Read about Edward Lorenz's work. He dealt with the weather, he was a meteorologist. He had modelled (rather over-simplified), the weather system as a set of simple differential equations. He had set out to achieve complete predictability, if not for the real weather, at least for the computer model. But surprisingly, he found that the computer model itself was unpredictable. He found Sensitivity to Initial Conditions (SICness). A situation, where even the best of computers cannot really predict the long-term behavior. Yes, by building powerful computers and algorithms, by considering more and more factors, and by doing more and more accurate calculations, we may be able to improve weather prediction from hours to days to probably weeks. But the system as such, in real long term, is unpredictable. I hope I do not sound out of context. Hmm... Maybe I do. I'll stop here.