Consultation and Evaluation Practices in the Implementation of Internet Voting in Canada and Europe
European Cases of Internet Voting
The following two sections provide more detail about three European country cases and six Canadian municipal and provincial cases where Internet voting has been introduced, was contemplated, or is being studied. In the course of discussing these examples, we examine the nature of consultation and evaluation methods used in these jurisdictions.
Estonia has the most comprehensive system of Internet voting in Europe. It has used Internet voting since 2005 during two sets of local elections (2005, 2010), two sets of elections for the national parliament (2007, 2011) and the European Parliament elections of 2009. Internet voting is used only during the advance polling period, originally for three days and more recently for seven days. The use of the Internet voting option has grown substantially over the course of these five elections, from 1.9 percent of all votes cast in the local elections of 2005 to 24.3 percent of votes in the 2011 parliamentary election. There are other methods of advance voting in Estonia, but I-voting constitutes a majority, 56.4 percent, of all advance votes (Madise and Vinkel 2011). Estonians living abroad cast external votes from 105 countries in 2011.
Ivar Tallo, Director of the e-Governance Academy in Tallinn, and formerly a Member of Parliament, gave the background of the establishment of Internet voting in Estonia (Tallo, October 16, 2012). The idea surfaced around 2000 in parliamentary committee discussions on constitutional affairs. The idea originated with the Minister of Justice at the time, who was the leader of the Reform Party. The Minister ordered preliminary reports on the feasibility and costs of an Internet voting option (Maaten, October 19, 2012). In his promotion of an Internet voting option, he was supported by the Social Democrats, of which both Tallo and his colleague, Liia Hänni, were members (Hänni, October 18, 2012). Not all parties were in favour; the Centre and People's Union parties were opposed at the time, and continue to be. (The People's Union party has since changed its name.)
The context in which Internet voting was originally considered in Estonia is important. Regular competitive elections were established only after the 1990 collapse of the Soviet Union, and 2000 marked a decade of use of electoral institutions. The Constitutional Committee of Parliament was reviewing a range of possible modifications to the electoral rules at that time, including the continuation of a personalized proportional representation electoral system, the electoral boundaries of the districts used to elect the members and the system of party financing. Advance voting was also being examined, and in this context the Internet voting idea was a relatively minor part of the total review (Tallo, October 16, 2012).
Voting turnout was an important factor in the consideration of I-voting in Estonia. "One of the declared aims of launching online voting in Estonia was to increase voter turnout, which perhaps could be described more realistically as broadening access possibilities and stopping the decrease in participation" (Madise and Vinkel 2011, 6). The Office for Democratic Institutions and Human Rights of the Organization for Security and Co-operation in Europe (OSCE-ODIHR) reiterated these points in its Election Assessment Mission Report for Estonia for 2007: "the main goals of introducing the Internet modality of electronic voting were to sustain and increase voter turnout, attract younger voters, and improve the convenience of voting" (OSCE-ODIHR 2007, 9).
The decision to proceed with Internet voting was made by the Estonian Parliament early in 2003, but the original goal of having the system implemented for the 2003 elections was postponed, and 2005 was made the target election instead. This had the added feature that the 2005 elections would be for the local level of government, and hence the system could be tried in a lower profile setting (Tallo, October 16, 2012). Project manager Tarvi Martens was chosen in 2003 to develop the project, and gathered a group of about 30 people with technical expertise to plan it (Martens, October 15, 2012). Consultations were limited to this group of experts, to the parliamentary committee and to the National Electoral Committee, which established a "supervisory board" to oversee the electronic voting development (Martens, March 26, 2013). As it happened, the 2003 parliamentary elections saw a large-scale turnover in membership, and an education process needed to begin again to renew the approval for the system to be implemented (Martens, October 15, 2012). The political parties were therefore the main entities consulted, and their primary concern was that they not be disadvantaged through the Internet voting system. Such an eventuality was feared if supporters of other parties were more likely to be Internet savvy. While, as already mentioned, there was some partisan opposition on this basis, the multiplicity of voting channels, both in advance polls and election day voting, meant that no one was required to use the Internet, and meant that the parliamentary majority was willing to try out the system, given some of the turnout and convenience goals.
