Oh the one where you have to wear funky gloves and wave your arms around a lot? The wide arm movements are a dead give-away. It's MADE FOR MOVIES. Not for actually using it.
Stop and think about that last point. If you're five or ten feet from a wall in your office and say you place a vector that runs perpendicular to your palm. It takes a surprisingly small amount of hand movement to get that vector to hit any portion of the wall.
By the time i had gotten used to the system, i could use it usually by performing completely minimal gestures so much so that people didn't even realize i was engaging it explicitly. This is similar to how you can see people playing wii tennis on their couch without standing up. [1]
If you read that sculley interview[2] you hear him talk about commoditizing the performance to create consumer products that react the way you expect them to. For me, gspeak convinced me that this was basically the case with a gestural interface, the hard part was figuring out something to do with it. I suspect this is a hard issue for many people because the majority of the complaints i see are basically "so what, i can't see myself using it" and fair enough, answering that question is really hard. But the underlying problem of how to provide that interface isn't your problem anymore, just like you don't have to figure out how to track touches or mouse movement anymore, you just build on top of that as a given. that seems like a net win to me.
A large part of gspeak for end users and designers ends up being the development of a descriptive and non-conflicting gestural language for the task at hand. The folks at usc have a good example of what i mean [3].
Say what you will about wild gesticulating or how lame it seems, the system (as any other) is only as cool as what you do with it.
[1]: http://www.penny-arcade.com/comic/2006/11/13/ [2]: http://www.cultofmac.com/john-sculley-on-steve-jobs-the-full... [3]: http://www.vimeo.com/8083372
I'm tempted to say "At least it would be good for games", but frankly the Wii has failed to revolutionize gaming as much as folks thought it would -- after waving your arms around for a bit the novelty wears off and you go back to wanting to mash buttons.
Not me. Button-mashing is tedious. I like standing up, pointing and waving and clicking and twirling. It's engrossing when done well; for example, Wii bowling or Wii table tennis. Or any game with a sword fight. :)
For example, Boeing has patents related to controlling a swarm of UAV's via a g-speak based system. There are more applications where that came from but this was the only example that is public.
Its collaborative features are a huge draw. Would you ever use photoshop via gestures, probably not, but you would use it for command and control systems with more than one concurrent operator.
I recommend that you go to oblongs website and watch some of the demo videos. A computers interface can be anything from manipulating a physical object on a table to "wearing funky gloves and waving your arms around". Its not limited to just a keyboard, mouse or touch pad.
The real pain comes when you try to apply 3D interfaces like g-speak to problems that are not inherently 3D. You get awkward interaction and a lot of wasted time. Real applications of 3DUIs include medical operation planning, virtual reality, scientific data visualization and architecture, just to name a few. Also, in most cases, gestural interfaces are not the most efficient way to do it.
If common desktop applications do, at some point, evolve to meaningfully take advantage of 3D visual interfaces, it could trickle down to a system that assumes a more natural posture (try resting your arms on your desk and gesturing above your keyboard: I think that could work). We've had hardware-accelerated 3D on the desktop for a decade, though, and have yet to see anything substantial come of it for traditionally two-dimensional UIs apart from sometimes-helpful transitional animations. Things could change, but I think applying gestures and multi-touch to 2D (a la 10/GUI, naturally) shows the most promising way forward.
For any extended period time, or for anyone with even minor physical disabilities, it's wretched.
Probably easier to teach new cops on the squad.