Until now, only password surveys have been analyzed, and I knew the results would be almost useless when compared to a real dataset of passwords because who in their right mind would donate one of their own passwords, or anything like it, to a survey about password statistics?