ABSTRACT
In some contexts, like for example complex games and puzzles, the search for solutions to problems leads to the discovery of different procedures, none of which can be considered ‘the best’ one. In these contexts, in fact, the different solutions can be compared only in some specific well-known domain of application, while in other domains the comparison is incomplete or vague, or the domain may have imperfectly known boundaries; it is therefore impossible to put the different procedures in a precise preference order over the entire domain of applicability. Moreover, these procedures may be ‘locally stable’ because small local changes in the instructions that define them do not lead to any improvement, and therefore individuals that discover one solution may remain locked in it without trying to search for alternative solutions.
