J
Jim Menegay
Guest
"John Edser" <[email protected]> wrote in message news:<[email protected]>...
> > > JE:- [snip] The reason why science is civilised is
> > > because the ideas that scientists invent, which are
> > > called theories, do their own fighting on the
> > > battlefield of ideas. To enter this battlefield any
> > > idea must be logically self consistent (free from
> > > contradiction) and refutable. Verification which
> > > provides the practical use of any idea is not the main
> > > weapon used to combat another idea. Science is a
> > > practical subject so, quite rightly, it mostly focuses
> > > on verification. However, the epistemologist, who must
> > > act as the battlefield referee has to check the amour
> > > of logical self consistency and the weapon of
> > > refutation for each combating idea _before_ the battle
> > > can even begin. If the epistemologist fails to do a
> > > good job, the chalk lines on a blackboard eventually
> > > transform themselves into battlefield trenches.
>
> JM:- Suppose the hypothesis is that 70% of the genetic (base-
> pair level) differences between **** and Pan are caused by
> drift, rather than by NS.
>
> JE:- The above hypothesis can be refuted but it cannot be
> verified. This is one of the harder propositions to
> understand. Even if a view can be refuted, unless it can
> be verified, it remains non testable. I will expand on
> this argument a little later. Firstly I have to cover
> some basics.
>
> Random patterns are just observed patterns that have a low
> probability of being a non random pattern, that is all.
> Consequently they are always thrown out as a non
> significant pattern of nature. This means the observation
> of a random pattern means that you have only perceived A
> ZERO EVENT, i.e. you were only looking at the white noise
> on a TV screen. Nothing of _significance_ is perceived in
> this zero event.
>
> We separate our experience of nature into perceptual
> patterns and invented concepts that explain these
> patterns, where proposed concepts cause the patterns and
> are tested by them. Since no concepts have been
> demonstrated to be heritable and the idea that concepts
> are granted by a god is not testable, then all concepts
> must have been invented by the human imagination unless
> somebody proposes a 3rd testable view.
>
> Unless a proposed causation of the 70% can be
> verified/refuted, it is not testable. The reason the
> proposed hypothesis above cannot be verified is because
> any random pattern can be validly supposed to be caused,
> at all times and at every instant, by either a random or
> non random process; no exceptions. Empirically, _non_
> random processes can be observed to cause both, random and
> non random patterns. To be able to separate a proposed
> random process from a proposed non random process, we
> cannot allow _both_ types of process to be able to produce
> random and non random patterns. If both can do so, then at
> no time can we validly test for either process, on its
> own. To this end, a hypothetical random process is
> proposed to only produce a random pattern. Now it becomes
> possible to refute the proposition that just a random
> process caused the 70%, if and only if, it is verified
> that the 70% is now a _non_ random pattern. This means you
> can never know that a random pattern is only caused by a
> supposed, random process. However gene centric Neo
> Darwinists do, and when they do, the error is very
> serious.
>
> JM:- Wouldn't that hypothesis be refuted by cumulative
> findings that 35% of the differences are adaptive (for one
> genus or the other), supplemented by a statistical
> analysis showing that differences caused by drift are
> extremely unlikely to result in adaptive differences as
> often as 5% of the time?
>
> JE:- The 70% remains non significant but the 30% is
> significant. It remains valid to assume the 70% is just
> random variation because this assumption does not alter
> the testability of any process that purports to cause the
> 30%, significant, _non_ random pattern. It does not matter
> if "differences caused by drift are extremely unlikely to
> result in adaptive differences as often as 5% of the time"
> or not, _unless_ you reclassify the 70% as now, a
> significant _non_ random pattern. However you can still
> argue, that because _rates_ of drift are selectable, this
> 5% figure should be maximal. Here the rate of random
> temporal variation is _reasonably_ expected to be selected
> to maximise and not minimise the rate of any random
> adaptation but only as a valid part of any proposed non
> random adaptive process. This link of random adaptation to
> both random and non random processes is very important.
> Any observed random adaptation may or may not have been
> caused by a non random process because non random adaptive
> processes such as selection can be observed to produce
> random patterns as well as non random patterns.
