Promiscuous reactions of natural enzymes and computationally designed enzyme activities offer promising starting points in the quest to achieve efficient enzymatic catalysis of new reactions. To date, these less-than-optimal reactivities have populated a middle ground in the continuum of proficiency between small molecule catalysts and optimized natural enzymes. We would like to understand quantitatively how these middle-ground enzymes attain rate accelerations beyond those of isolated side chains in solution, and further, what would be needed to bring these suboptimal enzymes to the level of natural enzymes. We have investigated the recently reported catalytic activities of computationally designed retroaldolases and compared these new enzymes to related small molecule catalysts in solution. From these studies, we can estimate contributions of catalytic factors to the rate accelerations of the retroaldolases. We compare the computationally designed enzymes to natural systems with promiscuous reactivity, and ask whether general principles emerge that might guide efforts to optimize desired catalytic activities.