The Politics of Levinson’s Constitutional Convention

Mark Tushnet*

Sanford Levinson asks us to imagine a constitutional convention called to consider revisions in structural provisions that he considers unsound and hard-wired. That the provisions are hard-wired is important to his argument, because he acknowledges that we’ve dealt by interpretation with other provisions he considers unsound.1 We can deal with the hard-wired provisions only by amending the Constitution.

Levinson also argues that at least some of the changes he proposes are unlikely to be adopted through the traditional amendment process, that is, by proposal from supermajorities in Congress and ratification by a supermajority of the states. The reason is that these changes would reduce the power of congressional incumbents — at least enough incumbents to exercise an effective veto over proposed amendments. So, he argues, the people of the United States should organize themselves outside the Constitution’s framework to adopt these revisions by means of a non-Article V convention and ratification process.

I have no quarrel with the theory of non-Article V amendment processes,2 But I have questions about the politics associated with Levinson’s proposals, on two levels: First, what are the political conditions under which his convention would convene,3 and second, what are the politics associated with the proposals the convention might advance for ratification by the American people? I suggest that Levinson’s convention would take place only if the American people came to believe that we were facing a crisis of governance caused by the structural defects Levinson identifies, and that the politics of the convention would be affected by concerns – risk aversion, the mobilization of affect by crisis, and more – that simultaneously suggest cautious pessimism and cautious optimism about the convention’s prospects.

Begin with the convention itself. European constitutional theorists distinguish between thepouvoir constituant – for present purposes, the people convened to create a constitution – and the pouvoir constitueé, the government and, specifically, the form of government called into being by the pouvoir constituant. The important theoretical point is that the pouvoir constituantcannot be controlled by positive law, because positive law is the embodiment of the pouvoir constitueé, even if the positive law is the constitutional provision for amending the constitution. The only tricky question is a practical not a theoretical one: How do you know when the pouvoir constituant has awakened and changed the constitution? The answer is provided by an analysis of social facts, as H.L.A. Hart’s jurisprudence suggests.4 So, the American people have the legal authority to bring about a constitutional convention outside the Article V framework if that’s what we want to do.

Consider these scenarios: (1) The Supreme Court holds that statutes making it a crime to burn the American flag as a means of political protest violate the First Amendment. And, lo and behold, people simply ignore the decision. Prosecutors continue to prosecute, juries continue to convict, jailors refuse to comply with judicial orders directing that they release the offenders, jurors refuse to award the offenders civil damages against their persecutors for violating their constitutional rights. How should we conceptualize this scenario? Maybe as one of widespread violation of fundamental law. The difficulty, though, is that the violation, if such it be, is too narrowly focused. Better, I think, to say that the people, acting as the pouvoir constituant, have amended the Constitution, redefining the scope of the First Amendment.

This example shows that we don’t have to think of the pouvoir constituant acting solely in a revolutionary manner, completely replacing an existing constitution with an alternative.5 Rather, the people can tinker with the constitution outside the bounds of the amendment process set by positive law.

(2) There is a groundswell of support for the election of Arnold Schwarzenegger as President, something plainly barred by the Constitution. But, fearing political retaliation by the voters, state election officials ignore court decisions saying that Schwarzenegger, being ineligible for election, should not be placed on the ballot, and do so anyway in thirty or forty or fifty states. Schwarzenegger wins (or loses narrowly, and perhaps only because he was not placed on the ballot in a few states). Here too, I think we should see this as a case in which political practice has amended the Constitution.

This second scenario provides the transition to my larger questions about Levinson’s enterprise. The people of the United States, acting as the nation’s pouvoir constituant, could convene ourselves into Levinson’s convention, and could consider making only small-ish adjustments in the hard-wired structural provisions he discusses. The Schwarzenegger scenario describes a set of political conditions under which the pouvoir constituant might mobilize itself to override a specific hard-wired provision of the Constitution with which it disagrees. What are the political conditions for the convening of Levinson’s convention?

