PRELIMINARIES

The rough thought we wish to make precise is that public officials promote democratic stability when they practice a certain kind of forbearance, refraining from certain actions even though, by virtue of their office, they have a formally codified right to take them. I will refer to this hypothesis as the forbearance conjecture. To formulate it with more analytical precision, the first step is to clarify the nature of the outcome—stable democratic institutions—that, per the conjecture, forbearance helps explain.

Roughly speaking, institutions are just sets of rules to which participants conform because of a common expectation that other participants will also conform to them (Greif and Kingston Reference Greif, Kingston, Norman and Caballero2011; Guala Reference Guala2016; Knight Reference Knight1992). The stability of an institution consists in the stability of the equilibrium in which the institution’s participants comply with its constitutive rules; democratic stability, in particular, refers here to a stable practice of complying with the rules and norms that constitute liberal democracy. These include the rules and norms that define free and fair elections as well as norms of respecting basic political liberties and the rule of law. By democratic backsliding, I mean an incremental, cumulative process of piecemeal deviations from and violations of democracy’s constitutive rules and norms (cf. Ginsburg and Huq Reference Ginsburg and Huq2018, 87, 96; Haggard and Kaufman Reference Haggard and Kaufman2021, 1; Przeworski Reference Przeworski2019, 172; Waldner and Lust Reference Waldner and Lust2018, 95).Footnote ³

To formulate the forbearance conjecture with more precision, the next step is to characterize the specific class of actions from which public officials forbear. According to Levitsky and Ziblatt’s conception of institutional forbearance, the specific class is the set of actions that respect the letter of constitutional law but violate its spirit (Levitsky and Ziblatt Reference Levitsky and Ziblatt2018, 106); according to the one presented here, it is the set of actions that comply with democracy’s constitutive rules but whose compatibility with those rules is not approximately common knowledge. Whether there is any theoretical reason to think the forbearance conjecture is true, and whether the conjectured mechanism behind forbearance has any explanatory power, depend on how one specifies the class of actions from which public officials are supposed to forbear. Some specifications turn the forbearance conjecture into a nontrivial but ad hoc and theoretically groundless generalization, while others render it trivially true and thus uninteresting.

Here is an illustrative example. As Przeworski (Reference Przeworski2019, 51) observes, poor institutional design partly explains some prominent cases of democratic failure, the most famous example being the Weimar constitution’s Article 48 through which the President could empower the chancellor to rule by decree during vaguely defined emergencies and which thereby allowed for a “constitutional slide into authoritarianism.” One might hypothesize that the Weimar case is just the clearest, most extreme example of a more general process of constitutional backsliding, facilitated by public officials whose actions are analogous to, if less dramatic and destructive than, Hindenburg’s exercise of his Article 48 powers. One might formulate the following definition to articulate this conjecture: to practice institutional forbearance is to refrain from exercising rights and prerogatives (like Hindenburg’s exercise of his Article 48 authority) the exercise of which would end democracy.

But this definition has an obvious shortcoming, which illustrates a criterion any satisfactory definition must satisfy. It deprives the forbearance conjecture of explanatory power. That a regime will not stably persist as a democracy if a public official exercises a right the exercise of which would end democracy is trivially true. Rather than a conjecture capable of being true or false in virtue of facts about the world, it becomes an empty tautology: a democracy will not persist if it stops being a democracy. The first criterion for our definition is that the forbearance conjecture retain explanatory power.

Another possible shortcoming of this definition is that it might unduly narrow the scope of the forbearance conjecture, which would become a hypothesis about a specific class of democratic regimes—namely those in which the constitution grants certain actors legal means of abolishing democracy. Generalizing from this observation, one might formulate a second criterion for a definition of “forbearance”: the mechanism of democratic backsliding prevented by forbearance should have broad scope, operating across a broad range of democracies, including democracies without any defects of institutional design.

Levitsky and Ziblatt’s conception of forbearance—“avoiding actions that, while respecting the letter of the law, obviously violate its spirit”—satisfies the scope criterion: their thought is that any democracy, however well-designed its institutions might be, creates opportunities for some public officials to take actions that respect the letter of the constitution even as they violate its spirit. Their distinction between the “spirit” and “letter” of constitutional law introduces other difficulties, however. In addition to the criteria concerning scope and explanatory power, a third criterion for a satisfactory conception of institutional forbearance is what I will call theoretical grounding: there should be theoretical reasons to expect forbearance to protect against democratic backsliding. Yet, on a natural interpretation of the spirit/letter distinction, there are often actions that in theory should not pose any risk of democratic backsliding at all even though they plausibly run contrary to the spirit of a rule or law. For example, the U.S. Senate, as a self-governing body, has the authority to abolish the filibuster, and if its members voted unanimously to do so, all agreeing that the filibuster serves the country poorly, they would certainly be respecting the “letter” of the constitutional provisions that define Senators’ rights and obligations. But would they be respecting their “spirit”? Seeing as the Senate is a countermajoritarian institution by design—its purpose is to curb the power of simple popular majorities at the national level—one might plausibly argue that countermajoritarian practices like the filibuster are not just compatible with but indeed emblematic of its spirit. If that is right, then Levitsky and Ziblatt’s conception would seem to imply that the Senators, by voting unanimously to repeal the filibuster, would be failing to practice forbearance, even though there would be no reason to expect its repeal to undermine democratic stability in the imagined circumstances.Footnote ⁴

