Linux at the end of the world (our 2012 predictions)

Little wonder then that capriciously changing those defaults might cause great waves of displeasure.

I'm sure the coders would tell you to s/capriciously/carefully and after extensive study/, no matter how you feel about the changes. I remember the massive storms of protest with default spatial Nautilus, for example, where the UI people told us that spatial was better and we were all wrong to want browser-style, and they had the studies to back it up. On the other hand, I've been generally impressed by some of the intrusive changes I thought I would find most annoying switching from GNOME 2 to GNOME 3.

For example, I expected to hate the full screen activities menu. Instead, I've found that I use it for tasks that are inherently disruptive, so taking over the whole screen isn't as annoying as I expected. Meanwhile, it makes good use of the full-screen space, so it really is easier to use than a launcher menu, task bar, or workspace switcher. And it goes away completely when I'm done with it, so I have less clutter and more space on my desktop the rest of the time. Maybe the guys who said it was a better way of doing things and wouldn't let me use my old setup actually had a clue of what they were talking about.

Also, the ability to comfortably use someone else's desktop only works if the features of that desktop are relatively stable from release to release.

Now here you're getting into to some murky waters. A new major version of a big project is going to have some big changes in it, and not all of those are going to get it right on the .0 release. Fixing your design mistakes means changing user visible behavior, while leaving them means allowing the mistakes to go uncorrected. Neither one is a perfect choice, but I'm inclined to accept that fixing the mistakes is the better course, especially for something that is intended to serve as a stable platform for a long time.