Media Briefings

Voting in 2020: logistics, safety, and ballot integrity

This media briefing was co-hosted by the AAAS Center for Scientific Evidence in Public Issues (EPI Center.)

Journalists: Get Email Updates

What are Media Briefings?

Experts have warned that a confluence of external factors and events will make the 2020-election voting process especially complicated. At SciLine’s media briefing, co-hosted by the AAAS Center for Scientific Evidence in Public Issues (EPI Center), three experts shared research-based insights about: the logistics of in-person voting during the COVID-19 pandemic; the potential impacts of a surge in mail-in ballots; and the related challenges of ballot auditing and integrity.

Journalists: video free for use in your stories

High definition (mp4, 1280x720)




RICK WEISS: Hi, everyone, and thank you for joining us. For those of you who may not be familiar with us, SciLine is a free service for reporters and scientists funded entirely by philanthropies and based at the American Association for the Advancement of Science, or AAAS. Our singular mission is to help get more research-backed evidence into news stories. We do that through an array of services, including a matching service that can connect you to a knowledgeable and articulate scientist for your story on deadline or as needed, as well as briefings like this one and an array of other offerings that I encourage you to check out on our website, Today’s briefing is co-hosted by a sister program at AAAS, the Center for Scientific Evidence in Public Issues, better known as EPI Center, which is led by my colleague Michael Fernandez. Michael, can you just take a minute to introduce EPI Center?


MICHAEL FERNANDEZ: Hello, everyone. And thanks again, Rick, for giving us the opportunity to co-host this briefing. Our mission at the EPI Center is to bring clear, concise and actionable scientific evidence to decision-makers on a wide variety of public issues. Like SciLine, we are committed to the value of evidence in public life. Our goal is to make it easier for policymakers to access and incorporate scientific evidence into their decision-making processes. One of our first focus areas has been on voting technology and elections security, and we’ve engaged with hundreds of election officials at the national, state and, most importantly, local level. So I encourage you to visit our webpages on the AAAS site for more information and useful resources. You’ll also find information there or by following us on Twitter at @AAASEPICenter on other public issues, including projects on fracking, PFAS chemicals, green infrastructure, racism and policing. So thanks again for joining us today, and I’ll hand it back to Rick so we can get started.


RICK WEISS: Thanks, Michael. And I want to note that near the end of this briefing, I’m going to ask EPI Center’s Steve Newell to add a few comments on what we’ve all heard. Steve’s been directing EPI Center’s year-and-a-half-long project on voting security and will be able to add some perspective based on what we’ve all heard from our speakers today. That should help you in very practical ways as you prepare to cover this topic in the weeks ahead. So we know that because of COVID-19 and other factors, this fall’s elections are going to run up against a number of technical, logistical and behavioral challenges, which means this year’s election, to us, is very much a science story, from the engineering of voting security to the social science of public trust in results. Just one example before we get started – you know, a poll last week revealed that 46% of registered voters say they’re uncomfortable going to a polling place this fall, and 6 in 10 would prefer to vote before Election Day, but only 3 in 10 say they’re very confident that if they did vote by mail or by some alternative means, their ballot would be counted accurately. To address the interrelated issues of voting safety, logistics and accuracy, we have three excellent expert sources with us today who will make brief presentations and then take your questions. I’m not going to take time to do full introductions for them. Their information is on our website.

But we will hear first from Dr. Bridgett King, who is associate professor in the Department of Political Science at Auburn University. We’ll focus on what the science says about the prospects for safe in-person voting during COVID-19. Second, we’ll hear from Dr. Nate Persily, a professor of law at Stanford Law School who will talk about mail-in ballots and related issues of election administration. And then third, we’ll hear from Dr. Alex Halderman, professor of computer science and engineering and director of the Center for Computer Security and Society at the University of Michigan, who will present on voting security, including some discussion of what to look for as reporters as you track the official audits of election results in November. So with that, let’s get started. I’m going to turn it over to you, Bridgett, to start.


Voting in Person during COVID-19


BRIDGETT KING: Thank you. Let me just pull up my slides real quick. OK. So to Rick’s point, one of the things we know is that the quality of the in-person voting experience is one of the measures that the public uses to determine how much confidence or trust they can have in the system and the overall process. There’s a variety of research that supports this claim, including my own, and that those in-person experiences can affect confidence at the state, local and federal level. In light of the 2020 election, and particularly COVID-19, what voters experience who choose to go to vote in person will be vastly unique and different from what they’ve previously experienced. And while there is a push for vote by mail, as Rick’s statistics point to, there is a considerable lack of trust with participating that way. So what we will more than likely see is voters who choose to still vote in person. So one thing that’s important for them to understand is the changes that are going to be made at polling locations and why their voting experience this particular year might look a little different.

So what I would like to highlight is some of the research that’s been done by URI VOTES, also a collaboration with myself at Auburn University and Dr. Jennifer Lather at the University of Nebraska-Lincoln, led by Dr. Gretchen Macht at the University of Rhode Island. So when thinking about what the interior of a polling location might look like, particularly in light of COVID-19, one consideration that has to be made is where people are standing. And as Rick noted, there’s concerns about being comfortable about being in physical spaces to cast a ballot. So one thing that can be considered when thinking about the layout and what voters might see is equipment that is adjusted for physical distancing. So what is on the slide is a 2-foot-by-2-foot consideration for individual voting space, but then an additional 6 feet from that specific voting space or an individual voter to make adjustments for physical distancing for voting in person. So in addition to considering, you know, where a voter is actually physically standing when they arrive to vote, whether that be in front of a ballot-marking device, in a voting booth – excuse me – at check-in or at an optical scanner, there is the need to, beyond that physical space, add an additional 6 feet while making sure that that 6 feet does not overlay too far with another person’s 6 feet.

So – and you can see the slide (ph), the red square is the voter space, and then the light red circle is the social distance ring. Beyond that, though, because you want to create an environment where there is clear direction with how voters are to enter and exit the system, another thing we’ve been considering in our conversations with local election officials over the summer is the establishment of a 3-foot unidirectional path so that you can effectively direct voters in and direct them out, similar to the arrows you might experience when you go vote at your local grocery store – not vote at your local grocery store. But if you go to your local grocery store, there might be arrows on the floor that suggest you should go in this way and out this way – basically to ensure that voters, in this case, wouldn’t necessarily be running into each other.


