Self-Coup and the Constitution

37 Const. Comm. 315

Daniel J. Hemel[1]


After the violence at the Capitol that left five dead and dozens injured on January 6, 2021, observers struggled to find a word or phrase that would fully describe the day’s bloodshed. Early reports characterized the attack as a “riot,”[2] a term defined at common law as “three or more persons . . . unlawfully assembled to carry out a common purpose in such violent or turbulent manner as to terrify others.”[3] But outside the common law context, “riot” typically implies spontaneity, and the more we learned about January 6, the clearer it became that the storming of the Capitol that day was a premeditated attack.[4] Commentators soon shifted toward “insurrection,”[5] a word that carries significant constitutional weight in light of the Fourteenth Amendment’s disqualification provision. But that, too, proved to be an awkward terminological fit. Setting aside the question of whether January 6 was an insurrection for constitutional purposes, the term “insurrection”—which classically refers to “a rising against civil or political authority”[6]—fails to capture the fact that on January 6, the leader of one branch of the relevant civil and political authority was in on the game.

As the days and weeks wore on, some social scientists began to describe the January 6 assault as a “self-coup.”[7] Although definitions vary, the term “self-coup”—or “autogolpe” as it is known throughout Latin America—generally refers to a sudden seizure of power by the president or other chief executive in contravention of a country’s laws. While the United States never had experienced a self-coup or attempted self-coup before 2021, several of its hemispheric neighbors already knew the phenomenon all too well. Peru was the site of a successful self-coup in 1992, and it would go on to survive an attempted self-coup in December 2022.[8] At least ten other countries in the Americas have experienced events that could be classified as self-coups or attempted self-coups since World War II.[9]

A little more than a dozen years ago, the notion that the United States might soon join those ranks would have struck many public law scholars as preposterous. As Adrian Vermeule observed then, “the risk of a coup by the executive—an autogolpe—is vanishingly low in America in 2010, or for the foreseeable future.”[10] When Professor Vermeule wrote those words during the middle years of the Obama presidency, he was an outlier only insofar as he deemed the remote risk of a self-coup to be worth discussing (though soon dismissing). Others didn’t even bother to contemplate the possibility.[11] In 2023, by contrast, the idea that an American president might encourage a violent attack on another branch of government no longer lies beyond the realm of the thinkable. Not only could that happen here—it did.

Although a self-coup in the United States now seems easy to imagine, the implications that follow from recognizing this risk are much harder to tease out. A constitution can prohibit self-coups, but such a prohibition would accomplish little because a self-coup, by definition, entails usurpation of the constitution. More modestly, certain legal and political institutions may be able to reduce the risk that a future President will attempt a self-coup or that, if he does,[12] he will succeed. But understanding the relationship between institutional structures and self-coup risk requires careful and sustained reflection.

Fortunately, we need not start on a blank slate. Although the term “self-coup” was unknown to the Framers, the risk of what we now call “self-coup” was not. When the Constitutional Convention assembled in Philadelphia in 1787, it had been less than a century and a half since King Charles I plotted against Parliament[13] and just fifteen years since Swedish King Gustav III seized power from the Riksdag[14]—an event mentioned by Alexander Hamilton in Federalist No. 22.[15] Writing under the pseudonym Publius, Hamilton highlighted two types of safeguards that would reduce the risk of a similar executive usurpation in the new United States: (1) presidential selection mechanisms that would favor virtuous candidates, and (2) incentive structures that would reward Presidents for pursuing power through electoral competition rather than political violence. Meanwhile, James Madison, sharing the same pseudonym, emphasized a different set of anti-self-coup safeguards: institutional checks and balances that would empower any two branches to keep the third in line.

From Publius’s polyphonic writings, we can draw out two models of self-coup and the constitution.[16] In the Hamiltonian model, legal and political institutions manage the risk of self-coup by reducing the probability that a President will attempt to seize power from the other branches. In the Madisonian model, legal and political institutions manage the risk of self-coup by distributing power broadly so that other actors can thwart a self-coup if the President ever attempts one. To be clear, the labels for these models are intended to organize analysis, not to imply any claim about intellectual biography. But whatever Hamilton and Madison believed in their heart of hearts,[17] Publius’s writings can inspire critical thought about historical and modern responses to the problem of self-coup.

This Essay seeks to develop the Hamiltonian and Madisonian models and to demonstrate their continuing relevance. Part I starts by defining self-coup and distinguishing it from the broader phenomenon of democratic backsliding. Part II turns to Hamilton and Madison’s writings in The Federalist, extracting from their ruminations two contrasting but potentially complementary models of self-coup. Part III assesses how the Hamiltonian and Madisonian models have fared over time. Although both men failed to foresee key legal and political developments, the brief sketches in The Federalist proved to be surprisingly prescient in important respects. Part IV considers whether the Hamiltonian and Madisonian models can serve as guides for twenty-first century institutional design. It shows how Hamiltonian and Madisonian perspectives can shed light on issues as varied as the rules of the presidential nomination process, standards for prosecuting ex-Presidents, the statutory framework for filling executive branch vacancies, and proposals for Supreme Court reform.

In the interests of time, space, and cognitive division of labor, the Essay focuses on legal and political institutions that affect the risk of self-coup rather than on broader economic and social trends that set the ecological conditions for democratic breakdown.[18] And even within the domain of legal and political institutions, the Essay does not aim to be exhaustive.[19] Its more modest goal is to trace the Hamiltonian and Madisonian logics through the centuries and to show how old ideas can point us toward new responses to the self-coup threat.

  1. Defining and Distinguishing “Self-Coup”

Despite its now-frequent invocation, the term “self-coup” lacks an agreed-upon definition. Canadian political scientist Maxwell Cameron wrote in 1998 that “[a]n autogolpe (‘self-coup’) occurs when a president closes the courts and the legislature, suspends the constitution, and rules by decree until a referendum and new legislative elections are held to approve broader executive powers.”[20] This narrow definition neatly fits two of the cases that inspired Cameron—Alberto Fujimori’s self-coup in Peru in 1992 and Jorge Serrano Elías’s failed self-coup in Guatemala the following year—but it squeezes out several others. For example, the January 6 assault on the Capitol did not involve any attempt to close the courts and so would seem to fall outside Cameron’s scope. More expansively, David Pion-Berlin and coauthors define “self-coup” to include all instances in which “a nation’s chief executive, in order to hold onto, consolidate or expand power, coercively interferes with or shuts down another branch or branches of government.”[21] That capacious definition, in contrast to Cameron’s, would include January 6 as an attempted self-coup, because it involved an effort to interfere with the proceedings of another branch and briefly forced Congress to shut down.

Even with this broader understanding of self-coup (which will serve as a working definition for our purposes), the term captures only a small subset of all cases of “democratic backsliding.” Nancy Bermeo defines “democratic backsliding” as “the state-led debilitation or elimination of any of the political institutions that sustain an existing democracy.”[22] So defined, backsliding entails an attack on democracy from the top (i.e., from the state’s leadership). Self-coup adds a horizontal dimension to backsliding’s verticality: not only does self-coup involve an attack on democracy from the top, but it also involves a side-to-side attack by one branch of government against one or more of the others.

As with most other social-science concepts, the outer bounds of the “self-coup” category are somewhat fuzzy. For example, Hitler’s seizure of authority from the German Reichstag through the Enabling Act of 1933 shares some features of a self-coup—an extraconstitutional consolidation of power in the executive—but the Reichstag, at least as a formal matter, voted to approve its own disempowerment. Or for a more recent borderline case, consider the arguable self-coup that occurred in Venezuela in 2017. In March of that year, the country’s Supreme Tribunal of Justice voted to strip legislative powers from the National Assembly and to exercise lawmaking authority itself. A majority of the justices on the tribunal were loyal to President Nicolás Maduro, who reportedly ordered the court’s actions.[23] In that respect, the Venezuelan case looks a lot like Fujimori’s textbook self-coup, with the President effectively dissolving the legislature. But Maduro had the imprimatur of the judiciary, making this—along the horizontal dimension—a case of two branches attacking one rather than one attacking two. Whether those distinguishing facts take the Venezuelan case outside the category of “self-coup” is, ultimately, a judgment call. The two organizations that maintain the most comprehensive coup databases disagree: the Center for Systemic Peace codes the Venezuelan case as a self-coup, while the Cline Center’s Coup d’État Project does not.[24]

Even without a bright-line boundary around the self-coup concept, focusing more narrowly on self-coup rather than the much broader phenomenon of democratic backsliding helps to draw our attention to two distinct aspects of the self-coup problem. First, self-coup necessarily involves the chief executive as a willing participant. The risk of self-coup therefore depends, at least in part, on presidential selection and presidential incentives. This is the key Hamiltonian insight: if a constitution can’t prevent the President from carrying out a self-coup, at least it can foster selection mechanisms and incentive structures that reduce the risk that the President will try. Second, self-coup—by definition—entails conflict among the branches. Self-coup thus invites responses that activate interbranch checks. We must not lose sight of the fact that self-coup is just one of many ways in which democracies die, but the narrower focus on self-coup here illustrates the analytical benefits of breaking the broad phenomenon of backsliding into bite-sized chunks.

  1. Hamilton, Madison, and Self-Coup

The concept of self-coup was not foreign to the Framers, who were familiar with historical examples of chief executives usurping other branches of government. The Framers’ writings and the records of their debates are replete with references to Julius Caesar, who consolidated power in the Roman Republic as “dictator in perpetuity” before his assassination in 44 bce,[25] and to Oliver Cromwell, who dissolved the Rump Parliament by force in 1653.[26] And in Federalist No. 22, Hamilton specifically mentioned the recent self-coup in Sweden, where—in his words—“the most limited monarch in Europe, in a single day, . . . became one of the most absolute and uncontrolled.”[27]

Yet the specter of self-coup does not dominate the Federalist Papers. Hamilton, Madison, and their coauthor John Jay certainly had strategic reasons to downplay the dangers of executive power, since one of the arguments made by the Antifederalists in the debate over ratification was that the new Constitution gave too much authority to the President. Hamilton, moreover, harbored his own expansive vision of presidential power and at one point during the Constitutional Convention suggested an “Executive for life.”[28] The risk that the executive would seize too much power may not have been salient to such an ardent presidentialist. The Federalist Papers themselves give another reason why the authors did not dwell on the problem of executive usurpation: they believed—or at least they claimed—that they had solved it. Through a combination of selection mechanisms, incentive structures, and institutional counterweights, the new Constitution had supposedly put the threat of self-coup in check.


  1. The Hamiltonian Model: Selection
    Mechanisms and Incentive Structures

The primary treatment of executive power in The Federalist spans eleven articles by Hamilton published in March and April 1788.[29] Hamilton took particular pride in the new Constitution’s provisions for presidential selection, of which he said that “if the manner of it be not perfect, it is at least excellent.” The constitutional architecture of presidential selection was, according to Hamilton, “almost the only part of the system, of any consequence, which has escaped without severe censure” from the opponents of ratification.[30]

Hamilton emphasized several factors that, in combination, would afford a “moral certainty, that the office of President will never fall to the lot of any man who is not in an eminent degree endowed with the requisite qualifications.”[31] One was the indirect mode of selection, with voters first choosing electors and electors then settling on a President. “The choice of SEVERAL, to form an intermediate body of electors, will be much less apt to convulse the community with any extraordinary or violent movements,” Hamilton wrote, “than the choice of ONE who was himself to be the final object of the public wishes.”[32] The two-stage process would combine the democratic legitimacy of popular elections with the stabilizing effects of elite intermediation.

A second factor, closely related to the first, was the delegation of decision-making to a “small number of persons,” who “will be most likely to possess the information and discernment requisite to such complicated investigations” into the candidates’ qualities and qualifications.[33] On this point, Hamilton’s remarks prefigured—by 170 years—economist Anthony Downs’s theory of “rational ignorance” in a mass democracy. Downs would go on to argue that as the size of the electorate increases, the probability that any one voter will swing the outcome diminishes. In a very large polity, Downs wrote, “the incentive to become well-informed is practically nonexistent.”[34] Anticipating Downs’s logic, Hamilton argued that delegation of decision-making to a smaller number of Electoral College members would help to solve the collective action problem of rational ignorance by vesting the ultimate selection of the President in a limited set of citizens who were more likely to make informed choices.[35]

Hamilton also hailed the fact that the presidential selection process—though not plebiscitary—was majoritarian in a different way: If no candidate commanded a majority in the Electoral College, then each state’s House delegation would cast a vote for President, with successive ballots until one candidate achieved a majority. Thus the President always would be chosen by a majority of some sort—either a majority of the Electoral College or a majority of the state delegations in the House. In Hamilton’s view, applying a majority requirement rather than a plurality threshold would reduce the risk that an unqualified candidate might prevail.[36]

Hamilton highlighted several other features of the presidential selection process that, in his view, contributed to its excellence. For example, the electors would meet in their respective states rather than all together in one national conclave. “[T]his detached and divided situation,” Hamilton wrote, “will expose them much less to heats and ferments, which might be communicated from them to the people, than if they were all to be convened at one time, in one place.”[37] Moreover, no Senator, Congressmember, or executive branch official could serve as an elector.[38] Thus, the President would not be beholden to the legislature, and the electors would not be beholden to the sitting President. Taken together, these factors would produce “a constant probability of seeing the station filled by characters pre-eminent for ability and virtue.”[39]

Hamilton’s optimism in Federalist No. 68 may strike the modern reader as extravagant, and it is difficult to believe that the hard-nosed Hamilton fully believed what he wrote. Possibly the best evidence that Hamilton didn’t fully believe what he wrote comes from Federalist No. 72, published a week after Federalist No. 68.[40] The immediate question in Federalist No. 72 was whether the President should be subject to term limits. Hamilton, who vehemently opposed term limits, argued that excluding the President from re-eligibility would produce “the temptation to sordid views, to peculation, and in some instances, to usurpation.”[41] He elaborated:

An avaricious man, who might happen to fill the office, looking forward to a time when he must at all events yield up the emoluments he enjoyed, would feel a propensity, not easy to be resisted by such a man, to make the best use of the opportunity he enjoyed while it lasted, and might not scruple to have recourse to the most corrupt expedients to make the harvest as abundant as it was transitory . . . .[42]

By contrast, a self-interested President with a chance at reelection might be less prone to plunder: “His avarice might be a guard upon his avarice,” in Hamilton’s words. The future Treasury Secretary continued:

An ambitious man, too, when he found himself seated on the summit of his country’s honors, when he looked forward to the time at which he must descend from the exalted eminence forever, and reflected that no exertion of merit on his part could save him from the unwelcome reverse; such a man, in such a situation, would be much more violently tempted to embrace a favorable conjuncture for attempting the prolongation of his power, at every personal hazard, than if he had the probability of answering the same end by doing his duty.[43]

Put more plainly, a term-limited President facing the end of his time in office might resort not only to corruption but also to self-coup. By contrast, a President who still had an opportunity to retain power through peaceful means would be more likely to pursue a path of electoral competition rather than risk the “personal hazard” (e.g., imprisonment or death) that self-coup, if unsuccessful, might entail. Contributing to Hamilton’s analysis was his very grim view of the post-presidency. Hamilton thought that former Presidents would have little to do—that they would “wander[] among the people like discontented ghosts, . . . sighing for a place which they were destined never more to possess.”[44] Term-limited executives might go to dangerous lengths to avoid such a dismal fate.

