Election Security

Election Modifications to Avoid During the COVID-19 Pandemic

By Michael Morley
Friday, April 17, 2020, 4:06 PM

As we approach the presidential election this November, election officials are developing plans to deal with the unique risks posed by the novel coronavirus that causes COVID-19. I have written about how states have grappled with past election emergencies and am participating in a nonpartisan task force and interdisciplinary working groups to offer recommendations to ensure that election officials are adequately prepared to face the challenge before us. The recent crisis with the Wisconsin presidential primaries demonstrates the importance of states having election emergency statutes that adequately empower election officials to respond to unexpected crises, as well as contingency plans for implementing that discretion. Just as important as discussing the affirmative steps that officials should take to address the COVID-19 crisis, however, is identifying those they should avoid. Because so many groups and experts, along with my previous work, have focused on the first task, it’s time to tackle the second.

Election officials will have at least three major goals concerning the November election. First, most basically, they must ensure that every legally eligible voter has sufficient opportunity to cast a ballot without facing a substantially greater-than-usual risk to their health (under the circumstances, some minimally heightened risk may be inevitable in virtually any activity). Second, they must maintain strong protections against fraud, mistake and other irregularities that may impact the results. Third, election officials must conduct the election in a way that assures the public it was fair and the results are accurate. The Supreme Court has repeatedly held that, when it comes to the integrity of the electoral process and governmental corruption, the appearance of propriety can be just as important as actual fairness.

Elections are fundamentally bureaucratic endeavors. Up to 213 million voters of varying ages, who speak different languages, with different degrees of physical and mental ability, literacy, understanding of the electoral process and voting experience have the opportunity to cast ballots over a period of a few weeks. Largely part-time or even volunteer personnel with limited training and experience staffing thousands of temporary polling locations collectively determine tens of millions of voters’ eligibility; provide them with the correct ballots; and then collect, store and often transport those ballots or resulting electronic records for tabulation. Opportunities for mistakes, misunderstandings or accidents are pervasive. Substantial unanticipated changes to the rules and procedures governing the process—particularly last-minute changes—only magnify the likelihood that such problems will arise.

Various states have adopted dramatically different approaches to election emergencies. Some states allow election officials to make only limited adjustments to respond to emergencies, such as relocating polling places if they become inaccessible or switching to paper ballots if electronic voting machines fail. Other states grant officials wider discretion to adjust or modify a wider range of election-related requirements. Still other jurisdictions, in contrast, allow officials to temporarily postpone elections if unexpected emergencies prevent them from being carried out. Instead of, or in addition to, election-specific statutes, several states more broadly allow the governor to suspend state laws or statutory deadlines during declared states of emergency when necessary to protect human life. As mentioned above, numerous guides have been, and are being, developed to assist officials in determining how best to exercise these powers. When exercising this authority, developing contingency plans, or even crafting new election emergency statutes (such as the bill recently introduced by Sens. Ron Wyden and Amy Klobuchar), election officials and legislators should avoid adopting any of the following potential courses of action.

Automatically Mailing Absentee Ballots Based on Voter Registration Records

Absentee voting—also called vote-by-mail in some jurisdictions—will likely play a substantial role in the 2020 presidential election. During the 2016 election, nearly a quarter of ballots were cast absentee, though this proportion varied dramatically among states. One obvious way of responding to the pandemic that election officials may be tempted to consider, in states that do not have laws requiring it, is automatically mailing absentee ballots to each registered voter at their address of record. Such a proposal, while well intentioned, would create serious problems and substantially undermine public confidence in the electoral process.

Rather than mailing out actual absentee ballots, election officials can minimize potential difficulties by instead taking the initiative to proactively mail out absentee ballot request/application forms to each registered voter in the jurisdiction. By completing and returning the form, voters can confirm their identity and continued eligibility to vote in the jurisdiction and specify the address to which they wish their absentee ballot to be sent. Distributing absentee ballot request/application forms, rather than actual absentee ballots, is especially necessary in jurisdictions that do not have a history of universal vote-by-mail. Such jurisdictions lack the procedures and policies that states that rely primarily or exclusively on mail-based voting use to attempt to enhance the accuracy of their registration rolls and reduce absentee ballot fraud. The Carter-Baker Commission on Federal Election Reform, the Bauer-Ginsberg Presidential Commission on Election Administration and even progressive groups such as the Brennan Center for Justice acknowledge that, when election fraud occurs, it usually arises from absentee ballots.