RICK WEISS: And, Bridgett, I’m going to interrupt just quickly. If you can move your mic away so it’s not, I think, touching your blouse. It’s making a little bit of noise. Great.

BRIDGETT KING: I’ll just hold it in my hand. Sorry about that.

RICK WEISS: All right.


BRIDGETT KING: OK. So it’s important to note that most voters vote in mid- to small-sized spaces, so all voters won’t necessarily have access to extra-large polling locations. There’s been a lot of attention given to using sports arenas. So here’s just – I’m just going to go through two quick examples. So in polling locations, considering the physical distancing requirements necessary for COVID, there may be reductions in the distribution of equipment in different polling locations. But it’s also true that if you adjust the layout, you can also have the same amount of equipment into locations that differ dramatically in terms of the actual square footage. And so the slide presented here is just an example of that. So in this scenario, you have eight voting booths, two check-in stations, one accessible ballot-marking device, a ballot box or an optical scanner, a sanitation area, two chairs for observers and, obviously, the needs for social distancing while including that 2-foot-by-2-foot voter space.

So the space on the left is a little over 4,000 square feet, and the space on the right is a little under 3,000 square feet. But what you can see is based on the way – or the layout decisions that you make in a given polling location, you might be able to actually include the same amount of equipment while keeping voters safe even though there are dramatic differences in the actual square footage of the space. It’s worth noting that these images do not include the use of plexiglass or any sort of partition. So they are specifically focused on how using the equipment that you have and taking into consideration a unidirectional path, how you can lay out polling locations of different sizes while taking voter safety into consideration. Here is just one more example. So it’s also true while that is true, in some instances while accommodating for physical distancing and the unidirectional path, you may actually have to reduce the amount of equipment. And here’s an example of that. So this is a library. It is a mid-sized polling location. And what we see here is under traditional layout where we don’t necessarily have the same considerations – or same specific considerations for space between voters and space between voters plus equipment, you have five check-in stations and 20 ballot-marking devices.

As you can see on the right, once you include the layout considerations for physical distance, you see a reduction in equipment from five check-in stations to two and then a reduction in ballot-marking devices from 20 to five. And so one thing that I think it’s important to communicate to voters is that these changes are being made in an effort to promote safety – or the specific safety of voters who, for a variety of reasons, choose to vote in person and that there is no – because no two polling locations are the same, what that equipment distribution from one polling location to the next, as experienced by individual voters, is going to look different this fall. So with that, I will stop.


RICK WEISS: Thank you, Bridgett. And obvious implications there for wait times and lines, which we can get into in the Q&A. I’ll turn now to Nate Persily. Oh, you’re on mute, Nate. You’re on.

Voting by Mail during COVID-19


NATE PERSILY: So thanks for having me. I should point out, first of all, I’m sitting next to my own Palm Beach butterfly ballot machine, which was the last election controversy at the presidential level that we all experienced. And while there are lots of concerns about in-person voting this time, and as Alex will talk about the machines themselves, most of the legal wrangling is focusing on absentee and mail balloting, so I’m going to focus my comments on that. The issue for this election – right? – is that we are moving tens of millions of voters to a new form of voting that they are not familiar with. Whether it’s, as Bridgett said, the new types of polling places or whether it’s mail balloting, we have never had to transform the American electoral infrastructure in such a significant way in such a short period of time. And every state, with the exception maybe of the all-vote-by-mail states – every other state is going to have some combination of voting by mail and retrofitting these polling places to ensure social distance.

This is not as easy as people think. We get a lot of commentary of people saying, well, why don’t we just mail everyone a ballot? Well, it’s not that easy just to, on a dime, transform an electoral system to vote by mail. The states that have had to do it have done it over a generation. And so we’re asking some states to do this over a four-month period. And I want to emphasize that this is a continuum, not a choice – not a dichotomous choice between two things – mail on the one hand and in-person voting on the other – because there are all kinds of choices in the middle, such as dropping off your absentee ballot in a drop box or at a polling place or at an election office. We even have the – we’ve developed the oxymoronic title of in-person absentee balloting, which has got a sort of existential feel to it, where people can go into a election facility and vote an absentee ballot and request it at the same time. So the constraints that jurisdictions are facing here involve, as in all things, time and money.

But at this stage, money, while scarce, is not as scarce as time. The election, as you know, is already underway. The ballots are out in the hands of the voters in places like North Carolina. And in the next week, in about six, seven other states, they’ll be receiving ballots. And we need more time to vote, more time to cast our ballots and more time to – for election officials to process and count them. There is a shortage of people and places and things – a shortage of poll workers for polling places, a shortage of polling places. And, as well, we need all kinds of things in the election administration pipeline, whether it’s PPE for poll workers or other kinds of equipment that deal with mail balloting. But the extremely decentralized nature of American elections, the sort of outdated contracting and procurement process that plays the election infrastructure like it does, the other areas of the economy dealing with COVID and the ill-fitted laws and regulations which are not prepared, in some states, for a massive transition to mail balloting all get in the way. And, of course, this is in an environment of extreme partisanship. When we had Bush versus Gore – right? – the controversy over the Florida recount, it was in an environment of relative partisan harmony. Right now, skin is very thin, and it’s not as clear that someone who loses, say, a fight over absentee ballots will go – that that candidate will go quietly.

And then finally, there’s just the unpredictability that we’re facing with the virus itself. How bad will it be over the next few months? How many people will vote by mail? How many will vote in person? Now, over the last, you know, 20 years, we’ve seen a rising trend toward absentee balloting and early voting. We should expect roughly half, I think, of votes this election to be cast by mail. And when we say by mail, we don’t mean necessarily through the post office, but at least half of the votes will be cast absentee – might be a bit closer to two-thirds, but probably, you know, between 50% and 60%. But that’s not going to be uniform across the country. You’ve got the all-vote-by-mail states of Oregon, Washington, Utah, Colorado and Hawaii. They’ve now been joined by Nevada and – by Nevada, California and Vermont and New Jersey that have now, in one way or another, moved toward mailing ballots to everyone. But like I said, some places, like California, will have polling places as well. But most of the battleground states will have no-excuse absentee balloting. And while there’s been lots of controversy about mail balloting, principally as delivered by the president’s Twitter feed, to some extent, the debate over mail balloting is really over because, as the president said and endorsed the mail balloting – the system of mail balloting they have in Florida, which is absentee balloting, that’s really the system that’s going to be taking place in all of the battleground states. But different states are better or less prepared to deal with mail balloting.

