policy not to respond to comments containing
> the H name. But I might mention the Naturalistic Fallacy
> as an alternative source of error.

JE:-
I can't imagine why you do "not to
respond to comments containing the
H name". Hitler employed human populations
as one gigantic, unethical experiment.
He refers to spurious survival of the fittest
experiments with animals to justify what he did.
Such dreadful things cannot and must not be,
ignored. Doing so means we risk a repeat
performance. These things have to be confronted,
understood and removed from possibility.

> > > JM:-
> > > Now Darwin
> > > did offer some real theories in "Origin of Species".
> > > There is for example, the theory of common descent with
> > > modification. There is also the theory that novel variation
> > > is selection-neutral (though he later backed away from this
> > > anti-Lamarckian theory).

> > JE:-
> > __________________________________________________
> > How come you never mentioned Darwin's theory
> > of evolution by natural selection?

> JM:-
> Because I have become convinced that NS by itself really
> IS a tautology. Popper was right.

JE:-
No, Popper was wrong because total
Darwinian fitness was never made
explicit. The establishment
was then and remains today,
incompetent. They still dish up
Spencer's mush for what Darwinism
actually is and remain recalcitrant
in doing so. Their hopeless attempt
at substituting headless models
for refutable theory is laughable.
 
> JM:-
> However, NS plus
> a source of unlimited variation gives you evolution,
> so perhaps unlimited variation is the true refutable
> empirical content of Darwin's NS theory.

JE:-
Unlimited variation cannot be measured.

 
> > How come
> > yourself, and almost everybody here just ignores
> > the experiment I proposed to halt all evolution
> > by Darwinian natural selection within a natural
> > population allowing my definition of total
> > Darwinian fitness which remains equivalent to
> > cmax (which was deleted from Hamilton's
> > Rule making it a headless model) to be tested
> > to refutation?
> > __________________________________________________

> JE:-
> I have mostly refrained from commenting on your proposed experiment,
> mostly because I think that there are so many practical difficulties
> in carrying it out, and so many possible sources of error, that
> any refutation or failure to refute would be subject to endless
> controversy regarding experimental error.

JE:-
Please provide just one of each
type of practical difficulty
that you argue exists.

> JE:-
> However, just a week ago, I did ask you a question regarding the
> *idea* of the experiment. You never responded to that question.
> The question was: "If the experiment shows that fitness is reduced,
> how do you know whether it was 'selected to be reduced' or whether
> it fell for some other reason."

JE:-
You question indicates you do not understand
the proposed experiment. Each total Darwinian
fitness IS being unnaturally selected to be
reduced. IF you define total
Darwinian fitness as I have defined it THEN
forcing all Darwinian total fitnesses within
one population to remain equal for a significant
period of time must halt all Darwinian evolution
within that population otherwise this definition
of total Darwinian fitness stands refuted.

> > >snip<

> > > JM:-
> > > I think that everyone that has read Popper would have to
> > > agree that Popper was hopelessly naive when it came to statistics.

> > JE:-
> > This does not matter. All statistics can do is
> > measure when a correlation becomes significant.
> > Popper was addressing theory of cause and effect.

> JM:-
> Once upon a time, I thought that accounting was only applied
> arithmetic, but then I took an accounting course and learned that
> there is a lot of theory about business administration there.
> Accountants are not just bean counters.

JE:-
They most certainly are not theory builders
or even allowed to be! Enron accountants became
exactly that and look what they "created":
total mayhem! The job of an accountant is to
account. Unless the rules of accounting are
absolutely applied mayhem results. The job
of statistics is to measure probabilities,
i.e. account for valid and invalid observations.
These include correlations. A correlation is
NOT a theory of anything. Statistics cannot create
one but it can suggest one. The classic misuse
of statistics is to make it fit a theory you
prefer via some selective use of it and then claim
all you needed were verifications/non verifications
of that theory. This misuse of statistics
is not the fault of statistics it is a rational
and scientific error.

> JM:-
> Once upon a time, I thought that statistics was only applied
> probability theory, but then I took a statistics course and
> learned that there is an lot of philosophy (mostly epistemology
> there. Statistics is applied epistemology, not applied
> probability.