Upon first glance, Federalist No. 72 seems to contradict Federalist No. 68.[45] If presidential virtue was a “moral certainty” given the selection mechanisms outlined in Federalist No. 68, then what need would there be for incentive structures that align presidential self-interest with the public interest? But if we set aside Hamilton’s most grandiose claims about the excellence of the presidential selection process, we can understand Federalist No. 68 and Federalist No. 72 as harmonious visions for the new presidency. In the first, Hamilton describes mechanisms to ensure that the presidency will be occupied only by good men. In the second, Hamilton describes an incentive structure designed for the case in which the presidency is occupied by the Holmesian bad man.[46] In combination, presidential selection mechanisms and incentive structures would reduce pressure on institutional checks and balances: it would either be in the president’s nature or his self-interest to play by the rules.

  1. The Madisonian Model: Institutional
    Checks and Balances

Madison, like Hamilton, recognized the risk that one branch of the federal government might usurp the others, though Madison focused primarily on the risk that the legislature rather than the executive would arrogate power. “In republican government,” Madison wrote, “the legislative authority necessarily predominates” while “the weakness of the executive may require . . . that it should be fortified.”[47] Nonetheless, later generations of scholars and jurists have extracted from Madison’s writings a theory of checks and balances that applies beyond the particular problem of legislative overreach. This “Madisonian model”[48] offers an approach to the problem of self-coup that both contrasts with and complements Hamilton’s.

Whereas Hamilton claimed that presidential selection mechanisms would ensure that the highest office was occupied by persons “pre-eminent for ability and virtue,” the Madisonian model does not entail a starry-eyed view of executive character. “If angels were to govern men, neither external nor internal controls on government would be necessary,” Madison famously wrote.[49] For Madison, the ultimate challenge of constitutional design lay in configuring “the interior structure of the government” so that “its several constituent parts may, by their mutual relations, be the means of keeping each other in their proper places.”[50] Through these institutional checks and balances, the non-angelic individuals who would inevitably emerge as leaders of the respective branches could keep each other in line. As Madison put it in Federalist No. 51:

[T]he great security against a gradual concentration of the several powers in the same department, consists in giving to those who administer each department the necessary constitutional means and personal motives to resist encroachments of the others. The provision for defense must in this, as in all other cases, be made commensurate to the danger of attack. Ambition must be made to counteract ambition.[51]

Federalist No. 51 is among the most cited and celebrated of Publius’s papers.[52] As Daryl Levinson and Rick Pildes observe, “few of the Framers’ ideas continue to be taken as literally or sanctified as deeply by courts and constitutional scholars as the passages about interbranch relations in Madison’s Federalist 51.”[53] But Federalist No. 51 is also as frustrating as it is famous. For one thing, as Levinson notes, Madison never explained why the branches would “tend to compete rather than cooperate or collude”[54]—why, for example, the legislature and judiciary would act as adversaries rather than accomplices to an overreaching president. For another, Madison—focused as he was on legislative overreach—never convincingly explained how the legislature and the judiciary would exert checks and balances against the one branch that has all the guns. In Federalist No. 48, Madison questioned how “parchment barriers” erected by constitutional text could guard “against the encroaching spirit of power.”[55] By Federalist No. 51, he seems to have forgotten his own question.

In his defense, Madison’s incomplete sketch of interbranch checks constituted only one component of a more complex model of political stability. “In the compound republic of America,” Madison wrote, “the power surrendered by the people is first divided between two distinct governments” (federal and state), “and then the portion allotted to each subdivided among distinct and separate departments” (the executive, legislature, and judiciary).[56] The vertical separation of powers between federal and state governments would provide a “double security”: if Congress and the federal courts failed to rein in a runaway President (or, more worryingly to Madison, if the President and the federal courts failed to constrain an out-of-control Congress), state governments would stand at the ready to exert a supplemental check.[57]

Taken together, Hamilton and Madison thus offer us two general frameworks for thinking about the self-coup problem. We can try to ensure that the President has neither the inclination nor the incentive to attempt a self-coup (Hamilton), or we can seek to guarantee that any self-coup attempt will be blocked by some combination of Congress, the courts, and the states (Madison). The Framers pursued both strategies, and while the Hamiltonian and Madisonian models have proven to be rickety in various ways, they held up well enough for the first 230 years of the Republic to forestall a self-coup.

III. Hamiltonian and Madisonian
Models Over Time

  1. The Fall and Rise of the Hamiltonian Model
  2. Selection Mechanisms

From the outset, presidential selection in the new United States looked nothing like the process laid out in Federalist No. 68. Indeed, Hamilton helped to break the very model that he extolled. As Marty Cohen and coauthors note, “Hamilton himself led the behind-the-scenes organizing in 1788 to ensure that colleagues in each state chose electors pledged to the prearranged candidates, George Washington and John Adams.”[58] The notion that electors, once chosen in their respective states, would conduct “complicated investigations” into the character and qualifications of presidential candidates proved fanciful.[59]

The rise of political parties in the early American Republic drove the Federalist No. 68 model even deeper into obsolescence. Both parties[60]—the Federalists and the Republicans—held caucuses in May 1800 to select their candidates for national office. Contrary to Hamilton’s expectation that members of Congress would be excluded from the presidential selection process, members of Congress were the exclusive participants in these party caucuses.[61] And further abjuring any pretense that the President would be chosen by a body independent of the legislative branch, the Federalists even held their 1800 caucus in the Senate chamber.[62] Congressional nominating caucuses, derogatorily known as “King Caucuses,”[63] gave way by 1840 to national conventions at which state party leaders rather than members of Congress dominated,[64] but the Electoral College—the institution so vaunted by Hamilton—never gained an independent vetting role.

In one sense, then, the evolution of presidential selection represented a failure of the Federalist No. 68 vision: the process, in all its details, deviated dramatically from Hamilton’s plan. But in another, the party-dominated process represented the triumph of the Hamiltonian model by other means. Throughout most of the nineteenth and twentieth centuries, the presidential selection process combined popular legitimation with elite intermediation—albeit in a different order than Hamilton anticipated. Federalist No. 68 foresaw that voters would choose electors from among the elite, who would then meet in their respective states to choose a President. Instead, elites in each party chose their nominees, and general election voters then chose from a limited menu of party-endorsed options. Starting in the early 1900s, some states instituted primaries to elect delegates to the national party conventions, but until 1968, most delegates were chosen by state party conventions or state party committees.[65] Summarizing the key features of this “mixed system” of primaries and party control that endured through 1968, Rick Pildes writes: “Although winning a primary could influence the selection process, the dominant power to determine the nominees continued to rest with the traditional party figures.”[66]

The intermediation of party leaders protected the presidential selection process from the popular “heats and ferments” that Hamilton so feared. The presidential candidates who emerged from the national party conventions were not always “characters pre-eminent for ability and virtue.” After all, this was the system that gave us Warren G. Harding. But they were, in the main, institutionalists rather than populists.[67] The requirement that a candidate needed to win a majority of convention votes—or a two-thirds supermajority in the Democratic Party until 1936[68]—further stacked the deck in favor of consensus builders. The process did not necessarily weed out crooks, incompetents, or even narcissists,[69] but it did seem to succeed in weeding out candidates with inclinations toward self-coup.[70]

This “mixed system” of primary input and party control turned out to be one of the Vietnam War’s many casualties. At their 1968 national convention in Chicago, Democrats nominated then-Vice President Hubert Humphrey—a reluctant supporter of the Johnson administration’s Vietnam policy—even though Humphrey had not participated in any of the fourteen state primaries that year. Frustration with Humphrey’s nomination spilled out into the streets, where antiwar protesters clashed violently with heavy-handed Chicago police. The Democratic National Committee responded by empaneling a twenty-eight-member commission to rewrite the party’s presidential selection rules. Forged in the crucible of Chicago’s chaos, the McGovern-Fraser Commission (as it came to be known[71]) would go on to carry out a “quiet revolution”[72] that forever changed the process of presidential selection.

On their face, the McGovern-Fraser Commission’s recommendations were modest. Under the commission’s guidelines, which the national Democrats adopted, state parties could continue to select delegates via state conventions as long as those meetings satisfied certain procedural requirements designed to promote public participation.[73] But an unintended consequence of the McGovern-Fraser Commission was to instigate a shift toward primaries nationwide. Rick Pildes offers one reason for the shift: “Apparently, Democratic state parties were worried that if they failed to implement the new rules properly, their delegations would be subject to credentials challenges. . . . Party leaders in many states thought primaries would be simpler and safer.”[74] Political scientist James Ceaser suggests a supplemental explanation: Party leaders realized that if they implemented the McGovern-Fraser reforms, activists on the left—who were less numerous but more motivated than moderates—would be able to take over the state conventions, which required a more significant time commitment from attendees than a single trip to a polling station. The party leaders, who were closer to the political center than the activists, came to favor primaries because they thought their preferred candidates would perform better at the ballot box than on the state convention floor.[75] In other words, according to Ceaser’s theory, Democratic leaders adopted primaries in order to defend—not to relinquish—their own power.

Oftentimes, Democratic-dominated state legislatures codified the switch to primaries via statutes that were binding on both parties. Thus, as political scientist Elaine Kamarck observed in 1987, “it could be said that the Democratic party has, in a roundabout way, ‘reformed’ the Republican party since 1968 by increasing the number of presidential primaries for both parties.”[76] But the reforms after 1968 did not spell an immediate end to party control. Even after most states switched to primaries, insiders in both parties still shaped presidential selection through what journalist Arthur Hadley described as “the invisible primary”[77]—the roughly year-long period before the Iowa caucuses in which Governors, Senators, Congressmembers, and other elected officials and interest group leaders vetted candidates, made endorsements, raised funds, and mobilized volunteers.[78] Although a veritable outsider, former Georgia Governor Jimmy Carter, nabbed the Democratic nomination in 1976, he was the last candidate from outside either party establishment to do so in the twentieth century.[79] According to Marty Cohen and coauthors, the candidate who emerged from the invisible primary with the largest share of endorsements from party leaders went on to win the nomination in every Democratic and Republican contest from 1980 through 2000.[80] The pithy and provocative title of a 2008 book by Cohen and coauthors summed up their assessment of presidential nominations in the final two decades of the twentieth century: “The Party Decides.”[81]

In the new millennium, however, two formidable forces challenged the parties’ decisional power over presidential selection: money and media.[82] The advent of Internet fundraising allowed candidates to raise millions through small-dollar donations even when they lacked support from the officials who controlled party-linked purse strings. Meanwhile, twenty-four-hour cable news, talk radio, Facebook, and Twitter enabled presidential aspirants to reach voters directly without relying on the party apparatus to carry their message. Both the forces of money and media were on full display in the 2016 Democratic and Republican contests. On the Democratic side, Bernie Sanders showed that a candidate fueled by Internet fundraising could mount a serious challenge to the party’s preferred standard-bearer. On the Republican side, Donald Trump proved that a social media-savvy outsider could not only come close to winning the party’s nomination, but could actually seize the crown.

Even after 2016, it would be premature to write off the “party decides” thesis altogether. In 2020, Democratic leaders ultimately coalesced around a single candidate, Joe Biden, and the party successfully fended off an outsider challenge from Sanders.[83] And already as of this writing, former President Trump has begun to follow the traditional playbook of currying favor from state party leaders in preparation for a 2024 bid[84]—a sign that, at least in Trump’s mind, “the party” still wields significant sway. But clearly, the rise of small-dollar donations and the proliferation of cable television, talk radio, and social media have weakened one mechanism that kept the threat of self-coup in check through most of American history: elite intermediation of the presidential selection process. And that significant development, as we shall soon see, places more pressure on American democracy’s other Hamiltonian and Madisonian defenses.

  1. Incentive Structures

Just as the presidential selection process diverged in practice from Hamilton’s blueprint, presidential incentive structures evolved in ways that Hamilton did not anticipate. Hamilton expected that the perpetual prospect of reelection would keep current Presidents in line—that if the President “could expect to prolong his honors by his good conduct,”[85] he might sacrifice his ambition for money and power. Yet in practice, and for a variety of reasons, the prospect of reelection has played a limited role in aligning presidential incentives with national interests.

