Originally Posted by Clear
The more data-points sampled the worse it gets, particularly in relation to anticipating and validating user behaviour. The issue is always that you're not dealing with absolute states (i.e. button-up, button-down) you're dealing with fuzzy behaviour due to user error and imperfections of form in performance.
Again, this is a critical flaw in an interface, because that "fuzziness" can be extremely undesirable based on circumstance. You can afford to be sloppy in a performance/play scenario like dance or certain sports, but in a fictitious "pressure" situation like operating a mech under fire... its never, ever, going to be optimal.
The whole idea of emulating physical mech controls with a non-physical gesture interface was strange (to say the least) from the beginning. And the parts where you have to push virtual buttons to change ammo when you already have buttons on your pad are borderline absurd.
I don't think the fact that the user may mess up his gestures in the middle of tense action is a problem in itself though (it's part of the skill required when playing the game, just like mashing buttons will get you killed in most games).
So Steel Battalion was flawed from the beginning indeed. There are a few good ideas though (judging from the demo), but most of them are based on being more intuitive, rather than having complex actions : holding your hand like a visor to activate the binoculars, grabbing and punching your fleeing partners... the periscope is ok, but most other commands aren't really adapted to motion controls. I think they really should have kept the motioin commands for minor tasks and QTEs.