I just received an update of articles posted to the cs arxiv at www.arxiv.org. For those of you who use blog readers for everything, Hein Roehrig maintains RSS feeds of most (if not all) of the topic areas maintained at the archive. Yet another email that I can eliminate from my mailbox; more time to read spam ;)
This spurred a thought that has been tickling me for some time now. Why is it that we in CS don't have a comprehensive archive of all papers published in the field ? INSPEC has often been the database search engine of choice, but it is usually accessible only from non-web-based interfaces, and is (in my opinion) secondary to google/citeseer searches nowadays. However, it has often happened that because I didn't look at the right sources or ask the right people, I missed a reference that would have been useful in my work.
Now if any of you have ever visited PubMed, you know that this is an incredible resource for biologists. Every paper from every journal (as far as I know) has a PubMed entry with an abstract, and can be searched for. There are also links to the actual document (usually via the journal web site). Biologists publish an awful lot of papers; when I was doing computational biology work every week there would be new papers that I'd have to track down and read, and having a resource like PubMed gave me the perfect one-stop-shoppping experience.
Math has MathSciNet, and physics (possibly parts of it) seems to have adjusted very well to the e-print server model. Why not CS ? One problem in my view is that the success of arxiv.org depends on people voluntarily submitting their papers. For PubMed (and possibly MathSciNet), this is done automatically (or via a third party) whenever new papers are published.
p.s Lance Fortnow has a hilarious description (with pictures) of how a Ph.D defense in Amsterdam is conducted.