First, the bar on a twice-elected President seeking a third term stymied Hamilton’s vision of perpetual reelection as a disciplining device. In the early years of the republic, the bar was an informal norm rather than a prohibition codified in the Constitution, and the robustness of the informal norm before the mid-20th century is a subject of some debate[86]: a different norm—a one-term tradition—arguably existed in the 1840s and 1850s,[87] and the two-term norm was not so crystalline as to prevent several second-term Presidents from toying with the idea of a reelection run.[88] In any event, the Twenty-second Amendment, ratified in 1951, largely assured that reelection would not be a constitutional option for second-term Presidents.[89] From then on, the idea that reelection incentives might serve as a check on presidential greed and ambition—that a President’s “avarice might be a guard upon his avarice”—applied only, if at all, in a President’s first four years.[90]

Hamilton hardly can be faulted for failing to foresee the Twenty-second Amendment, which was still more than a century and a half in the offing when Publius published his essays. But there was an even more basic flaw in Hamilton’s incentive-structure model: Hamilton appears to have ignored the interregnum between election and inauguration—a period when, term-limited or not, a President may know with certainty that the reelection doorway is shut. Once the votes are counted and the incumbent President learns that he has lost, he no longer encounters the prospect of prolonging his power “by doing his duty.” The Twentieth Amendment, which moved Inauguration Day forward from March 4 to January 20,[91] shortened the interregnum from approximately four months to two-and-a-half, but even this shortened period is unusually long by international standards. By contrast, the interregnum between election and inauguration of a new head of government is typically less than a week in the United Kingdom,[92] ten days in France,[93] and a little more than two weeks in Canada.[94]

Insofar as the prospect of reelection serves to incentivize Presidents to choose democratic competition over extra-constitutional means of maintaining power, the interregnum is—for Presidents who have lost their reelection bid—a perilous period of incentive misalignment. That is the period in which defeated Presidents around the world often have attempted self-coups—with recent examples including Laurent Gbagbo in the Ivory Coast in 2010 and Yahya Jammeh in Gambia in 2016.[95] The timing of Trump’s attempted self-coup follows in that rhythm.

Arguably, what is most remarkable about Trump’s self-coup attempt is not that it happened, but that it had never happened before in American history. Why did all previous U.S. Presidents relinquish office so willingly (or, at least, nonviolently)? For one thing, presidential selection mechanisms helped to weed out power-hungry leaders so that presidential incentive structures weren’t often stress-tested. Moreover, political parties helped to solve the “last period problem” for term-limited Presidents by aligning their interests with their party’s future.[96] But perhaps most importantly, the post-presidency did not turn out to be nearly as bleak as Hamilton foresaw. Outgoing Presidents faced strong incentives not to attempt a self-coup and instead to enjoy a post-presidential life of riches and glory.

Indeed, America has treated its ex-Presidents quite well. Two former Presidents, George Washington and Ulysses S. Grant, were recommissioned as Army generals after they left office.[97] Two more, John Quincy Adams and Andrew Johnson, would return to Congress.[98] Former President William Howard Taft ultimately achieved his lifetime dream of becoming a Supreme Court Justice.[99] And several former Presidents found happy homes at universities.[100] Three former Presidents—Martin Van Buren, Millard Fillmore, and Theodore Roosevelt—ran unsuccessfully for another term on a third-party ticket, and Grover Cleveland ran successfully on his old Democratic Party line,[101] but aside from that handful (and, of course, the ongoing example of Trump), most former Presidents have adjusted happily to post-presidential life.

Since World War II, moreover, former Presidents have discovered that life after the White House can be not only pleasant but also extraordinarily lucrative. Weeks after leaving office in 1953, Harry Truman sold his memoirs to Life magazine for more than $6 million in today’s dollars.[102] Truman nonetheless prevailed upon Congress in 1958 to pass the Former Presidents Act, which now provides ex-Presidents with a pension of $235,000 per year.[103] Even that amount—ample by upper-middle-class standards—is a pittance compared to what former Presidents can earn through other sources. For example, Ronald Reagan received a total of $2 million for a 1989 trip to Japan that involved two twenty-minute speeches.[104] Bill Clinton raked in more than $100 million in speaking fees in the first dozen years of his post-presidency.[105] And Barack and Michelle Obama signed a package deal for their memoirs with Penguin Random House for a $65 million advance—possibly the largest nonfiction book contract ever.[106]

Some commentators bemoan the commercialization of the post-presidency,[107] but from an incentives perspective, there is a potential upside to the phenomenon. Most relevantly for our purposes, a sitting President who foresees a future of great wealth may be less likely to roll the dice and risk it all on self-coup. The Sudanese-British billionaire Mo Ibrahim has put this same intuition to work internationally, offering a prize worth more than $5 million to former African heads of state or government who leave office at the end of their constitutionally mandated terms.[108] The free market has given rise to an informal version of the Ibrahim Prize in the United States—and on a supersized scale.

Viewed from this vantage point, January 6 presents something of a puzzle: why did the “first billionaire president”[109]—who could have added even more to his fortune through million-dollar speeches and MAGA branding opportunities—gamble it all on a self-coup attempt that very well could have (and still may) land him in jail? One possible answer is that even before the November 2020 election, Trump had reason to fear post-presidential prosecution and imprisonment. The New York Times reported on the eve of the general election that Trump was privately concerned about criminal probes—including in his former home state—if he lost at the ballot box.[110] The concern proved to be well founded: as of this writing, Trump already has been indicted in New York state court on felony charges related to pre-presidential payoffs to a porn star.[111] The post-presidency is obviously less alluring if one thinks it will entail prison time, and the marginal deterrent effect of criminal liability for self-coup is weaker if a President expects to face criminal liability after he leaves office anyway. [112]

Anecdotal evidence from abroad supports the hypothesized link between post-presidential prosecution and self-coup. In 1993, Guatemalan President Jorge Serrano Elías launched an unsuccessful self-coup in the shadow of potential corruption charges.[113] And in Peru, which has endured one relatively recent self-coup and an even more recent attempt, all six Presidents elected since 1990 have ended up in jail or facing a detention order.[114] In Federalist No. 72, Hamilton worried about former Presidents “wandering among the people like discontented ghosts.”[115] In Peru, former Presidents are fortunate if they can wander at all.

Of course, the United States—where only one former President has ever been indicted—remains a far way off from Peru, where post-presidential prosecution has become par for the course. Yet within days of the March 2023 Trump indictment in the New York state court, some prominent conservatives already had begun to float the possibility that Republican state and local prosecutors might pursue a tit-for-tat criminal case against Joe Biden after he leaves office.[116] (Never mind the fact that no one had yet identified any Biden crime.) And if the United States were to follow down the Peruvian path, the threat to the Hamiltonian model would be two-pronged. Most directly, a sitting President’s incentives to relinquish power would be weaker if the United States replicated Peru’s presidency-to-prison pipeline. More obliquely, a high risk of post-presidential prosecution would potentially influence presidential self-selection. The individuals who would seek the presidency under those conditions would probably be more power-hungry, less risk-averse, and more inclined toward self-coup.

To recapitulate: Hamilton cleverly intuited that presidential incentives—along with presidential selection mechanisms—could reduce the risk of self-coup. But the particular incentive on which Hamilton focused—the prospect of reelection—turned out not to be up to the task. That is partly because of the advent of presidential term limits, though also because Hamilton overlooked the problem of the interregnum. Nonetheless, the post-presidency has evolved in such a way that sitting Presidents have strong self-interested incentives to relinquish power: Leave the White House without a fight and you will be showered with riches and honors. Hamilton may have had the precise mechanism wrong, but his general insight regarding the importance of incentive structures proved to be astute.

  1. The Madisonian Model Inside Out

Just as the rise of political parties upended the Hamiltonian vision of presidential selection, so too did it transform the Madisonian system of separated powers.[117] The close relationship between the President and members of Congress from his own party—cemented through the King Caucuses in the early nineteenth century and the “invisible primary” in later years—significantly reduced the probability that legislators would exert a check against a co-partisan chief executive whom they had effectively chosen. As Daryl Levinson and Rick Pildes have observed, only in periods of divided government should we expect to see legislative ambition counteracting presidential ambition along Madisonian lines.[118] And divided government is a contingent feature of American politics—more prevalent than not in the post-Cold War era, but still no sure thing.[119]

The late political scientist Juan Linz famously argued that the “uniquely diffuse character of American political parties” blunted the full impact of partisan competition through the late twentieth century.[120] As Linz wrote in 1990, the unusual ideological heterodoxy of the modern Democratic and Republican parties meant that partisan competition in the United States took on a different valance than the more intense conflicts in other presidentialist states.[121] Inverting Henry Kissinger’s quip about the Harvard faculty, national politics in twentieth century America were less bitter because the ideological stakes were generally small. Yet the phenomenon of “partisan sort”[122] in recent decades has changed the terms of party competition. With conservative Southerners finally leaving their historic home in the Democrats’ “big tent,” the two parties have grown increasingly ideologically cohesive. By the 2010s, this partisan sort was largely complete—at least among elites.[123] In periods of divided government, this makes the Madisonian model more applicable—though also more combustible.

Not only has partisan competition transformed the Madisonian model of separated powers at the national level, so too has it redefined the relationship between the federal government and the states. “Put in only slightly caricatured terms,” Jessica Bulman-Pozen writes, “Republican-led states challenge the federal government when it is controlled by Democrats, while Democratic-led states challenge the federal government when it is controlled by Republicans.”[124] Bulman-Pozen notes that as compared to divided government at the national level, conflictual federalism is much less contingent on electoral ebbs and flows: with fifty states, some are always aligned with the President and others aligned against.[125] Arguably, partisan federalism thus bolsters the “double security” feature of the Madisonian model, though it also nearly assures that the states will never operate as a unified front.

Until the 2020 election, Madisonian’s compound-republic model was never stress-tested against a self-coup attempt—at least, not here in the United States. That fact is arguably a testament to the success of the Hamiltonian vision. With selection mechanisms that favored norm-bound leaders, as well as payoff structures that aligned presidential incentives with the peaceful transition of power, interbranch checks functioned as a failsafe rather than a first line of defense.[126]

When the test finally came, the Madisonian failsafe held up surprisingly well.[127] To be sure, Trump’s attempted self-coup took place at a time of partly divided government: the Democrats—rather than Trump’s Republican co-partisans—controlled the House. Divided government is the easier case for the Madisonian model of interbranch checks—the acid test of the Madisonian model is whether the branches will check each other when they all are under the control of a single party. Yet even in the upper chamber, where the GOP still retained a one-seat majority on January 6, forty-two out of fifty Republicans joined all of their Democratic-caucusing colleagues in rejecting the Trump-backed objections to both the Arizona and Pennsylvania presidential results.[128] Notwithstanding fears that partisan loyalties would dominate constitutional responsibilities, most Republican Senators checked Trump’s antidemocratic ambitions just as Madison had imagined.

Yet it would be dangerous to overread the results of the Senate vote. The relationship between President Trump and Senate Republicans was unusual: only one Republican Senator endorsed Trump in the 2016 primaries,[129] and that Senator—Jeff Sessions of Alabama—had since departed to become Trump’s ill-fated Attorney General. The extraordinary breakdown of the “party decides” model thus left a wedge between President Trump and Senate Republicans. Put another way, the Madisonian model arguably “worked” on January 6 in part because the neo-Hamiltonian mode of presidential selection by party elites failed four years earlier.

When one looks over to the House GOP side, moreover, the outlook for the Madisonian model is more ominous. The overwhelming majority of the GOP’s lower-chamber caucus—139 out of 204—voted to sustain either the Arizona objection, the Pennsylvania objection, or both.[130] When these House Republicans were faced with a conflict between their constitutional role as members of the legislative branch responsible for checking the executive and their partisan loyalty to President Trump, partisan loyalty prevailed. The phenomenon that Levinson and Pildes describe as “separation of parties, not powers” was on full display.[131]

It should be no great shock that the Trump-backed objections to certification fared better among House Republicans than among their Senate co-partisans. Geographic polarization and gerrymandering have generated a congressional map in which the vast majority of districts are “safe seats”—so solidly red or blue that they are unlikely to change parties even in a wave election.[132] For most House incumbents, a primary challenge is more worrisome than the remote possibility of a general election defeat. Political incentives thus push House members to cater to the base rather than to the center. The situation is somewhat different in the Senate: many states combine blue and red regions—and state lines obviously can’t be gerrymandered each census cycle—so Senators have stronger incentives to appeal toward the middle.

The judiciary, for its part, mostly played along with Madison’s plan. Judges behaved like judges and not like partisans in robes: of the thirty votes cast by Republican-appointed federal judges in post-election cases, twenty-nine went against Trump.[133] In one case, a federal district judge appointed by George W. Bush issued a temporary restraining order barring Georgia officials from erasing voting machine data, but the judge did not grant the much broader relief that the Trump-aligned plaintiffs requested.[134] All of the federal judges appointed by Trump himself voted against him. At least in the shadow of self-coup, John Roberts’s insistence that “[w]e do not have . . . Trump judges” proved true.[135]

States—even Republican-led states—also fulfilled their function of providing a “double security” buttressing interbranch checks. In Arizona and Georgia, two states where Joe Biden won the 2020 vote by a narrow margin, Republican Governors certified election results even as a President of their own party badgered them not to. Sometimes, these refusals occurred in an almost cinematic fashion. While he was in the middle of signing papers certifying the Arizona results, Governor Doug Ducey’s cell phone sounded with a “Hail to the Chief” ringtone, signifying that the caller was the sitting President. Ducey silenced his phone and proceeded with the certification.[136] Days later, as Republican Governor Brian Kemp of Georgia and his family mourned the death of a close friend, Trump called Kemp and berated him to overturn the Peach State vote.[137] Kemp nonetheless certified the state’s results—twice.[138] Trump would later not-so-subtly threaten Georgia’s Republican Secretary of State Brad Raffensperger with criminal prosecution if Raffensperger failed to “find 11,780 votes” for Trump. Raffensperger refused and then released a tape of the phone call to the press.[139] The partisan pattern highlighted by Bulman-Pozen—that state officials check the federal government only when those state officials and their federal counterparts are members of different parties[140]—turned out to be, at least in this case, thankfully wide of the mark.

