Switzerland & the Future of Elections
The Internet has had a profound impact on the way contemporary democracies work. Neither processes, such as electoral campaigns, nor actors, such as candidates, political parties or movements, are immune to the myriad challenges and opportunities offered by new media. The same goes for various fundamental institutions of democracy, such as parliaments and governments, which have adopted Internet-based strategies for both internal and external information and communication needs. Since vertical and horizontal forms of interaction among the elected and electors are increasingly impacted by the Internet, it is hardly surprising that one of the most fundamental acts of democratic life–voting–has been technologically upgraded in numerous countries.
Even prior to the deep societal transformations brought upon by the Internet, votes started to be cast electronically on voting machines, were transmitted to central election administrations by technological means such as telephones and fax lines, and were counted automatically. With the advent of the Internet, however, a more fundamental upgrading of the voting process to remote, Internet-based forms of casting the vote became a topic for many a government across the globe.
After an initial period of rather naïve enthusiasm, characterized by announcements of an imminent introduction of remote Internet voting for all citizens, the complexities of these projects became rapidly visible and, in particular, security concerns brought various projects to a halt. Initial trials in the United States, the Netherlands, the United Kingdom, Ireland, and elsewhere were stopped during the first decade of the 20th century. At the same time, successful experiences with remote Internet voting took place in two small European countries: Switzerland and Estonia. In Switzerland, three regional (“cantonal”) governments (Geneva, Zurich, and Neuchâtel) pioneered Internet voting under the auspices of the federal government. In over ten years of experience with Internet voting, ten more cantons have joined the early adopters. In the small Baltic country Estonia, since 2004 a member of the European Union, any citizen equipped with an electronic identification card can cast her vote in local, national, and European elections online since 2005. The continuity characterizing the Swiss and Estonian experiences stands in stark contrast with the “failed” innovations among the first movers. Today, Internet voting is gaining momentum once more, for instance in Canada and Norway.
With this contribution, we aim to track the history, promise, and challenges of Internet voting encountered by one of the two successful pioneers in the field: Switzerland. The Swiss case is of great interest to any student of Internet voting as it is to any policy-maker interested in the prospects, pitfalls, and solutions for avoiding the challenges of Internet voting within their respective contexts.
A Brief History of Internet Voting in Switzerland
Setting the Stage
Switzerland is a federal state comprising of 26 cantons, which have their own parliaments, governments, electoral systems, direct democratic institutions, fiscal resources, and say in most policy areas. The regularity with which Swiss citizens are asked to vote in direct democratic ballots–often four or five times per year–and the simplicity of these votes (yes/no/abstention) made Internet voting a strong candidate for exploration as part of Switzerland’s digital agenda. Consequently, the federal government decided in its 1998 “Strategy for an Information Society” to make use of this potential laboratory for innovation and offered cantonal governments the opportunity to take part in pilot projects aimed at introducing, testing, and evaluating forms of Internet voting.
In early 2000 and after the necessary adjustments of the legal framework at the federal level had been made, three cantons–Geneva, Zurich, and Neuchatel–volunteered to become part of these pilot projects and signed contracts with the Federal Chancellery to conduct legally binding tests of Internet voting systems during federal polling. The agreements included a series of basic ground rules as well as a commitment by the federal government to cover up to 80 percent of the extra costs associated with the Internet voting trials.
The first time Internet voting was introduced in Switzerland by a cantonal government was in January 2003, after a long series of technological, legal, and sociological studies as well as several tests of the system. The small commune of Anières in the canton of Geneva, near the French border and with only slightly over 1000 inhabitants, was chosen for the first trial. The object of this vote was a municipal referendum in which 44 percent of voters cast their vote over the Internet.
At least three factors contributed to Geneva’s role as a pioneer in Internet voting, which makes it today arguably the government entity with the richest worldwide experience in terms of binding decisions taken by the electorates. First, it has a centralized electronic voting registry, which has made authentication easier. Second, the legal framework authorized the cantonal authorities to experiment with new voting mechanisms in collaboration with the municipalities. Third, voters in the canton of Geneva already had a long-standing track record in distant voting. Together, these factors gave the canton of Geneva a head start in Internet voting.
