Here and there, constitutional analysts are discussing whether Congress could provide for a special presidential election as a sort of do-over for 2016, if both President Trump and Vice-President Pence were impeached and removed for complicity in Russia-related shenanigans that altered the outcome of the election.
The best answer, I think, disaggregates the issue into three different questions: (1) If Congress were to pass a law purporting to call a special election, should the courts strike that law down as unconstitutional? (2) Should a conscientious member of Congress who thought a special election would be good policy nonetheless vote against a special-election bill on constitutional grounds? (3) Is it plausible that the congressional lawmaking process, which is shaped both by formal constitutional structures and by the constitutional expectations of the players, could produce a statute calling for special elections to replace Trump and Pence?
I think the answer to (1) is “No,” but I also think that the answer to (2) is “Yes.” My most confident response is to (3), and the answer to that one is “No Way.”
In other words, I think that Congress has the legal authority to remove the President and Vice-President and to call a special election to replace them. But Congress’s authority to pursue that course of action is in important ways analogous to Congress’s authority to enact a 99% income tax, or to provide that government-printed calendars shall label the ninth month of the year July and the sixth month of the year September. Learning the principles of constitutional interpretation that lead to these conclusions involves developing a set of understandings and skills that can be useful as applied to non-hypothetical cases. But the hypothetical cases remain hypothetical, and figuring out the best answers in those cases is never ultimately the point of the enterprise. By a similar token, asking whether Congress can provide for special elections to replace Trump and Pence as if the question concerned an actionable possibility strikes me as among the worst uses of constitutional analysis. It’s a form of escapism.
Imagine asking a thousand American lawyers whether Congress had the power to call a special presidential election if the sitting President and Vice-President were both removed from office. Not having tried it, I can’t know for sure what the thousand lawyers would say. But I suspect a large majority would say Congress has no such power.
The sources of that dominant intuition—assuming that it is in fact the dominant intuition—aren’t hard to identify. We think of the Constitution as providing for a four-year presidential term. When Presidents have resigned or died in office, their successors have served the remainder of the four-year period for which the previous Presidents were elected. Never in the history of the Republic has an election occurred except on the every-four-years schedule. Indeed, the four-year schedule seems to most of us like as fixed a feature of the constitutional system as there is.
But considered carefully, the text of the Constitution leaves open the possibility that Congress could call an earlier-than-usual election if the President and Vice-President were simultaneously removed. The relevant language is at Article II, Section 1, clause vi. It reads as follows:
In case of the removal of the President from office, or of his death, resignation or inability to discharge the powers and duties of said office, the same shall devolve on the Vice President, and the Congress may by law provide for the case of removal, death, resignation or inability, both of the President and Vice President, declaring what officer shall then act as President, and such officer shall act accordingly, until the disability be removed, or a President shall be elected.
Focus on the last words, which provide that a statutorily specified successor acts as President “until . . . a President shall be elected.” On a conventional understanding, those words mean “until the next regularly scheduled election occurs.” That’s a perfectly plausible reading. In fact, if we assume that presidential elections occur only every four years, it’s the only possible reading. But nothing in the Constitution says that no presidential election can occur sooner than four years after the previous one. (Go ahead and look.)
If we relax the assumption that presidential elections occur every four years and never sooner, a different and equally plausible interpretation of the last words of clause vi comes into view. The words “until…a President shall be elected” could mean “until we can manage, ASAP, to elect a President.”
The ASAP reading would probably strike most constitutional lawyers as weird. We are accustomed to thinking of the four-year presidential term as hard-wired into the system. We’ve never held a presidential election except in a year divisible by four, for any reason. What’s more—and here I’m engaging in psychological speculation—it wouldn’t surprise me if the intuition that presidential elections are hard-wired into their familiar quadrennial form is strengthened by the fact that the four-year presidential election cycle happens to match (at least since 1904) the four-year cycles for leap years and the Summer Olympics, and that the years when all of those things occur happen to be years whose numbers are divisible by four.
That’s all a matter of coincidence. But it comes together to create a lived reality in which every four years, starting with the 00 year of the century, we have a sort of jubilee year, a year when we do Special Things that Happen Only Every Four Years. Whether or not those other four-year cycles reinforce the intuition that presidential elections are done every four years and no other way, we have a strong intuition, rooted in experience, that that’s the way presidential elections are done.
