Recently I was at the map interaction workshop to the Crisis Management Laboratory (inaugurated by the President of the European Commission José Manuel Barroso on 16 March 2012 link) of European Commission Joint Research Centre showing and talking about PixelSense and Kinect technologies and there was SR Labs that works from many years with eye tracking.
We was able to test the technology and first of all, for a better result, it needs some dedicated hardware that may cost a lot. You can also use a simple cam but eye tracking is all about image processing so with dedicated hardware there are other sensors that helps the image processing tracking eyes such us infra red camera.
Imho eye tracking tech is very good and important but used in a passive way and only in some particular cases in an active way.
What I mean for active is that using the eye tracking tech to move the mouse, scroll a document, or what ever that can be done with touch or mouse it is really not confortable, even it it is doable and we tested it, but with a mouse or a touch device you are one million time faster.
There are some particular cases where the use of this tech in an active way it is really important: just think to someone affected by amyotrophic lateral sclerosis or to a doctor doing a surgery that can't use the hand to interact with the system.
Using the eye tracking in a passive way could be really important just think to an operator that is looking to a map of a disaster: using eye tracking tech and a dedicated software the system could be able to determine witch areas are affected by the disaster and if the operator didn't look at it the system could highlight it.
For those of you interested on the argument here are some good resources: