The actual art of governing under our Constitution does not and cannot conform to judicial definitions of the power of any of its branches based on isolated clauses or even single Articles torn from context. While the Constitution diffuses power the better to secure liberty, it also contemplates that practice will integrate the dispersed powers into a workable government. It enjoins upon its branches separateness but interdependence, autonomy but reciprocity.... We may begin [any constitutional analysis] by distinguishing roughly the legal consequences of this factor of relativity. Justice Robert H. Jackson (1) Theoretically, Article I of the Constitution places the legislative power in Congress or, more accurately, jointly in Congress and in the Presidency, acting in accordance with the formalities of bicamerality and presentment. When this body--which we shall call the constitutional legislature--is capable of action, it will retain control of legislative authority.... Courts and agencies are capable of independent or autonomous action where the constitutional legislature is too fragmented to react. In such circumstances, lawmaking will tend to migrate to those bodies that are capable of decisive action. In these conditions we see the development of such familiar entities as the runaway bureaucracy and the activist courts. Professor John Ferejohn (2) I INTRODUCTION Explanations of judicial behavior that draw upon positive institutional theory have gained growing prominence within both the political science and the legal academies. (3) Within the latter, Professor John Ferejohn has been a leader in developing the neo-institutional approach to the study of the law and the courts. (4) His work has been particularly important in developing so-called models of judicial behavior. (5) These separation-of-powers models seek to explain judicial decisions by reference to exogenous constraints, such as the powers of Congress and the President to control federal courts, rather than by reference to endogenous constraints, such as the collegial nature of decision-making on certain appellate courts and the need to maintain a majority and avoid dissension. (6) Ferejohn's contribution to this symposium further develops this line of inquiry by outlining the possibility of a positive model of the judicialization of politics, and by suggesting institutional reforms that would reduce the level of the judicialization of politics in the United States. Although Professor Ferejohn's argument is still in its preliminary form, and some key features remain undeveloped, his argument nevertheless suggests important ways to think about both the relationship between law and politics in the United States and what should be done about the growing political role of the federal courts in U.S. democracy. This article begins by examining the central elements of Professor Ferejohn's argument. Part III specifically focuses on the relationship between the model of the judicialization of politics and Ferejohn's normative argument and proposals for reform. Part IV concludes by raising a few more general concerns about positivist models of judicial behavior. II FEREJOHN ON THE JUDICIALIZATION OF POLITICS Professor Ferejohn's essay draws together several complex and nuanced lines of analysis. Therefore, I begin by outlining the four key arguments that he makes. First, Professor Ferejohn recasts the debates about the relationship between law and politics in institutional terms. In doing so, Ferejohn presents the relationship between law and politics as a debate about the proper institutional location of the legislative function--the power to make binding, prospective norms and rules. (7) This approach draws from the constitutional theory of John Locke, who argued that, of the two analytically distinct functions of governments, legislating was the sovereign political power, while executing (including both ministerial and judicial functions) was a secondary, subordinate power: There can be but one supreme power, which is the legislative, to which all the rest are and must be subordinate. …