Small starting differences leading to big final outcomes are explained in Chaos Theory. Basically it stems from instability. It is like modeling the weather. If you start a model weather program with current stats at exactly 01:00 GMT and run it for a week; then come back and start the same program with the stats from one minute later at 01:01 GMT, you might not recognize the new result after another week of running. It is commonly called the "Butterfly Effect". If a butterfly flaps its wings off the west coast of Africa, then Florida gets hit with a hurricane three weeks later!
