Electronic Voting Experiments in Political Elections around the World

GERMANY | ARGENTINA | AUSTRALIA | BELGIUM | BRAZIL | CANADA | ECUADOR | SPAIN| UNITED STATES OF AMERICA | ESTONIA| FILIPPINES | FRANCE | GUATEMALA | NETHERLANDS | INDIA | IRELAND | ITALY | JAPAN | NORWAY | PARAGUAY | PORTUGAL | UNITED KINGDOM | ROMANIA | SWITZERLAND | VENEZUELA

 

EUROPEAN UNION:

NETHERLANDS (remote electronic voting at polling stations)

Voting systems as an alternative to paper voting have a large tradition in the Netherlands. Electoral legislation has allowed use of automatic voting systems since 1965. In 1974 transferring from mechanical devices to electronic devices was initiated. In 1989 around 3% of the municipalities used electronic voting machines manufactured by the Dutch company Nedap, and this figure rose to approximately 26% in 1990 and 30% in 1991, followed by a further rise to 60% in 1994, and from then on the number gradually rose to a figure of 90% in 2002.

In the May 2002 general elections, 95% of the Dutch municipalities had voting machines available and election by computer was possible in some polling stations.
In the 2004 European Parliamentary elections of 2004, voters located abroad on Election Day, who had specifically registered to do so, were able to use the Internet or the telephone to cast their vote electronically. Of those voters who had registered for remote electronic voting, 74% voted electronically, 4,871 via the Internet and 480 via the telephone.  An experiment was also carried out in these elections by 4 municipalities which allowed voting in any of the polling stations within the municipality and not just the one where the voter was registered. The possibility of voting in any polling station using the existing system was not considered workable and studies were set in motion to determine systems which could enable this.

In the 2006 parliamentary elections, electronic voting via the Internet was also carried out for voters who were located abroad and who had registered to do so, but voter registration was also carried out via the Internet. Voting using the telephone was not made available. Electronic voting either remotely or in the polling stations was carried out by more than 90% of the voters. 19,815 voters cast their votes via the Internet. Security problems were discovered in the SDU voting machines which were being used in 35 municipalities, and as a result these machines were banned one month before Election Day and several of these municipalities, including Amsterdam, elected to use paper ballots.

In the parliamentary elections held on 22 November 2006, the Dutch Government invited the OSCE – Organization for Security and Cooperation in Europe to send an Election Assessment Mission from their Office for Democratic Institutions and Human Rights to monitor the elections and their report, The Netherlands Parliamentary Elections 22 November 2006 – OSCE/ODIHR Election Assessment Mission Report contains a set of recommendations, including the following:

• Development of an open source version of the Internet voting system free of proprietary issues and secret components.
• Voting systems should be monitored by an independent entity with broad technical expertise.
• Routine testing of voting machines before elections.
• Introduction of a system for Voter Verified Paper Audit Trails (VVPAT) for voting machines.
• Software dependent vote recording mechanisms which do not permit a check by an independent body should be phased out.
• Discontinue use of systems which depend for their security on the secrecy of any part of their technical specifications.
• Publish the results of the vote according to polling station, including invalid ballot papers, votes awarded to each candidate, votes cast for no candidate and proxy votes (the proxy system of voting has a long history in the Netherlands, but the Mission recommended its revision). 
• The Mission observed that somebody who obtains access to voter authorization codes may alter their vote and noted that a voter may lose the secrecy of their vote by simply allowing themselves to be observed. In addition, it identified the problem that the list of voter authorization codes has to be stored until polls close, and there do not exist transparency mechanisms to ensure against the possibility that the list be altered, or that it is kept secret and the impossibility of its use to interfere with the votes.

On 4 October 2006 a group of Dutch citizens called “We do not Trust Computers” showed how to alter a Nedap manufactured voting machine on a television programme by making a simple and swift change (less than 5 minutes) to a component (a PROM) which was able to completely alter or simulate the results of the voting. A member of the group came on the programme later to state that with training this operation could be reduced to 90 seconds, and on 9 June 2007 a film was published showing the change being made in 60 seconds. On 10 October a film was published showing how a vote cast at a machine manufactured by Nedap could be observed  using the radio signals transmitted by the machine, which are detectable up a range of 25 m.

Following the subsequent debate, the Government set up two commissions in December 2006 to study electronic voting and vote counting in elections: one to examine how decisions had been taken to approve these voting machines for use and the lessons to be learnt from that analysis, and the other – an Electoral Process Advisory Commission– to examine the organisational nature of elections and to present proposals for improvements and any alterations which were considered necessary.

The Electoral Process Advisory Commission’s report, entitled Voting with Confidence was submitted on 27 September 2007. The main conclusions of this Commission were:

• Voting at polling stations should remain the main method of voting in the Netherlands, given that other voting methods, and in particular those via the Internet, telephone or postal vote, do not offer the same level of security.
• Voting at polling stations should be standardized.
• Paper balloting is preferable for reasons of transparency and verifiability. However, in order to overcome practical problems with the counting of paper ballots, electronic voting is acceptable as long as it produces paper ballots that can be checked by the voters themselves, can be produced by a printer and placed in a ballot box and electronically counted using an electronic voting counter once the polling station has closed. The alternative of an electronic voting machine with a memory which carries out the count at the end and a paper trail as a means of verification, whilst enabling a real-time count, is considered to have more disadvantages than advantages since it relies on software to ensure the correct memorisation; final counting of printed votes should only take between 15 and 30 minutes at each polling station, and in this way there is complete independence between the casting and the counting of the vote. The possibility of manual voting using pre-printed voting ballots, which are then scanned for the purposes of vote counting (as have been tested in the United Kingdom and used on a large scale in the USA) is advised against due to associated security weaknesses.
• Voting at any polling station within the same municipality, tested in previous elections, should be applied to all municipalities, but weaknesses detected regarding falsification in the act of voting need to be resolved through introduced authentication mechanisms. Voting at any polling station in any municipality should only be considered after a period of experience of voting at any polling station in the same municipality and on the condition that recommended security measures be adopted.
• Voting via the Internet should only be envisaged for voters resident abroad or who are absent due to work or business, and for the spouses or children of people in the aforementioned situation, as well as voters with special needs which prevent them from voting in a polling station, but the possibility of postal voting for these voters should be kept for those who do not have access to the Internet and cannot or do not wish to use it.
• Legislation should be passed to state the requirements with which voting equipment must comply to be used in elections and the transparency and verifiability of each electoral act should be improved by submitting it to a complete audit.

