Re: FWD [SK] Re: A Field Guide to Skepticism (3 of 3)

From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Thu Jun 05 2003 - 16:44:36 MDT

  • Next message: Harvey Newstrom: "RE: Martha Stewart and her Merrill Lynch Broker"

    Lee Daniel Crocker wrote:
    >
    > Even if the starbaby affair does reveal a certain naivite and
    > pig-headedness of Randi himself and others in CSICOP (who I am the
    > first to admit are a bit pathologically skeptical for my taste),
    > I think the unclaimed prize is still evidence that no psi claim
    > has yet met Hume's criterion (that we should not accept a miraculous
    > claim unless evidence forces us to conclude that /not/ accepting it
    > would be even more miraculous) and that it would even fail the more
    > relaxed criterion of an American civil jury.

    I suppose that, on a strictly Bayesian reading of the term, it is
    evidence; if psi existed, the probability of the prize being claimed would
    be greater than it is with psi not existing. Therefore, that the prize is
    not claimed, is evidence.

    But strong evidence or weak evidence? And is it evidence admissible by
    the standard of science, or even an American civil jury? I do not
    hesitate to hold CSICOP, self-declared carriers of the banner of
    rationality, to a high standard. They blew it with Gauquelin; they're
    gone. If psi did exist, it might appear on only a single occasion within
    a horde of fakes. No matter how many fakes you successfully detect, you
    cannot afford to get sloppy. If CSICOP rejects any claim, even a false
    claim, by a flawed process which would also have rejected a true claim,
    then they have blown it; all their rejections are worthless.

    > Gauquelin could easily force the issue by taking Randi and CSICOP to
    > court, and a million bucks on the line would make it worthwhile for
    > him to do so (and for some pro-psi lawyer to help). The fact that he
    > can't even meet /that/ test is significant, regardless of any personal
    > pecadillos of Randi and gang.

    I don't believe that Gauquelin was in pursuit of the Randi Prize as such.

    Even though I am personally still sure psi does not exist, the personal
    pecadillos of Randi and gang render the prize unsuitable as scientific
    evidence. They would not hesitate to dismiss a psi study on weaker
    grounds. The good guys must meet higher standards than the bad guys, and
    that is as it should be.

    I am sure that Gauquelin's results are either coincidental or indicative
    of something other than astrological influences. It is interesting how,
    if you search hard enough, you will eventually find *something*, or
    believe that you have... If psi is being held to a stricter standard than
    other areas of science, then this is indeed unjust. We should subject the
    other areas of science to equally hard scrutiny, or we are not safe. Psi
    demonstrates how easy it is to find effects hovering on the border of
    "statistical significance", if you are in pursuit of an ideology, if you
    run enough different experiments and don't count the failed ones as
    counterevidence against the validity of the field itself. Who knows how
    many other fields have accumulated the same false positives as psi, but
    without the skeptical scrutiny to which psi has been subjected? Consider
    psi as the control in a controlled study on the effectiveness of the
    standard scientific method: "We'll go hunting for something that
    *definitely* does not exist, and see if we can find evidence for it if we
    look hard enough and run enough studies." If psi researchers have been
    able to meet the criteria of science, then the standard scientific method
    is revealed as too lax. Perhaps we need to find a standard harsh enough
    to reject all psi experiments (as evidence for psi), then apply that
    standard to science as a whole.

    -- 
    Eliezer S. Yudkowsky                          http://singinst.org/
    Research Fellow, Singularity Institute for Artificial Intelligence
    


    This archive was generated by hypermail 2.1.5 : Thu Jun 05 2003 - 16:54:49 MDT