But it seems like the real problem he's trying to solve is that current ranking algorithms don't take into effect the fact that "users" are not one segment, but rather composed of different segments with differing political, religious, sexual, ethnic, etc. tastes. That is to say, Digg's algorithms are very good if you match a stereotypical Digg profile. If you weren't, well, it wasn't so amazing.
However, this is _hardly_ an unexplored area, and I would further submit that _Amazon_ is surprisingly good at this kind of thing. By analyzing what random samples of users bought (or, in other cases, ranked up or down), they're able to make (IMHO) often-insightful recommendations about what else you should buy. I've had thoughts about how you could make a site that would kick Digg's ass and probably be more valuable to advertisers using tagging, ranking, and some statistics, too.