On 16 May 2008 the Government decided to do away with electronic voting at the next elections and adopt a system of paper balloting. 

BELGIUM (electronic voting at polling stations)

Belgium has been the pioneer in carrying out electronic voting systems in Europe. This project dates back to 1989 and is based on a system of cards with magnetic bands containing data regarding the choice of candidates, with the vote being carried out on a screen with the help of an optical light pen connected to a personal computer, and then the magnetic ballot card is inserted with the vote recorded in a ballot box by the voter. The votes can then be used for an electronic recount. Electronic voting is only carried out in polling stations.

The first experiments in electronic voting took place in 1991. The Law on Automated Voting was adopted in 1994, and has been successively amended in the general, provincial and local elections of 1995, 1999, 2000, 2003, 2004, 2006 and 2007. In 1995 20% of the votes were electronic and in 2003 this number increased to 44% (3.2 million votes).  The system was used by more than three million Belgian citizens in the regional and European elections in June 2004.

A paper trail system known as “ticketing” was tested in two electoral cantons in 2003, but it was concluded that the technology used was not reliable enough. In the commune of Schaerbeek the electronic voting machines recorded 4,096 votes than registered voters.

In 2007 there were around 10,500 polling stations, around 4,000 of which had electronic voting systems.

The Belgian delegation invited the OSCE – Organization for Security and Cooperation in Europe to send an Election Assessment Mission from their Office for Democratic Institutions and Human Rights to monitor the parliamentary elections held on 10 June 2007 and their report, Belgium Federal Elections 10 June 2007 - OSCE/ODIHR Election Assessment Mission Report) contains a set of recommendations, including the following:

• Introducing Voter Verified Paper Audit Trails (VVPAT).
• Legislation to ensure the possibility of a meaningful post-election recount and audits of results.
• The transparency of the certification process of the voting system should be enhanced and tests covering all aspects of the system, based on a comprehensive list of required criteria.
• The practice that the auditing company is contracted by the vendor should be reassessed.
• More efforts should be made to clearly explain to the public the technical and operational aspects of the electronic voting system in order to ensure appropriate safeguards for transparency and accountability.
• The role of the College of Experts should be enhanced. Their mandate should start earlier than 40 days before election day and their members should be available full-time, at least during the election period. They may need to employ more staff for checking system components and verifying operational procedures. More external advisers and legal experts in the College of Experts should be considered. The activities of the College of Experts should be extended to include a recount of magnetic cards at randomly selected polling stations.  The electoral code should envisage that the reports of the College of Experts be published.
• The holding of a comprehensive vulnerability study of the entire system by IT security experts.
• Guarantee that the public authorities have adequate supervision and command of the e-voting system, in order to avoid excessive reliance on the equipment vendors.
• Increase in the number of people entitled to observe the process of making floppy discs from the masters.
• Reduction in the number of persons handling the floppy discs; ideally, only the Heads of the Main Offices of each Canton and the Chairpersons of the Polling Stations.
• Improvement in the security and integrity of the hardware and magnetic cards stored between elections.
• Publication of the detailed results of all the polling stations and counting centres as soon as the respective counts have finished.
• Inclusion of a ballot box transference document to record who transferred them to the counting centres.
• Changing the procedures for closing  polling stations so that the members of the polling board cannot leave without having been present when the ballot boxes’ floppy discs are coped, as well as signing the envelopes when they are enclosed. (in some cases the members of the polling board had left before this step, leaving only the Chairperson).
• The guarantee of the presence of at least one assessor during the transfer of the polling station documentation to the counting centre (the transfer is not witnessed, and in certain cases polling station Chairpersons left the ballot boxes in the polling stations, with the voting results, without verifiable documentation to verify a proper custody chain of this equipment).
• Systematic comparison of the counting results at the polling stations and the central counting centres with those received by the Ministry of the Interior (this verification was not carried out in many cases and there were situations of differences resulting from a software problem in the Ministry of the Interior's computers).

In 2006, the federal and regional administrations asked for an independent comparative study of the voting systems used and the drawing up of a list of requirements for the voting systems to be used in Belgium in the 2009 elections, and thereafter. This was awarded to a Consortium consisting of the following universities: Katholieke Universiteit Leuven, Universiteit Antwerpen, Universiteit Gent, Université Catholique de Louvain, Université de Liège, Université Libre de Bruxelles and Vrije Universiteit Brussel.

The study was divided into two parts, the first part considering the current electronic voting systems and the Belgium electoral system, and the second part making suggestions regarding the technical requirements for the voting systems to be used in the 2009 elections, and thereafter.  The report for the first part (BeVoting – Etude des systèmes de vote électronique, Partie I) was submitted on 15 April 2007 and the second part (BeVoting – Etude des systèmes de vote électronique, Partie II) on 4 December 2007.

The final report analyses 5 electronic voting systems, with the first being identified as that preferred by the Consortium:

• The member of the electorate votes using an electronic voting machine which prints out the vote cast by persons in a legible manner onto a ballot paper, which also registers the vote in such a manner that it is machine-readable, using a barcode or encrypted RFID. The voter then checks the printed vote, makes the machine-readable part visible and folds the ballot paper or inserts it into an envelope, hands it to the Chairperson of the polling station who checks for any unauthorised marks, and puts the ballot paper into the ballot box.
• The member of the electorate votes manually on a ballot paper which is then optically read to facilitate vote counting.
• The member of the electorate votes on an electronic voting machine connected to secure local servers though a local network and optionally creates a paper trail which allows the voter to verify the contents of his/her vote, without being able to alter it.
• The member of the electorate votes remotely via the Internet.
• The member of the electorate votes using an electronic voting machine located in a polling station and linked to a secure national network.

The Consortium concluded that completely automated electronic voting was not currently suitable for Belgium.

