As of seven a.m. on the morning of June 9, here’s how Georgia’s primary election, with a fancy new electronic voting system, was supposed to work.
Across the state, with roughly half the voters voting by mail, an estimated one million voters would still arrive at their assigned polling places. After quickly confirming each voter’s photo ID, election officials would access the state’s new electronic poll (e-poll) book—synced in real time with the state’s voter registration base—to verify the voter’s registration status.
Each voter’s personalized information would be quickly transmitted onto a software-encoded activation card that would be inserted into an ATM-like machine known as a “ballot-marking device,” or BMD. Voters would mark their choices on the BMD’s oversized touch screen. An attached printer would then generate a paper record, summarizing each voter’s electoral choices. Finally, voters would hand these paper printouts back for tabulation and return the activation cards, before leaving the polls.
Quick, secure, easy to use. And unlike old-style direct-recording electronic (DRE) machines that caused electoral controversy in Georgia and elsewhere because they didn’t print out paper ballots that can be recounted in case of close or questionable election results, the BMDs produce a paper record. All of this had been widely touted in a promotional video produced by the office of Republican Secretary of State Brad Raffensperger, with a $400,000 TV ad campaign paid for with federal election funds. Federal funds also financed some of the $107 million cost of Georgia’s 30,000 new BMD machines and accessories, purchased from Dominion Voting Systems, one of the nation’s two largest voting technology vendors.
Sure, BMDs had their critics, like Marilyn Marks, the president of the Coalition for Good Governance, and Richard DeMillo, a computer science professor at Georgia Tech. Marks, DeMillo, and a few others had waged a relatively lonely battle in Georgia, arguing, often in court filings, that BMDs have software-related security risks, that the paper ballots aren’t truly verifiable, and that it would be cheaper and less risky to have Georgia voters simply mark their ballots with a pen or a number 2 pencil like voters in most of America—with a voting machine or two reserved at each polling place for people with disabilities and others who might prefer to use them. Raffensperger’s office dismissed such concerns as “remote, unfounded speculation.”
Then the voting began—and almost everything went wrong. In hundreds of polling places, especially in and around Atlanta, voters waited in long lines for up to eight hours. BMDs malfunctioned, and printers broke down. Voters’ cards were inserted wrong. The e-poll books didn’t work as advertised. Many sites hadn’t printed out their voter registration lists, or they failed to have enough backup paper ballots available. How many voters grew so frustrated that they left without voting—disenfranchisement by election screw-up—will never be known.
Raffensperger, a Republican, blamed Democratic county election officials for botching the rollout. They blamed him and suspected voter suppression. Neither Democrats nor Republicans in the state, however, called for the obvious fix: Get rid of the BMDs and just use hand-marked paper ballots for most voters in the upcoming November general elections. A lawsuit to require just that, spearheaded by Marks, was defeated in federal court. But in her ruling, the judge nonetheless emphasized that Georgia’s BMDs had “caused severe breakdowns at the polls, severely burdening voters’ exercise of the franchise.”
Georgia wasn’t the only place where BMDs failed this spring. They also did so, spectacularly, in Los Angeles and several jurisdictions in Pennsylvania. Most of these debacles were overseen by Democrats. Indeed, Pennsylvania’s Democratic secretary of state is now getting an assist from the Trump administration, the RNC, and state Republicans in her legal fight against the NAACP to require all in-person voters in Philadelphia and two other counties to use BMDs rather than hand-marked paper ballots.
This November, 20 percent of the nation’s 200 million registered voters will be directed to cast ballots on BMDs if they go to a polling place—not just in Georgia and Pennsylvania but also in many large counties in Ohio, North Carolina, and Texas. The chances of more technical meltdowns in these places are real, but that’s not the BMD critics’ biggest concern.
The actual nightmare scenario involves the integrity and ultimate verifiability of the vote tallies generated by the machines. Recall the BMDs’ printer-generated vote summaries (technically called “scanning ballots” under Georgia law). Voters can review them, though they’re not required to, prior to turning them in for tabulation. The summaries themselves aren’t used to determine election winners and losers; instead, there is a bar code printed on the paper summary, or, as in Georgia, a QR code. Unintelligible to the voter, these codes are read by vote-counting machines.
What guarantees the integrity of these tallies? How would voters know that the BMD software accurately recorded and translated their choices into those codes and onto the paper ballot receipts? And who or what do they trust if there are meaningful discrepancies between the codes and the paper receipts?