So if you look at some of the battleground states, like North Carolina, Pennsylvania and Wisconsin that I’ve highlighted here with arrows, they have absentee ballot rates historically under 6% – right? – whereas Arizona has rates over 75%. And so for Arizona, they are better situated to deal with this rush to mail balloting than some of these other battleground states. So what are the challenges to voting by mail? We’ve spoken about fraud and perception of fraud. The president has certainly rang that bell several times. And we are noticing, you know, considerable partisan division on the issue of fraud and mail balloting. And as pernicious as actual fraud may be, which is quite low when it comes to, you know, the actual incidences of fraud in mail balloting, that there is this growing perception of fraud which itself casts doubt on the integrity of the election and the validity of the result. But just as important is to think about issues of disenfranchisement and bias in the absentee voting system. Because, you know, traditionally, we’ve had white, educated old folks have dominated the absentee ballot pool and that people who voted in person were more – racial minorities were more likely to vote in person, and for, you know, younger voters and the like.

What we’re seeing now is that there are sizable numbers of voters who are reluctant to change, that many are, as Bridgett said, are going to be voting in the polling place. Some – now we’re seeing a partisan division here because of the signals coming out of the White House that there’s distrust of the mail, and so you will end up with Republicans being more likely to vote in person, which actually was not historically the trend, at least through the first few, you know, elections of the – of this century. But when it comes to in-person voting, it’s a critical stopgap for voters with disabilities, for people who want to do same-day registration and as a fail-safe for any dysfunction in the mail process. Because one thing that we’ve seen in vote by mail is that especially when you make a rapid transition like this, and we saw this in the primaries, there can be some birthing pains. And a lot of people are not going to receive their ballots, and there could be errors in the mail balloting system, which will lead, really, thousands and perhaps hundreds of thousands of people to want to rely on polling places on Election Day or, in the early voting period, to remedy problems with absentee ballots.

If you go to see the research that we’ve put together at the Stanford-MIT Healthy Elections Project at, you will see research on the states where we look at which types of absentee ballots have been less likely to be counted. And we do find that younger voters, first-time voters, racial minorities are less likely to have their ballots counted, and that’s largely because of lateness – that if you’re a first-time – if you’re not familiar with voting by mail, a lot of times, you might miss the deadline. But there are also questions about how to verify signatures and how to make sure every vote is counted. And, you know, as a law professor, I can tell you that that’s an area where we’ll see quite a bit of litigation in the coming weeks. Thanks very much.


RICK WEISS: Thank you, Nate. Very interesting. I want to remind reporters that all these slides that you’re seeing today will be on our website after this briefing, as well as, within a day or so, video and transcript. Next, over to Alex Halderman.

Ballot Integrity in 2020


J. ALEX HALDERMAN: Hi. I’m Alex Halderman from the University of Michigan. I’m a computer scientist. Let me see if I can share my screen, which may or may not be possible for a computer scientist to figure out. And, indeed, it looks like I’m going to have a problem sharing my screen. I’ll try one more time here. All right, can you see my slides?

RICK WEISS: We do see it, yep.

J. ALEX HALDERMAN: Good. All right, so I want to talk about election cybersecurity and particular vulnerabilities and challenges that are facing the complex computer infrastructure that we use in the United States in order to conduct the voting process.

RICK WEISS: And, Alex, actually, we’re not seeing your slide now. I think we only saw your off-screen picture there for a moment.

J. ALEX HALDERMAN: Try that again. Are you seeing it now?

RICK WEISS: I’m only seeing your icon headshot.

J. ALEX HALDERMAN: Oh, goodness gracious. Give me – give me one second. I’m going to…

RICK WEISS: No problem.

J. ALEX HALDERMAN: …Fix it by doing this. Well, it doesn’t look like I’m going to be able to get my slides up, but I will nevertheless…

RICK WEISS: We’ll post them later.


J. ALEX HALDERMAN: I will, nevertheless, persevere. So we’ll provide the slides to everyone later. Let me start by talking about voting machines. So voting machines are at the core of our election technology. And whether they involve a paper ballot or they involve a purely electronic paperless system, they are the first place that I worry about vulnerabilities because they’re the ultimate arbiter of the election results. So…

RICK WEISS: And, Alex, if you want to go live on screen, you can turn your video camera on while you’re speaking.

J. ALEX HALDERMAN: Oh. I’m sorry.

RICK WEISS: There you go. Now you’re good.


J. ALEX HALDERMAN: All right. Very, very sorry about that. So over the last 15 years, security researchers like me have had the opportunity to study many of the voting machines used in the United States. And in virtually every case where independent scientists have had access, we’ve found that it would be possible for attackers to install malicious software on voting machines and actually change the election totals that they report. And it’s virtually every case. So we have an epidemic of insecure voting technology in the polling place, and going from individual voting machines to an attack that could potentially change the outcome of a close and contested national election is easier than most people would realize. Attackers could target the most contested swing states or swing districts and probe all of them, looking for weaknesses, as Russia did with voter registration systems in 2016. After that, it’s possible to spread from the Internet to voting machines in polling places through just a couple of steps by spreading from the infrastructure that local election jurisdictions use to program the machines for each election into the voting machines in the field. From there, attackers could install malicious software on the election equipment in order to shift just a small fraction of votes.

Our best defense against this is that many voting machines today produce a paper trail of record of the voter’s intent that can’t later be changed in a cyberattack. But the problem is that today, almost half of states have no procedures in place for routinely looking at that paper trail to confirm the outcome of the computers. And since attackers can predict where postelection audits or recounts that involve that paper will take place or won’t, an attack could be targeted in ways that could have national implications. So what do we do about all of this? Two years ago, the National Academies of Sciences, Engineering, and Medicine conducted a review of election security and produced what’s called a consensus report, the strongest category of recommendation from the scientific community reflecting the consensus of scientists and also election officials in this area. And what the academies recommend, and what I recommend, is that the core of our solution has to be based on good, old-fashioned paper ballots and making sure that those ballots are rigorously audited after every election. That may sound like a retrograde approach – computer scientists recommending paper – but it’s actually pretty sophisticated thinking. In other areas where we have very high-stakes security questions, say, like commercial aviation, if there’s a possibility of having a physical fail-safe, we absolutely want that to be in place.

