There's so much dispute about bypassing because, on multilayer boatds with ground and power planes, almost anybody's bypassing schemes will work. Usually, no bypass caps at all will work.
I have no hard theory or sim tools that let me understand these things, so I add SMA footprints to boards and TDR them unpopulated, or measure plane noise working. A power-ground plane pair, with no bypass caps, TDRs as close to a perfect capacitor as I can resolve, at the obvious calculated capacitance. Adding bypass caps *anywhere on the board* seems to make it look like a bigger perfect capacitor.
Any time I've done actual boards according to this theory, they've worked. ECL, FPGAs, GHz analog, whatever.
John