Steven posits that future aggregators should be heading in the direction of providing only relevant information rather than making us do all the filtering ourselves.

I don't know Steven. I can't say I'm interested in aggregators that make relevance decisions on my behalf. All I want is a single mechanism to look at all *potentially* relevant sources. I can only imagine how much I would miss if what I read became based on an algorithm, even if that algorithm is constantly updated to reflect my reading patterns. Maybe it would be helpful if the program was able to highlight items it believed were *more* relevant, leaving the ultimate decisions to the end-user. I think the future of librarianship is still predicated on the idea that humans will always be the most subtle filters of relevance. Now an aggregator that helps me locate and identify potentially relevant sources, that would be something.