Nice example of an algorithm that is simple enough to explain in a few lines, but still changes how you think about floating point errors.