Current legal scholarship on architectural regulation of software focuses on how its lack of transparency may frustrate public accountability or, by the same token, enhance its effectiveness. This paper argues that architectural regulation poses deeper dangers to the very concept of law. Ordinarily, we think of law as rules that a person thinks about when deciding how to act, and which human beings must decide to enforce. Law as architecture operates differently: instead of affecting our calculus of choice, it structure the very conditions of action, such as social settings and the resources available in those settings. Thus, architectural regulation operates surreptitiously and may not even be perceived as governmental action. Architectural regulation thus allows government to shape our actions without our perceiving that our experience has been deliberately shaped, engendering a loss of moral agency. Because our norms are often the product of social experience with and discourse about new technologies, architectural regulation poses the danger that government can distort the evolution of constitutional norms like privacy.