A Comparative Assessment of Electronic Voting
Part I: Methodology and Justification for the Cases Examined
The material for this report comes from government documents, academic books and articles, newspaper and magazine articles, personal interviews and communications, and survey data where available and applicable. Aside from examining the theoretical literature addressing remote electronic voting, the report also closely examines trials that have taken place in Canada at the municipal level, notably the cases of Markham, Peterborough and Halifax, and at the local, state and national level in Europe by drawing on the experiences of Estonia, Geneva and the UK. All of the Canadian trials were initiated from and administered by municipal governments. The European cases by contrast, regardless of the level at which the remote electronic voting pilot took place, were all launched and overseen at the national level, although local authorities did have input. The difference in scales, contexts and the magnitude of issues associated with both the Canadian and European examples justify examining them as separate sets of cases, even though some elements may be closely related.
While the report examines Canadian and some of the more prominent European instances of remote Internet voting, we can initially consider the situation in the United States. Although widely discussed, there has been no actual implementation of Internet voting in regular American elections. Furthermore, the debate surrounding Internet voting in the USA is considered poorly informed because of a lack of research (Alvarez and Hall, 2004). The bulk of the discussion focuses on the technical requirements of Internet voting, and has not proceeded to a real-world implementation of such a project. In fact, some argue that many of the problems that have occurred in the trials are the result of insufficient testing (Alvarez and Hall, 2004).
There has been an abundance of research and smaller trials of Internet voting in the USA, such as a state-wide straw poll of Republican party members in Alaska in January of 2000, the Arizona Democratic Party primary in March of 2000, as well as an experimental project (Voting Over the Internet Pilot Project) as part of the Federal Voting Assistance Program (FVAP) in conjunction with the 2000 presidential election, and the 2004 Michigan Democratic Party's Democratic caucus vote. Nevertheless, no larger scale projects have been implemented because of a culture of uncertainty surrounding the safety and security of Internet voting (Alvarez and Hall, 2004; Alvarez and Hall, 2008; Mohen and Glidden, 2001). In addition, the recommendations of the major American voting reports, the California Internet Voting Task Force report, the Report of the National Workshop of Internet Voting and the Caltech/MIT Voting Technology Project, all warn against the introduction of Internet-based voting methods given the perceived level of risk associated with them – notably the potential for fraud, vote buying and voter coercion as well as security issues and the threat of attacks (Alvarez and Hall, 2004). The advice of these reports serves to reinforce security concerns and acts as a deterrent from pursuing such projects or pilots.
All remote Internet voting projects that have been initiated for use in USA elections have been terminated in the planning stages of the projects. The 2000 California project is an important example of this. It was considered an ideal jurisdiction to trial Internet voting because the state possessed a high rate of Internet access (compared to the national average), an abundance of Internet related business and, according to survey research, general support for the notion of Internet voting. Prior to implementation however, the California legislature passed a bill, The Digital Electoral System Act, which required the state of California to assemble a task force to study the feasibility of using technology in elections (Alvarez and Hall, 2004). The final task force report concluded that Internet voting could not be used as a replacement for existing paper ballot procedures for a variety of reasons relating to security (notably computer security and voter identification) and made two major recommendations: (1) that, for the time being, Internet voting be tested solely within the absentee voting process; and (2) that it be very gradually phased in throughout the state, after considerable additional research. Overall, the report's findings regarding Internet voting were very tentative and any recommendations to proceed with an Internet voting program were laced with caution (Alvarez and Hall, 2004; California Secretary of State Bill Jones, 2000).
Created to target absentee electors, the one major initiative, the Department of Defense's Secure Electronic Registration and Voting Experiment (SERVE), which developed as an extension of FVAP, was created to test Internet voting as an alternative to traditional absentee voting for military personnel, their dependents and overseas citizens. Though more than 6 million citizens qualified for the program, it was decided to only offer it to 100,000 people for the 2004 primary and general elections to allow for adequate testing and evaluation. Even on that reduced scale, the project was cancelled before it could come to fruition due to opposition from a small segment of the scientific community. Specifically, SERVE was terminated not because of its system design or architecture, but rather due to concerns surrounding the Internet itself and the view that any transaction conducted over the Internet is not secure and considered vulnerable to system breakthroughs. It has been observed that the report that raised these concerns failed to note that the threats associated with Internet voting are analogous to threats surrounding traditional absentee voting such as vote selling, buying and coercion and denial of service attacks that occur without the Internet (Alvarez and Hall, 2008).
