Re: Howard theory - "cannot be computed"

On Mar 7, 11:56 am, Seanpit <sean...@xxxxxxxxx> wrote:
On Mar 6, 6:27 pm, "R. Baldwin" <res0k...@xxxxxxxxxxxxxxxxxxxx> wrote:



Seanpit <sean...@xxxxxxxxx> wrote innews:783655fd-9d14-4c96-9b10-3dda2980a178@xxxxxxxxxxxxxxxxxxxxxxxxxxxx:

On Mar 5, 9:10 pm, "R. Baldwin" <res0k...@xxxxxxxxxxxxxxxxxxxx> wrote:

In truth, you don't have a clue how likely or unlikely the
mechanism of RM/NS is when it comes to actually being able to do
the job in question.  The very best you have are bold declarations,
just-so stories, and baseless assumptions that this is the only
possible mechanism so it had to have done the job regardless of the
math.

The point that went whooshing over your head,Sean, is that not
everything subject to scientific observation is amenable to a
probability calculation.

All useful scientific theories produce predictive power - which does
in fact require some sort of probability calculation.  Observed facts
are not scientific theories.  It is not a "theory" that the Earth has
one moon.  That is as close to an observable fact as you can get - its
not really a theory.  There's a difference between observations and
hypotheses/theories that explain the observations and produce
predictive power.

While scientific theories do have predictive power, that does not
necessarily mean they produce probability calculations for a particular
situation. The number of variables may be too large.

Not true.  All useful scientific theories are based on calculable
probabilities that support one particular hypothesis to a statistical
degree of significance over other presented hypotheses.  

No, no, no, no! Probability theory *requires* that the underlying
mechanistic differences be due to chance alone. With probability
theory (as opposed to correlational statistics) you can *only*
'falsify' the expectation that the results are due to chance alone if
the empirical observations do not meet the expectations of chance
alone. You cannot determine what the causal bias was by falsifying
the expectation of chance alone.

For example, the probability expectation of an honest coin flip is
50:50 heads:tails *because* it is assumed that there is no causal bias
toward one or the other result. We can reject the hypothesis of "no
causal bias" or "random probability expectation" if we find that we
get significantly more heads than predicted on the basis of random
flips of an honest coin. [The argument is essentially the same wrt
Mendelian expectations for the gametes of a heterozygote, if you
prefer a biological example.]

But that 'falsification of random or probabilistic expectations' does
not tell us what the cause of "significantly more heads" was. It
could be due to an "intelligent flipper" who has good, but not
necessarily perfect, muscle control and always starts with the head up
and identical distances and surfaces. Again, all he needs to do is
produce a *significant* bias in the results, not guarantee that every
flip produces a head. Or it could be due to a "loaded" coin that
preferentially favors head up results. Or it could simply be that the
result is at the tail end of probability by chance (since some small
fraction of flips of an honest coin will produce a significant
difference). At this point, you have no way to tell it the reason for
the "falsification of the expectation of 50:50 heads:tails" is due to
the flipper, the coin, or bad luck.

The *only thing you have determined by your test of probability
expectation is that the coin flips do not appear to be within
expectations for heads for an honest coin.

Similarly, the only thing you have determined is that, *if* proteins
or sequences were assembled completely by chance from scratch (or the
mathematical equivalent) wrt function, it is highly unlikely that the
current non-random wrt function genomes we see would exist. I
agree.

Now we have to test explanations that involve *causal bias*. Not
merely assert that the only possible causal explanation involves a
magical assembly fairy.

For example, we can test, statistically, whether the *pattern* of
relationships between organisms living today and in the past fit the
expectations of historical divergence or common descent. Importantly,
we *do* find a few exceptions in eucaryotes of horizontal transfer
(and more examples in procaryotes), telling us that we can, in fact,
distinguish common descent from other forms of information transfer.
In fact, there is massive amounts of evidence at both the molecular
and morphological level for consistency with a *pattern* of common
descent. Now, like all science, this does not *conclusively* or
*mathematically* provide *proof* of common descent. It only tells us
the evidence is consistent with the expectations of common descent of
organisms.

Alas, there is no way to scientifically test whether the observed
*pattern* is due to your magical manufacturing fairy. Not because it
cannot be consistent. But because it could be consistent with the
pattern of common descent or consistent with the exact opposite
pattern or complete randomness. We cannot *test* an invisible magical
fairy, nor can we independently determine what its mechanism or goals
were.

The hypothesis that new functions arise by modification of old
structures that performed different functions has some other
predictions. These include persistence of traces of old sequence and
structure in the new function. There *will* be differences as well,
but some of the differences may arise rapidly by selection for changes
that optimize a new function. Others (in fact most) differences will
arise by neutral change over time. This makes the expected amount of
difference time-dependent on a geological time-frame and selection for
preservation of *some* features that the ancestral start point had
that the current sequence retains or changed. Thus a test of the
hypothesis involves the expectation of the existence of many proteins
with different but related function -- i.e. that there will be gene
families of related functionality. The second test is that there will
be all sorts of variation in the degree of interactions between
proteins, and much of these interactions will be due to specific aa
interactions rather than some generalized interaction. Of course,
*real* biological organisms and their proteins have been tested and,
to the extent that there is evidence, been consistent with the
hypothesis of protein common descent. That is, new proteins arise by
modification of pre-existing proteins and NOT by random assembly from
scratch or its mathematical equivalents.