This happy Madisonian story—Congress, the courts, and the states all keeping the President in line—is still incomplete in an important respect. Arguably, the most powerful check against Trump after the 2020 election came from an angle not foreseen by Madison: from within Trump’s own executive branch. Madison’s confidence that officials would do the bidding of their respective branches—that the “interest of the man” would be “connected with the constitutional rights of the place”[141]—blinded Madison to the possibility of intra-branch resistance, or what scholars now describe as the “internal separation of powers.”[142]

Much of the “internal separation of powers” literature has focused on formal mechanisms that shield executive branch officials from presidential influence—mechanisms such as civil service protections and for-cause removal restrictions. But the internal resistance to Trump’s self-coup attempt came primarily from officials with no formal insulation from the President. Attorney General William Barr and high-ranking officials in Trump’s Homeland Security Department publicly undercut the President’s claims of widespread election fraud.[143] After Barr’s departure, his replacement, Deputy Attorney General Jeffrey Rosen, resisted Trump’s efforts to involve the nation’s largest law enforcement agency in the President’s bid to retain power.[144] Meanwhile, the Chairman of the Joint Chiefs of Staff, Mark Milley, reportedly coordinated with high-ranking officials across the executive branch to ensure that the military would not come to Trump’s aid.[145] The result was that even though the President formally controlled the branch with all the guns, the famous Jacksonian quip—“John Marshall has made his decision, now let him enforce it”[146]—applied to Trump himself. The President made his decision to overturn the 2020 election results, but he could not enforce it with the nation’s military and law enforcement agencies aligned against him.

The “double security” that Madison imagined—with both the coordinate federal branches and the states exerting their own checks on a rogue President—thus turned into a “triple security,” with the executive branch bureaucracy adding its own extra layer of defense. Arguably that third layer should make us half again as confident in the Madisonian model: institutional checks on presidential power turned out to be even more robust than Madison thought. But it would be a mistake to take a victory lap without returning afterward to the field of institutional reform. The breakdown of peer review in presidential selection and the phenomenon of partisan sort make self-coup attempts more likely going forward. And as the next Part details, Trump’s self-coup attempt—even though it was beaten back by a combination of external and intra-branch checks—also revealed cracks in the Madisonian armor. A future President with inclinations and incentives toward self-coup may be able to exploit those fissures unless the United States fixes them first.

  1. Restoring the Hamiltonian and Madisonian Models

Publius’s key insight regarding self-coup was that a combination of selection mechanisms, incentive structures, and institutional checks and balances could reduce the risk of a President going rogue. This final Part seeks to translate that insight into recommendations for reform. Section IV.A considers changes in the Hamiltonian spirit; Section IV.B approaches the problem from a Madisonian perspective. The resulting recommendations will not eliminate the self-coup threat, but they still could matter near the margins.

  1. The Hamiltonian Model Redux: Reinvigorating Selection Mechanisms and Incentive Structures
  2. Selection Mechanisms

As explained above, the first element of the Hamiltonian model was a presidential selection process that empowered elites to exercise a “peer review” role. And although the Electoral College never came to serve that function, the parties managed to regenerate systems of peer review that guided presidential selection through the end of the 1960s and arguably into the early 2000s. Trump’s 2016 victory, which set the stage for his later self-coup attempt, illustrates the potentially catastrophic consequences of the recent shift toward a more plebiscitary mode.

Granted, a reversion to the 1960s status quo lies well outside the Overton window of political possibility circa 2023. As Elaine Kamarck writes, “it would be nearly impossible to turn back the clock” and allow the Democratic and Republican party establishments to choose their respective presidential nominees.[147] Primaries and broadly participatory caucuses have themselves become part of our lower-case-c constitution. As Kamarck observes, “the public regards its role in the nomination process as a right,”[148] even though that “right” is of relatively recent vintage.

Rather than a counterrevolution, Kamarck and other scholars have suggested incremental reforms to the nomination process that could make peer review somewhat more robust. One set of proposals focuses on “superdelegates,” who receive roughly one-sixth of the votes at the Democratic National Convention.[149] These superdelegates—sitting Governors, Senators, Congressmembers, elected members of the party national committee, and certain “distinguished party leaders” (e.g., former Presidents, Vice Presidents, House and Senate leaders, and national party chairs)—are among the few institutionalized vestiges of peer review still remaining. Under a 2018 rule change, superdelegates cannot vote on a contested first ballot at the Democratic convention; their votes become relevant only if no candidate captures a first-ballot majority.[150] To strengthen peer review, Democrats could restore the superdelegates’ ability to vote on the first ballot, and Republicans—who do not use superdelegates—could adopt the Democratic practice. [151]

In addition to changes at the national convention, parties also could try to inject formal peer review mechanisms earlier into the nomination process. Kamarck has suggested a pre-primary “national endorsing convention” at which party insiders would seek to coalesce around a single candidate.[152] Walter Shapiro has proposed that each party empower a “blue-ribbon group” to vet candidates for inclusion in televised debates.[153] To facilitate Shapiro’s suggestion, the Federal Election Commission would probably need to amend regulations that require debate staging organizations such as television networks to use only “objective” criteria for participation.[154] (Alternatively, the parties themselves—which are not subject to the staging organization limitations—could sponsor their own debates, albeit at their own expense.)[155]

Reasonable as they are, these proposals are all quite modest, and it is unclear whether—alone or in combination—they could have stopped Trump from capturing the Republican nomination in 2016. There is even a risk that robust party-based peer review could backfire and actually increase the likelihood of a presidential power grab. Parties are endogenous political institutions: parties shape presidential selection, but Presidents also shape their own parties. Trump—through his active involvement in state Republican committee races—has populated the GOP establishment with loyalists who echo his election denial claims.[156] At least on the Republican side, there is now less reason to expect that intermediation by party elites would favor candidates who are committed to norms of democratic competition. It could instead accomplish the opposite.

Worries about empowering a Trumpist GOP establishment should not necessarily deter Democrats from strengthening peer review mechanisms in their party. Although Republicans followed along with the Democrats’ shift toward primaries in the 1970s and early 1980s, the congruence between the two parties’ selection processes post-McGovern-Fraser reflected the fact that Democrats codified many of their reforms in state statutes. Since the reforms suggested by Kamarck and Shapiro would not be implemented via state-level legislation, there is less of a risk that Democrats’ changes would catalyze similar party-empowerment moves on the Republican side. The one exception is that any Federal Election Commission rule change to facilitate pre-screening of debate participants would apply to both parties. In that respect, empowering Democratic Party elites to exercise more of a peer-review role might well strengthen pro-Trump forces on the Republican side.

A second backfire-related concern arises from the interaction between the Hamiltonian and Madisonian logics: insofar as peer review tightens the link between the President and co-partisans on Capitol Hill and in Governors’ mansions, it potentially undermines the horizontal and vertical separation of powers. After all, when “the party decides,” sitting Senators, Congressmembers, and Governors are among the most influential deciders. As noted above, one collateral consequence of the demise of peer review in 2016 was the emergence of daylight between President Trump and other Republican elected officials—including Senators and Governors who did not endorse him in the pre-2016 “invisible primary.”[157] Whatever the ex ante benefits of party-based peer review in screening out candidates who might be inclined toward self-coup, there is a potential ex post cost in that Senators, Congressmembers, and state-level officials may be less willing to check a co-partisan President whom they played a hand in choosing.

In the end, the balance of considerations still may favor a more robust role for party-based peer review in presidential selection. Arguably, the primary threat to American democracy—more so than polarization per se or the possibility of a presidential power grab—is paralysis: the inability of government, even unified government, to get things done.[158] A tighter link between the President and co-partisans in Congress may make it easier for the United States to enact major legislation in times of unified government. But party-based peer review is no panacea, especially when one of the parties has become thoroughly Trumpified. While Hamilton thought that a well-designed presidential selection process would afford a “moral certainty” of a qualified and virtuous President, we would be wise to look beyond presidential selection for durable solutions to the problem of self-coup.

  1. Incentive Structures

In addition to his focus on presidential selection mechanisms, Hamilton highlighted incentive structures that would discourage Presidents from pursuing self-coup. While Hamilton’s emphasis on the perpetual prospect of reelection may have been misplaced, his underlying insight—that even a President who is hungry for money and power might be motivated to play by the rules of the constitutional game—still can lead us along promising paths toward institutional reform.

Hamilton’s analysis of presidential incentive structures anticipates a rational choice theory of self-coup: a President will pursue self-coup, Hamilton suggests, only if the expected value of a self-coup exceeds the expected value of the next best alternative. The expected value of a self-coup, moreover, must account for the “personal hazard”—the probability that the self-coup will fail and the punishment if it does. The next best alternative may entail running for reelection, or it may entail a quiet return to private life. The President in this analysis plays the role of the Holmesian bad man: “avaricious,” as Hamilton says, but not irrational.[159]

One implication of this rational choice approach is that the risk of self-coup is highest when pathways of electoral competition are closed off—particularly in the interregnum between a November general election that the incumbent has lost and Inauguration Day on January 20. In this period, the probability of retaining office through electoral means is effectively zero. A President may anticipate defeat even before votes are counted in November, but until ballots are tallied, an incumbent President may retain at least a scintilla of optimism about reelection. Only on Election Night (and sometimes substantially later) does the hope vanish.

As Sandy Levinson has noted in this journal (though not in the specific context of self-coup), Congress could shorten the interregnum between election and inauguration without any constitutional change.[160] The Twentieth Amendment, adopted in 1933, provides that presidential and vice presidential terms must end at noon on January 20.[161] No constitutional provision, though, requires that elections be held in November. Congress thus could shift Election Day to early January without a constitutional problem.[162] This would not eliminate the risk of a self-coup during the interregnum—an electorally defeated President still might strike fast—but it would cabin the peak risk of self-coup to a much shorter time frame.

The risk of self-coup is, to be sure, not the only factor for Congress to consider in setting the election date. A long interregnum potentially hamstrings the federal government’s response to crises foreign and domestic.[163] On the other hand, the only way to shorten the interregnum without a constitutional amendment (or a change to the calendars of the world’s major religions) would be to hold a vote during or immediately after the holiday season, when many voters would be distracted from their civic duties. Too, a midwinter election date would raise the risk of a snowstorm interfering with voting in northern regions of the country.[164]

Those seasonal constraints may be sufficient to outweigh the argument for a later Election Day. But even without any change to the length of the interregnum, American legal and political institutions could better align presidential incentives with the peaceful transition of power by making the post-presidency more attractive and self-coup less so. Given how lucrative the post-presidency is already, it might be difficult to do this with financial carrots—perhaps a large cash payout for Presidents who leave office without a fight would make sense in the United States, but it would likely require several more zeroes than the Ibrahim Prize.[165] A potentially more promising approach involves laws and policies around post-presidential prosecution and immunity, which can shape both carrots and sticks.

Post-presidential prosecution and immunity affect self-coup incentives in two conflicting ways. Immunity would lower the stakes of presidential succession and make the post-presidency more attractive, and for that reason, it would dull the incentive to attempt self-coup. On the other hand, criminal liability is itself a deterrent to self-coup. A self-interested President will be less likely to attempt to retain power through extraconstitutional means if he believes that criminal consequences may follow.

A possible middle ground is to adopt a policy of post-presidential immunity for all offenses other than crimes related to self-coup. Thus, a former President would not face the threat of prosecution for run-of-the-mill corruption, financial misconduct, and so on, but could face prosecution and conviction for “crimes against democracy.”[166] Under such a policy, former President Trump would be immune from prosecution for bank fraud and tax evasion, but still could face criminal charges for his involvement in the January 6 attack. Post-presidential prosecution would be designed principally to improve prospects for the peaceful transition of power—through immunity in one set of cases and liability in another.

This middle-ground approach to post-presidential prosecution and immunity seems sensible at first glance and again at the second, but it encounters at least three potential concerns. First, the category of prosecutable crimes would be difficult to define with precision. A violent attack on another branch of government would surely qualify, but what about a burglary of the opposition party’s national headquarters during an election year? Too narrow a definition of “crimes against democracy” would give the President impunity to pursue a wide range of offenses—some even involving political violence—so long as they fall short of a textbook self-coup. Too expansive a definition would undermine the whole point of differentiating between crimes against democracy and other offenses.[167]

Notwithstanding these definitional difficulties, there still would be clear cases on either side of the liability/immunity divide. Under the middle-ground approach, an ex-President would be immune from criminal liability for perjuring himself in grand jury testimony regarding a sexual relationship with a White House intern, as serious a crime as perjury may be. Likewise, an ex-President would be immune from criminal liability for lying about the size of his penthouse on a loan application.[168] Paying off a porn star and failing to report the payment to the Federal Election Commission would also fall on the immunity side of the divide, even though—in a strained sense—one could describe that as a “crime against democracy” because it hides information from voters.

But would the middle-ground approach be a stable norm? A second concern is that any liability/immunity distinction, however sensible, might not be honored. Even if the distinction were enshrined into the Constitution, the current President still might lack confidence that it would be respected. After all, a President who is contemplating self-coup himself surely must consider the possibility that his successor also might disregard the Constitution’s commands.

