@Frastlin this sort of technology does seem possible sinse it is based on us using specific interactions that the system can itself understand, and needing to learn to use those interactions, rather than the computer "reading your mind", though it would still be a good idega and as you said work in a lot of ways, though again a far distance from Neuromancer, Tad Williams' otherland etc.
@Cx2, I don't think saying "hgood science" or "bad science" is exactly the correct way of illustrating the problem, sinse the problem is far more serious than that as pretty much all! assumptions in science are questionable, including the assumption of the rules used to set those questions.
For example there is the problem of experimenters regress, see Good old wikipedia or basically how exactly when a scientist is expecting results from a given experiment, thos namely that in order to judge whether a theory is correct a scientist looks at experiments, however in order to judge whether an experiment has been performed correctly and that the results show what the scientist believes they show, the scientist will refer to the theory.
This isn't to say science is wrong, only that Dorkins is a really! arrogant knob .
I myself like the idea of Paul Feyerabend, that science itself is a constructed activity in which a bunch of competing theories are set against networks of ever more fundamental assumptions, and that the validity of those theories is about compatibility with the fundamental assumptions, which may themselves be wrong, but do at least give a coherent framework to measure the wrongness against, so rather than thinking of science as a continual progress up a hill towards some universal truth, it's much more like trying to weave a net with smaller and smaller fibres, occasionally having to discard the larger peaces, but hopefully coming up with something that will do the job in the end.
Getting back to less esoteric matters however, one development in control methods that I do believe will be coming quite soon would be a fully spoken interface. I don't mean vo, I mean an advanced form of Siri which could not just recognize the odd key phrase or word, but also act within individual applications to perform the same sort of actions as clicking icons. I actually think we'll start seeing this fairly soon, within the next few years, and what is nice is it will be a development which sighted users will have as standard, ---- sinse people are used to talking, and one which will also be great for access. How awsome would it be to be playing an rpg and say "Equip the ebony short bow and the dragon scale armor, then close the inventory and go to the level one dungeon"
or having an action game like Gma tank commander where you could drive and use the gun but could change weapons by saying "load the smoke canister" or the like.
With our dreaming and singing, Ceaseless and sorrowless we! The glory about us clinging Of the glorious futures we see,
Our souls with high music ringing; O men! It must ever be
That we dwell in our dreaming and singing, A little apart from ye. (Arthur O'Shaughnessy 1873.)