Alas, the magical fairy hypothesis remains untestable. Unless you
have such a test?

As for your assertion that gap sizes that do arise cannot be crossed
by RM/NS, you have yet to provide a coherent argument that that cannot
happen in any specific case. In fact, one problem is that all your
examples involve systems that evolved so long ago that it is difficult
to find any evidence pro or con because so many subsequent neutral
(mostly) and optimizing (some) changes have occurred. What you have
to do is present some "new" function that meets your fsaar or other
bull shit requirements and that arose in what *real* scientists claim
is the last few million years. Then determine the size of the gap
required to produce the new function and which changes are just
neutral drift that would be expected even if no new function arose.

It might help if you had some evidence for an alternative mechanism to
RM/NS or RM/neutral drift that actually does or did exist in nature in
the absence of humans.

If you cannot
do this for one particular hypotheses among several options, it cannot
be said that any one is a more "likely" explanation vs. any of the
others.

I never said it was a theory that the Earth has one moon, did I?
Try to read for comprehension.

Try to present something relevant as an argument.  Observations are
not theories.  Theories are explanations and all scientific
explanations do in fact require statistical support to produce the
idea that one hypotheses has greater "predictive power" vs. another.

And magical fairy hypotheses do not have any statistical support.
They have uniformly failed to explain anything other than their
proposer's level of personal ignorance. Was the magical fairy (aka
angels) hypothesis better able to explain the motion of planets than
gravity? Was it more useful? Better able to explain the motion of
other objects?


That has nothing to do with the question. How do you know that
"protein sequences are assembled by chance alone until the next
beneficial steppingstone in sequence space is actually discovered"?
What is your SUPPORT for this assertion?

That is your theory.  By your own theory of the mechanism random
mutations are in fact random, non-directed, mutations that are not
influenced ahead of time by natural selection.  The only time NS comes
into the picture as a positive selective force is when the random
mutations happen to hit upon a novel beneficial sequence "target".

No, that is not my theory. It is a complete misunderstanding of it on
your part. I gather, then, that your assertion that "protein sequences
are assembled by chance alone until the next beneficial steppingstone in
sequence space is actually discovered" is based on your failure to
understand biology.

You don't understand the mainstream theory of RM/NS then.  It is in
fact the mainstream position that natural selection plays no role at
all until random mutations do in fact discover a novel beneficial
sequence in sequence space. That is the mainstream theory.

Natural selection is never absent. Every change is tested for fitness
to local conditions. Sometimes selection is significantly positive,
sometimes selection is significantly negative, sometimes it is
selectively neutral. The *consequence* is that existing genomes from
which new 'functions' arise do NOT represent a random subsample of
total sequence space. One of the assumptions of your model is that
the starting point is a uniform or random sample of total sequence
space. Moreover, the model on which you calculate your probability
assumes that new functions are uncorrelated to the sequences
historically present in an ancestral species. It is hardly surprising
that such a model, completely divorced from empirical reality as it
is, gets falsified.

[snip]

.

Relevant Pages

  • Re: marginal expectation
    ... an infinite sequence of iid samples a_n, n=1, 2, ... ... from some probability distribution with expectation ... I guess I am asking if, in a sequence infinitely long, ... Or stated another way, whether, in a manner, the expectation ...
    (sci.math)
  • Re: Monkeys typing Shakespeare
    ... And endless sequence of throws has a countably infinite length. ... each throw it is possible to assign a probability of the required ... I'd question whether you can randomly select something from a continuous event space, because there is no limit on the amount of information required to specify what you have selected. ... The mere fact that you manage to specify it in a finite time makes it likely that the selection wasn't random. ...
    (rec.puzzles)
  • Re: Number Due
    ... Explain how the statement I made about, ultimately probability holds up ... The sequence of numbers in each instance will be different but the ... My program also shows quite clearly that the number of sevens (or any ... 2058 had 33 or more sevens (2 SD above expectation) ...
    (rec.gambling.craps)
  • Re: The Theory of Evolution is a mathematically irrational belief
    ...  That base has a probability close to, ... If you are talking about mutation and selection, ... sequence of bases which will satisfy a particular selection condition? ... trial and outcomes ...
    (talk.origins)
  • Re: Is it possible to assert that a subset formed off the random sampling of a random subset is rand
    ... seems that there will always be a small probability that even in random ... selection, then same number might be chosen, and this resulting subset ... I am wondering if it is not possible to assert ... is somehow less "random" than any other sequence of numbers. ...
    (sci.math)