The "desktop paradigm" you've described is important, but it is important because it is needed to foster FUTURE software development designed with the NIT platform in mind. I certainly hope the primary goal of Maemo 5 is not to enable reuse of existing desktop apps (ugh!).