Commentary on Shaffer et al.: A cluster analysis of harmony in the McGill Billboard dataset