in my experienceNow, infrastructure is a huge issue in any distributed and 'many to many' environment. Technology adoption and standardization are other concerns, and is why I don't have great hopes that I'll have more fine tuned semantic grouping of news/blog feeds. Fine tuned you ask? What I mean by that is grouping a bunch of feeds together in NetNewsWire does create some sort of semantic connection. But it's on a macro-semantic scale with sorting based on scale of a website, not the subjects of the containing content. Don't get me wrong, the groups idea is outstanding, but I want more.
What needs ("needs" = "what I want") to happen is to apply standard Dublin Core (DC) meta data to ALL news and blog entries, which means every content/document management software package needs to be updated, and people need to apply the DC elements to their content. The resulting feeds in RSS (preferably full RDF) would include this info and allow applications like NetNewsWire and AmphetaDesk to parse that data and allow the user to sort the info as they see fit.
Nice and easy, right?