I want to point out that Wolfram doesn't think the universe is a giant cellular automaton. He gives CA as examples of how easy it is to produce impermeable complexity. His more serious proposal is a graph-based system, though I haven't perused enough to know if that's also for the sake of example or if it's intended to be taken as a serious physical theory. Most likely it's an earnest attempt at a heuristic towards the goal of, in his terminology, "finding the program of the universe."