CCL: Science code manifesto - just publish the code

 Sent to CCL by: Konrad Hinsen [hinsen**cnrs-orleans.fr]
 On 18 oct. 11, at 20:32, Vincent Leroux vincent.leroux- -loria.fr wrote:
Wherever access to the code is necessary to reproduce results, it *must* be provided one way or another (e.g. as supporting information) if it is not freely available already, this is just plain obvious to me. 
 Wondering why a manifesto would be needed in order to do just that...
It is needed because "just that" does not happen today. I get a lot of computational science papers to review, but none of them had any code attached (not even small analysis scripts), nor any explanation of how I could obtain it. Even for my own articles, I rarely provide scripts because most journals don't want them. One day I'll put all that on my Web site.
As a reviewer, I insist that authors provide a full explanation of all algorithms they use (usually through literature references of course), and that they state which versions of which programs were used. I often ask them to consider making code and data available, but that has happened only in one case. I would like to be able to request such information more firmly, but I don't see much of a basis for that at the moment: it's neither a general habit nor something encouraged by the author guidelines of the major scientific journals. 
 --
 ---------------------------------------------------------------------
 Konrad Hinsen
 Centre de Biophysique Moléculaire, CNRS Orléans
 Synchrotron Soleil - Division Expériences
 Saint Aubin - BP 48
 91192 Gif sur Yvette Cedex, France
 Tel. +33-1 69 35 97 15
 E-Mail: research at  khinsen dot fastmail dot net
 http://dirac.cnrs-orleans.fr/~hinsen/
 ---------------------------------------------------------------------