Institutional Flip-Flops

Many people vigorously defend particular institutional judgments on such issues as the filibuster, recess appointments, executive privilege, federalism, and the role of the courts. These judgments are defended publicly with great intensity and conviction, but some of them turn out to be exceedingly fragile in the sense that their advocates are prepared to change their positions as soon as their ideological commitments cut in the other direction. For example, institutional flip-flops can be found when Democratic officials, fiercely protective of the filibuster when the President is a Republican, end up rejecting the filibuster when the President is a Democrat. Other flip-flops seem to occur when Supreme Court Justices, generally insistent on the need for deference to the political process, show no such deference in particular contexts.

likely effects of institutional arrangements. We explore how these problems might be overcome.
Introduction What is the legitimate authority of the President, Congress, and the Supreme Court? What is the proper relationship between the national government and the states? Many people have firm views about such questions. The puzzle is that in numerous cases, those responses seem to depend on the answer to a single (and apparently irrelevant) question: Who currently controls the relevant institutions? Because the answer to that question changes over time, history is full of what we shall call institutional flip--flops: judgments that shift dramatically with changes in the political affiliations and substantive views of those who occupy the offices in question.
Consider a few recent examples. Under President George W. Bush, Democratic senators aggressively defended the use of the filibuster, while Republican senators vigorously opposed it. 1 Under President Barack Obama, the two sides essentially flipped. Republican senators vigorously defended the use of the filibuster, which was sharply opposed by Democrats. 2 Or consider this question: Does the president have broad power to make recess appointments? Frustrated by Democratic opposition to many of his nominees, President Bush certainly thought so. 3 Many Republican senators agreed, while prominent Democratic senators did not. 4 Under President Obama, Republicans were outraged, and Democrats were supportive or quiet. 5 Similarly, President Bush issued a large number of signing statements, in which he asserted the constitutional prerogatives of the president as the basis for questioning legislation that, in his view, intruded on those prerogatives. 6 Democrats vehemently objected to the use of signing statements, in a way that seemed to signal a deep and enduring institutional opposition to them. 7 By contrast, President Obama's signing statements have not produced much protest from Democrats 8 (even though several involved contested issues with respect to presidential authority to protect national security). 9 There is a closely related type of flip--flop where a person's position on an institutional question does not depend so much on who currently controls the relevant institutions but on what particular political outcome will result. Consider a lawsuit brought by the attorneys general of Nebraska and Oklahoma, seeking to block Colorado's legalization of marijuana possession on the ground that federal law criminalizes possession. These same attorneys general have also argued that the Affordable Care Act is unconstitutional because it violates states' rights. Critics argue that the attorneys general have flip--flopped. "It is as if their arguments about federalism and state autonomy were not arguments of principle but rather an opportunistic effort to challenge federal policies they don't like on other grounds." Bush's use of signing statements: "Federal law is not some buffet line where the president can pick parts of some laws to follow and others to reject"). 8 Karen Tumulty, Obama Criticized Over 'Signing Statements', WASH. POST, June 3, 2014, at A8 (quoting a member of an American Bar Association panel criticizing the use of signing statements: "When Bush was issuing signing statements, the Republicans didn't care. When Obama was doing it, Democrats didn't care."). 9 For institutional judgments, there is a pervasive possibility of path dependence, as when people become committed to some such judgments at Time 1, and are therefore unwilling or unable to extricate themselves at Time 2, thus reducing flip--flops. A prominent example might be New Deal liberals of the 1930s and 1940s, who adopted a firm position against an aggressive judicial role, making it difficult for them to approve of the role carved out by the Warren Court. Justice Felix Frankfurter is probably the most famous case in point. Many theories of judicial review are an effort to response to a charge of institutional flip--flopping. In the context at hand, see JOHN HART ELY, DEMOCRACY AND DISTRUST (1983); RONALD DWORKIN, TAKING RIGHTS SERIOUSLY (1977). 10  On this view, they are "fair--weather federalists" 11 who promote or deride federalism based on their views of the substantive political outcomes at stake.
From these and other examples of real or apparent institutional flip--flops, it is tempting to conclude (for example) that when some people make broad and seemingly universal pronouncements about the importance of respecting executive power, what they mean is that they like and trust the current president. And when some legislators emphasize the importance of allowing Congress to check the power of the president as such, they mean that they dislike and distrust the current president. Their purportedly neutral institutional judgments are motivated and determined by the occupant of the office at the time that they are making those judgments.
We might also conclude that when people make an argument based on federalism or executive primacy in some domain, they really mean that they care about a specific policy outcome that happens to be advanced, in that particular instance, by the states (as opposed to the national government) or the president (as opposed to Congress). To say the least, it is disturbing if firm and apparently timeless statements about institutional authority are in fact an artifact of short--term judgments about substance, and not institutions at all. 12 Institutional flip--flops, real or apparent, also play a role within the Supreme Court. Within the federal judiciary, a flip--flop might result from an overriding interest in who controls relevant institutions, or more broadly on whose ox is likely to be gored -about the short--term substantive effect, in the case at hand, of one or another approach to a disputed issue. Should the Court defer to legislative fact-finding or to agency interpretations of law? Should it follow the Constitution's original meaning? Should it give the political process the benefit of every doubt? Should it give respect to the decisions of the states? On all of these questions, it is possible to find real or apparent flip--flops. 13 To take a prominent and much--discussed example, many people who abhor Lochner v. New York, 14 seeing it as a form of illegitimate judicial activism, are quite comfortable with Roe v. Wade. 15 In the same week in which Justice Antonin Scalia joined an opinion striking down section 4 of the Voting Rights Act, 16 he issued a wide--ranging dissenting opinion from the Court's decision to strike down the 11 Id. 12 We discuss below the complication introduced by the view that institutional judgments must, in the end, be defended by their substantive effects. 13 See, e.g., Thomas (1905). 15 410 U.S. 113 (1973). 16  Defense of Marriage Act, emphasizing the importance of judicial restraint. 17 Within the Court and the legal profession generally, are institutional judgments a product of an assessment of the likely composition of Supreme Court majorities? If so, is that a problem? Is it a problem if Supreme Court justices who are committed, in principle, to following the original understanding fail to do so in cases that involve standing, affirmative action, or commercial advertising (a species of institutional flip-flopping)?
In the same vein, debates over national power are replete with at least apparent institutional flip--flops. Those who believe that the federal government should have broad power to preempt state efforts in the domain of immigration may be reluctant to conclude that the federal government should have broad power to preempt state tort law. 18 Perhaps apparent inconsistencies of this kind can be justified, once we specify the appropriate principles, but in the area of preemption, as in other areas that involve the relationship between the federal government and the states, the driving force behind some people's institutional judgments seems to be the answer to this question: On the matter at hand, should we prefer the substantive judgments of the national government or instead the states? If that is indeed the driving force, institutional flip--flops are inevitable, at least over a sufficiently long period of time.
In principle, of course, the answers to institutional questions should not depend on short--term considerations, such as the political affiliations and beliefs of those who currently occupy particular offices. 19 The long--term effect on the public interest is what matters. Some kind of veil of ignorance might be invoked to provide enduring answers. The idea of the veil is that if people imagine that they do not know their party affiliation, or even many of their interests, they can more easily agree on the institutional norms that would advance the public interest. In many contexts, however, there is a pervasive problem: If we put ourselves behind such a veil, we might find it exceptionally difficult to identify clear answers to institutional questions. In the abstract, what is the right approach to the filibuster, recess appointments, signing statements, preemption, or the exercise of moral judgments by the Supreme Court? It is in large part because of the difficulty of answering such questions that people's judgments tend to be overwhelmed by short--term considerations of substance. More interestingly, the very idea of a veil of ignorance runs into plausible objections, because it seems to deprive people of indispensable 17 See United States v. Windsor, 133 S.Ct. 2675, 2697 (2013) (Scalia, J., dissenting) (decided June 26, 2013). 18 See infra note 48 and accompanying text. 19 We can imagine qualifications. Suppose, for example, that with a certain otherwise appealing approach to institutional questions, the result would be certain catastrophe in the short--term. If so, it might seem best to avoid the catastrophe. Of course there is a question whether judges or others could take account of potential consequences of that kind. See generally JOSEPH RAZ, PRACTICAL REASON AND NORMS (3d ed., 1999) (exploring "exclusionary reasons" as negative second--order reasonsreasons to refrain from acting for some reasons).
information. Nor is it exactly easy to encourage participants in institutional debates to focus on the long--term, because of the strong incentives they face to do otherwise. Some people might even wonder whether flip--flopping is a significant problem. It seems like a species of hypocrisy, and hypocrisy in politics is as old as politics itself. 20 In our view, however, flip--flipping causes serious harm, both because it leads to unproductive debates and because it produces bad outcomes (sometimes in the form of inaction). As we shall see, continuous flip--flopping reveals disagreements about institutional norms that ought to be resolved for the sake of good governance. As we shall also see, flip--flopping is often a form of bad reasoning that should be strongly discouraged. In the context of the judiciary, institutional flip-flopping can compromise rule of law values. Because accusations of flip--flopping are as promiscuous as flip--flopping itself, it is important to distinguish flip--flops from learning and other innocent forms of behavior.
Our principal goal in this Article is to elaborate the concept of institutional flip--flops and to explain why they are so pervasive. In Part I, we provide examples and derive some typologies from them. In Part II, the heart of the Article, we offer explanations. First, some flip--flops seem to be naïve, in the sense that people's views about institutional values vary depending on the short--term political outcomes they desire ----but they may well be unaware of that fact. In such cases, people's short-term substantive commitments have a kind of psychological priority, so much so that the supportive institutional judgment ("of course the President cannot have that authority") seems self--evident, even if the opposing institutional judgment would seem self--evident ("of course the President has that authority") if the substantive commitments were otherwise. In such cases, institutional flip--flops result from what we shall call merits bias, which amounts to a form of motivated reasoning. We offer some empirical support for our claim that merits bias is both real and important.