Pretty clearly, the political conditions cannot be created merely by popular agreement that Levinson’s proposals are good ideas (if they are). One can generate all sorts of good-government reforms, but good-government reforms are actually adopted only when the political circumstances are ripe. Reflecting on the theories associated with Charles Sabel and his colleagues, I suggest that the conditions for convening Levinson’s convention are two: First, nearly everyone agrees that something has to be done, and second, nearly everyone agrees that nothing that’s been tried so far works, and that nothing on the horizon holds out much prospect for success either. Mere policy gridlock won’t satisfy the first condition, because gridlock is fine if not enough people think that something has to be done. Perhaps the agreement that something has to be done occurs in connection with structures themselves,6but that need not be so. People might agree that something has to be done about some substantive matter. Here the second condition gets more complicated: People have to agree that nothing works, and that the reason for the failure lies in the hard-wired structures that we use to select among policy proposals, which prevent us from adopting proposals already “in the air” but not “on the table” because of those structures.

The “nearly everyone” could be the people of the United States, or it could be political elites, comprised mostly but not exclusively the people’s representatives. Consider a convention convened because elites conclude that something has to be done.7 Of course, members of Congress could have used the provisions in Article V for individual amendments – proposal of discrete amendments by Congress and ratification by the states — to do something about the hard-wired structures.8 Would an elite-driven convention adopt constitutional revisions that could not have been adopted through the individual-amendment Article V process? We can assume that membership in such a convention would overlap substantially with membership in Congress, that some of the non-congressional members would probably aspire to such membership, and that some of those members might even see participation in the convention as a springboard for their political careers. And, other members, we can assume, would be elites associated with existing power structures, including Congress.

Despite all this, the convention might produce revisions in the hard-wired provisions. And yet, perhaps no convention would be needed, were the conditions I’ve described to exist. Both as members of Congress and as members of the convention, elites should calculate what their prospects would be under the revised Constitution. The anchoring heuristic, risk aversion, and the status quo bias probably make it hard for members of Congress today – that is, when the conditions I’ve described don’t exist – to calculate accurately what their prospects are under a revised Constitution. They will anchor their estimates about what is likely to occur to what they already know, which is how existing institutions confer how much power on them. The anchoring heuristic suggests that their estimates about the impact of changes will be closer to the anchor than the impact would actually be; that is, members of Congress may underestimate the changes’ effects. Risk aversion might offset that, leading them to overestimate the effects. So, on balance, perhaps members of Congress – or, perhaps better, Congress in the aggregate – can indeed accurately calculate the effects of changing hard-wired provisions. At that point, the status quo bias, both on the individual level and as an aggregate effect, might kick in: Those more affected by the anchoring heuristic might say, “The changes aren’t going to be that large, so why bother?,” and those more affected by risk aversion might say, “The changes could be disastrous, so why bother?”9

Note, though, that the “something has to be done” assumption provides the answer to the question, “why bother?” That is, were the political conditions I’ve described to exist, political elites would have reason to bother with either making small but potentially beneficial changes, or larger but riskier ones. Crisis conditions mobilize fears and anxieties that can offset risk aversion and the status quo bias. Perhaps, then, if political conditions conduced to convening a convention, we wouldn’t need to hold one.10

The individual-amendment Article V scenario may be more promising than convention one for an additional reason: It makes possible consideration of a more limited number of amendments than in a convention. Recall that the pouvoir constituant can’t be controlled by positive law. If a convention understands itself to be exercising the constituting power – which is, as I’ve suggested, a matter of social fact, not legal theory – it can consider anything, not simply the proposals that motivated its convening.11

With the possibility of placing everything on the table, what would a convention do? For the reasons I’ve sketched, a convention dominated by political elites might confine itself to good-government proposals like Levinson’s – again, given the condition of widespread agreement that something has to be done. A more populist convention might well range more broadly, because its members would not have the detailed knowledge about how politics works within specific institutional contexts that elites, who conduct politics within those contexts, do have.

Among political elites, it is common to fear a populist convention because it would go out of (elite) control and repudiate fundamentals of constitutional democracy.12 Perhaps so, although developments in political technology provide mechanisms that, if used prior to or in conjunction with a convention, might show that such fears are exaggerated.13 A different skepticism is that a populist convention would end up doing almost nothing. Constitutional design is difficult, and the revisions Levinson proposes have their merits and faults. Even if Levinson’s proposals are on balance good ideas, a populist convention might conclude that coming up with a package that makes sense is too difficult.

And yet: Designing the 1787 constitutional proposal was difficult as well. But, in a setting where nearly everyone in the relevant elites and perhaps beyond did indeed think that nothing was working and that something had to be done, the Philadelphia Convention came up with a workable Constitution. Were those political conditions to recur, perhaps those called to deliberate on improving the original design might succeed.

