A different potential approach would be to use machine learning to train an algorithm to automatically determine the degree of helpfulness of each review. Such an approach would indeed involve less human effort, and could thus be applied to larger numbers of reviews. However, we could not draw the conclusions we would want to: any mismatch between the predictions of a trained classifier and the helpfulness ratios observed in held-out reviews could be attributable to errors by the algorithm, rather than to the actions of the Amazon helpfulness evaluators.