Because this kind of new visual interface is scaleable. You can operate 1000s of objects without a problem.
This is something your app centric user interface would start looking like Times-Square.
The Desktop-Metaphor is not scaleable with the physical space, because you need to remember the digital<->physical links in your own brain.
You can read a bit more about this problem here: http://openhybrid.org/learn%2c-setup%2c-operate.html