Also of course, my first post on this article says why I don't think "GNOME should learn from GNOME 2" is right. Unless you think the main goal of GNOME 2 should have been to avoid flames. GNOME 2 became much more popular than many desktops that did _not_ get flamed (as much). Taking a lesson away from GNOME 2 like "we shouldn't change stuff" or "regressions are never good" would be historically wrong, because either of those lessons would have doomed GNOME during 1.x -> 2.x. Knee-jerk "CADT bad" would have doomed GNOME too. Truly targeting mainstream users instead of Linux users would have doomed it too. There are lots of ways GNOME 2 could have gone wrong. But flames were non-fatal.
It doesn't mean that if you're getting flamed you're going to succeed, but it does mean that you _might_ be right to do something that gets you flamed. You also might _not_ be right. That's the hard part. One needs to figure it out.