If code had some kind of embedded information about its own complexity, could that be combined with other code complexity information at runtime to build a deterministic model of what will happen? If so, could a process scheduler do a game of Tetris with work and timeslices to do a better job than a more naive scheduler? Is this logically sound?
