Why upgrading voting machines is important for election integrity
Why upgrading voting machines is important for election integrity
Voting machines in the U.S. are largely built and sold by private vendors. And like with any market, there are big, established players, and then there are newcomers, like the San Francisco-based VotingWorks, a nonprofit that uses open-source software in its voting equipment.
But beyond the software, VotingWorks tries to use more off-the-shelf, standardized components in hopes that states and cities will be able to upgrade their aging voting equipment while not spending too much money.
“We think that the high cost of these machines leads to states being unable to upgrade as often as they should and want to, and it leads to security issues, reliability issues,” said Ben Adida, executive director of VotingWorks, during an interview with “Marketplace Morning Report” host David Brancaccio.
“So, for us affordable voting machines is really a core issue of democracy,” Adida said.
One of VotingWorks’ big projects in this year’s general election is assisting with the vote auditing process by using a rigorous, statistical approach known as a risk-limiting audit.
While only a few states will be using risk-limiting auditing after their polls close, Adida hopes their contributions will help that system become more streamlined and become a regular part of the post-election process.
“It’s the flossing of elections,” Adida said. “You don’t floss because there’s a real problem. You floss because it’s the right hygienic thing to do in the long term.”
Below is an edited transcript of Brancaccio’s full conversation with Adida on the importance of risk-limiting audits, building affordable voting machine infrastructure and why paper ballots aren’t likely to go away anytime soon.
David Brancaccio: I mean, there you are in the land of silicon and zeros and ones, yet you think sheets of paper should continue to be part of our system of recording votes?
Ben Adida: I do. I think paper is a critical piece of voting. And, importantly, in 2020, almost every state is going to be back to using paper.
Brancaccio: Paper because, what, you can go back and check?
Adida: The most important thing is exactly that, right? As the voter, you and the paper can be in a room just by yourselves, just you and that piece of paper, and you can check that your vote was properly recorded. That’s the key idea. You don’t need a computer, any software, anybody else to tell you how your vote was recorded. And because the ballot has to be secret, that’s really important.
Brancaccio: Then later, if there’s some issue that’s cropped up, there’s the piece of paper somewhere, hopefully.
Adida: Right. The other important thing is that once you drop that piece of paper in a ballot box, in a drop box, in the mail. Then we have a chain of custody process to make sure that all that paper gets managed properly, gets stored properly, can be recounted, can be audited. That’s exactly right.
Brancaccio: So you don’t see us moving away from paper anytime soon? I mean, two years, four years, no?
The importance of a secret paper ballot
Adida: Definitely not two or four years. One should always be careful about predicting things, you know, 10-plus years out. We could very well have some breakthroughs that change the way we think about it. But, again, the key issue in all of this is the secret ballot. It’s what differentiates voting from every other application online. And it’s the reason why we want that piece of paper that a voter can look at, with no intermediation, with no piece of software or person interpreting their vote for them. If we can find a way to recreate that electronically and to get that level of trust and to maintain the secret ballot? Maybe. But it’s a topic that I’ve been studying for almost 20 years, and I don’t think we have a way to get there yet.
Brancaccio: No, I’m seeing your point here in a sense, right? Let’s say you had a fancy flat-screen system, and a voter came in with less experience with technology. There might have to be someone holding that person’s hand. And is that person, you know, acting in good faith? Or is that person up to something? That’s where you start to worry.
Adida: There’s an analogy I like to use. If you think about banking on your cellphone, right, which a lot of people do. You transfer some money to a friend or you pay a bill on your cellphone. When you click send or when you click pay on that cellphone — on your smartphone — how do you know that the bank did what you expected it to? Well, you’re going to have an audit log, you’re gonna have a trace of that. If you sent money to a friend, hopefully, they’re going to receive it and they’re going to see it in their bank account. And if they don’t, you’re going to find out about it. If a bill doesn’t get paid, you’re going to find out about that, too. If something goes wrong, when you’re banking online, there’s an extensive audit trail that you can go into and discover that something went wrong.
Now, shift your focus to voting online. Imagine you have a cellphone, a smartphone application on your iPhone, right? You pick your candidate, Joe Biden or Donald Trump, you say “vote.” And then the system says “Thank you for voting.” And that’s it, you will never hear from the system again. You will never have a website you can log into to check that your vote was properly recorded because of the secret ballot. And so how do you actually know — in that moment — if you picture yourself watching that cellphone, tapping vote, and the screen says “thank you,” what’s giving you trust in that moment that your vote was properly recorded? I don’t feel that. There’s nothing after that action that’s going to give you that confirmation the way you would get if you were doing online banking.