When merits bias is at work, people's institutional judgments are motivated by their substantive commitments. But it is also true that many flip--flops among political insiders and judges are tactical. In such cases, institutional arguments are made opportunistically, and hence it is no surprise that they flip.
We also show that positions that may appear to be flip--flops may turn out to be nothing of the sort. What is true for objections to apparent hypocrisy is also true for objections to apparent flip--flopping: An investigation of people's actual beliefs might reveal that such objections have no merit. If a judge has a particular theory of constitutional interpretation, her apparently aggressive approach to one statute might not be inconsistent with her apparently passive approach to another. The governing principle might be complex and fine--grained, and once it is identified, an apparent flip--flop might disintegrate. In addition, and perhaps more interestingly, changes in institutional judgments might reflect one or another form of learning. The most complex setting involves people's adjustments of their institutional commitments as a result of learning that those commitments lead to unfortunate substantive outcomes (and not merely in the short--term), thus justifying those adjustments.
In Part III, we explore a way forward. We argue that flip--flops arise in the first place because it is difficult for people-as a matter of both psychology and politics-to differentiate optimal institutional design ("the rules of the game") and the specific, short--term political outcomes they care about, especially when in the midst of political debate. Yet the "gotcha" response is unhelpful; changing one's institutional views may well be justified. In general, the best way to adjudicate alleged flip--flops is to enlist the veil of ignorance-a device that forces people to evaluate institutional arrangements abstracted away from their short--term substantive commitments. But in the relevant contexts, the idea of a veil runs into genuine normative, conceptual, and empirical problems, in part because the veil might deprive agents of indispensable information about the likely effects of institutional arrangements. We explore what the veil of ignorance should be taken to mean in this context, and how those problems might be resolved.

I. Examples and Typologies
A. Flip--Flops: Real or Merely Apparent?
1. Relevant distinctions. An institutional flip--flop is a reversal of one's position on an institutional value based on partisan or political interests or substantive commitments. Here is a way of specifying the idea: People flip--flip when (1) at two distinct points in time they take different positions on the validity of a claim of institutional authority for a set of policy decisions, and (2) there are no relevant differences that would justify the shift. 21 We distinguish two types of institutional flip--flops. A partisan institutional flip--flop occurs when the reversal results from the change in party control of the relevant institution. A Democrat who decries presidential power when the president is a Republican and defends it when the president is Democratic engages in a partisan institutional flip--flop. A substantive institutional flip--flop occurs when the reversal results from differences or changes in the policies of the relevant institutions. A conservative who invokes federalism to defend states that restrict abortion but then invokes the supremacy of national power when Congress limits same--sex marriage engages in a substantive flip--flop (unless some principle is available to justify the different invocations). While we note this distinction, we think that these types of flip--flops are essentially the same phenomenon, and will use examples of both throughout our discussion.
It is important to begin by distinguishing between genuine flip--flops, in which there is no neutral justification for the flip, and less clear cases, in which a neutral justification can be identified, raising a question whether a flip--flop is involved at all. Suppose, for example, that a Democratic senator vigorously defends filibustering federal judges who are nominated by a Republican president, but later contends that it is unacceptable to filibuster federal judges who are nominated by a Democratic president. If so, a flip--flop is likely to be involved. In such cases, the institutional commitment appears weakly held or even non--existent: the senator's position on the legitimacy of the filibuster is wholly derivative of his views about the President; and it is hard to identify a relevant difference between the two situations that would justify the shift.
The matter is far more complicated if a judge votes in favor of upholding (say) the Affordable Care Act but not the Defense of Marriage Act. A critic might be tempted to argue that the judge has flip--flopped, perhaps by taking different positions on judicial authority, or on federalism and national power. But in cases of this kind, we may have an apparent rather than real case of flip--flopping. Different constitutional problems and provisions are involved, and no simple position on judicial authority is likely to cut across problems and provisions. In the example at hand, for example, a judge might endorse some form of democracy--reinforcing judicial review, 22 calling for a deferential approach to congressional action under the Commerce Clause (and hence to the Affordable Care Act), but a more aggressive approach to the Equal Protection Clause (and hence to the Defense of Marriage Act). So too, judges who vote to invalidate a physical invasion of private property under the Takings Clause need not be engaged in any kind of flip--flop if they vote to uphold restrictions on commercial advertising. Perhaps those judges are originalists, 23 and perhaps their view of the original understanding calls for invalidation of physical invasions but validation of restrictions on advertisements. If so, we have no flip--flop at all. 2. A possible objection. We will return to the role of neutral justification below, but at the outset, it is important to explore a possible objection. It is plausible to insist that on institutional questions, any particular position has to be defended on some substantive ground. We might favor an institutional judgment -say, a system of checks and balances of one or another kind ----on the ground that it will increase social welfare, or promote democratic self--government, or increase national security, or safeguard liberty. The senate filibuster, or presidential power to act unilaterally to protect the nation against immediate threats, might be justified on one or another such ground. Institutions are normally seen as arrangements that produce bundles of expected policy outcomes ----not as entities that are intrinsically 22  appealing or valuable in themselves. 24 It is hardly embarrassing to support an institutional arrangement with the claim that it will lead to good outcomes, all things considered. In this light, we might be inclined to raise a question about the whole category of institutional flip--flops. When people take one position on the filibuster under a Republican president, and other position under a Democratic president, is this apparent inconsistency due to divergent judgments about the consequences of filibusters? Is that a flip--flop at all? In what sense?
Suppose that someone's considered position is this: "I favor the filibuster under Republican presidents, but not under Democratic presidents." Or this: "I approve of the legitimacy of aggressive signing statements by Republican presidents, but not by Democratic presidents." On this approach, the expected (short--term) political commitments of the relevant institutions are part and parcel of the governing institutional position. And indeed, we shall give evidence that many people seem to think in these terms. But it is noteworthy that they do not explain themselves in this way, which suggests that views of this sort would be hard to defend publicly. In fact they would seem both self--serving and preposterous, because they would settle the rules of the game by direct reference to the political views of the relevant players. In public debate, people usually make institutional arguments that purport to appeal across partisan divides and across disagreements about policy outcomes; that is why no one would publicly argue that an institutional arrangement is justified because it favors one party or leads to a specific political outcome that is controversial.
On social welfare grounds, reasonable people might refuse to play by the rules of the game, and reject agreed--upon institutional norms, if and because they lead to extremely bad outcomes. For example, we could imagine a view that would support a shift in institutional arrangements if political power were suddenly obtained by fascists or communists. (Note, however, that if fascists or communists really obtained power, an institutional shift on the part of those who resisted them would be unlikely to take hold.) Some people think that democrats in Egypt, after supporting the move to elections, repudiated democracy when they saw that it could lead to Islamist rule. 25 Moreover, any position on (say) the authority of the president and the federal judiciary has to depend on the appropriate level of trust in those who occupy the relevant offices. As new information arrives -including new information about likely performance or preferences -judgments about appropriate 24 We are grateful to Adrian Vermeule for this formulation and for pressing this point. While it is not impossible that some people believe that (for example) democracy and other fundamental institutional arrangements are intrinsically, rather than, or as well as, instrumentally valuable, our argument does not depend on whether institutional preferences are instrumental or non-instrumental. 25 See, e.g., Jackson Diehl, Egypt's 'Democrats' Abandon Democracy, Wash. Post, July 21, 2013, http://www.washingtonpost.com/opinions/jackson--diehl--egypts--democrats--abandon-democracy/2013/07/21/58beace0--efc8--11e2--9008--61e94a7ea20d_story.html ("What happened to Egypt's young liberals? Five years ago, they were the most promising institutional authority might change, not least because predictions about the effects of one or another allocation might change. As we will see, shifts of that kind reflect learning. They might well be flip--flops, but there is nothing dishonorable about them -a point to which we shall return.
B. Political Debate Within Congress, unambiguous institutional flip--flops are easy to find. We have already catalogued a number of them. Many public officials have switched position on both filibustering and signing statements, taking the pro--executive side when their party holds the presidency, and rejecting that side when the presidency is held by the opposing party. An especially vivid example can be found in 2013 and 2014, when the Senate's Democratic majority enacted filibuster reform that it vigorously resisted under President Bush, 26 and that met exceptionally fierce resistance from Senate Republicans 27 -who themselves flipped to support the reform when they attained a majority. 28 Here is a confident prediction: If a Republican president is elected in 2016, we will see further flips on these issues.
