Having to take the mouse to an area of the screen that doesn't have any UI clues that something may happen is not a mouse paradigm. If I show you a window that doesn't have the red X in the top right corner, would you still take the mouse there and click anyway or would you not even bother?

It's not particularly a touch paradigm either. It's simply a "new" convention introduced by Windows 8. FWIW, moving to the mouse to the corners or edges to trigger "invisible" actions isn't unprecedented, for example that's how you show an auto-hidden taskbar (ok, so there's like a 1-pixel edge visible) or show tabs in a fullscreen IE in Windows, and Mac OS X has a "hot corner" feature that has you set actions for the corners.

Fitts's Law says that elements along the edges are easier to hit with a mouse, and elements in the corners are easiest to hit with a mouse.

There's nothing in Fitts's law that says that invisible UI elements are preferable to visible ones.