A final concern is that post-presidential immunity for offenses unrelated to self-coup would entail a sacrifice of accountability. Presidents might be less likely to engage in self-coup if they knew they would be immune from post-presidential prosecution for corruption, but by the same token, they would be more likely to engage in corruption.[169] And if post-presidential immunity applied to post-presidential offenses as well, it would give each former President a “get out of jail free” card for life. Decisions about post-presidential prosecution and immunity thus entail a difficult balancing among different evils. The tradeoff is real, and there is no easy out.

An advocate for the middle-ground approach might respond that if we are compelled to pick among poisons, the poison of presidential corruption is less toxic than the potentially fatal danger to the Republic of self-coup. To paraphrase Blackstone: It is better that ten Teapot Domes go unpunished than that we suffer one more January 6. This is, concededly, a contestable cost-benefit calculation. The rot of corruption can kill a democracy too, albeit at a more gradual pace than the sudden blow of self-coup. But it seems reasonable to say that in the United States in the 2020s, the risk of executive-initiated political violence is a greater threat to democracy than are bribery and graft. At the very least, the analysis here suggests that the effect on self-coup incentives should be one factor that federal, state, and local law enforcement officials carefully consider when deciding whether to prosecute a former President.

  1. The Madisonian Model Redux: Strengthening Institutional Checks and Balances
  2. The Vacancies Act and the Interaction Between the Internal and External Separation of Powers

As observed above, the failure of Trump’s self-coup demonstrated the strength of the Madisonian model: not only did both layers of Madison’s “double security” hold up, but a third layer—the internal separation of powers—showed its solidity. Yet the episode also shed light on a critical weakness in the Madisonian model—one that Trump initially tried to exploit. The flaw is this: The Senate, through its advice-and-consent power under Article II, is supposed to have a say in the appointment of the executive branch’s principal officers,[170] but as a practical matter Congress has given the President free rein to fill key posts such as Defense Secretary and Attorney General with cronies. The upshot is that a President can fire officials who resist a self-coup and replace them with loyalists who will go along for the ride. Not only does this take away an important external check on the President, but it potentially undermines the internal separation of powers too.

The primary source of this surprising vulnerability is the Federal Vacancies Reform Act of 1998, known colloquially as the Vacancies Act. That statute establishes the order of succession when an executive branch office becomes vacant.[171] As a default rule, the “first assistant” to the office (e.g., the Deputy Secretary of Defense in the case of the Secretary of Defense,[172] the Deputy Attorney General in the case of the Attorney General[173]) fills the office in an acting capacity.[174] But instead of following the default, the President may choose from any Senate-confirmed appointee, even one from outside the relevant agency, or any one of the hundreds or thousands of officers and employees within the agency who are paid at or above the GS-15 level and meet a minimum length-of-service requirement.[175] The presidential designee then may serve in the acting role for up to 210 days[176] (longer under some circumstances[177]).

Six days after the 2020 general election, President Trump took advantage of this wide latitude when he fired Defense Secretary Mark Esper via tweet and—in lieu of Esper’s deputy—named Christopher Miller, then the Director of the National Counterterrorism Center, to take Esper’s place.[178] Miller’s role in the runup and response to the January 6 attacks remains murky,[179] but Miller acknowledged in testimony to the House Select Committee on the January 6 Attack that at one point, he called a U.S. military attaché in Rome at the White House’s behest to investigate a baseless conspiracy theory that an Italian satellite had switched votes from Trump to Biden.[180] This admission arguably indicates that Miller was willing to participate—at least to some degree—in Trump’s efforts to undermine confidence in the election results.

For a brief period in late 2020 and early 2021, President Trump also considered using the Vacancies Act to install a loyalist atop the Justice Department. By that point, Trump’s second Senate-confirmed Attorney General, William Barr, had resigned—reportedly because Barr refused to validate Trump’s spurious allegations of election fraud and because Barr declined to publicize Justice Department investigations into Joe Biden’s son Hunter.[181] Under the statutory default rule, Barr’s first assistant—Deputy Attorney General Jeffrey Rosen—became Acting Attorney General after Barr’s departure. But Trump reportedly planned to install Jeffrey Clark, the acting head of the Justice Department’s Civil Division, in Rosen’s place. Per the plan, Clark would use his newfound powers as Acting Attorney General to pressure Georgia state lawmakers to overturn Biden’s electoral win in that state. Trump jettisoned the plan only after other senior Justice Department officials told the President that they would resign en masse if Clark replaced Rosen.[182]

The experiences at the Defense and Justice Departments in late 2020 and early 2021 should spur Congress to revisit the Vacancies Act. As Ben Miller-Gootnick argues in a superb student note, the statute’s legislative history gives little indication that members of Congress focused on the question of firings when they enacted the law in the late 1990s.[183] The text of the statute is itself ambiguous on this question. By its terms, it applies when an officer “dies, resigns, or is otherwise unable to perform the functions and duties of the office.”[184] At least arguably, a fired officer still is otherwise able to perform the functions and duties of the office. Nonetheless, the Office of Legal Counsel—the arm of the Justice Department that provides legal advice to the executive branch—has interpreted the Vacancies Act to apply when the relevant vacancy is created by the firing of the prior officeholder.[185] Absent legislative change or a judicial construction to the contrary, the Office of Legal Counsel’s interpretation is—for practical purposes—the one that matters.

Setting aside the statutory interpretation question, what would a normatively desirable version of the Vacancies Act look like? Anne Joseph O’Connell canvasses the competing considerations in a comprehensive study of acting officers. On the one hand, the status quo potentially allows the President to circumvent the Senate confirmation process by firing a Senate-confirmed officer and replacing her with someone whom the Senate hasn’t vetted for the post.[186] On the other hand, if the Vacancies Act didn’t apply to firings, “[t]he outgoing administration’s top officials” might “refuse to resign on January 20,” thereby hobbling the incoming President.[187] O’Connell suggests—as a “middle ground”—excluding firings from the scope of the Vacancies Act “but not in the first six months of a new administration as the incoming President gets their new team in place.”[188] Even a temporally narrower change might reduce the risk of self-coup. Congress could amend the Vacancies Act to prohibit the President from filling a principal officer post with anyone other than a Senate-confirmed first assistant during the period between Election Day and Inauguration Day. The President’s power to bulldoze over intra-branch checks would thus be weakest when the risk of self-coup is highest.

Indeed, Vacancies Act reform could be further limited to a small set of agencies and still have bite. The military is undoubtedly the institution whose support is most important to the success of a self-coup.[189] A key distinction between Fujimori’s successful self-coup in Peru in 1992 and Guatemalan President Jorge Serrano Elías’s failed self-coup the following year is the fact that Fujimori maintained military backing while Serrano did not.[190] While a President theoretically could go over the heads of military leaders to issue orders directly to the rank and file, the probability of a self-coup’s success is almost certainly greater if the President can control the armed forces through the regular chain of command. In that case, rank-and-file soldiers carry out self-coup simply by following orders that come from the top down. To limit that risk, Congress could amend the Vacancies Act to prevent the President from circumventing the default order of succession at the Defense Department and a small set of other agencies with significant law enforcement and intelligence responsibilities (e.g., the Justice Department, the Central Intelligence Agency, and the National Security Agency).

To be sure, a President who is prepared to carry out a self-coup also might be willing to disregard the Vacancies Act. Even if a statute says that the Defense Secretary can be replaced only by the Deputy Defense Secretary, the President might dismiss the Pentagon chief and designate someone other than the next in line to take the place. In that case, members of the armed forces would face a choice between following orders from the President’s pick or from the statutory successor. Although there is no guarantee that they would choose the latter, the statute can—at the very least—increase the probability that the military rank and file will recognize the President’s pick as illegitimate.

This discussion of the Vacancies Act reveals the fact that statutes, like constitutions, are parchment paper barriers. And the episode at the Justice Department in early 2021 suggests that statutory authority is not always outcome-determinative. The modest claim here is that members of the armed services, law enforcement agencies, and intelligence community are less likely to follow orders from a presidential designee who lacks statutory legitimacy than from an acting Defense Secretary, acting Attorney General, or acting Director of Central Intelligence whose authority is confirmed by law. But we should be clear-eyed about the fact that any effect here on the probability of self-coup operates at the margins. Again, all that law can do is reduce the risk of self-coup—and even then, maybe not by all that much.

  1. Congress and the Courts

Just as law cannot necessarily stop a self-coup, courts can’t always either. Recognizing as much, courts often capitulate when other political actors seize power through extra-constitutional means.[191] Still, courts can play an important role in shaping other actors’ responses to self-coup. For example, the refusal of the Guatemalan Constitutional Court to validate Serrano’s attempted self-coup in 1993 is sometimes cited as one reason why the military withdrew support.[192] Likewise, the U.S. Supreme Court’s refusal to intervene in the 2020 election almost certainly helped to seal Trump’s fate.[193] So long as courts are viewed by other institutional actors and by the general public as legitimate expositors of what the law is—and so long as those other institutional actors and members of the general public are inclined to follow what they perceive the law to be—judicial decisions can shape the success or failure of a self-coup even though courts typically lack any enforcement authority of their own.

Accordingly, a President who aspires to retain power through extra-constitutional means will often want to stack the courts with loyalists.[194] In the United States, the very fact that a President has appointed a particular Supreme Court Justice produces an empirically documented “loyalty effect”: the Justice is even more likely to vote in the appointing President’s favor than Justices appointed by previous Presidents of the same party.[195] The more Justices a President appoints, the more likely it seems that the Supreme Court will support the President in a self-coup. And with Supreme Court support, it becomes likelier that other institutional actors (including the military and law enforcement agencies) will fall into line.

The risk of self-coup thus leads to one argument against “court packing,” since court packing presumably would allow the sitting President to appoint a rash of loyalist Justices. And beyond court packing, the analysis of self-coup sheds light on two other ideas for judicial reform. The first is a much-discussed proposal to cap the tenure of Supreme Court Justices at eighteen years, with staggered terms ending every other year.[196] This would result in a President having appointed four of the Court’s nine Justices by the end of her second term—potentially five or more Justices in the event of an early death, resignation, or removal. Term limits would thus make it easier for a President to stack the Court with loyalists who might be likelier to validate a self-coup attempt.

While the analysis of self-coup offers one reason to reject eighteen-year term limits, it casts a more positive light on a separate proposal for the “decoupling” of Supreme Court appointments. Under that proposal, Justices would continue to enjoy life tenure, but each new President would be able to make two appointments per four-year term regardless of the number of vacancies that arise. Thus, vacancies and appointments would be “decoupled.” The new appointments, meanwhile, would not take effect until after the appointing President leaves office. Thus, the timing of Senate confirmation also would be “decoupled” from the time at which a new appointee ascends to the bench. The membership of the Court would fluctuate in size, growing by two Justices when a one-term President leaves office and by four Justices when a two-term President leaves office, and then likely declining over the next four to eight years as sitting Justices die or retire.[197]

Whereas the proposal for staggered eighteen-year terms generally would mean that a second-term President would have appointed a near-majority or majority of the Court, decoupling would mean that a sitting President would have appointed zero Justices to the Court.[198] Thus, the concern about the President stacking the Court with loyalists would be reduced. Justices still might turn out to be motivated by party loyalty, but they wouldn’t be animated by loyalty to the particular President who appointed them.

To be sure, decoupling—like every other judicial reform proposal—runs into the challenge of time consistency. Congress could enact the decoupling proposal today and then a future President, with the support of a narrow legislative majority, could choose to pack the Court (e.g., by passing special legislation to let all of her appointees-in-waiting ascend to the bench immediately). But here, decoupling enjoys an advantage over term limits. With term limits, the President can garner the support of a large minority—potentially a majority—of the Supreme Court without engaging in packing. Thus a President who pursues self-coup potentially can gain a stamp of approval from a Court that continues to command public legitimacy. By contrast, a President who acts against a legal backdrop of decoupling by packing the Court would, at best, gain the stamp of approval for his self-coup from a tribunal whose stamp no longer matters as much.

Ultimately, courts on their own can’t save a democracy from coup or self-coup. But courts can play a role in determining how other institutional actors respond to a presidential power grab. Self-coup risk is not, of course, the only consideration that should enter into discussions of Supreme Court reform. But it is, at the very least, a consideration relevant not only to court packing but also to more modest proposals for staggered and time-limited terms.


Donald Trump has reminded us, in violent fashion, of the risk that a President might try to shred the constitutional rulebook. Fortunately, Hamilton and Madison left us with a set of general strategies for countering future self-coup attempts. Time has transmogrified the Hamiltonian and Madisonian models, but the Framers’ focus on selection mechanisms, incentive structures, and institutional checks and balances remains as relevant today as 235 years ago. Following these Hamiltonian and Madisonian through lines across the centuries can help us understand how we as a nation arrived at January 6. Alas, Publius’s ghost will not tell us where to go from here. But he might at least be able to point us in the right direction.





        [1].    Professor of Law, New York University School of Law; The author thanks Jill Hasday, Daryl Levinson, and Rick Pildes for insightful comments on earlier drafts. Danny Levin provided stellar research assistance.

        [2].    See, e.g., Ted Barrett, Manu Raju & Peter Nickeas, US Capitol Secured, 4 Dead After Rioters Stormed the Halls of Congress to Block Biden’s Win, CNN (Jan. 6, 2021, 3:33 AM),

        [3].    Cohen v. State, 195 A. 532, 534 (Md. 1937).

        [4].    See, e.g., What Do We Call What Happened on January 6th?, Cline Ctr. for Advanced Soc. Rsch., (Jan. 8, 2021), (stating that “the Cline Center’s approach would exclude use of the term ‘riot’ to characterize the violence” on January 6 because according to the center’s definition, “a riot is a particular type of politically motivated violence that spontaneously erupts from an already tense situation” whereas with respect to the planned attack of January 6, “this requirement for spontaneity is not satisfied”).

