Is randomness a problem for evolution? Does randomness in
genetic changes prevent the development of complex and ordered
organisms?
ID theorists claim that random processes, even acting
cumulatively over billions of years, with all the "bad" choices
stripped away, could not produce the sophisticated complex
"designs" of living organisms.
>>
While this argument sounds convincing or at least plausible, they have to
deal with the reality of matter that show how information and complexity can
increase with just random mutations and natural selection.
<< If *selection* were random, there would be some truth to this
claim, because sophisticated complex order would be fairly rare
in the results (though it would or *could* still exist, because a
few "paths" of variations that lead to such complexity might be
missed by the random removal of organisms each generation --
certainly *complex* "organisms" would exist, if not functional
ones).
But selection is only marginally random, and then mostly in the
"middle" of any range of variations for a species. At the margins,
it is very strongly non-random; the mountain lion born with no
eyes will not usually reproduce, but the mountain lion *with*
eyes *may* still fail to reproduce if it gets hit by a meteorite.
With respect to variation, the argument seems to be that we could
not expect a random process to generate the variations needed to
produce the sophisticated results we see. After all, if we have a
computer generate random strings of letters, punctuation, and
spaces, we'll probably have a very long wait before we get even
one of Shakespeare's sonnets -- and many times longer to get
*Hamlet*.
But, evolution is not aimed at producing specific variations and
specific results. It is "aimed" (by selection factors) at producing
*any* variations that will benefit the reproduction of the genome
of the organism that has the variations. Thus, evolution was not
aimed at producing any of the organisms we see today. It would
have been "happy" with *any* collection of successful
organisms. Given the variety of organisms we *do* see, this
suggests that there are many possible variations on the basic
"genetic" nature of life and the patterns of species that may
develop from them. For example, it may well be that, some 3.8
billion years ago, a slight difference in one small part of a
molecule might have set a significantly different molecular base
pattern for all subsequent life on Earth, with the result that
*none* of the species we now know (including *us*) would ever
have come into existence. There might well be intelligent life,
flying life, water life, and so on, but the whole genetic encoding
method might be different.
>>
Random mutations generate the variability on which natural selection can act.
Also note that this does not mean that a mutation has to happen when it is
advantageous. If circumstances change later on, a neutral mutation could
become advantageous or even detrimental. The same applies to detrimental and
advantageous mutations. It's the environment that dictates what is neutral,
detrimental and advantageous.
<< Even so, there are limitations on what variations will work. How
*do* organisms produce the "right" variations. Are they
somehow influenced by some "designer" hiding in the wings and
whispering instructions to the DNA during that first replication,
telling it what variation to produce?
No. *Mostly* the "right" variations are *not* produced. In fact,
in general, there are no "right" variations. There may be, for any
evolutionary need of the moment, an indefinite and perhaps quite
large number of possible variations that would suffice. Any
*one* of these would suffice. And, sometimes, one of these
*does* occur.
How? Well, suppose the variation process was *not* random in
any significant way. Suppose Bertvan and the others are right in
thinking that variations must be directed. *But*, suppose these
folks were *wrong* in thinking that the direction is *intelligent*.
Suppose variations followed some rule that made certain kinds of
variations occur vastly more often than others. This would mean
that the "others" would get short-shrifted (and you know how
painful it is to have your shorts shrifted!). Would this help
evolution produce sophisticated, functionally complex
organisms?
>>
Interesting point. Non-random non intelligent mutations. Let me think about
this.
<< Please pause and ponder this for a moment before you answer,
because not every intuitive answer is correct (as we all intuitively
know).
Remember, we are talking about a heavily non-random skewing
of variations, but *not* in an intelligent way. Thus, there would
be no special reason for believing that the narrower effective
range of variations produced would just *happen* to be the kinds
of variations that a species needed at any particular moment in its
history, just *happen* to be the kinds that would contribute to
the organism's fitness.
>>
There would be a risk that the direction becomes detrimental leading to an
all out extinction versus it becoming benificial. The risk reward seems to be
biased towards risk here.
