I’m going to try to make the abstract argument against 3D interaction paradigms more strongly now.
Obviously when you are dealing with a 3D object, as with Medium, Quill, or a 3D data visualization, then you have a 3D interface, but I contend that the majority of browsing, configuring, and selecting interactions benefit from designing in 2D.
Splitting information across multiple depths is harmful, because your eyes need to re-verge (and, we wish, re-focus). This is easy to demonstrate if you have a convenient poster across the room in your visual field above your monitor – switch back and forth between reading your monitor and the poster, then contrast with just switching back and forth with the icon bar at the bottom of your monitor.
If 3D interfaces were a good idea, we’d already be using them.