A volume in the Advances in Electronic Government, Digital Divide, and Regional Development (AEGDDRD) Book Series

Transcription

1 A volume in the Advances in Electronic Government, Digital Divide, and Regional Development (AEGDDRD) Book Series

2 Managing Director: Production Editor: Development Editor: Acquisitions Editor: Typesetter: Cover Design: Lindsay Johnston Jennifer Yoder Allison McGinniss Kayla Wolfe Lisandro Gonzalez Jason Mull Published in the United States of America by Information Science Reference (an imprint of IGI Global) 701 E. Chocolate Avenue Hershey PA Tel: Fax: Web site: Copyright 2014 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by any means, electronic or mechanical, including photocopying, without written permission from the publisher. Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark. Library of Congress Cataloging-in-Publication Data Design, development, and use of secure electronic voting systems / Dimitrios Zissis and Dimitrios Lekkas, editors. pages cm Includes bibliographical references and index. ISBN (hardcover) -- ISBN (ebook) -- ISBN (print & perpetual access) 1. Electronic voting. 2. Electronic voting--security measures. 3. Voting--Technological innovations. 4. Votingmachines--Technological innovations. 5. Political participation--computer network resources. I. Zissis, Dimitrios, II. Lekkas, Dimitrios, JF1032.D dc This book is published in the IGI Global book series Advances in Electronic Government, Digital Divide, and Regional Development (AEGDDRD) (ISSN: ; eissn: ) British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library. All work contributed to this book is new, previously-unpublished material. The views expressed in this book are those of the authors, but not necessarily of the publisher. For electronic access to this publication, please contact: eresources@igi-global.com.

3 Susan Collard University of Sussex, UK Elodie Fabre Queen s University Belfast, UK This chapter presents an analysis of the unprecedented use of electronic voting by expatriates during the French 2012 legislative elections, when they elected their own representatives (referred to here as deputies ), to the National Assembly in Paris for the first time, in 11 newly created overseas constituencies. The study is presented within the broader perspective of electronic voting in France more generally, and in the historical context of extra-territorial voting by French expatriates. The authors discuss the main issues and controversies that arose during the 2012 elections, and in a final section analyse the results. The authors conclude by drawing attention to recent developments in electronic voting in France since the 2012 elections, which suggest that although there was much criticism expressed by experts of electronic voting as to the security and transparency of the system used, the official discourse that acclaimed the experience as a success, appears to have convinced its target audience. Until its introduction for the election of eleven deputies for newly created overseas constituencies for expatriate citizens in the legislative elections of 2012, electronic voting had barely been used in France. Debates and pilot schemes had previously focused on the use of electronic ballot boxes (sometimes known as voting machines or machines à voter in French), located in the polling stations. These derived their legal existence from a law passed in 1969 by the Minister of the Interior, Raymond Marcellin, allowing the introduction of mechanical voting machines in order to reduce electoral fraud. These machines had failed to deliver any improvements and had fallen into disuse, but the legislation was modified in 2003 to allow for the use of electronic ballot boxes, following on from developments in technology and the increasing use of electronic voting in other countries ( Machines à voter, 2003). This legal modification took account of three test schemes DOI: / ch009 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

4 conducted during the presidential elections of 2002, which were run in parallel to the traditional voting method, in Mérignac (Gironde), Vandœuvre-Les-Nancy (Lorraine) and in the 18th district (arrondissement) of Paris. Vandœuvre-Les-Nancy repeated the exercise for the legislative elections the same year. In September 2003, the association Forum for Internet Rights (FDI), founded with public funding, made recommendations to the then Minister for the Interior responsible for elections, Nicolas Sarkozy, regarding the future of electronic voting in France and the optimal conditions for its successful implementation ( Recommandations du Forum des droits sur l Internet, 2003). This report recommended that for all political elections, electronic voting should only take place using electronic ballot boxes located in polling stations, but that electronic voting from a distance (from private PCs) should be allowed for expatriate citizens voting in the elections for the Conseil Supérieur des Français de l Etranger (CSFE), which in turn elected 12 members of the Senate (see below for more details). Debate then began to polarise between supporters and opponents of electronic voting. One of the leading voices of the opposition camp was a newly created association called Recul démocratique (now called Ordinateurs de vote ), which organised a petition in a presidential election year - against the use of electronic ballot boxes, obtaining 85,000 signatures (now 107,161), on the grounds that they represented a risk to democracy ( Pétition pour le maintien du vote papier, 2007). In 2008, a book highlighted in detail all the potential technical problems, notably bugs, system failures, undetectable fraud, hacking, and the exclusion of citizen control over the ballot boxes (Perline & Noisette, 2008). Political parties started to take positions, with those on the Left and the ecologists leading the opposition, and the Right-wing UMP led by Nicolas Sarkozy being in favour. The Centre-Right mayor of Issy- Les-Moulineaux, André Santini, was another key supporter, who argued that electronic voting was the best and most progressive way of fighting fraud, saving time, and eliminating errors ( Vote électronique, 2009). He controversially used compulsory electronic voting in his municipality in the presidential and legislative elections of 2007, causing much public debate, and his detractors accused him of being in the pocket of the manufacturers of the electronic voting equipment. It was reported that 29.8% of the polling stations presented anomalies in the vote count in Issy-Les-Moulineaux, compared to an average of 5.3% for traditional voting, and discrepancies in other municipalities using electronic voting were also highlighted (Nymette, 2011). The formal response of the state to the issues raised came from the Constitutional Council, which published press statements in March and April 2007 confirming the legality and security of electronic ballot boxes, based on the 1969 law, but also acknowledging that long queues had built up in polling stations where they were in use (Conseil Constitutionnel, 2007a). In its final report on the elections in May, it reported the easing of this problem in the second round, and it made suggestions for increasing the number of electronic ballot boxes in future (Conseil Constitutionnel 2007b). It also discussed the reasons for the widespread opposition to this voting method, which it described as more psychological than technical, concluding that the dematerialisation of the act of voting had dispossessed citizens of a traditional ritual, causing a sense of loss for which no amount of technical reassurance would be able to compensate. It is against this background of partisan controversy that this chapter takes as its main focus the analysis of electronic voting in the 2012 legislative elections in the eleven newly created overseas constituencies for expatriate citizens. We will first set this case study into the broader context of the history of expatriate voting for French citizens, explaining the systems used for extra-territorial voting in countries of residence from 1981 onwards. We then discuss the main