For another example, consider the parties' positions on war powers. In 1999, many Republicans in Congress objected when President Clinton sent military forces into Serbia without congressional consent. 29 They argued that under the U.S. Constitution, the president may go to war only with the consent of Congress; and under the War Powers Act, the president must withdraw forces from hostilities if he has not received congressional authorization. 30 Many Republicans also complained when President Obama used military force without congressional authorization against Libya and threatened to do so against Syria. 31 Yet Republicans did not object when President Reagan used military force without congressional authorization in Granada, and when President George H.W. Bush used military force without congressional authorization in Panama. Meanwhile, many Democrats objected to these uses of unilateral presidential power by Republican presidents, 32 and did not complain when President Obama used military force without congressional consent in Libya, 33 nor when President Clinton did so in Serbia. 34 The parties also appear to flip--flop over the relevance of statutory constraints on other presidential powers. During the George W. Bush administration, prominent congressional Democrats complained that Bush disregarded statutes that restricted interrogation practices, surveillance, and criminal prosecution of detainees alleged to be terrorists. 35 During the Obama administration, prominent congressional Republicans have complained that Obama has disregarded immigration, education, and health statutes. In a short time, we have thus witnessed a dramatic partisan flip--flop over whether the presidency has become too powerful-whether an "imperial presidency" exists that needs to be reined in. 36 Of course some people have a general objection to what they see as excessive presidential authority, cutting across partisan divides. 37 What we are emphasizing here is that in the public domain, institutional flip--flops are pervasive. 31  --panama--reaction--congress--generally-supports--attack--but--many--fear.html (quoting Democrat Charles Rangel criticizing President Bush's military action in Panama: "As much as I would like to get rid of the bum in Panama, I don't see the legal authority of the use of the military."). 33 Jeff Zeleny, Airstrikes in Libya; Questions Back Home, N.Y. TIMES, Mar. 21, 2011, http://www.nytimes.com/2011/03/21/world/africa/21prexy.html ("The action against Colonel Qaddafi's forces drew support from many Democrats . . . ."). 34 See Dewar, supra note 28 ("As many Democrats see it, the Republican attacks on Clinton's Kosovo policy arise more from a sense that Clinton has largely escaped injury on domestic issues, leaving foreign policy as a weakness worth exploring."). 35  It is true that some of these cases may involve apparent rather than real flip-flops. Democrats objected that Bush exceeded his constitutional war powers. 38 Republicans tended to argue that Obama violated relevant statutes by underenforcing them in a manner inconsistent with prosecutorial discretion that is vested in the president. 39 With their various views on these particular issues, both Democrats and Republicans need not have flip--flopped. Without flip--flopping, one might believe that the President has broad authority over the use of force without having broad authority not to enforce domestic legislation. It is thus possible, depending on one's views about the president's constitutional authority, to believe that neither party flip--flopped (or that both did). But to say the least, politicians have not tried very hard to show that their positions have been consistent, and in some cases, the inconsistency has been palpable.
C. Constitutional Law Within the Supreme Court, it is also easy to find at least apparent flip--flops, especially on the question whether judges should defer to the political process. In United States v. Windsor, for example, a majority of the Supreme Court struck down the Defense of Marriage Act (DOMA), which denied federal tax and related benefits to same--sex married couples. 40 Justice Kennedy, writing for the Court, ruled that DOMA violated the Equal Protection Clause because Congress had singled out persons in same--sex marriages for disparagement. 41 In his dissent, Justice Scalia complained that the Court's holding would distort "democracy" by interfering with the public debate on same--sex marriage. "We might have let the People decide," he lamented. 42 And yet in two other cases decided the same week, Justice Scalia joined opinions that did not let the People decide but struck down duly enacted statutes or programs-section 4 of the Voting Rights Act in Shelby County v. Holder, and the affirmative action program at issue in Fisher v. University of Texas Austin. 43 In its various forms, the phrase "We might have let the People decide" seems to be used opportunistically.
Justice Kennedy launched his discussion of the merits in Windsor with a paean to the federalist system and the central role of states in defining marriage, 44 and yet in Fisher, he wrote a majority opinion that did not explore the possibility that Texas may have an interest in experimenting with affirmative action programs or that states play a traditional role in determining educational policy. 45  Ginsburg wrote an eloquent dissent in Shelby Country noting the importance of giving "deference" to congressional fact--finding while joining Kennedy's majority opinion in Windsor, which gave Congress' views on same--sex marriage no deference at all. 46 We surveyed colleagues for their favorite (or least favorite) example of institutional flip--flops in the Supreme Court. We offer them here not necessarily as real rather than apparent illustrations -no flip--flop is necessarily involved ----but as reflecting a sense of the terrain offered by specialists. Among the notable answers: Bush v. Gore and federalism. Chief Justice Rehnquist and Justices O'Connor, two of the Court's most vigorous advocates of federalism, 47 joined a majority opinion that disregarded Florida election law and handed the 2000 presidential election to George W. Bush. Dissenters, including Justices Ginsburg and Breyer, emphasized the value of federalism, which had not been a defining feature of their other opinions. 48 Federalism and preemption. Similarly, conservative justices who advocate federalism-in addition to Chief Justice Rehnquist and Justices O'Connor, Kennedy, Scalia, and Thomas-are the most likely to find preemption of state law when business interests are at stake, causing liberal justices who vote the other way to accuse them of flip--flopping, and exactly the same charge could be made against those same liberal justices, who are normally less enthusiastic about the rights of states. 49 Lawrence v. Texas and stare decisis. Justice Kennedy wrote a plurality opinion in Planned Parenthood v. Casey that provided an elaborate justification for stare decisis in explaining why he would not overturn Roe v. Wade. 50 But then in Lawrence v. Texas, he disregarded the recent precedent of Bowers v. Hardwick in the course of striking down a law that criminalized sodomy between people of the same sex. 51 Printz and Textualism. Justice Scalia is famous for his advocacy of textualism, the doctrine that judicially enforced norms must be grounded in the text of statutes or the Constitution. 52 Yet in Printz v. United States, he found that a statute that required state officials to enforce federal gun control regulations unconstitutional despite the absence of any textual ban on "commandeering." 53 Empirical research suggests that justices sometimes invoke interpretative canons opportunistically while achieving ideologically preferred results. 54 The First Amendment and Originalism. Justice Scalia is the Court's most vigorous proponent of the doctrine of originalism, and Justice Thomas has proclaimed himself an adherent of that doctrine on numerous occasions. They have written numerous opinions based on originalist arguments. Yet few scholars believe that they have given persuasive originalist justifications for their claims that the First Amendment blocks campaign finance regulations 55 or that the Fourteenth Amendment blocks affirmative action programs 56 or that Article III forbids Congress from conferring standing on citizens. 57 Frequently, they say nothing at all about the original understanding of the relevant constitutional provisions.
Boumediene and congressional authorization of emergency measures. In his one--page concurrence in Hamdan v. Rumsfeld, Justice Breyer explained that the Court's ruling that the executive lacked the authority to try Al Qaeda suspects in military commissions was based on a simple principle: "Congress has not issued the Executive a 'blank check.'" 58 But when Congress took out its checkbook and authorized military commissions in the Military Commission Act, Justice Breyer joined an opinion that struck down a provision in the Act on the ground that it violated the Suspension Clause of the Constitution. 59 Delegation and Chevron. In Clinton v. City of New York, the Supreme Court struck down the line--item veto statute as an impermissible delegation of legislative power to the president. 60 Yet the author of the majority opinion-Justice Stevensalso wrote the majority opinion in Chevron, which provided the legal foundations of the administrative state by holding that courts must defer to executive--branch statutory interpretations based on congressional delegations of legislative powers to regulatory agencies controlled by the president. 61 Rules, Standards, and Sebelius. In Massey v. Caperton, Chief Justice Roberts wrote a dissent criticizing the majority's holding that a state judge was required to recuse himself when his actions create a "probability of bias," pointing out that this standard "fails to provide clear, workable guidance for future cases," and then listing forty interpretative questions that the standard raises. 62 Yet Chief Justice Roberts' opinion in NFIB v. Sebelius that Congress' commerce power does not extend to "economic inactivity" creates at least as many interpretative questions. 63 (We can think of forty--one but will not list them.) II. Explanations A. Naive Flip--Flops, Merits Bias, and Motivated Reasoning 1. Empirical tests. A naïve flip--flop is one in which a person's institutional beliefs depend on the identity of the person or party currently in power, or on the short--term substantive outcome that follows from a particular institutional configuration. Recall the example of a Democratic senator who supports aggressive use of the filibuster when the president is a Republican, but who deplores such use when the president is a Democrat. If the senator genuinely believes that the filibuster is legitimate when the president is a Republican, and genuinely believes that the filibuster is illegitimate when the president is a Democrat, then the senator has engaged in a naïve flip--flop. 64 One might doubt whether the beliefs of real people could be so unstable. We conducted our own empirical test of that question. Using Amazon Mechanical Turk, we asked about 200 people this question: President Bush was often blocked by the Democratic Senate, which frequently refused to confirm his nominees. Frustrated by its intransigence, he resorted to "recess appointments," which bypass the Senate by installing nominees while the Senate is out on what President Bush considered a "Senate recess." Do you think that President Bush did the right thing? A strong majority of Republicans (58 percent) thought that he did. By contrast, a strong majority of Democrats thought that he did not (68 percent). 65 We asked a different group of people the same question, with just one difference: the name "Obama" was substituted for the name "Bush." With that change, the vast majority of Republicans (89 percent) opposed the recess appointments, whereas the strong majority of Democrats (66 percent) supported 62  them. 66 Though the result is not unambiguous, the best explanation is that people's institutional judgments were rooted in their beliefs about the merits. 67 We conducted a second empirical test using a different question: In the aftermath of the attacks of 9/11, President George W. Bush was sometimes concerned that legislation, enacted by Congress, intruded on his constitutional authority in the area of national security. Bush issued "signing statements," which set out his own views. Some of Bush's signing statements said that he would ignore congressional enactments that did, in his view, intrude on his authority. Do you approve of such signing statements? This time only 37 percent of Republicans said that they approved of signing statements -perhaps because of the current unpopularity of President Bush (even among Republicans), or perhaps because of a general concern, among Republicans during the Obama Administration, about presidential overreaching. But this was considerably more than the 20 percent of Democrats who said they approved of signing statements. 68 When the same question was asked about Obama rather than Bush, the difference became very stark: 68 percent of Democrats said they approved of signing statements, while only 16 percent of Republicans approved of signing statements. 69 As before, with both parties, a person was much more likely to support an institutional practice (signing statements) if it benefited a same--party president. While, unlike in the first case, a majority of Republicans rejected signing statements even when Bush used them, our results nonetheless show flip--flopping among numerous Republicans as well as Democrats.
