Although we imagine that the course of program running in a computer can always be predicted, this is actually to confuse projection with retrojection. It is one thing to say that whatever a computer does can be traced back to its program (along a very possibly torturous track indeed); it is another to say we could trace the program forward in the same way. These are problems of initial conditions and real-time realisations; unless we know precisely the state in which the (complex) system begins, and predict the outcome of every potential random fluctuation it takes into account faster than the process itself, tracing forward is not possible. therefore, if the initial conditions are vague and/or the path the program follows is torturous, or our simulation too slow, the only way to arrive at any distant future state of the program may be to run the program and wait. this is called 'solving the problem in real time'. On such a supposition, even if we were programs running in machines, it might still be impossible (given even slightly vague initial conditions) to state exactly how we might be in any given 'now' (and correspondingly impossible to specify the states of thousands of millions of semi-independent organismic 'units' at any future time). The specification of the exact state of a sufficiently complex computer system at some future time might, on this reading, be beyond even God.
-God and the Mind Machine, John Puddefoot