In September 2004, after several additional tests in the context of municipal ballots, the canton of Geneva introduced Internet voting for the first time for cantonal and federal ballots in selected municipalities and expanded the number of participating municipalities steadily over the subsequent years. Similarly, the canton of Zurich started–after a series of tests–the Internet voting trials with a municipal vote, which took place in the town of Buelach in October 2005. One month later, citizens in several municipalities were for the first time given the opportunity to cast their votes over the Internet.
In the canton of Neuchatel, finally, the first votes submitted electronically in a federal ballot were casted in September 2005. Unlike the cantons of Geneva and Zurich, the canton of Neuchatel did not limit the availability of the Internet voting trials to particular municipalities, but introduced a cap, determined by the federal government, on the overall number of permitted Internet votes. The reason for this cap was primarily guided by the experimental character of Internet voting in these early stages: by imposing such a cap, the probability of the overall outcome of the referendum vote being affected by a malfunctioning of the Internet voting system could be minimized.
Although the three pioneering cantons have continued to offer Internet voting ever since, the series of pilot projects coordinated at the federal level concluded in 2006 with a report by the federal government, in which it concluded that the tests had been successful and that there was neither a reason to stop further tests nor reason to extend them.
Expanding the Experiment
Soon after the initial trials concluded, which were targeted at voters living in Switzerland, the federal government acknowledged that Swiss citizens living abroad as a core constituency for Internet voting and decided to move forward and provide this service to citizens who live outside Switzerland. One year after an amendment of the federal law that governs voting by Swiss citizens living abroad in 2007, the canton of Neuchatel was the first canton allowing its registered citizens living abroad to vote over the Internet. This marked the first phase of significant expansion of Internet voting beyond the initial trials.
Another milestone in the evolving history of Internet voting in Switzerland was met in November 2009, when the canton of Basel-Stadt enabled all citizens who were registered in Basel-Stadt, but living abroad, to exercise their political rights online. Basel-Stadt’s entry marked the second wave of expansion beyond the initial trials, with several other cantons joining shortly thereafter.
A third important step in expanding the scale of Internet voting took place in November 2010, when the federal government approved for the first time the right of citizens of twelve cantons to vote via the Internet voting system of their respective canton in the context of a federal ballot. It should be noted that federal law currently sets an Internet voting cap of 10 percent for the federal electorate and 20 percent for the cantonal electorate (with certain exceptions for Swiss living abroad).
The success of internet voting trials prompted Swiss authorities to provide the option of Internet voting to a majority of out-of-country voters by the 2015 federal elections. In October 2012, Internet voting was added to the “Action Plan 2013” of Switzerland’s e-government strategy. This move signaled the federal government’s political and probably financial support for cantonal Internet voting systems development. Tellingly, the plan of action aims to enable Internet voting for all eligible voters.
Switzerland is among the pioneers of Internet voting. Over the past ten years of experimentation, a relatively rich body of experiences has accumulated, from which several observations and lessons learned can be distilled. The findings from these studies have resulted in a series of important recommendations, which will inform the implementation of the future Internet voting infrastructure. In the following section, we do not seek to provide a comprehensive overview of the lessons learned, but would rather like to share three high-level observations that might be of particular interest to an international audience.
Observation 1: The Swiss case demonstrates the benefits of a coordinated, but decentralized and bottom-up approach to the introduction of Internet voting.
As described in the previous sections, Switzerland has taken in incremental and bottom-up approach to introducing and testing Internet voting. The coordinated, but decentralized approach driven by cantons does not come as a surprise to anyone who is familiar with the Swiss political system and culture and, in fact, was a necessity vis-à-vis Switzerland’s 26 cantons with different voting systems, which also vary at the municipal level. The different cantonal approaches have resulted in a well-documented set of technologies, policies, and practices tailored to different circumstances. Three systems emerged:
•The Geneva system has been developed and managed by the canton of Geneva and is particularly suitable for cantons that have a central electoral roll or voter register. It has been used by the cantons of Basel-Stadt, Lucerne, and Berne.
