I’ve been working at a company which almost entirely uses the GNOME/Linux desktop for over four years. One thing that’s struck me is the extent to which desktop churn affects users, and I wonder if usability / UI people underestimate the cost of making changes.
As a truly trivial example, consider the GNOME Preferences system. This is pretty much just a menu with a list of about a dozen different categories of things you can configure, such as your proxy settings or desktop background. Building categories that make sense and presenting them to the users is hard, and each time we update Fedora release on the desktops things have been moved around as people try to improve matters.
These small changes generate a surprising number of support requests. The benefit of moving something between categories, or renaming a category is arguably outweighed by the problems it causes for existing users. A lot of people use computers by remembering by rote a string of menu selections (Preferences -> Desktop -> Appearance), quite a few people have notebooks with such things written in them.
I’m not trying to say all change is bad, but perhaps minor fiddling is probably not worth the cost. I wonder if this is one of those issues where the people doing the fiddling aren’t the people being phoned up as a result, so don’t care :-)