A modern commercial jet has a sophisticated satellite-guided computer navigation system, but it’s also required by law to have a magnetic compass in case the computers fail. It’s the same with paper and voting. Of course, we all know, too, that there’s a long and well-documented history of tampering with paper ballots in this country, going all the way back to the introduction of balloting on paper in the 19th century. But with modern optical scan voting systems, we have a good defense against that, which is that every ballot is initially counted by scanning it into a computer. So you have two records of each vote – one on paper, one in the computer – that we can check to make sure agree. And as long as they agree, cheating in the election and getting away with it would require both a large and a high-tech conspiracy. Now, the problem, though, is that you have to check to make sure the paper and the computer systems match. And here, again, science has a very good answer. We don’t have to look and count every ballot by hand the old-fashioned way.

We just have to sample enough ballots at random to be able to confirm the election outcome with high probability. This is called a risk-limiting audit. And techniques for risk-limiting audits have been perfected over the last decade by computer scientists and statisticians, and they’re now being piloted or even required by law in a number of states. So looking ahead to November, I see both a lot of progress since 2016 and some remaining challenges. Much of the progress has been vastly greater awareness of cybersecurity problems among policymakers and election officials. Election security is now at the top of everyone’s priority list. There’s also vastly more cooperation between the states and the federal government. And there have been substantial investments in new equipment and security testing based on nearly a billion dollars in new federal funding over the past three years. But, unfortunately, the challenges are also quite significant. It’s still the case that across the country, 14% of votes this year will be counted by election equipment that has no paper trail whatsoever, and another 20% of ballots will be counted on election equipment where the voter isn’t directly marking the paper, but it’s being marked indirectly through a computer device called a BMD, or ballot-marking device, that could also potentially be hacked to change the vote. With BMDs, voters do have an opportunity to review their ballot, and we should be encouraging them to do that quite rigorously.

But, unfortunately, research, including my own, has shown that voters, unprompted, do not do a good job reviewing their printed ballots. And, in fact, in a study I published earlier this year, only about 7% of voters caught an error that we deliberately introduced into their printed ballot by hacking the machines. States also, by and large, lack rigorous postelection audits. Only about 11 states have either piloted them or will require than this year, and 23 states have no statutory postelection audit requirements at all. This means that absent of a challenge through the courts, the computer results are not going to be verified. The expansion of vote by mail, as Nate has discussed, presents many additional and new challenges for elections, including security challenges. Vote by mail presents a different attack surface and has the possibility that attacks, including challenges through the Postal Service and their infrastructure itself, could be disruptive. Still, I think the most significant worry with vote by mail is just as Nate said; there’s so much new that’s being introduced this year, there are almost certainly going to be places where the system breaks down, and just for completely non-security-related reasons, that people suspect is related to an attack.

And that brings me to the last challenge. 2016 will very much be a question of legitimacy. And we know from 2016 and the Russian efforts to attack the election then that one of the big goals of attackers trying to infiltrate our election system is to undermine the legitimacy of democracy and public acceptance of election results. And for that reason, I think in reporting on events that happen on the ground in November, we all have to be very cautious. Many things that go wrong will have nothing to do with attacks, but it would be very easy for people wanting to spread misinformation and disinformation to paint them as if they were the result of hacking. We have to wait and see and let the experts investigate and find out what is just natural failure, what is human error and what may actually be evidence of computer hacking that needs to be more thoroughly explored. But we have to wait and see. November is coming. I think – what is it? – 49 days till the election today. So with that, I’ll be happy to take questions.


Are there recommendations to make with regard to HVAC, air circulation, and ceiling height for safe voting?


RICK WEISS: Great. Thank you very much, Alex. Thanks to all three of you for those opening comments. And we’ll get into Q&A now. As a reminder, you can hover over the Q&A icon at the bottom of your screen and type in your questions, and I will convey them. And we’re going to start here with a question for you, Bridgett, from Kate Gavaghan from Sci in the Tri in Durham, N.C., who’s asking, in addition to the spacing that you talked about, what about HVAC and air circulation, ceiling height or just overall density of people in a room? Are there recommendations to make with regard to those factors for safe voting?


BRIDGETT KING: Sure. So the American Society of Heating, Refrigeration and Air Conditioning (ph) early in October actually put together a list of guidelines for the selection of polling locations, obviously encouraging individuals to follow CDC guidelines. But beyond that, they do have guidance for polling HVAC systems, and it includes things like the selection of polling locations with really high ceilings, increasing ventilation in the polling location either by opening the doors and/or windows, having lines, if they do form, form outside as opposed to inside. So there are recommendations and guidance that do exist for local election officials. But I do think we also need to keep in mind that finding polling locations can be a challenge. And while there are clear opportunities for guidance in this election, as both of my colleagues have mentioned, we’re very close to the actual date of the election, and there’s a finite amount of time at this point to make changes and inform voters so that they know what is happening if there is or if there are, perhaps, better polling locations that adhere to the guidelines provided by ASHRAE.

How should journalists report on the vulnerability of voting machines while not stoking fear and doubt?


RICK WEISS: Thank you. Next question here is from Bret Jaspers from KERA in Dallas, Texas. How would you report on the vulnerability of voting machines while not stoking fear and doubt, which in itself contributes to undermining of trust in the election? I think that’s a question that all of you might be able to address, but, Nate, let me start with you and see what you think the approach is.


NATE PERSILY: Well, look. I think that Alex is right that we need to stand up a system of auditing and that that might, you know, go a long way to instilling confidence. I actually think that we’ve spent too much time focusing on the machines themselves and not the other electronic aspects of the electoral infrastructure and that we need reporters to focus on things like the registration system, the e-poll books, the election night reporting system – that there’s a lot of technology that’s involved in an election which is amenable either to dysfunction or to hacking. And so I think it’s important to look at all of those things. But, you know, as Alex mentioned, you know, well over 80% of the ballots will be cast on paper this time and almost all of the ballots in the battleground states. And so emphasizing that, I think, is quite important. And, you know, when you see election – when you see machine dysfunctions where you have, you know, the sort of vote-flipping problem that you see on some of these screens that often gets attention on social media, you know, it’s important to not look at these isolated instances and suggest that they are a systemic problem. I think, you know, we need to focus on the real vulnerabilities to the system and the real opportunities for exploitation by bad actors.