The e-Governance Academy was established in 2002, and Tallo and Hänni developed a code of conduct to govern I-voting. All political parties were invited to a conference on this document, and most signed it (Tallo, October 16, 2012). The e-Governance Academy has continued to consult with the parties every year after that (Tallo, October 16, 2012). Parties are also invited to training courses before each election; however, few of them choose to participate (Maaten and Hall 2008).
Accessibility for those with disabilities and external voting were two additional criteria for implementation of I-voting. According to Martens, the main group of people with disabilities consulted was the blind. He mentioned that some early problems for the blind were resolved after consultation by certain technical adjustments (Martens, October 15, 2012). Other groups of people with disabilities were not mentioned. The provision of a better channel for external voting was a result of the continued development of I-voting. Regarding the Estonian community in Canada, only recently (through suggestions of Martens to the embassy in Canada after a 2010 visit to Ottawa) has an importer brought in the inexpensive card readers needed to vote by computer. These Canadian Estonians are mostly in Toronto. Voting would be one of the few uses of an identity card for a citizen abroad, since most other services would not be relevant to someone living permanently outside Estonia. There appears to be some hope that the popularity of I-voting among Estonians abroad will in future allow some savings by curtailing other methods of external voting, but no changes have been made as yet. The national government is very sensitive to any potential charges that the Internet voting system is being required of anyone, or even privileged.
Given the goal of increased turnout, or at least stabilization of voter participation, it is no surprise that considerable attention has been given to the results, now that five Internet elections have been held in Estonia. The most extensive study addressing this was carried out by Trechsel and Vassil in 2010. This was before the 2011 parliamentary elections, but involved the four previous elections. Their conclusion was that "turnout in the 2009 local elections might have been up to 2.6 percent lower in the absence of Internet voting" (Trechsel and Vassil 2010, 63).
Formal published evaluation reports by OSCE-ODIHR were issued after the 2007 and 2011 parliamentary elections (OSCE-ODIHR 2007, 2011). These reports evaluated the elections in general, but a substantial portion of the review was devoted to the Internet aspect. The main criteria used were the legality, security, transparency and observability standards commonly used by this international organization. It should be noted that the OSCE-ODIHR is currently developing a handbook for the observation of Internet elections (Krimmer, October 26, 2012). The security aspect of the elections was also mentioned as an evaluation criterion by Madise and Vinkel (2011), as was the fact that the election was open for observers (Madise and Vinkel 2011). Related to this openness was the transparency of the operations for audit purposes (Madise and Vinkel 2011).
An evaluation criterion not specifically set out in reports, but still in the background, is cost. Given the Estonian long-term commitment to I-voting, the costs are seen as reasonable, much less than for a paper voting system (Martens, October 15, 2012). Since the I-voting operation was and is an add-on method, there is no overall saving for it, and in fact there is an extra cost. However, the more people use it, the more fixed costs are offset. Martens said that there would be no consideration to reducing regular polling stations to reduce costs, since this might be seen as disadvantaging those who wanted to go to a regular station and might have to travel a longer distance.
Satisfaction with Internet voting as measured by public opinion surveys is another evaluation criterion. Trechsel and Vassil (2010) demonstrate high levels of public support for I-voting. As might be expected, this support is much higher among I-voters themselves, but those who vote at the polling place are also favourable, as are non-voters. This pattern of support was found in all four of the elections studied (Trechsel and Vassil 2010).
In the 2011 OSCE-ODIHR evaluation report, a recommendation was made in the security area, regarding the verification of voters' electoral choice (rather than just the verification that a vote was received) so that an assurance can be given that their vote was not changed by some "malicious software" (OSCE-ODIHR 2011, 13). As a result of this recommendation, the constitutional affairs committee of Parliament set up an expert working group to develop recommendations to implement this verification system. A law was recently passed to do this, and also establishes a permanent e-voting committee, which reports to the National Electoral Committee and is charged with running the Internet voting part of the elections. Both committees are served by the Department of Elections, which actually conducts the regular elections (Martens, March 26, 2013). Martens is the chair of the e-voting committee, which has seven members. It is an operational group, composed of experts on topics such as security and equipment, and is charged with running the I-voting component of elections. The validation system will be tested in the next local elections, and probably also in the European Parliament ones in 2014. It will work in the following manner: "After making your I-vote, a code will be displayed on the computer screen. You take your mobile phone, start a verification app and take a picture of the code. Then the phone will connect to the voting server for data exchange and the choice will be displayed on the mobile phone screen" (Martens, March 26, 2013). For test purposes, only users of one phone operating system, Android, will be used (Martens, November 6, 2012).