Instead of distinguishing between “letter” and “spirit,” one might think to make a distinction between formal and informal rules, and say that public officials practice institutional forbearance when they refrain from actions that, despite satisfying all formal institutional rules, violate liberal democracy’s informal norms. Levitsky and Ziblatt often describe forbearance in these terms. But this definition threatens to sap power from the resulting explanation of democratic backsliding. Democratic backsliding just is an incremental process of piecemeal deviations from and violations of liberal democracy’s constitutive rules, some of which are informal. Actions that violate its constitutive rules—whether these rules are formally codified or not—are part of the phenomenon to be explained. Take democratic institutions in Britain as an example. As Levitsky and Ziblatt note, the Crown has the “formal” authority to select anyone as prime minister, but there is an informal norm, not codified in any formal law, requiring the Crown to select a member of Parliament who is able to command a majority in the House of Commons. By respecting this informal norm, the Crown is therefore practicing forbearance as Levitsky and Ziblatt conceptualize it (107). But this informal norm is constitutive of British democracy; if the Crown began breaking the norm and other actors stopped enforcing it, one could no longer describe Britain as a democracy—not because breaking the norm causes democracy’s collapse, but rather because compliance with the rule is part of what constitutes British democracy. Saying that the Crown would cause democratic backsliding if it deviated from the norm is like saying that getting a divorce would cause one’s marriage to collapse. That the Crown chooses not to exercise its royal prerogative in any manner that runs afoul of the norm cannot explain the persistence of British democracy because it is part of the phenomenon in need of explanation.

One can draw from these observations a more general conclusion about the right way to conceptualize institutional forbearance. Because violations of democracy’s constitutive norms—formally codified or not—are part of the phenomenon in need of explanation, the forbearance conjecture has explanatory power only if the actions from which political actors must forbear, per the conjecture, are actions that are in and of themselves compatible with the constitutive rules and norms of democracy, formal and informal alike. Otherwise the claim that democratic stability requires officials to forbear from taking these actions would be true but rather trivial, like the claim that preserving a marriage requires one to refrain from filing for a divorce.Footnote ⁵ Thus we are in search of a class of actions that (to satisfy explanatory power) are in themselves compatible with the constitutive rules and norms of democracy—whether formally codified or not—but that (to satisfy theoretical grounding) nonetheless could be expected in theory to contribute to democratic backsliding.

Whether one can satisfy either criterion, much less jointly satisfy both, depends on how one characterizes democracy’s constitutive norms and rules. An especially important question is whether these include a kind of master meta-norm requiring public officials to follow existing rules and procedures—whether these are established by formally codified law or informal custom—governing the permissible exercises of their prerogatives. Such a norm forbids public officials from making special ad hoc exceptions for themselves or their party and from exercising their power in arbitrary fashion; it requires officials to exercise their power only in ways that are compatible with whatever rules—formally codified or not—apply. Call it the norm of following applicable rules and procedures. Treating it, like the rule of law, as constitutive of liberal democracy has consequences for what we classify as democratic backsliding. For example, if there was an established informal rule requiring the Senate to hold hearings promptly after a President submits a nominee for its approval, then the Senate’s refusal to hold hearings would be a violation of the more general norm of following applicable rules and procedures. If this meta-norm is itself one of liberal democracy’s constitutive norms, then the refusal to hold hearings would be a form of democratic backsliding. More generally, if a norm of following applicable rules and procedures is a constitutive element of a liberal democracy, then it would be true, but only trivially so, that democratic stability requires public officials to forbear from actions that may respect the letter of the law but violate the informal norms governing the permissible exercise of their rights and prerogatives. Violations of such informal norms are part of the phenomenon in need of explanation, not a potential explanatory factor. In what follows, I will assume that the norm of following applicable rules and procedures, like the rule of law, is partly constitutive of liberal democracy, such that a decline in respect for this norm is itself a form of democratic backsliding.

It is worth noting in passing that if forbearance were supposed to be not causally explanatory, but instead just a symptom or indicator of democratic stability, then Levitsky and Ziblatt’s definition would be perfectly acceptable. The thought would then be that violations of informal democratic norms are diagnostic of democratic decline: they constitute democratic backsliding—and thus cannot explain it—but a particular form of democratic backsliding that political scientists should take greater notice of, perhaps because it portends more flagrant examples of democratic rule-breaking in the future. Levitsky and Ziblatt’s definition is unsatisfactory for my purposes only because I wish to consider the conjecture that forbearance contributes causally to democratic stability. Given the concept’s intended theoretical role, its connection to democratic backsliding must be an open question, not settled as a matter of definition. This is how Levitsky and Ziblatt introduce the concept as well—the norm of forbearance is said to be “critical to democracy’s survival” (106)—so I think they should also reconsider their choice of definition.

On the alternative definition I give below, officials practice institutional forbearance when they refrain from actions of “uncertain legitimacy,” actions that are, strictly speaking, compatible with liberal democracy’s constitutive norms, but whose compatibility is not common knowledge, or even approximately common knowledge: some observers will doubt their legitimacy, or expect other observers to doubt their legitimacy, or expect others to form this expectation, and so on. This conception satisfies the criteria of explanatory power, theoretical grounding, and broad scope. While there are surely alternative conceptions that do the same, I believe it has a further virtue of capturing the intuition that the informality of certain rules, indeed the impossibility of formally codifying all democratic norms, is important to the explanation of why forbearance can play a role in stabilizing democracies. It is not the distinction between formal rules and informal norms per se, but rather the distinction between actions with commonly recognized legitimacy and actions of uncertain legitimacy that matters. But in practice there will be considerable overlap between the two distinctions if it is harder to generate consensus on an action’s legitimacy when this turns on its compliance with an informal norm rather than a formally codified rule.

Before turning to this view, I note one more alternative that might occur to some readers. Democracy’s survival might sometimes depend on individuals’ refraining from actions that, despite their commonly known compatibility with the constitutive rules and norms of democracy, would provoke other bad actors to opt out of the democratic game or violate its rules. One might think to define “forbearance” broadly as refraining from such acts of provocation. A conception of this kind would also narrow the scope of the forbearance conjecture, however. It would apply only in circumstances in which there are bad actors whose willingness to comply with democratic rules is contingent on other actors’ refraining from exercising their democratic rights in ways that ought to be allowed per the rules of the democratic game. The mechanism of democratic backsliding I describe in the next section can operate across a broader set of circumstances.

