On Wed, Mar 21, 2001 at 06:11:11PM +0000, Tim Brody wrote: > As an example of an "Open Archive" that has had ample opportunity to be > filled with rubbish; (correct me if I am quoting wrong), arXiv has, in its > ten years, only had to delete 2 papers out of 160,000. This would suggest > that either arXiv has a very efficient staff or this is not really a > problem (or, as I suspect, both).
I would say that maintaining minimum standards in the arXiv is a problem, but one that has been solved. The arXiv mainly works on the self-respect system. Most serious authors have respect for their own work, and the arXiv takes certain steps to reinforce this principle. First, in order to register an author should provide some evidence that someone in the research community will be interested in his or her submissions. Any real academic affiliation is sufficient and the arXiv assumes it if the author has an academic e-mail address. It's not perfect since occasionally someone obtains an academic e-mail address without any affiliation whatsoever, but it works pretty well. Otherwise the author needs a letter of reference from someone in the research community. It doesn't have to be anything like a letter of recommendation for a job, just some kind of expression of interest in the author's work. Second, the component archives are moderated, mainly for the purpose of fixing misclassifications. At this level the arXiv doesn't keep out work that is merely wrong or weak; it might still be relevant to research. But there are separate categories for submissions that defy classification, and that includes material that makes no sense to the moderators. Third, submissions are irrevocable. You can always submit a new version, but all versions remain available. So you have to live with your mistakes; the best you can do is submit a withdrawal notice asking people not to read previous versions. This three-tier system has nothing to do with the handful of deleted submissions. Deleted submissions are things like conference announcements, garbled files, and duplicates. Besides that, I'm not sure what the original poster had in mind as "rubbish". Maybe half of the articles in the math arXiv are ho-hum works that would never interest me. But that's not the same as rubbish; most of these papers are legitimate but boring. Maybe 5% are so lame that I would be embarrassed to have my name on them. But even most of these are on-topic and publishable. On the other hand, the arXiv does have some excellent papers that will never be published, or that have even been rejected by a journal. I think it's important to put peer review after permanent archival. > Your suggestion, to me, does seem a rational one (and indeed currently > exists between arXiv and the APS - I believe the APS will accept > submissions using arXiv papers), that there are archives of "pre-print" > papers which are then picked up by validating services (i.e publishers) > which then "repackage" archives into validated subject/editorial content. This is a major controversy surrounding the arXiv right now. It is important for journals or other vehicles of peer review to validate research papers. But why bother repackaging them? -- /\ Greg Kuperberg (UC Davis) / \ \ / Visit the Math ArXiv Front at http://front.math.ucdavis.edu/ \/ * All the math that's fit to e-print *