The evaluation of the verification system will be performed by the e-voting committee, which will issue a report. It will monitor public contacts received through support and complaint lines, and assess the overall performance. If public opinion polls are held, some questions could be added about public understanding and trust in the system. If these complaints are not substantial and there is a lack of technical problems, those would be the evaluation mechanisms.
Switzerland has embarked on Internet voting trials for almost a decade. Its main locus has been at the level of the canton, where Geneva, Zurich and Neuchâtel have conducted referendums, and some elections, since 2004 (with a few preliminary tests in 2003). In Geneva, the best documented of the three cantons, there have been 28 electoral opportunities to date where Internet voting has been permitted (Geneva 2012). Almost all of these have involved referendums, rather than election of representatives and parties. They have involved a varying selection of communes (municipal areas) within the cantons; at times these differed, but Geneva has been offering Internet voting in the same communes since 2008 (Chevallier, February 18, 2013). Electoral legislation originally limited the percentage of voters using the Internet to 20 percent of the citizens residing in a canton and 10 percent of the Swiss population as a whole; this limitation was implemented by restricting the number of communes eligible for Internet voting. As of 2012, the federal government raised the limit to 30 percent of the citizens in the same canton eligible to vote online (Geneva 2012), and Geneva has conducted three referendums in which citizens of all 45 communes within the canton were eligible to vote online. On these three occasions, in May and November 2011 and October 2012, approximately 240,000 citizens were eligible to vote online. Given the fact that the turnout in these Swiss referendums varied between 30 percent and 40 percent overall, the number of Internet voters was moderate. In the referendum with the highest overall turnout (40 percent in May 2011), 21.8 percent of all votes were cast online (Geneva 2012).
Switzerland has proceeded cautiously in expanding the use of Internet elections, as witnessed by the limits mentioned above of the numbers of citizens who are eligible. In October 2011, however, there was an Internet voting trial in the elections for the Swiss Federal Assembly, one of the few occasions where elections (not referendums) were the subject of the voting, and where the election took place in the whole of Switzerland. There was a limited trial in this election, whereby citizens of Switzerland living abroad and registered to vote in 4 of the 26 cantons were eligible to vote by Internet. These cantons, Aargau, Basel-Stadt, Graubünden and St. Gallen, used Internet voting systems developed for the three original trial cantons. Eligibility only for Swiss living abroad in those four locations meant that 22,000 residents were eligible to vote by Internet (OSCE-ODIHR 2012a). The percentage of those eligible who used the Internet to vote in that election was not immediately available.
The concentration of the federal electoral pilot of 2011 on the Swiss living abroad in four cantons exemplifies one of the major purposes of the Internet voting experiments in Switzerland: an effort to improve the circumstances for external voting. The Swiss Abroad is a formal interest group, headquartered in Bern, the federal capital. This group is of substantial importance (Serdült, November 6, 2012) and has put pressure on all Swiss governments to allow Internet voting for Swiss citizens residing outside the country. In order to vote, Swiss citizens permanently living abroad must register to receive their voting materials by mail. Traditionally, there have been problems in receiving voting materials early enough to complete and return the ballots, problems which may increase in future given the uncertain state of the world's postal services (Driza-Maurer et al. 2012). The Swiss Abroad have been successful in getting citizens living outside the country exempted from the percentage limits on the number of people in any canton who are eligible to vote online. The Swiss Abroad has become something of a hub for others interested in expanding Internet voting to work through (Serdült, November 6, 2012). The group worked collaboratively with a youth organization responsible for developing a youth parliament to organize petition drives in 2012 asking for the expansion of I-voting (Taglioni, November 7, 2012). The Swiss Abroad petition achieved 15,000 signatures and was presented to the Federal Council.