On the vote-auditing process
Brancaccio: And take us through a bit of your work, briefly. You don’t necessarily have to count every single vote all over again, if there were questions in a given jurisdiction?
Adida: That’s right. Actually, the way to think about it, and the most important work we’re doing this year at VotingWorks, is in an area called risk-limiting audits. And the idea is, we have these paper ballots, it’s something that activists and election officials have been working on for 20 years to make sure everybody comes back to paper ballots. Then you take those paper ballots — that have been verified by every voter — and you put them through a scanner, which is nothing but a computer that will tabulate those votes and tell you who won. We need the computers for speed and accuracy. They can scan those ballots really fast. They can be objective and always interpret marks the same way. However, there’s a question of trust. Do you trust the computer to do the right thing? Maybe it was misconfigured. Maybe it was hacked. Maybe it’s not giving you the actual tally of those ballots. So you want to have an auditing process that helps you check that. You don’t want that auditing process, ideally, to require hand counting all the ballots because that would take many, many weeks and it’s extremely difficult to do that with a low error rate.
So instead, what a risk-limiting audit does, is it selects a sample of the ballots, a small subset of the ballots, and it does so using a rigorous statistical approach. And then that small subset is hand-counted. And because it’s a small subset, hand-counting it, taking all the time you need, 30 seconds, maybe a minute per ballot if you need to, is not a problem. And it can be done in the public eye with a bipartisan group of people looking at that ballot. And if the hand-count of that subset of ballots is sufficiently close to what you expect it to be, then you can have confidence that the tabulator did its job correctly. And that’s really, really important.
And, surprisingly enough, there has never been a risk-limiting audit in a presidential election before. There have been other kinds of audits that are not nearly as statistically rigorous. So they’re a little bit ad hoc, like, “hey, let’s pick 1% of the precincts and recount them.” They don’t necessarily provide you with that statistical confidence that — OK, yes, the tabulator really did the job. A risk-limiting audit does that and the first time one was carried out was in Colorado in 2017. This year in 2020, we expect to see five states run risk-limiting audits, and we’re going to be helping all of them except Colorado. Colorado’s got their own system to run it because they were the first ones. But all the other states running risk-limiting audits this year will be using software that we developed to help run those audits.
Brancaccio: So that means about five states have the capability to do that, or is it that you’re predicting, for whatever reason, there might be cause for doing it in certain states?
Adida: Ah, that’s a really great question. The important thing about a risk-limiting audit is you don’t do it because there’s a problem. You just do it as a matter of hygiene as part of checking the election. So, in fact, it can be done before the results are certified. So the five states that I’m thinking about, and it may be four, it may be six, we’re still working to figure that out, it’s because they have put in the time and effort and preparation work to be able to do it. It requires some practice, it requires preparation, it requires working with all the counties. So that’s the reason why they’re doing it, not because there’s any question in anybody’s mind. It’s, you know, it’s the flossing of elections. You should probably do it. You don’t floss because there’s a real problem. You floss because it’s the right hygienic thing to do in the long term.
The voting machine market and mechanics
Brancaccio: Now, you know this space better than most. You are one of I think it’s like about eight companies that does voting machines of whatever kind. Election Systems and Software Corp. is one, I think I’ve seen Hart InterCivic makes these things. But you would like to see in the future ways to foster more competition for this kind of stuff?
Adida: Absolutely. I think the most important thing to understand about the voting machine market is that it’s small. It’s about $300 million a year. It’s highly regulated. And it’s highly fragmented, different states have different requirements. And, obviously, that yields a market that is not very dynamic, right? Not a lot of players, not a lot of new players and not a lot of innovation in the voting systems that we have. So at VotingWorks, we’re taking a different approach. We are the only nonprofit in this space. And we are reusing a lot of existing technology, we build on off-the-shelf hardware, and we build on open-source software, so that we can move faster, be more customizable to each state’s needs, and just use better modern practices to be able to make a difference in this market that is otherwise very difficult to play in.
Brancaccio: I mean, it’s probably heavily regulated for a reason, right? I mean, we do want the votes to count. But you’re saying there may be a way to regulate appropriately, but still foster some competition.
Adida: I want to be clear that I think the regulation is absolutely appropriate. There are some details of how the voting systems are certified that I think can be improved. But the general idea of certifying systems and having standards is absolutely critical. I don’t think that should change. I just think that the approach to how we design products in the space needs to adapt to the fact that this is a small market that is also highly regulated, which means we can’t be reinventing the wheel. There’s just no room in this market to go reinvent technology that the Googles, Apples and Microsofts of the world have already created. So that’s why we build on off-the-shelf components, right? That’s why we don’t build custom plastic or custom screens or custom electronics or anything like that. We use off-the-shelf stuff and we’re able to move much faster and provide machines that are quite a bit cheaper where we expect our machines to be about one-third the price of the competition.
