Stepwise refinement and problem solving

The possibility of reexpressing the traditional notion of stepwise refinement as a combination of general problem-solving activities that are based on paradigms taken from artificial intelligence research is discussed. This reexpression can form the basis for a more explicit view of programming as a problem-solving activity. Experiments in which each step of the refinement process is encoded into problem solving activities are described. 26 examples of code implementation using the stepwise refinement of pseudocode have been analyzed. The presence of certain combinations of activities suggest that programmers are implicitly emulating certain paradigms that have proved useful in solving complex problems. Also, a particular paradigm and its associated activities seem to be applied often throughout the refinement sequence for a given problem. The nature of the problem to be solved influences the type of activities performed to achieve a solution, as well as the problem-solving paradigm that they implicitly support.<>