5 issues and controversies that arose concerning the use of electronic voting during the 2012 elections, and in a final section we analyse the results before drawing our conclusions. In order to understand why electronic voting was introduced for French expatriates in the legislative elections of 2012, it is necessary briefly to review the history of voting rights for French citizens resident abroad. Historically, the French state has had a rather ambivalent attitude towards its expatriates: after the Revolution of 1789, the exile of many aristocrats known as émigrés meant that they were considered as traitors, and expatriates developed from this legacy an unfavourable reputation. But when France developed a strong colonial presence from the 19th century onwards, the state adopted a more positive view of those citizens who took an active part in France s civilising mission that underpinned the colonial expansion of the Third Republic in the late 19th and early 20th centuries. The newly founded republican administration of this period set out to develop a strong sense of nationalism amongst its citizens, built on a close identification between nationality and citizenship. Consequently, a law was introduced in 1913, which allowed those (male) French citizens who lived overseas to retain their right to vote which had previously depended on residence in metropolitan France. They nevertheless still had to either vote by proxy (necessitating validation in person at the nearest Consulate), or by returning to France to actually exercise the right to vote, both of which meant that the new law had little practical impact. It was amongst this colonial population of republicans overseas that campaigning began for postal votes though the consular network, on the same lines as those granted in 1924 to civil servants sent to work in occupied regions of Germany (Garriaud-Maylam, 2010, p.107). This issue of postal voting was the driving force behind the creation in 1927 of the association of French expatriates, the Union des Français de l Etranger (UFE), which was supported in this endeavour by three other associations representing the key components of the French presence overseas: teachers (la Fédération des professeurs français à l étranger: FPFE), war veterans (la Fédération des anciens combattants français résidant hors de France: FACS), and chambers of commerce (l Union des chambres de commerce françaises à l étranger: UCCFE). The development of these four associations fostered a sustained dialogue between the republican state and representatives of the French expatriate community around the world, whose determination to be heard was a key factor in establishing their formal recognition. Indeed, one of the specificities of the French state s attitude towards its expatriates has been the quest to maintain communications with this distant population, monitored closely using registers which are still essential to the functioning of the contemporary consular network. Their claims for political representation were however not heard until after the Second World War, and even then, only partially. French expatriates had played a critical role in the Resistance and the Liberation, and two of their high profile representatives were members of the Constituent Assembly, which met to draw up the constitution of the Fourth Republic. They achieved two things: first, the right to retain their civic and political rights, regardless of their place of residence, was enshrined in Article 4 of the Fourth Republic s constitution of 1946, and second, they were given political representation through the Senate, then called the Conseil de la République. But they did not win direct representation to the National Assembly)as has now been established in the 2012 legislative elections), and they were not granted the right to vote by post from overseas, even though certain categories of mainly military personnel were allowed to do so. Instead, French expatriates were given three dedicated Senators out of a total of 320, based on geographic criteria: one for

6 Europe, one for America and one for Asia-Oceania (Africa was the subject of separate agreements due to colonial arrangements). The Senators would at first be appointed by the National Assembly, though the four associations were allowed to put forward their own candidates for nomination. They also obtained further representation in 1948 through the creation by the Ministry of Foreign Affairs of the Conseil supérieur des Français de l Etranger (CSFE), consisting of elected representatives from the French expatriate community. This body would play a consultative role to the government. It consisted of 55 members, including the 3 Senators (Conseillers de la République), the president and director of the UFE, and the three presidents of the other associations; five more members were appointed by the Ministry for their reputations and competence relating to French expatriates, and a further 42 were elected by the members of the four associations. The first elections to the CSFE took place in 1950, in 70 countries straddling the three zones represented by the Senators. The right to vote in these elections was dependent on registration on the consular lists and on membership of one of the associations. By imposing these two conditions, the state hoped to encourage their grouping together into identifiable communities and to obtain information about its expatriate citizens. The CSFE continued to press for the introduction of voting by postal vote, or even at the ballot box in the consulates themselves, but a range of practical difficulties, as well as fear of fraud and problems relating to the potential alleged infringement of the sovereignty of certain host countries, were invoked by the state against these ideas (Garriaud-Maylam, 2010, p.111). With the establishment of the Fifth Republic in 1958, Article 24 of the new constitution stated that The French established outside France are represented in the Senate, and the number of senators was increased to six, with three new ones representing Africa. In the wake of further decolonisation in the early 1960s, the imbalance between the continents was addressed by the addition of three more senatorial seats, making a total of nine. The CSFE, whilst retaining a purely advisory role to government on issues of particular concern to expatriates (school fees, access to consular services, tax and inheritance complications, access to health care and social benefits, and questions of security in certain countries) became the sole electoral college for the senators and its membership increased to 84. It was not until that laws were introduced establishing polling stations in the consulates of consenting countries, enabling expatriates who chose to do so to vote in their country of residence in national referendums, presidential elections and elections to the European Parliament (EP). These European elections took place in 210 overseas polling stations for the first time in Expatriate voting in legislative elections was still either by return to France or by proxy, under complex regulations which generated accusations of partisan manipulation, and which posed difficulties for those who no longer had close relations in France to act as proxies. The motivation behind these new laws permitting extra-territorial voting for certain elections was partly a response to longstanding demands from expatriates, but there were also political considerations, since this population was widely reputed to be favourable to the Right, who were in power at the time. The proof of this came in the presidential election of 1981, when the Right-wing candidate Valéry Giscard d Estaing won a massive 70% of the overseas vote in the second round, compared to 30% for the socialist candidate François Mitterrand, who nevertheless won the election overall with 51.76%. The election of a socialist president inspired a group of left-wing expatriates to create an association which would challenge the UFE, which they saw as a covertly right-wing association. Indeed, although the associations of expatriates were supposedly apolitical and non-partisan, the fact that all the senators had affiliations with parties of the Right belied a rather different reality. The Association Démocratique des Français à

