Well, we could argue fruitlessly about whether I'm missing anything (obviously the motion and gunk I described aren't bothering contact wearers or they'd complain), but the muscles and motion of the eye are spectacularly weird. The superior oblique muscle connects to the eyeball through a tiny bone stirrup to produce the kind of rotation that you mention, which is the weirdest anatomical tidbit I've run across. http://en.wikipedia.org/wiki/File:Eyemuscles.jpg

I'll have to respectfully disagree with at least the first and last of your rebuttals. I've seen lots of eyes blown up on video screens while tracking, and contacts absolutely float around on the cornea. I'm painfully aware of this because the edge of small contacts (gas-permeable?) frequently interferes with eye-tracking in one of two ways: creating changing refractions of the edge of the pupil; and/or reflecting light itself and creating false positives when the tracker attempts to detect a corneal reflection. This is not to say that contacts are completely untethered, but they certainly move around quite a bit, and I imagine more than enough to make a tiny embedded screen useless without sophisticated control. As to keeping contacts clean by blinking, that's just silly. Why would they need to be cleanly nightly if they were being cleaned thousands of times each day? Believe me, there's plenty of crud on contacts that blinking doesn't budge; it's the very first thing we ask people to do when there's an issue, and it only helps sometimes. As to the orientation maintenance, that's news to me, but I'm pleased to hear that somebody cracked it. It also makes the left eye/right eye near/far correction I encountered even less comprehensible.

I love the idea, but having conducted many eye-tracking sessions I think that there are a number of basic challenges that the contact-lens implementation would need to overcome. First, contacts drift all over the cornea. It's really quite disconcerting to watch, especially when you've got obvious markers like vanity-coloring for irises. Station-keeping over the pupil would require technology that's not mentioned here. Alternatively, you might monitor the portion of the contact that's positioned over the pupil at any given moment using photocells on the underside to catch light reflected off the retina, but then you'll need a bigger display and use up valuable real estate. Second, similarly and perhaps even more importantly, orientation maintenance. If there was a good way for contact orientation to be maintained, bifocal contact lenses would be a reality already. Instead you encounter bizarre stopgaps like a reading lens in one eye and a distance lens in the other. Third, eyes are a fairly disgusting environment. Crud on regular contacts doesn't seem to impede vision, but the delicate electronics in a device like this might object to working in a wet, salty, bacteria-ridden setting unless very carefully insulated. Nothing bad about reaching for the dream, but I'll take a pair of prescription Virtual Light-style glasses while we're waiting.

Could we just switch it to GnIMPh and be done?