Re: Entropy in crystalization: up or down?

On Oct 20, 5:39 pm, "R. Baldwin" <res0k...@xxxxxxxxxxxxxxxxxxxx>
wrote:
"Seanpit" <seanpitnos...@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message

news:1192896442.109108.174790@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

On Oct 19, 5:36 pm, "R. Baldwin" <res0k...@xxxxxxxxxxxxxxxxxxxx>
wrote:

For a critique of Sean's article, to which Sean never responded,
seenews:FJ2Eh.273$aO6.29@trndny06.
I haven't checked whether he has correced
any of the errors.

Like your erroneous notion that the past history of a string cannot be
used to predict the future of the string to any useful degree?

That is not my notion. My notion is that the fact of a single string itself
is not sufficient, one must have at least some additional statistical data,
because there are categories of data on which no prediction scheme will
succeed. Background data tells you whether you are likely to be in or out of
such a category. And this notion is not erroneous.

The string itself provides statistical data. Additional data is
equivalent to making the string longer since all additional data could
be represented by a longer length of string. Although the predictions
based on the string's history can never reach perfection, they can be
reasonably proposed a supported to a useful degree of predictive
power. The hypothesis that the string is in fact the result of a
biased and therefore predictable source can be reasonably accepted and
the null hypothesis of a random origin can be reasonably rejected
based only on prior history with the string itself - upon a prior
history of successful predictions of a chosen martingale or
algorithm.

Somehow you manage to argue that because no algorithm or martingale
can reach statistical perfection that all such propositions are
worthless in call cases regardless of the degree of past success over
time. That notion of yours is simply ludicrous. It goes against the
basis of science itself - of inductive reasoning.

The question is, at what point do can one reasonably reject the null
hypothesis? What P-value is "significant" for you? You seem to
reject the whole concept of a p-value and therefore of a host of
scientific experiments that form the basis of scientific disciplines
like medical science for instance.

Or,
your related notion that because all patterns could be the result of
random production, that no particular pattern of a string can be used
with any degree of confidence to say anything about its likely random
or non-random origin?

Again, that is not my notion. You are misrepresenting what I've said, or
never understood it.

I doubt it. You've specifically said that you would not bet on any
sequence or martingale regardless of how many times it had worked
successfully in the past to predict the next digit in a string or
sequence. I've given you several scenarios, all of which you've
rejected arguing that the sequence could still be the result of random
generation and the algorithm could still be wrong.

You need to work on your concept of what it means to accept or reject
a null hypothesis without ever being able to know with 100% certainty
if you are actually correct. You need to identify a P-value or point
at which you would reject the random hypothesis as being most likely
false, based on the past history of the pattern itself without any
additional information. That's your biggest problem as I see it. You
seem to have difficulty with the very foundation of science itself -
i.e., the use of inductive reasoning.

Not at all. The issue has to do with having properly formulated hypotheses
to begin with. What, specifically, is "the random hypothesis" in meaningful
mathematical terms? We've never had this from you. Simply saying "the random
hypothesis" is mathematically meaningless. I've presented several arguments
to you, based on what it appears to mean, but the meaning seems to keep
changing as objections are presented to you.

Obviously, the "random hypothesis" is the hypothesis that a growing
sequence is the result of a truly random source vs. a predictable
biased source. Which hypothesis is most likely true? Can one tell
with a reasonable degree of usefulness based only on past experience
and the success of a predictive algorithm over the course of time? I
say yes while you say no.

The mathematical "examples" or arguments you've listed off previously
have been confused in that they haven't grasped the concept of
statistical significance, of accepting or rejecting opposing
hypotheses and at what point one can reasonably accept one hypothesis
and reject the opposing null hypothesis (i.e., what p-value is
"significant").

I take it, you either don't understand the criticism of your article, or are
not prepared to accept the possibility you might have erred?

I've gone over and over your "criticisms" with you in dozens of posts
now. How you can say that I've not done this is beyond me. Sure,
you've pointed out a few minor errors or ways to improve clarity, but
overall your main arguments have been completely off base as far as I
can tell. Ultimately, your arguments undermine the very basis of
scientific reasoning itself.

Sean Pitman
www.DetectingDesign.com

On Oct 20, 5:52 pm, "R. Baldwin" <res0k...@xxxxxxxxxxxxxxxxxxxx>
wrote:
"Seanpit" <seanpitnos...@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message

news:1192893985.643449.140780@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx





On Oct 19, 6:58 pm, wf3h <w...@xxxxxxxxxxxxxxx> wrote:
On Oct 19, 10:35 am, Seanpit <seanpitnos...@naturalselection.

As I mentioned in my first post to this thread, the 2nd Law of
Thermodynamics is NOT violated and can never be violated. The problem
with the ToE is that the proposed mechism violates the concept of
*informational* entropy. Again, informational entropy is not the same
thing as thermodynamic entropy. They are somewhat related but
distinctly different concepts.