The justifications for not opting for any of the four final systems, in preference to the first, can be summarised as follows:

• System 2. It has the inconvenience of consuming considerable computer resources, with high costs, and it is relatively slow, susceptible to reading errors, and produces large quantities of information, which it is necessary to preserve to be able to carry out verification.
• System 3. It has the inconvenience of requiring the servers to both record an individual voter record (the eID of the voter or a card supplied by the polling station Chairperson) to ensure he/she votes just once, and also to record his/her vote. Checking the eID of the individual is clearly the most practical method in most cases but the system has to ensure that no link of any kind is established between the eID and the vote. This is difficult to achieve and it would be difficult to prove to voters that voter anonymity is guaranteed. A further inconvenience is that the polling station servers constitute a single point of vulnerability.
• System 4. It has the inconvenience of not allowing a paper trail, of requiring blind trust in the system software, and partially depends on the client computer software directly used by the voter, the security of which cannot either be ensured or verified due to the diversity of viruses, variations in operating systems and the bugs which are to be found in a personal computer. At present the necessary infrastructure to ensure the security of remote voting in non-controlled environments does not exist. Besides this, such a type of voting is susceptible to coercion, vote purchasing, loss of secrecy, etc. It could, however, be interesting to make this type of voting available for certain specific categories of voters, such as overseas residents.
• System 5. It has the inconveniences of not envisaging a paper trail and requires blind trust in the system software.

On 11 July 2008, the House of Representatives voted that the electronic voting system used in 2007 should be used in 2009 if it was still in good working order, given that it was too late to introduce a new system.

FRANCE (electronic voting at polling stations)

French electoral legislation has allowed the use of electronic voting machines since 1969, although a 1972 regulatory decree stipulated this was only possible in municipalities with more than 30,000 voters, a value which was reduced to 3,500 in 1998.

However, the first experimental piloting of electronic voting was only carried out in the 1994 elections for the European Parliament, involving around 4,000 voters in Strasbourg. This was followed by further experimental piloting in Issy-Les-Moulineaux in 1995, in Lyon in 2000, and in Mérignac in 2002.

In 2001, an Internet voting experiment took place in Voisins-le-Bretonneux using a kiosk which was set up in the polling station. Further experimental piloting took place in Vandoeuvre-les-Nancy in the 2002 Presidential elections. The 2002 local elections in Issy-les-Moulineaux also had experimental piloting, which involved a reduced number since only 1,449 persons enrolled to take part in this type of voting, of whom only 939 picked up their secret code and of these 860 voted.

Regulations concerning the requirements which electronic voting machines had to comply with came into force in 2003 and this introduced a certification process for these machines.

In 2003 remote internet voting for French citizens resident in the USA was permitted for the election of members of the Superior Council of Français de l’Étranger. Around 8% of the more than 61 thousand registered voters (60% of the electorate) voted via the Internet. In the corresponding elections held in 2006 remote voting via the Internet was possible in every country, but few voters chose this form of voting. The Association Démocratique dês Français de l’Étranger – Français du Monde asked for an evaluation of the voting system on 7 June 2006 from François Pellegrini, Maître de Conférences en Informatique à ENSEIRB – École Nationale Supérieure d’Electronique, Informatique et Radiocommunications. The report resulting from this evaluation noted the following:

• There was a serious problem of violation of the principle of the secrecy of the vote with regard to certain voters, given that the Chairperson and each Voting Office's aides received lists of the corresponding voters and the totals of votes cast for each candidate and in some offices there was such a small number of voters casting their vote via the Internet, with cases including just one or two registered voters.
• The system adopted did not offer any kind of guarantee which would be expected from a voting system, given that the software used third party libraries and the source code which would have enabled its correction was not available, although this task would always be difficult given the software complexity which had been adopted, and given that the hardware was also considered secret and not available for verification.
• Voting aides were supposed to check that it functioned correctly, but with the dematerialisation of the process with no hard copy this function could not be fulfilled.
• Voting via the Internet is subject to the results possibly being destroyed or falsified by a small group of individuals, such as distributed denial of serve attacks, and DNS poisoning, issuing a virus to infect the voter machines or alter the voting.
• The use of a system which as a whole provides no hard copy of information relating to suffrage should be legislated against.

 On 1 July 2003, the National Commission for Information Technology and Liberty (CNIL) adopted a recommendation (Délibération n° 03-036 du 1er juillet 2003 portant adoption d'une recommandation relative à la sécurité des Systemes de vote électronique) concerning the security of electronic voting both in polling stations and remotely - especially via the Internet - which included the following:

• Separation of data relating to the name of the voter from the votes file in distinct IT systems and encryption of the electronic vote from its emission at the terminal.
• The impossibility of staff encharged with managing or maintaining the IT system from accessing information concerning the counting of the votes, which should be encrypted with encryption keys and the decryption information conserved in a sealed form. Remote servicing should be forbidden during the counting of votes and until the end of the period when appeals may legally be submitted.
• Access to the software’s source code and the use of public encryption algorithms should be available.
• The system should provide a complete trace of its internal operations to provide a solid basis for external audits, and should ensure effective monitoring of electoral processes by the competent national bodies, members of the polling board, candidates' representatives and their delegated experts.
• The servers and other central IT resources should be located in national territory.
• There should be public verification of the initial state of the system before initiating the counting of votes.

The Forum des Droits sur l’Internet, a non-profit private association, whose members are public bodies, associations and companies (there were around 70 members in 2008), was formally set up in May 2001 through the initiative of the French State, with the aim of ensuring its independence from interest groups. It is mainly funded by a grant of around 1.1 million Euros from the State  (this provided 83% of its budget in 2007). On 26 September 2003 it published the report Quel Avenir pour le Vote Electronique en France? , with its main recommendations concerning political elections being:

• Remote electronic voting should not be used in practice, except for residents abroad, but is it desirable that any voter can vote at an electronic voting machine in any polling station.
• The Government should provide the specific accounting regarding the cost of the equipment and electronic voting systems.
• There should be absolute separation of the IT management of electoral records and electoral ballot boxes.
• Access by nominated experts to the source code.
• An audit of the voting system after each election.
• Preservation of voting date until the appeals period has expired.
• Location of voting system servers within French territory.
• As regards remote electronic voting, it should be possible to carry this out over several days and complete this before Election Day, but each electronic vote must be final. The vote cannot become final without dual validation by the voter, who should be able to alter his/her vote up to final validation.
• An electronic voting observatory should be set up, with the aim of centralising information and lessons to be learnt from electronic voting experiments as well as monitoring voting experiments abroad; to advise the Government about electronic voting systems, ensure that electronic voting systems are evaluated, and to publish an annual report on these activities.

