Review of "The tracker: a threat to statistical database security, by D. E. Denning, P. J. Denning, and M. D. Schwartz", ACM Trans. Database Systems, 1979

The authors point out that queries to the system compute raw statistics for a query set. Questioners with prior knowledge of a given individual can divide their queries into parts such that when reassembled into a special characteristic, formulas will isolate the proprietary information about the individual. These formulas are called trackers, The paper reviews the literature on the extent and limits of trackers, provides excellent information to show how they can extract individualized information and so subvert the confidentiality of database files. The paper also reviews individual trackers and defines ranges within which they work, and shows how statistical responses to queries do compromise confidentiality of information, positively or negatively; positively if the individual falls into the given category and negatively if not. The authors then proceed to develop a general tracker and show how it can be applied within restricted subranges to extract individualized information. Under conditions where this fails, they show how a double tracker can be used. In general, trackers for most statistical database systems exist within the database, or can be easily obtained, thereby compromising the confidentiality of the data.