RICK WEISS: Alex, do you want to add to that?


J. ALEX HALDERMAN: Yes. I think it’s important for the public to understand that there are understood vulnerabilities in the systems. That doesn’t mean that our systems are definitely going to be attacked. It doesn’t mean that our systems are as secure as they need to be. But a public understanding that there are vulnerabilities we understand, I think, will add credence to investigation of any problems that happens after the polls in November. So I think the most important thing is to emphasize that not everything that looks like it goes wrong on Election Day can be attributed to attacks and that we do have experts in government and in civil society who are standing by and ready to try to sort through problems that do appear.

So public awareness of the ongoing vulnerabilities in election systems – both the machines and, as Nate said, many other aspects of our election technology – is going to be really important for driving further improvements to that security in the next – under the next administration, whoever wins. So I don’t think that we should avoid talking about the real problems, but we have to avoid scaring people and hyping people at the same time. So maybe the most important thing to remind people is that the problems with election security can be solved given the political will and that we have made progress. And with continued political will, this is a grand cybersecurity challenge we can overcome.


RICK WEISS: Thanks. It sounds like one of the more immediate things reporters can do when they do learn of glitches that are happening here or there is to ask the question of the appropriate experts, is there any evidence that this is actually a large-scale problem, or are these a bunch of one-offs so we can keep things in…


J. ALEX HALDERMAN: And, frankly, 99% of the time, the experts are going to tell you, no, that’s not hacking.

RICK WEISS: That’s right.

J. ALEX HALDERMAN: Right. That’s just something that happens occasionally normally. It’s human error or it’s a non-hacking glitch.

Is there a shortage of poll workers for the 2020 election?


RICK WEISS: Great. We’ve got questions, actually, from two reporters that are pretty similar. Both Lucy Perkins at WESA in Pittsburgh and Kate Payne at Iowa Public Radio are asking about the apparent shortage of poll workers given concerns about health and other issues. Bridgett, is that something you have focused on at all, and do we know how that’s being dealt with?

BRIDGETT KING: Not directly. But I have some thoughts.



BRIDGETT KING: So according to the 2018 Voting and Administration Survey conducted by the Election Assistance Commission, 70% of the 6,500 jurisdictions in all 50 states reported that it was very difficult or somewhat difficult to get enough poll workers. So the challenge of recruiting poll workers isn’t necessarily new this year. It’s just been exacerbated by COVID-19. In response to the question of where the shortages were, I think different counties in different states are having different experiences and are responding to these challenges differently. So, for example, in March, in Tennessee, there was legislation passed that lowered the age requirement to serve as a poll worker. In Cuyahoga County, Ohio, the council approved a resolution – excuse me – that would allow county employees to receive paid leave if serving as poll workers in the November election. Even here in Alabama, there’s been a push to get more poll workers, and I believe they’ve increased the pay for serving as a poll worker. So I haven’t seen anything that says there is explicit trends with respect to whether the shortages are in urban or rural areas, but – so COVID-19 has made a challenging part of election administration all the more challenging. I will say we do see a lot more responsiveness on the part of administrative authorities in creating new avenues to encourage people to serve as poll workers that also vary.


NATE PERSILY: I can jump in here, too, Rick, since we’ve been working quite a bit on poll worker recruitment. Many of you are familiar with, which is a new organization that has sprung up in order to try to solve this problem. There – Bridgett is right, which is that there’s no – it’s hard to say systematically that you have more – you know, that whether it’s an urban or rural problem. I can tell you Detroit right now really needs a lot of poll workers. Philadelphia did need over 5,000 as of a month ago but right now is doing pretty well – still probably needs a thousand or so more poll workers. Most of the big cities in Florida, North Carolina and Wisconsin have said – and Georgia, for that matter – have said they’re OK, but I think one thing that we’re worried about is that people are going to flake a lot on Election Day. And so they need to build up greater capacity to make sure that they have people to substitute in. But one thing, also, to think about with the poll worker problem is that you’re going to have this whole new crop of poll workers who are going to be staffing the polls this year because of the older poll workers are not going to be serving. And we need a massive poll worker training program unlike anything that we’ve done before because, you know, it’s not intuitive to be a poll worker, and we need to make sure that these folks are really well-trained, especially given the new issues they have to deal with in the COVID environment.

How likely is blockchain-assisted mobile voting in the near future?


RICK WEISS: I’m glad you addressed that since that was part of the question I hadn’t conveyed from one of the reporters. Alex, did you want to add anything to the poll worker issue? Great. OK. So we have a question here for Alex from Janet Babin from the “Future Of Everything” podcast based in New York. What do you think of blockchain-assisted mobile voting? It’s happening in a limited way for military and disabled in West Virginia, for example.


J. ALEX HALDERMAN: Right. So blockchain in general is an exciting technology, but it’s not magic. It provides a good way to have decentralized public accounting of a set of events, but voting requires much more than that. We need not only a public ledger, but we need to a secure way of getting information onto that public ledger. We need a way of authenticating the voter to make sure they are the authorized person and not an attacker. And the blockchain-based or so-called blockchain-based app that has been used in West Virginia has actually been studied by computer scientists. A group from MIT did a security review that was published just last month, and what they found was a series of really devastating weaknesses, including that the app doesn’t actually interact with the blockchain at all. It just sends your vote up to a server that then does things you can’t see to it. Basically, the bottom line about blockchain is, like, blockchain protects your vote from manipulation about as well as Bitcoin can protect your money from being stolen. That is, it’s not a be-all and end-all solution. There are many ways that your cash can be stolen if it’s stored in Bitcoin, whether through compromise of your computer or the exchange where it’s stored, theft of a password, et cetera. Really, it’s the same with your vote, even with blockchain technology. I think it’s going to be many years before the state of computer security advances to the point where we can vote over the Internet securely, blockchain or not. And, unfortunately, it’s going to require some further breakthroughs.

