Unfortunately, there still isn't one way to do perfect search of all important parts of any of these databases. When searching social bookmarking databases, I start with http://tagcentral.net which, similar to Gataga, searches tags over a large number of tag employing databases.
Furl's native search, however, does offer some important advantages. I've been thinking lately about how important it is to search not just tags, but other fields as well. Furl uses a weighted search algorithm that includes search of user applied keywords, the comment and clipping fields and the full text of the article itself. When you search your own archive in Furl, you also search the full text of the cached copy. This helps with document recall by going beyond the limited tags already applied by users, something that evolves over time and is a good compliment to full text search rather than a substitute for it.
One other feature of Furl is that it allows you to view your archive and the archives of other users by entry dates, but also rating and number of page views. I know that articles from my blog, for example, having been furled by either me or another furl user, have sometimes been viewed via Furl search results up to 30 times. I can easily see which ones are being viewed most and least via Furl in my arhive with one click. This is part of the reason I Furl my own articles (I save in delicious too), so as to get them into those databases right away. No need to wait for someone else to submit/save them.
I wrote about this as a part of my blog posting routine
here.
Yay for this article's top placement in TechSoup again. It's so interesting. I like to read industry mags like KM World and InfoWorld, and think "how is the nptech community striving to achieve the same functionality with much lower budgets?" Social Bookmarking is one part of that, I think.