I liked Rubrick's link explaining hysteresis. It does a good job of explaining it. A "hysteresis loop" is what I was thinking of when I first read it, where if you perform an action, and the perform its opposite, you don't end up where you started, because of hysteresis.

Imagine dragging a reasonably obedient dog around by a 6' leash, and that you start with the dog 6' to your left. First you walk him 20 feet to the right. This means, because of the leash, he's only moved 14 feet to the right. Then you go back to where you started, 20 feet to the left. Buddy the dog is still going to be 6' to your right when you finish, because his rule is "always walk behind my owner at the full length of the leash". So although you have taken yourself exactly back to where you started, because of some lag in the system (where the system = you + Buddy the dog), the whole thing is not quite back to its initial state.

And of course, you can think of ways to imagine this problem where you do both end up in the original state, but there are also many starting points, combined with the "lag" produced by the rope and the fact that this obedient little dog only trails behind you , where you don't both end up in the same position. Does that make sense?