Put in terms I’ve used elsewhere, the four-year cycle is part of our constitutional expectations. Constitutional expectations are, to summarize, our deeply rooted intuitions about how the system is supposed to work, intuitions that are the joint product of experience, socialization, and principle. When we have a stable set of constitutional expectations, we tend to read the text of the Constitution as if it dictated the arrangements we expect to see. As applied here, our expectation that presidential elections occur only on a pre-set four-year cycle leads us to read Article II, Section 1, clause vi as if it directed that an Acting President serve until that regular cycle ran its course, rather than noticing that the language would also be satisfied by a special election held on an earlier schedule.
The stability of many of our constitutional expectations is a key element in keeping the constitutional system workable. But it’s also true, and important to remember, that constitutional expectations sometimes change over time. In the first years of the Republic, before Americans had grown accustomed to a pattern of elections every four years and never more frequently, the idea that presidential elections could occur only in every fourth year wasn’t yet a dominant view. On the contrary, Congress in 1792 enacted a law providing for early elections in the event that the Presidency and Vice-Presidency became simultaneously vacant. What’s more, the new President chosen at the early election would serve a full four-year term, rather than simply filling out the remainder of the last President’s term. So in the contemplation of this early Congress, if a President and Vice-President elected in the year 1800 were killed or removed in 1803, a special election could be held in 1803, and the next regularly scheduled election would be in 1807.
The fact that Congress in 1792 deemed early elections appropriate doesn’t prove that such elections are constitutional. Maybe Congress passed an unconstitutional law. (James Madison, then serving in the House, seems to have thought the law was unconstitutional. Then again, Madison argued that lots of laws Congress passed over his objection in the 1790s were unconstitutional.) Maybe—not for sure, but maybe—there are good structural arguments against early elections.
But it’s also plausible that Congress in 1792 made a constitutionally valid choice. As a matter of structure and of democratic theory, it’s not crazy to think that the best way to identify a President is to elect one. And nothing in the constitutional text clearly precludes, or even cuts heavily against, holding an early election. As noted before, Article II says that an Acting President serves “until…a President shall be elected,” not “until a president is elected at the next regularly scheduled election.” And to deepen a point made earlier, no constitutional text prevents a presidential election from being held less than four years after the previous one. Consider this comparison: Article I, Section 2, clause 1 provides that members of the House of Representatives shall be “chosen every second year[.]” But Article II, Section 1, clause iv simply says that “Congress may determine the time of choosing [presidential] electors, and the day on which they shall give their votes[.]” It doesn’t say that no election for President can occur sooner than four years after the previous election. If it wanted to, Congress could provide that the election to decide who will become President on January 20, 2021, will be held in December of 2017.
To be sure, Congress can’t replace a President midway through a term just by calling an early election: the power to set the date of an election is not the power to end a term early. If Congress wants to replace a President who has served less than four years, it has to use impeachment and removal. But if the office is vacant, or if it is occupied by an officer who is authorized to serve only “until . . . a president shall be elected,” then electing a new President to take office right away does not prematurely end anyone’s rightful time as President. In short, Congress’s ordinary power to decide when presidential electors will be chosen and cast their votes seems perfectly adequate for ordaining earlier-than-usual elections under the special circumstances described in Article II, Section 1, clause vi.
None of this proves that calling early elections in the case of a double vacancy would be a good idea. Today, presidential succession is governed by a 1947 statute under which statutorily designated Acting Presidents are supposed to serve until the end of what would have been the previously elected President’s four-year term. If I were a Member of Congress, I’d be reluctant to go back to the possibility of off-cycle elections, because I’d be loath to unsettle the now-prevalent constitutional expectation that legislatures do not alter the timing of the electoral cycle. Keeping that expectation settled helps prevent certain kinds of unhealthy political gamesmanship that occur in democratic systems with variable election cycles: Once Congress gets used to the idea that there can be an election sooner than four years after the last one, it will also start thinking about timing elections for partisan benefit, rather than leaving them to occur on the same schedule in every cycle.
