Computing indicators of creativity

Abstract

Currently, the most common measurement of creativity is based on tests of divergence. These creativity tests include divergent thinking, divergent feeling, etc. In most cases the evaluation criteria is a subjective appraisal by a trained "rater" to assess the amount of divergence from the "norm" a particular submitted solution has to a presented or discovered task. The larger the divergence from the standard, the more creative the solution is. Although the quality and quantity of the solutions to the task must be considered, divergence from the accepted "norm" is a significant indicator of creativity. Using the current model for showing creative divergence, a method for evaluating the divergence of programming solutions as compared to the standard tutorial solution, in order to indicate creativity should be in line with current creativity research. Instead of subjective "rater evaluations" a method of calculating numerical divergence from programming solutions was devised. This method was employed on three separate class conditions and yielded three separate divergence patterns, indicating that the divergence calculation appears to demonstrate, not only that creativity can be shown to exist in programming solutions, but that the calculation is sensitive enough to differentiate between different class learning conditions of the same teacher. So based on the idea that creativity can be shown through divergence in thinking and feeling, it stands to reason that creativity in programming could be revealed through a similar divergence to a standard norm through calculating the divergence to that norm. Consequently, this divergence calculation method shows promising indicators to inform the measurement of creativity within programming and possibly other scientific areas.