Its a matter of use cases. This is clearly intended for much more accurate, gesture control. Kinect has much more range, and has 2 camera inputs including depth, (color and infra red). Kinect is much more appropriate for full body or half body interaction modes, face detection, etc. The leap strikes me as a much more nuanced, fine tuned and highly accurate device meant to augment and replace the mouse.

I'm sure you can get some sort of raw depth out of it, and can possibly turn it sideways to do face recognition, but time will tell. Kinect Near Mode is more accurate, but this is claiming 100x accuracy. Even if its 10x better, thats a huge improvement for lots of use cases over the kinect.