Re: The Pitman CSI Formula

Seanpit wrote:

On Jul 16, 4:31 pm, John Harshman <jharshman.diespam...@xxxxxxxxxxx>
wrote:


If all you're saying is that homologous sequences resemble each other
more than expected by chance, that's nice but hardly interesting.

It is actually very interesting. It is the basis for your notion of
common ancestry.

It's a useless statement of some small fraction of the basis of my
notion of common ancestry.

Yet, without sequence homology in varying degrees you'd have no basis
at all.

Presumably you mean "sequence similarity". Sequence homology is an
inference from similarity.

I also use this evidence as the basis for my belief
in the common origin of such biosystems. The only thing we differ on
is the nature of the origin; not the fact that that origin was common.

And this is your major obfuscation here, intentional or not. You are
confusing the origin of variation with the origin of the nested
hierarchy that contains that variation. This notion of "common origin"
is not useful, largely because of that conflation.

You believe that the origin of the variation was random mutation and
natural selection. This mechanism, acting over time, produced the
nested hierarchical pattern. That is in fact your position. No reason
to hide it.

This is not true. The hierarchical pattern results from fixation of
changes on a branching tree. The processes that caused the fixations is
irrelevant to the existence and explanation of the hierarchy.

While the mechanism of random mutation and natural selection most
certainly can produce a nested hierarchical pattern, that isn't the
only feature seen in living things. The pattern itself, by itself,
isn't a problem. The problem comes with trying to explain the
functional aspects of the systems within the pattern. Lower-level
systems aren't a problem. The systems beyond these very low levels
quickly do become a problem for your proposed mechanism.

Oh, I know, you don't care about the "mechanism". You argue that
common descent from a common evolutionary ancestor is evident without
any need for knowledge about the mechanism. You assume that a nested
pattern would only be the result of common descent. Yet, this
assumption is not based on evaluation of the potential or limitations
of deliberate vs. non-deliberate processes. A nested pattern could
easily be the result of common design.

How? Since any pattern whatsoever, including a complete lack of pattern,
could be the result of "common design", your statement says nothing.
Present a reason why we would *expect* a nested hierarchy from common
design or abandon it as an explanation for the pattern, because an
explanation does indeed require reasons, more than "well it could have".

It is well within the range of
possibility. You reject this possibility, not on the basis of
scientific demonstration of the limits of what intelligence can
achieve, but upon your notion that an intelligent agent simply
wouldn't have done it that way.

Again, I reject it not for that reason, but because we have no reason to
expect any particular pattern from "common design". An explanation that
explains anything you could conceivably have observed is no explanation
at all. You need to deal with this.

That, my friend, isn't a conclusion
based in science, but in subjective conjecture - a feeling of how you
think an intelligent designer would or should create. That's very
very weak "science".

I suppose it would be, if indeed that was my reasoning. Note that I'm
not saying here than an intelligent designer didn't create the diversity
of life. I'm only saying that if he did, he used common descent as part
of the process. If we can't tell this, then scientific investigation of
the history of life is impossible. Is that your position?

I'm
not even sure of you definition of "biased"; it seems to be something
similar to "unlikely by chance".

Biased means not random.

Not as I understand it. The roll of a loaded die is biased, but it's
still random. The random distribution happens to differ from that of a
fair die.

Random actually means non-predictable with better than even odds of
success given the assumption that certain numbers are likely to fall
in a certain distribution pattern.

That is not the definition of "random" that I am familiar with.

Anything that predictably deviates
from this assumption is the result of some sort of non-random bias.
There are all kinds of potential biases. A biased distribution,
however subtle, is more predictable than a non-biased distribution. A
loaded die could be loaded in any number of different ways. It could
be loaded with a computer chip that biases the next roll based on the
outcome of the previous roll in any number of ways. It could be
biased to produce a 2 after a 6 or even a specific series like pi or
the square root of 2.


Again, that's hardly interesting.

I guess what's "interesting" is in the eye of the beholder . . .


We knew that.

The question is, how did you know that? You had to have some basis
for "knowing" that the origin was most likely biased . . .