        [5].    Then-Senate Majority Leader Mitch McConnell was among the first to use the I-word. See Nicholas Fandos, Emily Cochrane & Marc Santora, Congress Confirms Biden’s Election, Hours After a Pro-Trump Mob Stormed the Capitol, N.Y. Times (Jan. 7, 2021), (quoting McConnell’s description of the attack as a “failed insurrection”).

        [6].    “Insurrection,” Am. Dictionary of the Eng. Language (1828).

        [7].    See, e.g., David Pion-Berlin, Thomas Bruneau, and Richard B. Goetze Jr., The Trump Self-Coup Attempt: Comparisons and Civil–Military Relations, Gov’t & Opposition 1, 4 (2022); Charles T. Call, No, It’s Not a Coup—It’s a Failed ‘Self-Coup’ That Will Undermine US Leadership and Democracy Worldwide, Brookings: Order from Chaos (Jan. 8, 2021),; Fiona Hill, Opinion, Yes, It Was a Coup Attempt. Here’s Why, Pol. Mag. (Jan. 11, 2021, 3:15 PM),; Christopher Ingraham, How Experts Define the Deadly Mob Attack at the U.S. Capitol, Wash. Post (Jan. 13, 2021, 6:00 AM),; Joshua Zeitz, Ask the ‘Coupologists’: Just What Was Jan. 6 Anyway?, Pol. Mag.
(Aug. 19, 2022, 4:30 AM), (quoting New York University professor Ruth Ben-Ghiat).

        [8].    See Mitra Taj, Peru’s President Tried to Dissolve Congress. By Day’s End, He Was Arrested, N.Y. Times (Dec. 7, 2022),

        [9].    See Monty G. Marshall & Donna Ramsey Marshall, Coup D’État Events,
Ctr. for Sys. Peace (Jan. 7, 2022),; Buddy Peyton, Joseph Bajjalieh, Dan Shalmon, Michael Martin & Jonathan Bonaguro, Cline Center Coup D’état Project Dataset (Cline Center for Advanced Social Research 2020). This count does not include instances of “term limit evasion,” which are much more common than outright self-coups. See Mila Versteeg, Timothy Horley, Anne Meng, Mauricio Guim & Marilyn Guirguis, The Law and Politics of Presidential Term Limit Evasion, 120 Colum. L. Rev. 173, 194 (2020) (defining “term limit evasion” as “[w]hen a leader successfully stays in office past the end of his term”); id. at 199 tbl.1 (tallying 60 attempts at term limit evasion around the world since 2000). According to Mila Versteeg and coauthors, two-thirds of term limit evasion attempts involve efforts to “[a]mend the constitution to eliminate term limits or extend [the] number of terms,” see id., whereas the Center for Systemic Peace and the Cline Center code events as self-coups only if they involve extra-constitutional measures. See id. at 3; Buddy Peyton, Joseph Bajjalieh, Dan Shalmon, Michael Martin, Jonathan Bonaguro & Scott Althaus, Cline Center Coup d’État Project Dataset—Codebook, Ctr. for Advanced Soc. Rsch., 1, 5 (Feb. 23, 2023),

      [10].    Adrian Vermeule, Concepts in Law: Regulating Political Risks, 47 Tulsa L. Rev. 241, 243 (2011) (reviewing Bruce Ackerman, The Decline and Fall of the American Republic (2010)).

      [11].    A search of the LexisNexis database of U.S. law reviews and journals churns up just 37 articles before 2021 that include both the terms “self-coup” and “constitution” in their body text—and 36 of those articles are focused on self-coups or the risk of self-coup outside the United States (Vermeule’s article is the one exception).

      [12].    Mila Versteeg and coauthors have compiled a database of every democratically elected leader who sought to stay in office beyond the end of his constitutional term since 2000 and note that “every one of the leaders that attempted to overstay was male.” See Versteeg et al., supra note 9, at 182–83.

      [13].    On King Charles I’s participation in the abortive “Army Plot” of 1641, see Conrad Russell, The Fall of the British Monarchies 1637–1642, at 291–95 (1995).

      [14].    On Gustav III’s coup, see generally Michael Roberts, The Age of Liberty: Sweden 1719–1772 (1986).

      [15].    The Federalist No. 22 (Dec. 14, 1787), 18th_century/fed22.asp (Alexander Hamilton).

      [16].    I use the term “constitution” with a “lower-case c” to refer to “the set of rules, norms, institutions, and understandings” that make up our government. See Richard Primus, Rulebooks, Playgrounds, and Endgames: A Constitutional Analysis of the Calabresi-Hirji Judgeship Proposal, Harv. L. Rev. Blog (Nov. 24, 2017),

      [17].    On the relationship between the Federalist Papers and the Framers’ true beliefs, see Clinton Rossiter, Introduction to The Federalist Papers, at xv (1961); and Quentin Taylor, The Mask of Publius: Alexander Hamilton and the Politics of Expediency, 5 Am. Pol. Thought 55, 62–63 (2016).

      [18].    Cf. Aziz Z. Huq & Tom Ginsburg, Democracy without Democrats, 6 Const. Stud. 165, 166–67 (2020) (distinguishing between “gestalt” and “granular” approaches to the study of democratic backsliding).

      [19].    For example, while the Essay discusses rules regarding presidential primaries and caucuses, it does not discuss voting procedures in general elections. See Richard H. Pildes, Election Law in an Age of Distrust, 74 Stan. L. Rev. Online 100 (2022), (discussing institutional reforms designed to increase public acceptance of election results). This omission is not meant to minimize the importance of election administration to the peaceful transition of power. One likely reason why Brazil avoided a self-coup in 2022—despite suggestions from then-President Jair Bolsonaro that he might reject the results—is that the country, which uses an electronic voting system, counted almost all of its ballots within hours of polls closing.
See Philip Bump, The Uncomplicated Reason Brazil Can Count Its Ballots So Quickly, Wash. Post (Oct. 31, 2022, 12:54 PM),

      [20].    See Maxwell A. Cameron, Self-Coups: Peru, Guatemala, and Russia, 9 J. Democracy 125, 125 (1998).

      [21].    See Pion-Berlin, supra note 7, at 1.

      [22].    Nancy Bermeo, On Democratic Backsliding, 27 J. Democracy 5, 5 (2016).

      [23].    See Mariana Alfaro, Juan Guaidó, Venezuela’s Opposition Leader, Was Just Sworn In as the Nation’s Interim President Amid Huge Protests Against Nicolas Maduro, Bus. Insider (Jan. 23, 2019),

      [24].    See Marshall & Marshall, supra note 9; Peyton et al., supra note 9.

      [25].    Caesar’s consolidation of power is a borderline case of self-coup because the legislature acquiesced in its disempowerment: the Roman Senate voted to make Caesar “dictator perpetuo.” See Paul Chrystal, Rome: Republic Into Empire—The Civil Wars of the First Century BCE 117 (2019).

      [26].    See, e.g., The Rump Dissolved, UK Parliament (Jun. 4, 1787), parliamentaryauthority/ civilwar/overview/rump-dissolved/ (last visited May 6, 2023).

      [27].    See The Federalist No. 22, supra note 15.

      [28].    See Madison Debates, The Avalon Project (Jun. 18, 1787),

      [29].    These run from Federalist No. 67 through Federalist No. 77.

      [30].    See The Federalist No. 68 (Mar. 14, 1788) (Alexander Hamilton),

      [31].    Id.

      [32].    Id.

      [33].    Id.

      [34].    Anthony Downs, An Economic Theory of Political Action in a Democracy, 65 J. Pol. Econ. 135, 146 (1957).

      [35].    In a country with a population of less than 4 million in 1790, elites in the Electoral College also were more likely to be acquainted personally with the presidential candidates. Cf. 1790 Fast Facts, fast_facts/1790_fast_facts.html (last visited Apr. 5, 2023).

      [36].    The Federalist No. 68, supra note 30 (noting that “it might be unsafe to permit less than a majority to be conclusive”).

      [37].    Id.

      [38].    See U.S. Const. art. II, § 1, cl. 2. (“no Senator or Representative, or person holding an Office of Trust or Profit under the United States shall be appointed an Elector.”).

      [39].    See The Federalist No. 68, supra note 30.

      [40].    See The Federalist No. 72 (Mar. 21, 1788) (Alexander Hamilton),

      [41].    Id.

      [42].    Id.

      [43].    Id.

      [44].    Id.

      [45].    Cf. Alan Z. Rozenshtein, The Virtuous Executive, 108 Minn. L. Rev. (forthcoming 2023), (manuscript at 21) (“[T]he tensions between a politics of virtue and a politics of self-interest was just that, an ongoing tension, in which neither alone can fully explain the worldview of the Founders.”).

      [46].    Of course, the Holmesian bad man wouldn’t appear on the legal-theory stage for more than a century. See Oliver Wendell Holmes, The Path of the Law, 10 Harv. L. Rev. 457, 459 (1897) (“If you want to know the law and nothing else, you must look at it as a bad man, who cares only for the material consequences which such knowledge enables him to predict, not as a good one, who finds his reasons for conduct, whether inside the law or outside of it, in the vaguer sanctions of conscience.”).

      [47].    The Federalist No. 51 (Feb. 8, 1788), 18th_century/fed51.asp (James Madison).

      [48].    See James MacGregor Burns, The Deadlock of Democracy 6 (1963).

      [49].    The Federalist No. 51, supra note 47.

      [50].    Id.

      [51].    Id.

      [52].    As of 1998, Federalist No. 51 ranked fourth among the Federalist Papers by Supreme Court citation count. See Ira C. Lupu, The Most-Cited Federalist Papers, 15 Const. Comment. 403 (1998).

      [53].    Daryl J. Levinson & Richard H. Pildes, Separation of Parties, Not Powers, 119 Harv. L. Rev. 2312, 2313 (2006).

      [54].    Daryl J. Levinson, Parchment and Politics: The Positive Puzzle of Constitutional Commitment, 124 Harv. L. Rev. 657, 670 (2011).

      [55].    The Federalist No. 48 (Feb. 1, 1788) (James Madison),

      [56].    The Federalist No. 51, supra note 47.

      [57].    Id. To be sure, Madison’s model of the vertical separation of powers suffered from the same flaws as his horizontal one: he did not give a convincing account of the motives for states to check the federal executive or the means through which they would do so. See Marci A. Hamilton, The Elusive Safeguards of Federalism, 574 Ann. Am. Acad. of Pol. & Soc. Sci. 93, 99 (2001) (observing that “political parties are national organizations that demand allegiance to their national agenda’s stars,” and that elected officials at the state and local level “are driven to subservience to the national party”); Edward L. Rubin & Malcolm Feeley, Federalism: Some Notes on a Neurosis, 41 UCLA L. Rev. 903, 928–29 (1994) (noting that “[o]ur federal armed forces have the capacity to turn any state, any segment of humanity, or humanity in general into a thin gas . . . in about the space of half an hour” (internal quotation marks omitted)).

      [58].    Marty Cohen, David Karol, Hans Noel & John Zaller, The Party Decides: Presidential Nominations Before and After Reform (2008).

      [59].    See Chiafalo v. Washington, 140 S. Ct. 2316, 2320, 2326 (2020) (noting that “[e]lectors have only rarely exercised discretion in casting their ballots for President,” and holding that states may penalize electors who vote for someone other than the winner of the state’s popular vote).

      [60].    Concededly, applying the term “party” to the turn-of-the-nineteenth century Federalists and Republicans is somewhat anachronistic. Historian James Roger Sharp describes the Federalists and Republicans as “proto-parties,” adding that “it was not until James K. Polk’s victory [in the election of 1844] that it could be said that a balanced, resilient, and mature party system had finally came of age.” James Roger Sharp, American Politics in the Early Republic: The New Nation in Crisis 8, 287 (1993).

      [61].    See David W. Houpt, John Adams and the Elections of 1796 and 1800, in A Companion to John Adams and John Quincy Adams 142, 152–53 (David Waldstreicher ed., 2013).

      [62].    See William G. Morgan, The Origin and Development of the Congressional Nominating Caucus, 113 Proc. Am. Phil. Soc’y 184, 186 (1969).

      [63].    See Caitlin E. Jewitt, The Primary Rules: Parties, Voters, and Presidential Nominations 23 (2019).

      [64].    See Richard H. Pildes, The Historical Development of the U.S. Presidential Nomination Process, in The Best Candidate: Presidential Nomination in Polarized Times 36, 39–40 & n.10 (Eugene D. Mazo & Michael R. Dimino eds., 2020).

      [65].    See Richard S. Katz & Robin Kolodny, Party Organization as an Empty Vessel: Parties in American Politics, in How Parties Organize: Change and Adaptation in Party Organizations in Western Democracies 23, 36 (Richard S. Katz & Peter Mair eds., 1994).

      [66].    See Pildes, supra note 64, at 41.

      [67].    See Cohen et al., supra note 58, at 88. The one notable exception is Andrew Jackson, who ascended to the presidency while the party system was still in formation. See Eric A. Posner, The Demagogue’s Playbook: The Battle for American Democracy from the Founders to Trump 86 (2020).

      [68].    See Harold F. Bass, Jr., Presidential Party Leadership and Party Reform: Franklin D. Roosevelt and the Abrogation of the Two-Thirds Rule, 18 Presidential Stud. Q. 303 (1988).

      [69].    On presidential narcissism, see generally Ashley L. Watts et al., The Double-Edged Sword of Grandiose Narcissism: Implications for Successful and Unsuccessful Leadership Among U.S. Presidents, 24 Psychol. Sci. 2379 (2013).

      [70].    See Posner, supra note 67, at 152 (“The candidates whom the parties made available for public election had to survive a careful screening process through which no demagogue could pass.”).

      [71].    Senator George McGovern of South Dakota chaired the commission until he stepped down to run for President in 1972, after which Representative Donald Fraser of Minnesota led the panel.

