'Disorder' is probably a better way to think about it than energy and perhaps information (I'm weak on information theory; I know it has similar concepts though

'Disorder' is probably a better way to think about it than energy and perhaps information (I'm weak on information theory; I know it has similar concepts though). So, if the gasses simply reversed places, there's no net gain in disorder. But it's certainly proper to think about it in terms of microscopic, reversible processes that tend to result in the most statistically likely final states. The second law "could" be violated, it's just not very damn likely.