7 l Etranger (ADFE) was thus created in November 1980, based on a more progressive vision of the expatriate community, committed to the values of the broad republican Left, the universality of Human Rights, democracy, secularism and opposition to all forms of discrimination. It was instrumental in persuading President Mitterrand to include the promise of representation in the National Assembly in proposition 48 of his manifesto, though this pledge was not honoured once in power (unsurprisingly, given the election result). It also engineered a reform of the CSFE in 1982 which introduced the direct election (including by postal vote) of its 155 members by all expatriate citizens rather than indirectly through the auspices of the main associations, a system which had been strongly denounced as undemocratic by a group of socialist senators in A further reform in 1983 confirmed the directly elected CSFE as the electoral college of the senators representing expatriates, whose number was increased from nine to twelve, as part of a complex internal negotiation relating to the political balance of power within the membership of the CSFE, designed to allow the Left a greater chance of electing their own senators (M. Cerisier ben Guiga, personal communication, June 13th, 2012).1 This happened for the first time in 1983 with the election of Jean-Pierre Baylé, followed by that of Guy Penne in 1986, and a subsequently regular flow of socialist senators confirmed the end of the domination of the Right over these seats: since the senatorial elections of September 2011, there are four socialists and eight from the UMP. The creation of the ADFE thus brought about a more overt politicisation of the representation of the expatriate community both in the Senate and in the CSFE, as it became bi-polarised on the model of national politics, despite both organisations officially maintaining the fiction of having no formal links with any political party. One of the most significant but unanticipated side effects of this development was a gradual reduction in the level of participation in the elections for the CSFE, which dropped to 24% in 1997, and 19% in This high level of abstention prompted, under the left-wing Jospin government, the creation of a commission to look into reforming the CSFE, which came about in after the Right had returned to power. It was renamed the Assemblée des Français de l Etranger (AFE), and the electoral map on which the election of its members was based was modified to take account of demographic changes. The number of unelected members was reduced, but no new powers were granted and its attempt to be recognised as a territorial collectivity was not taken up. In an attempt to increase participation in these elections, electronic voting was used in a pilot scheme in 2003 in the two districts in the USA where voting was taking place; the turnout remained very low at 15%, of which 61% voted by Internet, 34% by postal vote and 5% in the polling stations ( Le vote par Internet, 2006). The experiment was nevertheless repeated in all areas where elections were taking place in 2006 (Europe, Middle East and Asia): only 6% of voters (28,000) registered to vote by Internet, of which only half (14,000) confirmed their electronic registration, with only 10,201 finally using their vote (Ibid). The two main associations both commissioned reports into this use of the electronic vote (Pellegrini Report for the ADFE and Lang Report for the UFE) and both drew damning conclusions, particularly with regard to the verifiability, transparency and sincerity of the vote (Pellegrini, 2006; Lang, 2006). The Pellegrini Report (2006) recommended use of the postal vote for expatriate voters rather than electronic voting. Another report in English by Andrew Appel, an American professor of Computer Science at Princeton University who was in France in 2006, focused on the role and training of the assessors in these elections, concluding that The clear consensus of computer science experts around the world is that Internet elections cannot be trusted the voters and political parties cannot audit the operation of the software and hardware that serves as the real bureau de vote (Appel,

8 2006, p. 9). But despite these reports, electronic voting was once again used in the AFE elections of 2009 and 2010, and a decree was issued under President Sarkozy in 2009 authorising more general use of the electronic vote, because it was seen as progressive, cost-effective and a practical solution to the specific case of expatriate voting. The main reason for the extension of electronic voting must be understood in the context of the decision by Sarkozy s government to introduce a major reform of political representation of its expatriate citizens, giving them directly elected deputies in eleven newly created constituencies abroad. These new global constituencies were created during the general redrawing of constituency boundaries carried out in 2009, and this major institutional innovation was one of a package of reforms that were the subject of the major constitutional revision passed by the Congress (National Assembly and Senate) in Sarkozy could justify this initiative as a response to long-held demands by the French abroad for better political representation, as we saw earlier. But it was also the case that he was sensitive to the growing size of the expatriate community (Duchêne-Lacroix, 2005), as well as to the fact that it was reputed to be sympathetic to the Right, as confirmed in election results since The reform caused considerable political controversy, with the Leftwing opposition claiming that it was simply a manoeuvre by the Right to win more seats in the National Assembly (Collard, 2013). But despite a robust challenge to the government s proposals, the opposition failed to prevent the passage of the Figure 1. Map of French overseas constituencies and number of voters on electoral register 2010, Guardian News & Media Ltd. Used with permission.