DEMOCRACY’S SELF-ENFORCEMENT MECHANISM

Why think that institutional forbearance, as I have proposed we understand it—officials’ refraining from actions that are compatible with liberal democracy’s constitutive norms but whose compatibility is not approximately common knowledge—contributes to democratic stability? The conjecture rests on the thought that the prospect of coordinated resistance to abuses of power is an important mechanism of democratic stability, and that failures to practice forbearance produce situations that sow seeds of doubt about who is prepared to enforce democracy’s rules, impairing the deterrence mechanism of coordinated resistance.

In a stable democracy, incumbent ruling parties choose to hold elections, and to relinquish power when they lose, even though there is no other agent who can use the coercive powers of the state to compel them (seeing as they already wield those powers themselves). What explains how compliance with the rules of democracy comes to be a “self-enforcing equilibrium” in this sense? Some elected officials may have principled commitments to democracy, but at least some, and perhaps most, will have only conditional and instrumental reasons for respecting its rules. Their willingness to relinquish office after losing a free and fair election will reflect a calculation of expected benefits and costs, not a principled commitment: refuse to step down, risking the dangers of the resulting social unrest, or pass control of the state over to the opposition in the expectation that they will someday pass it back (Przeworski Reference Przeworski2019).Footnote ⁶ Their calculations will come out in democracy’s favor only if they believe transgressions are sufficiently likely to trigger some kind of effective resistance. But effective resistance requires the opposition to mount a coordinated response. If the President has declared martial law and suspended elections, no one wants to be the lone dissident who takes to the streets to protest. Generally one prefers to join a collective effort to resist the incumbent official’s transgression only if enough other individuals are expected to do the same.

In what we might call the Lockean models of self-enforcing democracy, it is the prospect of mass rebellion that deters incumbent officials from transgressing against democracy’s fundamental rules (Weingast Reference Weingast1997; Fearon Reference Fearon2011). But one could also entertain more elite-centered versions of the theory, in which fear of coordinated resistance among more powerful political or social elites creates the deterrent, or class-centered versions, in which it is the fear of coordinated resistance from a particular social class. For the sake of concreteness, I will assume a theory like Weingast’s or Fearon’s, which focuses on the role of mass resistance, but one could entertain a more abstract version of the argument, or one tailored to these other versions of the theory, so long as one accepts the assumption that effective resistance involves coordination among some set of actors who have the ability to observe and resist abuses of power. This assumption implies that any mechanism that disrupts their successful coordination might undermine the democratic equilibrium. The seeds-of-doubt mechanism, described below, is one example, and this idea is the basis for the forbearance conjecture.

Maintaining a credible threat of coordinated resistance to abuses of power depends, first of all, on there being enough citizens who are at least conditionally willing to defend democracy, citizens who value participating in collective action to resist transgressions if they believe enough other people will participate to make the action successful. Democracy’s self-enforcing equilibrium will break down if there are too few of these “conditional defenders” and too many who are either democratic free-riders—they value democracy but do not value participating in its successful defense—or antidemocratic partisans, happy to tolerate or even support undemocratic behavior by their own party but ready to resist attempts by the other party to exercise power. Thus a democratic equilibrium may break down if economic and social change, by increasing partisan polarization and the costs of turning power over to the opposition, increases the numbers of antidemocratic partisans who are willing to support, or at least tolerate, an incumbent party’s violations of democratic norms (Graham and Svolik Reference Graham and Svolik2020; Kaufman and Haggard Reference Kaufman and Haggard2019; Luo and Przeworski Reference Luo and Przeworski2023; Simonovits, McCoy, and Littvay Reference Simonovits, McCoy and Littvay2022; Svolik Reference Svolik2020). That direct effect of polarization may suffice to undermine the democratic equilibrium, but polarization may have a second-order effect as well. Anticipating this possibility and fearing that not enough other citizens will participate in collective efforts to defend democracy, its remaining conditional defenders may choose not to participate themselves.Footnote ⁷ That is, even democracy’s conditional defenders may not participate in efforts to resist abuses of power if they fear that increasing polarization has produced a critical mass of antidemocratic partisans who will tolerate the incumbent’s behavior.

The coordination equilibrium may also break down as a consequence of a change in beliefs, without any change in preferences. If a sufficient number of citizens have the preferences of the conditional defender, then there is an equilibrium in which they all protest, but still one in which they do not. Common knowledge that they all prefer the outcome of coordination does not suffice for successful coordination. If a conditional defender comes to fear that other conditional defenders will stay home, they will also choose to stay home.Footnote ⁸

One reason for this fear may be uncertainty about how many of their compatriots are even aware the incumbent officeholder has violated democracy’s rules. Even well-formulated rules will have ambiguous requirements in certain contexts, and partisan observers, under the influence of ideologically motivated reasoning, may form different beliefs about what the incumbent has done and whether the action violates the rules (Ahlquist et al. Reference Ahlquist, Ichino, Wittenberg and Ziblatt2018).Footnote ⁹ Moreover, even if the rules were unambiguous, everyone perfectly rational, and all this common knowledge, observers would still sometimes form different beliefs about whether a public official has violated a rule. Whether an action counts as a violation may depend on facts that are hard to ascertain, no matter how clearly defined the rule may be. For example, whether a government prosecutor has violated the rule of law by charging a member of the opposition party with a crime depends on the circumstances of the case—is there evidence that the indicted politician did in fact commit the crime, would the evidence and circumstances of the case motivate the prosecutor to file charges even if the politician were not a member of the opposition, etc.? Outside observers cannot easily answer these questions. For all these reasons, even a citizen who recognizes that a public official has violated one of democracy’s fundamental norms may wonder how many other citizens have formed the same belief, or wonder whether other citizens will have such doubts, and so on. In other words, public officials may flout the law or break informal democratic norms but without its being common knowledge they have done so. They might also comply with all relevant norms without their compliance being common knowledge.

