"Still, Scott's advice is far from useless. It can be applied to contexts far afield from those that concern him here. His case studies help explain, say, why national regulation tends to work better when it consists of altered incentives rather than flat commands. Some of the most successful initiatives in American regulatory law have consisted of efforts to increase the price of high-polluting activities; and some of the least successful have been rigid mandates that ignore the collateral effects of regulatory controls. Scott's enthusiasm for metis also suggests that certain governmental institutions will do best if they act incrementally, creating large-scale change not at once, but in a series of lesser steps. We might think here not only of common law, but also of constitutional law. Many judicial problems derive from a belief that judges can intervene successfully in large-scale systems (consider the struggles with school desegregation in the 1960s and 1970s), and many judicial successes have come from proceeding incrementally (consider the far more incremental and cautious attack on sex discrimination in the same period). And Scott also offers larger implications. A society that is legible to the state is susceptible to tyranny, if it lacks the means to resist that state; and an essential part of the task of a free social order is to ensure space for institutions of resistance. Moreover, a state that attempts to improve the human condition should engage not in plans but in experiments, secure in the knowledge that people will adapt to those experiments in unanticipated ways. Scott offers no plans or rules here, and a closer analysis of the circumstances that distinguish success from failure would have produced greater illumination. But he has written a remarkably interesting book on social engineering, and he cannot be much faulted for failing to offer a sure-fire plan for the well-motivated, metis-friendly social engineer."