9 bill which was finally passed in April 2011, and in the following month a State Secretariat in charge of the French Abroad was created. Once the political decision had been taken to introduce overseas deputies, the question was posed as to how to ensure a reasonable turnout. Indeed, although the number of registered voters has continued to increase since 1981, the rate of abstention has also risen, as can be seen in Figure 2, which charts the evolution of turnout among French expatriates registered to vote abroad and compares it to overall turnout in France. It shows the considerable increase in the number of French people registered on the consular electoral list (LEC) and who chose to vote abroad in presidential elections.2 However, as the number of registered voters increased, turnout decreased quite considerably. This graph shows that whereas turnout in presidential elections has remained relatively stable across France, turnout among expatriates decreased from 75-78% in 1981, which was a level close to that of the rest of France, to just above 50% in In the last three presidential elections, turnout has averaged at around 40%, half of the participation rate of France (displayed here as an average of the two rounds). This suggests that the main driver for increased registration was a large increase in the number of ex-pats, whose registration to vote abroad may not always have been fully thought through, since the request to vote simply involves ticking a box as part of a more general registration process which is a requirement for all French citizens abroad seeking access to other consular services (such as renewal of identity papers, marriage certificates, etc.). The graph illustrates the fact that expatriates are an electorate that is difficult to mobilise, which explains in large part why e-voting was introduced in the new overseas parliamentary constituencies. Thus, since participation rates for presidential elections amongst expatriates had typically been about half that of Metropolitan France (around 40%), it could be anticipated that if the same ratio was reproduced for the legislative elections, for which national participation rates were in decline Figure 2. Number of registered voters abroad and expatriate turnout in presidential elections,

10 since 1978, the turnout may not even reach 30%. The main problem in the new overseas constituencies would be that of proximity to a polling station, especially in those constituencies such as the 11th (Asia and Oceania), covering vast swathes of territory. Although the number of polling stations across the globe was increased from 583 in 2007 to 783 in 2012, there would clearly be many thousands of eligible voters for whom voting in person was simply not practical, hence the importance of electronic voting for these elections. From the perspective of the consular network, the challenge would be to organise two rounds of voting offering four different methods of voting, as decided by the government in order to maximise turnout: in person, by proxy, by post or electronically ( Rapport du directeur des Français à l étranger, 2012, pp ). For the central administration in Paris, the task would involve first putting in place a legal framework to secure the implementation of the 2008 constitutional law creating the new constituencies. Second, it had to establish the necessary electronic infrastructure to enable electronic voting to take place. The administration responsible for French expatriates is the Direction des Français à l étranger et de l Administration consulaire (DFAE), which is part of the Ministry of Foreign Affairs (MAE), but since elections come under the responsibility of the Ministry of the Interior, considerable inter-ministerial cooperation was needed. Clearly, this whole operation constituted a major administrative operation for the DFAE, involving considerable new resources at a time of increasing budget austerity, and a special Steering Committee ( comité de pilotage ) was constituted to oversee the various tasks involved, including the introduction of electronic voting. It conducted an extensive evaluation of security issues, using the EBIOS (Expression of Needs and Identification of Security Objectives) method for assessing risk management ( EBIOS 2010, 2010). The coordination of this exercise was carried out in-house by teams from the DFAE and the MAE responsible for the security of Information systems (RSSI). Independent auditors were also invited to assess the systems before, during and after the vote, in accordance with the recommendation by the National Commission for Information Technology and Liberties (CNIL), which updated its guidelines regarding privacy and security requirements for Internet voting in a deliberation of October 2010, after input from an independent auditor. The CNIL expressed reservations over the use of electronic voting for political elections, and made several detailed recommendations relating to the need to introduce a number of technical changes to ensure greater security in the organisation and verification of the vote ( La CNIL met à jour la recommandation vote électronique, 2010). At the end of the Steering Committee s assessment, some changes were indeed introduced to minimise risks, and these were all minuted in a report. The final version of the software was digitally signed in April The provider of the server and software was the group ATOS-SCYTL, selected after a tender open to all European companies: ATOS is a French company which provided the server, located in a bunkerised data centre at Vendome in France, and SCTYL is a Spanish company which provided the software, Pynx, which was used in the AFE elections of 2009 and 2010, and which the DFAE bought in The choice of provider was determined as a result of what is described as a competitive dialogue organised by the DFAE ( Choix du prestataire informatique, 2012). Four companies were pre-selected out of twelve responses, and the successful companies were selected because they offered the most effective system at the lowest cost. The technical specifications ( cahier de charges ) were agreed at interministerial level, as was the risk analysis. The contract with SCTYL stipulated that it could not make available online the source code, and no company was able to offer an open source system which met the technical specifications. Also, a