But if Congress were to depart from our current set of constitutional expectations and provide by law for early elections in the event of the simultaneous removal of the President and Vice President, it would be easy to defend the constitutionality of that law in court. The text of the Constitution empowers Congress to set the dates of presidential elections. Nothing in the text prohibits earlier-than-usual elections, and Article II plausibly contemplates such early elections in the double-vacancy case specified in Section 1, clause vi.
One could argue against a special-election statute by saying we have a long history of holding elections only every fourth year, and there might be wisdom in such an argument. But under the currently normal conventions of constitutional discourse, arguments from that sort of tradition are strongest when the tradition in question goes all the way back to the beginning, and Congress in 1792 quite clearly provided for early elections. (I’m talking here about the conventions for arguments that courts take to be sufficient reason to invalidate legislation. In a legislature or within the executive branch, the fact that we’ve done something a certain way for a very long time is often an excellent argument for proceeding as if things need to be that way, even if “a very long time” is only a century or so.) And arguments from constitutional structure are available on either side. Yes, there’s something healthy about treating the election cycle as an invariable feature of the system—and as a legislator, I might consider that point dispositive. But in the face of a contrary legislative decision, it might be hard to withstand the democratic-theory point that the presidency is supposed to be an elected position and the separation-of-powers point that as a general matter, our system regards legislatures, not courts, as the bodies competent to schedule and administer elections. And in particular, Article II, Section I, clause iv gives Congress the power to schedule elections for the office of President.
At the beginning of this post, I wrote that the question of whether Congress could call early elections in the event that President Trump and Vice President Pence were both impeached and removed should be understood as three different questions, as follows: (1) If Congress were to pass a law purporting to call a special election, should the courts strike that law down as unconstitutional? (2) Should a conscientious member of Congress who thought a special election would be good policy nonetheless vote against a special-election bill on constitutional grounds? (3) Is it plausible that the congressional lawmaking process, which is shaped both by formal constitutional structures and by the constitutional expectations of the players, could produce a statute calling for special elections to replace Trump and Pence? So far, I have addressed (1) and (2). It’s now time for (3).
Even if a majority in Congress came to believe that as a general matter it would be good policy to hold immediate presidential elections when double vacancies occurred, it would be exceedingly difficult to get Congress to pass a law providing for such elections. Consider first the obstacle created by constitutional expectations. By the time people get to be Senators and Representatives, they tend to have strong intuitions about how the system is supposed to work. To be sure, they may have no clue about crucial features of American government, like OIRA or cooperative federalism. Or NATO. But at the level of civics-book constitutionalism, the deeply held intuitions of normal American adults can be pretty robust. Given the confidence with which most of us take the four-year presidential term to be hard-wired, and perhaps also given fears that special elections are the sort of things that happen in countries with socialized medicine and multicolored currency, it might be extremely difficult to persuade Congress that providing for special elections is even on the table, no matter what the actual words of the Constitution might say and no matter what actually happened in 1792. (For good and for ill, settled constitutional expectations have a way of resisting arguments based on careful parsings of constitutional text or unfamiliar aspects of American history.)
Next, consider a problem created by constitutional structure. Passing a special-elections statute would mean repealing at least in part the currently governing law, the Presidential Succession Act of 1947. Under that statute, in the event that the Presidency and the Vice Presidency both become vacant, the Speaker of the House acts as President. In other words, the person with the most to lose if the current system is varied is the person who presides over the Lower House of Congress. Needless to say, it’s hard to pass major reform legislation without the Speaker of the House on board. And how motivated do we imagine the average Speaker would be to mobilize the political capital necessary to pass such legislation, given that one of the most direct consequences of such a reform would be, in the event of a double vacancy, to prevent the Speaker from serving as President for the full remainder of the existing four-year term?
The point goes beyond the personal incentives of the single most powerful Member of the House. It extends also to the incentives of the major political parties. The Speaker is always a leading figure in one of those parties—specifically, the party with majority control of the chamber. Under the current arrangement, the majority party in the House knows that in the event of a double vacancy, one of its own will move in to the Oval Office. Why would a majority party want to ditch that guarantee for the uncertainties of a special election, in which victory by a candidate from the other party is, speaking roughly, about a fifty-fifty shot? Or to come all the way back to our present reality, why would Congress as it is presently composed call an election in which some Democrat (Hillary Clinton or otherwise) might win the office, rather than giving the job to Paul Ryan and calling it a day? Not every Republican Member of Congress is a fan of Ryan’s, to be sure. But as between the guarantee of Ryan and the possibility of Clinton—or really any Democrat—it’s hard to imagine the Republicans’ thinking twice.
