I've yet to see a use of this that didn't look or feel like a gimmick. The cube game is the first thing I think I've ever seen game-wise that I might not get bored of in 10 minutes.
While I like the idea of adding such a feature, I hope it doesn't come for a while yet (unless someone else outside scirra adds it - Construct is open-source after all.).
Sounds like a post-1.0 feature.
Construct already has webcam support, and i've linked to a great computer vision framework that is open source -- integrating this open source framework shouldn't require nearly as much effort as them developing their own implementation on their own.
the PS3 game Eye of Judgment actually brings legitimate advantages through this technology. computer vision is what makes it possible to use the same cards in local matches against other players, local matches against computer AI opponents, and online matches.
this technology is largely unexplored for games, but i think there's definitely potential for cool things. what mainly prompted this thread's creation was me playing with the webcam object in Construct and seeing that it is basically useless -- you cannot manipulate or utilize the video input in any way... it is just there. and this computer vision stuff is open source -- if i knew my way around c++, i'd try to do it myself. sure, it's not something that should be a top priority, but i think it should be a priority. it would not only expand on the types of games and applications construct can make, but it would expand the potential audience of construct. there is a lot of value in an intuitive development environment that can utilize a wide variety of inputs.
edit: there's even a python client using the open source framework --
http://code.google.com/p/pytuio/