Re: [sdpd] New poll for sdpd

["ROBIN SHIRLEY (USER)" <R.Shirley...@surrey.ac.uk>]

> Any member can propose a poll on this mailing list, and more
> alternatives are possible. But the question has to be precise.

The problem is not whether or not the question is precise, but that
it is absolute.

You ask about trusting *fully* and offer alternatives *always* and
*never* (plus a satirical one).  No room here for degrees of trust.

But in the real world there are degrees of confidence rather than
absolutes.  Thus it may make good practical sense to let a program
estimate the relative probabilities of a number of possible
spacegroups if its algorithm is a good one.  Similarly we may choose
to save time by relying on the judgement of a human expert, if we
think he or she is a sufficiently good one.

How much anyone will then rely on these findings will then depend on
how much time they will waste (or other real consequences they will
suffer) if the program's finding is wrong - just as it would with
the judgement of some other human crystallographer.

In the end, we all have to rely on our *own* judgement, even if that
judgement is that we should accept some other program or person's
evaluation.

Robin Shirley

-------------------------------------------------------------------

To:            sdpd...@egroups.com
From:          Armel Le Bail <armel...@fluo.univ-lemans.fr>
Date:          Fri, 27 Oct 2000 14:33:53 +0200
Reply-to:      sdpd...@egroups.com
Subject:       Re: [sdpd] New poll for sdpd


>I can't respond to this poll because I don't think the question is a
>reasonable one, nor are the alternatives offered sufficient

Any member can propose a poll on this mailing list, and more
alternatives are possible. But the question has to be precise. The
question could have been similar by replacing "software" by "human" :

Would you fully trust (without verifying by yourself) a space group
proposed from powder  diffraction data by a human being ?

And alternative answers may have been : 
     Always, 
     Yes if Robin Shirley or etc,
     Never.

In theory, science is not a matter of polls. In practice you may
need expert advices because >1000000 scientific texts are
published per year, all scrutinized by reviewers (=2 experts only).
How reviewers can let publish > 50 algorithms for Kalpha-2
elimination  is a mystery for me. A poll could ask to users which
Kalpha-2 algorithm elimination they prefer among the 50 published, 
and this could be a good information for beginners.

Just an example more. If a beginner finds an article published
in "Advances in X-Ray Analysis" in wich it is said "The 
deconvolution treatment is a very powerful tool for improving
the angular resolution", he may conclude : "wow, no need for
a synchrotron", when looking at the original powder pattern
and at the deconvoluted one, presenting FWHMs 10 times
smaller...

Because the reviewers accepted that paper, a poll
would be useful... (I may send the full reference on demand ;-). 

Being published is not sufficient proof (nor are polls), such is Science.

Best,

Armel Le Bail
http://www.cristal.org/course/

PS - Even from single crystal data, I prefer to have a look
myself, not trusting any software. That step is absolutely
crucial : wasting days or weeks is the consequence of an error.
So, what need for a software if you always need to verify
manually ?







-------------------------- eGroups Sponsor -------------------------~-~>
eGroups eLerts
It's Easy. It's Fun. Best of All, it's Free!
http://click.egroups.com/1/9698/14/_/40740/_/972657865/
---------------------------------------------------------------------_->