It’s an open source project by the talented Samuel Clay, who has been working on it for 15 months now. He’s currently working on an iPhone version, but the regular version is very impressive in itself.
The sources are hosted over on GitHub, and the project uses the Django Web framework for Python. This means that you can probably install it on your own server and break free from the big G. But unlike Fever, which is another server-based reader, NewsBlur has built-in multi-user support, and it’s free. So, you don’t have to install it on your own server to use it; you can just go to the NewsBlur site and open an account.
NewsBlur lets you import your current feeds from Google Reader. It lets you vote articles up and down, and it uses your input to figure out what you might like to read in the future. It then marks articles appropriately so that you can (hopefully) read just the stuff that you’re interested in out of the sea of RSS feeds that you’ve been reading.
To really see if the intelligence works, you have to use it for a while, which is something I haven’t done. This is a very impressive project, but I feel that the lack of documentation (not even an About page) is holding it back. I would have given you a complete screenshot tour of it, but it’s very difficult when you don’t know what anything does.