A Comparison of the Accuracy and Completeness of Problem Solutions Produced by Example-Solutions and Program Code

摘要

Use of an inductive processor to convert example-solutions into implied logic resulted in errors of omission no different from that which occurs with use of program code; however, such a technique provided a significant reduction in errors of commission. Feedback-aids in conjunction with example-solutions enabled the programmer to engage in more complex problems with few errors of commission; those aids, to be most useful, need to include the implied logic. The feedback-aids that were optimal for the initial example-solutions were not suitable for the revision of incorrect example-solutions. In the population of programmers who participated in these experiments, the number of programming languages and the number of operating systems that the individual knew were established as the best predictors of success in developing example-solutions, as well as for writing program code. (Author)