Attempts by Russia to interfere with US elections have been headline news in the last year. But the problems with the election process in the United States goes deeper than the public generally realizes and includes obsolete, proprietary systems, a lack of funds for upgrades, and near monopolies on voting machines. As the 2020 US elections near, academics are working to provide solutions to these issues – and open source software and hardware are at the core of these solutions, together with modern interface design. One of the most promising solutions is Prime III, being developed by Juan E. Gilbert (Figure 1), a computer scientist who heads the Human Experience Research Lab at the University of Florida [1].

Figure 1: Juan E. Gilbert has been developing open source voting machines for almost two decades.

The problems being addressed by academics such as Gilbert go back to the 2000 presidential elections. In Florida, poor ballot design combined with difficult-to-use punch card voting machines resulted in an usually high number of voters choosing either too few or too many candidates, with ambiguous results [2]. In the ensuing debate, electronic voting became the leading solution, and in 2002, the Help America Vote Act was enacted to reform the voting process [3]. Unfortunately, the machines used in the 2004 election themselves caused problems; since then, federal funds for updating voting machines have not been available. In practice, American voters often use hardware that is obsolete by three or four generations.

Today, Neal McBurnett notes that the problems continue. In the 2016 elections, recounts in three states were done on the same machines that had produced questionable results, and auditing results were complicated by the fact that the four main vote-handling formats used throughout the United States are proprietary and do not easily communicate with one another [4].

"Everyone recognizes that we have an antiquated system of voting, with machines [that are] out of date," says Lee C. Bollinger, cochair of the Committee on the Future of Voting at the National Academies of Sciences, Engineering, and Medicine. "It is really imperative that the system be sure enough that it is a system that is credible, and that we can believe in. To that end, you have to do certain things: You have to have an upgradable system; you have to have a paper ballot; you have to [have] audited systems that can reliably tally the vote" [5].

To produce such results, Michael A. McRobbie, the other cochair of the Committee on the Future of Voting recommends a human-readable ballot and auditing of the voting process from start to finish, as well as regular updates of voter databases [6]. With another election less than two years away, the race is on to implement these much needed changes, but time is running short, and upgrades are being done on a county-by-county basis. At best, the next major American election seems likely to be conducted using a variety of technologies, including many of the machines that have created problems since 2004.

As Seth Rosenblatt wrote in 2018, "The industry that provides the hardware and software for the election process has been scarcely studied and often is opaque, even to election administrators, policymakers, and representatives at other governmental and nongovernmental organizations that support or directly participate in the election process" [7].

Third Generation Technology

In the current environment, most solutions are coming from academics like Juan Gilbert rather than industry. Gilbert remembers that, shortly after the 2000 election, "my students and I were at [a] conference where we heard a speaker talk about how computers could not be used in voting. They were saying that there were security issues. Essentially, they offered no recommendations on how to fix voting, but explained what couldn't be done. My students and I were very discouraged by this and decided to do something about it. We decided to fix voting in the USA. We went to the lab, brainstormed, and invented the Prime III" – the third generation of voting technology (Figure 2), with the first being paper and pencil, and the second being the current lines of voting machines [8]. Development has continued ever since.

Figure 2: The screen of a test version of Prime III.

Gilbert and his students identified the major problem was voting machines that stored votes – a tactic that posed security risks and made the results vulnerable to tampering. To overcome this risk, they made the Prime III stateless. That is, it does not store votes. Instead, it produces a paper ballot for each vote cast that is the official ballot of record. Because each ballot is individual, vote tampering is more difficult and likely more obvious, since a stack of forgeries is more noticeable than alterations to a computer file. Just as importantly, to avoid confusion, the paper ballot shows only the office being contested and the voter's choice, not the entire list of candidates. Voters can see that the ballot accurately reflects their choice, and no ambiguities complicate the vote counting.

Another major feature of Prime III is its accessibility. Prime III allows voters to cast ballots by tapping a touchscreen or speaking into a microphone. Those who can't articulate a candidate's name have the option of blowing into the microphone, and headphones and audio instructions are available for those who have trouble reading the screen.

Gilbert adds, "Prime III is open source because we believe it's important that everyone can see the code and determine it's secure. [Moreover, because] Prime III is stateless, it could be run using machines that don't have a hard drive. Prime III could be run from read-only mediums such as DVDs to ensure the code is what it is. From the hardware perspective, Prime III runs on commercial off-the-shelf components. You can go to Best Buy and order machines and use them in the election. If the hardware is malfunctioning, replace it. The hardware has very little to do with the actual ballot of record." Typically, the hardware consists of a standard computer or tablet, a keyboard, a headset and microphone, and a printer (Figure 3).

Figure 3: Prime III is assembled from standard computer components.

In addition to Prime III, Gilbert is also involved in developing systems for other voting issues. For example, VoterPass allows voters to make an appointment to vote, using web browsers, phone apps, or touch buttons on phone. Voters can opt to receive a reminder, and, at the polls, join a dedicated VoterPass voting line [9].

Gilbert is also working on Video Verification (vSquared) for voter ID [10], an issue that has been raised by both conservatives and liberals. "Essentially," Gilbert says, "You are your ID for voting. When you arrive to vote, you say your name and address to the poll worker. They will record a short video clip of you saying your name and address. They will compare you to previous video(s). If someone is going to pose as you, they will have to look like you, sound like you, and know your name and address. This approach is far superior to voter ID, [since] our studies suggest that only TSA agents and probably college bar bouncers are the only people who can accurately detect fake ID[s]. Our studies have shown that poll workers can't do this very well. However, with vSquared, they are much better" (Figure 4).

Figure 4: vSquared is a video system for voter verification being developed by Gilbert.

"We are also working on a new project for absentee voting. We are going to fix the issues of signature verification. Stay tuned," Gilbert says.

Barriers to Use

Prime III has pilot studies dating back to 2004. Two of these studies have been done at the state level in Wisconsin and Florida in 2014 [11]. According to Gilbert, "our pilots have consistently informed us that Prime III works. It works for voters from diverse populations, people with disabilities, and those from different age groups. Whenever we discover an issue or something we can improve, we simply implement it and move on to the next pilot study."

This verification hardly seems surprising. The logic behind efforts like Gilbert's is convincing to anyone with even a superficial understanding of computer issues. Also, although Gilbert does not mention the fact, open source solutions like his are apt to be considerably cheaper than the existing technologies. For example, San Francisco county will pay $21 million in the next three years for a contract with Dominion Voting Systems. By contrast, according to Chris Jerdonek, president of the San Francisco Elections Commission, implementing an open source replacement would cost about $6 million and could be serviced without vendor lock-in. In the ongoing absence of federal assistance, the difference in cost could very well make the difference between correcting existing problems and allowing them to continue.

However, as Gilbert admits, the struggle does not lie in creating technical solutions. Rather, "the struggle has been in getting them into practice." Gilbert identifies a lack of support as a potential problem, saying, "we are not voting machines vendors; therefore, if someone wants to use Prime III, they have to be technically savvy. We help them, but we are not vendors."

To all appearances, the concerns about voting in the United States are legitimate, and practical solutions are ready or in development. The main question now is whether the will exists to implement the solutions.