Re: ID and the Difference Between Spheres and Cubes

nightlight wrote:

It is relevant against the unconditional claim (to which it responded)
that mutations are absolutely not correlated with any future states of
the organism's environment.

Of course present events are correlated to future events. It's likely
that they are not caused by them, since, as far as we know, physics
laws are causal.

It's widely understood that quantum effects are not
relevant at the scale of organic molecules, like dna,
for a phenomenon called "decoherence".

That is a crude, low resolution statement. The 'decoherence' washes out
the aspects of QM which cannot even in principle be simulated by a
classical model, such as violations of Bell inequality. But this was
not the aspect being used in my argument. The QM aspect I used was QM
indeterminism, i.e. that even the theoretically most precise knowledge
of a state does not allow you in general to predict the outcome. { The
Bell's theorem and its tests are used to support QM claim that this
indeterminism is fundamental.} A larger system in which decoherence
would normally be significant, would not violate Bell's inequality,
hence it could be simulated with some classical indeterministic system
(e.g. where classical noise may play role of QM indeterminism). But my
argument only required necessary indeterminsm, which would apply to
exact quantum model as well to its classical simulation.

It's ok, but then I don't understand why you need to use quantum
mechanics, since statistical mechanics, which deals with indeterminism
at a classical level, would be fine.

...

Anyway I think that you are thinking of something
slightly different. You are asserting that the dna
mutation events are biased towards favorable outcomes.
This would imply that both statistical and quantum
mechanics were wrong, since they predict that the
random events follow certain probabilty distributions
that are inconsistent with persistently mostly
favorable outcomes.

It wouldn't mean they are wrong. For example if QM predicts on average
F of some event E out of sample of N measurements, then if you repeat
the N measurements tests, you won't get exactly F events E with each
N-sample. You will get binomial distribution with an average number of
events F, but any number of events, call it K from 0 to N, has some
nonzero probability:

P(K)=p^K*(1-p)^(N-K)*C(N,K),

where p=F/N and C(N,K)=N!/(K!*(N-K)!). Additionally, QM doesn't tell
you which K measurements out of N will produce event E. There are
C(N,K) ways to pick these K out. Hence you can satisfy the
probabilistic constraints and still have plenty of freedom to send
message in a sequence of outcomes.

Note that this is similar to observing empirical probabilities of
various letters in English texts. For a given writer these tend to be
fixed (even word frequencies remain fixed and can be used to help
identify the author). Hence a series of books by a writer will have
letter & frequencies which comply with his signature probabilities, yet
the texts are entirely different.

Further, market research and social sciences use routinely
statistical/probabilistic methods, even though their subjects seem free
to choose what to do. A good poll of 5000 'likely voters' will
generally predict election outcome even though the remaining 300
million - 5000 are free to decide as they please.

There is no way for an external force to influence
favourably the tosses and still adhere to the
theoretical binomial random distribution.

An agency with memory which keeps track of frequencies and can keep
blocks of multiple symbols before producing adjusted output can send
messages without violating binomial distribution. This is also trivial
to do if you set the encoded message to grow proportionately to
sqrt(N), where N is the length of the full sequence. Similarly, if you
compress your message optimally then encrypt it, it would be
indistinguishable (except to a cryptoanalyst) from a genuine Bernoulli
sequence.


I don't clearly understand why you started talking about messages. This
doesn't seem an appropriate formalism unless you explain better how
"messages" (and Information Theory, I suppose) are related to
favourable/neutral/unfavourable outcomes.

Quantum Events cannot be biased towards favourable outcomes and follow
exactly the theoretical distributions, because there would be certain
regions of the outcome space (the favourable regions) more probable
than the other regions.

Anyway I agree that Quantum Events might not be following the
theoretical distributions but some other probability distributions or
even deterministic algorithmic processes, immensely more complex, that
produce results indistinguishable from the theoretical distributions by
any know statistical test.

In fact, deterministic processes that generate apparently random
sequences, the pseudo-random number generators, exist and are widely
used in computer science.

However, such algorithms are usually relatively simple, while an
algorithm capable of generating pseudo-random sequences of mutations
biased towards some outcomes which are favourable for an organism,
would be extremely complex.

Thus, the Occam's razor tell us that, since there is no empirical
evidence to belive that such a complex process is at work, we stick
with the simpler "pure random" (unbiased) distributions.

Thus, unless you provide adequate evidence that the
dna mutation are biased, we should hold that our
present scientific knowledge is correct and you are wrong.

As explained above, there is no need to violate QM or statistical
mechanics in order to have directed mutations. After all the directed
mutations already exist (e.g. produced by molecular biologist) and that
instance violates no natural laws. There is no a priori reason why
other (besides the brains of molecular biologists) intelligent
networks, from biochemical reaction networks within cells through
ecowebs, could not do the same. They all can interact at the physical
level (through any number of intermediate interaction chains) with the
cellular DNA, just as the brain of a molecular biologist does.

This seems confusing. I thought you were claiming that the source of
"intelligence/purpose/foresight"
in dna mutation was a bias in the underlaying quantum random events.

