The methodology we employ in the domain of tic-tac-toe is generally applicable to many, if not all, domains of interest to researchers in artificial intelligence, and not just generally applicable to some, but not all, of the domains of games. Applications in every domain of interest to researchers in artificial intelligence have choices. Some of these choices lead to failure, however failure is defined. Whether the choices made are optimal or not, intelligent software should avoid making choices which lead to failure. One way to avoid failure is to not repeat the same or similar mistake made in the past. A database, however represented, can store these past mistakes, and the database can be accessed by intelligent software to determine if a current choice available to the intelligent software might lead to a failure. In this way, choices available to the intelligent software can be excluded, leaving only choices which, we hope, will not lead to failure because the given choice is not represented in some form in the failure database. After this process is completed, another artificial intelligence system can then make a choice between the remaining options, with the options leading to failure, hopefully, already removed.