Will the real designer please stand up?

From: Chris Cogan (ccogan@telepath.com)
Date: Tue Sep 12 2000 - 09:09:19 EDT

  • Next message: Nelson Alonso: "RE: Will the real designer please stand up?"

    Many ID advocates scoff scornfully at the evidence of
    animal breeding, mathematics, information theory,
    empirical genetics, and rigorously designed computer
    evolution of information strings of various sorts, that all
    demonstrate (beyond even a remotely sane objection to
    anyone who actually knows anything about how these
    things work and their results) that information *can* (and
    indeed *does*) sometimes increase (by *any* measure of
    information) via random variations in an existing string
    of information.

    The criticism that I've seen most often is that the
    experimenter/animal breeder/computer programmer/etc.
    is playing the role of the designer in these cases.

    But, is this true (and, if it is in some cases, does it
    *need* to be true for the same type of results to occur?)?

    As usual, the short answer is No.

    A longer answer is First, ID theory typically claims that
    *their* designer is either creating new species (or new
    "forms" of life -- as if a new species was not a new form)
    -- or manipulating the *variations* in an otherwise
    self-running process. We have yet to see any new forms
    spring from nothing, so this whole line of argument is
    obviously without evidentiary basis (with one caveat that
    I may deal with later).
    The other line, that the designer manipulates variations to
    insert otherwise-impossible information into the DNA,
    combined with their claims that selection is unimportant
    (which, in a sense it is) is, however *not* compatible
    with their critique of the evidence mentioned above,
    because the flaws in *those*
    (if any) are in the *selection* mechanism, which they
    also claim *cannot* put information into the DNA.

    So, then, if people breed animals entirely by selection
    (which, until modern knowledge of DNA, they *did*), it
    is obvious from their one premises that this method does
    *not* constitute *design* in their sense.

    Further, since they are *right* in claiming that selection,
    as such, does not insert information into DNA, we have
    the question of where the new "genetic" information *does*
    come from in animal breeding.

    Obviously, until very recently, *no* animal breeder knew
    enough about DNA to manipulate it and to *direct* variations
    to suit his purposes. So, what's a poor animal breeder to do?

    Simple: He *waits*. He waits until the information he's
    looking for appears *spontaneously*, and then *selects*
    for it. He does not create the information, he does *not*
    manipulate the DNA.

    Therefore, by the premises of this kind of ID theory, he is
    *not* a designer.

    Further, it is easy to show that given enough generations
    and *no selection at all*, the *same* kinds of variations
    will occur (because selection does not effect the variation
    process, and it doesn't effect it because it occurs *after*
    the variations).

    That is, if *all* variations are saved and enabled to
    reproduce, *some* of them will match selective criteria
    even if there *is* no selection, and it doesn't much matter
    what the criteria are, as long as they are not so severe
    that, *if* they were applied, they would eliminate too
    many variations. Indeed, if *all* variations are saved and
    enabled to reproduce (*regardless* of natural viability,
    etc.), the result will *typically* be that given complex
    results will be achieved *sooner* than if there were no
    limitations imposed by selection.

    To see how this works, consider the following string of
    letters a Suppose we replicate this many times, with, each
    time, a small probability that one of its "replications" will
    either not include the "a" at all, or will include one other
    letter as well. Now, given a large enough number of
    generations, *all* of the combinations of two letters
    beginning or ending with "a" will be produced, including
    zillions of copies of the letter "a" by itself. If we sort
    and eliminate duplicates, we'll have,

            a, aa, ab, ac, ad, ae, af, . . . . ba, ca, da, ea,
            . . . za.

    Note, please, that all of these (including "aa" and
    *except* "a") represent an *increase* in information content.

    Given enough generations and a truly random
    introduction of one-letter variations, *all* possible
    two-letter combinations will be produced. Enough
    further generations will produce all *three-letter*
    combinations as well, including, by pure chance, all
    three-letter words in the English language (or in any
    other language). And, note again that all of these
    combinations will be increases of information over that of
    the initial letter "a". Note also that, by any measure one
    chooses, *most* of these strings will be more *complex*
    than the string containing only the letter "a" by itself.
    But, will they be *orderly*? Yes, *some* of them will be
    orderly, by *any* principle of order for three letters.

    At this stage, none of the strings will be particularly
    complex. But, suppose we have disk space to store
    billions of generations of strings of letters, even very
    long ones. The result will be that some of the strings of
    letters will be as complex and/or as orderly as anyone
    might wish.

    The key here is the rate of variation and the "alphabet"
    from which the variations come, and the sizes of allowed
    variations.

    But, isn't this just a variation on the "six monkeys" idea?
    No, because the number of "monkeys" increases along
    with the population of strings, and the frequency of
    changes (per string) increases with length (and stays
    constant per letter). This would only be matched by the
    monkeys if they could exponentially increase their typing
    speed, so that, at some point, they were turning out a
    book once every millionth of a second.

    Further, *all* but the first string are based on existing strings, so the
    "monkeys" are, in effect, given "books" already written, and all they do is
    use a machine to mindlessly *copy* each book and possibly make a few
    changes. Since none of the later strings is typed from scratch, if it
    already has "meaningful" (to us) information in it, there is a high
    probability that most of that "meaning" will remain, and a reasonable
    probability that at least a *few* of the changed copies will contain even
    *more* meaningful information. Thus, if most of "Moby Dick" is already
    present, among the zillions of copies will be some that contain more of the
    original book, and some that contain other meaningful additions.