The fact that interest groups are important in making policy demands in the I-voting area is in keeping with the operations of Swiss democracy. Little original consultation is centrally organized, but commentary and demands are expected to emerge from civil society organizations. These are not regulated in terms of finance or advertising. They may create policy initiatives that are put to referendums, and referendums are also used for votes on government legislative measures. In 2009, Geneva passed a cantonal constitutional provision in a referendum with over 70 percent of the vote that authorized Internet voting.
Increased voting turnout was an important goal for the introduction of Internet voting in Switzerland. The introduction of the postal voting option, which occurred at various times in the Swiss cantons between 1978 and 2005, is estimated to have increased the voting rate by over four percentage points (Luechinger, Rosinger and Stutzer 2007). Postal voting has become so popular in Switzerland that the vast majority of votes are cast by this method. In the absence of an Internet option, up to 95 percent are cast by post, with the remaining 5 percent being dropped into the ballot box in the limited time that the polls are open. There were hopes that the introduction of an Internet voting option would accomplish a further increase in the voting rate (Chevallier, November 8, 2012).
An early analysis was very hopeful that the voting rate would increase in Geneva when Internet voting was introduced (Auer and Trechsel 2001) and later analyses continue to be optimistic (Gerlach and Gasser 2009). The Internet voting option is used, though at somewhat lower rates than when first available (Serdült, November 6, 2012). The elections in Geneva show that the Internet voting option generally constitutes between 15 and 20 percent of votes cast (Geneva 2012). In a statistical analysis of the November 27, 2011, cantonal votes, the communes within the city varied between 26 percent and 11 percent in their use of Internet ballots, with an average of 17.2 percent (Geneva 2012).
Similarly, the Internet voting option was seen as potentially appealing to young people, who vote at lower levels in Switzerland, as in most places where a turnout decline or low turnout rate is present. For example, in a tabulation of voting participation rates by age group, under-25-year-olds voted at less than half the rate of those in their sixties at four referendum opportunities in 2012. A study of the voting by means of the Internet in the referendum of November 27, 2011, however, showed that those in their twenties were only slightly below those in their thirties in the likelihood of using the Internet to vote. There is also some evidence that Internet voting particularly appeals to "occasional voters" (Chevallier 2009, 35) who also, of course, are more likely to be found in younger age groups.
Several other factors have led to the widespread acceptance of Internet voting in Switzerland, as exemplified by the 70 percent approval in the Geneva referendum mentioned above. The OSCE-ODIHR report in 2011 referred to the high trust that the remote voting system involving postal voting enjoys. There is also an underlying goal of improving Swiss performance in the whole information technology (IT) area, which given the democratic proclivities of the country seemed to fit together well with an I-voting operation. In particular, Geneva hopes to make itself an IT centre. Political science contributions to a feasibility study on this started as early as 1998 (Chevallier, November 8, 2012; see also Gerlach and Gasser 2009).
Switzerland has a formal consultation process between the federal and the cantonal governments. In fact, the Internet voting project (the "vote électronique") is called by the Federal Chancellery "a joint project between the Confederation and the Cantons." The co-operation is carried out by a working group of federal officials and those from the cantons. This working group is concerned with the exchange of best practices and with improving the voting experience for those with disabilities, particularly the blind, who may be better able to use an enhanced electronic voting computer screen than a paper postal ballot (Taglioni, November 7, 2012).
The federal-cantonal working group is following a vote électronique "roadmap" (Taglioni, November 7, 2012) that has five main areas of concentration: the establishment of a joint strategy between the confederation and the cantons, security, expansion, transparency and costs. One of the main areas of concern for this working group is security. A security subgroup has been established, which has instructions to develop a vote choice verification procedure that assures the I-voter that his or her choice has been correctly registered, rather than just a vote received. The leader of the vote électronique project, Geo Taglioni, believes that a successful resolution of the verification question will be a key to any expansion of Internet voting in Switzerland. The Swiss Federal Council will make an analysis of the trial period (2006 to 2012) and outline the development of the project. The third report was to be published in July 2013.
Switzerland has had a relatively sporadic set of evaluations of the I-voting trials. The Federal Chancellery produced in 2006 a lengthy report on the Internet voting pilot projects. This report emphasized the costs of the three trials in Zurich (contracted to global information technology company Unisys), Neuchâtel (contracted to global electronic voting company Scytl) and Geneva (which runs its own system). It examined the legal basis, the security aspect and the uses of the three trial systems.