Are there any concerns with hand sanitizer affecting the readability of paper ballots for people voting in person?


RICK WEISS: Thank you for that encouragement about the future. So I want to let reporters know, by the way, that we have permission from our panelists to go about 10 minutes beyond the top of the hour. We have a fair number of questions backed up, so prepare yourself to hang in there longer if you can. I have a question here from Clark Merrefield from Journalist’s Resource. Are there any concerns with hand sanitizer mucking up paper ballots for people voting in person? Anecdotally, I’ve heard this could be an issue. Anything about hand sanitizer – yeah, Nate. I see your hand.


NATE PERSILY: So the CDC now has guidelines on sanitizing voting equipment, and the Election Assistance Commission does on how to clean your voting equipment correctly. And so the – you know, depending on the technology that’s being used, there are particular protocols that the poll workers have to follow to sanitize it after each voter comes through. And so yeah. Look. As with anything with paper, if you’re only relying on the paper ballot and you dump an entire thing a hand sanitizer on it – right? – it could destroy the paper ballot. But, you know, you would still at least have the paper record there. And this is the kind of thing that we get with absentee ballots all the time, where you look to see whether – in a recount situation – whether the voter’s intent is clear. But yeah, this is something, you know, that we should be vigilant about. I wouldn’t overstate the significance of it. I’m a little more concerned about new poll workers over sanitizing the electronic voting equipment that is – or the ballot-marking devices, which won’t lead to votes probably being lost, but it could lead to some dysfunction with the equipment itself and leading to shorter – fewer machines in operation and maybe longer lines.

RICK WEISS: Thank you.


BRIDGETT KING: I just wanted to add one thing real quick. I read somewhere – can’t remember where it was – that in terms of sanitizing the equipment, that that should be left to the poll workers and voters themselves should not, for example, take an alcohol pad and wipe it over different surfaces that are electronic because there are specific cleaners for specific machines. And like Nate just said, if a machine does break down because of some sort of issue with the cleaner, that could then lead to longer lines and a less positive voting experience for them is all.

Are there known cases where someone actually has hacked a voting machine and changed the results?


RICK WEISS: Sounds like a good educational point to include in stories there. Thank you. A question here from Sherman Smith at the Kansas Reflector – are there known cases where someone actually has hacked a voting machine and changed the results? Also, what do we know about the vulnerability of voter rolls?

J. ALEX HALDERMAN: Well, I guess I can…

RICK WEISS: Alex, can you start?


J. ALEX HALDERMAN: Yeah, I guess I can address some of that. So we have plenty of evidence from the laboratory that voting machines are hackable. But actual cases where attackers have broken in and tried to change something are fewer. We know in 2014 in Ukraine that attackers linked to the Kremlin broke into the central reporting system used by the Ukrainian government to announce election results and had rigged the system to announce the wrong results. And that was only detected shortly before the results would have gone live. We also have – probably the most famous election to have been subject to attempted hacking tabulation was the 1994 election of Nelson Mandela in South Africa, where votes were tabulated by computer and where someone hacked into the computer network where the tabulation was occurring and attempted to disadvantage Mandela’s party.

The interesting thing about that is that it was detected by election officials and they stopped counting and went back and counted everything again to make sure that they would get the right result. But they swore everyone involved to secrecy, to never say what had happened. And only in very recent years, around the time of Mandela’s death, have some of the U.N. officials and other international observers involved started discussing it publicly and writing about it in their memoirs. So there are lots of reasons why we might not find out if attacks like that took place.


RICK WEISS: Great. And – but it sounds like no – sounds like no one is documenting an actual case of a hacking of a vote in this country.

J. ALEX HALDERMAN: That’s right. That’s right. And in fact, the 2016 election, people often have questions. We know Russia did attack voter registration systems, at least one election system vendor and, if you believe Bob Woodward’s book that’s out this week, perhaps even installed malware in registration systems in Florida. But we don’t have any evidence that they got as far as being able to change actual votes. And the evidence from the 2016 recounts, I think, supports the proposition that they did not alter the results of the national contest.

RICK WEISS: This – Sherman is also asking about the vulnerability of voter rolls. Nate, anything on that?


NATE PERSILY: Well, this has been – as Alex mentioned, that was – of the areas that we know where the Russians were probing in 2016, the voter registration database – when they say voter rolls, I assume they’re talking about voter registration databases. Those – that was the area that they – the Russians were most active in, at least as far as we know. The – now, saying that they probed a voter registration database doesn’t mean it was hacked, right? It’s just that there was – that they were peeking around. And the – I can tell you from conversations with folks at DHS that there’s been a lot of improvement on the security of the online voter registration databases over the last four years. That’s an area where I think they knew that there were problems and that they’ve tried to bring a lot of assistance. I think there are still some concerns, but the states have backup systems in place.

But just to reiterate one thing that I think both Alex and I are saying, which is that there is – there are vulnerabilities in the system, and then there are sort of perception hacking that can result as well – that if you undermine confidence in the system by, you know, slowing down the reporting of results or casting ambiguity in the way that different votes are cast and counted – right? – that even if everything works, then, you know, the perception that it hasn’t been working could be just as dangerous.

For absentee and mail-in ballots, how is voter privacy protected in terms of the confidentiality of who each voter casts a ballot for?


RICK WEISS: It’s a super important point for our audience here of reporters that they not contribute unduly to that sort of lack of trust, so thanks for bringing that up. Question from Bonnie Juettner a freelance reporter based in Milwaukee. For absentee and mail-in ballots, how is voter privacy protected in terms of the confidentiality of who each voter casts a ballot for?


NATE PERSILY: Well, so the – so different states have different processes in dealing with this. And so in general – right? – every state has an envelope in which you place the ballot. The signature is placed on the exterior envelope, and the voter’s identity is validated. And then the ballot is taken out and separated from the envelope, which has the voter’s name on it. And so, you know, in theory, you could have someone who gets a hold of ballots and sees the name and then attributes – and opens it up at the time. But the processes that are in place are there to protect confidentiality and the voter’s identity so that the validation of identity is a separate process from the counting and even the opening of the ballots.

Do mail-in ballot tracking processes allay some fears of fraud?