The possibility of such higher-order uncertainty about whether a rule-violation has occurred—uncertainty about not just what has happened but also others’ beliefs about what has happened, others’ beliefs about beliefs, and so on—is one possible explanation of the observation that democracies often deteriorate gradually. In the literature on democratic backsliding, the cases that scholars treat as paradigmatic are cases of incremental decline, and the incremental nature of the process is part of the causal mechanism that is thought to explain democratic backsliding (Bermeo Reference Bermeo2016, 14; Ginsburg and Huq Reference Ginsburg and Huq2018, 118, 119; Haggard and Kaufman Reference Haggard and Kaufman2021, 1–7, 61–73; Przeworski Reference Przeworski2019, 178–83). As Fearon observed of his account of democracy’s self-enforcement mechanism, the central role of coordination in that mechanism may explain the gradual nature of democratic decline.

Some citizens may not observe a “small” transgression of democratic norms and laws, and those who do may wonder how many other people are paying attention. Or they may have a higher order of uncertainty, wondering whether other citizens will wonder how many people are paying attention, and so on. Uncertainty of any order undermines the confidence among democracy’s conditional defenders that enough of their compatriots would join an effort to resist the transgression. The fear that too few citizens are willing to resist then becomes self-fulfilling, as an individual’s participation is pointless if too few participate.

Failures to resist are not the only corrosive effect one might reasonably predict uncertainty about rule-violations to have. One might also expect it to produce failures to accept legitimate exercises of power, such as the January 6th insurrection after the 2020 U.S. presidential election. Citizens form their beliefs about the legitimacy of elections largely on the basis of chains of testimony—journalists report on their interviews with investigators who interviewed eye witnesses, and ordinary citizens see journalists’ reports or get second-hand accounts of what journalists have reported. Donald Trump’s supporters in the 2020 election could not directly observe the actions of poll workers and election officials. The officials’ compliance with the law and democratic norms could only be inferred from indirect sources of evidence, testimony in particular. One would therefore expect Trump and his surrogates to be able to convince a segment of his supporters, who were already suspicious of mainstream media outlets but who (like everyone else) could form beliefs about the existence of electoral fraud only on the basis of testimony from sources they trusted, to doubt the legitimacy of Biden’s victory (Clayton et al. Reference Clayton, Davis, Nyhan, Porter, Ryan and Wood2021). The effort to resist Biden’s assumption of power is then just what the theory of self-enforcing democracy would predict of citizens who were convinced the election was stolen.

These, then, are the two direct effects of (first- or higher-order) uncertainty about norm violations. There are failures to resist norm-transgressing actions because some citizens do not realize democracy is under attack. Fewer citizens are then prepared to defend it even if they believe others are going to protest. Among those who recognize the transgression and are still conditionally prepared to join in its defense, their knowledge that not everyone recognizes the attack for what it is reduces their confidence that enough people will participate in its defense to make their own participation worthwhile. And there are failures to accept a public official’s norm-compliant actions when some citizens falsely believe the actions are wrongful and embark on a misguided “defense” of democracy that, in reality, merely interferes with the lawful and legitimate exercise of power. Under the leadership of cynical elites, their misguided campaign may then attract the antidemocratic partisans who are indifferent to questions of democratic legitimacy.

Higher-order uncertainty plausibly has a further indirect effect on democratic stability through what I refer to as the seeds-of-doubt mechanism, and this is the hypothesized effect that figures centrally in the conception of institutional forbearance I present below. Suppose an officeholder takes an action of uncertain legitimacy, in the sense that its compatibility with the rules—its legitimacy—is not common knowledge. The conditional democrats who observe it and are committed to respecting and enforcing democracy’s rules conditional on an expectation that enough other citizens will do the same will have divergent reactions to the officeholder’s conduct: some of them, believing the action satisfies the relevant rules, will respond by acquiescing to or condoning his conduct, while others, believing it violates a relevant rule, will resist—publicly denouncing it, or joining protests, and so on—creating a failure to accept the legitimate (albeit only dubiously so) exercise of power. These divergent reactions are, however, also what one would expect if some citizens are motivated purely by narrow partisan advantage. Thus some citizens acquiesce (or resist) because they are conditionally committed to complying with and enforcing the rules of democracy, and they sincerely believe the officeholder has complied with (or violated) its rules; but others might acquiesce (or resist) because they prioritize partisan victories over democratic fair play. The effect is that the officeholder’s action, by inducing these divergent reactions, sows seeds of doubt about the breadth of fellow citizens’ commitments to democracy. Having revised downwards their beliefs about the share of the population willing to put respect for democracy’s rules above partisan advantage, conditional democrats will be less inclined to participate in coordinated enforcement of the rules in the future, weakening democracy’s self-enforcement mechanism.

Consider, for example, someone who sincerely believes that Donald Trump did not engage in a criminal scheme to overturn the 2020 election, as Special Counsel Jack Smith’s indictment alleged, and who, moreover, believes there was insufficient evidence to justify bringing the indictment in the first place. From their perspective, the Special Counsel’s conduct—and perhaps also Attorney General Garland’s decision to appoint the counsel—appears to be a politically motivated abuse of power and a violation of the rule of law. Seeing other citizens support or at least acquiesce to the Special Counsel’s conduct, they might entertain the charitable hypothesis that these other citizens have merely formed (in their judgment) an incorrect belief about the quality of the evidence and the merits of the case, but more probably they will suspect many people of being indifferent to the rule of law and happy to condone abuses of power that favor their preferred party or candidate. Many observers have of course drawn the same inference about Republican politicians and rank-and-file voters in light of their indifference to or support of the January 6th insurrection in particular. This inference is the core of the seeds-of-doubt mechanism by which actions of uncertain legitimacy affect democratic stability: conditional democrats, having observed what, to them, appears to be one or the other mode of failure—failures to accept legitimate exercises of power or failures to resist abuses of power—and as a result having revised downwards their beliefs about how many of their fellow citizens are conditionally willing to defend democracy, then become less inclined to defend it themselves, seeing as their willingness is conditional on an expectation that a sufficient number will join the effort and make it successful. The result is cascades of ever deepening mistrust and cynicism, undermining citizens’ ability to coordinate their responses to genuine abuses of power.