The OSCE-ODIHR Election Assessment Mission Report for 2011 looked at the legality, security, management and testing of the Internet voting in the four test cantons, and recommended that future evaluations be undertaken by an independent body (OSCE-ODIHR 2012a).
In Geneva, after the referendum authorizing Internet voting was passed in 2009, an implementation law followed in 2010 (Chevallier, November 8, 2012). It established a Central Electoral Commission, which in turn charged the longstanding Commission for the Evaluation of Public Policy with implementing an evaluation of the I-voting law that was to take place within three years. Professor Pascal Sciarini of the University of Geneva is in charge of the evaluation, indicating collaboration between the university sector and the government.
Sciarini's evaluation study is currently underway, and will be evaluating the Geneva I-voting operation on four criteria (Sciarini, November 7, 2012). These are, first, the impact on participation, particularly turnout. The second criterion is the effects on subgroups, particularly the young, but also "selective" (occasional) voters, as well as late deciders. There is always a rush at the end of the advance polling period with postal ballots to mail the ballot so that it arrives in time to be counted. With Internet voting, late deciders can wait a little longer, up to noon on Saturday, whereas the mail ballots need to be mailed by Thursday (Sciarini, November 7, 2012). The election itself takes place on Sunday morning. Additional evaluation criteria for Sciarini are the characteristics and attitudes of Internet voters, and finally a model of the determinants of online voting, which will investigate, among other things, whether the effects of the I-voting system are neutral among the political parties or the referendum sides.
Norway is a relative latecomer to Internet voting trials, but even so it took seven years of planning. The Ministry of Local Government and Regional Development appointed a working committee in 2004, and received a report in 2006 (Nore et al., October 22, 2012). This committee, chaired by political scientist Bernt Aardal, was composed of a variety of experts including some political scientists, technical people and some from the local governments and the Ministry. Some were there for practical expertise and some because they were in a policy-making position. It encompassed a variety of positions of support and included some who were skeptical of the idea. This committee recommended a slow step-by-step approach, involving successive pilot projects, to deal with any technical problems and to build support for the reform (Aardal et al., October 23, 2012). Initially, the thinking was to have Internet voting take place at central locations, but it was then reasoned that any security breaches were actually less likely to be widespread if the voting took place from personal computers. That way, if viruses existed or security was breached, it was more likely to be limited to the individual computer, rather than a central server affecting many (Nore et al., October 22, 2012).
A variety of goals precipitated the initial interest in Internet voting in Norway. Accessibility was a major focus (Nore et al., October 22, 2012). Turnout was in the background, but the general literature consensus that Internet voting did not provide a big boost to turnout mitigated against it being made a major goal. There was also an aspect of efficiency in vote counting as a goal. The Norwegian electoral system includes not only a choice of party (for a proportional representation system) but also a choice of candidates within the selected party list. There are different systems for national, county and municipal elections. It is possible to vote for individual candidates in national elections, but in practice this has no effect at this level. At the county and municipal level, in contrast, personal votes partly determine which candidates are elected. In municipal elections, voters can also give personal votes to candidates from other party lists than the one they vote for. Although such a system (known as "panachage") may be counted by hand with difficulty (as it is in Switzerland), in Norway the ballots are often scanned in and counted by computer. It was hoped that Internet voting would improve the speed and accuracy of the counting process (Nore et al., October 22, 2012).
When the Aardal committee released its report in 2006, it was distributed to stakeholders and anyone interested, and there was a public hearing on it (Nore et al., October 22, 2012). The Ministry emphasizes that this is the normal process, and that it is open to anyone who is interested to get a copy of the report and participate by sending a comment on it. In 2008, a budget was given to the Ministry of Local Government and Regional Development to start preparing a trial. In 2009, there were two reference groups established. The first was a group of political party representatives and the second was a professional group, with some repeat members from the original Aardal committee, and others added for a total of about 35. The intention of the Ministry was to coordinate plans with both groups; there were presentations (by prospective vendors), a website, and a blog. However, it is the opinion of the Ministry interviewees that this consultative process did not work very well in practice, as there was little feedback, a shortage of time once the trial was decided on for 2011 and a need for more staff to organize the consultations.