Binding electronic voting in polling stations became possible with a Governmental decree on 18 March 2004 which authorised municipalities in the area of Brest to carry out electronic voting. Electronic voting was carried out using machines from the Dutch company Nedap. In the June 2004 European elections, experimental piloting took place in 18 municipalities, which was binding in some cases, such as in Vandoeuvre-lès-Nancy.

Experimental piloting also took place in the referendum on the European Constitution of May 2005, in 60 municipalities, with voting in some of them binding. 75% of these municipalities used the Nedap voting machines.

Electronic voting machines were used in around 60 municipalities in the 2007 Presidential elections with prior experiments having been carried out in 82 municipalities. Around 800 polling stations were equipped with around 1,200 electronic voting machines. Problems with some voting machines not functioning correctly during this election caused some municipalities to stop electronic voting in the second round of the elections (Amiens, St Malo, Le Perreux, Ifs).

Although municipalities are authorised to use 3 of the 4 types of machine certified by the French Interior Ministry, around three quarters of the municipalities use Nedap machines.

Given the high level of use of Nedap machines in France, the controversy over these machines which happened in the Netherlands is worth recalling. On 4 October 2006 a group of Dutch citizens called “We do not Trust Computers” showed how to alter a Nedap manufactured voting machine on a television programme by making a simple and swift change (less than 5 minutes) to a component (a PROM) which was able to completely alter or simulate the results of the voting. A member of the group came on the programme later to state that with training this operation could be reduced to 90 seconds, and on 9 June 2007 a film was published showing the change taking 60 seconds. On 10 October a film was published showing how a vote cast at a machine manufactured by Nedap could be observed  using the radio signals transmitted by the machine, which are detectable up a range of 25 metres. On 16 May 2008 the Government decided to do away with electronic voting at the next elections and adopt a system of paper balloting.

SPAIN (remote electronic voting and at polling stations)

Various experimental pilots concerning electronic voting have been carried out in Spain since 1995 when a system involving cards with electronic bands was tested in 2 polling stations during the elections for the Catalan Parliament which involved a total of 851 voters. Experimental piloting of electronic voting also took place in the Parliamentary elections in Galicia in 1997 and in 2005, in the Valencia Community in 1999, in the Balearic Islands in 2003, in Catalonia in 2003 (in 5 polling stations involving around 1,500 voters who took part in an experiment involving optically read touch screens), and Andalusia in 2004.

Internet voting was experimented with in November 2003, which involved just over 730 voters.

In March 2004 experimental piloting of electronic voting took place in various Spanish municipalities. In Jung, near Granada, out of the 800 voters who were able to remotely vote electronically, 400 cast their vote via the Internet and 197 through SMS. Other experimental electronic voting pilots via the Internet were carried out in Zamora and Lugo, where 274 voters experimented with voting via the Internet.

On the referendum for the Constitution of the European Union, held in February 2005, there was a large-scale piloting experiment involving 52 municipalities with around 2 million voters, of whom only 10,534 participated (0.5% of the electorate). An independent association set up by the Institution for Automation and Manufacturing of the Universidade de Leon - Electronic Voting Observatory – published a report on 21 February 2005 which stated that the system was not reliable.

On 4 March 2008 the Catalan Government signed an agreement with the University of Barcelona to study, analyse and assess the use of ICT in the electoral process, with the project to be concluded by the end of 2008.

On 29 July 2008 the Spanish Government announced that it was studying the possibility of introducing voting via the Internet for Spanish individuals resident abroad.

PORTUGAL (remote electronic voting and at polling stations)

The first experimental piloting of electronic voting in Portugal took place in the 1997 local elections, involving several polling stations in a ward in Lisbon.

The electronic voting system used a voting machine designed by a Portuguese company with a touch screen where the ballot appeared and voting was carried out. An electronic card reader with a chip (like a smart card)which was formatted and identified as being a voting key, and empty of votes, was given to each elector to enable them to vote using the electronic machine. There was also a mobile phone telecommunications interface which remained disconnected during voting and which was used, once a printout of the final voting results had been made, to communicate to the centre for vote counting. This communication was also possible by landline. The vote cast was recorded on the electronic card, which was then placed within an electronic ballot box, and after reading and recording the vote to the ballot box’s memory, the voting information was erased, thus enabling the card to be reused by the next voter. In this way, the system allowed for the use of two independent vote counts, one in the voting machine itself and another in an electronic ballot box.

In the 2001 local elections, experimental piloting was carried out in polling stations in a ward in the Lisbon electoral district and another in the Oporto electoral district. The voting machine was essentially the same as that in 1997, with some improvements which had in the meantime been carried out. This time the software system was made available for auditing by anyone who wished to carry this procedure out.

In the 2004 European elections, three different electronic voting technologies were tested in polling stations in 9 wards of different councils from various regions on the mainland, which had a total electorate of 150,000, of whom 61,000 voted with 9% of these taking part in the experiment. The technology was supplied by two multinational suppliers – Indra and Unisys – and a national consortium – Multicert/PT Inovaçao.

The three systems consisted of a voting machine with a touch screen where it was possible to vote once a smartcard had been given by the Chairperson of the polling station and which was returned to him/her after voting. The main differences were:

1. The Indra system consisted of a set of electronic voting machines which were totally independent and functioned as separate electronic ballot boxes. The Unisys system was also a system in which the voting machines were separate voting machines but included a vote supervision system monitoring the polling station machines and informing the polling board if the votes had been cast and saved correctly. In the Multicert/PT system the voting machines were used to record an encrypted vote on the smartcard and were not electronic ballot boxes, but independent electronic equipment available in the polling station booth.
2. In the case of the Indra and Unisys machines, the vote was not recorded on a smartcard and this could not be reused for other votes since information was recorded on it that it had been used to cast a vote although the vote itself was not recorded there. In the case of the Multicert/PT machine (considering a possible future development of the system to allow remote voting, where the voter casts his/her vote in a different polling station to where the vote is counted, the smartcard was initialised with a verifiable digital signature, the vote was encrypted in the machine using digital signatures and recorded on the smartcard, which was then inserted by the voter into an electronic ballot box card reader located in the polling booth where it was stored, once more encrypted and then the smartcard was cleaned of any information regarding the vote and configured with a new verifiable digital signature, so that it could be used for voting by another member of the electorate, and only after the voting had finished were the votes uncrypted, the stored votes uncrypted and the final count carried out.
3. The Indra and Unisys systems did not include any electoral register management component, as this was managed in a conventional manner, while the Multicert/PT system had an electoral management system linked to the voting.

