I like what Marco is saying here. I don’t totally agree with him about the fallacy of the river approach, but otherwise he’s spot on. The river style RSS content flow is relevant to high-noise (i.e. high output) sources. There will be some articles you don’t want to miss but the majority of content you can skim or skip. But I don’t think that it disqualifies these feeds from your reader. A good reader should be able to differentiate the high value content from the low (for any particular user). Dave Winer loves the river paradigm, and handles the value proposition by creating multiple rivers.
An alternate approach is to use folders or tags to segregate content, though you still end up with Inbox Overload Syndrome. I like a recent suggestion by a selfoss user to be able to specify some feeds as “noise.” These would be available in the reader but not as part of the unread articles. Thus the interesting stuff doesn’t get drowned out by the river.
The best system would be able to dynamically adjust what’s shown based on user preferences and usage history. Kind of like Google Reader’s old “magic” reading order or the way it showed your most read feeds on the home page. Not many reader projects have the resources to fulfill a request of that nature. For now I’ll settle for tagged feeds that can be filtered of “noise.”