I'm pretty sure EEG's never gonna be good enough for the kind of cursor control gamers are thinking about. People have been trying to work with the signals for a long time, and even though they can get crappy 2D control now, the signal to noise ratio is just too high for anything much better.
Whoops, looks like I misremembered - the patient in that movie wrote by blinking at a grid, not via P300. My bad.
This system has been around for a while; I've seen it demonstrated live twice, and it didn't work at all either time. In my opinion, even in best conditions (bald patient, shit-tons of electrodes, professional setup, well-trained subject) it doesn't work well enough to fuel science-fiction fantasies, and probably never well. For locked-in patients, who can do nothing but move their eyes, though, it's an awesome technology. They made a movie recently about such a patient who spent years using it to write a novel: http://www.imdb.com/title/tt0401383/.
Here's how it works - metal electrodes on the brain (EEG) pick up an analog signal, and *any* stimulus which is particularly salient to a subject creates a spike in the signal 300ms after that stimulus appears (this spike is called the P300, there's a good wiki article on it). If you have a dude staring at a grid of letters, you can tell which one he's looking at by hunting for the big spikes 300 ms after the right letter flashes. The only problem is the signal-to-noise ratio, which is notoriously terrible in EEG, though of course there are people out there working on improving it.
"Today's robots are very primitive, capable of understanding only a few simple instructions such as 'go left', 'go right', and 'build car'." --John Sladek