It wouldn't be hard, probably, to create pre-programmed surround with, as you say, only half the available channels as a result. But trying to get the system to dynamically decide which sounds are where and then do the timing on the fly would waste a lot of CPU cycles. That kind of work is better suited to a DSP. Further, surround sound is much more useful for 3D games, IMO. Why do you need to know what's behind the player when left and right pretty much represent the spectrum of sound locations quite well?