I think that would improve chatter -- or at least make the following mechanism more relevant. But chatter is not a feature I often use. As for feeds, I have been thinking for a while on unfollowing people and sticking to domains and tags. I have decided I'll never do it, for two reasons: one, I use hubski to discover new domains and can't do this well without following users; two, tags are too heterogeneous for me to ignore all the new interesting ones that crop up, which I would never run across if again I didn't follow actual users. [A profile toggle allowing us to change our feeds to just domains/tags is however a Good Idea.]

My followed list rarely changes. At this point it might work to prevent me from seeing content rather than ensuring that I do.

I really doubt it. You follow 50 people, including at least 10 who are "prolific sharers." There's probably no way to put a hard number on this, but I bet at least 90 percent of all hubski posts (that get shared at all) are "eligible" for your feed, either through direct following or indirect sharing. That's just a function of the current community size.

EDIT: at some point I would love a post from you summarizing your experience to-date running a content aggregator. Did you anticipate the amount of thought you'd have to put into every single feature decision? It's fascinating that the dilemma isn't whether people will use tags or how difficult the programming will be, but rather what each small change does to the site's unique dynamic.

EDIT2: sorry this is getting long, I genuinely think it's all worth reading. Having now read kleinbl00's comment -- the next conversation hubski needs to have is about tag compartmentalization and homogenization. Having six different tags that all vaguely mean 'news' isn't stable. The solution isn't a dropdown menu where you can only pick one of ten things, but also the longterm solution isn't the current "anything goes." Like said above, content is too hard to find.