Well change seems inevitable because developers want the same multi-touch apps for all the new phones to work on desktops without redoing the interface. So the PC is going to need multitouch. So either the screen goes multitouch (which it has in some cases), or the input devices do. Since touchscreens have issues with things like smearing and comfort distance, that leaves the interface devices. Multitouch pads have been done, but most people still prefer mice. They're more precise due to the size of the working area, and easier for certain tasks like dragging because of the extra degrees of freedom on the arm/elbow which frees up the fingers for clicking instead of overloading them for both position and input.
Of these candidates, the cap mouse is most likely to win out, followed by the orb mouse, which may see a competing run in the high end. Why? Let's see:
FTIR mouse: This is basically an internal reflecting material like a lightpipe or fiberoptic cable. The problem is it limits the mouse because it requires this kind of material (think the demo uses acrylic), and design such that the camera can always see it. The shape has poor balance, CG, and drag properties, and will probably result in breaking or issues sliding for many people. The restrictions to mouse design will annoy existing manufacturers, unlike say optical sensors, which were just drop in replacements for mouse balls.
Articulated mouse (Arty): Not happening, for a simple reason - people won't want to readjust to left/right click being thumb/forefinger instead of index/middle. It sounds stupid, but believe me it will be a showstopper. Plus the design is a bit fragile, and I'm not sure on the ergonomics of having to extend the finger and thumb like that, seems like an RSI issue waiting to happen.
Side mouse: This has some potential, but it will be plagued by unintentional inputs. Any time you drum your fingers impatiently, drop a pen on the desk, move the camera too close to something sitting on the desk, it will go nuts. It might be useful in cases where you can't build a touchpad into a device, but in most of those cases the device is so small you want to hold it not rest it on a desk anyway, so there'd be no surface for the side mouse to track on.
Now for the showdown between the two serious contenders.
Orb mouse: Really nice input image. Can easily do a variety of applications with it, since there's so much area. Datacenters sometimes use illuminated vein pattern recognition for biometrics, which can be efficiently integrated with this, and it's a better solution than those stupid touchpad fingerprint readers. But for more conventional apps it's got the most area, the best shape to exploit the use of all fingers, and in deference to the mention of clock-based positioning on the Gizmodo article about it, will probably be the easiest for people to extend thinking to. The main showstoppers are cost (not sure) and bulk/shape issues. People may not find the bulgy shape appealing though I suspect it will test well with male audiences.
Cap mouse: Probably going to win, despite the low resolution sensor image. Why? That "$1 gesture recognition" on the video says it all. Not the gesture support part, the $1 part. Cost wise it's probably cheapest, and it seemed to work sufficient for the apps in the demo. It's also just a bolt-on to existing mouse designs. No need to modify the existing shape or ergonomics to accommodate it, which means it's the path of least resistance. If it's also the path of least cost, which given most of the rest need a camera-quality sensor it most likely is, then the winner seems pretty obvious.