After a lot of digging around for information, I wrote up a quick primer for New York on the problems the usual sources of authoritative information will face in “calling” the 2020 election results, particularly on Election Night:

You are undoubtedly aware that Election Night on November 3 could be unlike any other we’ve seen before. Heavy, heavy pandemic-driven voting by mail; laws in some states allowing mail ballots postmarked by Election Day to be counted afterwards; problems (legal or logistical) many states will have in efficiently processing a glut of mail ballots could produce a much slower count nationally than is normal. But at the same time, a big partisan split in how and when voters choose to vote could produce some herky-jerky results. Democrats fear that Republicans’ preference to vote in person on Election Day will create an ephemeral lead for Donald Trump that he will use to prematurely declare victory.

Americans are accustomed to being told who has won major elections quickly. With the exception of the infamous 2000 contest, we’ve known the winner of every presidential election in this and the last century by the day after Election Day. In the post-1960 era of exit polling and media network competition to “call” races, we usually know much earlier than that, sometimes moments after polls close in enough states where the overall “winner” can claim the necessary 270 electoral votes to be president.

So how will the system work in this insane election cycle? Here, some key issues to think about.

Who “Calls” Elections, Anyway?

We generally think of the authenticators of presidential, congressional, and statewide election results as the major television-broadcast and cable networks (ABC, CBS, NBC, CNN, and Fox) along with independent media services, preeminently the Associated Press, now supplemented by outfits emphasizing speedy results like Decision Desk HQ. All these authorities employ “decision desks” of election wizards, typically locked away in deliberate isolation from their employers and other potential sources of influence, who utilize carefully developed “models” to make projections (or, in the AP’s case, “declarations”) of who will ultimately win each contest. For presidential elections, of course, these “calls” are state by state (and also district by district, in the case of Maine and Nebraska, which let each congressional district cast one electoral vote) to reflect the reality that electoral votes determine the overall winner. None of these worthies will “call” the presidential race until they have “called” enough states to give the winner the requisite 270 electoral votes.

In addition to the “official” arbiters of the results, other media organizations will be collecting data for their own “unofficial” analysis, some of it quite influential, as anyone who associates Election Night 2016 with the infamous New York Times “needle” can tell you.

Anyone, of course, with access to a camera, a microphone, or a social-media account can “announce” an election result, real or imagined, and it is important to keep in mind that what we think of as “official” authenticators of the results are not necessarily authoritative to everyone. Let’s go ahead and look at the “red mirage” scenario, in which Donald Trump claims victory at midnight on November 3 based on sure-to-be-reversed leads in very partial results. It’s extremely unlikely that any of the television networks or the AP would verify such a claim. But whatever Fox News’ official “decision desk” decides to do or not do (and the network has gone out of its way to insist upon its independence), it’s likely that highly influential Fox News pundits would echo Trump’s claims anyway, along with vast armies of social-media warriors and Russian bots.

So it’s possible that “official” and “unofficial” calls of the election will compete for attention, with the unwillingness of the “official” outlets to make pronouncements based on incomplete data perhaps undermining their authority.

How Can You Do Exit Polling in a Year Like This One?

Back in the day, when voting by mail was rare and exceptional, Election Night “calls” depended on models that verified exit-poll data of people who had cast ballots in person with raw votes from key precincts, making quick decisions possible in all but really close races. Since 2004, exit polls have been conducted by Edison Media Research, backed by a consortium of media companies (currently ABC, NBC, CBS, and CNN).

Exit polls have a good recent record of reliability, but there have been serious problems historically. In 2002, the whole system broke down and no exit-poll data was released, and in 2004, exit polls significantly overestimated Democratic voting, leading John Kerry’s staff to prematurely celebrate a victory he never actually won.

The steady growth of voting by mail has been a particular problem for exit-poll-based decision models. That’s one reason the AP and Fox News withdrew from the Edison consortium in 2018, deciding to rely entirely on the preelection polling everyone uses to get a handle on those who vote by mail. With mail ballots spiking this year, and with the partisan skew in voting methodologies, the particular value of exit polls (often used to interpret elections results as well as to project them) may decline further. Often the major media organizations will publish exit-poll findings (not the horse-race numbers but highly suggestive answers to questions other than candidate preference) even before polls close, but this year it may be necessary to take them with a shaker of salt.

Aside from the doubts the pandemic has sowed in the adequacy of exit-poll data, it has affected the mechanics, too. Here’s a description from ABC News of how exit interviews will work this year:

“Some exit polling procedures have been modified this year to help ensure a safe experience for the interviewers and voters. Typically, the exit poll interviewer walks right up to the selected voter and hands them the exit poll questionnaire and a pen. This approach has been modified.

“This year, the interviewer, who will be wearing a mask at all times, will approach the voter from a distance of at least 6 feet. Voters who agree to fill out the exit poll will be directed to a nearby table to get the questionnaire and a single use golf pencil.”

While ABC assures us that “these procedures were tested successfully during the recent primaries,” you do have to wonder if they will affect interview participation patterns in a general election where groups of voters across the country may react differently to the risks involved.

When Is the Presidential Race Likely to Be Called?

Again, the “official” arbiters of election results have, since at least 1980, followed the practice of refraining from any “call” of the presidential election until individual state “calls” have been made awarding one candidate or the other 270 electoral votes. In a highly informative interview with FiveThirtyEight, ABC’s Dan Merkle admits that the virtual certainty of slow counts in Michigan, Pennsylvania, and Wisconsin makes it very unlikely that a winner will be indicated before November 4, if not later.

This is an important point to keep in mind. There has been a lot of talk about results from faster-counting states such as Arizona, Florida, and North Carolina telling us what we need to know about the ultimate outcome. It’s true that if Joe Biden wins all three on Election Night, the odds of him ultimately winning are prohibitively high. But if Biden hasn’t nailed down 270 electoral votes, he won’t be officially “called” the winner, even if the pundits are all concluding Trump is done.

The time gap between the first “takes” on Election Night on the results and their ultimate authentication is a window for great mischief, particularly on the part of a president who has repeatedly said he cannot lose unless the results are “rigged.” And if the race really is very close, with the results legitimately in doubt, Americans may relive the twilight experience of 2000, when every morning we woke up to wonder if we had dreamed Election Night had never quite ended.