> It's a contradiction! I mean, it's basically an infinite regression: either something happens outside of your control or you control what you control what you control...
But you can have a process interacting with itself in a nonlinear way, and the outcome of those types of processes are difficult to predict.
Even if you write a small non linear program, with each instruction completely simple and clear, it's still very difficult to predict the outcome without running the program.
Now imagine that at a scale ten or twenty orders of magnitude higher.