On 6/14/2015 6:48 PM, Bill wrote:
>> Really? They don't show lottery draws where you live, and
>> >don't play cards either?
> Can you show how that has any bearing on anything anyone has
> said?
You don't understand the use of the term "random" as used to describe
mutations.
Where I live, the winning lottery numbers are drawn using a contraption
full of numbered ping-pong balls. I'm sure you've seen them, but look it
up if you haven't.
You seem to be arguing that the trajectories of each ball (and
ultimately which ones rise to the top of each chute) is entirely
controlled by known principles of Newtonian Mechanics; thus, but for the
very large number of variables, there is nothing random, nothing
conceptually different from computing the path of a cannonball.
That, to my knowledge, is true. And irrelevant.
What matters here is the randomness - and thus "fairness" - of the
result. And it does seem to be random. Perhaps we could find a way to
make it less so; by weighting some of the balls, or making some slightly
bigger or smaller than the rest. But if the balls really are essentially
identical, the machine should produce a random result.
What is more, we can test that result, even if we know nothing of the
process that produced the numbers. Run the machine over and over. See if
certain numbers come up more often than others. Counting is all that is
needed, albeit for a great number of repetitions.
When people use the term "random mutations", they do not mean that the
process by which the mutations come to be is fundamentally inscrutable.
Like the balls, there are known causes, and presumably some yet unknown,
that involve chemical reactions and radiation effects.
Further, standard biology does not claim that the results (unlike the
lottery) are even mathematically random; meaning that every conceivable
outcome has an equal probability. It is known that they do not.
What is claimed is that the *results* - like the lottery - seem to be
random with respect to the phenomenon we are interested in. In the
lottery, that's "who wins". In biology, that's "how 'useful' is the new,
mutated result" as compared with the unmutated one.
That too, we can test for, even if we had no knowledge of how mutations
are produced. Again, by counting. The first and most obvious thing to
leap out of the data is that the vast majority of mutations seem not to
affect the fitness of the creature at all. Most in fact do not seem to
change any functions or characteristics. So, as a first approximation,
"randomness with respect to fitness" seems like a reasonable bet.
But could there be a subtle "signal" to be found amidst the very great
amount of neutral "noise"? To my (very limited) knowledge, no test has
yet shown such a "bias". Tests have been done to see if bacteria
subjected to a toxin produce more of the "saving" mutations than a
population not subject to the toxin. [here the biologists among us can
correct or refine my fuzzy recollection if necessary] So far that does
not seem to be the case, although an overall increase in the mutation
rate has been observed.