On Mon, 11 Dec 2000, Greg Kuperberg wrote (in part; the subject line was Re: The preprint is the postprint):
GK> However, it is not quite true that the arXiv is completely unfiltered. GK> Rather, the system has the absolute minimum of filtering needed for GK> self-sustaining quality. [snip] GK> So what kind of filtering is there? Each category has moderators, and GK> I am one of about 30 in the mathematics section. We have a few hours GK> to review submissions. If we do nothing they are automatically GK> posted. We are not allowed to censor any remotely relevant submission, GK> no matter how wrong or trivial it appears to be, nor do we want to. GK> GK> However, we can reclassify a submission if it is off-topic. We can GK> reject a submission if we can't recognize it as research at all, for GK> example if it is pornography or a non-mathematical autobiography. We GK> can reject a submission if it has the wrong form, for example if it is GK> only an abstract or an unannotated bibliography or unannotated data. GK> And we can intervene against spam, e.g. an author who divides one GK> self-contained manuscript into 100 submissions. GK> GK> Most of the moderating decisions reclassify legitimate submissions GK> with strange classifications or excessive cross-listings. [snip] GK> The system is similar to the conventions for informal seminars in the GK> non-electronic world. Anyone can attend, except for people who are so GK> clueless that they would disrupt the talks. I found this information from Greg Kuperberg to be *very* interesting. So, should one criterion for the evaluation of the quality of preprint/ postprint servers be the existence of (as a minimum) a "filtering system" analogous to the one described by Greg? I'm assuming that, at present, the arXiv network of servers provides one appropriate "gold standard" against which other preprint/postprint servers should be compared. I'm also assuming that preprint/postprint servers will continue to provide one valuable way (but not the *only* way!) to free the peer-reviewed research literature. Jim Till University of Toronto
