As for two specific examples: Learning Ruby on Rails has been a long, slow, struggle for me. A number of reasons for that, but one of the main ones being that I don't have many other programmers in my area that I can talk about it with. Although there are scores of blogs and tutorials on Rails, it's hard to know which articles are good, and nobody has the time to read them all. So if I was looking for a post on, say, integrating jQuery with Rails, I could do a quick search at http://www.delicious.com/tag/rails+jquery, and I'd find a bunch of articles that had been selected by individuals, for their own use. It was great for finding high-quality content.
Another example deals with niche interests. Delicious made it easy to suss out who is talking about $randominterest. Just by going to http://delicious.com/tag/randominterest, you can see who else is bookmarking it, who's writing about it, and so on. Especially when dealing with, say, fringe programming languages, or uncommon design details, Delicious makes it easy to find people.
I wonder if this could be implemented as an API of websites? (Or perhaps on behalf of websites?) Instead of one single site, you'd have a p2p network of sites using protocols similar to those used for digital cash schemes. Think: a distributed open marketplace of middlemen. A sort of Diaspora middle tier.
I have more than 2k entries there collected over the course of the last five years. I only bookmark sites that are for some reason important to me, and have some lasting value (eg. I don't bookmark most HN submissions). I tag everything I save and some of my tags have special semantics that is probably only useful to me.
The difference between Delicious and news aggregators like Reddit / Digg / HN is that the rating of links is a side-effect, not a conscious action.
As a result ranking on Delicious does not prefer yellow journalism, obscure interests or vanity displays; much like Google's ranking but small enough to not be a target for promotion through SEO.
In other words, the whole social aspect.
The main problem was that the RSS feed was limited to 15 entries.