FORBEARANCE RECONSIDERED

If this theory is right in broad outline, then it suggests a conception of institutional forbearance that would meet the desiderata set out earlier. The idea, roughly put, is that public officials practice institutional forbearance when they refrain from actions of uncertain legitimacy, that is, actions that respect the constitutive rules (formal or informal) of democracy but whose compatibility with those rules is not common knowledge. The forbearance conjecture is the hypothesis that actions of uncertain legitimacy reduce—and forbearing from such actions thus contributes to—democratic stability. Such actions are likely to produce divergent reactions that sow seeds of doubt about the number of citizens who are committed to respecting and enforcing democracy’s rules. Some citizens, wrongly believing the official has violated democratic norms, will draw cynical inferences about their compatriots’ willingness to enforce democratic norms, while others, recognizing the legitimacy of the official’s conduct, will despair at their compatriots’ unwillingness to accept legitimate democratic outcomes that do not go their preferred way. Failing to practice forbearance thus indirectly foments mistrust and higher orders of uncertainty about the breadth of citizens’ willingness to defend democracy, thereby weakening the foundations of its self-enforcement mechanism.

The case of Senate Republicans’ refusal to hold hearings on Merrick Garland provides an illustration. With the 2016 election looming, Senate Republicans refused to hold hearings on President Obama’s nomination of Merrick Garland to the Supreme Court. Did they thereby violate a constitutional rule governing the permissible exercise of their advice-and-consent power? They did not violate any formally codified rule, but not all constitutional rules are formally codified. If they violated an applicable informal constitutional norm, then they also violated the meta-norm of exercising official prerogatives according to the applicable rules, a norm that I assume is constitutive of liberal democracy, like the rule of law. In this case, the right description of the case is not that they failed to practice forbearance, but rather that they violated one of liberal democracy’s constitutive norms. But this description is not obviously correct. As Levitsky and Ziblatt themselves observe, Senate Democrats had earlier begun “to stray from the norm of forbearance in the area of advice and consent, obstructing an unprecedented number of President Bush’s judicial nominees, either by rejecting them outright or by allowing them to languish by not holding hearings” (152). So the rule “Hold hearings in a timely manner when the President nominates someone to fill a vacancy on the Supreme Court” had not been violated prior to Senator McConnell’s gambit, but the rule “Hold hearings in a timely manner when the President nominates someone to fill a vacancy on a federal court” had been. The past practice of promptly holding hearings on Supreme Court nominees was consistent with each of these rules, but, plausibly, it was motivated by respect for the more generally formulated rule rather than the Supreme Court-specific rule. Once Senate Democrats repeatedly violated this more generally formulated rule, there was presumably no longer a shared expectation that the Senate would promptly hold hearings on judicial nominees in the future. And if there was no such shared expectation, then one might argue there no longer remained a constitutional norm requiring the Senate to do so, seeing as it is built into the concept of a norm that people follow the rule it prescribes in expectation that others will do the same (Bicchieri Reference Bicchieri2017). Conceivably, then, Senate Republicans’ obstructionist strategy did not violate any standing (as opposed to defunct) constitutional norm governing the exercise of their advice-and-consent powers.

But even supposing, for the sake of argument, that the strategy was legitimate, its legitimacy was not common knowledge. Far from it. And to apply the conception of forbearance I have proposed, uncertainty about its legitimacy is enough to conclude that McConnell’s gambit was a failure to practice institutional forbearance. Some people may have felt the norm from which Senate Democrats had deviated a few years earlier was still valid—an informal norm can continue to exist even after some (indeterminate) number of violations have occurred. Moreover, even if their behavior had wrecked the custom of promptly holding hearings for lower-court nominees, some may have thought the Supreme Court-specific version of the custom was still intact. In short, whether the Senate Republicans’ actions complied with the relevant standing procedure was far from common knowledge. Refraining from their obstructionist strategy would thus have been an example of institutional forbearance, as I am proposing we understand it.

A skeptical reader might wonder whether it is generally possible for officeholders to practice forbearance. Many citizens are ignorant of officials’ actions and thus not in any position to form beliefs about their legitimacy, one way or the other. Moreover, the average citizen often lacks the knowledge of the relevant laws and constitutional norms that would permit them to judge the legitimacy of the officeholder’s action even if they do observe it. A skeptic might object on these grounds that no officeholder’s action is ever commonly known to be legitimate, making it impossible to practice forbearance.

A second objection points to the same conclusion even in cases where everyone does observe an officeholder’s action and is reasonably well-informed about the relevant rules. An event is common knowledge if everyone knows it has occurred, everyone knows everyone knows it has occurred, and so on indefinitely (Aumann Reference Aumann1976; Lewis Reference Lewis1969). Requiring that the action’s legitimacy be common knowledge within a large population may seem too demanding. In the case of a single citizen, knowledge of (as opposed to mere belief in) an action’s legitimacy may be demanding. To require this of every member of a large population may seem impossibly demanding.

The initial characterization of institutional forbearance is best understood as a simplification, however. Given the underlying theory, it would be more accurate, albeit somewhat more cumbersome, to characterize forbearance in terms of an approximation to common knowledge. Say that the legitimacy of an officeholder’s action—its compliance with the relevant rules, formal and informal, governing the officeholder’s choice—is approximately common knowledge if almost everyone who is aware of the action confidently believes in its legitimacy, confidently believes that almost everyone who is aware of it confidently believes in its legitimacy, and so on. Note that these beliefs need not be based on the individual’s own knowledge of the relevant rules. An ordinary citizen may confidently believe that an officeholder’s action is legitimate simply because the action did not elicit outcry from the actors who would invariably call foul if the action were illegitimate. If virtually no elected Republicans are questioning the constitutional legitimacy of the process by which Democrats enacted the Inflation Reduction Act, then an ordinary observer can safely infer that it was enacted by legitimate means.

