I’ve been at this since Suse 7 (ppc) with kde2.0, circa 2000:
http://www.kde.org/screenshots/images/large/kde2b3_5.png
KDE is now, finally, kicking butt. At first I’m like “Nepomuk, Strigi, Akonadi…yeahright”. Now, not so much. The Personal Information Management (PIM) stuff is working extremely well. There are several KDE desktop/laptops here running Kontact, all using akonadi to centrally store (remote MySQL db) calendar events, to-dos, journal entries, addressbook data… _damn, does that solve a huge data sync problem. Enter an appointment or to-do on the laptop, it shows up on the desktops (after a refresh). It works.
Lot of complaints out there about Strigi file indexing consuming processor cycles, and OS X users complaining about Spotlight. To mitigate indexing load, reduce the scope of the indexes (System Settings > Advanced > Desktop Search > Advanced Settings) and/or renice your nepomukstorage processes to 19 (htop is handy for this sort of thing). Or, it may not even be a good time to bother with Strigi, since it’s got a lot of work to go yet. What’s cool is semantic technology, and laying groundwork to leverage it.
http://techbase.kde.org/Projects/PIM
http://techbase.kde.org/Projects/PIM/Akonadi
http://www.omat.nl/tag/akonadi/ (developer blog)
http://techbase.kde.org/Projects/Nepomuk
http://strigi.sourceforge.net/
http://nepomuk.kde.org/discover/user
http://techbase.kde.org/Projects/Decibel
http://wiki.kolab.org/index.php/Kontact_for_Windows_%28beta-huge-debug%29