There is No Such Thing as a “Trial and Error Strategy”

Abstract

The concept of “trial and error strategy” is pervasively used in the literature on robot programming: in particular, it is often claimed that some novices use a trial and error strategy, while others use a more rational and reasoned strategy. Here it is argued that the concept of “trial and error” is of little use for the analysis of the strategies adopted by novices and experts in programming robots. Indeed, in a certain sense, all problems and sub-problems faced by programmers are addressed in a trial and error and rational fashion. Moreover, that very concept does not play any meaningful role in revealing the richness of the criteria used by novices and experts to orient themselves in the vast space of the possible solutions to their problems. These claims will rely on an artificial-intelligence-inspired construal of the concepts of “problem” and “strategy” and will be substantiated by a preliminary analysis of the various sub-problems involved in robot programming.