All the popular aggregators are starting to look the same, and don't do much beyond supporting feed reading. I've probably said this somewhere before, but it would be nice to have an open, domain specific, scriptable (internally and externally) platform for slurping, slicing, and showing webfeeds and equally important meta information and analyses of these feeds. Much in the way that Emacs is not only a text editor, but a platform for building applications that do a lot of text munging.
Memo to self: find more round tuits.