•The Zurich system is arguably the most sophisticated model from an organizational and technical angle as it was designed for a canton that follows a decentralized approach to the management of elections. The system has been developed and managed by the private company UNISYS for the canton of Zurich, which has licensed its software since 2010. Today, the cantons of Aargau, Fribourg, Graubünden, St. Gallen, Schaffhausen, Solothurn, and Thurgau are part of a consortium building upon the Zurich system (consortium system). However, the consortium system, unlike the Zurich system, is currently limited to Swiss citizens living abroad.
•The Neuchatel system has been part of a broader digital initiative called “Guichet Unique”, which can be used – as a “virtual government window” – for a broad range of transactions, including tax management, license plate applications for cars, etc. Given the embedded character of the Neuchatel Internet voting system, it can currently only be used by this canton.
The fact that cantons can pick and choose from at least two systems, depending on the local conditions and requirements, when introducing Internet voting to their citizens is one of the key benefits of the decentralized approach Switzerland has taken. But the approach has admittedly resulted in a lower level of cost efficiency and a lack of standardization, common practices, and collaboration when compared to an alternative top-down approach such as the one adopted in the Estonian case.
Observation 2: It is too early to assess the impact of Internet voting on participation – but initial numbers from Switzerland do not give reason for enthusiasm.
Swiss voters are frequently called to the polls–at least four times a year. At the same time, Switzerland has one of the lowest levels of voter turnout among established democracies. Various studies have explored why only relatively few citizens participate in the political process through voting. One of the factors identified relates to the costs associated with access to the ballots: the higher the transaction costs, the lower the likelihood of participation. This hypothesis has been supported by the introduction of postal voting as one mechanism to reduce transaction costs, which significantly increased voter turnout rates—according to some studies a double-digit increase. Against this backdrop, the hopes have been high in Switzerland that the introduction of Internet voting would have similar effects, especially as surveys indicated that convenience was among the factors that citizens associated with Internet voting.
A few years into the Swiss experiment, a number of observations can be shared. First, it is interesting to look into the adoption rates of Internet voting, which was introduced in addition to all other forms of (remote) voting. While data are limited, recent experiences from the canton of Geneva suggest that Internet voting is particularly popular among Swiss living abroad with about half voting via Internet, whereas Swiss residents still have a preference for postal voting. This picture is also confirmed by earlier data from the Zurich pilots, where about 20 percent of the votes were collected via the Internet, as compared to 65 percent making use of postal voting. Another interesting fact relates to demographics: statistics from the cantons of Neuchatel and Zurich indicate that Internet voting–contrary to expectations–has not been particularly popular among younger populations and was actually more heavily used in the canton of Neuchatel by people aged 30 to 65. Early studies from the Geneva trial suggest that higher levels of computer literacy, faster Internet connections, and higher levels of trust in Internet communication are some of the key factors that make it more likely for a voter to prefer Internet voting over alternative forms of participation.
Similarly and secondly, the Swiss experience to date is quite sobering with regard to question whether Internet voting increases voter participation. In the canton of Zurich, where statistics are available, the introduction of Internet voting has not led to an increased turnout. Earlier studies from the Geneva trials confirm this picture. While these indicators are arguably not particularly encouraging, it is in our view too early to draw final conclusions regarding the impact of Internet voting on participation. In many respects, we are at the beginning of history, not its end, and as Internet voting grows more mature and might get fully embedded into our personal digital environments–including voting apps and social media–the long-term impact might look different from what the short-term experience would suggest.
Observation 3: Security and transparency issues remain the key challenges of Swiss Internet voting systems.
As noted earlier, Internet voting builds upon and contributes to a complex ecosystem of legal, technical, organizational, and even cultural factors. It needs to be considered, designed, and improved in the context of a more comprehensive transition towards e-government services. Given this complexity and given what is at stake, Internet voting for everyone is an unlikely vision. Switzerland has taken a cautious approach by introducing and testing Internet voting step-by-step in a highly controlled environment. With this approach, it has produced a significant body of data, technical know-how, and expertise.
