Since GNOME began making usability a priority in the first years of this century, it has improved out of all recognition. However, even the best design principles can be rigidly or inconsistently applied, and usually lose out when they conflict with policy. Here are seven places where GNOME could use improvement:
No Control of Notifications: In KDE, you can control which notices appear as pop-ups, and which of four broad categories display. By contrast, in GNOME, you have to endure the whole barrage. The result is that, in some GNOME-based distributions, the messages can come so quickly that at times you might almost imagine yourself on Windows.
Obviously, you want to be notified of any serious problems, and to avoid interrupting processes that are still running, but many notifications are irrelevant to the average desktop users and can only distract them from what they are doing.
So why not give users some control over notices?
Unfortunately, however, the decision to limit the default menus to a single sub-menu means that the second level of the menus is simply an alphabetical list, and can quickly grow long as applications are added. In some cases, the Debian menus, which can be four or five levels deep, are actually quicker to navigate because, although they require more clicks, they are better organized.
Moreover, within in the System menu, the distinction between Preferences and Administration is frequently vague -- even the rule that Preferences refer to the settings for the current account and Administration to those for the entire operating system is not strictly enforced. As for the Places menu, it could be replaced by a good file manager, if only it could be deleted.
In the past, GNOME has emphasized applications like Beagle or Trackers as ways to locate files, and, in GNOME 3.0, apparently intends to feature Zeitgeist, an application that locates files by dates. All these efforts might enhance a file manager, but as replacements they seem far more complicated than the system they are intended to replace.