Information theory?

From: Wesley R. Elsberry (
Date: Fri Jan 14 2000 - 12:19:27 EST

  • Next message: Cliff Lundberg: "Re: evolution and stewardship"

    Steven Jones wrote:


    SJ>Johnson is summarising a complex issue for a *newspaper*
    SJ>interview. I cannot speak for Johnson but I believe, from
    SJ>discussions in another forum, that he would probably
    SJ>concede that mutation and selection might be able to create
    SJ>small amounts of information in a specialised, technical,
    SJ>Information Theory sense. For example, when a gene is
    SJ>duplicated, according to Information Theory, there is no
    SJ>new information *content*. But then if one of the pair of
    SJ>genes subsequently mutates, that is technically the
    SJ>production, of one new bit of information content, in an
    SJ>Information Theory sense.


    By either Shannon or Kolmogorov information measures, a
    message comprised of two identical subunits does contain more
    information than is in the subunit, so long as the subunit is
    not composed entirely of repeats of one symbol. The
    highlighting of "content" makes it look like Steven is
    attempting to discuss meaning, which is not a topic of
    Information Theory.

    The information change due to a point mutation doesn't reduce
    to "one bit" so easily. Under Shannon, one could develop an
    expectation based upon the fact that one base instantiates one
    out of four possible states. Then one could modify this by
    reference to an ensemble of possible messages. Of course,
    other types of mutations may cause larger changes in measured
    information by either Shannon or Kolmogorov measures.

    William Dembski took a stab at estimating a maximum information
    increase per generation due to natural selection. See it in


    This archive was generated by hypermail 2b29 : Fri Jan 14 2000 - 12:08:47 EST