There are more doubts I have with their methodology: "A VM will get a 16GB disk drive, [..] 128MB of VRAM". These seem byzantine and may trigger all sorts of be

There are more doubts I have with their methodology: "A VM will get a 16GB disk drive, [..] 128MB of VRAM". These seem byzantine and may trigger all sorts of behavior that isn't tested for or should be tested for. Also, memory consumption is measured with `free`, which is even more evidence that not just the DE is measured, but the entire system. They only test Fedora Spins, but that is no assurance the rest of the config is identical between them. Something about lies and statistics ;) If the proof of the pudding is in the eating, then simply use the DEs of your choice for a while and see what's optimal for you. Don't do silly things like measure memory consumption with `free` on a VM once, and draw general conclusions.