It depends which point of view. If I made a movie of the simulation and then played it backwards, the gas would go into the box. There is nothing in your expe

It depends which point of view. If I made a movie of the simulation and then played it backwards, the gas would go into the box. There is nothing in your experiment that says this is wrong. In your experiment, you must wait, and therein lies the problem. You've implicitly put a forward arrow in there.