Turning to constitutional law, we asked a large group of subjects the following questions: (1) Do you support same--sex marriage? (2) Do you believe that the Constitution permits Congress to ban same--sex marriage throughout the country? The first question calls for a substantive judgment; the second calls for an institutional judgment with a constitutional foundation. Our hypothesis was that people who support same--sex marriage will believe that Congress cannot ban it, and that people who oppose same--sex marriage will believe that Congress can ban it. 66 N=230; p<0.01. 67 The reason that the result is not ambiguous is that the words "the right thing" can be taken in more than one way. Some respondents might have thought that the question invited a substantive judgment, rather than an institutional one. But the structure of the question strongly suggests that an institutional judgment was sought. 68 N=203; p<0.01. 69 N=239; p<0.01.
The results strongly support this hypothesis. Among same--sex marriage supporters, only 10 percent believed that Congress could ban it; 90 percent believed that Congress could not ban it. Among opponents, 49 percent believed that Congress could ban same--sex marriage; 51 percent believed that Congress could not ban it. 70 We also asked the inverse of question (2), namely, do you believe that the Constitution permits Congress to require all states to recognize same--sex marriage? Again, supporters of same sex marriage were much more likely to believe that Congress possesses this power (69%) than opponents (6%). 71 2. Motivated reasoning and institutional judgments. We interpret these results to mean that people's views on the merits influence and sometimes decide their positions on institutional questions. Republicans are more likely to believe that George W. Bush had the legal power to make recess appointments or issue signing statements than Barack Obama does, because Republicans are more likely to trust Bush to use those powers wisely. Democrats thought similarly. People's view on the merits of same--sex marriage strongly influences their positions on the institutional question of congressional power.
We suggest that naïve flip--flops reflect the phenomenon of merits bias, through which people sincerely accept an institutional position that fits with their substantive commitments. 72 In such cases, the institutional position is motivated in the sense that it is a product of those commitments; but agents are not aware of that fact. If, for example, members of Congress are genuinely hostile to the incumbent president, and believe that his own positions are threatening to freedom and self-government, they are far more likely to accept, and even to find self--evident, an institutional position that fits with those views. But if they trust that president -if, for example, Democratic senators are asked whether they want to deny authority to a Democratic president -that same institutional position might seem preposterous. If a Democratic president acts unilaterally, Democratic legislators might think that he is reasonably exercising his authority in the fact of a "broken system." But if a Republican president acts unilaterally, Democratic senators might insist that he is violating the system of checks and balances. In fact, both the flip and the flop might seem self--evidently correct.
Similarly, when people believe that Congress supports their views (about same--sex marriage, abortion rights, and so on) and the states do not, they are likely to reject objections to national legislation based on federalism. When people believe that the states take their views more seriously, they are likely to invoke federalism as a reason that Congress should not override state law. In our view, merits bias is 70 N=200; p<0.01; using Mechanical Turk. 71 N=201; p<0.01; using Mechanical Turk. 72 As discussed below, any judgment about an institutional issue must, in the end, turn on some kind of substantive judgment. We are understanding the idea of "substantive commitments" in a narrower sensenot as connoting an ultimate justification for an institutional arrangement, but as separate and often shortterm commitments, to some kind of outcome, that influence an institutional judgment. pervasive in politics and law, and it helps to account for institutional flip--flops and other puzzling judgments, plausibly including the majority opinion in Bush v. Gore. 73 Merits bias is consistent with a significant body of psychological research, emphasizing the pervasive role of motivated reasoning. 74 If, for example, fans of a particular football team -say, the New England Patriots -watch a game, they are likely to be systematically biased in their assessment of the referee's neutrality. 75 Typically they will see the referee as favoring the opposing team. 76 Their judgment to this effect is entirely sincere, even though they would have a quite different assessment if they rooted for that opposing team. Notably, they are blind to their own bias. 77 Motivated reasoning can be found in countless contexts, including politics and law. 78 What we are suggesting here is that people's assessment of institutional issues is often motivated as well.
Related research finds that people's willingness to accept outcomes, based on institutional legitimacy, is overridden when those outcomes are truly objectionable. In such cases, the fact that the "right" institution reached that outcome drops out as a normative consideration. 79 Similarly, people tend to emphasize the substantive valence, not institutional considerations, in deciding whether Congress or instead courts should produce certain outcomes. 80 When people care greatly about the substance, their institutional judgments do not much matter. Fontana & Braman's findings-that people's cultural priors, rather than their institutional preferences, determine their support for political outcomescould also explain certain kinds of flip--flopping. 81 We should acknowledge that it is possible, at least as a matter of logic, that the causal direction could go the other way. People start off with a strong position about how much Congress can legislate about marriage, and this position influences their beliefs about the merits of same--sex marriage. On this interpretation, people suffer from a cognitive bias that causes their abstract commitments to congressional power or federalism to influence their views about substantive issues like same--sex marriage, abortion, and the like. This interpretation strikes us as deeply implausible for a simple reason: Most institutional commitments (for example, a belief that Congress can overrule the states on marriage) imply exactly nothing about one's substantive views (for or against same--sex marriage, abortion rights, the death penalty, minimum wage laws). 82   We suspect that naïve institutional flip--flopping is more common than naïve substantive flip--flopping. At least on the very largest issues, people's substantive views are not highly malleable, and on such issues, they may well be impervious even to reports about the views of party leadership. 88 Most people who oppose the death penalty will continue to oppose it even if the party changes its mind. 89 By contrast, institutional values are far less robust, because they do not trigger immediate or strongly held reactions in the abstract, tend to be derivative of substantive views, and often depend on complex tradeoffs.
Of course it is true that some flip--flops are palpable inconsistencies and might produce a significant degree of embarrassment in the agent (who might have to offer an explanation in terms of learning). But sometimes the embarrassment is tolerable (for reasons explained below 90 ) and sometimes it is avoidable. In our example above, the flip--flopper might be able to reason to herself that different values are at stake in the case of nominations and statutes, and so her changing position about the filibuster is not really inconsistent.
B institutional arguments might be invoked for purely strategic or opportunistic reasons. The agent recognizes that her credibility may suffer but believes that the tactical gain outweighs any long--term loss. She might believe that people's memories are short and hence that she has little to lose from the flip--flop. Or she might believe that the relevant audience consists mainly of naïve flip--floppers who will not believe that the agent's flip--flop is tactical. Use of institutional arguments can be good strategy. It is clear that if an agent invokes an institutional or legal position that is independent of the merits, she might be able to obtain agreement from people who fiercely disagree with her on matters of substance or politics. For example, a lawyer might persuade judges, or a judge might persuade colleagues, to vote to uphold an agency's interpretation of the Affordable Care Act under the Chevron principle 91 ----and thus bracket disagreements that might break out in the absence of that principles. 92 The problem is that if an institutional position is held only for tactical reasons, there is a pervasive risk of flip-flopping and hence a loss of credibility.
2. Politics. Tactical flip--flops are ubiquitous in politics. Recall that when Republicans hold a majority of the seats in the Senate, many of them decry the use of the filibuster by the Democratic majority, claiming that it is antidemocratic; 93 but when Republicans are in the minority, many of them claim that the filibuster is sanctioned by the Senate's traditions. 94 Many Democrats make exactly the same arguments according to their position as majority or minority party. We have given other examples of this phenomenon-recess appointments, court--packing, signing statements, unilateral executive action, and so on. In some such cases, flip--flopping is naïve, but in others, it is tactical. While legislators self--consciously invoke institutional considerations, they do so opportunistically, seeking to enlist some apparently principled argument (about checks and balances or the need to break a logjam) in the interest of a substantive goal, which is all that they really care about.
Some tactical flip--flops are in a sense shameless, but others are more subtle. Suppose that a leader of a political party wants to attack a policy initiative from the president, who is from the opposing party. Suppose too that the substantive issue (immigrations, relations with Cuba, same--sex marriage, climate change) is one on which the leader's own party is divided. The leader might press an institutional claim in the hope of building a large coalition. Some members might be far more willing to agree that "the president has exceeded his authority" than that "the 91 93 See, e.g., Hulse, supra note 1. 94 See, e.g., Peters, supra note 2. president's policy preferences are objectionable." To be sure, such members would be, in the circumstances, immune from merits bias. But we could easily imagine circumstances in which a party leader attempts to enlist people who are willing to press a claim about institutional overreaching alongside those who object to the president's action on the merits.
Tactical flip--flops are easy to understand, certainly within the political domain. (We will see that the judicial context is more complex.) The overriding concern of individual legislators is to be reelected, 95 and the goal of each party is to control the government. 96 If legislators want to be reelected, it might well be in their interest to reject the initiatives of a president of an opposing party, lest they be accused of capitulation (and render themselves vulnerable to a primary challenge). At the same time, legislators face a number of pressures, electoral and otherwise, to support the initiatives of a president of the same party. It is true that the electoral self--interest of individual legislators will sometimes lead them to support a president of the opposing party or to oppose a president of the same party; but the general tendency is clear. A Republican member of the House of Representatives in a majority Republican district is unlikely to have much to gain by supporting a Democratic president, and might have something to lose.
Some members of Congress are partisans, either by choice or because of the influence exerted by party leadership. To a greater or lesser degree, they want to maximize the authority of their own party. To partisans, of course, the ideal world is one in which their party exercises full control when it holds the majority and can block the other party's attempt to govern when it does not hold the majority. Thus, partisans routinely advance generous interpretations of institutional constraints when out of power and narrow interpretations when in power.
