All you people complaining about the state of Akonadi and the interconnectedness of KDE are missing an important part of the picture.
No matter what OS or desktop or GUI apps you have been using for the past 20 years, all of them have an implied conceptual data model underlying the collection of apps that you are using. For the most part, the "implied" bit has followed the old X11/Windows 2.0 "shared nothing" model, where bits and pieces of the data has been either replicated or ad-hoc shared (apps explicitly being aware of other apps and sharing data with them). This has worked okay for a long time, since users have assumed that this is the way it will always work.
Some of you folks may remember PalmOS, or are using Symbian and/or Android phones now. These platforms make the data model explicit, and all apps are aware of them. On an Android phone, the SMS, IM, Calendar and Contacts apps all know how to access the user's address book without knowing anything at all about each other. If you write a contacts sync backend, you don't need to care about any of those apps - you know how to access the contact data already from the database API.
What ends up happening is that YOUR DATA becomes the central focus of the platform. Not the app's data, or its storage format or anything like that. The apps are relegated to being clients of the larger data model. This is the correct approach for the future of the UI.
The KDE project has had its eye on this model of the future UI for a decade now. There has been bumbling and disorganization, miscommunication, dozens of key people coming and going, cat-herding and unfinished code. This is hardly surprising considering the size and nature of the KDE development team and its organization. Despite your irritations with the current state of a given KDE release, I think it's worth remembering what is being attempted here.
Of course, everyone's threshold is different and you may decide that being part of this massive experiment is not worth the headache. For you folks there are plenty of alternatives and I don't think anyone is going to be upset if you go back to the 80s-style UIs of other desktops such as KDE 3.x.
At the same time, don't be surprised if the KDE project doesn't make it a high priority to enable picking bits and pieces to run elsewhere; it's worthwhile, but it comes with a high opportunity cost that dilutes effort on the larger goals of the project.