The thesis that judges could (voluntarily or not) promote efficiency through their decisions has largely been discussed since Posner put it forward in the early 1970s. There nonetheless remains a methodological aspect that has never (to our knowledge) been analyzed in relation to the judges-and-efficiency thesis. We thus show that both promoters and critics of the judges-and-efficiency thesis similarly use a definition of optimization in which history, constraints and path-dependency are viewed as obstacles that must be removed to reach the most efficient outcome. This is misleading. Efficiency cannot be defined in absolute terms, as a global ideal that would mean being free from any constraint, including historically deposited ones. That judges are obliged to refer to the past does not mean that they are unable to make the most efficient decision because constraints are part of the optimization process; or optimization is necessarily path-dependent. Thus, the output of legal systems cannot be efficient or inefficient per se. This is what we argue in this paper.