Nonsense. There is no way to describe anything as violating the concept of
informational entropy. Informational entropy is simply a measure. It isn't a
concept to be violated. The only restriction of Shannon's theory has to do
with channel capacity.

I'm talking about a different concept of "information" here. I'm not
talking about Shannon's definition of information. I'm talking about
meaningful or useful information. The concept of "entropy" does
indeed relate to the concept of meaningful information.

I've also proposed a measure for useful information when it comes to
biosystems. All biosystems are not created equal. Different types of
systems have different minimum structural threshold requirements. If
these minimum size and specificity requirements are not met, the
system will not work at all to produce a particular type of function -
like flagellar motility or lactase activity etc.

In short, you can't put significantly more information into a system
than is already there. In fact, the information that is already there
tends to get corrupted and lost over time (hence the term
informational "entropy"). The only way to add information to a system
is from an outside source of greater information.

That is not what informational entropy means, and it is perfectly possible
to add information to a system. Noise adds information to a system, as long
as a receiver pays attention to the noise. Inputs add information to a
system. All real-world information systems are affected by external
influences.

Biosystems obtain information from their environment.

The environment or "Nature" does have information, but it cannot
transmit information to a biosystem in any sort of deliberate or
guided or goal-directed way. It cannot directly tell the biosystem
how to survive better or how to be more reproductively advantageous.
The only thing Nature can do, as a mindless entity, is select
biosystems that happen randomly upon an improved change. Until purely
random changes happen to hit upon something that is functionally
*more* reproductively beneficial than what came before, Nature is
powerless as a guiding light.

The problem with this limited ability of Nature to only see functional
changes and only select those functional changes that produce some
reproductive advantage is that higher level systems are more widely
separated in sequence/structure space than are lower level systems.
Each increase in the minimum structural threshold requirement for a
system to work translates into a linear increase in its average
distance to the next closest potentially beneficial system in sequence/
structure space. This linear increase in the non-beneficial gap
distance translates into an exponential increase in the average random
search time needed for the same size population to achieve success
when it comes to evolving or discovering a higher-level beneficial
system.

At this point, some might posit Nature as an outside source of higher
information. The problem with this notion is that the forces
generally proposed as being used by nature are mindless, non-
deliberate, and have no goal. Such forces do not bring the quality of
higher-level information to the table that is needed to improve the
information content of a system - to include all mechanical systems be
they made of biological or non-biological building blocks (computers
or amoebas). It makes no difference.

Information, as defined in information theory, is neither qualitative nor
goal-oriented. There is no "higher level" in information theory. You are
introducing your own personal nonsense to a field that was developed without
any relationship to biology and stands on its own independent of biology.
There is no reason for anyone to accept these unsupported assertions you
keep making.

There is plenty of reason to recognize the obvious concept that there
are different levels of meaningful/useful information just as there
are different levels of Shannon information or Kolmogorov/Chaitin
complexity or algorithmic entropy etc. It is overwhelmingly clear
that different biosystems have different minimum structural threshold
requirements. That's a fact. It is undeniably true for anyone who
considers the evidence with a candid mind.

Sean Pitman
www.DetectingDesign.com

.

Relevant Pages

  • Re: OT (was: Re: Letter to US Sen. Byron Dorgan re unpaid overtime)
    ... time-complexity involved in repeatedly calculating the length of a string), ... if Jos Horsmeier or Programmer Dude were to state something that I ... You wouldn't know good reasoning if it bit you on the nose (which, ... If you make a mistake and then say "oops", then nobody cares two hoots about ...
    (comp.programming)
  • Re: Two Questions about "strlen", "strcat" and "strcpy"
    ... The reasoning is that unsigned values have a larger maximum value and ... incremental string concatenation is simplified ... fact calls to strlen, implicit or not, is the real performance ... People may write strcpy() in "hand coded assembly language" if they ...
    (comp.lang.c)
  • Re: Programmers unpaid overtime.
    ... > problem in this reasoning, ... > Chris Sonnack presented some interesting numbers. ... Over and above the individual cache for a string ... > a PowerString instance and its string. ...
    (comp.programming)
  • Re: [PO] Re: Can a regular Turing Machine provide Protected Memory?
    ... String B has infinitely recursive self ... That is also why the Liar Paradox is not a paradox ... >> You are clearly now a proven liar. ... I merely used his own reasoning. ...
    (comp.theory)
  • Re: [PO] Re: Can a regular Turing Machine provide Protected Memory?
    ... String B has infinitely recursive self ... That is also why the Liar Paradox is not a paradox ... >> You are clearly now a proven liar. ... I merely used his own reasoning. ...
    (sci.logic)