Dembski's detecting

From: Wesley R. Elsberry (welsberr@inia.cls.org)
Date: Mon Jan 03 2000 - 13:43:20 EST

  • Next message: MikeBGene@aol.com: "Re: Many Worlds?"

    MikeBGene wrote:

    [...]

    MB> For example, Bill Dembski does not seek out a vital force, he
    MB>seeks out empirical detectors of a mind's ability to implement
    MB>a plan.

    [...]

    What I have read of Dembski does not indicate this. Dembski's
    TDI proposes that we can detect the "design" which an
    intelligent agent has left behind, but Dembski's Design
    Inference tells us precisely zip about the intelligent agent.
    IMO, DI does not even get us as far as knowing that an
    intelligent agent existed, much less acted. Dembski has
    recently been struggling with the issue of what evolutionary
    computation can do, and has introduced the concepts of "actual
    CSI" and the "appearance of CSI" into the discussion. "Actual
    CSI" is the CSI that intelligent agents leave behind, while
    "apparent CSI" is that stuff that algorithms or processes
    leave behind. It is interesting that in order to characterize
    evolutionary computation as a "probability amplifier" and thus
    incapable of producing actual complexity, Dembski abandons the
    formula for measuring complexity that he advanced in TDI.
    Instead of calculating the conditional probability of an event
    given a *chance* hypothesis, Demsbki now says that for
    algorithms and processes one should use a conditional
    probability given a *non-chance* hypothesis, or just the odds
    that the algorithm or process should give the event in
    question. That's bogus. If one calculates the complexity of
    event using two different measures depending upon what one
    wishes to find, one shouldn't be surprised to find that others
    may not wish to buy the argument. If Dembski were being fair
    about this, either complexity would always be measured with
    respect to the chance hypothesis, *or* complexity would always
    be measured with respect to the likelihood that its proximate
    cause would yield it. Omniscient, omnipotent entities would
    never produce complexity under the latter yardstick, and the
    complexity achievable by other less capable intelligent agents
    would still be very low indeed. If the latter measurement
    option is taken, it becomes problematic as to how to measure
    the complexity of an event when we do not have information
    concerning the nature of its proximate cause. Yet Dembski has
    urged us to explore and resolve events as being due to
    intelligent agents where such information is missing. How
    this can be supported I do not know.

    I mentioned proximate causation above, and that also is
    interesting in connection with Dembski. Even if we were to
    grant the claim that the Design Inference can identify
    intelligent agency, the Design Inference is incapable of
    distinguishing between CSI proximately caused by an
    intelligent agent and CSI that was ultimately caused by an
    intelligent agent removed one or more steps from the event
    being analyzed. Even if Dembski is right about what his
    Design Inference does, Dembski has done no more than provide
    another argument for Deism.

    My Dembski link page points to commentary on Dembski.
    <http://inia.cls.org/~welsberr/evobio/evc/ae/dembski_wa.html>.
    If anyone has further suggestions for links, please send
    them to me.

    Wesley



    This archive was generated by hypermail 2b29 : Mon Jan 03 2000 - 13:40:18 EST