Describe a method in which one can generate a dataset that satisfies k-anonymity, has the same size as the input dataset, offers reasonable utility, yet offers very weak privacy protection.