The most obvious and benign possible reason we might see delayed results on Election Day is what political scientist Charles Stewart and I call the “overtime count.” This term refers to ballots that are counted after election night, as the election officials canvass returns before finally certifying their results. Before 2002, when Congress passed legislation in response to the 2000 Florida recount, ballots tallied during the “overtime count” period represented a relatively small percentage of total ballots cast and usually could be ignored on election night without worrying about the accuracy of “calling” the election results that night. Of course, valid “overtime” ballots would continue to be counted, in order not to disenfranchise eligible voters, but the winners and losers virtually never would change, and so the public largely ignored the official declaration of results a week or two later.

But as a result of electoral reforms passed since 2000—well-intentioned responses to real problems that needed fixing—our electoral system now relies much more heavily on the “overtime count” phenomenon. For instance, provisional ballots—ballots whose validity is uncertain at the time they are cast—are now required not to be counted on election night. A review of these ballots is necessary during the canvassing of returns to determine their eligibility. (Although the federal requirement to use provisional ballots is nationwide, states vary in the degree to which they employ them.)

We have also seen a rapid rise in reliance on absentee or mail-in ballots. This trend accelerated after long lines began presenting problems in presidential elections, which prompted states to encourage more voters to use mail-in ballots. In the old days, it was necessary to have an “excuse” to vote absentee—a business trip, hospitalization or something else that physically prevented you from going to the polls on Election Day. Now, in many states, absentee voting is a common option, and the choice is the voter’s. Some states permit absentee ballots to be counted as long as they are postmarked by or on Election Day; they can arrive at local election offices several days later and still be eligible for counting. Obviously, these absentee ballots cannot be part of election night returns, and as they grow in number, they increase the likelihood that close elections won’t be decided right away.

In the 2018 midterms, for the first time, the overtime count made a difference in nationally important races. In Arizona, it flipped the outcome in the Senate race in favor of Democrat Kyrsten Sinema. In Florida, the overtime count caused election night leads for Republican Senate and gubernatorial candidates to slip significantly during the canvassing of returns. Although the Republican candidates held on in both cases, they and their supporters grew nervous as they saw their leads diminishing. (Just as the overtime count is not yet well understood, the fact that it tends to work to the advantage of Democrats—a phenomenon now called “the blue shift,” based on some of my own research—is also perplexing to many, with some jumping too quickly to the conclusion that shenanigans are at work.)

The problem in Iowa’s 2020 caucuses was not caused by provisional or absentee ballots and, thus, is not technically an overtime count. But an event nonetheless occurred that prevented knowing the winner on election night. At least so far, officials have made clear that there were no improprieties or hacks. Instead, the results are taking longer than expected due to a glitch in an app that was developed to report precinct tallies, followed by phone lines getting overloaded once the app failed. It’s understandable that these unexpected delays would cause confusion, but all signs suggest that, thanks to paper records, the caucuses will still succeed in their fundamental purpose: accurately assigning delegates for the Democratic nominating process. It is unfair to label Iowa an apocalyptic failure just because of a delay.

What Iowa’s 2020 experience shows us is that Americans—in particular, the media—have developed unreasonable expectations about getting results on the night of an election, when our voting system is trending in the direction of longer vote-tallying periods. We have become accustomed to Iowa, as the first state to hold a primary contest, immediately setting a narrative heading into New Hampshire and subsequent primaries. But we should drop this pretense—not just for Iowa but for the general election, too. If we don’t know the winner of the presidential race on election night—which is quite likely—it is dangerous for the media to spin it, without evidence, as a sign that our elections have been tainted. It is also dangerous for new outlets to call the race before the results are verified—an understandable temptation, given the competitive pressure to be “first,” but one with adverse consequences to the public’s understanding.

What should the media do? Here’s one idea. On election night, most networks visibly display a “percentage of precincts reporting” number, which starts at zero and works its way toward 100. At that point, viewers are inclined to think 100 percent of ballots have been counted—which is often not true because of the lagging “overtime count.” The media does not explain this, causing confusion. But perhaps, between now and November, outlets could develop a new metric, such as “percentage of ballots counted” or “estimated percentage of ballots counted.” They could even publish an estimated number of ballots in the “overtime count” for each state (or each locality within the state). And they could report the number of provisional ballots cast, and make clear they haven’t yet been counted. Likewise, the media could report the number of absentee ballots that have yet to be counted on election night. All of this would reinforce that we shouldn’t expect election results to be finalized on Election Day.

Based on the reaction to what happened in Iowa, news anchors, cable commentators and reporters appear completely unready for the possibility of a prolonged reporting of the general election results. That’s what needs to change before November. It might be a tough addiction for media to kick; they seem to need election night results like a drug. But some form of rehab has to happen soon.