It's funny. I was around from KDE 1.1 to about 4.7 and some of those decisions were things I was involved in directly... Like the branding shift towards KDE as a community (people sharing vision and development infrastructure) as opposed to KDE as a monolithic desktop environment. I haven't been involved for ages now. I did some coding too, but not a tonne.

The KDE 4.0 release messaging was one of my core tasks. We had a release party in Mountain View -- and we invited all the packagers for all the distros to the event. Linux community "luminaries" like Patrick Volkerding were there and it was a great party. But we thought that, by bringing all the packaging types there, we had the messaging problem bottled -- and KDE 3.5 and 4.0 would be offered alongside each other as though they were different desktops entirely. (Like Gnome, or whatever... Just choose what to launch in your session manager.) What happened instead is that 3.5 was dropped like a hot potato and users fled 4.0. Distros didn't want two versions of libraries installed, so running a 3.5 app in a 4.0 environment was difficult, but not all the 4.0 apps had been ported yet. Yikes! This is a huge reason for the subsequent split between version numbering of Desktop releases (later Plasma) and things like "Frameworks".

Side note: we had even considered the idea of KDE (as a community) offering multiple desktop interface offerings, each with their own branding. So you could run Plasma, Kicker (a hypothetical KDE 3.x desktop environment ported to the current frameworks), etc. alongside applications from multiple versions of desktops. This was the reason the session management code was in KRunner rather than Plasma, for example. This would allow highly experimental user interfaces to be developed around the KDE libraries. But that never happened, as far as I can tell.

Anyway, for 4.0 -- so much for Linux applications and the mantra "release early, release often." Lesson learned. Linus, I'm so sorry for disappointing you. ;)