The primary reason that election officials should automatically distribute absentee ballot request/application forms, rather than actual absentee ballots, is that many states’ voter registration databases are outdated or inaccurate. A 2012 study from the Pew Center on the States concluded that approximately 24 million records—one out of every eight—contain inaccurate or outdated information. A 2010 study from the Caltech/MIT Voting Technology Project similarly estimated that nearly 9 percent of voter registration records are invalid. Although many states have taken steps to improve their records’ accuracy over the past decade, problems with the validity of voter registration records continue to be identified. Voter rolls may be problematic for a variety of reasons. Most commonly, a person’s voter registration record may not have been cancelled after the individual moved, particularly to a different state or sometimes even county. In other cases, voters have died or been incarcerated for felonies.

Several jurisdictions with automatic voter registration systems have inadvertently registered ineligible noncitizens to vote. And numerous voter registration records are completely fraudulent—often as the result of third-party voter registration drives where participants are directly or indirectly paid based on the number of applications they submit. Both progressive and right-leaning groups have been caught engaging in systematic fraudulent registration schemes, including attempts to improperly register thousands of celebrities, sports figures or completely fabricated people. Attempts to update or correct voter registration rolls are often challenged in court as partisan attempts to “purge” voters. And federal law prohibits states from systematically correcting their voter registration rolls within three months of federal primary or general elections, which effectively limits such efforts to odd-numbered years.

Consequently, there are likely too many inaccurate, outdated or fraudulent voter registration records for election officials to rely on their voter registration databases as the basis for automatically mailing out absentee ballots. There are more than 213 million registered voters in the United States. If even 10 percent of voter registration records are problematic, that means more than 21 million absentee ballots in the presidential election would be sent to outdated addresses, erroneous addresses, nonexistent people or other problematic destinations. Moreover, many voters who have moved would likely request that replacement ballots be sent to their new addresses, leaving millions of duplicate ballots unaccounted for. Having tens of millions of absentee ballots outside of both election officials’ control and the hands of the voters who are supposed to be casting them raises a serious threat to the actual and perceived integrity of the electoral system. Sending an absentee ballot request/registration form to each potential voter instead helps ensure that ballots are sent only to actual voters, the information in those voters’ registration records is correct (or is updated based on the information they send back), and the correct ballots are sent to the right places.

In addition, even when voter registration records are accurate, voters may not be staying at their addresses of record. If stay-at-home orders remain in place, voters may be staying with family or friends for the duration of the quarantine. If the orders are lifted, voters may be away from home at school or for work. Particularly given the major economic and social dislocations to which COVID-19 has given rise, there may be an even greater need than usual to send voters’ absentee ballots to alternate addresses. Unilaterally mailing absentee ballots to voters’ addresses of record will likely result in millions, and perhaps tens of millions, of ballots being systematically sent to the wrong places. And, again, many voters who are away from home will request that replacement ballots be mailed to them at the addresses where they are staying. Thus, automatically mailing out millions of ballots to addresses where voters are not present will lead to millions of unaccounted-for ballots, which may facilitate ballot theft and can lead to inadvertent or intentional double voting.

Furthermore, a substantial fraction of voters who would receive absentee ballots do not wish to vote. Unilaterally distributing tens of millions of unwanted, unrequested absentee ballots would pose serious risks for the electoral process. The U.S. Election Assistance Commission reports that approximately 65 percent of registered voters cast ballots in the 2016 presidential election, meaning that over a third of the population did not participate. Undoubtedly, many registered voters who wished to vote were prevented from doing so by circumstances beyond their control. Nevertheless, a substantial portion of people who did not vote in 2016 inevitably did so as a deliberate, voluntary choice. A Knight Foundation survey of 12,000 persistent nonvoters found that the most common primary reasons why people decline to vote are that they don’t like the candidates, don’t know the candidates or issues, or believe their vote doesn’t matter.

