Xref: utzoo sci.bio:1366 sci.misc:2164 sci.research:432
Path: utzoo!utgpu!attcan!uunet!husc6!cs.utexas.edu!ut-emx!ethan
From: ethan@ut-emx.UUCP (Ethan Tecumseh Vishniac)
Newsgroups: sci.bio,sci.misc,sci.research
Subject: Re: Strange results in Nature article
Summary: Huh?
Message-ID: <4536@ut-emx.UUCP>
Date: 27 Jul 88 18:56:19 GMT
References: <10465@lll-winken.llnl.gov> <20850@beta.lanl.gov> <2444@cxsea.UUCP> <2301@boulder.Colorado.EDU>
Organization: The University of Texas at Austin, Austin, Texas
Lines: 87

In article <2301@boulder.Colorado.EDU>, pell@boulder.Colorado.EDU (Anthony Pelletier) writes:
> In article <4520@ut-emx.UUCP> ethan@ut-emx.UUCP (Ethan Tecumseh Vishniac) writes:
> >I see in today's newspaper that an investigation by Nature  discovered
> >that there were at least two problems (fatal problems) with the
> >laboratory protocol.  First, the experimenters knew at all times
> >which solutions were which (i.e. control and experiment) and
> >this is well known to produce biased results.
> 
> This does not sound right.  I hate to suggest that you do someting radical
> like read the original paper, but they state in the paper that each initial
> tube was coded with two different codes by two different pairs of observers
> (not actually doing the experiment).  They do point out that, not long after
> the experiments were begun, it was possible for the experimenter to tell
> which was which from the results.  This is a problem that is hard to overcome,
> believe me.  Even if someone else labels your samples in code, the experimental
> sample can have an appearance so distict that you can tell which it is.
> From that point on, you know which tube is which.

I have read the original paper.  My statement was based on the article
in the NY Times.  I am aware of the conflict between the two.  I don't
see any way for me to judge who is telling the truth except by the
credibility of the sources, in this case a group of fairly credible people
on all sides.


> 
> >Second, the laboratory
> >notebooks revealed a large number of cases that produced negative
> >results but were not included in the statistics. 
> 
> This could be a big problem; or it could be that it does not work all the time.
> Do your expermints work all the time?  Unless you have never tossed a
> data point, don't be too quick to judge.  
> But, this requires some explanation.

I'm a theorist.  My experiments never work (or they always do) :-)
I have also listened to a large number of very impressive experimental
talks whose results have evaporated in subsequent years.  I try to
be both slow to believe and slow to disbelieve any particular result.

> 
> >This is all supposed to come
> >out in Nature in the near future.  The author (Benveniste ?) is
> >standing by the work and has denounced the investigation as sloppy
> >and unprofessional.
> >
> >As with any truly startling result, the sensible thing is to believe it
> >only when such questions have been cleared.
> 
> No argument there.  I am a bit surprised that someone who seems to be proud
> of his skepticism has assumed that "the Amazing Randi" and Co. have
> the definitive, final word on the accuracy of a scientific work.
> What did they have to say about the fact that 6 independent labs got the
> same result?  Or, does that mot matter?  

I never said that the Amazing Randi has had the last word.  I implied
his criticisms need to be convincingly answered.  Surely this is a modest
request.

As for the independent labs, again there seems to be conflicting information.
The investigation team from Nature ascribed almost all the positive results
to a single researcher working in two different labs.

> Let's wait until all the data are in--I don't think they are yet.

They never will be.  However, I remain quite willing to alter my stance
in light of further information.
F
 i
  l
   l
    e
     r

      F
       o
        r

         t
          h
           e 
              machine
-- 
 I'm not afraid of dying     Ethan Vishniac, Dept of Astronomy, Univ. of Texas
 I just don't want to be     {charm,ut-sally,ut-ngp,noao}!utastro!ethan
 there when it happens.      (arpanet) ethan@astro.AS.UTEXAS.EDU
    - Woody Allen            (bitnet) ethan%astro.as.utexas.edu@CUNYVM.CUNY.EDU