We know that humans can mutate a dna "intelligently" but we have no
reason to belive that this imply any sort of this "quantum bias". In
fact, as far as we know, the human intelligence is just an emergent
property of large number of random events that happen in the human
brain, which are, themselves, unbiased.

So the fact that human biotechnolgist can make "intelligent" mutations
by no way imply that such biased event exist.

Finally, I have explained several times the explicit quantitative
criteria which distinguishe whether the "random" mutation model is
capable of reproducing the observed rates of evolutionary novelties or
not. Hence, whether such computation is presently practical or not, the
mere existence of a precise quantitative criteria implies that the
question of ID vs RM is a perfectly valid scientific question. There
may be other more accessible tests, but they are not necessary to
falsify the neo-Darwinian claims that ID vs RM is not a legitimate
scientific question). It is and there exist algorithms which
distinguish between the two.

I think that ID is not science in general because it's always
impossible to prove or disprove that something was designed by an
"intelligent" entity/process, without asserting any proprety of the
process, and without even defining the term "intelligent", so ID is, in
general, unfalsifiable.

However, individual claims of ID can be falsifiable.
The test you proposed, which I think is too simplistic, could be in
principle carried out.
Even if it wasn't simplistic and if it proved that our model of RM is
not true, it would be difficult to claim that this had proved ID.

It could be simply that our model of Random Mutation, and perhaps even
Quantum Mechanics were not correct. This would be a very interesting
scientific result, but claiming that this was the proof of Intelligent
Design would be an Argument form Ignorance.

(And here I apologize for my Ignorance of English :) )


To put it more precisely, it is the RM conjecture that excludes any
statistically significant "excess" of favorable mutations. And the
"excess" is understood to be with respect to the rates that the RM
model would predict. Thanks for agreeing that this is a perfectly valid
scientific question which can in principle be decided (once we can
compute what the prediction of RM are and how do they compare to the
empirical rates of favorable mutations).

Ok.

You're trying to weasel out by switching to debate about definition of
"biologically evolved". I am simply saying that "natural processes"
exist which do perform purposeful mutations. RM conjecture must take
that fact into account. The RM which agrees with empirical facts cannot
be defined or stated without introducing messy exceptions (which are
too vague to be a scientific theory).

No. The Theory of Evolution (RM+NS) doesn't claim that ALL living
organism MUST be generated by RM + NS. It proposes a reasonable process
which can explain the origins of organisms without introducing unknown
forces/entities/processes. If we have the evidence that some organism
was "designed" by some known entity (for instance an human
biotechnolgist), we accept the latter explanation. This by no way
contraddicts the theory.

Again, it's an example of the usage of the occam's razor.

But the RM conjecture is testable, in principle. Should it fail the
test, the remaining options is ID (intelligent benevolent designer) or
MD (malicious designer, which makes the favorable empirical rates lower
than the RM model prediction).

As I said before, implying Intelligent (or Malevolent) Design from the
failure of RM, would be an Argument from Ignorance.

The point was that you cannot use any statistical patterns which can be
found in evolution of technologies, languages, religions, human
societies,... as a proof that the biological mutations are
undirected/random, when you find analogous patterns in the biological
domain. If you bring up some statistical pattern from biology as a
proof or indicator in favor of RM conjecture, and if someone can point
out to an analogous pattern from these domains, your proof is invalid
and your indicator is irrelevant.

Ok.

Why would the intelligent agency/agencies guiding evolutionary
mutations be any smarter or more accurate? Are you perhaps trying to
sneak in the "perfect designer" strawman?

No, I was just saying that using a random mutation model in these
fields doesn't lead to an high approximation error.

The official public threats to destroy careers of anyone supporting ID
are no secret. There has been plenty of discussions on that on the pro
ID sites such as Dembski's:
http://www.uncommondescent.com/

where someone suggested that such blatant discrimination in conjunction
with recent court decisions which labeled ID as religion, could be used
to sue universities for religious discrimination.

It seems to me that these guys are more interested in legal questions
than scientific ones.

Until you have some evidence that falsify evolution
science should assume that evolution is correct.

Strawmen. The RM conjecture is not equivalent to "evolution". The RM is
falsifiable (at least in principle).

Until you have some evidence that falsify RM, science should assume
that RM is correct.

You can't use teleological argument (as you do) to eliminate such
configurations upfront. The test was for RM conjecture (not for natural
selection), hence all such mutagenic events have equal a priori
probabilities, regardless what may happen later to the cell. You can't
use teleological reasons within RM (the model whose predictions are
being tested) to eliminate such events from counting. When counting all
DNA configurations reachable via a single mutagenic event from a fixed
initial DNA state, the DNA change which subsequently kills the cell has
equal weight as the most favorable one.

I was not using a teleological argument. I was just saing that the
events that break the dna, changing it in something that isn't dna
anymore, and thus killing the cell, should go under the "probability of
dying" factor, that seems not present in your model.

...

