However, the information in many cases is out there. With LiveJournal's FOAF information, I can build a graph of who my friends are - that's simple enough. I'm in the center, and they're all at points equidistant around me. The next step is to add in the next level of Friendship, or knows data. For example, I know jpallan, and she knows pie_is_good, who I also know. In fact, jpallan and I share over 50 "friends" - people we both know/read. For those in the non-LiveJournal world, this is like sharing 50 of the same links on a "blogroll". So, obviously she and I are closer than, for example, myself and danbri, who share no links. Drawing the graphs is not the difficult part - TouchGraph already does that with information available via FOAF and other sources.
So, what to do with this data? We now have a representation of the "community" around me - drawn out to 2, 3, 4 levels, or deeper, depending on how far out you want to go. Now, depending on that community, I may want to read different things. For example, I may be interested in technology today - so, I'll stroll over to the danbri side of the fence, or over to the group surrounding jito. Tomorrow, I may be interested in reading more personal things - so I'll look at jpallan's entries, and from there wander out to pie_is_good.
Right now, everything in my aggreagator comes linearly, chronologically. However, this isn't always the way that I want the information. I want groupings, communities and neighborhoods. Geocities had it right - group people by interest, and let them stroll through them. The community and neighborhood aspects of the site back in the mid 90's might have been ahead of their time then, but now, with the massive amount of information, it's perfect.
Generate my friends mappings graphically. A 3d map, that I can stroll through, click on, and have it show me content. A grouping of people discussing, with the ability to listen to them individually. Something that allows me to view the content I want in the order I want it. Rather than linear, I want groupings, I want selections. I want the ability to seperate the content by relationships between people, not just between people and me. The data is out there, and it can be used in this way. Bring in more content - let me find something new through my trusted friends, rather than through some random google link. Use RSS + FOAF as transmittal scheme, use graphing techniques like touchgraph to create the information set. Then let me move through the information in a way that is non conventional. Let me browse through people, read what they have to say - and read more from people like them.
The idea of walking up to a group of people that I know appeals to me much more than reading a list of entries written by people I'm not entirely interested in. Show me a community based network model of aggregation, and I have a feeling you'll be showing me a killer app for the semantic web and how it works.