11 decision was taken not to publish the application code for reasons of security (fear of hackers) but rather to have it controlled by independent auditors The overall supervision of the electronic voting was carried out by the Electronic Voting Bureau (BVE), an independent organisation presided by a Councillor of State (a senior civil servant); it comprised a representative of each of the three administrations concerned: the Ministry of Foreign Affairs (MAE), the Ministry of the Interior (MININT) and the National Agency for the Security of Information Systems (ANSSI)3, as well as three elected members of the Assemblée des Français de l Etranger (AFE), making a total of seven members. The process was also observed by a delegation from the Organisation of Security and Cooperation in Europe (OSCE, 2012) and by observers from Switzerland, Norway and Russia. Meetings were open to candidates, proxies, and citizens from the eleven new constituencies, and minutes of all meetings were made available to the OSCE / ODIHR, and after considerable discussion, also to voters and proxies. Once these systems were in place, a massive communication campaign was launched to inform French expatriates of the new arrangements and to encourage them to register to vote. The consular network undertook a major operation to update its electoral registers and to request individual addresses and mobile phone numbers. This was necessary because identifying numbers and passwords would be sent by text and to all those who registered to vote electronically. ( addresses were also used by candidates to send election campaign propaganda, much to the annoyance of most voters, leading to complaints to the CNIL). At the same time, advertising was diffused on media networks TV5 Monde, France 24, Radio France Internationale (RFI) as well as on Google, Facebook and in the French national press. A first campaign took place in autumn 2011 and a second in the Spring of 2012, informing French citizens of the various methods of voting in the upcoming elections. A new consular Website was also launched in November 2011, MonConsulat. fr, which proposed the possibility of electronic voting. It offered a promotional video showing French citizens how to update their personal details online so as to be able to use the electronic vote. In January 2012, the DFAE launched a test vote by Internet, by inviting 15,200 of those registered on the consular electoral lists to participate in a dummy run, but only 30% of participants were able to cast their vote (Prissette, 2012; Thompson, 2012). An evaluation questionnaire completed by 5,000 participants provided valuable feedback and suggestions for improvement, but concerns were also raised by the socialist senators for French expatriates in a communiqué dated 23 February, which expressed fears that technical problems would result in very low turnout ( Vote par correspondence, 2012). Despite these warnings, the government felt compelled to continue with its plans in order to maximise participation, given the high political stakes involved. One press report of policing the consulates claimed that the MAE had put huge pressure on its consulates to make the turnout respectable, quoting a diplomatic telegram as stating that participation rates will be looked at very closely (Soubrouillard, 2012). Political considerations aside, it is clear that the MAE had a great interest in securing at least a reasonable level of participation (even if not necessarily with the desired results) because of the cost involved in the operation: during a hearing of the Commission of Finance of the National Assembly, the general director of the Department of modernisation of the Ministry of Foreign Affairs, Stéphane Romatet, responsible for the elections for the French abroad, revealed that the elections for the French Abroad would cost around 15 million euros. Most of this cost, which related to both rounds of the presidential and legislative elections, would be borne by the Ministry of the Interior under the general organisation of elections, but the MAE would take on specific expenses linked to communication, the preparation of electoral lists and

12 consular services, and there would be 2.2 million euros of specific additional costs for the creation of a software programme to run the new option of electronic voting. In fact, when the accounts were published, the total cost of the presidential and legislative elections was 16.6 million euros; whilst some costs (about 5 million) are calculated jointly for both elections, the specific cost of the legislative elections was 4,623,800 euros, of which 1,807,400 was attributed to the electronic vote in 2012, in addition to the 1,469,400 euros spent on this in the 2011 budget (Assemblée Nationale, 2012). When it came to the actual election, registered voters had first to verify that their computer conformed to the norms set out under the label MonVoteSécurisé : this meant that they could see whether they had the right version of Java to be able to vote. If it failed, instructions were given as to how to ensure compatibility and from April 1st onwards, a telephone assistance line was made available.4 Voters were advised that they needed version of Java Virtual Machine (JVM), which had been tried and tested, however, problems arose because a new version (1.7) was subsequently released which was not compatible. Those with the new version were asked to uninstall it and reinstall the older version, which caused various technical difficulties and was clearly impossible for anyone not using their own computer. Once the problem was detected, an applet capable of reviewing the computer s configuration and rolling back to the adequate version of JVM was published on the Website, but judging by comments in the blogosphere, many were still unable to vote electronically. In the case of unresolved incompatibility, voters could register before May 1st for a postal vote instead. The procedure for voting for those who had supplied their addresses (roughly 210,000 citizens; see OSCE, 2012, p.10) was that they were sent a unique login identifier by post or by SMS and a password was sent by a few days before each round. Digital IDs were not issued because France does not yet issue ID cards with electronic chips (CNIe) that could be verified at the point of voting. The introduction of electronic IDs for this specific purpose would have been a very costly and complex exercise involving technical and logistical difficulties not considered appropriate in the circumstances (Government plans to bring in electronic ID cards, first mooted in 2001, were shelved in January 2013 due to both the high annual cost estimated of 85 million euros, and to insufficient confidence in security systems). An online video demonstrated the whole procedure for explanatory purposes. Electronic voters had a week in which to cast their vote: for the first round, this was from Wednesday 23rd May to Tuesday 29th May (the vote took place in France on 10th June), and for the second round, from Wednesday 6th June to Tuesday 12th June (17th June in France). Because of electronic voting, the calendar for expat elections was different from that in France: the 2 rounds were 2 weeks apart, as for presidential elections, whereas legislative elections in France are only one week apart. Electronic voting was available for a week, with voting ending on the Tuesday or Wednesday (depending on the time zone) before the physical ballot. So French ex-pats were voting electronically in the second round before the first round ballot in France, and curiously, results were made available immediately. E-voting thus created a change in the election calendar which led to a certain amount of confusion for voters as we will see below. The reason for the early electronic vote was to enable people to vote at the polling station if there were technical problems. Voting could take place once a secured connection was established to the voting portal using the login and password provided. A two key system was used, with a public key ensuring the encryption of the data and a private key ensuring its decoding (Pinault & Courtade, 2012). The two keys were generated at the beginning of the poll, when the electronic polling station was opened, and both were necessary to start the counting of

