$newsid = ''; ?> del.icio.us is a "social bookmarks manager" and the hot new web-based app of the moment. It's a bookmarks manager because you can use it to record and annotate web URLs as your surf (a bookmarklet makes the process as easy as saving bookmarks within your browser). It's "social" because your bookmark list is visible to other del.icio.us users and as an RSS feed.
Did I say an RSS feed? Actually del.icio.us lets users or RSS subscribers see the bookmarks arranged in numerous different ways. The top del.icio.us page is a ticker tape of whatever users have been bookmarking most recently. There are views by keyword (e.g. CSS, language or art), by user (riddle or blogal villager), and by a combination of the two (my CSS bookmarks). In its most "social" application, you can even use it to see who has bookmarked a particular item (say bookmarks for the Rosetta Project) and perhaps find people with common interests.
del.icio.us is by Joshua Schachter of memepool and GeoURL fame. del.icio.us is in "pre-pre-alpha" state so it would be premature to criticize any missing features. My main concern is that, as with GeoURL, del.icio.us appears to have several problems of scale from the outset. One is that it is aggressively chronological in its presentation: that's how you want to see things if your interest is in catching the latest memes to hit the shore, but not if you want to use it as a reference across a reasonably long time span. Another, of course, is that del.icio.us currently has maybe dozens of users; assuming it can be made to function for thousands of users ("a simple matter of programming" as they say, and hardware) there's still the usability question of how to make a Mississippi of bookmarks into something from which one can take a satisfying drink.
Subject access is strictly on the basis of whatever keywords users enter without even an accommodation for spelling errors, let alone a controlled vocabulary or a subject hierarchy. Of course the simplicity of raw text keywords can be seen as a feature rather than a bug from the point of view of an individual user, who can use whatever set of subject buckets he or she prefers and won't be slowed down when making a bookmark by having to navigate someone else's complex and all-encompassing subject system. However, the idiosyncrasies of dozens or thousands of individual keyword choices severely limit the collaborative value of the system. (If you say "socialSoftware" and I say "social_software" how can we collaborate?) A thesaurus, perhaps even an automatically generated one, might go a long way to solving this problem.
Regardless, Joshua has come up with an ingenious proof of concept and a highly addictive toy. It's especially fascinating right now while the early adopters are putting in a lot of high-quality links (there's not even any spam yet!) Take a dip and you'll have trouble not checking back throughout the day to see what new goodies have been bookmarked.