Abstract: A system effect arises when the properties of an aggregate differ from the properties of its members, taken one by one. Familiar examples include Condorcet’s paradox and the Prisoners’ Dilemma. Public law is rife with system effects that are more important and less familiar. Although such effects are sometimes recognized in local contexts, they have a common analytic structure and can profitably be analyzed in global terms. The failure to recognize system effects leads to fallacies of division and composition, in which the analyst mistakenly assumes that what is true of the aggregate must also be true of the members, or that what is true of the members must also be true of the aggregate. Examples are (1) the fallacious assumption that if the overall constitutional order is to be democratic, each of its component institutions must be democratic, taken one by one; and (2) the fallacious assumption that if judges are politically biased, courts must issue politically biased rulings. In these cases and many others I will discuss, system effects are an indispensable analytic tool for legal theory. A systemic approach to constitutional theory implies what I will call second-best constitutionalism. Stated abstractly, suppose that at least some of the conditions necessary to produce a given ideal or first-best constitutional order fail to hold. Even if it would be best to achieve full satisfaction of all those conditions, it does not follow that it is best to achieve as many of the conditions as possible, taken one by one. Rather, multiple failures of the ideal can offset one another, producing a closer approximation to the ideal at the level of the overall system. Although the idea is abstract, we will see that problems of second best are chronic in real-world constitutional systems, including our own, because such systems are always partly constrained by technology, economics, and politics.