>
> It is observed rates of documented random adaptation that
> are important, but it is exactly this that has become
> invisible, because temporal variation has been wrongly
> classified as "evolution". Intense study on selectable
> _rates_ of random drift effects on random adaptation have
> not been undertaken because drift has _incorrectly_, been
> allowed to contest Darwinian selection. This was always
> invalid because the proposition that drift caused
> evolution can be refuted but cannot be verified so it
> cannot validly contest selection which can be both
> verified/refuted.
>
> All Neo Darwinists seem to do is crow about how drift is
> supposed to be able to contest and win against selection.
> They can never prove any such contest even exists let
> alone know who won it! Drift rates should be selected to
> become adjusted to _maximise_ random adaptation. If drift
> rates can only lower adaptation then the drift rate should
> be selected to always be _minimal, i.e. the lowest
> possible rate of sampling error. Since drift rates can and
> do vary, then they are probably not the lowest possible
> rate. Thus drift rates may be varying only because the
> maximal rate of random adaptation that is possible varies
> between gene groups and/or is increasing. It still remains
> heresy to talk about how rates of variation can be
> selected let alone link this to rates of random
> adaptation, because people have misused simplified models
> of what causes drift within evolutionary theory as
> independent theories of causation when they are not.
>
> We separate our experience of nature into perceptual
> patterns and the proposed concepts that cause them. The
> concepts that cause them must be on the table. Different
> concepts are invented to explain the _significant_ (non
> random) patterns that we observe. Non significant patterns
> are all the same: just a random pattern. A random process
> is suggested to cause a random pattern, only. Nobody has
> ever seen a random process they have only perceived a
> random pattern. Concepts are invented to cause the
> patterns that we can observe in nature, (the observed
> patterns are not said to cause the concepts that are used
> to explain them).
>
> JM:- It seems to me that it is the hypothesis - that 70%
> of the differences are caused by NS - that is unrefutable.
>
> JE:- Yes, that is why the 70% random pattern must be
> thrown out as, not significant. The view that the 70% is
> caused by ONLY a random process is also thrown out,
> because it cannot be verified.
>
> JM:- For the defender of this hypothesis only has to point
> out that just because an adaptive explanation of the
> difference has not yet been found, that doesn't mean that
> one won't be found in the future.
>
> JE:- Yes.
>
> JM:- In fact, Maynard Smith, in the third (1975) edition
> of The Theory of Evolution, quoted with approval exactly
> this argument against Kimura. (Which just goes to show -
> the pope ain't infallible!)
>
>
> JE:- Yes, we cannot know that may be discovered in the
> future, but we can know what epistemology will produce it.
> Allowing random sampling error to become "evolution" is a
> misuse of gene centric, over simplified models.
>
> My apologies for inadvertently deleting the most
> important part of your post. Working at 3am has its
> disadvantages :-(
JM:- Apology accepted, though I would point out that even if
you are constrained to work at 3am, that doesn't mean that
you have to respond at 3am of the same day that you receive
a post. Darwin waited 20 years to publish, allowing his
arguments time to gel. Largely as a result, his ultimate
product was a rhetorical masterpiece.
Also, I regret to note that you seem to have dropped your
experiment using a newsreader. Assuming that you did this
because of delays in seeing your replies in print, I beg you
to reconsider. Now, on to substance.
John, I hesitate to say this, but I don't think that you
understand the genius of Popper's epistemology. He
emphasized refutability, rather than verifiability, for a
reason, and your attempt to add verifiability as a second
hurdle to be crossed before entering the "arena" is just out-and-
out wrong.
I fully understand your "random pattern" / "random process"
argument. (So, I believe, does everyone else. It is
blindingly obvious. I suspect that the problem that you have
is that you express your arguments in such ideosyncratic and
eccentric terminology that people like BOH are drawn into
quibbling and disputing with you. I will avoid this trap).
The hypothesis of drift cannot be verified. Full stop.
But the fact that this hypothesis cannot be verified does
not mean that is false. And, since this hypothesis
conceivably MIGHT be true, that means that any rational
collection of epistemological rules cannot rule it out as a
scientific hypothesis. Popper understood this, IMO. That is
one of the reasons why he insisted only on consistency and
refutability as tickets into the arena.