Brancaccio: I was reading here that a lot of the voting machines in America are getting a little old. I don’t mean ancient, it’s not like we’re you know carving our votes in runes on the ever-living rock or something, but like some from early-ish part of the century?
Adida: That’s right. So voting machines are currently pretty expensive. And states and counties, and most of the time it’s counties and sometimes states that do the purchasing, they don’t have a lot of money to buy these machines. So the last giant wave of machine purchasing that happened was, after Florida 2000, the federal government passed the Help America Vote Act, allocated $3 billion to the states to buy voting machines and then the states went shopping and they bought new voting machines. And there’s a whole story about how many of those voting machines were sub-optimal. Notably, they didn’t have paper trail. But, beyond that, even the ones that did, they were bought in the early 2000s. And many states and counties simply haven’t had the money to upgrade since then. So if you think about technology that’s, you know, closing in on being 20 years old, yeah, some of it is not great simply because it didn’t have a paper trail. But then other technology is just falling apart, right? It’s almost 20 years old. How much technology do you have in your house that’s that old? So, we think that the high cost of these machines leads to states being unable to upgrade as often as they should and want to, and it leads to security issues, reliability issues and whatnot. So, for us, affordable voting machines is really a core issue of democracy.
Brancaccio: I mean, you think this issue of aging machines is going to be a pressing issue this Tuesday?
Adida: I think it could be, absolutely. I think we are helped by the fact that most of these states have moved to paper ballots. So even if there are breakdowns and slowdowns, you know, you still have the paper to go back to. But, absolutely, I think in some counties you’re going to see some aging machines that are problematic. I don’t think it’s going to be a massive disaster around that topic, around that issue. But you’ll probably hear some stories about it.
The biggest threat? Disinformation. Here’s what you can do about it.
Brancaccio: But at some level, you’re thinking that if we wanted to prioritize what we worry about going into Election Day about the system of voting, voting machines, you know, breaking down or getting jammed or miscounting, is not actually your top worry?
Adida: That’s right, it’s not my top worry this year. The reason it’s not my top worry is, one, because we’ve moved to a lot of vote by mail that’s on paper. I think in any election — you have to think of elections as a very complex dance. And in any complex system, you have a lot of little failures. So you’re going to hear some of those failures. But if you stack rank all the things that could go wrong this year, I think the one I’m most worried about is disinformation causing folks to lose trust in the electoral process itself.
And you’ve already seen that. There was, you know, this threatening email sent, apparently, by Iran, that tried to scare a number of voters in a number of states about the legitimacy of the election. And I think the problem, the core difficulty of an election, is because it’s so complex, you always have small problems that come up. If you take those small problems and you amplify them, and you make them out to be these things that scare voters, or make them lose trust in an election, that in and of itself, is an attack on democracy, because it reduces voters trust, it reduces their motivation to vote if they don’t trust that their vote is going to be counted properly. It’s my largest concern this year, because I think it’s in the ether, right? We are all talking about, should we trust elections? And should we count the results for a few days after election day? Of course we should. That’s the kind of thing that makes voters question whether their vote matters. And I hope we can all do something to stop that.
You know, as an individual voter, the most important thing you can do is, don’t retweet, or don’t Facebook share, that scary story right away. You might be an unwitting participant in spreading disinformation. Read it, sit with it, don’t amplify it necessarily, right? It doesn’t help. News organizations can also go at this by just being a little slower in how they process this. Don’t over-rotate on every single issue that comes up.
And, of course, don’t expect the results to necessarily be known on election night. Because with all the vote by mail that’s happening, a lot of ballots simply aren’t going to come in until a day or two after Election Day. So, of course, they can’t be counted until then.
Brancaccio: Yeah, I mean, that’s a good reminder, right? That even in a generally smooth Election Day, because of so many votes by mail where the deadline would have been on Election Day, some states can’t even start counting those until those drift in.
Adida: That’s right. And in some cases, some states are not even allowed to start counting votes until Election Day itself. So they’ve got this backlog of vote-by-mail ballots that are sitting there, and they can’t process them yet. So in this entire setup of our elections, we have to remember just how much work and dedication our election officials, our clerks, our poll workers are doing. It’s really heroic, especially this year with all the complications from COVID.
There’s a lot happening in the world. Through it all, Marketplace is here for you.
You rely on Marketplace to break down the world’s events and tell you how it affects you in a fact-based, approachable way. We rely on your financial support to keep making that possible.
Your donation today powers the independent journalism that you rely on. For just $5/month, you can help sustain Marketplace so we can keep reporting on the things that matter to you.
