If our value-based method performs similarly to random sampling, the sophistication of multi-dimensional ranking would be unjustified. We hypothesize our approach significantly outperforms random ...