JE:-
Theory building requires inductive reasoning.
Epistemology, like anything else, also requires
same. No inductive assumption can be created by
a machine or by "statistics". Popper's deletion
of induction remains a terrible error. All he
needed to say was that it produces refutable absolute
assumptions that must contest each other. I repeat,
these are not the same as absolute dictates. In
mathematics absolute assumptions can be defined
as any constant term.

> > > JM:-
> > > If he had read his contemporaries Pearson and Fisher with any
> > > comprehension, his own work would have been much less painful
> > > to contemplate. The fact is that biologically oriented
> > > statisticians had already confronted and resolved the problems
> > > of epistemology that Popper dealt with.

> > JE:-
> > No they have not. They just ignore them
> > and then suggest they have solved them.
> > They "fix" low oil pressure in a
> > car engine by ripping out the red warning
> > light and then proclaim they have fixed the
> > car.

> > > JM:-
> > > Physicists and
> > > philosophers in Popper's day were as naive about statistics as
> > > Popper himself. But modern physicists (and some philosophers)
> > > are more sophisticated - they learned this epistemological/
> > > statistical sophistication from biology and the social sciences!)

> > JE:-
> > You are only attempting to redefine black
> > as white to suit a prejudiced viewpoint.
> > Science employs refutable theories that remain
> > entirely based on an inductive assumption.
> > No machine (let alone statistics!) can invent
> > such inductive assumptions.

> JM:-
> As a believer in strong AI, I
> disagree. But lets not go there.

JE:-
Here is my prediction: AI will
continue to fail unless a machine
can be built that can make a unique
inductive inference.

> > JE:-
> > When Popper deleted
> > induction he cut off his own head.

> JM:-
> I'm not sure that Popper actually objected to induction in the
> broad sense that you mean it. He rejected a particular
> philosophical approach to Hume's problem that claimed that
> an accumulation of confirming evidence moves you "closer"
> to certain knowlege, in some sense.
> Popper quite properly IMO, pointed out that failed attempts at
> refutation are better evidence for a scientific theory than
> an accumulation of confirming, but riskless experiments. He
> also claimed that we don't get "closer" to certainty using
> either approach.

JE:-
My approach is an evolution of Popper

> > > JM:-
> > > So, does Hamilton's rule, considered as a theory, have a parent
> > > theory, in the sense of formal cause? Yes, in a sense it does.
> > > But the parent is not really a very good theory. The parent
> > > is a kind of handwaving belief that altruism (which is arguably
> > > fairly common in nature) must be an adaptation (for some level
> > > of selection). Hamilton's rule turns this belief into a good
> > > theory (i.e. one that provides a testable mathematical
> > > relationship between possible measurements). Hamilton's model
> > > justifies this theory as a reductionistic consequence of
> > > lower-level processes that are fairly well established - for
> > > example Mendel's genetics.

> snip repeats<

> > > JM:-
> > > Can Hamilton's rule, considered as a Popperian theory, be refuted?
> > > Of course it can! Simply find a species in nature, exhibiting
> > > some social behavior for which r, b, and c can be measured.
> > > If rb>c, yet the behavior, over time, decreases in frequency
> > > in the population, then Hamilton's rule stands refuted.

> > JE:-
> > Typically you confuse a verification/non verification
> > with a refutation. If no altruism is observed when rb>c
> > then the rule is NOT refuted it is just non verified.

> I didn't say it was not observed. I said it was observed to
> decrease in frequency.

JE:-
Yes, but this only constitutes
a non verification. The rule
prohibits nothing at all because
it purports measure every possibility.
In reality this only appears so
because the rule is just 100% relative.

What is the rule 100% relative to?

Either state that the rule is not
relative or state what it remains relative
to. Refusing to answer is not a rational
option.

> > JE:-
> > Something can be non verified for a million different
> > reasons. You can close your eyes and verify nobody
> > else exists. Verification and non verification are
> > NOT definitive.