13 the ballots after the close of the election. Confirmation of a successful vote would be provided as a receipt with a validation code in printable format, though as was pointed out in the report referred to above: This confirmation, however, did not ensure individual end-to-end verifiability as it could not provide the voter with a proof that the vote was cast as intended, recorded as cast, and counted as recorded (OSCE, 2012, p.10). In terms of overall participation, the electronic vote was used by 57% of expatriate voters (126,947) in the first round, and 54% (117,676) in the second round. These figures were considered by the DFAE in its Annual Report to be a real success, and along with its claims that there had been no breaches of security, it suggested that its use of Internet voting would lead France to become a reference point for the organisation of other electronic elections on a European level ( Rapport du Directeur des Français à l étranger, 2012, p.51). Despite the self-congratulatory tone of the French administration s assessment of its achievement, there were many critics of the process in the media, especially with regard to its security and transparency (see Echo des medias, 2012) for a comprehensive list of articles in the media relating to problems with electronic voting). The criticisms were led by the newly created Pirate Party, which presented four candidates in the overseas constituencies, and which had published a press statement on May 21st, warning that there was a lack of transparency in the procedure for Internet voting for these elections ( Le Parti Pirate, 2012). They complained that the entire electronic voting procedure was being monitored by a single electronic voting bureau ( Bureau de vote électronique or BVE), dealing with potentially 700,000 expatriate voters, equivalent to the cities of Lyon, Toulouse and Nice put together. This was compared to French voting bureaux which typically only monitored 1000 voters. The seven members of the BVE were seen as woefully inadequate and lacking in expertise, and as a result, most operations were carried out by private service providers, not subject to the same degree of democratic transparency. Although all parties contesting the election were allowed to send delegates to monitor operations, only five actually attended the opening of the proceedings on May 18th, four of them from the Pirate Party and one from the Front de Gauche. They claimed that the generation of electronic keys, allowing access to the electronic ballot boxes after the vote, were not secure and could have been duplicated by anyone accessing the relevant computers. Consequently, they created a dedicated Website where people could publish their comments on their experience of Internet voting in order to give citizens a voice on these matters (Parti Pirate, 2012a). 245 did so, many of them complaining that they had not received the login identifier in the post, or that they had not received the password by , and suggesting that addresses entered by voters by hand on a form had then been wrongly recopied by the consular services. The detailed findings of this exercise were collated by the Pirate party in a press dossier dated 12 June, which took the French constitution, the Geneva based Inter-parliamentary Union, and the Council of Europe s rules for defining transparency, verification and responsibility, and reliability and security as its legal basis for evaluating the voting process (Parti Pirate, 2012b). First, it claimed the vote was not transparent since there were insufficient observers, and that insufficient information had been given to them; in particular, they were not given the source code. Second, it criticised the fact that private companies had played too important a role in all the procedures, beyond democratic control, as illustrated by the updating of Java described above. Third, it complained that it was theoretically possible for members of the BVE to match a vote to a voter, thus depriving them of anonymity. Fourth, it endorsed the risks, highlighted by a computer expert, of vote manipulation resulting from a man in the middle attack which exposed the whole system as insecure (Grégoire, 2012). Fifth, it lamented

14 the fact that a considerable number of people had not been able to vote, as witnessed on the party s Web-site. Finally, it urged publication of data which would enable estimates to be drawn up of the number of people who were not able to vote for technical reasons. It was later reported that of 244,623 voters over the two rounds, 12,893 or approximately 5%, had tried to connect to the voting site but had not cast their vote (see Assemblée Nationale, 2012). The problem of Java incompatibility was later taken up by Axelle Lemaire, elected deputy to the 3rd constituency, in a written question to the Minister in charge of the French Abroad, dated March 5, 2013: she also raised the question of why so many addresses had been wrongly transcribed by the consular services (Lemaire, 2013). The Minister s reply repeated the official assertion that the electronic vote for French expatriates had been a success and had fully met all the necessary security guarantees. It claimed that all voters had received a letter in November asking them to verify the address given to the consulate, and that numerous reminders had been sent by post and by and on the special consular Web-site votezaletranger.gouv.fr, telling people that they could modify their details, including addresses, on the MonConsulat. fr site until just before the elections. The minister also confirmed that for the two by-elections that would take place in May and June 2013 (as a result of the cancellation of those elections by the Constitutional Council on grounds of irregularities in campaign accounting), the system had been updated to the latest version of Java. These problems were also picked up in the final report by the OSCE (2012) referred to earlier, which noted that the update of Java Virtual Machine (JVM) during the first round voting period (which was totally beyond the control of the BVE) meant that some people were unable to cast their vote. The same report also pinpointed a small anomaly in the second round, when there was a one vote disparity in the final count due to the corruption of one digital certificate (the Pirate Party also used this to argue that the whole system was invalid). In terms of recommendations, it proposed that a strategic voter project be launched in order to enhance and promote citizen participation in the evolution, organisation, control and adoption of new voting technologies and to increase transparency of the procurement process (OSCE, 2012, p.12). It also suggested consideration of measures to permit voters to recast their votes in case of error and to limit the possibilities of voter intimidation, and recommended use of a verifiable Internet voting scheme or an equally reliable mechanism for voters to check whether or not their votes were cast as intended, since there was no official certification process for Internet voting in these legislative elections. The report by the Constitutional Council on the legislative elections of 2012 briefly discussed the problems raised by what it referred to as use of the Internet, but simply concluded that there was no need for any new legislation (Conseil Constitutionnel, 2013). Thus there was a wide gulf between the officially expressed satisfaction with the proceedings and results, and the criticisms voiced by voters and experts in electronic voting. We turn now to examine the results obtained in the electronic vote in the 2012 legislative elections in overseas constituencies and the two 2013 by-elections. It is important to set these results into the context of the trends shown earlier, of increasing numbers of voters on the electoral register, but decreasing levels of turnout. Voting and turnout in the 2012 legislative elections also have to be placed within the wider context of differential turnout: the fact that some elections are perceived as being more important and therefore

