Main problem is there is that as all information used by the leap is continuous in nature there is essentially no (forseeable) way to unequivocally convey the intended start and end of a gesture.

I personally imagine an interface not unlike a wand - people might carry around a pointing stick of some kind, with a button built into the handle to confirm gestures either on or off (at least until we can start doing crazy things like gauging intent - what if the computer only responded to your motions if it knew you were looking at it?).

Probably a terrible idea.