> JM:-
> Perhaps you should explain (again?) the difference between
> a refutation and a non-verification. And, if you would,
> please apply it to my example above regarding the orbit of
> Mercury, which I claimed was a refutation..

JE:-
A refutation is an excluded observation
but a non verification is just an observation
that never eventuated. What are you arguing
does orbit of Mercury purported to refute?

In Hamilton's rule any number can
be allocated to any variable including -r.
Unless a number can be allocated to a variable
that is not a verification of the rule then
quite obviously, no possible refutation can
even exist.

> > As I have argued so many times
> > (why do you never supply any counter argument?)
> > the rule remains 100% relative only allowing the
> > diagnostic sign of c to be arbitrary. As O'Hara
> > confirmed, a reduced positive c (a reduced act
> > of altruism) cannot be differentiated from
> > a negative c (an act of _non_ altruism) within
> > Hamilton's Rule since both can be negative when
> > fitness remains incomplete, i.e. total fitness
> > remains a deleted constant.
 
> And, as I have said many times, your questions are meaningless
> and that fact probably indicates that you don't understand
> what "c" means. Respond to the "Perplexed in Peshawar"
> thought experiment so that I can see what you mean by "c"
> and I will respond to your argument regarding relativity
> and arbitrary signs.

JE:-
My argument is you think you know what c means
but you do not. The proof is that you refuse
to define it.
__________________________________________________
For the 2nd time: please define
it here and now in order to settle the matter.
__________________________________________________

> > >snip a mass of non verifications incorrectly
> > employed as refutations<

> > > JM:-
> > > One additional point, dealing with usenet etiquette and efficiency
> > > rather than with biology and epistemology: I have noticed that
> > > when you ask a question that you think is crucial, and receive
> > > the answer "Your question is based on false assumptions, sir,
> > > or is otherwise meaningless", then your common response is to
> > > place the question in a box and repeat it to everyone you
> > > encounter. This is, IMO, both rude and unproductive.

> > JE:-
> > I ask questions. They remain unanswered.

> JM:-
> They receive answers. The answer is: "Your question is based on
> false assumptions, sir, or is otherwise meaningless."

JE:-
Rather _obviously_, it is NOT sufficient to
just SAY so otherwise you are just dictating.
You are required to isolate the contradiction that
exists within the question OR answer the question.
If you have no answer you are required to say
so. To do nothing is not a rational option.
 
>snip repeats<

> > > JM:-
> > > Accept
> > > the fact that other people don't share your assumptions or
> > > your way of using technical terminology.

> > JE:-
> > I do, if and only if, they supply
> > rational alternatives.

> JM:-
> Well, as a rational alternative to the ideosyncratic technical
> terminology that you use, why not try using the standard
> terminology?
> Let's start with the term "reciprocal altruism", which has a
> standard meaning in sociobiology. Yet everytime someone uses
> this term to refer to the thing that it was designed to refer
> to, you go off on a political diatribe against Trivers. Then
> you confuse the situation even more by claiming that the
> proper term for "reciprocal altruism" is "mutualism".

JE:-
Reciprocal altruism IS MUTUALISM.

It is childish to label something
is a deceitful way!

> John, if you still refuse to respond to "Perplexed in Peshawar",
> please at least answer this. When you use "mutualism", do
> you mean that:
> a. Organism A helps organism B, AND organism B helps A.
> b. Organism A helps organism B, BECAUSE organism B helps A.
> c. Organism A helps organism B, and helps himself directly in
> the process.
> d. None of the above.
> e. All of the above.

JE:-
You have NOT supplied ANY objective
measure (TOTAL FITNESS) to which help
can be measured against so as to be
able to separate "help" as a donation
from "help" as a mutualised investment.
Therefore I have no other choice but
to choose d.

> > > JM:-
> > >..it might help if you bumped the
> > > conspiratorial and/or malice-based hypotheses to the bottom
> > > of the stack. These should be hypotheses to be tested only
> > > as a last resort. Placing these hypotheses first in the "queue"
> > > is not only rude, it may be a sign of incipient mental illness.