Based on these statistics, even if we have record turnout for the 2020 election, it is reasonable to assume that a substantial number of voters will remain uninterested in participating. Automatically mailing absentee ballots to tens of millions people who do not wish to receive or cast them would irrevocably undermine the integrity of the electoral process. It would create a substantial risk of ballots being given away, sold, traded, collected by activists, or stolen from trash cans or mailboxes. Election officials would be able to do very little to prevent fraud or other criminal mishandling of so many ballots sent to uninterested recipients. These risks can be reduced greatly by simply sending out absentee ballot request/application forms instead.

Finally, sending absentee ballots (rather than application/request forms) to all voters without request will likely nudge a substantial fraction of the electorate to vote absentee when they otherwise would have voted in person. For elderly voters, immunocompromised voters and other particularly vulnerable populations, this may be the best or even only safe choice. For other voters, however, it is important to recognize that voting absentee—like all methods of voting—involves trade-offs. The U.S. Election Assistance Commission reports that, nationwide, nearly 0.75 percent of absentee ballots are rejected for reasons such as late return or the voter’s failure to sign the secrecy envelope. And even when a ballot isn’t rejected, it still may not be treated as a legally valid vote in particular races. The voter may inadvertently undervote or overvote by marking the ballot too lightly, insufficiently erasing an erroneous vote, accidentally making stray marks or marking the ballot in incorrect places. In most jurisdictions, voters who cast their ballots at polling places will be warned of such errors and given the opportunity to cast a new ballot. Election officials are available to help them if they are unable to correctly complete their ballots due to age, disability or other challenges. And absentee ballots unavoidably run the risk of being lost in transit or being misplaced by election officials or accidentally damaged. While absentee voting will be a critical component of our response to the pandemic, it comes with trade-offs of which election officials should be cautious.

In conclusion, election officials seeking to bolster participation in the presidential election should send out absentee ballot request/application forms, rather than actual absentee ballots, to each person on the voter registration rolls. This step is already well within election officials’ discretion in most jurisdictions, and it best balances the need to facilitate absentee voting due to the pandemic with ensuring the integrity of the electoral process.

Third-Party Ballot Harvesting

Another measure that election officials should reject is the use of third-party “designated persons”—frequently referred to as “ballot harvesters”—to collect absentee ballots from voters (except in jurisdictions where state law expressly authorizes their use). Any voter may return an absentee ballot by mail to election officials. Many jurisdictions allow a member of a voter’s family or household to pick up a blank absentee ballot for the voter and then return the completed ballot to election officials on the voter’s behalf. Third-party ballot harvesting involves members of a political campaign; political activists affiliated with outside groups; or, depending on the jurisdiction, even paid personnel going door-to-door to collect absentee ballots. This is especially concerning when third parties who are not related to the voter—and who may not even be known to the voter—are permitted to harvest unlimited numbers of ballots, frequently without having to identify themselves to election officials or note their identity on the ballots’ envelopes.

Election officials should not expand the use of third-party ballot harvesting, particularly as a response to the pandemic. As noted above, absentee ballots are particularly susceptible to election fraud, and many previous scandals—involving both Democratic and Republican operatives and supporters—have involved ballot harvesting. Ballot harvesting gives third parties who may be completely unknown to both the voter and election officials the opportunity to potentially tamper with absentee ballots. Harvesters may pressure voters into giving them blank ballots or casting their votes a certain way. When a person has voted for the “wrong” candidate, the harvester may surreptitiously change the vote, include additional votes to void the ballot, or simply dispose of the ballot rather than returning it. Third-party absentee ballot harvesting unnecessarily exacerbates the risks of absentee voting, undermining public confidence in the electoral system.

Moreover, allowing third-party ballot harvesting is particularly dangerous given the nature of COVID-19. Completely asymptomatic people may carry and be able to transmit the virus. Encouraging people to go door-to-door to collect absentee ballots, particularly from older or infirm voters, creates a tremendous unnecessary risk of spreading COVID-19 throughout some of our most vulnerable populations. Our nation’s primary method of slowing the spread of the virus has been to promote social distancing, frequently enforced by stay-at-home orders. Election officials should not undermine these strategies by incentivizing candidates, activists and third-party groups to go door-to-door to collect absentee ballots.

