open-source software for bioinformatics (was Re: Unix vs Linux - the movie.)

Kevin Karplus karplus at cse.ucsc.edu
Tue Aug 15 05:44:04 EST 2000


I disagree with the following:

>   Given that:
>   Peer review is done to make sure the conclusions of papers are
>   'correct'.

Peer review is not ever going to ensure that papers are "correct".
What peer review can (and does) check are that papers are reasonably
clear, that the information presented supports the conclusions, that
some attempt at citing relevant previous work has been made, that the
work appears to be original, and that no obvious errors were made.
Theoretical papers may be checked a little more thoroughly for errors
than experimental ones, since mathematical derivations are easier to
check than experimental results.

In some cases, where an experimental design is particularly weak, or
where conclusions are made that are not supported by the evidence
given, peer reviewers will suggest additional work that is needed to
justify publication.

Only in cases where results are astonishing are peer reviewers
expected to ask for further supporting evidence.  

Reviewers are asked to review dozens of papers a year, with no
remuneration.  It takes 1-10 hours per paper for an adequate
review---nowhere near enough time to read 100,000 lines of code, as
John Anderson thinks reviewers should be doing.




-- 
Kevin Karplus 	karplus at cse.ucsc.edu	http://www.cse.ucsc.edu/~karplus
life member (LAB, Adventure Cycling, American Youth Hostels)
Effective Cycling Instructor #218-ck
Anything below this line is junk added by others without my approval.







More information about the Bio-soft mailing list