    Now, remember, this is all without *any* selection at all.
    Each "book" is simply a copy of a previous "book" but
    possibly with some one-letter changes scattered here and
    there. In sufficient time, this process will, if we can store
    the results, give us copies of every *possible* book of
    the length of "Moby Dick" (allowing or missing
    punctuation and spaces and such, which I left out to
    make the description simpler). Since all *possible*
    books of that length have been generated at this point,
    *one* of them *has* to be "Moby Dick" (along with
    astronomical numbers of books that differ from "Moby
    Dick" by small amounts, and even many more that differ
    by progressively larger differences.

    Also, since *every* book of this length (or shorter, let's
    say) is produced, then a great many other known books
    will also be included.

    Does this mean that there is some mysterious "design" at
    work? No, it's just that *all* possible strings of that length
    are produced, and since all known books of that length are
    also strings of that length, they must be included *somewhere*
    in the great morass of gobbledy-gook.

    Can we improve things by selection?

    No, except in that *then* we can get by with much less
    disk space, because we only allow strings to continue to
    exist and produce variations if they contain at least some
    visible part of a known book. Thus, while *all* the other
    variations are "out there" (logically, "virtually"), we only
    need to produce ones that seem to be headed toward
    producing something we recognize. For example, we
    might, after a few generations, throw out all the strings
    that do not start with the letter "C" (as in "Call me
    Ishmael."). Eventually, a string with "a" as the second
    letter will appear, and we then keep *that." And so on.

    Now, at this point, we are clearly introducing
    intelligence, but not design, and *especially* not design
    in the ID sense of gene-manipulation. Further, the degree
    of intelligence is *incredibly* low. The intelligence needs
    only to recognize the letter "C," the letter "a" and so on,
    blindly and stupidly *selecting* strings that match "Moby
    Dick." But, this is okay, because these ID theorists claim
    that *selection* is ineffective in producing new results, so
    we can do it all we want without affecting the demonstrative
    value of the results.

    And they are right, because it is the variation process that
    gives selection the raw variations to select.

    However, the result could easily be "Moby Dick." But,
    isn't this cheating, matching strings against an
    *already-existing* string ("Moby Dick" as a string of
    letters)? No, because this same process will work to
    produce *any* string and, more importantly, any *kind*
    of string. Why? Because, again, *all* possibilities will
    eventually be generated, and the standard of selection
    only determines which ones "survive" for the next
    generation.

    Thus, *if* animal breeding is the same as being a
    *designer* in the ID sense, then I'd have to say that the
    designer is pretty pathetic, because his role can so easily
    be performed by humans and because it is such a
    mindless process. It hardly needs any intelligence in the
    ID sense, because matching strings against other strings,
    or even against more-sophisticated criteria, does not
    require much intelligence. Children can do it.

    What's really going on is this:

    A new string is produced. Does this string match certain
    criteria (which may be as mindless and mechanical as you
    please)? If so, keep it for future generations. If not, get rid of
    it. Repeat the process *many* times.

    What do the "surviving" strings contain at the end of
    many repetitions? They contain a specification of the
    nature of the selective forces in their environment, the
    same selective forces that prevented so many other
    strings from surviving.

    Their environment would include any phenotype (body)
    they would produce if they produced one.

    And, this information includes implicit information about
    the laws of physics as embodied in local conditions
    during the string's history.

    Thus, as I've said before of genes, the strings will have
    an order of their own and will generate orderly
    phenotypes, if any, because such order is a survival
    requirement.

    Throughout the evolutionary history of all organism's, at
    *every* step, *only* those changes that are sufficiently
    orderly in functional terms (if only to the extent of not
    being too harmful) will be the ones that remain after the
    less-orderly ones are trimmed away by the consequent
    failure to reproduce.

    Selection is not design. It is not planning. It is not
    arrangement of parts by a mind for a purpose. It is the
    elimination of variations that don't sufficiently
    "represent" the selective factors involved.

    Random variation incrementally produces all types of
    strings, a little at a time, some portion of which happen to
    contain information that reflects selective factors
    sufficiently well that they continue to reproduce. No
    designer is necessary to produce random variations some
    of which happen to be successful, because the process
    will simply keep on producing variations until some are
    "right" enough to reproduce.

    Jones claims that the animal breeder is a designer. But
    the animal breeder need be no more a designer than a
    person who simply goes out and *finds* animals that suit
    his purposes. The breeder, regardless of Jones' claims,
    has historically not been able to manipulate DNA to
    control variations. Even choosing which animals to
    breed together (to mix traits, etc.) is *selection* of
    variations that *already* exist, in the *hope* that the
    result of breeding them together will result in a *further*
    suitable variations for him to select from in the future.

    And, as I said above, if *this* is what is meant by
    "designer," then ID's designer is a pretty pathetic designer
    indeed. Such a "designer" obviously does nothing more
    than what Nature can do completely mindlessly by killing
    off locally "unfit" genomes and thereby "selecting" which
    organisms get to breed together to produce further
    variations of those same genomes.

    Designer theory of this type confuses design with
    selection. But, in the attempt to claim that such selection
    *is* design, ID-ers reduce their own alleged designer to
    the level of the mindless failure of some genomes to
    survive in an environment that is unfavorable to them.
    Thus, by claiming that an animal breeder is a designer,
    they effectively negate their own claim that the designer
    *designs*; he merely *selects*.

    Poor thing.



    This archive was generated by hypermail 2b29 : Tue Sep 12 2000 - 09:13:16 EDT