The point of my post is that its already being done better with today's technology and that this proposal is not an improvement. Currently, mechanical designers & engineers are using a combination of a 2D & a 3D mouse. The 3D mouse handles the pan/scan/zoom on a large-screen LCD and the 2D handles the fine control challenges. The 3D mouse is not technically necessary, as the same functionality can also be achieved using modifier keys in combination with the standard mouse, but it does make things smoother. So really this system isn't solving any problems.
If we really want to do it better than today, we need some sort of eye-tracking and/or brain-scanning to eliminate the lag of accurate selection. But there are still a lot of challenges, such as gorilla arm which posters noted. There's no clear solution to the fact that humans aren't design to hold up their arms for 8 hours a day...except maybe modifying ourselves!