In the past 24 hours,* calls for Donald Trump to abandon his bid for the White House have swelled. While the odds of him doing so willingly still seem long, they are not negligible; who knows what they will be by early next week, after the second debate between Trump and Hillary Clinton. No matter the odds, the very real possibility that the Republican Party might replace its nominee for President of the United States less than one month before Election Day has invited reflection concerning a variety of issues, both political and legal.

Many others, including my Election Law @ Moritz colleague Ned Foley here, have already offered thoughts and analyses about what a Trump withdrawal at this point might mean to this year’s election. One issue that so far has received relatively less attention, however, is what this late-breaking development might say as a policy matter for future elections about the appropriate duration of the periods of early in-person voting and absentee voting. More specifically, the present moment helps to highlight the hazards of encouraging voters to cast their ballots too far in advance of Election Day. As described in the paragraphs below, decisions about how much pre-Election Day voting to allow need to balance its convenience with the problem of voters’ casting ballots on the basis of materially different information – including changes in candidates – across a span of time.

Each state makes its own decisions about how to conduct even federal elections, except that Congress has specified that federal elections are to occur on the first Tuesday after the first Monday in November of each even calendar year. For most of U.S. history, these federal elections were one-day affairs, conducted in-person on the specified day by requiring voters to present themselves at a designated polling location to cast their ballots, under rules established by each state. The exception, of course, was absentee voting, which began primarily as a method for members of the military to vote while serving in the armed forces. Absentee voting then slowly spread to other categories of voters who could claim some real impediment to voting on the designated Election Day, but still accounted for only a tiny fraction of total voter turnout.

However, over the past two decades, early voting has grown dramatically, as some three-dozen states have either opened their absentee voting process to all voters, or have developed options that allow voters to vote in-person before Election Day. It is now the case that as many as one third or more of the total votes cast in federal elections can be expected to be cast prior to Election Day. Unfortunately, most empirical research suggests that these opportunities to vote prior to Election Day have done little to increase voter turnout (except in some local, non-federal elections for which turnout is much lower to begin with), and instead seem largely just to have made voting more convenient for voters who would vote anyway even if the more convenient early voting options were not available.

The states with early in-person voting or open absentee voting have no uniform date when these options begin. Federal law now requires that absentee ballots for the relatively small number of military and overseas voters be available 45 days before Election Day, and some states choose to make absentee ballots for regular voters available at that same time. More typically, regular absentee voting begins 30 days before Election Day. Meanwhile, early in-person voting (which occurs at a county clerk’s office or other designated early voting center) usually begins later, with the average period of early voting beginning nineteen days before Election Day.

It is easy to understand why states should provide absentee ballots to military and overseas voters well in advance of Election Day, given the difficulties involved in transmitting voting materials to and from other nations, especially into and out of military theaters. It also is easy to understand why election jurisdictions might favor a uniform period of absentee voting for everyone, and therefore send absentee ballots to all voters once their military ballots are ready, 45 days in advance. Or, for jurisdictions that do not make that choice, it also is easy to understand why, once the ballots are ready, they might still feel pressure to get them into the hands of voters relatively soon, say 30 days in advance (the date when the voter registration window closes in many states). Voters themselves presumably want more rather than less time to cast their votes; the campaigns of candidates at the top of the ticket likewise view it as helpful to be able to begin harvesting – and thereby locking in – votes from their supporters sooner rather than later. Similar pressures can affect the choice of how soon to begin early in-person voting.

Yet the events of the past 24 hours simply have to make us think carefully about how these periods are set. Many absentee and early voting ballots have already been cast. But the voter convenience and administrative ease that these votes represent come with costs, including the impact on these voters of a subsequent Trump withdrawal. Were that to occur, it would mean that the act of casting an early ballot is effectively disenfranchising for any voter for whom a Trump withdrawal is material to their voting preference: the voter would have lost the opportunity for meaningful participation in the election. This reality leads some to argue that once early voting begins it may be too late for a political party to replace their nominee on the ballot.

One response (for instance here) is that all early voters choose to take this disenfranchisement risk when they decide to vote sooner than Election Day. But voters may do so without much knowledge of or reflection concerning this risk. Moreover, it is quite a different matter for the government to be needlessly permitting (even encouraging) this early-voting risk taking, if it produces relatively little benefit to voters and the election process generally. Early voting (whether absentee or in-person) undoubtedly has advantages. But extending the period of early voting further and further in advance of Election Day produces diminishing returns while exacerbating the potential that voting early will disenfranchise an early voter because of subsequent events or information important to the voter. This potential disenfranchisement could be reduced (though not eliminated) by shortening periods of absentee and in-person early voting that are longer than ideal.

Although empirical research suggests that most of those who take advantage of the opportunity to vote early are committed partisans who are the voters least likely to change their views in light of typical examples of late-breaking information, that fact does not alter the reality that sometime even these voters will encounter new information or new circumstances that do matter to their vote, including on occasion the replacement of a candidate. Moreover, if the early voters are the committed partisans, they will vote regardless of the length of the early voting period. Although it may be the highest profile such event in generations, the potential implosion of the Trump campaign is hardly the first time that an election has been thrown into chaos in its final weeks. In addition to the self-inflicted implosion of a candidate or campaign, elections can also be affected by late-breaking news about a candidate’s ties to business or industry, as with yesterday’s release of hacked Clinton campaign emails; by sudden new developments in international or domestic affairs, such as the outbreak of a new war; or by a candidate’s illness or death, as when three weeks before Election Day Mel Carnahan was killed in a plane crash while campaigning to be one of Missouri’s U.S. Senators, or when Senator Paul Wellstone of Minnesota similarly died in a plane crash 11 days before Election Day. The potential for these various types of disruption have received relatively little attention as states have expanded their windows for early voting.

One of the American Law Institute’s current projects is to develop Principles of Election Administration. (I serve as the Associate Reporter for this project.) Earlier this year, the ALI approved a portion of this project concerning early and absentee voting. Recognizing the sound arguments that may lead a state to offer some period of early voting, one of the ALI Principles is that for states that choose to offer early voting, “a uniform statewide period of early in-person voting should begin by the 10th calendar day before Election Day, and should continue daily through the second calendar day before Election Day.” Although the ALI has not taken a position on what the maximum period of early voting should be, a Comment accompanying this Principle notes that, because of the potential for late breaking developments, “A prolonged early-voting period therefore is undesirable (as also is a prolonged period of open absentee voting, which similarly encourages voters to cast their absentee ballots well before Election Day). At the same time, the early in-person voting period needs to be long enough to provide a critical mass of voters with a meaningful alternative to Election Day voting.”

Obviously, any decision to permit a meaningful period of early voting – at least 10 days, under the ALI Principle – means the possibility exists that some voters will have voted by the time some truly late-breaking event occurs. But that possibility doesn’t negate the importance of minimizing the risk by properly confining early voting periods. That is a balancing task that involves careful consideration of how much added benefit a marginal additional day of early voting provides, beyond some critical minimum period of 10 days, or two weeks, or whatever a state thoughtfully concludes. The current attention to the potential disruption that a Trump withdrawal would cause is an appropriate invitation for states to reconsider whether their absentee and in-person early voting periods are longer than they should be.

* This post was originally published on October 8, 2016, on the Election Law @ Moritz website.

Steven F. Huefner is the Alumni Society Designated Professor of Law at The Ohio State University Moritz College of Law and a Senior Fellow at Election Law @ Moritz. The views in this Comment are the author’s alone, and not those of either Election Law @ Moritz or the American Law Institute.