Re: Evidence for Big Leaps?


nightlight wrote:
Richard Forrest wrote:

It would be an evidence that neo-Darwinian model (RM+NS) is
an unlikely explanation of the observed adaptation and that
a more efficient (than random) search algorithm is responsible
for the adaptation. By convention, we can call this more
efficient algorithm an 'intelligent' or guided mutation,
or generally an ID algorithm.



No we don't.
Not in science, anyway.
If, in science, we don't know what is responsible, we say
"We don't know what's responsible. How can we find out?"

If some algorithm is more efficient than a dumb/random search
it is perfectly reasonable to call it _by convention_
'intelligent' search.

Why? Unless there is evidence for intelligence, there is no reason to
apply that term to an algorythm.

There is nothing to "research" and
"find out" about a perfectly natural naming convention,
which merely indicates the basic property (greater efficiency)
of such alternative algorithm.


Perhaps not, but there is no reason to invoke the intervention of an
intelligent agent to explain the unknown unless there is specific
evidence for such an agent.

And please don't go down the line of asserting that one can have
intelligence without an intelligent agent. That's simply nonsensical.


In ID, which isn't science, they say
"We don't know what's responsible, so GodImeananintelligentdesigner
must be responsible. Well, now that's settled, let's go and spend those
fat cheques the DI has sent us."

You seem to have mistakenly cut and pasted something from your
debate with Pat Robertson. Or maybe you just prefer to wrestle
with your little strawman.


What straw man is that? That ID claims to be science when it isn't? Or
that "GodImeananintelligentdesigner" is not the default explanation for
science?

it would be an evidence that the 'intelligent agency' (IA) model
is not necessary to explain that particular adaptation.


Oh, I see.
So the position is that there *must* be an intelligent designer, but if
any particular test doesn't uncover his actions, it's the wrong test.

Hard to falsify.

Not at all. I am simply pointing out that one can imagine a
hypothetical experiment or observation which provide no data
which can discriminate between the various models.

In which case the only possible conclusion is "I don't know", not
"GodImeananintelligentdesigner did it".

That is
not equivalent to saying that no observation can discriminate
between the ID and ND models.

And what ID "model" would that be?
Do you mean the one which says that an "intelligent designer", of
unkown but possibly supernatural powers has interfered with normal
evolutionary processes using unknown but possibly supernatural methods
for unknown and unknowable motives?

I can't think of anything such a model could *NOT* explain.


Or do you have some other ID "model" in mind?

After all, this whole series of
posts is arguing precisely the opposite -- that there exist
ways to falsify either of the two models of the evolutionary
algorithm.

Which two models?
There are several different models which have been investigated by
science, and the factors which control evolutionary development are a
complex interaction of genetics and environment. This is why biologists
research concepts such as neutral drift, allopatric and sympatric
speciation, natural selection, genetic constraints on morphology and so
on.

Are you suggesting that ID offers a scientific alternative - i.e. a
model which can be falsified by investigation?

If so, perhaps you can tell us where such a model is to be found. The
DI don't seem to know about it.


To form a theory, you have to start by testing hypotheses. So far, no
IDer has tested a single hypothesis about the action of any intelligent
designer.


The hypothesis is that there is an 'intelligent agency' (IA) guiding
the mutations (or generally all transformations of genetic networks
across generations).

And how can this "hypothesis" be falsified?

This results in a more efficient search of the
space of all "possible" (=consistent with the laws of physics &
chemistry) DNA configurations for the 'favorable' configuration than
the neo-Darwinian search algorithm (random mutation, unbiased toward
possible mutations which may turn out favorable later) can achieve.

And the evidence on which you base this conclusion is what?

Hence the testable hypothesis, as explained at length in previous
posts, is that the observed rates of the new 'favorable'
configurations should be greater than those that a random search
algorithm would find.

Bearing in mind that "favourable" is a term which is meaningless except
by reference to the environment in which a population of organisms
lives, how on earth would you test this assertion?


The single nucleotide mutation described in the paper being discussed
is a particularly convenient case for discussing what is meant above
since the search space is relatively simple to define - it is a space
of all DNA configurations which differ from the initial state by a
single nucleotide change (of any kind of change i.e. the change need
not result in a valid nucleotide, it needs only to be consistent with
the laws of physics/QM and chemistry applied to the initial state).