RICK WEISS: Great. And a follow-up question from Kate Gavaghan (ph) in Durham – do mail-in ballot tracking processes allay some fears of fraud?

NATE PERSILY: Well, I think they’re really quite important. This is something that Charles Stewart and I recommended early on when the pandemic hit, that there are several technologies that are available to track the mail ballot through the system – through the postal system, through the counting system and the like. And different states have – actually, sometimes it’s by county, where different counties will have different technology to track the ballots. So I think for the voter who is really interested and vigilant to track their ballot, that, yes, I think it does promote confidence. And even the availability of that technology then is seen as at least one way – one additional point of security so that, for example, when the president says in North Carolina that you should vote by mail and you should vote in person just to make sure that – to see whether you – your ballot has been actually counted – right? – well, the answer to that is, first of all, that that’s a crime. Secondly, that if you do that, that there are – North Carolina has ballot tracking, so you can actually figure out whether your ballot has made its way through the system and been counted.

How can reporters prepare the public for the possibility that results will not be available on election night?


RICK WEISS: I have a general question here from Scott Thistle from the Portland Press Herald. Do any of the panelists have suggestions on how we reporters prepare the public for the possibility that we will not have results on election night? I don’t know who might want to jump on that first.


NATE PERSILY: I was part of a – if you’re interested, we issued a report through this group at UC Irvine on the sort of media recommendations. And also, The Aspen Institute is doing a whole session on this soon if you’re interested. But I will say this. First of all, that – just so people understand the way elections are called – right? We do not have a national election authority in the United States that is in charge of validating the vote nationwide as to how many votes each candidate received. That’s done at the state level, and it’s done through a process where the counties canvass the ballots, report the results and then there’s a certification at the state level.

The individual counties will, depending on the number of absentee ballots and the number of ballots that are in dispute, will take, you know, some time after the election, depending on their rules, to count all the ballots. I am one who believes that within a day of the election that we will have a good idea as to who won this election. That’s because – although we will have perhaps tens of millions of ballots that will remain outstanding, we will have enough evidence from the ballots that have been cast and counted by that point to make good inferences about who has won. Now, there’s a very big caveat there, which is that if the 2020 election is like the 2016 election, in which case everything comes down to Michigan, Pennsylvania and Wisconsin and we have millions of outstanding absentee ballots that haven’t been counted and the election is close, then we may need to wait to week or more to find out who the eventual victor is. But barring a repeat of 2016, we could be in a very good position to know who the winner is within 24 hours of the election.


J. ALEX HALDERMAN: I would add to that. And that’s all the stuff I very much agree with that Nate said. But I would add to it that it’s important not to pressure the candidates to say that they are going to accept the results on election night, as well, because one of the important parts of our process is the process of canvassing, of auditing, even of challenges and recounts when necessary when results are very close or there’s evidence that something may be wrong. And I think in 2016, Hillary Clinton and her campaign boxed themselves into a corner by vocally criticizing the Trump campaign for not saying that they would concede the result on election night. All of those things that candidates have as options after the election are part of the process or within the rules. And we have to let the rules play out before the candidates themselves concede, as well.


RICK WEISS: Thanks. Reporters, reminder – we’re going to go about 10 minutes extra here. Bridgett, anything to add on advice to reporters should things not happen instantly, from your perspective?


BRIDGETT KING: I mean, I just sort of – what is it? – rearticulate the sentiments of my colleagues here. I think it’s important to communicate to the public that different states have different laws and different timelines and because something it’s taking longer in one state versus another does not mean something nefarious is afoot. They are required by law to follow their processes. And you know, on the conversations we’ve been having about integrity and confidence in the process, even though the public might want to know who won, I think there’s also something to be said about local jurisdictions and states sticking to their processes in spite of the protocol and perhaps public pressure they might be under to announce a winner.

How many swing states do not have backup paper ballots, or do not conduct audits?


RICK WEISS: Great. Let’s see. I have a question here I wanted to get back to from Laura Lundquist from Missoula Current – Montana. For Alex, of the states that either don’t have backup paper ballots or the 23 that don’t do audits, how many belong to the dozen or so swing states – thinking Arizona, Florida, Maine, Michigan, Minnesota, Nevada, New Hampshire, North Carolina, Pennsylvania, Wisconsin?

J. ALEX HALDERMAN: All right. Well, rather than getting into state-by-state, why don’t I point you to Verified Voting, which is an organization that tracks the state of election equipment and auditing nationwide? They have a wonderful online tool called The Verifier that will allow you to get that kind of information for each jurisdiction about what equipment and procedures they have in use.

What are the greatest concerns of foreign interference at this point in the election cycle?


RICK WEISS: Perfect. I’m going to ask one more question here, and then I’m going to ask Steve Newell from EPI Center to make a few remarks based on what we’ve been hearing. Then, I’m going to come back finally for quick wrap-up statements from each of our panelists. The question here is from Tracy Brown, producer at Associated Press. Can the panel speak about the greatest concerns of foreign interference at this point in the election cycle? What lessons of potential foreign interference haven’t been learned from 2016? We’ve talked about this a little bit already. Anything extra people want to add on foreign interference?

BRIDGETT KING: I mean, I guess the one thing I would add is – you know, obviously, I am not a super tech person (laughter). But I guess one of the bigger concerns I’ve had even before 2016, though, was the – are the negative consequences of misinformation and disinformation and how rapidly it can spread, particularly on social media. So I think one thing to do in your reporting and discussing these issues is to perhaps encourage voters to go straight to the source – so their chief election official in their state or their county elections office, as opposed to relying on information that they might see on the internet. Because even though there are obviously actors who are well-intended, they, too, might find themselves in a position of spreading misinformation if they particularly don’t consider how the rules are so vastly different from state to state, as we just saw with the Postal Service.


RICK WEISS: Great. Nate or Alex, anything to add on foreign interference?

J. ALEX HALDERMAN: Well, I think that it’s important to remember that the goal of very much foreign interference is to weaken our democracy rather than so much to manipulate it in favor of a specific candidate or other. And for that reason, attacks on the legitimacy of the process or attacks that just try to increase political divisiveness can be extremely effective. So just remember – you don’t have to hack anything in order to convince people that the election is a fraud. You just have to make it look like you did or merely suggest that you did. And I think for that reason, that’s such an easy kind of thing for an attacker to pull off. It’s much more likely that that will happen than that actual hacking to change the results will take place.