In the 2005 Parliamentary elections, the previously described voting platforms were used, with certain improvements, including the use of support technologies for citizens with special needs who voted. Experimental piloting was carried out on these in the polling stations of 4 wards in the district of Lisbon and 1 ward in the council of Covilhã, involving a total of 38,000 votes, of which around 26,000 voted with 33% of these participating in the experiment. An Internet voting experiment was also carried out, offered by the Portuguese company Novabase for around 147,000 voters registered to vote abroad, of which 3% took part in the experiment.

Both the results of the experiments carried out in 2004 and 2005 were published with independent auditing reports carried out by university teams (see 2004 European Elections  and  2005 Parliamentary Elections). One of the conclusions of these audits was that that electronic voting systems were considered to have very few advantages over traditional voting systems and that no investment could therefore be justified, even small, to generalise the use of these technologies, arguing that in the Portuguese case it would only make sense to adopt an electronic voting system which would enable a voter to vote at any polling station and not just the one where he/she is registered to vote.

GERMANY (remote electronic voting and at polling stations)

From a legal point of view, every municipality may carry out electronic voting in the Federal Parliamentary elections and in the European elections but not for local elections as not all the Länder have passed legislation to enable this.

The only electronic voting machines officially sanctioned in Germany are those manufactured by the Dutch company Nedap.

The first experimental piloting of electronic voting in Germany took place in the 1999 European elections in Cologne, using 600 Nedap voting machines. Due to the introduction of electronic voting machines the number of polling stations was able to be reduced from 800 to 540 and the number of people allocated to helping with the voting process in each polling station reduced from 7 to 5.

In May 2002 the Federal Interior Ministry announced that in a first stage all the polling stations would be ,computerised before the 2006 Parliamentary Elections so that in those elections any member of the electorate could vote in any one of them, and so bring about the so-called "mobile vote". In a second stage this would be made possible through public terminals linked to the Internet and in a third stage voting could be carried out using any Internet access point. However, this plan was abandoned due to doubts over the reliability of electronic voting and also for financial reasons, given that the experiments carried out in Cologne in 2002 had shown that the costs were high.

In 2002, 1 million voters cast their votes in the Federal Parliamentary elections using Nedap machines installed in around 2,000 of the 87,000 German polling stations, namely those in Cologne and the following cities which had in the meantime obtained machines - Dortmund, Neuss, Cottbus, Koblenz.

In 2003 the polling stations were networked and an electronic electoral register was set up.

In the 2005 Federal Parliamentary elections and the municipal elections in Cottbus in October 2006, electronic voting machines were used, but as it was then known that they could be corrupted special security measures were implemented, including the comparison of the software in each machine with that uploaded using a reference EPROM prior to the voting.  Around 2 million people cast their vote in electronic machines, approximately 5% of the electorate.

In the same Federal Parliamentary Elections the city of Hamburg also experimented with simultaneous conventional paper voting with the help of a "digital pen", which saved what was written on a paper ballot into its internal memory with a pattern which enabled a scanner installed in the pen to provide information to enable the coordinates of the tip of the pen on the paper to be calculated (see figure below). The contents of the pen’s memory was electronically read by the Chairperson of the polling station at the time when the voter used the paper ballot, so that there was an electronic and a paper ballot cast at the same time which enabled verification counts to be held.

On 4 October 2006 a group of Dutch citizens called “We do Not Trust Computers” showed how to alter a Nedap manufactured voting machine on a television programme by making a simple and swift change (less than 5 minutes) to a component (a PROM) which was able to completely alter or simulate the results of the voting. A member of the group came on the programme later to state that with training this operation could be reduced to 90 seconds, and on 9 June 2007 a film was published showing the change taking 60 seconds. On 10 October a film was published showing how a vote cast at a machine manufactured by Nedap could be observed  using the radio signals transmitted by the machine, which are detectable up a range of 25 metres. The debate which ensued in the Netherlands led the Government to ban the use of Nedap voting machines in the Netherlands and return to paper balloting. Given that the machines used in the elections in Germany are also from the same manufacturer, there was also doubt regarding the security of electronic voting in Germany.

In the local and Länder elections of 2006 and 2007, voting machines were used in various municipalities, and in Mainz voting with the "digital pen" was also tested, as had previously been done in Hamburg.

There have been a number of court cases against the use of electronic voting machines in elections, one of which reached the Federal Constitutional Court in February 2007. The problems raised have mainly been the lack of transparency as regards the functioning and security of the machines and the lack of a paper trail which allows for controls through the recounting of votes.

Photography of the "digital pen":

Photographs of paper with the appropriate pattern to be read by the "digital pen" (with a small rectangle magnified to give an idea of the standard):

 

ITALY (electronic voting tests in polling stations)

The earliest experiment with electronic voting in Italy, with no legal effect, took place in San Benedetto del Tronto in the local elections of 16 April 2000 and involved inviting almost 900 voters to take part in a voting experiment using machines with touch screens, after they had cast their vote in the conventional manner; 363 voters participated in the experiment.

The next electronic voting experiments took place in a constitutional referendum on decentralisation in Avellino in 2001 in which 206 persons cast their votes in voting kiosks as part of the e-Poll project and financed with European Union funds. An identical experiment took place during the provincial elections in Campobasso in May 2002 involving 403 persons.

On 17 November 2002, Cremona carried out a simulated referendum with electronic voting kiosks located in three points of the city in which a thousand people took part.

In the June 2003 regional elections experimental piloting of electronic voting using voting machines with a touch screen was carried out in 4 polling stations in different municipalities in the Friuli-Venezia Giulia region which had a total registered electorate of 3,800 persons.

In September 2004 electronic voting tests using kiosks in several points of the city took place during the municipal elections in Avellino, Campobasso and Ladispoli. In this latter experiment, electronic kiosks were placed in many parts of the city, such as supermarkets and bars. A similar vote was carried out in Specchia in 2005.

