Real science, hot off the web

A while ago I commented on an Economist article claiming that Web 2.0 tools "were beginning to change the shape of the scientific debate."  My contention was that the web wasn't so much changing the debate as changing the means of publication.  In particular, there had always been a trade-off between speed of publication and thoroughness of review, and the web was becoming a publishing mechanism of choice on the lightly-reviewed end of that continuum.

More recently, looking for something I no longer recall, I ran across Cornell's arxiv.org (I assume the x is meant to represent a Greek χ), a repository for "Open access to 703,281 e-prints in Physics, Mathematics, Computer Science, Quantitative Biology, Quantitative Finance and Statistics."  The number 703,281 was current when I scraped it. It's probably higher by now.

That's a lot of articles, but does anyone use it for anything important?  Well, one recent entry is Superluminal neutrinos in long baseline experiments and SN1987a (Cacciapaglia, Deandrea, Panizzi et. al., yes, those neutrinos).  Indeed, there seems to be a lot of activity in the experimental high-energy physics section overall, which makes sense.  It's useful to have experimental results available quickly, bearing in mind that there can be quite a bit of calibration, number-crunching and checking before an experimental paper is ready for public consumption (months, in the case of the neutrino paper).

Submissions to arxiv.org "must conform to Cornell University academic standards".  It's not immediately clear to me what process is in place to ensure this, but from a little browsing it's clear that these are serious academic papers.  It also seems reasonable to assume that most of the papers have not been through the full process of peer review required for publishing in a major journal.  Indeed, the published version is almost certainly not going to appear on such a site, if only for reasons of copyright.

It seems like a good niche to fill.  If you have a significant result that you're comfortable sharing with the world and staking your reputation on, there should be some way to make it immediately available, with the implied tradeoff between speed and thoroughly careful vetting.  Publishing under the aegis of a major university gives everyone some assurance that you're at least doing real research.  I notice from random sampling that the reference sections generally don't cite arxiv.org, giving some indication of the preliminary nature of the publications.

With that in mind, arxiv.org looks like a great resource not only for working academics but for the curious layperson as well.