As an amendment to the original formulation, say that an officeholder practices institutional forbearance if they refrain from actions that are legitimate, but whose legitimacy is not approximately common knowledge. There are no grounds for skepticism about the possibility of practicing forbearance, so understood. Suppose, for example, that McConnell had decided to hold hearings on Garland’s nomination soon after it was submitted. Many citizens would have been oblivious, but among those who were aware of his action, none or almost none of them (excepting a possible scattering of cranks) would have believed that the decision violated constitutional law or norm. Moreover, almost all observers would have been not just confident of its constitutionality, but also confident that almost everyone else who observed the action was confident of its constitutionality, and so on. The decision to hold hearings would have met with widespread acquiescence to its constitutional legitimacy among conditional democrats (even if some Republican partisans would have objected on other grounds), and their acquiescence would not have raised doubts about their commitments to upholding democratic norms. By contrast, acquiescence to McConnell’s refusal to hold hearings did contribute to such doubts, precisely because its legitimacy failed to be even approximately common knowledge.

Democratic officeholders can usually choose among various actions whose constitutional legitimacy is approximately common knowledge in the stipulated sense. Elected legislators can choose to propose legislation according to the rules, or refrain; they can vote for or against proposed legislation; officials can choose actions that are clearly and unambiguously in compliance with court rulings. In all these spheres, public officials have available to them actions whose constitutional legitimacy is uncontroversial—even if their choice is bound to provoke other kinds of controversy. Perhaps they cannot help but act in ways that many citizens will denounce as unjust, immoral, self-serving, and worse. But the question raised by the skeptic’s objection is not whether democratic officials can act without courting controversy; the question is rather whether they can act without provoking controversy specifically about whether their actions violated constitutional laws or the informal constitutional norms respect for which is constitutive of liberal democracy.Footnote ¹⁰ Under normal circumstances they clearly can (I discuss exceptions to the rule below). The public reception of these actions, contentious as it may be, should not raise significant doubts about the number of citizens who are prepared to resist unconstitutional abuses of power.

To preempt a different kind of objection to the argument, it is important to distinguish between the direct and total effect of an act of forbearance. The theory I have sketched supports a conjecture about the direct effect of an act of forbearance on democratic stability, mediated by its effects on beliefs and mutual trust. But an act of forbearance might exert countervailing effects along other causal pathways, so its total effect on democratic stability is theoretically indeterminate. The forbearance conjecture is the hypothesis that, holding fixed the intermediate outcomes through which other causal pathways might link an act of forbearance to democratic stability, and considering only the effect that travels along the path involving the mechanism of mutual trust and coordinated responses to abuses of power, the effect is to improve stability. However, in some circumstances the total effect of a public official’s act of forbearance could be to further democratic backsliding, even supposing the forbearance conjecture is true.

To illustrate the idea, suppose a future Senate majority leader employs an unusual parliamentary maneuver to effect a general repeal of the filibuster by simple majority vote (perhaps they use “the nuclear option,” deployed by Senate majorities to limit the filibuster in 2013 and 2017, or perhaps a more arcane maneuver like the hypothetical example Shepsle (Reference Shepsle2017) discusses). Let us suppose that the maneuver, whatever it is, is legitimate—compatible not just with the Constitution but also with existing Senate rules, and thus compatible with what I earlier described as a meta-norm of following applicable rules and procedures—but its legitimacy is contested and far from common knowledge. Thus, refraining from the maneuver would be an exercise of forbearance, as I have characterized it. When the majority leader repeals the filibuster by this method, some observers, believing it violates the Senate’s rules, may interpret their colleagues’ acquiescence as evidence of their willingness to condone breaches of democratic norms when it suits them. The forbearance conjecture is that the direct effect of the procedurally dubious repeal, mediated by its corrosive effect on the shared confidence that undergirds coordinated responses to violations of the rules, is to weaken democratic stability. Yet we can imagine circumstances in which the act would simultaneously affect democratic stability by other causal pathways, and the sum-total of these disparate effects might therefore be uncertain even if the forbearance conjecture is true. For example, the filibuster might facilitate so much obstructionism that its repeal, even by a maneuver of uncertain legitimacy, would, on net, contribute to democratic stability—ending gridlock and reducing the political dysfunction that fuels antidemocratic populists, even as it also chips away at the shared expectations that underlie the enforcement of democratic norms.

This conception of forbearance satisfies the three criteria discussed earlier. Most importantly, it preserves the potential explanatory power of forbearance as a mechanism of democratic stability. Recall that the conjecture has explanatory power only if the actions from which public officials must forbear, per the conjecture, are actions that are in and of themselves fully compatible with democracy’s constitutive rights, procedures, and norms—including the meta-norm of following all applicable rules and norms governing the exercise of their rights and prerogatives. The proposed conception of forbearance—refraining from actions that have only uncertain legitimacy, in the sense that it is not even approximately common knowledge that they respect the constitutive rules and norms of liberal democracy, even if they do in fact—satisfies this condition. The same actions, in counterfactual circumstances in which their normative status were approximately common knowledge, would have no deleterious effects on democratic stability at all, and there would be no reason for a democrat to forbear from them. The conjecture is that when their status is not approximately common knowledge, they contribute to a spiral of mistrust that weakens democracy’s self-enforcement mechanism. The magnitude of the conjectured effect may be greater or smaller, depending on the circumstances, or it might not exist at all if the theoretical argument about coordination games and the seeds-of-doubt mechanism is wrong. The forbearance conjecture is not trivially true, and the argument for it has testable implications. In line with much of the theoretical literature on antiregime protests (e.g., Bueno de Mesquita Reference Bueno de Mesquita2010; Fearon Reference Fearon2011; Kuran Reference Kuran1991; Little Reference Little2016), the argument assumes that citizens’ decisions to resist breaches of fundamental democratic rules, like an incumbent’s suspension of elections, are strategic complements: a citizen is more inclined to participate in resistance the greater the expected number of other participants. But this is a testable claim, which might fail to hold in certain contexts (Cantoni et al. Reference Cantoni, Yang, Yuchtman and Zhang2019). Apart from testing such assumptions of the theory, one could also test the seeds-of-doubt mechanism. If one could design survey experiments that reliably measured the beliefs and attitudes implicated in the theory, one could test the hypothesis that observing other citizens’ acquiescence to an officeholder’s illegitimate exercise of power undermines the respondent’s confidence that the other citizens would be willing to resist future abuses of power.