Suppose, for example, that a president makes a series of recess appointments in circumstances in which people reasonably dispute the question whether he has the legal authority to do so. Most people are unlikely to have clear views on the underlying question. The underlying issues are highly technical, and without a great deal of work, it is not easy to end up with a firm conviction. To be sure, dedicated guardians of legislative power, taken as such, might be expected to insist on the importance of advice and consent; they might have an institutional conviction that outruns short--run considerations about whether a particular president is likely to be able to ensure that his preferred people are in place. And it is possible to identify some legislators who have had some convictions. 97  incentives faced by those who run for public office, and who seek to keep their jobs, it is highly unusual to emphasize, and to stand by, such convictions. For this reason, tactical flip--flopping is not exactly surprising.
From another perspective, however, tactical flip--flops are also puzzling. The flip--flops are so common that one might wonder why anyone ever believes these institutional arguments-and if no one believes them, why politicians would ever both to make them. One possible explanation is that no one believes them; it is all theater or "cheap talk," 98 similar to polite conversation. But there is another possibility, which is that many people (including many voters) have short memories, and the sheer plausibility of an institutional argument, on the merits, can ensure that it is not entirely ineffective.
If Republican legislators decry the "imperial presidency" during a Democratic administration, and if their objections have apparent force, it might not much matter that they defended (similar) presidential authority under a Republican administration. And even if no one is actually persuaded by their objections, at least they might have the functions of intensifying the commitment of the like--minded and appealing to the beliefs of constituents. Under these circumstances, it might well be in the interest of many legislators to flip at Time 1 and to flop at Time 2. From the standpoint of individual agents, the benefits of both actions exceed their costs. The benefits of the initial flip are clear, and the costs of the flop might be small or even zero.
At the same time, it is reasonable to suppose that institutional arguments sometimes do exert some force ----but usually not a large amount, at least when the underlying question are genuinely difficult. Senators from both parties, for example, jointly benefit from common institutional rules that protect political minorities. A current majority is aware that if it adopts rules that greatly weaken the authority of the current minority, it might itself be disempowered in the future. And legislators of both sides are likely to have some respect for the traditional institutional prerogatives of the national legislature, thus ensuring that it will preserve a "core" of those prerogatives. "Turf protection" can sometimes unite legislators across partisan lines, reducing flip--flops.
It is a nice question how a degree of institutional self--protection interacts with the electoral concerns of individual legislators; we could easily imagine and even find cases in which electoral self--interest argued in favor of little interest in institutional questions, even within the "core. can be meaningfully invoked in debate, even if they exert influence only on the margin.
3. Courts. Clear tactical flip--flops are much less visible in the courts than in the political arena, but the concept is not unfamiliar in discussions of the judiciary, and tactical flip--flops can sometimes be found within the legal system. To take an admittedly extreme allegation, begin with Bush v. Gore, where some people flatly accuse the majority of ruling for Bush in order to ensure that the next several appointments to the Supreme Court would be Republicans. 99 On this view, the Republican justices who voted for Bush feared becoming a minority as a result of Gore appointments over the next four to eight years, and thus losing their power to shape American law. The justices were willing to lose some credibility, to adopt an adventurous interpretation of the Equal Protection Clause, or relax their commitment to federalism, in order to ensure that they had enough power in order to advance their legal and ideological goals another day.
We do not mean to endorse this set of claims, which we think unfair, accusatory, and wrong. But we suspect that within the federal judiciary, tactical flip-flops are much more common than generally recognized, albeit less dramatic (and far more interesting) than the alleged tactical flip--flop in Bush v. Gore. One likely reason that justices are unable to make fully consistent institutional arguments in all opinions that they join is that it is necessary to form majorities with justices who may agree on the outcome but disagree with the reasoning. A justice who does not have a weak commitment to federalism, or who is unenthusiastic about originalism, might be willing to join an opinion that shows a strong commitment to federalism, or a keen interest in originalism, as a way of preventing undue splintering within the Court. It is common for judges or justices to join opinions with which they do not wholly agree, and the result can be a degree of institutional flip--flopping from one case to another.
Judicial opinions often reflect incompletely theorized agreements, in the form of low--level principles on which diverse people can agree, notwithstanding their disagreement on fundamental issues. 100 When incompletely theorized agreements are in place, there will be no flip--flopping. But sometimes incomplete theorization is not possible, perhaps because it is insufficient to resolve the relevant controversy, perhaps because some members of the winning coalition want to offer some reasoning with which other members do not fully agree. In such cases, the opinion might not fully reflect the views of all those who join it, and a degree of flip-flopping will eventually follow.
For example, suppose that in Windsor, Justice X believed that DOMA was unconstitutional at least in part on federalism grounds while Justice Y believed that 99  DOMA was unconstitutional entirely on Equal Protection grounds (and hence that federalism was irrelevant). Perhaps Justice X was assigned the opinion, and Justice Y wished to avoid writing a concurrence because she did not want to complicate the law, or increase uncertainty, and perhaps also feared that X might flip his vote (or perhaps would similarly refuse to join X's opinions in future cases). Accordingly, Justice Y joins an opinion whose reasoning may be inconsistent with another opinion written by Justice Y, resulting in what appears to be a flip--flop, and indeed is a tactical flip--flop. Justice Y suppresses her reasoning and takes a possible hit to her reputation in order to preserve stability and comity within the Court and her relationship with Justice X (enabling Justice Y to exert greater influence on the law in some future case). If so, Justice Y will engage in tactical flip--flopping, in the sense that she has committed herself to a view with which she does not fully agree, and from which she will retreat in time.
C. Non--Flip--Flops Many cases of alleged or apparent flip--flopping are nothing of the kind. Charges of flip--flopping -like charges of hypocrisy more generally -often disintegrate once we specify the beliefs of the relevant agent. The problem with the charges is that they may depend on taking those beliefs at a high (and obtuse) level of abstraction, when those who hold them do not take them in that way.
Suppose, for example, that a judge is charged with flip--flopping if she favors invalidation of affirmative action programs but has no objection to discrimination on the basis of sexual orientation, or if she is willing to uphold gun control legislation but is unwilling to uphold restrictions on commercial advertising. At a certain level of abstraction ("does the judge believe in judicial activism?"), such a judge might be accused on flip--flopping, but the charge might well be baseless in light of the theory of interpretation that the judge actually holds. A judge who believes that the Equal Protection Clause requires racial neutrality, and no more, is hardly inconsistent if she votes to invalidate affirmative action programs but to allow discrimination on the basis of sexual orientation. A judge who votes to strike down an agency interpretation of law on the ground that it violates the text of the underlying statute is not inconsistent if he later votes to uphold an agency interpretation on the ground that it does not run afoul of any statutory text. We suspect that within the court system, accusations of flip--flopping are frequently and perhaps generally misplaced, because they fail to specify the theory under which the accused judges are operating.
Indeed, it is quite clear that in the contexts of substantive due process and federalism, Chief Justice Rehnquist and Justice Kennedy can raise plausible defenses against the flip--flop accusations, certainly if the decisions discussed above are treated in isolation. More broadly, and apart from theories of interpretation, there are two basic responses to a flip--flop accusation that fall well within the conventions of legal reasoning. Weighting. In the context of substantive due process, Justice Kennedy could argue that in both Casey and Lawrence, he took seriously the principle of stare decisis, but the principle had more weight in Casey than in Lawrence. In Casey after all, the precedent, Roe v. Wade, was almost 20 years old. Countless judicial opinions had relied on it, as had many state legislatures. Private actors had as well-by, for example, moving their households to conservative states or setting up abortion clinics in those states. Politicians had built their careers defending or criticizing Roe, suggesting that the Court's decision had supplied a settled background for a longstanding public debate. 101 In Lawrence, the precedent (Bowers v. Hardwick 102 ) was nearly as old, but states generally did not enforce their anti--sodomy laws, and it is therefore hard to believe that anyone relied on the holding in Bowers. On this ground, Justice Kennedy was justified in giving relatively little weight to Bowers, while giving much weight to Roe. No one believes that stare decisis is absolute. Sometimes the arguments in its behalf are very strong, and sometimes they are weak. It is certainly arguable that the claims for stare decisis were stronger in Casey than in Lawrence.
Similarly, Justice Rehnquist could argue that federalism concerns in Bush v. Gore were far less weighty than the federalism concerns in Gregory. On one view, Florida's local law and its (arguably questionable) local judiciary made decisions that would have significant national import. In Bush v. Gore, it was contended that the Equal Protection Clause called for a principle of equality in national elections that was not being respected by Florida officials. Retirement ages for state employees, by contrast, really are local in nature, and so federal legislation that attempts to control them runs afoul of weighty federalism concerns. Whether or not this line of argument is ultimately convincing, it is certainly plausible to say that federalism concerns deserve different weights in different contexts -and hence that no flip--flop is involved.
Omitted institutional values. One institutional value must be weighed against others, and so even if the value about which a court allegedly flip--flops is equally weighty in two cases, the outcomes would still not necessarily be the same. Suppose, for example, that Justice Kennedy believed that gays and lesbians were a politically vulnerable group, one that on the logic of Carolene Products 103 was entitled to a special level of judicial protection because its members could not defend themselves in the political arena. 104 This institutional factor could cut against stare decisis.
As for Bush v. Gore, it is possible to speculate that the Supreme Court intervened in order to prevent partisan Florida judges from throwing the election for Gore. 105 If that was a motivation for Chief Justice Rehnquist's vote, he may well have believed that one institutional value-preserving free and fair elections at the national level-superseded the values of federalism that controlled the outcome in cases that did not involve national presidential elections.
