This practice sounds like complete the opposite of the principles of scientific research.
Of course. Publishers aren't in this for the science.
The only thing that KDE4 has accomplished to date is to offer less features than 3.5, and make everything slower and a little more mouse dependent.
Agreed. I just finished a 10-day trial of Kubuntu 9.04, after having used KDE3 versions of Kubuntu since their first release. It was pretty, for sure, but really felt like every visual enhancement came at the sacrifice of some functional enhancement. Example: Icons of folders containing pictures in KDE4 have tiny thumbnail pictures on them, but no longer could I hover over a picture to see its dimensions or JPEG comment. Granted, KDE 3.5 might be the most feature-packed, customizable desktop environment of all time (for better or worse). Recreating every one of its features is a huge task, and it's going to take time. I can keep using KDE 3.5 while I wait.
Another megabytes the dust.