Signature-Match Requirements

Legislators or election officials may also consider the possibility of diluting or even waiving signature-match requirements for absentee ballots as a response to the pandemic. In many states, voters are required to sign the envelope used to return their completed absentee ballots. These jurisdictions confirm that the absentee ballot was actually received and completed by the voter entitled to cast it by comparing the voter’s signature on the envelope with his or her signature on file. Depending on the jurisdiction, the exemplar signature may be from the voter’s original voter registration application or absentee ballot request. Some jurisdictions go further, comparing the signature to an electronic database of all variations of the voter’s signature from various election-related forms submitted over the years.

In jurisdictions that rely on it, signature match is an important check to ensure the integrity of the electoral process. Election officials can neither exercise control over absentee ballots once they are mailed out to voters, nor ensure that they have been received and cast by the voters entitled to do so. Perhaps more importantly, signature match is an alternative to potentially even more burdensome requirements, such as mandating that voters’ ballots be witnessed or notarized or directing voters to include a copy of their identification card with the ballot. Particularly given the nature of the pandemic, signature-match requirements raise far less public health concerns than other ways of confirming a voter’s identity that require personal interaction.

Like almost any verification mechanism, signature match is not without problems. Election officials are not trained signature analysts, and an irreducible element of subjective discretion will inevitably exist. While some jurisdictions have promulgated detailed guides for determining whether signatures match, others treat it as a primarily holistic endeavor within most people’s ordinary experience, like determining whether someone’s face matches a photograph on an identification card. And people’s signatures frequently change over time. Some people—whether due to age, infirmity, physical handicap or another condition—are unable to reliably reproduce their signature. Most jurisdictions allow people to update their signatures at any time, however, even by mail. And if election officials send out absentee ballot request forms to each voter, the law of many states allows officials to compare the signature on the absentee ballot to the request form, ensuring they are comparing recent signatures. As discussed above, other states permit election officials to compare a signature on an absentee ballot to all of a voter’s signatures on file. Several jurisdictions go even further, notifying voters if there is a problem with their signature and offering them an opportunity to provide a replacement. In short, many states’ enhanced reliance on absentee voting as a response to the pandemic only underscores the need to enforce reasonable measures to ensure that absentee ballots were cast by the voters entitled to do so and to prevent mistakes, fraud and irregularity.

Internet-Based Voting

Another tempting response to election emergencies is for states or election officials to use the internet to send ballots to voters and for voters to return their completed ballots the same way. Officials in New Jersey purported to authorize such measures in the wake of Superstorm Sandy; a county supervisor of elections in Florida did so after Hurricane Matthew. Both decisions raised serious legal, technological and administrative questions. Although federal law authorized a pilot program to explore internet voting for military and overseas voters, current technology is far too insecure to consider internet voting for domestic voters, particularly in the context of a presidential election that is subject to a substantial risk of foreign interference. The National Academies of Sciences, Medicine, and Engineering concluded in 2018 that “the risks associated with Internet voting are more significant than the benefits. Secure Internet voting will likely not be feasible in the near future.” It later elaborated that “Internet voting should not be used in the future until and unless very robust guarantees of security and verifiability are developed and in place, as no known technology guarantees the secrecy, security, and verifiability of a marked ballot transmitted over the Internet.”

The Association for Computing Machinery similarly cautioned about the “insoluble security problems that are inherent to casting ballots online, including server penetration attacks, client-device malware, attacks to emailed and faxed ballots in transit, denial-of-service attacks, disruption attacks and the challenge to reliably authenticate voters.” Other groups such as the National Institute of Standards and Technology and the Verified Voting Foundation have echoed these concerns.

Most experts’ skepticism has generally focused on the use of the internet to transmit voted ballots. With the serious threat of hacking backed by foreign militaries or intelligence services, however, allowing blank absentee ballots to be downloaded or transmitted over the internet raises similarly serious concerns. Given our experience with actual and attempted cyberintrusions during the 2016 election cycle, we should not further increase our vulnerability by expanding the use of the internet for the transmission of blank or completed ballots—particularly as a hurried way of attempting to mitigate the impact of an election emergency.


When a natural disaster or terrorist attack directly impacts an ongoing or impending election, election officials may have no choice but to postpone or even cancel it. For most types of elections, such postponements or cancellations have serious, but limited, consequences. In the context of a presidential election, in contrast, even a short postponement would raise critical issues and likely have destabilizing effects far beyond the state that made the decision to postpone. As I’ve argued elsewhere, the Supreme Court’s recent ruling in Republican National Committee v. Democratic National Committee, arising out of litigation over the Wisconsin primaries, is best understood as recognizing that a federal court must have an especially compelling justification to order an election postponement as a remedy in an election emergency case. At a minimum, other types of modifications to the election’s rules must be insufficient to ameliorate the threat to, or burden on, constitutionally protected voting rights.