In the language of statistics, drift is a "null hypothesis".
Null hypotheses have a peculiar status in epistemology. They
are, almost by definition, refutable, and hence they are
admitted to the arena. But, they are unverifiable, so how
can they win? The answer may surprise you, but it shouldn't.
A null hypothesis, on entry into the arena, is immediately
and by default enthroned as the reigning champion. It,
unlike other theories, doesn't need to be verified - it only
needs to resist refutation.
What entitles the null hypothesis to this privileged
epistemological status? A principle of epistemology that
long predates Popper - Occam's razor. Null hypotheses are,
by definition, simpler than their challengers. In the
language of statistics, they have fewer degrees of freedom.
Now, in everyday statistical argument, it is usually the
fate of the the null hypothesis to be used as a straw man.
It is set up only to be knocked down. It ain't easy being
null, everyone is always gunning for you. Most null
hypotheses are knocked down quickly. But, occasionally one
resists overthrow for a considerable period. Kimura noticed
that the null hypothesis of drift, while certainly damaged
by the hypothesis of NS, had never been convincingly
overthrown. Hence, he suggested, lets take this particular
null hypothesis seriously. We can never verify it, but we
ought to take its default status on the throne seriously and
admit victory by NS only in those rare cases where it is
undeniable.
Please note, that in Popper's prefered area of study,
physics, null hypotheses frequently stand up to their
challengers. Is the universe overall flat, or is there
curvature? Flat is the null hypothesis, and so far it is
holding up well. Is general relativity right, or one of its
more complicated challengers? While GR cannot be verified to
be superior to its challengers, it can definitely be refuted
by them, but has so far resisted.
I repeat from my original post, John. If you wish to have
epistemologists treated as guardians of scientific
civilization, they have to be fair referees that abide by
the received rules. Those rules specify refutability, not
verifiability, as the ticket for admission to the arena. If
you wish to argue against drift, you cannot do so while
wearing your epistemologists hat. Get out of uniform and
duke it out on the merits, like everyone else.
> > > JE:- [snip] The reason why science is civilised is
> > > because the ideas that scientists invent, which are
> > > called theories, do their own fighting on the
> > > battlefield of ideas. To enter this battlefield any
> > > idea must be logically self consistent (free from
> > > contradiction) and refutable. Verification which
> > > provides the practical use of any idea is not the main
> > > weapon used to combat another idea. Science is a
> > > practical subject so, quite rightly, it mostly focuses
> > > on verification. However, the epistemologist, who must
> > > act as the battlefield referee has to check the amour
> > > of logical self consistency and the weapon of
> > > refutation for each combating idea _before_ the battle
> > > can even begin. If the epistemologist fails to do a
> > > good job, the chalk lines on a blackboard eventually
> > > transform themselves into battlefield trenches.
>
> JM:- Suppose the hypothesis is that 70% of the genetic (base-
> pair level) differences between **** and Pan are caused by
> drift, rather than by NS.
>
> JE:- The above hypothesis can be refuted but it cannot be
> verified. This is one of the harder propositions to
> understand. Even if a view can be refuted, unless it can
> be verified, it remains non testable. I will expand on
> this argument a little later. Firstly I have to cover
> some basics.
>
> Random patterns are just observed patterns that have a low
> probability of being a non random pattern, that is all.
> Consequently they are always thrown out as a non
> significant pattern of nature. This means the observation
> of a random pattern means that you have only perceived A
> ZERO EVENT, i.e. you were only looking at the white noise
> on a TV screen. Nothing of _significance_ is perceived in
> this zero event.
>
> We separate our experience of nature into perceptual
> patterns and invented concepts that explain these
> patterns, where proposed concepts cause the patterns and
> are tested by them. Since no concepts have been
> demonstrated to be heritable and the idea that concepts
> are granted by a god is not testable, then all concepts
> must have been invented by the human imagination unless
> somebody proposes a 3rd testable view.