As Emily Levy, founder of Scrutineers, a new networking group for the election integrity movement, puts it, citizens should be asking election officials in universal-use BMD states this very pointed question: “When you are called upon to prove that these election results are correct, how is it that you think you’re going to be able to do that?”
Let’s say the November election is close, and Donald Trump comes up short in Pennsylvania or Georgia or North Carolina, or all of them, and loses the Electoral College vote. He has already sent every possible signal that he would contest such a result as fraudulent, would refuse to concede, and would litigate (or worse) to stay in office. It would make perfect sense for him and his lawyers to seize on the ambiguities of BMDs to argue that the voting was rigged and illegitimate. And Democrats would be hard pressed to prove Trump wrong—especially since some of their own elected officials and allies have been relatively quiet about the BMDs’ shortcomings, and in some cases have even been at the forefront of pushing for the machines’ use.
Even in the wake of the Georgia primary debacle, election reform groups like Common Cause and New York University’s Brennan Center aren’t demanding that all voters get hand-marked paper ballots in November. Instead, they’re urging that even more BMDs be supplied, especially in minority districts. In Pennsylvania, the Democratic secretary of state, Kathy Boockvar, is fighting an NAACP lawsuit that seeks to overturn the scandal-tainted decision of Philadelphia County to adopt universal BMD voting. (In August, the Trump campaign intervened in the case to back Boockvar, though it objected to other demands in the lawsuit, such as expanded vote by mail and early voting.)
What is the risk if the 2020 election hinges on the verifiability of millions of BMD-generated ballots marked not directly by voters, but by printers obeying the algorithms embedded deep within the machines’ proprietary software? Terms like BMDs and QR codes could end up making the 2000 constitutional crisis over Florida’s “hanging chads” look like child’s play.
In Georgia, skeptical Democrats were simply outnumbered as Republican leaders enacted legislation that required all of Georgia’s counties to adopt BMDs. Local governments didn’t even have the option to make their own choice. But in Los Angeles County, it was prominent Democratic officials, and progressive advocacy groups like Common Cause, that championed a new $300 million BMD-based system that failed so spectacularly that California has since mandated that every county mail all their active registered voters a ballot.
There are legitimate reasons for BMDs to play some role in today’s elections. For example, in Los Angeles County, software-programmed BMDs can—if they run properly—give voters the option to view instructions and candidate information in a language other than English. BMDs also can accommodate certain disabled voters for whom the traditional paper ballot isn’t a viable option. Even after the failure of LA County’s “Voting Solutions for All People” system in March, Kathay Feng, then the executive director of California Common Cause, defended the universal use of BMDs for all voters because it reduced stigma: “We didn’t want to create a separate and unequal system that people with disabilities or language disabilities had to vote on.” Disability access also remains a key selling point touted by voting machine companies, along with some, but not all, disability rights groups.
But ballot-marking devices were originally developed as alternative “accessible” voting machines for people with disabilities, not as universal-use equipment foisted on all polling place voters. This is the BMD critics’ central complaint. “Even if computer-marked paper ballots are necessary to accommodate those who are unable to hand-mark their ballots, it is utterly irresponsible to make them the primary system for all voters,” Jennifer Cohn, a leading election integrity advocate and journalist, says.
That’s not to mention the risks of debuting universal-use BMDs in the middle of an already fraught presidential election year, even if it were not being complicated by the coronavirus pandemic. That’s the commonsense view of Jim Dickson, the co-chair of the National Council on Independent Living’s voting rights committee, and a champion of accessible voting machines for people with disabilities. “Trying to install a new voting system for everybody during a presidential election,” he says, “is worse than changing tires on a moving car.”
Perhaps nothing better illustrates the silence and even complicity of certain Democratic elected officials—and major progressive allies—than what has happened in Pennsylvania.
In early 2018, Governor Tom Wolf declared that all counties had to jettison their old DRE machines and replace them with ones that used voter-marked paper ballots. So far, so good. But where Wolf and his secretary of state, Kathy Boockvar, went astray was in buying into what DeMillo, Cohn, and others deem the “false equivalency” at the core of this controversy.
The nation’s foremost election equipment vendors sell both major types of voting systems—ones that revolve around paper ballots marked directly by voters, and the far more complex (and more software-dependent) BMD-based systems that also include printers. Both types are often linked to other error-prone—and, many argue, also potentially hackable—technologies such as e-poll books. (Indeed, postmortems of the Georgia and LA County problems found that e-poll book were as much at fault for long voter lines as BMDs, if not more.)