Mathematically, this problem is equivalent to the problem of finding
whether the algorithm I am using in the number guessing game
performs better than the random guessing. If you only know that I
guess the hidden number in 10 tries on average, my claim is that
this fact alone is not enough to decide whether my guessing
performance is better than a random guessing algorithm. You
also need to know what is the range of allowed numbers, before
arriving to such conclusion. The basic neo-Darwinist claim here
(the crowing about how the mice paper confirms the neo-Darwinist
RM conjecture) is equivalent to claiming that from just knowing
that I use 10 tries per correct guess one can conclude that my
guessing performs no better than the "random guessing" algorithm.

In the mice adaptation example, knowing the empirical rate of mutations
at the site where they found the 1 nucleotide mutation, is not enough
to conclude that the neo-Darwinian RM algorithm (the random search
in the space of configurations 1 nucleotide away from the initial
state) is the model which explains the observed adaptation. Namely,
all that the empirical mutation rates (taken together with the
number of organisms and the time for adaptation to arise) give
you is the number of tries that the search algorithm had to find
the favorable adaptation. The number of tries alone is insufficient
data to decide whether the search algorithm performed better
or worse or equally as the random search. You need to know what
the search space was, before declaring that the empirically
observed performance is no better than random search. Hence,
there is nothing there for neo-Darwinists to crow about.





Consider an 'intelligent agency' which we know to exist in nature
(human brain) applying its efforts to stock market trading. One
could trade using all his knowledge and foresight and not do
any better than chance on any particular day or a week or
throughout the whole trading career. Could you declare that
he was picking randomly if his gains don't exceed random
ones for some given span of time? You can't. You would
need more data and possibly different kind of data (such as
direct observation of his trading or an interview) to support
such conclusion.


Quite what the relevance of this is to evolution is a mystery to me.
I supose that if you have not one scrap or tittle of evidence, then all
you can do is to argue from analogy.


The intent was to illustrate just how far the neo-Darwinists
are willing to go in twisting the common language and concepts
in order to uphold their religiously/ideologically motivated
dogma (of 'random search' being the sole search algorithm used
to explore the space of DNA configurations).

I suggest that you take the time to educate yourself in what
"neo-Darwinists" actually write, and how they go about their research
before you make such stupid assertions about religious and ideological
motivation.

By placing their
absurd criteria into an analogous context but in which they don't
have so strong emotional investment as they do in the context
of evolution, makes the absurdity of the rigged criteria more
obvious.

No, it merely shows that your analogy is complelely inappropriate.



So what was the hypothesis which ID is testing?

The hypothesis is that the random search algorithm for
favorable configurations in the space of all physically
possible DNA configurations would under-perform the empirically
observed evolutionary algorithm i.e. the random search would
take much longer to find the successful evolutionary novelties
than the observed rates of such novelties.

And what would this tell us?


As explained
earlier, the single nucleotide adaptation with the mice,
greatly simplifies the formulation of the hypothesis and
its test criteria.

And if this were demonstrated, what would it tell us?


On the other hand, there is plenty of other evidence to support
evolution by natural selection, and no evidence whatsoever to support
the idea that an "intelligent designer" occasionally interferes with
normal biological processes.


You are again pasting here from your debate with Pat Robertson or
some such. Nothing I said here or in other threads implies the
strawman "theory" you are mocking above.

And what straw man would that be?
That if we don't know. an "intelligent designer" must be responsible?


First, "intelligent designer"
is quite loaded term, implying among others a deistic perspective
(which I think is wrong). I prefer term "intelligent agency" (IA),
which is more consistent with a pantheistic perspective which I
find more coherent.

Quibbling about the label does not change the fact that there is no
evidence whatsoever for the intervention of such an agency.

Namely the IA is continuously active, not just
in evolution, but also in forming and upholding the uniform physical
laws across the universe, from incredibly finely tuned physical
constants (e.g. google on "anthropic principle") down to the
elemental properties of our space-time.

So you mean God.
Fine, but don't pretend that this is anything to do with science.

A bit more about the
possible model for IA is given at the end of an earlier post here:

http://groups.google.com/group/talk.origins/msg/651222ff530cbe4e


Yup. You do mean God.
What has this to do with science?


I bring up the general view underlying my arguments only to
avoid various caricatures being tossed in by the neo-Darwinists
here as their presumed best guess as to what perspective one
must have to make those arguments.