This conception of forbearance satisfies the criterion of theoretical grounding, too. The forbearance conjecture is not merely a description of an empirical regularity, or an untheoretical and ad hoc attempt to fit a curve to observed data points. Existing theories of democratic stability—theories that assume that the prospect of coordinated resistance to abuses of power is one of democracy’s stabilizing mechanisms—support the conjecture that public officials’ failure to practice institutional forbearance would contribute to democratic backsliding.

Finally, the definition of “forbearance” preserves the broad scope of this conjecture, satisfying the last of our three criteria. The causal link between forbearance and democratic stability is not contingent on quirks of institutional design, like the Weimar constitution’s Article 48. The argument for the forbearance conjecture applies in any circumstances where the mechanism supporting a self-enforcing democratic equilibrium involves the threat of coordinated resistance from some set of observers (ordinary citizens, elites, etc.), who cannot be certain that all the other members of the set are committed to defending democracy. It is thus broadly applicable.

In thinking about its scope, the reader might wonder whether it applies under realistic assumptions about public officials’ motivations. The word “forbearance” has connotations of virtuous self-restraint. According to the theory outlined here, one should indeed regard a disposition to practice forbearance as a civic virtue; one can see it as a particular aspect of the more general virtues of moderation and political cooperation that liberal and republican accounts of democratic stability emphasize (F. Lovett Reference Lovett2022, 194–202; Rawls Reference Rawls2005, 157). One might hope politicians would exercise forbearance out of respect for democracy, but if that were the only possible motive, one might also doubt whether it could explain extended periods of democratic stability. The theory put forward does not assume ethically motivated politicians, however. Notwithstanding its connotations of virtuous self-restraint, “forbearance” refers here to conduct that could have various underlying motives, including the crass motives one expects of garden-variety democratic politicians. Assume, for the sake of argument, that politicians care only about the rewards to accumulating and exercising power. By taking actions whose compatibility with democratic norms is not approximately common knowledge, the politician contributes to the gradual erosion of the epistemic foundations of citizens’ ability to launch coordinated responses to true transgressions against democratic norms. Such a policy might seem attractive to amoral, power-seeking incumbent officeholders; it is, after all, a means of whittling away at the ultimate constraint on their power. No doubt it will be attractive to many of them; the theory underlying this account of forbearance implies that aspiring autocrats have instrumental reasons to seek out actions of uncertain legitimacy, as these are opportunities for spreading mistrust and dividing the popular coalitions that might attempt to resist them. But for the more circumspect democratic politician, such actions are a double-edged sword, like any other strategy for weakening democratic institutions. Free and fair elections are the incumbent party’s safest route back to power if it suffers an electoral defeat. It would thus be a mistake to think that forbearance can operate as a mechanism of democratic stability only in an idealized polity whose leaders value democracy for its own sake, or who are motivated by other moral considerations. Just as rational calculation of self-interest may lead politicians to practice other democratic virtues that are conducive to stability (Basu Reference Basu2021), it may also recommend practicing forbearance out of caution, given the risks that democratic backsliding entails (cf. Przeworski Reference Przeworski2019, 20, 21). Moreover, under favorable conditions acts of forbearance can serve the politician’s short-term interest in winning the next election even if the politician is indifferent to their long-term impact on the survival of the regime. A politician may prudently forbear from actions whose compatibility with basic democratic norms is open to doubt if he fears that they would cost him electoral support.

A different question is whether forbearance helps explain democratic stability only in nonideal circumstances. As observed earlier, one way to limit the scope of the forbearance conjecture is to define “forbearance” as refraining from actions that, despite being fully compatible with the constitutive rules and norms of democracy, would provoke other actors to abandon their tenuous support for democracy. On that characterization, forbearance is like the indulgence that a parent shows her child when she refrains from making the best move in a game of chess, knowing that the child would refuse to continue playing if his prospects of winning significantly diminished. Interpreted this way, forbearance would be unnecessary in a hypothetical population whose members were all conditional democrats, willing to comply with democratic norms provided that others do the same; such an idealized democracy would then fall outside the scope of the forbearance conjecture. As I have characterized forbearance, however, the conjecture could hold even in these idealized circumstances. Even if everyone has idealized motivations there will be some actions whose compatibility with the relevant rules is not approximately common knowledge. These actions can still precipitate the destabilizing cascades of mistrust if citizens are uncertain about each other’s motivations. To place a hypothetical society outside the scope of the theory, one would have to idealize not only its members’ motivations but also their information about each other. They must be not merely democratic saints, but implausibly well-informed saints besides.

THE DEMOCRAT’S DILEMMA

If democracy is valuable and the preceding argument is right, then institutional forbearance is also valuable, all else being equal. The ceteris paribus qualification is important, however. The value of democracy does not always tell decisively in favor of practicing forbearance. Take for example President Obama’s policy of deferred action for childhood arrivals (DACA), which allowed individuals who entered the country as undocumented immigrants as children to receive renewable two-year periods of “deferred action” from deportation. In January 2025, the Fifth Circuit Court of Appeals upheld a lower court’s decision that found parts of the DACA unlawful, and the case may still be appealed to the Supreme Court. Clearly, whether President Obama’s action was a legitimate exercise of his constitutional authority is far from even approximately common knowledge. Thus the preceding argument implies that the value of preserving democracy—specifically, the mutual trust that facilitates coordinated responses to abuses of power—was a pro tanto reason for President Obama to forbear from exercising his power in this way, even supposing it was fully legitimate. But it is only a pro tanto consideration. There were also other values at stake—of justice, decency, economic efficiency—which may in the aggregate have outweighed the concern with democratic stability. The claim that forbearance is instrumentally valuable as a means of preserving democracy does not imply that public officials ought always to practice forbearance. If democracy is not the only good, then one ought to weigh these other values against the value of democratic stability.