> > JE:-
> > I am arguing that Hamilton's view is
> > politically biased. Apparently this is not
> > allowed even though it remains a refutable
> > proposition. The politics of evolutionary
> > theory is open to rational investigation
> > just like anything else.

> JM:-
> Agreed. Though I would prefer that political observations not
> suddenly appear in what had been a nice technical discussion.

JE:-
They appear because they compliment the argument
in a TOTALLY OBJECTIVE WAY. At no time do I
include such a reference where this is not the
case. I do not care-a-less if some hopelessly
biased idiot does not like it.

> JM:-
> I very much enjoyed Segerstrale's book on the politics surrounding
> "Sociobiology" and its aftermath. There were giants in the
> mainstream of evolutionary biology contesting some very important
> philosophical points, with important political ramifications.

JE:-
Yes. I suggest you put forward some of these
arguments as sbe threads in their own right.
I only employ accusations of political bias
within reasoned arguments re: Hamilton's
Rule where they are applicable.

> JM:-
> I also enjoy discussions regarding the recent presidential
> elections in my country. Again, we had ... well, I can't call
> them giants, but, anyways ... important people contesting
> important points.
> What I don't enjoy is getting into a discussion with some
> fringe leftist who tells me that the presidential elections
> were not really politics; that politics is the failure of
> that fringe group to get a fair public hearing.
> And, in biology, I don't enjoy ... well, you figure it out.

JE:-
The FACTS REMAIN:-

1) Fisher was a right wing bigot.
2) Hamilton was very left wing
apologist.

Evolutionary theory was destroyed by
Spencer's right wing nonsense from
its very inception. Slowly but surely
evolutionary theory is now being dominated
by the political left. Science is
desperately attempting to live
as meat in this stale sandwich.
It is hopelessly nieve to argue that
a researcher's politics does not
influence their research.

> > JE;-
________________________________________
> > If you agree that Hamilton's rule was
> > simplified/oversimplified you must supply
> > what it was simplified/oversimplified from
> > and show how this was undertaken.
> >
> > If you agree that Hamilton's Rule is
> > 100% relative you must supply what
> > it remains 100% relative to.
> >
> > I still do not have any rational answer
> > to either of these questions.
> > __________________________________________

JM:-
> Don't tell me what I *must* do.

JE:-
You simply don't get it.
It is not _me_ that is requiring
you must answer this basic question it
is the scientific method. You know,
the one you _agreed_ to: Popper's
refutation _requirement_. A proposition
cannot be 100% relative to just nothing
at all and remain refutable.
__________________________________________
EITHER junk Popper or answer the question:

If you agree that Hamilton's Rule is
100% relative you must supply what
it remains 100% relative to.

___________________________________________

>snip gratuitous comments<

Regards,

John Edser
Independent Researcher

PO Box 266
Church Pt
NSW 2105
Australia

edser@tpg.com.au
 

Relevant Pages

  • Re: Fundamental theorems, dilemmas, fitness, and information.
    ... > you could call it a simplification. ... To get from an Edser fitness to a Fisher fitness, ... I feel obliged to point out that growth rates and offspring counts ...
    (sci.bio.evolution)
  • Hamiltons Rule: A Matter of Integrity
    ... a simplification and an over simplification ... same theory of fitness. ... to refutation my claim of total Darwinian ... _both_ models have deleted gene fitness epistasis. ...
    (sci.bio.evolution)
  • Re: Absolute or just relative fitness?
    ... > (any actual Darwinian maximum) with just ... >> it is quite possible for natural selection to drive populations close ... > Unless this relative fitness selection ... > just a simplification of a testable ...
    (sci.bio.evolution)
  • Re: Absolute or just relative fitness?
    ... Darwinian fitness, within a _testable ... from via the process of modeling simplification. ... It is not possible for an absolute ...
    (sci.bio.evolution)
  • Re: Ernst Mayr: Where Are We (1976)
    ... > the contribution of any one gene to fitness, ... organism fitness altruism is not, ... is indeed a model simplification which is better and more accurate than ... >> a function of its parents' genomes, ...
    (sci.bio.evolution)