The weighting and omitted--values arguments are potentially available to the justices to avoid the flip--flop charge, but they are not necessarily valid. Some people may agree that the weight that should be given to Roe is greater than the weight that should be given to Bowers; others may disagree. To resolve this disagreement, one must investigate further the underlying theory of precedent. A very simple theoryolder precedents are given more weight-could genuinely constrain flip--flopping. Others-for example, better--reasoned precedents are given more weight-might themselves be too spongy to prevent flip--flopping. Similarly, there must be a boundary on the type of institutional values that justices might invoke to rationalize their decisions. We might be more willing to accept weighting and omitted--values arguments when justices candidly make those arguments, but they do not always do so-indeed, they did not Lawrence and Bush.
D. Bayesian Updating Agents sometimes change their minds about institutional values. When they do so, their decisions may appear to be a flip--flop. But whatever one calls their decisions, the negative connotation of the word "flip--flop" does not seem fair. The change in positions is not naïve; no merits bias and no motivated reasoning need be involved. Nor is it tactical or opportunistic; the agent is being sincere. But the change in position is genuine, in the sense that at time 2 the agent invokes a value that she repudiated at time 1. There is no hypocrisy here, and no naïveté, and nothing tactical. The agent has changed her mind.
Consider, for example, a hypothetical proponent of executive power who has consistently argued in favor of broad executive authority in general, and such specific manifestations as signing statements, unilateral war--making, the line--item veto, and so on. Her beliefs are based on her study of history; perhaps she believes that time and again the executive saved the day, while feckless legislatures and courts tried to hold it back. One day, this proponent realizes that her institutional assumptions were wrong. The executive is actually more dangerous than the other institutions, and cannot be trusted. She now opposes signing statements, unilateral war--making, and the line--item veto.
Or consider a person, living in the 1920s and 1930s, who reveres the Supreme Court and the rest of the federal judicial system, seeing it as the great bulwark against political corruption or tyrannical majorities. But as the Court repeatedly blocks much--needed social legislation, she begins to think that it is a reactionary institution that holds the nation back. By 1937, she is willing to support FDR's court--packing plan, having repudiated institutional values that would have caused her ten years earlier to oppose it. She adopts an approach associated with James Bradley Thayer, arguing in favor of a strong presumption of constitutionality, which she once believed to be indefensible and absurd. 106 She thinks that her new view reflects hard--won wisdom.
We could imagine a similar reversal in more recent decades. Suppose that in the 1960s, observers (and perhaps judges as well) supported an aggressive role for the federal judiciary, seeing courts as indispensable safeguards for politically weak groups. 107 Perhaps the Warren Court seemed to be a desirable model. Perhaps the Supreme Court appeared to be "the forum of principle" in American government. 108 But suppose that in the last decades, people who were once inclined to this view have become disaffected, not on the ground that the Warren Court was wrong, but because of a belief that it represented an unusual and perhaps unique period in American history. Perhaps they believe that the Court cannot protect politically weak groups, because it is unwilling to do so. Perhaps such people now think that the Court cannot be the forum of principle, because it is not good at moral theorizing. Good Bayesians could reject their previous position on the ground that the justices are likely to use judicial power in ways that they would reject -and that a restrained judicial position is therefore better. Perhaps they end up embrace some version of Thayerism. 109 Alternatively, we could imagine people -once frustrated and even appalled by the approach of the Warren Court, and insistent on a restrained judicial rolechanging their view in light of the possibility of originalist judging or firm judicial protection of liberty as they understand it. Those who reject progressive constitutional law, of the sort exercised by the Warren Court, might flip from an embrace of Thayerian to strong support for judicial restraint on the power of Congress under the Commerce Clause or on what might be considered to be takings of private property. Of course such people could not fairly be accused of an institutional flip--flop if they held the same view all along. But if they shift from Thayerism to some other approach, they are engaged in an institutional flip--flopnot for tactical reasons, but because they have learned over time. In fact we can easily imagine numerous shifts, at least over a sufficient period of time. People with a particular set of substantive views who favor an institutional position at Time 1 might flip at Time 2, and flip back at Time 3, only to flip again at Time 4, all because of what they learn at relevant periods. It is possible to accuse people who engage in such shifts of massive flip-flopping across numerous policy areas. But in the cases we have described, the agent will argue that the term is not appropriate because of its negative connotations. She has legitimately changed her mind. Open--mindedness is a virtue, and it is both important and honorable to learn over time. Still, it is not clear that this type of global flip--flop is always or necessarily less troublesome than the policy-specific flip--flop. We might be less inclined to trust someone who could be so wrong (by her own lights) for so long.
In any event, we call this type of flip--flop "Bayesian updating" in order to acknowledge that people will rationally update their beliefs about institutional values as they obtain more information about how institutions work and of what they achieve. As they do so, they may change their beliefs about the desirability of specific policy outcomes that are grounded in specific institutional values. A person might think that signing statements are good until she loses trust in the executive as an institution, and then rationally believe that signing statements are bad. A person might think that an aggressive judicial role is good until she loses trust in the capacities of the judges, and then rationally believe that an aggressive judicial role is bad. Recall that institutional positions require some kind of substantive justification, at least over the long term, and reduced levels of trust certainly bear on substantive justifications.
In some contexts, of course, serious questions might be raised about the legitimacy of Bayesian learning. Suppose that the right theory of constitutional interpretation is originalist, and suppose that that theory leads to a particular set of conclusions about certain issues involving executive power, voting rights, and the Commerce Clause. If so, any flip--flopping, from one era to another, would be illegitimate; the agent's role is to follow the original understanding. If the agent has learned something about the capacities and propensities of judges, she ought not to flip. She ought to continue to follow the original understanding. The general point is that with respect to institutional matters, Bayesian learning would be legitimate only if what people are learning is legitimately relevant to their decisions.
In the context of executive--legislative relations, some of the underlying questions do not involve constitutional law; they involve issues of policy. And it is possible to think that judgments about constitutional law, or even interpretation as such, depend, in the end, on assessments of consequences, so that understandings about institutional performance, and the capacities of judges and legislatures, really should inform one's choice of interpretative theory. 110 If so, institutional flip--flops may in fact reflect a form of permissible updating on the basis of new information. For those who believe (as we do) that an approach to interpretation cannot possibly be blind to consequences, it is not objectionable for someone to shift in the direction of Thayerism when new information supports Thayerism, or to shift away from Thayerism for the same reason. Tactical flip--flopping would not be admirable within the judiciary (putting to one side the practical issues faced by multimember courts), but nothing is wrong with learning from experience.
Judges have on occasion changed their views about legal doctrine as a result of what they saw as learning. 111 For example, Justice Blackmun reversed his stance on the death penalty after voting to uphold it for many years. In an earlier decision, he had said that the wisdom of the death penalty was for legislatures to determine, while it was possible for courts to ensure that racism and other invidious factors did not influence capital punishment decisions. 112 Many years later, he argued that experience had taught that legislatures could not create rules that ensured that the death penalty could be administered fairly, and thus it was unconstitutional. 113 It was the experience of the intervening years that caused him to change his mindand he appeared to change his mind about institutional values or capacities (the capacity of legislatures to ensure that capital punishment is administered fairly) rather than on the policy itself.
Some people, particularly intellectuals, seem to undergo conversion experiences. They resist the accumulating evidence until it becomes irresistible and then repudiate their old views in one great burst of anguish. 114 Government officials, politicians, and judges seem to change their views more gradually. 115 It is frequently noted that Supreme Court justices "drift" over time, becoming more liberal or more conservative. 116 This may reflect Bayesian updating. A more concrete example is the claim that Republican Supreme Court justices lost trust in the executive branch and began turning against the Bush administration in the war--on--terror cases after disclosures of executive--branch ordered torture of suspected terrorists. 117 In that context, there is a plausible argument about judicial learning.

III. Beyond Flip--Flops? A Way Forward
Flip--flopping is so common that it has generated its own meta--discourse, in which journalists, politicians, academics, and judges not only disagree on the substance, but also accuse each other of institutional flip--flopping. These responses generally amount to little more than "gotcha!", with implicit accusations of hypocrisy and bad faith. As noted, many of these allegations are unjustified. We think that there are more productive ways to approach this debate, though they raise problems of their own.
A. Identifying the Problem Flip--flopping is not inevitable. Indeed, from one perspective, it is surprisingly rare. Imagine that President Bush had announced in 2008 that because of the war on terror, a state of emergency existed that (1) allowed him to stay in office beyond the end of his term; or (2) entitled him to stand for election to a third term; or (3) allowed him to jail political opponents who undermined the war effort by criticizing his policies. We are confident that if President Bush had made any of these claims, Republicans as well as Democrats would have opposed him, just as Democrats (and Republicans) would object if the president who made such claims were a Democrat. In this context, we think it plain that the parties would not flip--flop-that they would oppose such actions regardless whether they were taken by a co--partisan or a member of the opposite party. In the face of clear constitutional norms, we do not expect to see flip--flopping.
Flip--flopping also is surprisingly rare for a range of longstanding norms that lack a constitutional foundation. It is an established practice for the majority party in the House or Senate to permit members of the minority party to sit on committees and question witnesses. We think, again, that if a leader of the majority party proposed depriving members of the minority party of their traditional seats, that leader would be opposed by co--partisans as well as by members of the opposite party. Here again there is no flipping or flopping.
A key precondition of flip--flopping thus seems to be ambiguity as to whether a constitutional or institutional norm exists. Consider whether a norm exists that permits filibusters of presidential nominees to executive branch offices. There is no consensus that the norm exists-or that it does not exist. Disagreement prevails. In this setting, it may not be surprising that politicians flip--flop. The problems with a norm permitting filibusters may be obvious when one's party holds the presidency, and much less so when the other party holds the presidency. Since disagreement exists as to the existence and value of the norm, people may genuinely come to see costs or benefits that they previously overlooked when a different party held control of a relevant institution..
