President Trump has had a bad week. Last week’s firing of FBI Director James Comey was followed last night by reports that the President has revealed classified information to the Russians. This news has added to the steady drumbeat of criticism from President Trump’s domestic critics, many of whom have consistently called for his impeachment (see here, here, here, and many other places; this is, of course, an old genre of anti-Trump writing). Recently, Evan Osnos published a long essay in the New Yorker entitled “How Trump Could Get Fired.”

How should we think about the possible end of the Trump administration? The analyses above target the case for impeachment, the partisan politics of impeachment, or the Constitutional provisions for impeachment or removal via the Twenty-Fifth Amendment. These are useful but ultimately narrow perspectives. Analytically, it is more useful to consider the problem of presidential impeachment through the lens of democratic transitions.

Democratic transitions are the bread and butter of comparative politics. They are highly visible, dramatic political moments, as a Trump impeachment or removal would certainly be. Because most political scientists share a normative commitment to democracy, democratic transitions also carry significant meaning to those who study them—again, as a Trump impeachment would. Those of us who have studied transitions (and failed transitions) are humbled by the acts of heroism by ordinary people who stand up for rights and liberties, and by movements who organize collectively to press for political change.

A transitologist’s perspective, can, however, put those actors and movements in perspective. Here are two important things to note in the current U.S. political moment.

First, popular movements write histories in which they are the agents of change. One pertinent example is the People Power movement in the Philippines. This is a classic case of democratic transition via social movement, and has served as a template and an inspiration for other pro-democracy movements since, such as the so-called Color Revolutions of the 2000s. And yet the details of the final days of Ferdinand Marcos point to key decisions made by internal security forces and external patrons such as the United States as the proximate drivers of regime breakdown. Moreover, as political scientist Benedict Anderson observed long ago, Corazon Aquino came from the same network of colonial and postcolonial elites as did Marcos. The transition was made easier by the general sense among that faction of Filipino elites that democratization would ultimately not be that radical of an exercise.

Second, “heroes” are situational. For example, in my view, one of the heroes of Indonesia’s 1998-99 democratic transition was Soeharto’s final vice president, B.J. Habibie. Habibie is an unlikely hero because he was the handpicked subordinate of a brutal dictator, but he did oversee a transition that he had not planned but might have stopped had he cared to try. Habibie’s importance for Indonesian democratization does not depend on whether or not he was a good person, nor on whether or not his intentions at the time were noble.

These observations suggest that if there is to be an actual non-electoral mode of transition in the United States—and that is, after all, what impeachment or constitutional removal would be—mass opposition to Trump may be the background condition, but it is unlikely to be the proximate causal factor that sets in motion the end of the Trump presidency. Making sense of that prospective causal sequence will require looking more soberly to the incentives of different actors and the political game that they think that they are playing. The literature on democratization has a language for this: hardliners and softliners, pacted transitions, elite settlements, uncertainty, and contingency.

The classic statement is Guillermo O’Donnell and Philippe Schmitter‘s Transitions from Authoritarian Rule (pdf). Anyone who is serious about understanding how to build towards impeachment or constitutional removal should read this short book carefully. It makes several observations about democratic transitions: that transitions are moments of high uncertainty, that splits within the regime are the necessary precondition, and “transitional justice” is inevitably an issue. Applied to the problem of contemporary U.S. politics, the key question is not “how much will the Congressional GOP tolerate?” or “will the GOP’s principles ever kick in?” but rather “can the Congressional GOP be convinced that it has a future?”

This point has uncomfortable implications. The extreme polarization of U.S. politics means that Democrats see the only way forward as using President Trump as a symbolic cudgel, to defeat every Republican elected official and then to hold them to account for their actions. The threat of electoral defeat, so the thinking goes, might be sufficient to force GOP elites to break with the administration.

But that is not how you build a coalition to move President Trump out of office. Rather, if one were to chart out a transition pathway that follows the process model of O’Donnell and Schmitter, it would involve identifying softliners within the Trump administration and the Congressional GOP who are “biddable” on impeachment or removal. It would then involve a pact of some sort in which those softliners participate in impeachment or removal, almost certainly in return for something of value to them—perhaps health care, perhaps federal judiciary positions, perhaps support fending off the inevitable primary challenge. That pact is a political compromise which makes change possible by allowing the softliners to free themselves from their dependence on the Trump administration.

It should be obvious here that an underappreciated problem in mobilizing GOP softliners against President Trump is the problem of mobilizing their Democratic counterparts to accept such a pact. Who is willing to trade partisan principles in favor of broader constitutional stability?

Transitions from Authoritarian Rule is full of other relevant gems. For example, on soft-liners:

they may be indistinguishable from the hard-liners in the first, “reactive” phase of the authoritarian regime. They may be equally disposed to use repression and to tolerate the arbitrary acts of the appropriate ministry or security agency. What turns them into soft-liners is their increasing awareness that the regime that they helped to implant, and in which they usually occupy important positions, will have to make use, in the foreseeable future, of some degree or some form of electoral legitimation.

And on the problem of clemency for past crimes:

Where [the guilty parties] cannot prevent the transition, they will strive to obtain iron-clad guarantees that under no circumstances will “the past be unearthed”; failing to obtain that, they will remain a serious threat to the nascent democracy.

There are limits to this analogy. O’Donnell and Schmitter were writing mostly about hard authoritarian regimes, often led by juntas who murdered their opponents. President Trump is not a hard authoritarian. “Transitional justice” is not the question of whether or not to convene a truth and reconciliation commission to investigate death squads, but rather owning the President’s broadly unpopular legislative agenda and legacy, and allowing some in the GOP to claim credit for its success.

Stepping back, however, the general lessons from transitology are broadly applicable to the present moment. Transitions from authoritarian rule are political processes, not simply constitutional procedures. Put differently, the Constitution has procedures for executive removal that are easily enacted, but only after the political settlement has been reached. Political transitions almost inevitably require some degree of compromise among self-interested political actors, not the discovery of hidden democratic heroes with impeccable democratic credentials Who Have Finally Had Enough.

The smart but unpleasant move for those who believe that President Trump is unfit to serve is to start thinking about what kind of pacted transition would acceptable. The alternative is to wait for the 2018 elections. But if you thought that waiting until 2018 were preferable to a pacted transition now, you probably would not have read this far.