But as with new automobiles, there’s typically a higher profit margin when vendors can sell higher-end, more complicated systems. (In Georgia, for example, the simpler ballot system, also offered by Dominion, would have cost nearly $80 million less.) And by convincing election officials to make all voters use BMDs, they could sell whole fleets of the machines.
Central to the vendors’ strategy was convincing election officials to see BMDs as good, if not superior to, systems that relied on hand-marked paper ballots. These key players included a prominent Philadelphia-based national organization, Verified Voting, as well as the Brennan Center and Common Cause. All three groups, in a major May 2018 advisory report for election officials, emphasized, as many experts continue to do today, the importance of having a voter-marked paper ballot, whether by computer or by hand—based on the now-debunked myth that voters checked those printouts. In fact, recent research has shown that 93 percent of voters don’t catch BMD errors on the printouts, when they do occur, and the system may be especially vulnerable to hacking.
Two-thirds of Pennsylvania’s counties eventually chose hand-marked paper ballots with scanners, though they also bought the requisite number of BMDs to comply with federal law on ensuring access to voters needing disability or language assistance. However, Democratic-led Philadelphia County, along with officials in GOP-leaning Northampton and Cumberland Counties, chose to embrace the more expensive Express-Vote XL BMD model sold by the nation’s dominant manufacturer of such equipment, Election Systems and Software.
These three outlier counties brushed aside expert and citizen objections over the XL’s cost, security, and near certainty of failures. For Philadelphia alone, the initial cost was $29 million for the 3,750 machines. (Only later was it revealed that this decision might not have been based solely on the machine’s merits; the nearly $3 million in hidden lobbying and campaign donations by the company likely contributed.)
The November 2019 debut of ExpressVote XLs in local elections vindicated critics’ concerns. Reuters later exposed that in roughly 40 percent of Philadelphia’s polling places, the machines malfunctioned. In Northampton County, according to The New York Times, the system’s multiple failures included tallying just 164 votes, out of 55,000 ballots cast, for a leading Democratic judicial candidate.
This face-plant led the county’s bipartisan election board to unanimously approve a no-confidence motion for the new equipment. But Boockvar and the Democratic county executive insisted that Northampton stick with the technology. Meanwhile, the Democratic Party remained mum at both the state and national levels. (Spokespeople for the national DNC, the Pennsylvania secretary of state, and the Georgia Democratic Party all declined to answer questions about their voting technology stances for this article.)
Richard Garella, a citizen activist who founded the nonpartisan Protect Our Vote Philly, says that after the Northampton debacle, “we were expecting major advocacy and political groups including Democrats to see the danger this posed for disaster in November 2020 and to do something about it, and we are disappointed that they didn’t.” The state Democratic Party’s tacit acceptance of BMDs was allegedly fueled, he argues, by influential Philadelphia pols, including a newly indicted city councilmember whose protégé led the election board while getting donations from Election Systems and Software.
At the national level, Oregon Senator Ron Wyden is one of only a handful of Democrats who have directly taken on BMDs and the election vendors who promote their use with tactics that have included hundreds of thousands of dollars in donations, trips, and lobbying, as chronicled by NPR and other outlets. Wyden is a chief cosponsor of a bill to rein in risky voting equipment while requiring a hand-marked paper ballot option for all American voters. “An election system is really only as strong as its weakest link. A single state or a handful of counties can determine the outcome of the 2020 election,” he says.
Some leading Democrats and progressive voting rights activists at the state and local levels in Georgia, Pennsylvania, and elsewhere have also been fighting the universal use of BMDs. But while Pennsylvania’s NAACP chapter filed a lawsuit against Boockvar’s office over the universal-use BMDs and other voting barriers, neither the state or national Democratic Party has entered the lists in support of the lawsuit. However, another major party is actively working to thwart it. The Trump campaign, the RNC, and the Pennsylvania Republican Party filed in late July a motion to intervene in support of Boockvar’s effort to insist on the use of BMDs this fall.
“Right now, Secretary Boockvar is fighting this case alongside the Trump campaign and opposing the NAACP,” notes John Bonifaz, the president of the reform group Free Speech for People, which is coordinating several similar lawsuits around the country.
