I understood what the article was saying, but it seems I didn't understand what you were saying. If I understand it now, you are saying that to prove his case he has to not only show that entropy can't directly decrease by, essentially, undoing the process and getting back to the original state, but he also has to show that entropy can't decrease by the system ending up in a state different from the first, but yet which is also of lower entropy? It seems to me that you are correct.