<< In fact, in general, if genetic variations were skewed in such a
severe and general way, nearly every species would be severely
deprived of many of the options for variation that might just
allow it to survive or to evolve into a new niche, etc. Such a
restriction on variations would not, in general, help the evolution
of complex order in organisms. It would normally *hinder* such
evolution by eliminating a large number of otherwise valuable
ways of achieving fitness, some of which would include
increasing complexity (though, as has been pointed out,
complexity, as such, is hardly favorable trait; it is merely part of
the cost, sometimes, of having traits that *do* promote fitness).
Heavily skewing the distribution of variations would force the
evolutionary process often to "make do" because of the much
more limited "diet" of variations. Genetic information ultimately
comes from such variations. Limiting the range of variations
limits the kinds variations and can severely interfere with
evolution. Suppose you had to "write" one of Shakespeare's
sonnets by flagging appropriate letters as they were displayed on
a computer screen (so, if you needed an "s", and an "s" was
displayed, you'd click your mouse button, say). Now, suppose the
letter generator software was biased so as to display the letter "e"
only once in every ten thousand letter-display cycles. Wouldn't
you *rather* have a truly random letter generator that would
produce the letter "e" much more often?
Bertvan's apparently "intuitive" understanding actually *reverses*
the facts. The fact is that, given that there is no pre-knowledge of
what will contribute to fitness and what won't, we *can't* do any
better than to use a random method to generate "tries." Without
great luck in choosing the non-random process, it will almost
always do *worse* than pure randomness in this case.
Since the variation process is apparently not intelligently
directed, the best "strategy" is to make the variation process as
nearly perfectly random as it is possible to make it.
ID supporters would be right in saying that a random process
would be sub-optimal only if they were *also* right in asserting
that there is a guiding intelligence with knowledge of what is
good for the organism and what is not, *or* if there were
something else that could effectively ensure a higher rate of
successful variations.
Is intelligence the only thing that can guide the process
"intelligently"?
It may seem odd, but the answer is a qualified "yes."
*Initially*, pure randomness is the best strategy for making many
small incremental steps resulting eventually in complex order
(plus an *awful* lot of dead variations that didn't get passed on).
But, the DNA can become repositories of past successful
variations, even ones that would not be successful at the moment.
Success builds on success, not on failure (nothing builds on the
dead variations, because they stop varying). Thus, if these
"warehoused" variations are allowed to be tried out occasionally,
it may be that circumstances have changed to make them useful
again (this seems to be the way the famous finches changed beak
size so quickly: they already had gene-size genes in the genome
and only needed to make it dominant by selecting *for* it, thus
increasing the rate at which it was expressed). But, not only can
this enable an organism to evolve quickly (because it has stored
information for a particular variation), but it can produce totally
new combinations of morphological traits, and these variations
can serve as the basis for new variations that can enable the
organism to evolve more quickly in directions it has *never*
gone in the past.
In general, since some variations are *extremely* unlikely to be
advantageous, an organism may develop ways to make their
occurrence very rare indeed. Another example: If two species
both had a gene for producing large amounts of botulism toxin in
each cell in the body, the one that had an effective means of
keeping this gene extremely well suppressed would have an
advantage (other things being equal) over the one that simply
allowed this gene to be expressed randomly from time to time.
Further, mechanisms can be found that limit some kinds of
variations (such as those that would damage the reproductive
mechanism itself), and possibly to increase the rate of some other
kinds. The alleles thus displaced may not like it, but the
reproductive genes take first priority, so, if any genes support
them by producing successful variations in still *other* genes,
who are they to complain?
There will always be an element of randomness in this process,
because the genes simply *can't* "know" the future sufficiently
well from the past to enable them to dispense with randomness
altogether. But, they can still provide a basis of pre-tested
variations that can enable the organism to try out variations that
have a higher-than-random chance of being successful again, or
in new combinations, or with some modifications. If whales were
to be driven toward evolving to be land animals (again), they
could do so *much* more quickly than an animal otherwise
similar that had no land-animal genetic background. Of course,
the new land animal might be as different from the previous land
animal as it is possible for two land animals to *be* different,
but there is still enough in common between mammalian land
animals that the old genes could be revamped fairly quickly,
much more quickly than creating a good set of land-animal genes
from scratch would be.
We have initial possibilities to consider:
1. Random variation.
2. Variation strongly guided variations (intelligent design).
3.Largely non-random variation without intelligent guidance.