RICK WEISS: You just have to get reporters to say, don’t trust it.

J. ALEX HALDERMAN: Well, skepticism of all claims, both claims of legitimacy and claims of illegitimacy, is important, as in any scientific endeavor.

RICK WEISS: Nate, last comment…


NATE PERSILY: There’s a lot of good reporting, briefings from First Draft Media (ph) and others about how to cover disinformation. One thing that concerns me is how the legacy media institutions end up amplifying disinformation inadvertently by sort of repurposing and repeating claims that are made. If you want to find examples of foreign disinformation this cycle, you’ll be able to find them. But they’re going to be – as a total share of the disinformation that you’re going to see in this election – right? – we’ve got plenty that’s homegrown at this point, right? Disinformation that’s foreign-sponsored is really a drop in the bucket compared to what our own culture and politicians are putting out there.

RICK WEISS: I’m going to turn here to Steve Newell from EPI Center, who’s been studying some of these issues for the last year and a half. A couple of quick comments from you, Steve, and then we will wrap it up with final closing remarks from each of the three of our panelists.

Epi Center: Perspectives on Voting Security


STEVE NEWELL: All right. So to start, I would really echo Rick’s comments from the beginning of the briefing. You know, science is more vital than ever in considering how to ensure elections are fair, accurate and secure and how to cover these events of intense local and national interest. The decentralized nature of American elections really can make these issues challenging to cover. I think Nate’s comments on voting changes being on a continuum are exactly on the mark. Really, every jurisdiction is doing things slightly differently. Luckily, there are, you know, amazing scientists doing incredible work across the country to address these types of concerns. So really, scientists work on every aspect of voting in America, from Bridgett’s work on reducing COVID-19 transmission through polling place administration to Alex’s work on things like interventions aimed at ensuring voters check their printed ballots. I would also recommend everyone visit the Stanford-MIT Healthy Elections Project at, especially its vote by mail resource guide.

Given the considerable changes in voting systems in some places, as our speakers mentioned, we really can expect some delays in results. It’s also important remember that there’s lots of great evidence out there on things like vote by mail to help us understand the process. And recall that fraud in any case in elections is exceedingly rare. As multiple of our speakers mentioned, we also need to pay particular attention to voting equity this fall. You know, a 2019 study found that turnout among voters with disabilities was 5% lower than individuals without disabilities in the 2018 midterm election. That’s equivalent to roughly 2.4 million votes. Another 2019 study looked at phone data to calculate that, relative to entirely white neighborhoods, residents of entirely Black neighborhoods waited about 30% longer to vote and were 74% more likely to spend more than 30 minutes waiting at their polling place. And this was prior to COVID-19.

There’s also a considerable amount of scientific literature examining issues like confidence and perceived legitimacy of elections by people like Bridgett, Charles Stewart and others. For example, highlighting science on how voter confidence is affected by changing in – changes in voting systems can help your readers and listeners better understand our elections and what to expect on both election night and after. Like all elections, we will inevitably see some issues here, but scientific expertise can help us separate the hype and conjecture from reality. As speakers mentioned, isolated incidents really should be not treated as malicious behavior or hacking unless there’s underlying scientific evidence coming out from trusted sources. So with that, I’ll turn things back over to Rick.

What are some key takeaways for journalists covering the 2020 election?


RICK WEISS: Thanks. Thanks, Steve. I encourage people to check out the EPI Center website within the AAAS website. Just to wrap up, I’m going to ask each of you to just offer up a take-home message. What’s the most important thing you want reporters to walk away with today if they can only grasp one or two things at the end here? Nate – or, actually, Bridgett, let me start with you.


BRIDGETT KING: I guess my takeaway would be that elections are complicated. They are not the simple processes that we often talk about or taught about in schools. I think reporters and those in the media are in a unique position to help the public have greater confidence in our processes and procedures and understand them better. So what I would encourage them to do is – particularly those who are covering elections in a local jurisdiction, if they haven’t already – is to reach out to their local election officials. Have conversations with them. Ask questions. Read as much as you can about process and procedure, and just recognize that they are working to facilitate democracy under circumstances and in a situation that I don’t think any of us were prepared for with limited resources and in a very finite amount of time.

RICK WEISS: Thank you. Nate Persily?


NATE PERSILY: So I think, you know, my one admonition to the press this time is to remain calm unless otherwise noted. There are a lot of sky-is-falling scenarios that are being pumped out there. And there’s a risk that these become self-fulfilling prophecies and that we need to be – you know, there are – there’s the risk of violence in polling places. There’s a risk of lots of lost votes with mail. There’s a risk of, you know, contagion in polling places and the like. But I think for now, you know, it’s extremely important not to focus on the aberrant mishap or unlikely danger and to really focus on the process and how it’s going to work. There’s going to be plenty of time after the election if it’s close for us to focus on these controversies. But for now, I think it’s really important to try to build confidence in the system that’s working right now.

RICK WEISS: Thanks. And Alex Halderman?


J. ALEX HALDERMAN: So the thing that voters can most do at this point to increase both the ultimate legitimacy and the odds that the 2020 election is going to proceed without problems is to participate, is to vote and is to make sure that they are following the correct instructions that vary by locality about how they can do that this year. So if we can encourage voters to participate, and especially if that results in an election that is perhaps less close than one with less turnout and participation, that will do more than anything else to boost confidence, to boost legitimacy, to boost security.


RICK WEISS: I love that final point to get out and vote. Thank you all very much for the reporters on the line here. Again, this will all be on our website over the next day or so. Please do follow us on Twitter at @RealSciLine. And as you log off today, you’ll note a prompt for a very short three-question survey. We would very much appreciate it if you would take just the half a minute it takes to fill out those questions and give us some feedback so we can keep bringing you the most useful kinds of media briefings to help you on the job. Thank you all for attending today’s briefing, and thanks to our panelists for their contributions. So long. See you next time.

From the AAAS Center for Scientific Evidence in Public Issues (Epi Center)

Panelist presentations


Video: high definition

(mp4, 1280x720)


Video: standard definition

(mp4, 960x540)