There is no better proof for simulation than the fact we all have to sleep and still can not explain why.
It's obvious: Someone is trying to save on simulation time. You don't sleep? You get punished for using extra simulation time. If you do sleep, you are fed faulty simulation data that is not good for anything else and also some random a/b trials. These highly localized simulations are much cheaper in terms of processing, as details aren't usually important, and can sometimes be skipped entirely depending on the individual.