      [72].    See Byron E. Shafer, Quiet Revolution: The Struggle for the Democratic Party and the Shaping of Post-Reform Politics (1983).

      [73].    See Democratic Nat’l Comm., Mandate for Reform, A Report of the Commission on Party Structure and Delegate Selection (1970), reprinted in 117 Cong. Rec. 32908 (Sept. 22, 1971).

      [74].    Pildes, supra note 64, at 49.

      [75].    James W. Ceaser, Presidential Selection: Theory and Development 263 (1979). Consistent with Ceaser’s suggestion, political scientist Scott Meinke and coauthors find that states were more likely to adopt Democratic presidential primaries in the wake of the McGovern-Fraser report where the ideological distance between party elites and a state’s citizens was narrower. Scott R. Meinke, Jeffrey K. Staton & Steven T. Wuhs, State Delegate Selection Rules for Presidential Nominations, 1972–2000, 68 J. Pol. 180 (2006).

      [76].    Elaine C. Kamarck, Delegate Allocation Rules in Presidential Nomination Systems: A Comparison between the Democrats and the Republicans, 4 J. L. & Pol. 275, 276 (1987).

      [77].    Arthur T. Hadley, The Invisible Primary (1976).

      [78].    Marty Cohen, David Karol, Hans Noel & John Zaller, Party Versus Faction in the Reformed Presidential Nominating System, 49 PS: Pol. Sci. 701, 702–03 (2016).

      [79].    On Carter’s outsider status, see Austin Ranney, Farewell to Reform—Almost, 24 Society 30, 30 (1987). Carter won the Democratic nomination again in 1980, but by then he was—as President—the ultimate insider.

      [80].    See Cohen et al., supra note 58, at 175–78.

      [81].    The claim that “the party decides” immediately raises the question of who constitutes “the party.” Cohen et al.’s conception of the party includes not only elected officeholders and party officials, but also leaders of interest groups that are allied with and influential in one of the two parties (e.g., organized labor in the Democratic Party and the religious right on the Republican side). See id. at 6.

      [82].    See id. at 703–04.

      [83].    See Adam Hilton, Sanders Is Out. Does That Mean That ‘The Party Decides’ After All?, Wash. Post: Monkey Cage (Apr. 10, 2020, 7:45 AM), sanders-is-out-does-that-mean-that-party-decides-after-all.

      [84].    See Michael Scherer, Josh Dawsey & Maeve Reston, Trump Works State-by-State To Improve Chances at Republican Convention, Wash. Post (Feb. 24, 2023, 4:17 PM),

      [85].    See The Federalist No. 72, supra note 40.

      [86].    See Bruce G. Peabody & Scott E. Gant, The Twice and Future President: Constitutional Interstices and the Twenty-Second Amendment, 83 Minn. L. Rev. 565, 85 (1999) (concluding that, “despite statements by scholars to the contrary, the custom of a two-term limit on presidential service appears to have been upheld somewhat contingently” before FDR).

      [87].    See Paul G. Willis & George L. Willis, The Politics of the Twenty-Second Amendment, 5 W. Pol. Q. 469, 469 (1952).

      [88].    See Charles W. Stein, The Third-Term Tradition: Its Rise and Collapse in American Politics 77–81, 237–49 (1943) (citing the examples of Ulysses S. Grant and Woodrow Wilson).

      [89].    See U.S. Const. amend. XXII, § 1 (“No person shall be elected to the office of the President more than twice, and no person who has held the office of President, or acted as President, for more than two years of a term to which some other person was elected President shall be elected to the office of the President more than once.”).

      [90].    Arguably, the language of the Twenty-second Amendment still allows a two-term President to be elected Vice President and then to ascend to the highest office upon the sitting President’s removal, death, or resignation. See Peabody & Grant, supra note 86, at 568.

      [91].    U.S. Const art. XX, § 1.

      [92].    Elections customarily occur on a Thursday, with Parliament reconvening the following Wednesday. See Gus O’Donnell, Letter to the Chairman of the Committee from the Cabinet Secretary, 23 February 2010, UK Parliament, pa/cm200910/cmselect/cmjust/396/396we02.htm (last accessed May 16, 2023).

      [93].    See Barbara Tasch, Polls Show Emmanuel Macron Is Now Leading the First Round of the French election—Here’s How France Chooses Its President, Bus. Insider (Mar. 9, 2017, 3:29 AM),

      [94].    See CNN Editorial Research, Justin Trudeau Fast Facts, CNN, (Dec. 9, 2022, 4:27 PM).

      [95].    See Neil Munshi, Ex-Ivory Coast Leader Laurent Gbagbo Acquitted of War Crimes, Fin. Times (Jan. 15, 2019),; Dionne Searcey, Why Democracy Prevailed in Gambia, N.Y. Times (Jan. 30, 2017),

      [96].    See generally Alberto Alesina & Stephen E. Spear, An Overlapping Generations Model of Electoral Competition, 37 J. Pub. Econ. 359 (1988) (explaining how political parties may solve the last-period problem with respect to outgoing presidents).

      [97].    See Edward G. Lengel, General George Washington: A Military Life 360 (2007); Max J. Skidmore, After the White House: Former Presidents as Private Citizens 4, 12, 84 (2004).

      [98].    See Joseph Wheelan, Mr. Adams’s Last Crusade: The Extraordinary Post-Presidential Life of John Quincy Adams 68 (2008); Skidmore, supra note 97, at 76–77.

      [99].    Skidmore, supra note 97, at 106–08.

    [100].    See id. at 5, 26, 32 (noting examples of Thomas Jefferson, James Madison, James Monroe, John Tyler, James Buchanan, Rutherford Hayes, Grover Cleveland, Benjamin Harrison, Taft, and Jimmy Carter). See Skidmore, supra note 97, at 5; Eli Watkins, Jimmy Carter Granted Tenure at Emory University After 37 Years of Teaching, CNN (Jun. 3, 2019, 4:58 PM),

    [101].    See Thomas F. Schaller & Thomas W. Williams, ‘The Contemporary Presidency’: Postpresidential Influence in the Postmodern Era, 33 Pres. Stud. Q. 188, 193–94 (2003). In addition, Grant briefly sought the Republican Party nomination in 1880, four years after leaving office. See id. at 194.

    [102].    See Paul Campos, The Truman Show: The Fraudulent Origins of the Former Presidents Act, 2022 Mich. St. L. Rev. 1, 35 (2022); CPI Inflation Calculator, U.S. Bureau of Labor Statistics, (last visited Jan. 2, 2023).

    [103].    See Campos, supra note 102, at 7 (describing Truman’s lobbying efforts); see also 3 U.S.C. § 102; Salary Table No. 2023-EX: Rates of Basic Pay for the Executive Schedule (EX), Off. of Personnel Mgmt., (last visited Jan. 2, 2023).

    [104].    See Elisabeth Bumiller, Ronald Reagan, Toast of Tokyo But Controversy Mars His Symbolic Trip, Wash. Post (Oct. 28, 1989), archive/lifestyle/1989/10/28/ronald-reagan-toast-of-tokyo-but-controversy-mars-his-symbolic-trip/11cdc265-ee18-4682-8e1f-5de188dc3ee8.

    [105].    See Philip Rucker, Tom Hamburger & Alexander Becker, How the Clintons Went from ‘Dead Broke’ to Rich: Bill Earned $104.9 Million for Speeches, Wash. Post (Jun. 26, 2014),

    [106].    See Elizabeth A. Harris, Obama’s Memoir ‘A Promised Land’ Coming in November, N.Y. Times (Sept. 17, 2020), 09/17/books/obama-memoir-a-promised-land.html; Barack Obama Memoir: How Celebrity Book Deals Measure Up, Times of India (Nov. 19, 2020), Other post-Presidents have served on corporate boards and in corporate advisory roles. See Assoc. Press, Gerald Ford on the Board, N.Y. Times (Jun. 9, 1983),; Leslie Wayne, Elder Bush in Big G.O.P. Cast Toiling for Top Equity Firm, N.Y. Times (Mar. 5, 2001),

    [107].    See, e.g., Richard Cohen, Opinion, The Selling of the Post-Presidency, Wash. Post (Jun. 1, 2015),; Jerome Karabel, Op-Ed, $400,000 for One Speech? For Ex-Presidents, It Is Now the Norm, N.Y. Times (Apr. 27, 2017),

    [108].    See Adam Taylor, The Sad Story of Africa’s Most Prestigious Prize, Wash. Post (Mar. 4, 2015), The analogy is not exact, as Ibrahim Prize recipients also must have “demonstrated exceptional leadership,” a criterion that U.S. ex-Presidents do not need to satisfy in order to cash in. See Ibrahim Prize for Achievement in African Leadership, Mo Ibrahim Found., (last visited Jan. 3, 2023).

    [109].    See Ben Steverman, U.S. Billionaires Got $1 Trillion Richer During Trump’s Term, Bloomberg (Oct. 30, 2020),

    [110].    See Maggie Haberman, Alexander Burns & Jonathan Martin, As Election Day Arrives, Trump Shifts Between Combativeness and Grievance, N.Y. Times (Nov. 2, 2020),

    [111].    See Ben Protess, Jonah E. Bromwich, William K. Rashbaum, Kate Christobek, Nate Schweber & Sean Piccoli, Trump Is Indicted, Becoming First Ex-President to Face Criminal Charges, N.Y. Times (Mar. 30, 2023), 30/nyregion/trump-indictment-hush-money-charges.html.

    [112].    To be sure, this incentives-based theory of Trump’s election denialism could be supplemented by a psychological account, which also may carry explanatory force. See, e.g., Anna Medaris, What It Will Take for Trump to Concede, According to a Psychologist, Insider (Dec. 28, 2020, 1:12 PM),

    [113].    See Maren Christensen Bjune & Stina Petersen, Guarding Privileges and Saving the Day: Guatemalan Elites and the Settlement of the Serranazo, in Presidential Breakdowns in Latin America: Causes and Outcomes of Executive Instability in Developing Democracies 165, 168 (Mariana Llanos & Leiv Marsteintredet eds., 2010).

    [114].    See Marcelo Rochabrun, Peru Is Running Out of Space to Keep its Jailed Ex-Presidents, Bloomberg (Feb. 24, 2023, 9:00 AM),

    [115].    The Federalist No. 72, supra note 40.

    [116].    See Steven Nelson, Conservatives Call for Charges Against Biden After Trump Indictment, N.Y. Post (Mar. 31, 2023, 4:17 PM),

    [117].    Id. at 2313.

    [118].    See id. at 2344.

    [119].    In 1992, political scientist Morris Fiorina declared that we now live in an “era of divided government.” See generally, Morris P. Fiorina, An Era of Divided Government, 107 Pol. Sci. Q. 387 (1992). Since then, the White House and both chambers of Congress have been under unified party control roughly 40 percent of the time. See Party Government Since 1857, Off. of the Historian, U.S. House of Representatives, (last visited Mar. 29, 2023).

    [120].    Juan J. Linz, The Perils of Presidentialism, 1 J. Dem. 51, 53 (1990).

    [121].    For decades, scholars have debated whether and when the American party system switched from conflictual to consensual, with E. E. Schattschneider and Walter Dean Burnham famously attributing the change to the aftermath of William Jennings Bryan’s defeat in 1896. For a critical review of the literature on the “system of 1896” thesis, see David R. Mayhew, Electoral Realignments, 3 Ann. Rev. Pol. Sci. 449, 455–56, 469–70 (2000).

    [122].    See Matthew Levendusky, The Partisan Sort: How Liberals Became Democrats and Conservatives Became Republicans (2009). For a brief summary of the literature on partisan sort, see Brian D. Feinstein & Daniel J. Hemel, Partisan Balance with Bite, 118 Colum. L. Rev. 9, 26–29 (2018).

    [123].    See, e.g., Edward G. Carmines et al., Who Fits the Left-Right Divide? Partisan Polarization in the American Electorate, 56 Am. Behav. Sci. 1631 (2012).

    [124].    Jessica Bulman-Pozen, Partisan Federalism, 127 Harv. L. Rev. 1077, 1080 (2014).

    [125].    See id.

    [126].    See Levinson & Pildes, supra note 53, at 2319.

    [127].    Arguably, the fact that January 6 even happened is a strike in favor of Linz’s theory. Linz thought that presidentialist regimes with ideologically cohesive political parties would be particularly prone to political violence. In the United States in the early 21st century, we finally got ideologically cohesive political parties—and then, in short order, we got political violence.

    [128].    See Karen Yourish, Larry Buchanan & Denise Lu, The 147 Republicans Who Voted to Overturn Election Results, N.Y. Times (Jan. 7, 2021), interactive/2021/01/07/us/elections/electoral-college-biden-objectors.html.

    [129].    See Francine Kiefer, A Man or a Movement? Answer on Trump Baffles GOP, Christian Sci. Monitor (May 14, 2016), Politics/2016/0514/A-man-or-a-movement-Answer-on-Trump-baffles-GOP.

    [130].    See Yourish et al., supra note 128.

    [131].    See Levinson & Pildes, supra note 53, at 2312.

    [132].    As of February 2023, the Cook Political Report rated 366 out of 435 House seats (84 percent of the total) as either “solid Democrat” or “solid Republican.” See 2024 House Race Ratings, The Cook Political Report with Amy Walter (Feb. 28, 2023),

    [133].    See Russell Wheeler, Trump’s Judicial Campaign to Upend the 2020 Election: A Failure, But Not a Wipe-Out, Brookings Inst.: FixGov (Nov. 30, 2021),

    [134].    See Pearson v. Kemp, No. 1:20-cv-4809-TCB, 2020 U.S. Dist. LEXIS 226348 (N.D. Ga. Nov. 29, 2020), appeal dismissed, 831 Fed. Appx. 467 (11th Cir. 2020).