So yes, Congress likely has the legal authority to call a special election. If Congress impeached and removed the President and Vice President and passed a law calling an immediate election, there would be pretty good arguments against any attempt to have that legislation struck down in court. But there’s no end to the outlandish laws that Congress has the authority to pass, were it bent on doing so, and which no mature observer of the constitutional system should take at all seriously as actual possibilities. Congress is not going to enact a 99% income tax, though it would be entirely constitutional for Congress to do so. Nor, to offer a more analogous case, is the Ryan/McConnell Congress going to pass a law—with the concurrence of the relevant state legislatures, as specified in Article IV—merging the ten reddest states in the Union into a single state, thus reducing the membership of the Senate to 82 and throwing control of the chamber to the Democrats.
It’s no more likely to provide for special elections to replace Trump and Pence. For one thing, special elections go heavily against the grain of how people in Congress expect the system to work, and not even the most elegant theorizing and precise textual reading by pointy-headed law professors can be counted on to overcome the force of constitutional expectations. For another, there is no reason to think that the person who exercises the greatest agenda control in one-half of Congress, or his political party more generally, would want to hold a special election in the event of a double vacancy. The Speaker’s likely lack of enthusiasm for the idea has, for seventy years, been built into the incentives that come with occupying his office. The majority party’s attitude is just as guaranteed.
There’s also a third reason to think that legislation providing for a special election is exceedingly unlikely, even if Congress has the authority to pass it. Unless Congress actually thought it likely that Trump and Pence were going to be impeached and removed, there would be no occasion for exerting the considerable energy that would be required to draft and pass this kind of legislation. And people who now engage the special-election question as if it were actionable might be vastly overestimating the likelihood of a double impeachment. Yes, there are investigations, and yes, it’s likely that those investigations will find hands in inappropriate cookie jars. And yes, people have talked about a Trump impeachment since January 20—indeed, for longer than that. But no American President has ever been impeached by a House of Representatives controlled by his own political party, and at least to this point there’s little sign of Republican appetite to break new ground on that front. Not to put too fine a point on it, Trump and Pence might not be going anywhere.
For what it is worth, I suspect that at least some of the current conversation about special elections needs to be understood in light of the extant disjuncture between talk of impeachment among Trump’s opponents and the actual likelihood of impeachment by Congress. Ever since Trump emerged as a leading candidate in the Republican primaries, many people horrified at the idea of him as President have indulged a series of fantasies about his disappearance from the scene. November 2015: He’ll fade before New Hampshire. February 2016: He has a 30% ceiling of support within the Republican electorate. April 2016: There will be a brokered convention. August 2016: Trump never thought he could get this far, and maybe he doesn’t even really want to be President, and he’ll get out while he can. October 2016: Gosh, this Access Hollywood tape is awful—mainstream GOP officials will have to break with him, and he’ll crash completely. December 2016: The Electoral College was built to stop something just like this. February 2017: Now that he’s President and taking it on the chin every week, he’s going to stop having fun and resign.
The idea that Trump and Pence will be impeached and removed, and that Congress could then provide for a special election for a new President, strikes me as the latest iteration of this end-the-nightmare fantasy. It’s a way of imagining that something gone terribly wrong can be made right. We’ll all come to our senses. Our allies will forgive us, and the Russians will stop trying to hack into Western politics, and the white supremacists who have been energized will crawl back into their holes, and we can all either say we’ve learned from the experience or, better yet, pretend it never happened. Maybe Neil Gorsuch will resign and offer the seat to Merrick Garland. How could he do otherwise, really, given that everyone understands what a travesty this all was, and how important it is that we behave decently and move forward together?
Constitutional law is most valuable as a species of practical wisdom, not as a framework for fantasy. To be sure, thinking well about constitutional law requires the ability to design and work through thought experiments, and those thought experiments often traffic in far-fetched facts. But the most valuable thought experiments are the ones that shed light on actionable possibilities within the world of constitutional governance. If I were teaching students about the basic moves of constitutional interpretation, I could do it by reference to the question of the constitutionality of a statute providing for a special election, and the exercise might even be fun. What’s your initial intuition, and where does it come from? What argument are you tempted to make about historical practice, and did you bother to check whether the actual historical record resembles the one you assume, and how do you react to discovering that the history isn’t what you expected? Can you see how the text permits an answer that you intuitively think is wrong, and if so how do you reconcile that tension? Skilled constitutional analysts put themselves through these paces with respect to far-fetched scenarios in order to hone their abilities for analyzing real ones.