15 attract a higher share of voters than other elections (Reif and Schmitt, 1980). In France, presidential elections are the classic first-order elections, seen as being the most important ones, with the highest stakes (Mény, 1993, p. 52). Figure 3 illustrates the increasing gap between on the one hand presidential elections and on the other hand, other types of elections. The figure shows that European Parliament elections have the lowest level of turnout, at the same level as the ex-pat vote in presidential elections since Turnout in legislative elections has declined quasi-continuously since Legislative elections have taken place a month after the presidential election since 2002 and the synchronisation of both elections after the presidential term was reduced from 5 to 7 years, with turnout since then losing another 10 points compared to 1997, suggesting voter fatigue. Indeed, the gap between the average of the two rounds in the presidential election and that of the two rounds of the legislative elections was 14 points in 2002 and 25 in This suggests that expatriate turnout in the 2012 legislative elections was also likely to suffer from this combination of perceived lower stakes of the legislative election and voter fatigue. Combined with the traditional low turnout among this electorate, it was safe to assume that turnout would be rather low. Figure 4 confirms that turnout in the 2012 overseas parliamentary constituencies was considerably lower than in France, with a gap of over 30 percentage points. Overseas turnout in legislative elections was half of that for presidential elections, with only 1 in 5 voters registered overseas casting a ballot either in person, by proxy, by postal ballot or by e-voting. In the by-elections, turnout could have been influenced by three factors with contradictory effects. Voters in these two constituencies would have been more familiar with the election of Figure 3. Turnout in French elections,

16 Figure 4. Turnout in the 2012 legislative elections and 2013 by-elections Note: the by-elections were in the 1st constituency (US and Canada) and the 8th constituency (South Eastern Europe: Italy, Greece, Cyprus, Israel and Turkey) (see Figure 1 for a map of the geographical boundaries of the overseas constituencies). overseas deputies and may have been more aware of their existence as well as of the election. This could in theory contribute to a higher turnout. However, by-elections tend to have a lower turnout, as voters perceive their stakes to be lower (Mény, 1993, pp. 52-3). In the cases at hand, the fact that after a number of by-election defeats in France the government majority was at risk of being further reduced if the PS lost these elections gave these by-elections a national stake that could increase turnout. In fact, turnout in both by-elections was particularly low, in particular in the 8th constituency.5 As the election data shows, there were noticeable regional variations in turnout. Figure 5 compares turnout in all 11 constituencies in both rounds of the 2012 elections. Overall, turnout was on average 20%, with standard deviations of 3.85 and 3.99 for the first and second rounds respectively. In all constituencies turnout was below 30%, and it went as low as 13.8% in the first round (9th constituency) and 12.8% in the second round (8th constituency). In this last case, this drop is particularly important given the fact that this constituency had registered the highest level of turnout on the first round (28.8%). There is no overall trend in turnout between the two rounds, as 5 constituencies saw turnout increase in the second round, 5 saw it decline, and one stayed the same. This suggests that campaign-specific factors, such as the candidates left in the second round, might have affected the evolution of turnout between the two rounds. This data suggests that e-voting, which was supposed to encourage turnout by making it easier to vote and limiting the need to travel to a possibly distant polling station, failed to maintain turnout at the same level as that of the 2012 presidential

17 Figure 5. Turnout by constituency in the 2012 legislative elections elections, let alone exceed it. The next sections analyse e-voting in these two sets of legislative elections. So far we have only discussed overall turnout. This section turns to electronic voting in the 2012 parliamentary elections and its contribution to the overall turnout as well as the impact of the constituency and country where voting took place. Figure 6 reports the shares of electronic voting, polling station (in-person) voting and postal voting in the elections 2012 elections. A first observation is that of the three voting methods, the postal ballot was very seldom used outside Europe. Even there, only 3.9 and 4.3% of voters chose a postal ballot respectively in Switzerland (6th constituency) and in the 7th constituency, which include Germany and Central Eastern Europe (CEE). In contrast, only 1 postal ballot was cast in the whole second constituency and very few in the 9th, 10th and 11th, where some countries postal services may not be entirely reliable. Internet voting was overall very popular among voters, as on average 53% of all ballots were cast through the Internet. The graph above shows that there are some regional variations. The two African constituencies (9th and 10th, the latter also includes the Middle East (see map of overseas constituencies in Figure 1) are, perhaps unsurprisingly, the constituencies with the lowest shares of electronic voting and the highest shares of in-person voting at the polling station. The 2nd constituency (Mexico, Central and Latin America and the Caribbean) is the only other constituency where fewer than 50% of ballots were cast by the Internet. This suggests that the ability of voters to vote using the Internet is dependent on the Internet infrastructure of the country they live in. Indeed, the share of Internet voting is highest in