We explore below the extent to which the problem of flip--flopping results from a failure of enforcement (the topic of Section D, below). What we are emphasizing here is that at least in the United States, flip--flopping is a product of ambiguity, at least in the sense that ambiguity is a necessary whether or not sufficient condition. The ambiguity concerns the existence and contours of a purported norm, that is, the extent to which a norm has really been followed or not.
For example, there is a long history of filibustering; what is relatively new is the filibustering of executive branch nominees (or perhaps the filibustering of a large number of such nominees as opposed to isolated outliers). It may be unclear whether the norm that permitted filibustering of bills extended to executive branch nominees or not. Or, to consider another example, the principle of federalism dictates that state governments legislate with respect to local matters, but it is often ambiguous what counts as a local matter and what does not.
In the face of such ambiguity, merits bias is likely to have a great deal of influence. The same is true of tactical changes of position to gain political advantage. It is in the face of ambiguity -rather than clearly established lines -that learning can produce shifts in institutional commitments. And yet, at the same time, norms do come and go, 118 and so it is possible that what once was clear can become ambiguous, and vice--versa.
B. Optimal Institutional Design Flip--flopping, then, is connected to the problem of evaluating ambiguous rules of institutional design (including both constitutional and "sub--constitutional" or institutional rules or norms) in the "midst" of normal politics. Parties recognize that some rules define the "game" of political conflict; those rules must be applied impartially to both parties. That is why everyone agree that the 22 nd amendment applies to Republican and Democratic presidents alike. But, as noted, many rules are ambiguous (including filibustering rules); it is also possible that everyone would recognize the value of new rules. Indeed, new rules are being proposed all the time. Consider, for example, rules that now govern campaign contributions (at least the relatively uncontroversial ones), or that require the GAO to determine the budgetary impact of proposed laws. The idea behind these rules is that they improve politics, or benefit the nation as a whole, without giving an advantage to a specific party; thus, if the rules are implemented, they must be enforced impartially.
In an ideal world, flip--flopping would be avoided if politicians honestly evaluated these rules and with the requisite information and impartiality, supported those that improved the political system. Evaluation of the rules would then depend on some kind of social--welfare analysis. For example, a typical argument for campaign finance reform is that existing rules allow the rich to bias political outcomes in their favor and away from the public interest. 119 If so, public--spirited politicians should support reform. Or consider a narrower justification. Unlimited campaign finance harms both parties by locking them in an arms race where they spend all their time raising money and no time governing. 120 It also harms donors because their increased donations are cancelled out by donations on the other side. Campaign finance reform that put limits on donations benefits both parties without harming anyone.
In a more realistic world, agreement on rules that advance the public welfare may be stymied by a number of factors. First and most familiarly, politicians do not necessarily act in the public interest; they are more likely to act in their own electoral interest. In the standard formulation, members of Congress try to maximize the chances of reelection. 121 Thus, we are more likely to see agreement on rules that are mutually beneficial for those who agree to them than on rules that advance the public interest, though to be sure many rules may do both. Second, politicians often face a conflict between their short--term electoral interest and the long--term interest of their parties. For example, politicians may worry that campaign finance reform may prevent them from winning the next election even if they agree that it would help their party, or even themselves, if they survive that election. Third, and our particular focus here, politician may have difficulty, both because of the pervasiveness of uncertainty and the role of merits bias, figuring out whether a purported rule that serves her interest when she is in power but harms her when she is out of power (or vice versa) is in aggregate beneficial or harmful for her political interests. It is this last factor that may cause flip--flopping.
Similar points can be made about judicial flip--flops. Consider a judge who strongly opposes abortion but also takes his judicial role seriously and acknowledges that as the law now stands, abortion rights are protected by the U.S. Constitution. Nonetheless, faced with a novel controversy-consider, for example, recent laws that restrict abortions based on purported or real concerns about women's health 122 -the judge may be tempted (perhaps unconsciously) to allow his beliefs about abortion to influence his decision. He thus may uphold the laws by citing principles of federalism even though in another context he strikes down a state law notwithstanding principles of federalism. The judge flip--flops on federalism, driven by his substantive beliefs.
C. The Veil of Ignorance 1. In general. A (potential) solution to these problems is the veil of ignorance. The veil of ignorance was made famous by John Rawls, who used it to motivate his theory of justice. 123 Rawls' veil of ignorance, like the state--of--nature constructs used by earlier political theorists, allowed us to imagine the ideal constitutional norms that people would choose under conditions of impartiality. Because people did not know their positions in life, they would choose norms that advanced the public good or met other criteria of justice rather than their own self--interest.
The type of veil we have in mind, however, is less ambitious but for our purposes more familiar and helpful. 124 A politician who considers the merits of a norm against filibustering, or campaign finance reform, or a similar constitutional or institutional reform, should ask herself whether she would support the norm if she did not know whether she was a Republican or Democrat. The veil of ignorance deprives her of knowledge of her party. Under the veil, the politician would support the norm or reform if and only if she believes that it would advance the nation as a whole.
Similarly, the judge in our example above might ask himself what the principle of federalism implies in a world in which he does not know whether he supports or opposes abortion, or indeed any of his political positions. This type of thinking encourages the judge to think about the merits and disadvantages of federalism abstracted away from a specific outcomes. Indeed, this type of impartial thinking is exactly what we expect judges to do.
Both naïve and tactical flip--flopping is a sign that an agent does not apply the discipline of the veil of ignorance to her positions. The problem with merely accusing the agent of flip--flopping is that it doesn't tell us which position the agent should adopt; it is just an accusation of bad faith or naiveté. The goal should be not just to shame the agent, but to force her to announce a position on the norm based on an argument from behind the veil of ignorance.
2. Puzzles. The idea of a veil of ignorance is both time--honored and appealing, but in the context at hand, it raises a series of puzzles. The first is normative. To see the problem, consider a stylized example. Suppose that a nation has two parties. Party A is usually in the majority, and it will be in the majority for the relevant or foreseeable future (two years? four years? eight years?). Party B represents the minority. Party A is committed to a program that would reduce both welfare and liberty; Party B strenuously resists that program. Let us stipulate that the veil of ignorance deprives both parties of knowledge about which party is likely to be in power. If both were placed behind the veil, they might, on plausible assumptions, choose a resolution that gives the majority a great deal of power. If so, Party B should be entitled to object that the veil leads to an unfortunate and possibly even catastrophic outcome, and that if it is entitled to consider both its enduring minority status and the welfare--reducing, liberty--impairing program of the majority, it would insist on its current prerogatives.
On this view, the veil of ignorance is unhelpful, because it deprives people of essential information. It is only with knowledge of the circumstances (i.e., not behind a veil) that one knows what the right course is. Or to put it another way, the veil of ignorance treats optimal constitutional design as an exercise in ideal theory, where it might be better understood as an exercise in nonideal theory. It might follow that the use of a veil of ignorance is helpful only if we benefit from having a degree of (unrealistic?) agnosticism about the future electoral success of the parties and also about which party is correct on the merits. 125 Is it clear that the veil should deprive the parties of that knowledge? On imaginable assumptions, the veil could produce outcomes inferior to those that would emerge if both parties know what is in their self--interest.
The second problem is conceptual. Behind the veil, how can the parties resolve their differences? What are the particular goals that a successful resolution would promote? To narrow the inquiry, let us focus on a discrete question that might be thought to be relatively easy: whether and when the Senate should be authorized to filibuster executive branch nominees. To answer that question, it seems clear that the parties would have to decide how to balance the interest in allowing the President to select his own team against the interest in ensuring against incompetent or extreme officials. While these are the primary considerations, the parties might have to consider other factors, such as the value (or not) of allowing "holds" as a means of extracting executive branch concessions on other matters ("extraction authority"). To come to terms with the full range of considerations, the parties would have consider, at a minimum, the following:  whether and to what extent it is important for the President to have his own nominees in place, or whether "acting" officials can essentially do the job;  whether and to what extent one or another adjustment to the filibuster rules would increase the likelihood that the President could have his own nominees in place, and to what extent;  the seriousness of the risk that without certain uses of the filibuster, incompetent or extreme nominees would be chosen or confirmed;  the opportunity costs that come from time spent closely examining executive nominees;  the costs and benefits of allowing the Senate to have the extraction authority that is associated with the power to hold nominees. In the abstract, it is not simple to provide a disciplined answer to these questions. On certain (plausible) assumptions, the Senate should essentially allow the President to pick his team, and no kind of minority veto is justified. Political safeguards constrain the President's choices, and if a nominee is genuinely incompetent or extreme, the Senate as a whole will not confirm him. If so, there is no need to retain the ability to filibuster. On other (plausible) assumptions, the filibuster is an important safeguard, and the nation benefits from it. If political safeguards are otherwise imperfect, the filibuster is a check against bad choices. Perhaps it is true that behind the veil, Senators could agree that, all things considered, the filibuster should be used exceedingly rarely in these cases. That judgment seems at least as reasonable as any other. But it would reflect something like an informed hunch about costs and benefits.
The third problem involves motivation and feasibility. Why, exactly, would Senators want to adopt a veil of ignorance? 126 And exactly when, with what kinds of discount rates for the future? In a specific year -say, this one -the veil might not seem appealing to legislators who are concerned above all with their own electoral prospects, or who are focused above all on the short--term consequences (avoiding disastrous consequences in the next two years). If so, perhaps Senators could agree that they will adopt a specific resolution of an institutional question, but only in a certain year -say, three years hence. The advantage of this approach is that it would abstract from current controversies, and no Senator could know whether he would be helped or hurt. A settlement by a future date would create the functional equivalent of a veil of ignorance.