But they shouldn’t mistake the far-fetched for the real. And the present constitutional reality is this: Donald Trump is President of the United States. That’s a disastrously bad thing. I firmly believe that our constitutional system gives us the tools to mitigate the damage and eventually to set ourselves on a better course. But with a lot of sweat and pain and shoe leather, rather than with a magical act of expiation.
 Though it doesn’t quite say that. See Richard Primus, Unbundling Constitutionality, 80 U. Chi. L. Rev. 1079, 1103 & n59 (2013).
 Soccer’s World Cup—Men’s or Women’s—was not part of the consciousness of the present generation of the American decisionmaking class when it was coming of age and forming its worldview. And the Winter Olympics are nice, but if you think their being off-cycle is all that significant, you and I may simply stand on opposite sides of an unbridgeable conceptual chasm.
 Richard Primus, Constitutional Expectations, 109 Mich. L. Rev. 92 (2010)/
 Almost as if his being part of a congressional minority opposed to a pending bill tended to prompt him to argue that the bill was unconstitutional.
 A statute providing for earlier-than-usual elections might call for an election to fill the office until the date on which the previous President would have left office on the regular schedule, or, as the 1792 statute did, it might provide for a term of as long as four years. Given the 20th Amendment, a statute providing for a term that ran longer than four years from the prior President’s inauguration would either have to provide that the new President would take office on January 20th or else provide for a term that was not a whole number of years. The latter option seems textually permissible, if highly counterintuitive under current constitutional expectations (see Richard Primus, Unbundling Constitutionality, 80 U. Chi. L. Rev. 1079, 1103 & n59 (2013)). The former option seems permissible as well, though it might highlight the fact that Article II, Section 2, clause vi says that an Acting President serves until a President is elected, not until such a President is inaugurated. In other words, someone could make a textual argument that an Acting President must leave office as soon as the next President is elected (and that the portion of the current succession statute providing that the Acting President serves out the prior President’s complete term is unconstitutional to the extent that it keeps the Acting President in office between the day the Electoral College votes and January 20). But I doubt that any of these things would be insuperable obstacles in an environment in which Congress actually wanted to hold special elections, both because there are available legalistic workarounds and because of the strong and usually sensible tendency among constitutional interpreters to avoid making trouble by insisting on textual readings that are linguistically accurate but governmentally otiose. Not that everyone endorses that mode of behavior, of course.
 As Brian Kalt describes in Chapter Four of his book Constitutional Cliffhangers: A Legal Guide for Presidents and their Enemies, some smart people have worried that this aspect of the current arrangement is itself unconstitutional. The point of departure for these worries runs like this: Article II, Section 1, clause vi says that Congress’s power to provide for a statutory line of succession is the power to decide what “Officer” shall act as President, and the Speaker of the House is not an “Officer of the United States.” (“Officer of the United States” is a designation belonging only to members of the Executive Branch.) For what it is worth, I’m not worried, in part because I don’t think that “Officer” in clause vi needs to be read as if it said “Officer of the United States,” and the Constitution in Article I, Section 2, clause v clearly recognizes the Speaker as an “Officer.” To be sure, one could still find reasons to argue against the constitutionality of the arrangement, perhaps on originalist or structural grounds. But no argument of which I am currently aware has troubled me enough to make me doubt the validity of the now-settled expectation, easily reconcilable with the text, that the Speaker succeeds after the Vice President.
 There still remains the possibility of a Congress enacting a special-elections statute not because it will immediately face the double-vacancy situation but simply to better provide for government in the long term. In that scenario, the incentives I’ve described would not obtain. But Congress is not in the habit of enacting significant reform legislation in the abstract. It usually takes a crisis to motivate that sort of lawmaking. So except at a time when Congress is thinking about the prospect of a double vacancy in highly particular terms, it’s unlikely that a special-election proposal would rise to the top of the agenda.