The ideal *initial* possibility (from some points of view) is 2,
intelligent design. The worst is 3, largely non-random variation
without intelligent guidance.
But, over time, and in an environment in which such things as
the laws of physics remain constant or nearly so, and in which
such variables as temperature and resources vary but not so
hugely or so rapidly as to make life impossible, genomes can
evolve ways of modifying variation rates for certain kinds of
variations, thus effectively incorporating "knowledge" gained
from the past about what things may be needed in different
environments. This is not, of course, real intelligence, but it has
many (though by no means all) of the effects of real intelligence.
A real intelligence guiding the variation process could
presumably make real predictions about the future environment
of the organism and thus pre-select variations to suit, and there
would be no need to allow old genetic code to be occasionally
"tried out" to see if it was again useful. There would be no need
to allow occasional "throwbacks" for example, in which some
older genes are expressed but some newer ones are not.
>>
So it's apparant design by apparant intelligence?
<< Naturalistic evolution starts out "dumb" but may over time
accumulate bits of "smarts" that make it more closely
approximate intelligently guided evolution, by enabling the
organism to evolve more quickly to a new state and do so with
fewer casualties due to needless "attempted" variations.
Are there ways to distinguish this empirically from intelligent
design? Well, yes, but only *if* the intelligent design theory is
bold enough to have empirical implications. If, as is usually the
case, the ID theorists refuse to specify anything empirically
significant about their alleged designers, the empirical facts are
simply irrelevant to it. Without empirical implications, they can
*claim* to explain anything at all, regardless of *what* the
empirical facts are. If we opened up every organism on the planet
and found it to be a mess of randomness, they would still just
say, "Well, that's just the way the designers chose to do it. And
besides, maybe it merely *looks* random." >>
Comments made with respect to Hox genes come to mind. I am amazed to hear the
claims in "mere creation" that suggest that Hox genes are a problem for
evolution and actually show design. Not according to Behe's IC criterium and
it has not been shown that Dembski's design inference does better.
<<
You see, their strategy is to be able to claim success completely
without regard for whether their theory is true or not, without
regard for *any* empirical facts. It is totally divorced from
reality in this respect. They cannot specify *any* empirically
observable fact, which, *if* it were observed, would actually
show that ID is false. There are dozens of such conceivable facts
with respect to naturalistic evolution (for example, if animal
breeding did not in fact work, or if there was empirical
geological/paleontological proof that dinosaurs arose *before*
the first bacteria), but not even a single *one* with respect to ID.
But, while it is *designed* ( ;-) ) not to be falsifiable, it can
be shown not to be particularly *reasonable*. For example, it has
taken nearly for *billion* years for life to get from its earliest
forms to many of today's species. Obviously, really good
designers *could* have done it in *much* less time. Unless ID
theorists can give us a specific and *sensible* reason why the
designers would take so long, their appeal to the unknown
purposes of the designers is meaningless evasion. If the whole
process had taken five million years, say, ID might be
supportable. But four billion years is a *major* fraction of the
entire age of the *Universe*! Thus, this is yet another potential
evidence of design that goes belly up. Naturalistic evolution with,
initially, only occasional variations (*very* short genomes or
whatever functioned as genomes), would make sense. Design is
possible but not plausible in this case.
The erratic and often extremely indirect way in which life
evolves also suggests (but, I admit, does not prove) that design is
*not* at work. The Panda's "thumb" is a good kludge, but not
something a reasonably intelligent and knowledgeable designer
with lots of genetic-manipulation tools at hand would be
expected to produce. The Panda's "thumb" makes sense
evolutionarily, but is, again, implausible under the intelligent
design hypothesis.
Finally, if we look at today's existing living organisms, from
body structure to the level of cellular structure and function, we
see hodge-podges, kludges, and plain *silliness* that make sense
given naturalistic evolution but which can only be pathetically
*excused* given alleged intelligent design.
>>
Of course this does not deny a designer either. After all an unidentified
designer with unspecified goals and unspecified mechanisms can do anything.
Even make it appear totally naturally. But then we return to my thesis that
ID even if it were able to detect design reliable does not allow us to
eliminate natural designers.
This archive was generated by hypermail 2b29 : Sun Sep 10 2000 - 22:43:05 EDT