But Senators who are now in the majority might well be ambivalent about that approach. It would not do anything to resolve their current problem, and if they anticipate being in the minority in the future, it might be profoundly unappealing. And what incentive would the current minority have to resolve the disagreement at that time? If members of the minority believe that they will continue to be in the minority, they would not be drawn to the idea of a settlement. Perhaps they would like the idea if they anticipate being in the majority -but if so, the current majority may well anticipate being in the future minority, in which case the question of feasibility returns.
In some respects, an immediate solution might seem more appealing. At least from the standpoint of the majority, such a solution would eliminate the existing problem, and would be attractive for that reason. Of course the majority will be aware that any solution might end up harming it if it loses its majority status. But that very awareness creates a kind of veil of ignorance, whose effectiveness and reality turn on how much the majority discounts the future.
In some cases, the parties will resolve a disagreement about a norm by engaging in a deal. Bipartisan deals of this kind (or deals between branches) are familiar from disputes about the budgetary process, the sharing of executive--branch information with Congress, and even filibustering. 127 What is the difference between 126 See id. at 428--32 (describing how veil rules in the Constitution control the legislature by reducing the institutional "energy" of the legislature, assuming that self--interest is the "principal spur to action," id. at 429). 127 See, e.g., Ashley Parker & Jonathan Weisman, Congressional Leaders Reach Deal on Spending, N.Y. TIMES, Dec. 9, 2014, http://www.nytimes.com/2014/12/10/us/politics/congressional--leaders-reach--deal--on--spending.html (describing the recent spending package to fund the federal government for the rest of this fiscal year, which included "more leeway for banks and other financial deals and the veil of ignorance? Deals take place when the two parties believe that resolution of the dispute about the norm serves both parties' interests (or the interests of party leaders or majorities in each party). Thus, the deals may not necessarily advance the public interest (though they may). A deal to limit campaign finance may, for example, benefit incumbent members of both parties while harming challengers of both parties. By contrast, the veil of ignorance encourages agents to take a more public--spirited approach. It might also be seen as a device for helping agents to see the perspective of their opponents.
3. A procedure. While we recognize these problems with the veil construct, it nonetheless seems to us a useful method for thinking about flip--flops. To be more specific, we propose a two--step procedure for identifying and evaluating flip--flops.
In step one, we decide whether a purported flip--flop-an apparent inconsistency in the positions about institutional values taken by an agent-is a real one. In some cases, the inconsistency is only apparent, not real: the agent can identify some additional institutional value, fact, or factor that explains the apparent inconsistency. In other cases, the flip--flop is genuine-it is naïve or tactical. The existence of a naïve or tactical flip--flop-or multiple flip--flops by numerous agents in some policy domain-is of importance because it shows that a constitutional or institutional norm is unsettled.
In step two, we demand that agent provide a justification for her institutional position. Typically, the agent will provide the justification from behind the veilthat is, by appealing to public values rather than to the agent's self--interest. By applying the veil, we transform a debate mired in political advantage--taking into a debate about constitutional and institutional values that transcend party affiliation. If a consensus emerges about the optimality of a rule, then this is a strong argument in favor of that rule.
We do not contend that this two--step process will always work. For the reasons we have sketched, merits bias and short--term incentives might prevent members of Congress from taking the relevant questions seriously. For analogous reasons, judges themselves might encounter some difficulty in using the veil. But if unjustified flip--flops are to be avoided, the process provides the best imaginable safeguard -with the exception of more formal enforcement mechanisms, to which we now turn.
D Institutional flip--flops occur when people invoke institutional values in order to support ideological or policy outcomes, changing their positions on institutions whenever doing so supports immediate goals. Such flip--flopping is troublesome because it suggests that the institutional rules do not matter. But if they do not, the gains from optimal institutional design are lost, and, indeed, in the extreme one may wonder how political organization can be possible. Thus, we turn in this section to the question of how institutional flip--flopping can in fact be constrained (putting the veil of ignorance to one side). In doing so, we collect and summarize some earlier observations, and survey some alternative theories as well. Legal enforcement. We begin with, but will immediately dismiss, one possible theory, which is that courts and other legal institutions should constrain flip-flopping. The problem with this theory is that it is question--begging. In the domain that interests us the problem is precisely that people must cooperate in the absence of legal enforcement. If senators avoid flip--flopping on the filibuster rules, it is not because the threat of judicial enforcement. Indeed, judicial enforcement is in effect endogenous to the problem that we study. When people object to court--packing and other forms of political control of the judiciary, but then turn around and engage in just this activity when in power, courts themselves cannot constrain them-the degree of independence of the courts will itself be the outcome of the flip--flopping or absence of it.
Reputation. A more fertile theory is that agents avoid flip--flopping (or excessive flip--flopping) because flip--flopping may damage their reputations or credibility. The mechanism is easy to understand. If Senator X criticizes the filibuster when in the majority, and lauds it when in the minority, people may stop believing anything that Senator X says. And if people do not believe Senator X, then she will have trouble of persuading them to support her proposals in the future. This mechanism is powerful enough that when confronted with these contradictions, Senator X will always respond by trying to distinguish the two settings. The plausibility of the distinctions that she draws will help determine her credibility in the future.
Repeated interaction. A related theory is that agents avoid excessive flip-flopping so as to maintain cooperative relationships with other agents. 128 Suppose, for example, that senators do not know whether they will be in the majority or in the minority after the next election. All senators believe that they gain more through retention of the filibuster (protecting their interests in the minority) than they lose (allowing them to prevail more often when in the majority). The next majority may then retain the filibuster, expecting that if it does not, the other party will retaliate by refusing to recognize the filibuster when it reaches power. This theory will be even more powerful for institutions like the Supreme Court, where a small number 128  of people must interact with each other over a long period of time, and risk losing cooperative gains if they flip--flop.
The reputation and repeated--interaction mechanisms are subject to well-known limits. They tend to work best in small groups, and when people expect to interact with each other well into an indefinite future. As noted, smaller institutions (courts) may avoid flip--flops more effectively than larger institutions (say, the House).
Internalization. Agents may avoid flip--flopping out of a simple sense of self-respect. We admire people who abide by principle ("statesmen") and disparage those who do not ("hacks"). It would not be surprising if these public values were internalized by some or many people. Internalization may not be subject to the small--group limits of the reputation and repeated--interaction mechanisms. However, it may also be the case that unprincipled people are drawn into the political arena (a selection effect) or that even principled people find it difficult to resist immediate pressure and may rationalize their flip--flops to themselves (cognitive dissonance reduction).
Public opinion. Public opinion may constrain the ability of agents to flip--flop. If the public cares about institutional rules, then it will disapprove of those who try to manipulate them. But does the public have institutional values? One study suggests that public approval of the Supreme Court depends on whether the public approves of particular outcomes, not on the jurisprudential quality of the opinions. 129 This seems hardly surprising. Still, politicians try to avoid the "flip-flopper" label, suggesting that the prospect of public disapproval plays at least some role in constraining flip--flopping.
Foreign involvement. Political agents may try to constrain flip--flopping by seeking enforcement from foreign countries. A well--known theory posits that when eastern European countries made the transition to democracy in the 1990s, the initial wave of liberal leaders worried that reactionary elements would initially agree to liberal constitutional reforms but then flip--flop when those elements finally came to power. 130 To forestall such flip--flopping, these countries entered human rights treaties, hoping that the threat of foreign involvement would deter the reactionaries. Another example is the use that some former British colonies make of the Privy Council to settle internal disputes. 131 Foreign involvement as an enforcement mechanism must itself be subject to limits, given that the foreign countries that are empowered to intervene may themselves refuse to obey the rules. For this reason, most countries are not enthusiastic about yielding sovereignty to foreign countries or international organizations.
Conclusion It is tempting to dismiss politicians, judges, and commentators as hypocrites who constantly shift positions on institutional norms so as to advance their electoral, political, or ideological interests. The temptation should be resisted. In most cases, there is a consensus about institutional rules, which anchor debate and allow government to function. But when the rules are ambiguous, flip--flops do arise. Unfortunately, ambiguity and hence flip--flopping are present in many important settings. We have seen that real or apparent flip--flopping can be found in recurring debates over the authority of the President and the Supreme Court.
In our view, many flip--flops are a product of motivated reasoning and in particular merits bias. Often the merits seem clear, and if so, one's judgments about institutional issues may be decisively influenced by one's judgments about the merits. We have identified highly suggestive evidence of merits bias; we suspect that the phenomenon accounts for a large number of flip--flops. But it is also true that institutional arguments are often opportunistic and hence tactical. Because politicians are not usually punished for making purely tactical arguments, and because they have strong incentives either to oppose or support the incumbent president, a degree of institutional flip--flopping is inevitable. Within courts, the constraints of multimember tribunals, and the occasional difficulty of achieving a five--person consensus on a single opinion, also ensure a significant amount of flip-flopping.
Some of the most interesting flip--flops are a product of learning. In principle, it is not simple to distinguish between motivated reasoning and Bayesian updating. When people change their evaluations of a powerful presidency or judiciary, the two may be simultaneously involved. But there is no question that when some people flip, it is because of a period of disappointment with a particular allocation of institutional authority. Sometimes, of course, the relevant learning is not pertinent to the question at hand, and hence does not justify the flip--flop. But to the extent that an assessment of consequences legitimately bears on judgments about the allocation of authority, flip--flops may turn out to be honorable.
In theory, there is a right answer to the question of what institutional arrangements should be-whether, for example, it advances the public interest if the Senate can filibuster presidential nominees. In practice, however, the answer is often obscure, and agents are not always well--motivated to implement that answer even if they can identify it. A fundamental problem arises because an agent's short-term substantive or electoral interests may be inconsistent with reform of suboptimal institutional arrangements. These problems give rise to both naïve and tactical flip--flops.