* Mark Tushnet is William Nelson Cromwell Professor of Law at Harvard Law School.
[1] Some have been interpreted virtually out of existence, such as the Contracts Clause. See, e.g., Home Building & Loan Ass’n v. Blaisdell, 290 U.S. 398 (1934). But see Allied Structural Steel Co. v. Spannaus, 438 U.S. 234 (1978). Others have been rendered essentially toothless, such as constitutional protection of core property rights. But see Eastern Enterprises v. Apfel, 524 U.S. 498 (1998) (Kennedy, J., concurring in the judgment and dissenting in part).
[2] In my view, U.S. constitutional theorists have struggled too hard to make sense of what is basically a simple theory. See, e.g., Akhil Reed Amar, Philadelphia Revisited: Amending the Constitution Outside Article V, 55 U. CHI. L. REV. 1043 (1988); Akhil Reed Amar, The Consent of the Governed: Constitutional Amendment Outside Article V, 95 COLUM. L. REV. 457 (1995); BRUCE A. ACKERMAN, WE THE PEOPLE: FOUNDATIONS (1991).
[3] Some criticisms of Levinson’s proposals might arise because he presents the proposals as good-government reforms, floating free from any politics at all. But, as I suggest, once we identify the political conditions under which the idea of convening a convention might have some traction, we might understand why Levinson’s proposals would not be merely good-government proposals in that environment.
[4] See H.L.A. HART, THE CONCEPT OF LAW 112 (2d ed. 1994) (describing the “rule of recognition” as a social fact).
[5] Or, if you like, we could conceptualize the scenario, with what I think is unnecessary complexity, as one in which the people wiped out the old Constitution entirely, then re-enacted it with the single change with respect to flag-burning.
[6] This would be so, for example, if Levinson’s belief in a substantially more populist form of democracy than the one we have in the United States became quite widespread.
[7] They might do so to forestall the convening of a non-Article V convention. See also note 9infra.
[8] I believe this to be true even of the equal representation of the states in the Senate, for reasons identical to those I sketched in connection with the flag-burning example. Briefly: (1) Consider the use of Article V in two steps, in the first of which Congress proposes and the states ratify the repeal of the restriction on possible amendments contained in the last phrase of Article V, and in the second of which Congress proposes and the states ratify a population-based apportionment of the Senate. (2) Now collapse those steps into one, in which Congress proposes an Article V amendment with two provisions, the first of which is the repeal and the second of which is the apportionment. (3) Now eliminate the first provision, on the ground that the only sensible way of understanding an apportionment proposal is as an implicit repeal of the restriction on amendments.
[9] I have framed my argument with reference to members of Congress, but I think it holds as well for political elites in the states during the ratification process.
[10] This argument is, I think, independent of another, which might also have some force: Members of Congress would propose constitutional amendments to forestall the convening of a convention in which they might have less influence than they would over proposing amendments themselves.
[11] For discussions, see William W. Van Alstyne, Does Article V Restrict The States to Calling Unlimited Conventions Only?, 1978 DUKE L. J. 1295; Walter E. Dellinger, The Recurring Question of the “Limited” Constitutional Convention, 88 YALE L. J. 1623 (1979).
[12] In the context of Levinson’s work, see Cass R. Sunstein, It Could Be Worse, The New Republic, Oct. 16, 2006, at 32.
[13] A more complete account would have to explain the political conditions under which these mechanisms would be used prior to or in conjunction with a populist convention. One currently available technology is the deliberative poll. For a description, see JAMES S. FISHKIN, DELIBERATIVE POLLING®: TOWARD A BETTER INFORMED DEMOCRACY, Another model is the “citizen assembly” of 160 randomly chosen voters convened to consider changes in the electoral system in British Columbia. For a description, see CITIZENS’ ASSEMBLY IN ACTION, The assembly proposed a major change, which would have gone into effect upon approval by 60 percent of all voters and by simple majorities in 60 percent of the province’s 79 electoral districts. It was approved by majorities in 77 districts, but received only 57.69 percent of the total vote, and therefore was not enacted. For the results, see ELECTIONS BC, FINAL REFERENDUM RESULTS: REFERENDUM ON ELECTORAL REFORM – MAY 17, 2005, available at

Old Paper by