It is not the same. With KDE 4.0, a few months before it was renamed to a developers release. From what I heard, it was pretty buggy and not everything was complete. This is not how I see GNOME 3.0.
However, there is a big change and comparing such a big change with the 16th release of the 2.x cycle, and not with 2.0 for me is very logical that it doesn't work out. I see 3.0 as really good and way better quality than I expected.
You're comparing to KDE 4, bringing up ME, and suggesting things. Then stating that data collection doesn't matter. With any big change, there is pain. You have loads of articles written about the effects of change, etc.
We've had pretty much the same feedback during GNOME 2.0.
Without any data, no idea how 3.0 compares to 2.0.