One day Lorenz wanted to print out a previous sequence again. To save paper, he made it start reprinting in the middle of the sequence instead of the beginning. He entered the first three decimal places of the sequence and left it to print. When it finished printing, he looked at it and found out that the sequence that was printed was different from the sequence he expected to get. In fact, it started accurately, but slowly diverged to a different pattern until at last it ended up completely different. .
After studying the problem, he realized that the reason this happened was because he only entered the first three decimal places accurately, 0.506 while the computer recorded up to six decimal places. (The number of the sequence he wanted was 0.506127.) So the difference, 0.000127, was enough to change the output significantly. .
From this, Lorenz determined that his program had sensitive dependence on initial conditions. That means that, if in a calculation, you enter perfect information except for one tiny error, it will be compounded on and eventually the end result will be wrong. If the initial conditions in an experiment result are not precise, the end may turn out completely wrong. And for initial conditions to be precise, all variables must be entirely accurate. However, Lorenz decided it is impossible for all variables to be measured to complete accuracy! For example, even if you measure something to the nearest millionth of an inch, it is not entirely accurate because it can still be measured to the nearest billionth, trillionth, et cetera. From this idea, Lorenz determined that it is impossible to accurately predict long-term weather.
However, since Lorenz was the first to come up with this idea, he decided to test it. He oversimplified the equations in his program to three equations as opposed to twelve, to the point of surrealism. Hence, the program's purpose was no longer to predict the weather, but to create a formula that was sensitive to initial conditions and seemed to give rise to chaotic behavior.
