Back when I was in university, one of my professors described a similar approach he used for a vision system he built to classify objects in the 1980s; and it was primarily used as a way to talk about ways to work towards solving problems that were far from being solved. Essentially, the concept is to take a problem that is too complicated to be solved and removing all complexity until you could solve a much simpler problem, and then slowly reintroduce complexity and solve that problem; and in his particular case it involved painting objects with bright neon spray paint and placing them in an room that had nothing but the objects in it.
In this case in particular, I would doubt that anyone would ever make a product that required people to wear such silly gloves; but if they can slowly reduce the number of colours that are required, and determine what the core problem is with analyzing the position of the hands, they will be far more likely to actually produce a gesture system that works 5 or 10 years from now.