National legislation has authorised the use of electronic counting since the 2004 Elections for the European Parliament, the 2005 regional elections in Liguria and the 2006 Parliamentary Elections, but not electronic voting, except in the case of the regions of Friuli-Venezia Giulia and Trentino-Alto Adige which have allowed electronic voting in local elections.

In the European Elections of 4-5 April 2005 an experiment involving counting and electronic transmission of results took place in 47 polling stations in Liguria, running in parallel with traditional methods.

In the 2005 municipal elections in the regions of Trentino-Alto Adige experimental piloting was carried out in 5 municipalities in May and in 1 municipality in November. 59% of the electorate participated in the first experiment and 89% in the second one. There was additional experimental piloting in the municipal elections of May and November 2006. Around 10,000 persons participated in these experiments.

In the national elections of 9 and 10 April 2006, the municipality of Cremona experimented with Nedap voting machines in 4 polling stations involving 3,000 voters. In the same elections, the Italian Government carried out an experiment involving the electronic counting of votes in 4 regions (Lazio, Liguria, Puglia and Sardegna) with regard to 11 million voters. It had been arranged that at the end of the elections the manual and electronic counts would be compared and, in the event of there being a difference, the manual count would prevail and only the electronic count for Liguria would be transmitted electronically, but in contrast to what had been stipulated the electronic count for all the regions involved in the test was actually transmitted. In addition, the Ministry of the Interior refused to publish separate data for regions and electoral districts. Given that the difference between the two Presidential candidates – Romano Prodi and Silvio Berlusconi – was very narrow (0.1%, just 24,755 votes out of 38 million) and this had steadily gone down throughout the counting period after an initial difference of 8%, this raised certain suspicions in relation to the counting of votes and the role of the electronic counting. The count was challenged in the Lower House and the Senate requested the recounting of 4 million votes.

As a result of this sequence of events, on 29 November 2006 the Minister of the Interior announced that electronic voting would be dispensed with.

IRELAND (electronic voting tests in polling stations)

Ireland started planning electronic voting experiments in 1999, the year in which it passed the necessary legislation, and carried it out in its 2002 elections in 3 of the 42 electoral districts, with a total number of 138,011 electronic votes cast. It had been planned to use electronic voting using touch screens in polling stations for the July 2004 Local Elections and European Parliamentary Elections.

Following the publication of the article Electronic Voting: A Safety Critical System by two researchers at the Computer Science Department of the National Union of Ireland in Maynooth, who questioned the security of the Nedap/Powervote electronic voting system, which had been acquired for this purpose, the Government set up the Electronic Voting Commission on 1 March 2004 to prepare reports on the level of secrecy and precision of the aforementioned Nedap/Powervote system, with the remit to publish one or more reports by 1 May 2004 concerning the applicability of this system to the elections scheduled for 11 June 2004 and to present a follow-up report or reports on the performance and experience of electronic voting and counting in elections. On 29 April 2004 the Commission presented an Interim Report which stated that it could not recommend the use of the system for the June elections, although this conclusion was not based on any discovery concerning any mal-functioning of the system, but in not having proved by the time of publication that the system functioned correctly. This report led the Government to cancel electronic voting in the 2004 elections.

On 15 December 2004 the Electronic Voting Commission published its 1st Report and on 4 July 2006 its 2nd Report, which concluded its work, and the body was wound up on 4 September 2006. In the final report the Commission made recommendations concerning the voting and electronic counting equipment chosen to be used in elections in Ireland, subject to the modifications which it also recommended. It considered the security of the proposed computer to be inadequate for the preparation of elections and aggregating the counting of votes, which needed to be improved. It considered that security improvements had to be made with regard to storage, transportation and access to the votes following voting. It could not recommend the software for election management and aggregating votes since it was not stable, and the project documentation and software development were insufficient to ensure its reliability. Tests had also shown programming errors. The Commission did however mention that although it could not foresee it being realistic that this software could be made suitable, other election management software could be developed at a reasonable cost which would be compatible with the voting machine and the other system hardware and software components. It also concluded that the system tests as a whole which had been carried out until that date and the certification of its components were insufficient to ensure secure use in elections in Ireland, such that more work was necessary in that area. In comparison to the process of paper voting, the Commission concluded that, in terms of secrecy and precision, the paper system was moderately superior to the electronic system which had currently been chosen and that, with the implementation of the Commission’s recommendations, the system chosen had the potential to provide greater precision and to provide a level of secrecy similar to that of the paper system.

Meanwhile, the voting machines bought by the Government at a cost of 52 M€ from the Dutch firm Nedap would remain in storage, with an annual storage cost estimated at 0.8 M€. The cost of software improvement was estimated at 0.5M€.

On 4 October 2006 a group of Dutch citizens called “We do not Trust Computers” showed how to alter a Nedap manufactured voting machine on a television programme by making a simple and swift change (less than 5 minutes) to a component (a PROM) which was able to completely alter or simulate the results of the voting. A member of the group came on the programme later to state that with training this operation could be reduced to 90 seconds, and on 9 June 2007 a film was published showing the change taking 60 seconds. On 10 October a film was published showing how a vote cast at a machine manufactured by Nedap could be observed  using the radio signals transmitted by the machine, which are detectable up a range of 25 metres.

On 16 May 2008 the Government decided to do away with electronic voting at the next elections and adopt a system of paper balloting.

In April 2006, the Government of Ireland decided that there would not be electronic voting in the elections on 24 May 2007.

UNITED KINGDOM (remote and polling station electronic voting tests)

A commission was set up in January 2000 in the United Kingdom to study innovation in voting systems, the recommendations from which led to the Representation of the People's Act 2000, which enabled experimental piloting in local elections. Amongst other innovations electronic voting systems were tested in local elections in England and Wales in May 2002, 2003 and 2007.

In May 2002, 9 local authorities organised experimental piloting involving electronic voting, 4 requiring the voting to be carried out in polling stations and 5 enabling remote voting, which was used by 14.6% of the corresponding voters. The evaluation report of the Electoral Commission was published in August 2002 with the title Modernising Elections – A strategic evaluation of the 2002 electoral pilot schemes. The Electoral Commission observed that none of the data collected suggested that electronic voting would lead the electorate to vote in significantly higher numbers than with conventional voting systems; it questioned the cost-benefit relationship of electronic voting (more than 3 million pounds financed by the DTLR – Department of Transport, Local Government and the Regions, in addition to the costs borne directly by the local authorities); it expressed the view that experimental piloting of electronic voting should not be considered a priority in the future and considered it premature to suggest that the Government was well on the path towards its aim of an election carried out electronically sometime after 2006.

