In order to establish a search asymmetry, one must run an experiment with a symmetric design and get asymmetric results. Given an asymmetric design, one expects asymmetric results, and such results do not imply an asymmetry in the search mechanisms. In this paper, I argue that a number of experiments purporting to show search asymmetries contain built-in design asymmetries. A saliency model of visual search predicts the results of these experiments, using only a simple measure of target-distractor similarity, without reliance on asymmetric search mechanisms. These results have implications for search mechanisms and for other experiments purporting to show search asymmetries.