Federal law requires states to choose their presidential electors on Election Day (that is, the Tuesday following the first Monday in November; in the present case, Nov. 3, 2020). It further specifies, however, that if a state has held an election that day but “failed to make a choice,” it may appoint its electors on a subsequent day. This provision gives states flexibility to postpone or extend a presidential election in the extreme case where a severe, unexpected emergency makes it impossible or unreasonably dangerous to complete balloting on Election Day itself.

A state’s ability to postpone a presidential election is quite limited, however. Federal law requires electors to cast their electoral votes on a specified day in mid-December (Dec. 14, 2020). And the Constitution specifies that all electors throughout the nation must cast their votes on the same day. Thus, any postponed election would have to be held far enough in advance of that date to allow ballots to be counted, provisional ballots to be processed, and, potentially, any recount or election contests to be resolved. A substantial delay in a particular state would either preclude other states’ electors from casting their electoral votes or raise serious questions about whether that state’s electors would be permitted to cast belated votes.

Federal law further provides that the chambers of Congress must meet in joint session on Jan. 6 to count electoral votes. Electors who have not been selected by this date will not have their votes counted. And, of course, the Constitution states that the president’s term expires on Jan. 20. If the chambers of Congress do not resolve a presidential election by that date, the offices of president and vice president become vacant, resulting in appointment of the speaker of the U.S. House of Representatives as acting president under the Presidential Succession Act. Having such an interim figurehead would likely jeopardize national security and America’s standing in the world and inflame partisan tensions. In short, while states have a very narrow window during which they may postpone a presidential election, they should do so only in the most extreme circumstances, if carrying out voting on Election Day becomes impracticable or impossible due to an unexpected disaster or crisis. In the extremely unlikely event a state is forced to exercise that authority, it must complete its election at the earliest possible opportunity—within the next two weeks, if at all possible—to avoid impacting the Electoral College.

100 Percent Mail-Based Voting

Finally, some commentators have argued that, to avoid exacerbating the public health risks posed by COVID-19, the November election should be conducted entirely by mail—an option some states have adopted for elections this spring. A few advocates have gone even further, arguing that our experience with COVID-19 demonstrates that all elections should be 100 percent mail-based. These suggestions are fundamentally misguided. As the events of the early 21st century alone demonstrate, our election system is vulnerable to interruption by a wide range of disasters, including hurricanes, public health crises, and terrorist attacks, to say nothing of earthquakes, floods, nuclear accidents, cyberwarfare, widespread power outages, strikes and blizzards. To be sufficiently robust to withstand such a sweeping range of potential threats, an election system must incorporate multiple distinct vehicles for voting. Having numerous options—potentially including in-person voting on Election Day, a limited period of early voting, absentee voting, curbside voting, and facilitated voting for people confined to medical or assisted-living institutions—ensures the electoral system is flexible enough to withstand nearly any threat. Adopting only a single method of voting, in contrast, subjects the system to tremendous systemic risk. A 100 percent mail-based system, for example, could be disrupted by anthrax attacks, a national postal strike, the U.S. Postal Service’s financial instability or destruction of central mail-sorting facilities. Neither the upcoming November election nor future elections should rely solely on a single channel for voting. A diverse electoral system is a robust system.


State election emergency laws often give election officials substantial discretion to modify the rules governing the state’s electoral process. General state-of-emergency laws often give the governor even broader authority. When legislatures draft such laws or executive officials develop plans for implementing them, they must identify not only the most effective courses of action to take—an issue my previous work discusses—but also particularly problematic alternatives that should generally be avoided. Knowing what not to do is an important, yet largely overlooked, aspect of contingency planning. Measures such as automatically sending absentee ballots (rather than absentee ballot request/application forms) to every person in a state’s voter registration database, permitting third-party ballot harvesting, diluting or eliminating signature-match requirements, transmitting blank or completed absentee ballots over the internet, and postponing or extending the election unless absolutely necessary should be rejected.