Other major American research projects, such as the National Workshop on Internet Voting and the Caltech/MIT Voting Project, also take negative positions on the introduction of remote Internet voting and have helped to affirm the legitimacy of fears surrounding Internet voting (Internet Policy Institute, 2001; Caltech/MIT Voting Project Report, 2001). Despite recognizing its potential to enhance accessibility for certain groups of electors, these reports come to similar conclusions that such systems pose "significant risk," such as potential attacks to the elector's computer, the server and/or network as well as issues of ballot legitimacy and secrecy, and conclude that Internet voting should not be introduced on a large scale (Alvarez and Hall, 2004:23). A culture of uncertainty in the USA surrounding the notion of Internet voting has prevented serious research and testing.
While there is lack of testing and practical research in the USA, Europe by contrast has been a breeding ground for Internet election projects and, as a whole, can be considered to have advanced the furthest with respect to Internet voting technologies and approaches. The important differences between the USA and Europe in this regard are aptly highlighted by Alvarez and Hall in their most recent book, Electronic Elections (2008), on the applicability of electronic elections in the USA:
When we published Point, Click, and Vote: The Future of Internet Voting in January 2004, we had little idea that we should have been publishing the book in Europe, not the United States. The road map we lay out in the book is being followed, just not in the United States. Instead, it is in countries like Estonia, Switzerland, the Netherlands, and France that e-voting experiments are being conducted (71).
Though the concerns raised in American literature are important, they are taken into consideration in Internet voting experiments in Canada and Europe. Furthermore, in these settings, the potential problems are examined in the context of real-life examples and pilots, rather than in theoretical discussions. Therefore, the remainder of this report concentrates on European as well as Canadian trials.
What is Meant by Electronic Voting and Internet Voting? What Types are Examined in this Report and Why Do We Predominantly Focus on Remote Internet Voting?
The term electronic voting is a blanket term used to describe an array of voting methods that operate using electronic technology. There are three primary types of electronic voting, namely machine counting, computer voting and on-line or Internet voting.1 With respect to the last of these types, there are four kinds of electronic voting that use the Internet; these include kiosk Internet voting, polling place Internet voting, precinct Internet voting and remote Internet voting (Alvarez and Hall, 2004). Kiosk Internet voting typically involves the use of a computer at a specific location that is controlled by election officials. This differs from electronic machine voting because, among other things, the ballot is cast over the Internet. Polling place Internet voting is conducted at any polling station through the use of a computer that is controlled by election representatives. Precinct Internet voting is analogous to polling place voting except that it must occur at the voter's designated precinct polling place (Alvarez and Hall, 2004). Remote Internet voting is voting by Internet from a voter's home or potentially any other location with Internet access.
The following section discusses the benefits and risks associated with Internet voting in general, primarily concentrating on remote Internet voting. This is because, in the first place, in most of the literature addressing electronic voting the term 'Internet voting' has become synonymous with remote Internet voting and is addressed as such (Mercurio, 2004). Furthermore, remote Internet voting has the greatest potential to positively impact accessibility for voters. Internet voting machines that are either located at a polling station or another central location still require electors to travel to the poll or location. While in some cases travel to a central location such as a mall or supermarket may be convenient, their use still requires additional effort that voting from home or work does not. Finally, remote Internet voting is most consistent with the development of other political aspects of society that have changed with technology. While kiosks and machines can be useful, people are now using home computers to conduct more transactions than ever before and this will only increase in the next decade.
1 Machine counting requires voters to punch a hole in their ballot which is then scanned and counted by a central computer. Computer voting or direct-recording electronic voting machines involve the use of either a keyboard, touch screen or some kind of pen or pointer and computer terminal and are immediately factored into the tally of votes (Parliamentary Office of Science and Technology, 2001).