Ivan is right on the money in saying we must build exploring software rather than searching software. You can red blogs and their contents on bloglines but can you find and extrapolate what you're searching for?
I'm convinced we must target a company like Microsoft and given them a way to find and compare data contained in blogs. On a corporate intranet which I saw, they were clearly wanting to collate data from blogs, but it was useless.
I'll be honest and say that reBlogger is good. It uses include and exclusion of keywords to ensure it's always on signal (there's no interferance) but what I've come to see is that when I'm trying to find lots of info on a particular topic, there is still so much info on there, it's hard to collect next to each other exactly what I want.
So we're always on topic and on signal – but the signal waxes and wanes so much. We need a way to bring good content to the fore and drop bad content to the back. Yahoo knows this and is buying up social applications left right and center, because people are indentifyng what content is good and what is bad.
Technorati also sees the need to find what sources are better than others. They have an authority slider. Interesting perspective here. But it's a filter. It reminds me of this awesome Amazon AJAX Diamond search.
No, I really think filtering and popularity is not far enough up the tree. We need to go higher.