Despite its relative success, a number of challenges remain. Even if risk assessments are likely to change over time, the security and integrity of the Internet voting infrastructure is a continuing concern. Recent reviews of the Internet voting systems—other assessments are underway—have revealed two potential points of failure in particular. The first set of security challenges relates to the process of voting. A 2012 assessment, for instance, criticized that the materials used for voter identification, which are delivered to a voter via mail (with recent experiments in the canton of Neuchatel sending the codes via text message), were produced by a single person with minimal oversight. Weaknesses have also been identified with regard to encryption and the transmission of electronic votes. The “consortium system” encrypts the votes once the servers receive them, while the Geneva system has a higher level of security by encrypting the vote on the voter’s computer before it is transmitted.
The second set of security problems that was diagnosed relates to the management of cryptographic material, especially access to the private keys used to decrypt the electronic ballot. The report recommends Switzerland implement a number of measures to increase the security level on this score, including recommending that the private key be generated at a public meeting, be divided into separate parts, and shared by at least two people “who are unlikely to collude.” A series of academic studies recommend additional, including more fundamental, changes to the Internet voting infrastructure in Switzerland.
A second cluster of challenges that emerges from the Swiss experiments with Internet voting is related to transparency. While security and secrecy are important factors for the individual voter, Internet voting systems need to be transparent in order to be trusted by citizens. Swiss scholars have pointed out the importance of sharing information about the security mechanisms, organizational processes underlying the voting system, software and program code, and evaluation reports with the public.
In this respect, Switzerland is not a leader in the field when compared, for instance, to the more transparent approach Estonia and Norway have taken. Especially in the context of the “consortium system” experts have raised transparency concerns: the strong involvement of a private company and the debate around the topic of proprietary versus open source software were at the base of these concerns.
In this contribution we have retraced the introduction and gradual diffusion of Internet voting in Switzerland. As an early adopter with a continuous experience at hand, the Swiss case can offer important insights to policy makers elsewhere. Instead of summarizing the already highly condensed observations contained in the preceding pages, we propose to emphasize three additional points: first, the feasibility of Internet voting and its effects cannot be assessed in isolation from the institutional context. Second, to be successful, the introduction of Internet voting must be treated as a process rather than an abrupt and potentially disruptive event. Third, the seemingly banal but often underestimated importance of the political will and cross-partisan support for the introduction of Internet voting is essential for its success.
Regarding the first point, we sustain that the general voting system with its electoral administrative processes is a strong conditioning factor for the success of Internet voting. The latter has to deal with a diverse set of challenges, ranging from security to transparency issues, which need to be addressed at the technical, organizational, and legal layer. The update of the legal infrastructure in the case of Switzerland, for instance, required significant amendments to federal and cantonal laws. At the same time, policymakers wisely refrained from enacting strongly detailed provisions, opting for an enabling and principled approach instead. The same goes for the technology adopted: the three pilot cantons could develop their own systems that interacted with the local legal norms and requirements. Through such a decentralized and flexible approach, fostered by the federalist structure of the Swiss polity, the interaction between the legal system and technology, but also between the public and private sector could shaped according to the needs and opportunities of the local context.
Our second point emphasizes the gradual, incremental approach to innovation in the realm of electoral technology. In Switzerland, the general understanding prevails that Internet voting should be considered as “work-in-progress.” We strongly believe that this pragmatic perspective, opposed to a perfectionist’s approach, is a condition sine qua non for the success of Internet voting. The intersection between technology and democracy is itself conditioned by the larger societal context that is constantly evolving. The prudent but dynamic approach chosen for the gradual implementation and diffusion of Internet voting in Switzerland was clearly more successful than any “big bang approach.”
Finally, the consensus-fostering political style of Swiss policymaking allowed for a large, cross-party political coalition favoring the experimental introduction of Internet voting at an early stage. A clearly visible political will at all levels of the state, coupled with a measurable desirability for Internet voting within society, is a powerful ingredient for success. It is also a necessary condition for necessary developments and future adaptations of Internet voting.