I appreciate the sentiment behind a lot of the media narrative over the last couple of weeks—and the much broader effort of the election law and political science communities long before that—about the lack of certainty that we may have on “Election Night.” That said, I do think there’s a risk of overstating the claim, or of conflating a few ways of thinking about the claim, which I wanted to parse out.

First, of course, formally, the election isn’t over until it’s over. But that’s actually very late, depending on your view of when it’s “over.” A presidential election isn’t really “over” until Congress counts the votes (this year, January 6); a congressional election isn’t really “over” until Congress seats that member (this year, January 3), as it has the authority to judge the elections and returns of its own members.

But stepping back a bit, a state “finalizes” a result when it certifies an election result, usually by a set deadline in late November, but it could be earlier in some states or stretch into December in others. That’s a pretty good deadline from the state’s view, at least, that the election is over.

Stepping back yet again, however, we do see states offering various stages of counting. Counties may canvass the results by mid-November, and that canvass is usually enough to trigger the opportunity for a recount. It’s pretty “final” for that reason—someone appears to have lost and has the right to challenge that result through a recount; or, the margin is so close an automatic recount is triggered.

And stepping back to Election Night itself, there’s a canvass that night that provides a pretty clear picture of the results. “Unofficial,” sure. But it’s also the foundation for the later canvasses.

Of course, media outlets can “project” a winner, regardless of how many votes are in or not. They’ve improved dramatically since 2000, when they had problems in Florida not just in the projection but in “calling” states before all the polls in the state had closed. Since then, they’ve been much more careful about projecting results and have become much more sophisticated in their modeling.

We also know that “conceding” an election is a legally meaningless act. Al Gore “conceded” in 2000 to George W. Bush, allowing Mr. Bush to claim victory, only to retract it later. Stacey Abrams still refuses to concede her 2018 Georgia gubernatorial loss, to no legal effect.

“Claiming victory” is legally meaningless, as, again, Mr. Bush experienced in 2000. That can even happen without official results—think about how both Pete Buttigieg and Bernie Sanders claimed victory after the 2016 Iowa caucuses, well before any “official” results were released.

Both, of course, conceding and claiming victory can have popular reactions or responses. And maybe two presidential candidates simultaneously claiming victory leads to greater public perception problems.

Now, of course, we have several complications about Election Night this year.

The first is the sheer volume of absentee and vote-by-mail ballots. While most jurisdictions will allow some pre-processing of those ballots to allow them to be counted quickly on Election Night, some (like Pennsylvania) won’t until later. That volume—coupled with perhaps distorted political preferences between those voting by mail and those voting in person Election Day—could undermine the certainty of an Election Night conclusion.

The second is the dramatic increase in jurisdictions accepting ballots received after Election Day as long as they are postmarked by Election Day (or, in Pennsylvania’s case, those without a postmark at all), and the longer period of time in many jurisdictions. With the increase in absentee voting (but, it appears many are submitting their ballots early!), we might see a larger wave of such ballots, which means we don’t even know how many ballots we have on Election Night. We might know the delta of how many ballots have not yet been returned, an idea about the maximum number of outstanding ballots. And depending on the contest, it might not be enough to matter much.

The third is an anticipation of post-Election Day litigation. If the claim is that there is increased election uncertainty because of litigation, it’s a reason to distrust even “unofficial” results Election Night. It’s particularly true in states like Pennsylvania and Minnesota with segregated ballots received after Election Day; or in any recount, contests and changes to vote totals can occur.

I’m somewhat concerned about “normalizing” these last two narratives. I don’t know it’s great to think about ballots trickling in several days after the election—I know that there are good reasons, especially for military and overseas voters, to have some late-arriving deadlines; and I accept that, in a pandemic, we may want some concessions due to high volume of mail. But I prefer some finality of ballots received by Election Day, in that we know what the vote totals are. And maybe that means some better and earlier voting opportunities to get those ballots out. I also don’t think it’s great to expect or anticipate litigation as something to change outcomes.

Finally, it’s worth noting that we do sometimes have lightning-fast certification. Consider Jason Smith of Missouri, who was sworn into Congress less than 18 hours after the polls closed in 2013. True, Missouri had an unusually fast certification window for special elections.

In short, I think a lot of the framing is right—we should not have an expectation of finality of an election on Election Night. That said, we’ll still know a lot about a lot of elections, including many if not most slates of presidential electors. But, I think it overstates that many times we know, in a less formal sense, about the winners; that this election does, admittedly, present some greater uncertainties; but, I hope, we can seek out greater certainty and finality closer to Election Night in future years. We should count all the votes, and we should know that takes time. But we should also know that we often “know” the winners relatively quickly.