Well, when it comes down to it, surface computing will not be optimal for the typical paradigms that we have grown to love/hate with keyboard/mouse interaction. A single-touch device is basically applying/mapping an already solid interface (the mouse) to a lesser approach. The whole point of multi-touch technology is to break away from the typical one-process-at-a-time task, and move into the realm of a computing adapting itself to the user's preferences.
I've had the chance to play around with the MS surface table... It's true that there isn't much there besides (for me anyway, an advanced computer user) the flash. However, watching the more typical user (and especially kids) interact with it, was a blast. Most multiplayer games require separate controller-type gameplay, but with surface we had our game immersed into a combined space.
In short, I think that multi-touch is a much better approach to computational collaboration that the single node per user approach... but one has to realize that it's still in its infantile state (for a typical, off-the-street user).