In May 2003, 17 local authorities carried out experimental piloting of electronic voting in different forms (in polling stations, remotely via the Internet, through digital TV, and by SMS). Around 21% of the registered voters cast their votes electronically in the 17 locations. The Electoral Commission made similar observations to those it had already made for the 2002 May elections. As far as costs were concerned, Stratford on Avon District Council reported costs of £120 per vote via the Internet, £22 per vote in an electronic booth and £1 per conventional vote.

In the May 2007 elections experimental piloting of electronic voting was carried out by 5 local authorities, all involving remote voting via the Internet within a period of approximately two weeks prior to Election Day, along with voting by telephone except for one of the experiments, and also voting in the polling station on Election Day in one of the locations. Voting machines were not used and every case involved the prior recording of voters on paper.

The August 2007 report of the Electoral Commission (Electronic Voting) highlighted that security and quality assurance in the experimental pilots in the May 2007 elections were insufficient and observed that there was an overall lack of transparency concerning the technology adopted and its use. From 2.4% to 17.7% of the electorate voted electronically in each of the 5 locations, 33% to 58% of those who had registered for an electronic vote. The Electoral Commission  considered that the cost per electronic vote was extremely high (between £100 and £600, whilst the cost for each registered voter also varied greatly but was between £1.8 and £27). It also underlined the fact that, even if best practices with regard to security and quality assurance were satisfied, there were risks associated with remote electronic voting which required special attention, including the possibility of:

• Voting instruments, such as personal computers, being compromised by viruses and other malicious software;
• Attacks by people with privileged access to the system (those involved in its development or management, with this latter risk also significant for electronic voting in supervised locations);
• Selling of votes due to the ease with which electronic voting credentials may be exchanged and the added vulnerability in remote voting.

The Electoral Commission further recommended that no more experimental piloting of electronic voting be carried out while the following was not available:

• A complete framework for electoral modernisation which included answers to the questions of transparency, trust and costs;
• A central process enabling the choice of solutions which have been previously tested and approved;
• The preparation of electronic votes at least six months before the contracting of the service from the supplier and the Election Day.

ROMANIA (voting tests via the Internet)

In the referendum on the revision of the Constitution on 18 -19 October 2003 Romania carried out an Internet voting test for members of the armed forces on duty abroad. Of the 1,600 voters, 97% participated in the experiment.

ESTONIA (remote electronic voting)

The idea of voting via the Internet in non-controlled environments was introduced into Estonia in 2001. In the summer of 2003 the National Electoral Committee (CEN) started the project and the first voting took place in the 2005 local elections. The system uses the Estonian electronic identity card with digitally verified signatures. Voting via the Internet is only possible in a period prior to the Election Day (from 4 to 13 days prior). Multiple votes are allowed, with only the last one counting, in order to mitigate any possibilities of influence, coercion or vote purchasing.

In the elections held on 16 October 2005, only 0.9% of the voters cast their vote electronically, despite it having been estimated that 80% of the electorate had the means to do so. In the Parliamentary elections of 4 March 2007 only 3.4% of the voters cast their vote electronically when electronic voting was made available to 100% of the electorate.

In the parliamentary elections held on 4 March 2007, the Estonian Government invited the OSCE – Organization for Security and Cooperation in Europe to send an Election Assessment Mission from their Office for Democratic Institutions and Human Rights to monitor the elections and their report, Republic of Estonia Parliamentary Elections 4 March 2007 - OSCE/ODIHR Election Assessment Mission Report) contains a set of recommendations, including the following:

• Introducing legislation concerning security for the Internet voting system, responsibility of the institutions involved and sanctions in the event of failures.
• Revising the process of counting and announcing votes via the Internet in order to ensure that all the devices used meet suitable security measures (in particular given that the computer used to read the CD containing the results is connected to the Internet during part of the counting process).
• Carrying out a complete audit of the system before and after each election, including all the components and transactions, and also carrying out a real voting test with predetermined results to compare this with the final count, and for all reports to be made public (in particular as the source code had not been tested by the audit designated by CEN, nor had any audit been requested after the election).
• Modifying the Internet voting system to not record the actual moment of the final vote of each voter but to keep a record of a list of the final votes cast via the Internet available to observers for the purpose of transparency.
• Monitoring of and system responses to potential security threats via the Internet and response planning to these threats with clearly defined roles for each of the Institutions involved; improving the monitoring process of the server which stores the votes to further ensure that there has been no unauthorised Internet access which could affect the integrity of the data (in particular, there was no formal plan in place to deal with risks such as that from denial of service).
• Greater involvement of the political parties and civil society in monitoring the Internet voting system so as to provide plenty of opportunities to identify weaknesses and security concerns (confidence in the system should be based on a full understanding of security issues and the effective  transparency of the Internet voting system; for example, the idea that the system was comparable to Internet Banking  was frequently stated by representatives of the political parties and civil society without understanding that the two dealt with substantially different processes given that the recording of Internet Banking transactions are linked to the events relatable to each of the actors as opposed to electronic voting which has to ensure that these relations cannot be made).
• Greater division of tasks between the technical officers involved in the Internet voting system such that there is no single person dealing with the system as a whole.
• Reconsidering whether Internet voting should be used in such a widespread manner, in the event that all the issues raised cannot be completely resolved, or if as an alternative it can be used in a limited or not complete manner.
• Consistent use of procedures to transfer the votes cast early and to ensure that unauthorised individuals do not deal with these votes (for example, the Chairperson of one of the polling stations took the votes home in unsealed envelopes; in another polling station a municipal employee was involved in the transference of votes, which was not supposed to happen and, in addition, this individual was a candidate in the election.
• Publishing the results of the voting in each counting location immediately after the counting has been approved, in order to improve the transparency of the process.

OTHER EUROPEAN COUNTRIES:

SWITZERLAND (remote electronic voting)

In general Switzerland has between 4 and 6 elections or referenda per year, such that the possibility of introducing electronic voting started to be considered in 1999. In January 2002 the Federal Chancery presented a report on the reliability of electronic voting. From then onwards, various binding remote electronic voting tests have been carried out in the canton of Geneva.

