For a start, it precludes a sane implementation of focus-follows-mouse.

Secondly, for frequently used commands, keyboard shortcuts will be faster anyway (I'm aware of the study you're probably going to mention, but plenty of other studies have had different conclusions).

Lastly, and this is pure conjecture on my part, wouldn't having the menu bar at the top of the screen result in a much narrower arc from the current mouse position to the desired entry? The theory also seems to ignore the fact that in most cases you'd want to move the mouse back to the original window, or hit some specific (and often quite small) target in that window. Which is now made more difficult by the mouse being farther away. Just some idle wonderings, they were quite possibly addressed in that study.

I know that you started your post with "Whether people like it or not isn't really relevant. It's that this is the correct way to do it.", and thus are unlikely to be swayed from your opinions; it's worth considering though that every choice has pros and cons, and sometimes the total value of each side very much depends on the person.

(although I share your "Screw the people, this is /correct/!" view when it comes to tau: no-one's perfect)