Wired:

Several decades ago, Edward Lorenz noticed something seemingly minor: his computational weather models yielded very different results if he changed the input variables only a tiny bit. Lorenz published a paper on this topic in 1963 and that led to the modern field of chaos.