It's probably a good idea to concentrate maintenance efforts on fewer kernel trees. There should probably be some kind of synchronization going on, i.e. declaring a new "stable" kernel when there are enough major Linux distros currently basing on it.
This is a good idea. I have been on 2.6.32 forever in Debian unstable, and it's starting to get old. Yes, I could compile my own newer kernel, but then udev breaks, and then I've wasted 5 hours :)
This is very good. Sometimes it was some kind of guessing game to choose a kernel release to use in our environment.