Legal efforts in Georgia were joined this year by new pandemic-related cases in North Carolina by the Lawyers’ Committee for Civil Rights Under Law, and in Texas, where suits were filed in mid-July by the Texas NAACP and Mi Familia Vota. (Courts in Pennsylvania, Texas, and North Carolina have ruled against these lawsuits in recent weeks, but the civil rights attorneys have either mounted appeals or expect to do so shortly.) Kristen Clarke, the president of the Lawyers’ Committee, clearly sees the discriminating impact of this risky voting technology. Clarke is supporting a lawsuit by the North Carolina NAACP against the rollout of the machines in the Charlotte area, noting, “It’s no coincidence that the gravest problems with outdated and hackable voting equipment are
in states with a history of suppressing the votes of people of color.” Clarke and others point to an example in Shelby County, Tennessee, which includes Memphis. Republican state legislators and their appointees are trying to force the universal use of BMDs in that predominantly Black county—where in a 2015 election 40 percent of the machine votes from predominantly black precincts weren’t counted by the county’s central tabulator.
Why are so many Democrats and their allies not vigorously opposing the proliferation of BMDs across the electoral landscape? It may be out of fear that even raising such questions might end up reducing turnout by shaking Democratic voters’ already weak confidence in our election system. BMD critics say they’ve heard this argument—usually voiced privately—but find it specious. “There is no indication that talking about election security reduces voter participation,” says Lulu Friesdat, the president of Smart Elections, another election integrity group fighting BMDs. Friesdat cites a 2018 Harris poll showing that voters were, in fact, more likely to vote if they were worried about hacking.
There is clearly an ongoing tension between election integrity proponents focused on accurate and verifiable vote counts and the larger, more influential voting rights groups that emphasize increased turnout and pushing back on laws they view as aimed at voter suppression.
Whatever the reason, several of the nation’s best-known election watchdog groups haven’t been barking very loudly, if at all, about the security risks and logistical dangers of BMDs. Two of these groups are Verified Voting, based in Pennsylvania, and NYU’s Brennan Center. Both have played prominent roles in helping muffle objections to the BMDs, and even have lent their credibility to the promotion and rollout of these systems in Georgia, Pennsylvania, and other key states.
Even before the disaster in Georgia, BMD critics were characterizing the Brennan Center as “missing in action.” “They’ve got a passive attitude,” Marilyn Marks says, “and just don’t consider it a priority.” She notes that Brennan chose not to support her lawsuit—or even intervene in Georgia when the state board of elections fined the Athens-Clarke County election board $5,000 a day for wanting the option to use hand-marked paper ballots instead of BMDs.
Lawrence Norden, director of the Brennan Center’s Election Reform program, has also been mostly silent about the negative press and malfunctions of BMD-based systems in their widely read and well-respected cybersecurity reports. While Brennan has been reassuring the public that the old DRE machines generally won’t be used this fall in battleground states, when it comes to BMDs, Norden admits, “it’s not been a priority.”
Verified Voting has perhaps done more to help kill paperless DREs than any other national organization. In early 2019, it was urging Georgia to use hand-marked paper ballots. But then, critics alleged, it gave the machines its blessing by participating in audits that were touted by Raffensperger and Dominion as proving the accuracy and validity of the BMD vote tallies.
That characterization was derided as invalid by two of Verified Voting’s election security experts, who then resigned in late 2019 from the organization: Georgia Tech’s DeMillo and the board member Dr. Philip Stark, a Berkeley statistics professor and an elections expert. Marian Schneider, Verified Voting’s president, says of the departed critics, “We don’t agree with them that helping elected officials do a better job at election administration is inherently a bad thing.” After the resignations and the ensuing negative publicity, Verified Voting clarified last December that it was still firmly opposed to universal-use BMDs.
The situation in Georgia illustrates the risks that trouble-prone voting technology may pose for the November elections and beyond. The commonsense reforms pushed by Marilyn Marks’s group could be widely applied in some swing states before the federal lawsuit is resolved. One option would be to remove most of the BMDs, give voters the same kind of paper ballots being sent to absentee voters, and then use the scanners already on hand to tabulate them.
Of course, printing millions of additional paper ballots in states like Georgia, Pennsylvania, North Carolina, Ohio, and Texas will cost money. But surely it would be worth it. Otherwise, we face the real possibility that Trump’s lawyers will be battling Biden’s over the legitimacy of the election for weeks or months, while Trump remains in the White House, tweeting about how he won’t let Democrats and voting machine companies collude to steal the election.
A version of this piece appeared online on September 23, 2020.