One also has to estimate the magnitudes of the different effects that an action has and scale the weight one gives to the different values accordingly: perhaps the effect of DACA on democratic stability was small, while its effect on the welfare of the undocumented immigrants it targeted substantial. Given the theoretical argument behind the forbearance conjecture, one would expect the effect of a particular act of forbearance to vary in magnitude depending on the circumstances. For example, the greater the set of individuals among whom the action’s legitimacy is approximately common knowledge, the smaller the set of citizens whose resistance to the action will raise doubts about their willingness to accept legitimate exercises of power. For this reason among others, one would not expect all failures of forbearance to be equally damaging to democratic stability. The normative argument for forbearance must be context-specific.

Even if one prioritized democratic stability over competing values, there would be yet another reason for sometimes excusing public officials for not practicing forbearance. As explained earlier, the conjecture that acts of forbearance contribute to democratic stability is a claim about a direct effect, but the same action might exert countervailing direct effects along other causal pathways, such that the total effect of an act of forbearance might actually be to worsen democratic backsliding. I gave the example of a simple Senate majority employing dubious parliamentary procedures to repeal the filibuster, but another illustration of the point, of greater interest to normative democratic theory, is a class of situations I will call the Democrat’s Dilemma.

When a public official or political party has violated a democratic norm, other actors must choose how to respond, and they will typically have a range of options that are more or less retaliatory. Among the more aggressive forms of retaliation there will be actions that are themselves violations of democratic norms and would be licensed by a principle of “negative reciprocity” or tit-for-tat strategy.Footnote ¹¹ Because this strategy, by “feeding spirals of punitive normative breaches,” may further destabilize democracy, Andreas Schedler argues that democrats must instead practice “democracy-preserving reciprocity,” which tries to steer a middle path between the “Scylla of mechanical retaliation and the Charybdis of passive endurance” (Schedler Reference Schedler2021, 258).Footnote ¹² Its most basic principle is to prioritize democratic means of resisting attacks on democracy when these are available: holding bad actors accountable through normal democratic channels (260). Schedler is concerned with the risk of an escalating spiral of reciprocal norm-violations that a simple principle of negative reciprocity would license, but even the “tools of democratic accountability” carry risks of similar destabilizing processes when there is not a shared consensus about whether democratic norms have been violated in the first place. The Democrat’s Dilemma arises in just such a situation, where the normal strategies for defending democracy conflict with the principle of forbearance.

For example, a candidate for office may attempt an unlawful intimidation of an election official in a bid to subvert a free and fair election. Government prosecutors may decide to bring charges against the politician. By permitting its prosecutors to do their job and let justice run its course, the incumbent party does not violate any democratic norms; indeed, it might even be considered a violation of the rule of law, a core principle of liberal democracies, if the prosecutor declined to indict the opposition candidate, or if the incumbent chief executive pressured the prosecutor to drop the investigation. Even if it would not be a violation of any democratic norm per se, failing to punish a party’s attempt to subvert an election might strengthen incentives for future attempts. The value of democracy is a reason to hold the politician accountable. Yet the same actions would be a violation of democratic norms, the rule of law in particular, if the opposition candidate were innocent of the charges, there was little evidence to suggest he was guilty, and the prosecutor, either at his own initiative or under pressure from the chief executive, launched the investigation as a means of discrediting the candidate. Unfortunately, whether there is compelling evidence of criminal conduct, of the kind that would lead an impartial prosecutor to bring charges, is not likely to be common knowledge, or anywhere close to it, when citizens are polarized along partisan lines and do not have common sources of trusted information. The prosecutor is in a bind where either course of action might imperil democracy. He can forbear from exercising his right to bring charges against the politician, but his forbearance might embolden future bad actors to undermine democratic institutions; or he can attempt to hold the politician accountable, knowing that a significant segment of the population will then draw the inference that he and his supporters are willing to flout democratic norms in pursuit of partisan advantage. This is an example of the Democrat’s Dilemma.

A particularly severe form of the dilemma arises when a public official must choose between two actions and neither action’s legitimacy is approximately common knowledge. In such a case it is not even possible to practice institutional forbearance. Indeed, some readers might interpret the prosecutor’s situation as just such a case: if one part of the public believes that respect for the rule of law, a constitutive norm of liberal democracy, requires that the prosecutor bring charges, while another believes that it prohibits him from doing so, then the prosecutor cannot help but choose a course of action whose legitimacy fails to be approximately common knowledge.

With enough information, the Democrat’s Dilemma would become a well-defined choice problem with a straightforward solution—not a true dilemma at all. The prosecutor’s choice of action has the potential to affect future bad actors’ beliefs about whether they will be held accountable for criminal attempts to subvert elections; it also has the potential to contribute to partisan mistrust and the erosion of the coordination mechanism that sustains democracy over the long term. If the prosecutor can assign probabilities to these different possible effects and compare their expected magnitudes, and the goal is simply to preserve democracy, then one or the other action will be best, in expectation, and there is no true dilemma. Obviously this is easier for the decision theorist to describe in the abstract than for the real-world political actor to calculate in practice. Lacking good information about the possible effects of the two courses of action, a situation like the prosecutor’s feels like a genuine dilemma, where the value of democracy seems to counsel against each action simultaneously. But the point I wish to underscore is that in principle the dilemma might resolve in favor of not practicing forbearance. The theory put forward here implies only that forbearance is valuable all else being equal, not that public officials ought always to practice it.