The announced plan to go ahead with a trial in the 2011 local elections (for municipal and county councils) was debated in the Norwegian Parliament in November 2010 in the context of a private member's motion to stop the trial. Comments made by those in opposition to Internet voting mainly involved questions of its legality, the possible improper influence on voters where the process of voting from remote computers was not secret, the cost and the lack of evidence that it would improve voting turnout. Those speaking in favour mentioned the need for a trial to see whether there were positive effects, the expected improvement in accessibility, and the need to try to involve young people and keep up with the times (Parliament of Norway 2010).
The ten Norwegian municipalities that used the Internet in the 2011 trials are scattered around the country. After they were chosen from among those who applied, the Ministry further undertook a boat trip to consult with those along the coast. The local press was invited to meet the Ministry and local authorities at each stop to provide publicity about the upcoming event. The local municipalities were required to do a pre-pilot test of the system with some local question of interest on the ballot, but there were low turnouts to these (2–10 percent). Local politicians were also informed about the system, but there were no formal training courses (Nore et al., October 22, 2012).
Through a competitive bidding process, a contract for the evaluation of the Norwegian Internet voting trial was awarded to the Institute for Social Research (ISF) in Oslo. A component of the evaluation was also conducted by the International Foundation for Electoral Systems (IFES). The IFES evaluation was designed to ensure that the Internet voting met "internationally accepted norms and standards for democratic and electoral rights" (ISF 2010, 3). It looked at such things as the efficient counting and transparency of the results, and conducted focus groups with stakeholders and interviews with election administrators (IFES 2010).
The ISF report is available in detail only in Norwegian, but an English summary gives results of a public opinion survey undertaken at the time of the election. It concludes that the short-term effects of the trial on voting turnout are unclear and indemonstrable, but that "the people who voted online are very happy with Internet voting and report that it was easy to cast a ballot in this manner" (Ministry of Local Government and Regional Development 2012, 2). Interviews with participants with accessibility problems revealed that, though there were some operational issues, these voters with disabilities were extremely positive toward Internet voting. Trust in, and approval of, the system are reported as being very high, even among those who did not vote on the Internet. The survey also revealed that public opinion did not consider a lack of secrecy to be an important problem.
The OSCE-ODIHR also prepared an evaluation report on the Internet voting pilot project part of the Norwegian election of 2011. A number of implementation recommendations were made in this report but the general verdict was positive (OSCE-ODIHR 2012b). The report dealt with the testing and set-up of the system, the production of the polling cards, the voting, the counting, and the disposal of the data; recommendations to clarify these procedures were offered in the report. Several aspects of the security of Internet voting were also examined, and recommendations were offered in the report. It should be noted that a number of features of the Estonian Internet voting system were adopted here, including the repeated possibility to vote on the Internet, with each subsequent vote cancelling out the previous one. Also, an election day paper ballot would cancel any previous Internet votes. A major innovation in Norway was the use of a “return code” whereby the voter received a cellphone message giving a code that could be matched with the voter's personalized voting card to verify the party the voter had voted for. The operational success of these return codes has motivated demands in Estonia and Switzerland, previously mentioned, for the development of similar verification systems in those countries. The OSCE-ODIHR report recommended that a review of the return-code system be carried out. The IFES evaluation of the compliance of the Norwegian electronic voting in 2011 concluded that it was compliant with most of the 112 recommendations for the electronic voting standards (many of them technical) issued in 2004 (Segaard, Baldersheim and Saglie 2013).
There is still a division of opinion in Norway about the Internet voting method. A recent paper points out that the cleavage lines behind this difference of opinion are not necessarily predictable. The authors posit that this split falls along a centre-periphery dimension, with parts of the "national elite" in the capital and larger cities very doubtful and concerned about secrecy, and those on the ground in the municipalities actually undertaking the Internet voting extremely positive toward it (Baldersheim, Saglie and Segaard 2012). The municipalities felt this was a positive element in their image, and gave them the reputation of being more technologically advanced. The generally positive reception of the 2011 local election trials in Norway has resulted in the government announcing that further trials will be held in September 2013 during the national parliamentary election. The same locations will be involved, and two new ones added (The Norway Post 2012).