Not at all. It only may appear "incredible foresight" to our present
molecular biology. But not to the intelligent agency itself (which may
well be, at least in part, implemented in the "junk" DNA). After all,
it would seem incredibly difficult for our present technology to create
a live cell from scratch. Yet cells do it with ease -- you put in one
cell with the required substratum and you check later and there are
millions of cells. Now you take in the same substratum and put in (or
around) the whole biotech industry and science, all the best brains and
biggest computers and labs, but without allowing them to call a cell
for help or to borrow cell's tools and materials (hence they can't use
anything produced by cells or viruses; that would be cheating), and you
can wait all you want, there won't any live cells in the dish. There
won't be even an organelle.

So the "Intelligent Designer" , if exists, is more intelligent and
expert that modern biologist.

we observe a lot of unfavourable mutations
that any silly human would never make. ..This seems
a problem for your theory.

It is not problem at all. Different intelligent agencies have different
computational capabilities and different optimization algorithms and
utility functions being optimized. To a cell all molecular biologists
would appear inept (after the thought experiment described above).

Still he wouldn't have done gross errors like breaking the vitamin C
gene in the human/chimp ancestor.

...
Occam's razors rules out biased mutation until it's
proven necessary to explain evidence.

The Occam's razor is relevant when selecting among theories which are
equally correct but differ in complexity. But the ND and ID cannot be
both correct since there exists an algorithm (practical or not) which
can distinguish between the two. Hence Occam's razor is inapplicable.

Until the test is carried out, they are both coherent to empirical
evidence and RM is much simpler than ID.

...
The science doesn't deal with it because it is not advanced enough, not
because it doesn't exist. It will be figured out eventually. My
conjecture is that all objects, from elementary particles and up, have
mind-stuff.

But, until you can define "mind-stuff" in scientifical terms, this
remains a metaphysical conjecture.

That's called "artificial selection".

So? The same events can belong to multiple, related and unrelated
patterns of purposeful actions. The events of you reading this sentence
at your computer is also part of a process of your ISP making money
from selling you internet access. The one doesn't exclude the other.

It's not "guided mutation" but "guided selection" on quasi-random
mutations.

I am using the term "intelligence" to mean process which uses
foresight, look-ahead to optimize some gain/utility function. The
neural networks, which are a mathematical abstraction capturing the
common regularities of variety of adaptible networks (complex systems)
in nature, are intelligent agencies.

I studied neural networks extensively in Artificial Intelligence and
Robotics courses. These are not models of natural processes but
mathematical objects "inspired" by nature.
And, in general, they have no foresight or look-ahead.

Simply by being exposed to
'sensory' inputs from environment and to punishments & rewards from
utility function (which slightly modify individual links), they
converge to a state which optimizes the punishments/rewards.

If you are lucky.

In this
"learned" phase, they internally model their environment and the 'self'
and play what-if look-ahead game to discover the actions of the self
actor which optimize their punishments & rewards. Our brains are just
one example of such "intelligent" system.

Currently, the only one known that works.

Even languages (natural and
artificial such as mathematical formalisms) form such intelligent
networks, living on humans as their substratum. There is even a paper
by Eugene Wigner titled "The Unreasonable Effectiveness of Mathematics
in the Natural Sciences"

http://www.dartmouth.edu/~matc/MathDrama/reading/Wigner.html

marvelling over the apparent intelligence of mathematical formalism,
which somehow anticipates by decades or longer seemingly unrelated
empirical discoveries in physics. Mathematicians themselves merely
follow the aesthetics of the formalism as a motivation for
developments, without any particular knowledge or anticipation of the
phenomena in physics which get discovered decades or centuries later
and for which the already developed formalism (for its beauty) just
happens to come out as a perfect model.

These theories about "macro-organisms" (like Gaia) are surerly
fascinating, but they seem irrelevant to the debate.

.

Relevant Pages

  • Re: A fully developed creature can evolve?
    ... > I'm not sure what you mean by "minor genetics changes". ... > I'm going to assume that you mean mutations. ... the usual assumption is that a mutation happens in a single cell. ... > the parent before sex, rather than in the fertilized egg after sex. ...
    (sci.bio.evolution)
  • Re: Question:)
    ... billion years in any given ancestral lineage is explained by population ... If you allow under "mutations" such things as a potential meal thriving ... becoming one with the cell, or gene exchange (or DNA strand exchange) ...
    (talk.origins)
  • Re: A fully developed creature can evolve?
    ... I'm not sure what you mean by "minor genetics changes". ... I'm going to assume that you mean mutations. ... the usual assumption is that a mutation happens in a single cell. ... If a mutation happens in a single cell of an adult, ...
    (sci.bio.evolution)
  • Re: ID and the Difference Between Spheres and Cubes
    ... the mutations are, after they occur. ... resistant bacterium into single 'daughter' dish R and no resistant ... the R dish is that the reproduction rate of the resistant bacteria was ... mutation of the protein in the E. coli cell wall that T1 binds to. ...
    (talk.origins)
  • Re: Cell division creationism
    ... hence little to no evolution. ... somatic mutations can be quite important in ... Evolution of a germ line has to ... A mutation in a lung cell never gets ...
    (talk.origins)
Loading