    [135].    See Adam Liptak, Chief Justice Defends Judicial Independence After Trump Attacks ‘Obama Judge,’ N.Y. Times (Nov. 21, 2018), (quoting Chief Justice Roberts’ statement that “[w]e do not have Obama judges or Trump judges, Bush judges or Clinton judges”).

    [136].    See Jonathan J. Cooper, Arizona Governor Silences Trump’s Call, Certifies Election, Assoc. Press (Dec. 2, 2020),

    [137].    See Amy Gardner, Coby Itkowitz & Josh Dawsey, Trump Calls Georgia Governor To Pressure Him for Help Overturning Biden’s Win in the State, Wash. Post (Dec. 5, 2020), 12/05/fd8d677c-3721-11eb-8d38-6aea1adb3839_story.html.

    [138].    See Max Greenwood, Kemp Recertifies Biden Win in George After Recount, The Hill (Dec. 8, 2020, 8:17 AM),

    [139].    Michael D. Shear & Stephanie Saul, Trump, in Taped Call, Pressured Georgia Official to ‘Find’ Votes to Overturn Election, N.Y. Times (May 26, 2021),

    [140].    Cf. Bulman-Pozen, supra note 124, at 1080 (“States oppose federal policy because they are governed by individuals who affiliate with a different political party than do those in charge at the national level, not because they are states as such.”).

    [141].    See The Federalist No. 51, supra note 47.

    [142].    Rachel E. Barkow, Institutional Design and the Policing of Prosecutors: Lessons from Administrative Law, 61 Stan. L. Rev. 869 (2009); Neal Kumar Katyal, Internal Separation of Powers: Checking Today’s Most Dangerous Branch from Within, 115 Yale L J. 2314 (2006); Gillian E. Metzger, The Interdependent Relationship Between the Internal and External Separation of Powers, 59 Emory L.J. 423 (2009).

    [143].    See Jonathan D. Karl, Inside William Barr’s Breakup with Trump, The Atlantic (Jun. 27, 2021, 6:00 AM), william-barrs-trump-administration-attorney-general/619298; Press Release, Joint Statement from Elections Infrastructure Government Coordinating Council & the Election Infrastructure Sector Coordinating Executive Committees, Cybersecurity & Infrastructure Sec. Agency (Nov. 12, 2020),

    [144].    See Michael Kranish, New Details Emerge of Oval Office Confrontation Three Days Before Jan. 6, Wash. Post (Jun. 14, 2022), 2022/06/14/inside-explosive-oval-office-confrontation-three-days-before-jan-6.

    [145].    See Susan B. Glaser & Peter Baker, Inside the War Between Trump and His Generals, New Yorker (Aug. 8, 2022), 2022/08/15/inside-the-war-between-trump-and-his-generals.

    [146].    Andrew Jackson probably never uttered those words, though he expressed similar sentiments. See Alfred A. Cave, Abuse of Power: Andrew Jackson and the Indian Removal Act of 1830, 65 Historian 1330, 1349–50 n.70 (2003).

    [147].    See Elaine C. Kamarck, Returning Peer Review to the American Presidential Nomination Process, 93 N.Y.U. L. Rev. 709, 723 (2018).

    [148].    See id.

    [149].    See id. at 724–25; Raymond J. La Raja & Jonathan Rauch, Voters Need Help: How Party Insiders Can Make Presidential Primaries Safer, Fairer, and More Democratic, Brookings Inst. (Jan. 31, 2020),; Thomas E. Mann & Norman Ornstein, Delegates of Steel: Why Superdelegates Should Be Welcomed, Not Feared, Brookings Inst. (Feb. 15, 2008),; Frances McCall Rosenbluth & Ian Shapiro, Bring Back the Superdelegates, Wash. Post: PostEverything (Aug. 1, 2019, 7:59 AM),; see also Bruce E. Cain & Cody Gray, Pluralist Parties by Design: Party Reform in a Polarized Era, 93 N.Y.U. L. Rev. 621, 639 (2018) (considering options for superdelegate reform).

    [150].    As a formality, Democratic superdelegates can vote on the first ballot if one candidate already has a convention majority based on pledged delegates. As it happened, Joe Biden did command a majority based on pledged delegates, thus allowing superdelegates to vote on the first ballot. See Caitlin E. Jewitt, Restoring Trust and Reducing Perceived Influence: Superdelegates and the 2020 Democratic Nomination, 57 Society 680, 684 (2020).

    [151].    As Hans Noel has proposed, parties also could increase the chances of reaching a second ballot by changing their allocation rules for pledged delegates. As Noel notes, some Republican primaries remain winner-take-all, while Democrats award pledged delegates proportionally to candidates who get at least 15 percent of the vote. Noel suggests a lower threshold (e.g., 5 percent), which would make it harder to cobble together a first-ballot majority and would thus favor candidates who can build coalitions at the convention. See Hans Noel, Make ‘Contested Conventions’ a Routine Event, in How to Un-Break the Primaries, Wash. Post (Feb. 28, 2020), outlook/2020/02/28/how-un-break-primaries.

    [152].    See Kamarck, supra note 147, at 726.

    [153].    See Walter Shapiro, Rationalizing the Presidential Nomination Process, Brennan Ctr. for Justice (Oct. 28, 2021),

    [154].    11 C.F.R. § 110.13(c); see Bob Bauer, A Debatable Role in the Process: Political Parties and the Candidate Debates in the Presidential Nominating Process, 93 N.Y. U. L. Rev. 589, 597–98 (2018).

    [155].    See Bauer, supra note 154, at 598–99.

    [156].    See David Siders & Stephanie Murray, ‘Get on the Team or Shut Up’: How Trump Created an Army of GOP Enforcers, Politico (Jul. 13, 2021),; Nicholas Riccardi & Joey Cappelletti, Failing at Polls, Election Deniers Focus on State GOP Posts, Assoc. Press (Feb. 26, 2023, 4:30 AM),

    [157].    One of the Republican Governors who pushed back against Trump’s efforts to overturn the 2016 election results, Governor Doug Ducey of Arizona, did not endorse any candidate in the 2016 Republican presidential primary. See Mike Sunnucks, Top Republicans Doug Ducey, Jeff Flake, John McCain Not Endorsing Before Trump-Focused Arizona Primary, Phoenix Bus. J. (Mar. 18, 2016), news/2016/ 03/18/top-republicans-doug-ducey-jeff-flake-john-mccain.html. Governor Brian Kemp was Georgia’s Secretary of State in 2016; he does not appear to have backed a candidate in that year’s primary either.

    [158].    See Richard H. Pildes, Democracies in the Age of Fragmentation, 110 Cal. L. Rev. 2051, 2052–53 (2022).

    [159].    See The Federalist No. 72, supra note 40.

    [160].    Sanford Levinson, Presidential Elections and Constitutional Stupidities, 12 Const. Comment. 183 (1995).

    [161].    U.S. Const. amend. XX, § 1.

    [162].    See Levinson, supra note 160, at 184.

    [163].    See Jack M. Beermann & William P. Marshall, The Constitutional Law of Presidential Transitions, 84 N.C. L. Rev. 1253, 1281 (2006); Nancy Armoury Combs, Carter, Reagan, and Khomeini: Presidential Transitions and International Law, 52 Hastings L.J. 303, 305 (2001).

    [164].    One reason why some other countries can manage a faster transition is that their main opposition parties typically form a “shadow cabinet” that is ready to take office almost immediately after an election. See David Fontana, The Permanent and Presidential Transition Models of Political Party Policy Leadership, 103 Nw. L. Rev. Colloquy 393, 394–98 (2009).

    [165].    To be sure, the optics of a multimillion-dollar post-presidential golden parachute would be problematic if the payout were funded with taxpayer money. And if a private philanthropist were to put up the funds, then the arrangement could raise legitimate concerns about influence buying.

    [166].    Cf. Lance deHaven-Smith & Matthew T. Witt, Preventing State Crimes Against Democracy, 41 Admin. & Soc’y 527, 527 (2009) (defining “state crimes against democracy,” or SCADs, as “actions or inactions by government insiders intended to manipulate democratic processes and undermine popular sovereignty”).

    [167].    An alternative approach might be to create a mechanism whereby sitting Presidents remain susceptible to criminal prosecution. Fears about, for example, corruption charges might do less to motivate a sitting President to pursue self-coup if retaining office did not confer immunity either. On the pro-democracy role of public prosecutors who are empowered to bring charges against chief executives, see Huq & Ginsburg, supra note 18, at 172–73. A norm of prosecuting current Presidents, though, encounters serious concerns about stability. A President who is considering an extraconstitutional self-coup presumably will consider ousting the public prosecutor too.

    [168].    See Jonah E. Bromwich, Ben Protess, William K. Rashbaum & Matthew Haag, Hyperbole or Fraud? The Question at the Heart of Trump Investigation, N.Y. Times (Jan. 19, 2022),

    [169].    For thoughts on a similar tradeoff in international human rights law, see Michael P. Scharf, The Amnesty Exception to the Jurisdiction of the International Criminal Court, 32 Cornell Int’l L.J. 507 (1999).

    [170].    U.S. Const. art. II, § 2, cl. 2.

    [171].    5 U.S.C. § 3345(a).

    [172].    See Exec. Order 13963, 85 C.F.R. 81,331 (2020).

    [173].    See Exec. Order 13787, 82 C.F.R. 16,723 (2017).

    [174].    5 U.S.C. § 3345(a)(1).

    [175].    Id. § 3345(a)(3).

    [176].    Id. § 3346(a)(1).

    [177].    Id. §§ 3346(a)(2), (b), 3349(a).

    [178].    See Missy Ryan, Dan Lamothe, Paul Sonne & Josh Dawsey,
Trump Fires Defense Secretary Mark Esper, Wash. Post (Nov. 9, 2020, 11:30 PM),

    [179].    See Amanda Carpenter, Trump’s Jan. 6th National Guard Lie Crumbles, The Bulwark (Jul. 27, 2022, 5:30 AM),

    [180].    See Zachary Cohen, Trump’s Defense Head Called Attaché in Rome to Investigate Baseless Election Claim About Italian Satellites, CNN
(Jun. 23, 2022, 5:25 PM),

    [181].    See Katie Benner, William Barr Is Out as Attorney General, N.Y. Times
(Dec. 14, 2020),

    [182].    See Katie Benner, Trump and Justice Dept. Lawyer Said to Have Plotted to Oust Acting Attorney General, N.Y. Times (Jan. 22, 2021), us/politics/jeffrey-clark-trump-justice-department-election.html; Michael Kranish, New Details Emerge of Oval Office Confrontation Three Days Before Jan. 6, Wash. Post (Jun. 14, 2022, 6:00 AM),

    [183].    See Ben Miller-Gootnick, Note, Boundaries of the Federal Vacancies Reform Act, 56 Harv. J. on Leg. 459, 475–81 (2019).

    [184].    2 U.S.C. § 3345(a).

    [185].    See Guidance on Application of Federal Vacancies Reform Act of 1998, 23 Op. O.L.C. 60, 61 (1999); Designating an Acting Director of the Bureau of Consumer Financial Protection, 41 Op. O.L.C. , at *4 (2017); Designating an Acting Attorney General, 42 Op. O.L.C., at *4 n.1 (2018).

    [186].    See Anne Joseph O’Connell, Actings, 120 Colum. L. Rev. 613, 709 (2020).

    [187].    Id. at 675.

    [188].    Id. at 710.

    [189].    See İpek Çınar, Riding the Democracy Train: Incumbent-Led Paths to Autocracy, 32 Const’l Pol. Econ. 301, 309 (2021) (noting that “the secondary literature on self-coups reveals that the level of military backing an incumbent is able to secure for his autogolpe is crucial for its subsequent success”).

    [190].    See id. In a similar vein, lack of support from the military appears to be one factor that dissuaded Brazilian President Jair Bolsonaro from attempting a self-coup after he lost his reelection bid in late 2022. See Jack Nicas, Refusing to Accept Defeat, Bolsonaro Backers Call on Military to Intervene, N.Y. Times (Nov. 2, 2022),

    [191].    See Tayyab Mahmud, Jurisprudence of Successful Treason: Coup D’Etat & Common Law, 27 Cornell Int’l L.J. 49, 138–39 (1994) (“Unfortunately, when confronted with a successful coup, most courts have opted for the worst choice, namely, validation and legitimation of extra-constitutional usurpation.”).

    [192].    See, e.g., Maxwell A. Cameron, Latin American Autogolpes: Dangerous Undertows in the Third Wave of Democratisation, 19 Third World Q. 219, 226 (1998); Tim Golden, Guatemala’s Counter-Coup: A Military About-Face, N.Y. Times (Jun. 3, 1993),

    [193].    See Kyle Cheney, Josh Gerstein & Nicholas Wu, Trump Lawyers Saw Justice Thomas as ‘Only Chance’ to Stop 2020 Election Certification, Politico (Nov. 2, 2022, 3:09 PM),

    [194].    See, e.g., Kai M. Thaler, Nicaragua: A Return to Caudillismo, 28 J. Democracy 156, 159 (2017).

    [195].    See Lee Epstein & Eric Posner, Supreme Court Justices’ Loyalty to the President, 45 J. Legal Stud. 401 (2016).

    [196].    See Presidential Commission on the Supreme Court of the United States, Final Report, The White House 111–145 (Dec. 2021),

    [197].    See Daniel Hemel, Can Structural Changes Fix the Supreme Court?, 35 J. Econ. Persp. 119, 136–38 (2021); Submission by Vicki C. Jackson to Presidential Commission on the Supreme Court of the United States, The White House 2, 16–17 (Jul. 20, 2021),

    [198].    One exception would be the Grover Cleveland–type case in which a President serves a term, leaves office, and then serves a second term.