I'm not convinced it's worth _anyone's_ time to do so.
1. Master the wonky interface of some search engine.
2. Be disappointed by results of aforementioned wonky search engine.
3. Instead find results through rapid iteration of search terms.
I've worked on an in house search engine for the last year. User have been conditioned by google to just keep trying search terms till you find what you want. No interface (that I've seen) can beat the rapid iteration of search terms with fast response times.
A good reference: Information Architecture for the World Wide Web http://oreilly.com/catalog/9781565922822
Absolutely. Human curation feels right, because that's what we've been doing for so many thousands of years. We are multi-faceted, and so are our decisions on what information is useful and when.
At the thinner end of the wedge, though, are those information addicts. I worked on an auto-tagging system for developer content (live at http://coder.io/) for some time. It's turned out not to be as good as I had hoped (but at least I know why) but I'm not convinced that even developers are.. and I'm picking my words with trepidation here.. sophisticated enough to put together queries, filters or even collections of topics in the main.
Solely my opinion, though, but through these projects (and others) I'm finding folks prefer branded, curated digests of knowledge rather than filterable, 'queryable' firehoses. Either that, or I'm just better at providing the former ;-)