>
> Unless a proposed causation of the 70% can be
> verified/refuted, it is not testable. The reason the
> proposed hypothesis above cannot be verified is because
> any random pattern can be validly supposed to be caused,
> at all times and at every instant, by either a random or
> non random process; no exceptions. Empirically, _non_
> random processes can be observed to cause both, random and
> non random patterns. To be able to separate a proposed
> random process from a proposed non random process, we
> cannot allow _both_ types of process to be able to produce
> random and non random patterns. If both can do so, then at
> no time can we validly test for either process, on its
> own. To this end, a hypothetical random process is
> proposed to only produce a random pattern. Now it becomes
> possible to refute the proposition that just a random
> process caused the 70%, if and only if, it is verified
> that the 70% is now a _non_ random pattern. This means you
> can never know that a random pattern is only caused by a
> supposed, random process. However gene centric Neo
> Darwinists do, and when they do, the error is very
> serious.
>
> JM:- Wouldn't that hypothesis be refuted by cumulative
> findings that 35% of the differences are adaptive (for one
> genus or the other), supplemented by a statistical
> analysis showing that differences caused by drift are
> extremely unlikely to result in adaptive differences as
> often as 5% of the time?
>
> JE:- The 70% remains non significant but the 30% is
> significant. It remains valid to assume the 70% is just
> random variation because this assumption does not alter
> the testability of any process that purports to cause the
> 30%, significant, _non_ random pattern. It does not matter
> if "differences caused by drift are extremely unlikely to
> result in adaptive differences as often as 5% of the time"
> or not, _unless_ you reclassify the 70% as now, a
> significant _non_ random pattern. However you can still
> argue, that because _rates_ of drift are selectable, this
> 5% figure should be maximal. Here the rate of random
> temporal variation is _reasonably_ expected to be selected
> to maximise and not minimise the rate of any random
> adaptation but only as a valid part of any proposed non
> random adaptive process. This link of random adaptation to
> both random and non random processes is very important.
> Any observed random adaptation may or may not have been
> caused by a non random process because non random adaptive
> processes such as selection can be observed to produce
> random patterns as well as non random patterns.
>
> It is observed rates of documented random adaptation that
> are important, but it is exactly this that has become
> invisible, because temporal variation has been wrongly
> classified as "evolution". Intense study on selectable
> _rates_ of random drift effects on random adaptation have
> not been undertaken because drift has _incorrectly_, been
> allowed to contest Darwinian selection. This was always
> invalid because the proposition that drift caused
> evolution can be refuted but cannot be verified so it
> cannot validly contest selection which can be both
> verified/refuted.
>
> All Neo Darwinists seem to do is crow about how drift is
> supposed to be able to contest and win against selection.
> They can never prove any such contest even exists let
> alone know who won it! Drift rates should be selected to
> become adjusted to _maximise_ random adaptation. If drift
> rates can only lower adaptation then the drift rate should
> be selected to always be _minimal, i.e. the lowest
> possible rate of sampling error. Since drift rates can and
> do vary, then they are probably not the lowest possible
> rate. Thus drift rates may be varying only because the
> maximal rate of random adaptation that is possible varies
> between gene groups and/or is increasing. It still remains
> heresy to talk about how rates of variation can be
> selected let alone link this to rates of random
> adaptation, because people have misused simplified models
> of what causes drift within evolutionary theory as
> independent theories of causation when they are not.
>
> We separate our experience of nature into perceptual
> patterns and the proposed concepts that cause them. The
> concepts that cause them must be on the table. Different
> concepts are invented to explain the _significant_ (non
> random) patterns that we observe. Non significant patterns
> are all the same: just a random pattern. A random process
> is suggested to cause a random pattern, only. Nobody has
> ever seen a random process they have only perceived a
> random pattern. Concepts are invented to cause the
> patterns that we can observe in nature, (the observed
> patterns are not said to cause the concepts that are used
> to explain them).
>
> JM:- It seems to me that it is the hypothesis - that 70%
> of the differences are caused by NS - that is unrefutable.
>
> JE:- Yes, that is why the 70% random pattern must be
> thrown out as, not significant. The view that the 70% is
> caused by ONLY a random process is also thrown out,
> because it cannot be verified.