There are plenty of ways that an 'intelligent agency' could exist
and perform directed mutations. I merely pointed out couple
possibilities which are not a priori excluded by the presently
known laws of physics.


So propose a test which could *falsify* the intervention of an
intelligent agency.

Well, showing that the neo-Darwinian random search algorithm can
perform at the level of the empirically observed evolutionary
algorithm, would make the IA model subject to Ockham's razor.


No it wouldn't.
Perhaps you can give me an example from anywhere in any other field of
science in which the falsification of a particular hypothesis has led
the researchers to the conclusion that an "Intelligent Agency" or an
"Intelligent Designer" has to be responsible.

On the other hand, if the random search algorithm were to
perform better than the observed one, that would imply a
'malevolent' (to life) IA, while the performance of random
search worse than the observed would imply a 'benevolent' IA
(this is the variant assumed by default as IA).


Oh I see!
So you presume a priori that GodImeananIntelligentAgency is
responsible, and interpret the motives of GodImeananIntelligentAgency
based on the outcome of your analysis.

Doesn't look much like science to me.

One can also easily flip upside-down these value loaded
attributes (malevolent & benevolent) by viewing the emergence
and evolution of life as a shortcut or a short-circuit speeding
up the approach of the whole system to the maximum entropy state,
the heat death of the universe (since the processes of life
accelerate the total entropy generation for the whole system).

What was that scientific alternative to evolution by natural
> selection again?

The alternative that I have in mind is a hypothesis that
the neo-Darwinian 'random mutation' search algorithm would
under-perform the actual/empirical evolutionary algorithm.

That's a test of whether or not mutation is random in respect of
benefit. Falsifying that does not provide any support for your
assertion that GodImeananIntelligentAgent is responsible.


Since that core aspect of the neo-Darwinism was never put
to test (it may be computationally too complex at present to
test decisively),

If you think that, I suggest that you read the literature on the
subject.

all three types of algorithms, the neo-Darwinian,
the benevolent IA and the malevolent IA, are equally open
and falsifiable hypotheses, none so far shown or known to be
better than the others.

So what is your model again?
If we can't explain it under the "neo-Darwinian" model, the only other
possible explanation is that GodImeananIntelligentAgency is involved?

That's not a scientific model.

It's an unfounded assertion.


It isn't ID - Michael Behe, one of it's leaders conceeded under oath
that it isn't science, so you must be talking about something else.

He is welcome to believe and say as he wishes. I don't share that
particular view.

What is *your* education in science, by the way?

Theoretical physics. Since the grad school (Brown Univ.),
though, I have been working in industry, R&D, mostly in
research and design of practical combinatorial and optimization
algorithms (used for forecasting, compression, search,...).

Well, I suggest that rather than making assertions about the current
state of development of evoluionary biology which demonstrate little
except your ignorance of the literature in that field, you take the
time to research the primary literature and talk to evolutionary
biologists about what they are doing and how they are doing it.

RF

.

Relevant Pages

  • Re: IDs semantic manipulations
    ... capable of modeling the observed rates of evolution. ... observed site are "random" or a part of some anticipatory algorithm ... computed by the cellular biochemical network. ... This is exactly the same point that the number guessing ...
    (talk.origins)
  • Re: Evidence for Big Leaps?
    ... a more efficient search algorithm is responsible ... for the adaptation. ... the neo-Darwinian search algorithm (random mutation, ... Quite what the relevance of this is to evolution is a mystery to me. ...
    (talk.origins)
  • Re: Human design and natural "design"
    ... >> This is also true for some GA's; for example in the checkers playing GA ... >> Abstract-An evolutionary algorithm has taught itself how to play the ... >> suggest that the principles of Darwinian evolution may be usefully ... The GA had a fixed set of built-in fitness functions ...
    (talk.origins)
  • Re: Minimization principal for evolution
    ... Part of evolution is ... etc.) rather than arms-race ... But a disadvantage when genetic algorithms are ... You use a genetic algorithm to do the actual optimizing, ...
    (sci.bio.evolution)
  • DNA in nonliving systems
    ... Is my tooth brush made out of DNA? ... of the design of either a living or nonliving system. ... non-compartmentalized models of evolution. ... search algorithm dubbed Directed Exhaustive Search. ...
    (sci.bio.evolution)