18 Figure 6. Share of the different voting methods by constituency in the second round of the 2012 legislative elections North-West Europe (3rd constituency) and in the US and Canada (1st constituency). Figure 7 plots the relationship between the percentage of ballots cast by e-voting per constituency and the share of each constituency s population who used the Internet in 2012 (also as a percentage, from data often based on an estimate by the International Telecommunication Union ITU or, in some cases, national statistical agencies or government ministries). The graph shows a positive relationship between individual use of the Internet in a constituency and the share of e- voting in this constituency. Data about individual Internet use is used in this graph as a proxy for the comprehensiveness of the country s Internet infrastructure and the availability of Internet connections that French citizens may have used to cast their ballot online, but this does not in fact mean that this measure indicates secure individual access. This may be better measured by data on the share of the population with Internet at home. The data from the ITU (not shown) confirms that fewer people have an Internet connection at home than have access to the Internet. Indeed, in some countries, Internet cafes or Internet at work may be some people s main point of access to the Internet. Table 1 evaluates the relationship between Internet availability at home (usually measured by survey and also compiled by the ITU) and the share of e-voting in the 2012 elections (measured by the average share of e-voting in the two rounds) per country. The relationship is similar to

19 Figure 7. individual use of the Internet and share of voting per constituency (2012) Table 1. Correlation between Internet at home and share of electronic voting **.Correlation is significant at the 0.01 level (2-tailed). Share electronic voting Pearson Correlation Internet at Home.605** Sig. (2-tailed).000 N 121 Table 2. Correlation between turnout and e-voting Average turnout Pearson Correlation Share of e-voting (Average 2 Rounds).077 Sig. (2-tailed).334 N 161 that between individual use and e-voting, with a positive and significant relationship between the two variables. Although we have established that turnout was low in those legislative elections, we may wonder whether there is a correlation between turnout and the share of e-voting, that is, whether turnout was higher in those countries where the share of e-voting was higher. Table 2 shows a weak and insignificant correlation between these two variables, confirming that turnout was not higher in those constituencies where e-voting represented the most important voting method. Overall, the data shows that electronic voting did not prevent the inevitable decrease in turnout in the 2012 legislative elections compared to the presidential elections that had taken place a month previously. The results also suggest that e-voting may have replaced in-person voting in some constituencies (mostly in the most affluent

20 ones) rather than provide an opportunity to vote for previous non-voters, as those who would have voted at the polling station voted from their computer instead. However we lack individual-level data to support this hypothesis. This section focuses on the two by-elections that took place in May and June 2013 in the 1st and 8th constituencies. Figure 8 shows that turnout declined between the 2012 election and the byelection in both constituencies, most dramatically in the 1st (on average, -7 points in the 1st against -2.6 in the 8th). Overall turnout was extremely low, with abstentions reaching 90% in the 8th constituency on the second round. Electronic voting did not therefore offer a solution to this problem. Figure 9 compares the share of the different voting methods in both constituencies and shows very little change in voting patterns between the two sets of elections. One may have assumed that since by-elections have a lower level of salience among the electorate fewer people would have bothered to waste time to go to the polling station and that e-voting would have been the voting method of choice. The data show a slight increase in the already high share of e-voting in the 1st constituency, but the share of e-voting decreased in the 8th between 2012 and This suggests that some people either did not bother to register again for e-voting for the by-election or that information about the need to register again for the by-election may not have been sufficient or clear enough in a context of low voter interest. Turnout may have been even lower without e-voting, but the ease of e-voting did not have an effect on turnout in the 2013 by-elections. So taking all the above into consideration, was the experience of electronic voting for French expatriates a success or a failure? And what impact will these elections have on the future of electronic Figure 8. Turnout in the 1st and 8th overseas constituencies in the 2013 by-elections

21 Figure 9. Share of the different voting methods in the 2012 legislative elections and 2013 by-elections in the 1st and 8th overseas constituencies Note: R1 and R2 mean first and second round respectively; the average of the two rounds is displayed for 2012 voting in political elections in France? As we have seen, even though turnout was embarrassingly low, the official discourse tells a story of success: this is unsurprising, given the financial and political stakes involved. Indeed the practice of electronic voting seems to have gained ground, with three important developments since June First, electronic voting was adopted by the UMP party to elect its leader in November 2012, and again in June 2013 to select the candidate for the mayoral elections in Paris in Other parties will be doing the same in the run up to the municipal elections of Second, a law was proposed to the National Assembly on 19th July 2013 by Thierry Mariani, deputy for the 11th overseas constituency, to extend electronic voting in future to presidential & EP elections in overseas constituencies. Third, a major reform of the AFE was passed by the National Assembly in June 2013, which involves the enhancement of electronic voting as a result of the abandonment of the postal vote for these elections, leaving a simple choice between voting in the polling stations or electronically (much to the annoyance of some of the Senators representing the French Abroad, since postal voting was traditionally the main voting method used in these elections) ( Loi relative à la representation des Français établis hors de France, 2013). So our analysis would seem to suggest that electronic voting has a secure future, at least for expatriate voting, but also in mainland France in certain types of political election (especially party primaries) and in professional elections. Yet critics of electronic voting still voice strong concerns that it is not good for democracy and that it represents a significant challenge to the security, anonymity and integrity of the vote. It is indeed hard to see how some of these technical issues can be easily resolved. But problems of election security and fraud are not exclusive to electronic voting, and electronic voting does at least seem to promise a reduction in cost. With specific reference to elections involving expatriates dispersed around the globe, electronic voting might seem to offer a good solution in theory to the problems of distance that prevent so many from voting, and it