[UPDATE, October 31: If you read this post and conclude that my argument is that “both sides do it” then you have missed my point, which is found not in paragraph 1 but in the remainder of the essay.]

The 2016 American presidential election has caused many political scientists to ask whether American democracy is fundamentally vulnerable. Importantly, the source of democratic vulnerability for most political scientists is not the identity of either of the candidates, nor even (usually) their policy positions, however potentially illiberal they might be. What is actually threatening to democracy itself is that citizens would come to believe that democratic institutions do not work. Trump’s even raising the possibility that he might not concede the election is one example. My view on this is unpopular, but the immediate uproar from the American left to FBI Director James Comey’s announcement of new emails related to the Clinton email scandal is a structurally identical example. The allegation in both cases is that when agents of the state act in ways that might affect one political party’s electoral interests, then state institutions are fundamentally biased.

How to think about this theoretically? I think it most helpful to recall insights from Dankwart Rustow’s landmark 1970 article “Transitions to Democracy: Towards a Dynamic Model” (PDF). This article is famous for rejecting just about every argument about the so-called “social preconditions” for democratic stability, and instead focusing on how citizens come to habituate themselves to democracy, even in deeply divided societies.

But Rustow does identify on important precondition for the emergence of democracy.

The model starts with a single background condition—national unity. This implies nothing mysterious about Blut und Boden or daily pledges of allegiance, about personal identity in the psychoanalyst’s sense, or about a grand political purpose pursued by the citizenry as a whole. It simply means that the vast majority of citizens in a democracy-to-be must have no doubt or mental reservations as to which political community they belong to…Democracy is a system of rule by temporary majorities. In order that rulers and policies may freely change, the boundaries must endure, the composition of the citizenry be continuous.

The idea here is even simpler than he puts it. If you don’t agree on who is able to participate in a democracy, you cannot have a system in which those participants allocate power to one another temporarily. Group A cannot have the position that Group B does not have the right to participate. Group B need not always, or ever, win—but the question of its participation in the first place must be off the table permanently. That’s the deal.

This seems quite a leap, from citizenship and the boundaries of the political community to whether democratic institutions are functioning. But a logical implication of Rustow’s argument that Group A must not question Group B’s participation in democracy is that Group A must also believe that it itself can participate in democracy. That is also part of the deal. If Group A believes that it cannot participate in democracy even if it wants to, then this upends any argument about why it ought to tolerate democratic procedures.

The point is not that all parties in a democracy must agree. The point is that all parties, all groups in a stable democracy agree to disagree by rules, and that is only feasible when those rules are acceptable to all parties. It requires trust in those rules and procedures through democracies are governed. From Rustow,

…new issues will always emerge and new conflicts threaten the newly won agreements. The characteristic procedures of democracy include campaign oratory, the election of candidates, parliamentary divisions, votes of confidence and of censure-a host of devices, in short, for expressing conflict and thereby resolving it. The essence of democracy is the habit of dissension and conciliation over ever-changing issues and amidst ever-changing alignments. Totalitarian rulers must enforce unanimity on fundamentals and on procedures before they can get down to other business. By contrast, democracy is that form of government that derives its just powers from the dissent of up to one half of the governed.

Trust is nothing more than the agreement, and the self-understanding, that one is part of the community empowered to participate in democratic government. The absence of this is a case like Thailand, in which democratic procedures that encompass equal participation have been decisively rejected.

There are obvious caveats to this understanding of citizenship, trust, and democracy. Many democratic regimes can survive for a time when a significant minority of the population government by the regime does not possess full citizenship (and hence full participatory) rights. But these are cases in which it is hard to argue that that system is itself democratic, rather than something else. And such regimes thrive only so long, as both the American Civil War and the Civil Rights Movement showed. It is also worrying that questions about citizenship and participation are being raised directly, although these are only sometimes explicit (immigrants, Muslims, etc.) and often implicit (voter fraud).

Holding those caveats aside, though, adopting Rustow’s view has important implications. Like just about every political scientist, I fear what happens if Trump does not concede the election, or tells voters that he didn’t really lose. But I also fear how the election has politicized government institutions even among moderate Democrats. UPDATE, October 31: Yes, even among moderate Democrats. That supporters of Secretary Clinton seem to trust U.S. political institutions is not informative about the partisan distribution of trust in institutions. We don’t learn about trust in institutions from those who think that those institutions will deliver them an electoral victory. That is my point regarding the Comey scandal.

One surprising implication of Rustow is the following. If Republican voters increasingly believe that they cannot win presidential elections (either because they really can’t, or because their party elites tell them so), then to safeguard democracy over the long term, Democrats should want Republicans to nominate moderate successful candidates that they would be willing to lose to. Democrats should want to bury Trump and salt the earth from which he emerged, but they should also want to nurture a party that represents those whom they do not. They should be wary of their own messages that every political defeat means that democracy is broken. Indeed, believing that might actually make it true.