Your use of language is bizarre. The basis is simple enough, and it's
simple probability, under any distribution you may prefer. No need to go
into anything like CSI.

Detecting bias isn't always as simple as you make it appear. If you
didn't know about pi, the pattern produced by this algorithm would
indeed seem quite "random" to you since its distribution follows that
produced by other seemingly random sources of sequences - i.e.,
radioactive decay, etc. In fact, this little subtly is the reason by
no sequence taken by itself, not even an infinite sequence, can ever
be proven to be the result of a random process - even though it can be
proven that the vast majority of potential sequences are in fact
random or have maximum Kolmogorov Complexity.

I see no point to this digression.

Similar sequences do not arise by chance, and nested
hierachies of similar sequences do not arise by chance.

How do you know? The fact is that they could have arisen by purely
random "chance" as you put it. It is possible. It just isn't
statistically likely.

It's unlikely enough that we can afford to ignore the possibility. On
this we are all agreed, so I can't imagine why you belabor the point.

The question is at what point can you be confident in your hypothesis
of a non-random origin? Some cases are easy because the distribution
pattern is so obviously biased. However, other cases, like pi, are
not so obvious and may in fact be extremely difficult to discover even
though they were in fact produced by a very simple algorithm.

Conventionally, P<.05 is taken as the threshhold of confidence. And
there are tests that allow one to reject particular models of randomness.

They demand
explanation. But we know the explanation: descent with modification and
branching. Rarely, the explanation for some similarity is not common
descent but common function, as with a few important residues in the
lysozymes of some folivores. Don't see where this gets you, though.

There is a difference between detecting bias and detecting the actual
origin of that bias. We agree on the bias part. What we don't agree
on is the origin of the bias.

The real problem comes with the fact that those systems with
increasing "irreducible complexity", as a measure of the minimum
structural threshold requirements needed to produce the functions in
question, show less and less CSI (as I defined the term) compared to
anything else in the gene pool - as a fraction of the maximum possible
CSI. It is the significant decrease in CSI relative to each increase
in IC that gets the ToE into real trouble. Increasing the CSI at
higher levels of minimums structural threshold requirements is a real
challenge for the proposed mechanism of random mutation and function-
based selection that rapidly reaches the point of incredulity.

Once again you conflate common descent, which is what I'm interested in,
with the mechanism of change, which is what you are apparently
interested in. I have very little interest in arguing about whether
mutation and selection are sufficient to account for adaptation. I'm
talking about the cause of nested hierarchy, which has nothing to do
with selection except as selection is one underlying cause of fixation.

Again, the cause of a nested pattern is very much related to the
proposed mechanism. Different mechanisms of change may or may not
predictably produce such a pattern. Also, different mechanisms may
produce such a pattern besides common descent.

You are misinformed on both counts. But feel free to elaborate, with
examples.

.

Relevant Pages

  • Re: Sean Pitman and nested hierarchy
    ... means that every key aspect of the pattern was designed. ... you don't know that the original goal wasn't to create a NHP ... example of a nested hierarchy is an army. ... common descent, but is not what would be expected from special ...
    (talk.origins)
  • Re: Sean Pitman and nested hierarchy
    ... means that every key aspect of the pattern was designed. ... you don't know that the original goal wasn't to create a NHP ... example of a nested hierarchy is an army. ... common descent, but is not what would be expected from special ...
    (talk.origins)
  • Re: Testability and Science
    ... any sort of pattern, or no pattern at all, any time you like. ... It is relevant because the argument of nested hierarchy as evidence ... for the common descent-only hypothesis is based entirely on the notion ...
    (talk.origins)
  • Re: Testability and Science
    ... that you can produce a nested pattern any time you like. ... any sort of pattern, or no pattern at all, any time you like. ... for the common descent-only hypothesis is based entirely on the notion ... an intelligent designer could easily have done it that way. ...
    (talk.origins)
  • Re: Common Descent non-Falsifiable?
    ... in the absence of any other credible explanation. ... Branching descent produces a pattern that cannot be sensibly exlained ... common descent would have to be at work. ... descent to be the best explanation for the nested hierarchy of life. ...
    (talk.origins)
Loading