West Virginia and Denver say mobile voting pilots increased turnout

Recent experiments by election officials allowing overseas voters to cast ballots through an encrypted mobile app are being credited with increasing participation in elections. This is despite continued skepticism from election-security advocates and computer scientists questioning the reliability and security of collecting ballots over the internet.

The claims come from West Virginia, which last year became the first state to allow some of its expats to vote on their phones, and Denver, which followed suit earlier this year during its mayoral and city council races. In both places, officials credited the app they used, called Voatz, with boosting the number of ballots collected from members of the military posted in foreign countries, and that those ballots held up to scrutiny by extensive auditing processes.

In total, 144 West Virginians on military deployments or living abroad otherwise used Voatz to transmit their ballots back home last November. Though a tiny fraction of the nearly 600,000 people statewide who voted in 2018, it was enough to increase the participation rate of overseas voters by 3 to 5 percent, wrote Anthony Fowler, an associate professor at the University of Chicago’s Harris School of Public Policy.

While the top-line result was encouraging to West Virginia Secretary of State Mac Warner, Fowler was measured in his full analysis of the prospects of mobile voting.

“Voter turnout is often low and unequal, but the opportunity to cast votes on a mobile device could drastically lower the cost of democratic participation,” he wrote after analyzing West Virginia’s experience with the Voatz app and considering its use against the larger voter universe.

Just 31 of the state’s 55 counties accepted Warner’s invitation to offer their overseas voters the mobile-voting option last year, and of those, 24 actually collected ballots using the app. Fowler looked at the number of overseas ballots those 24 counties had collected in five previous elections and calculated a sample of 1,754 voters who were likely be living outside the country last November and eligible to vote.

Fowler said the overseas-voting process did not veer drastically for the 144 people who eventually used the mobile app. Under the Uniformed and Overseas Citizens Absentee Voting Act, eligible voters must first send a postcard-sized application for an absentee ballot back home through physical mail, fax or email. Fowler surmises that because overseas voters were familiar with the application process, West Virginia offering a mobile app did little to grow the overall voter pool — but he estimates that making the application itself digital might have.

“The people willing and able to submit an [application] are also presumably able to vote by mail, fax, or email, so the effect of mobile voting will be lower among this subpopulation,” Fowler wrote. “The effects of mobile voting might be notably greater if registered voters could use the mobile app without first having to request it or if they could request it using only their mobile device.”

Warner, who’s previously suggested the state may repeat it in future elections, touted the modest turnout improvement Fowler’s study attributed to the Voatz app.

“West Virginians are proud that we are investing in new technology to encourage and help our military members stay civically engaged,” Warner said in a press release.

An opaque system

Voatz has come under criticism in other recent analyses. While the app advertises security features like facial recognition to verify users and transmits completed ballots back to election officials using a distributed ledger, the five-year-old company has not allowed its technology to be closely scrutinized by outsiders, according to a paper published in May at the University of South Carolina.

“The system has not gone through federal certification, or any public certification to our knowledge,” the paper says. “The company has not disclosed its source code nor allowed its system to be examined openly by third party experts, as other Internet voting systems have.”

The paper’s five authors, including Duncan Buell, a South Carolina computer science professor who has studied digital voting systems extensively, questioned Voatz’s facial-recognition and device-verification methods, as well as the security of the blockchain architecture it uses to record and transmit ballots.

“Facial comparison is done via machine learning processes that must be trained to match faces,” the report says of the app’s user-authentication process. “Exactly what training set was used, and how large and diverse was it?”

The researchers raise several more issues surrounding the use of blockchain technology in West Virginia’s app: “In West Virginia, was there a separate blockchain for each county, or one unified blockchain for the entire state? Could an insider at Microsoft or Amazon” — the two companies that hosted the 32 blockchain servers in West Virginia — “destroy the whole election by deleting the blockchains and server processes?”

Many of the South Carolina paper’s criticisms were raised in a Slate article last month that elicited a forceful response from Voatz. On its website, the company said it undertook several methods to secure the ballots it processed, including allowing West Virginia’s county clerks to conduct pre-tabulation audits, which the company says found no errors. The counted votes were subjected to four independent audits. Voatz also pointed to its enlistment of cybersecurity firm HackerOne to run a bug-bounty program that rewards individuals who find flaws or vulnerabilities in the app’s programming.

Denver claims success

The Voatz app has spread beyond West Virginia as more governments have partnered with Bradley Tusk, a venture capitalist who has funded the mobile-voting projects in West Virginia, Denver and elsewhere. On Monday, Tusk and officials in Denver, which offered its overseas voters the app in recent municipal elections, credited Voatz with doubling expat participation over the city’s 2015 contest. In total, Denver collected 119 ballots from 36 countries through Voatz for the May 7 primary, and 112 from 34 countries in a June 4 runoff.

Those ballots were subjected to a livestreamed audit by the National Cybersecurity Center, a Colorado Springs nonprofit, and a report published Monday shows the organization found “no issues with the tabulation and recording of the ballots.”

“Today’s existing infrastructure is outdated and vulnerable, and instead of continuing to use systems that we know are prone to hacking and interference, we are looking at advanced solutions such as blockchain and identity verification that offer an anonymous, secure, and transparent way to vote,” Tusk said in a press release. “Twice now we have proved the best way to move forward is accompanying blockchain voting with an auditable paper trail to increase voter participation and security measures for upcoming elections.”

Voatz’s next test will come in Utah County, Utah, which will offer it to its overseas voters in local elections this fall.

Despite early claims of success from Tusk and others, internet-based voting still has many doubters in high places. The U.S. Senate Intelligence Committee’s report last month on foreign hacking attempts during the 2016 election stated that “states should resist pushes” to move their elections online, even with the sympathetic example of the civically engaged soldier posted far from home.

“One main argument for voting online is to allow members of the military easier access to their fundamental right to vote while deployed,” the committee wrote. “While the Committee agrees states should take great pains to ensure members of the military get to vote tor their elected officials, no system of online voting has yet established itself as secure.”