>
> JM:- For the defender of this hypothesis only has to point
> out that just because an adaptive explanation of the
> difference has not yet been found, that doesn't mean that
> one won't be found in the future.
>
> JE:- Yes.
>
> JM:- In fact, Maynard Smith, in the third (1975) edition
> of The Theory of Evolution, quoted with approval exactly
> this argument against Kimura. (Which just goes to show -
> the pope ain't infallible!)
>
>
> JE:- Yes, we cannot know that may be discovered in the
> future, but we can know what epistemology will produce it.
> Allowing random sampling error to become "evolution" is a
> misuse of gene centric, over simplified models.
>
> My apologies for inadvertently deleting the most
> important part of your post. Working at 3am has its
> disadvantages :-(
JM:- Apology accepted, though I would point out that even if
you are constrained to work at 3am, that doesn't mean that
you have to respond at 3am of the same day that you receive
a post. Darwin waited 20 years to publish, allowing his
arguments time to gel. Largely as a result, his ultimate
product was a rhetorical masterpiece.
Also, I regret to note that you seem to have dropped your
experiment using a newsreader. Assuming that you did this
because of delays in seeing your replies in print, I beg you
to reconsider. Now, on to substance.
John, I hesitate to say this, but I don't think that you
understand the genius of Popper's epistemology. He
emphasized refutability, rather than verifiability, for a
reason, and your attempt to add verifiability as a second
hurdle to be crossed before entering the "arena" is just out-and-
out wrong.
I fully understand your "random pattern" / "random process"
argument. (So, I believe, does everyone else. It is
blindingly obvious. I suspect that the problem that you have
is that you express your arguments in such ideosyncratic and
eccentric terminology that people like BOH are drawn into
quibbling and disputing with you. I will avoid this trap).
The hypothesis of drift cannot be verified. Full stop.
But the fact that this hypothesis cannot be verified does
not mean that is false. And, since this hypothesis
conceivably MIGHT be true, that means that any rational
collection of epistemological rules cannot rule it out as a
scientific hypothesis. Popper understood this, IMO. That is
one of the reasons why he insisted only on consistency and
refutability as tickets into the arena.
In the language of statistics, drift is a "null hypothesis".
Null hypotheses have a peculiar status in epistemology. They
are, almost by definition, refutable, and hence they are
admitted to the arena. But, they are unverifiable, so how
can they win? The answer may surprise you, but it shouldn't.
A null hypothesis, on entry into the arena, is immediately
and by default enthroned as the reigning champion. It,
unlike other theories, doesn't need to be verified - it only
needs to resist refutation.
What entitles the null hypothesis to this privileged
epistemological status? A principle of epistemology that
long predates Popper - Occam's razor. Null hypotheses are,
by definition, simpler than their challengers. In the
language of statistics, they have fewer degrees of freedom.
Now, in everyday statistical argument, it is usually the
fate of the the null hypothesis to be used as a straw man.
It is set up only to be knocked down. It ain't easy being
null, everyone is always gunning for you. Most null
hypotheses are knocked down quickly. But, occasionally one
resists overthrow for a considerable period. Kimura noticed
that the null hypothesis of drift, while certainly damaged
by the hypothesis of NS, had never been convincingly
overthrown. Hence, he suggested, lets take this particular
null hypothesis seriously. We can never verify it, but we
ought to take its default status on the throne seriously and
admit victory by NS only in those rare cases where it is
undeniable.
Please note, that in Popper's prefered area of study,
physics, null hypotheses frequently stand up to their
challengers. Is the universe overall flat, or is there
curvature? Flat is the null hypothesis, and so far it is
holding up well. Is general relativity right, or one of its
more complicated challengers? While GR cannot be verified to
be superior to its challengers, it can definitely be refuted
by them, but has so far resisted.
I repeat from my original post, John. If you wish to have
epistemologists treated as guardians of scientific
civilization, they have to be fair referees that abide by
the received rules. Those rules specify refutability, not
verifiability, as the ticket for admission to the arena. If
you wish to argue against drift, you cannot do so while
wearing your epistemologists hat. Get out of uniform and
duke it out on the merits, like everyone else.