Did I hear him say "gaze detection" near the end of the video? that's also a big deal I haven't heard anything about anywhere else.
I certainly get the impression from the way he says it that this is something they are looking at the moment in a more advanced way and is a common conversation point in their software/hardware dev teams at the moment. VERY exciting.
I wanted to use my eyes to create 3D models via eye tracking. Looked into it for about a week before concluding it was probably impossible, or at least way less accurate than good old-fashioned hand movements.
Dunno what the Oculus guys are doing with gaze detection, but I doubt they're doing fine-grained detection such as being able to pick out a particular target with your eyes. (I would love to be wrong about this! That'd be an awesome feature. I just doubt whether it's technically possible.)
The way to convince yourself your eyeball isn't moving is to touch your actual eyeball with your fingertip -- like press lightly against the left corner side of your left eyeball with your left index finger -- and then switch your gaze back and forth between the two letters. As long as your viewing distance is correct, you won't feel any movement. (You may have to back your head away from the screen. I'm viewing from a distance of arm's length.)
On the other hand, if you look back and forth between adjacent words, then you can feel physical eye movement.
I'm assuming you've seen some peer reviewed research that fixed peoples' heads and used high resolution eye tracking to support your hypotheses, though?
I've trained myself to be able to do this with vertical lines, like window blinds.
Both are about $100 and have dev kits/SDKs, although I haven't looked closely at either.
I can't find it by googling, but a long time ago I used to read far too many Flight magazines.
One of the problems of immersive VR is that you need very high precision near the fovea ( = at the point of screen you are looking at), and you need to render very fast at the edges of vision. This means that you need to render at a very high resolution and at a very high frame rate. This is, of course, a problem.
However, while the vision at the fovea is very sharp, it is also very slow, and while the vision at the edges is very fast, it is very blurry and imprecise. With eye tracking, this is exploitable. The practice is called foveated rendering, and the idea is to only draw full precision near the position of the screen you are actually looking at, and elsewhere render at a much reduced resolution but at a higher frame rate.
In tests, this has been found to be impossible to distinguish from just drawing the full screen at full resolution and speed.