Well, another developer and I were discussing the general presentation of the modern application and the statement “A GUI should not be fun.” was made. I find this quite interesting, especially because the examples that we have had in the past, and the more modern examples we are seeing today. Such as Google.com, it's just a really smart box that you can query for information. When speaking on Operating Systems, I personally agree with the aforementioned statement, I think that a user interface, should be functional, and then it should be pretty. “Go before show” is a phrase that I've used with my friends in conversation involving fixing, upgrading, updating, etc. cars.
On the topic of three-dimensional operating system graphical user interfaces, I have done much research. I read through Microsoft Research’s' approaches at the issue. I have read papers by many professors and researched the source of a couple of versions of 3D windowing systems in Linux. One thing I have found in common with most of these systems, besides the fact that everything is in 3D, is that they are quite fun. By fun, I mean that they all contain workspaces that are entertaining in some fashion by the arrangement of 3D objects considered your desktop. It may look like something, or some place else that you have seen. A desktop that is a desktop, or a hallway that your desktop is composed of, even a living room, how fun.
My issue with this fun way of computing is that the people trading stocks or running massive projects that involve over a hundred different physical sites don't need the cute features of one of these systems. I don't need a fuzzy bunny to deliver the email nor do I want to learn to translate from a 3D coordinate system (some would say four, some would say n) to a 2D coordinate system (with a mouse) back into a 3D coordinate system. Especially if that 3D coordinate system has the freedom of movement and selection that it required for a usable 3D GUI.
Gloves always seem like the natural progression when it comes to moving to more advanced interfaces. The interface however is too free and will have to be limited to be useful as possible. Series of simple movements could express multitudes of data, and keyless and dummy keyboards would realistically be created along with this device, if not something in software, something on screen. Why not use the freedom of all of this space to provide well-researched, well-thought out and even faster methods of organizing on-screen data as well as faster methods of expressing and manipulating data? I'm sure we some day will.
Really the entire subject is in its infancy, but I have spent much time contemplating it all... Maybe I should write something moving in this direction...