Federal law was altered on 21 June 2002 and this was followed by legislation which was approved in September 2002, to allow electronic voting experiments, which took effect from the start of 2003.

A pilot project was started to which the Government invited all the cantons and in the end selected the cantons of Geneva, Zurich and Neuchâtel.

The canton of Geneva was the first to begin the project, chiefly because it already had an electronic register of voters and its legislation already foresaw the possibility of these types of tests. It selected the bid made by Hewlett Packard and by Wisekey. In a period of 3 weeks prior to Election Day, the electors were able to vote by post or via the Internet or on Election Day itself, with voting carried out at a polling station.

The voters in Anières were the first to be able to vote via the Internet, in a municipal referendum held in January 2003 (1,162 voters, 323 votes via the Internet). Other experiments then followed in Cologny (November 2003, 2,521 voters, 432 votes via the Internet), Carouge (April 2004, 9,049 voters, 1,024 votes via the Internet), Meyrin (June 2004, 9,170 voters, 2,723 votes via the Internet), Vandoeuvres (October 2004, 1,382 voters, 240 votes via the Internet). In 2004 voting took place with included more extended Internet voting; in April 2005 a vote on a popular initiative took place which included 7,911 votes cast via the Internet out of a total electorate of 88,082 persons.

From September 2005 onwards several elections or referenda were held in the canton of Neuchâtel which included voting via the Internet, with the number of voters able to use this medium to cast their vote gradually being increased, reaching 1,494 votes via the Internet in voting in June 2007.

In the canton of Zurich, voting via the Internet or via SMS was available in 3 of the 171 communities in October 2005, with 1,461 votes cast via the Internet and a number of other votes having been made from this period onwards.

A list of all the votes in which it was possible to vote via the Internet in the cantons of Geneva, Neuchâtel and Zurique up to 17 August 2007 is available: (Tableau synoptique des essais de vote électronique ayant eu lieu en Suisse de 2003 à 2007).

In the Federal referenda of 27 November 2005 and 11 March 2007, 41% of the voters used remote electronic voting, a much higher percentage than that which had normally been the case until then, with the figure not usually going over 30%.

A Report from the Council of State on the electronic voting project in Geneva was delivered to the Senate on 24 May 2006. On 31 May 2006 the Federal Government presented a report to the Federal Parliament summarising the experiments which had been carried out, and arguing for the introduction of remote electronic voting in stages for all elections and referenda (Rapport sur les projets pilotes en matière de vote électronique).

In the Federal referendum on 17 June 2007 17% of the voters cast their vote electronically, 97% of whom via the Internet and 3% via SMS.

In September 2007 the Federal Government presented its draft acts regarding electronic voting via the Internet enabling its extension throughout Switzerland, and these were approved.

NORWAY (electronic voting tests in polling stations)

4 pilot projects have been carried out: three in the municipal elections of 15 September 2003 (10 polling stations with a total of 9,700 voters) in which 30% of the electorate voted electronically, and one for local administration on 26-27 October 2003 (1 polling station with 1,300 voters) in which the vote was electronic except where the voter asked to vote in the conventional manner, and 91% of the voters cast their vote electronically. The electronic voting was carried out in touch screen kiosks using an electronic identification card as means of voter authentication. Evaluation of these pilot projects stated that questions of security required further clarification.

On 26 May 2004 the Government assigned a Working Party to consider the issues concerning electronic voting and recommend if this type of voting should be adopted in Norway and, if this was to be the case, how it should be implemented based on an evaluation and recommendation of the rules and technical standards which should be applied. This group presented the report Electronic Voting – challenges and opportunities in February 2006, with some of its main recommendations being:

1. Keeping the two stage system currently in force, the first for early voting in designated places, and the second on Election Day.
2. Allow electronic voting only in the first stage and keeping unaltered the paper balloting procedures in the second stage.
3. Enabling a voter to be able to vote several times in the first stage, with only the last, final vote counting and enabling those who vote electronically in the first stage to vote in the traditional way in the second stage.

The Working Party underlined that the principle of secret voting challenged the idea of voting in non-controlled environments, whether this vote was made in an electronic format or not, such that it considered that electronic voting on Election Day was incompatible with the principle of giving each voter the right to a secret vote. It underlined that it could of course be the case in non-controlled environments that voters may be influenced, coerced or bribed, such that the right to cancel or annul a cast vote should be given to the voter.

The Working Party also recommended that electronic voting be the only form of voting allowed in non-controlled environments, such that the use of postal voting should continue to be as restrictive as at present, or that is, carried out in public service offices pre-set up in a controlled manner, and justified this by saying that repeating an electronic vote is easy and cost-free, while repeating a paper ballot through a postal vote involves added costs.

the Working Party also mentioned that the introduction of electronic voting in controlled environments is very expensive and brings very few advantages: election results can be obtained more rapidly and with fewer human resources, but the cost of the equipment and the accompanying logistical organisation would be much greater than the current system, and stated its belief that only if the electronic equipment used to vote were looked after by the voters themselves would there be any potential cost savings. It also stated the advantage of keeping a tried and tested traditional system of voting in the second stage, not only as a safety measure in the event that problems with electronic systems occurred in the first stage.

NON-EUROPEAN COUNTRIES:

UNITED STATES OF AMERICA (electronic voting in polling stations)

Electronic voting in the United States of America (USA) started early on. Oklahoma was the first state to do it, when in the 1990s it introduced an electronic optical reader system  for paper ballots. In the 1996 Presidential Elections, systems of this kind were used by 25% of the registered voters.

Of note is that the perforated card voting systems are often counted with electronic optical sensors. The perforated carding voting systems were first introduced in the USA in the Presidential Primary Elections in 1964 in two counties in Georgia and in another two counties in California and one in Oregon in the Presidential Elections of that year. Voting using perforated card grew rapidly in subsequent years. In 1980 voting using perforated cards was made available to 29% of the registered voters. In the 1984 elections Iowa banned the use of perforated cards due to counting errors, but, following data from Election Data Services, Inc. the availability of this type of voting system continued to grow up to a maximum of 35% of registered voters in 1986, and from then onwards gradually decreased to 30% in 1998 and 26% in 2000.

The possibility of voting via the Internet has been the subject of various studies. One of the first official analyses in the USA was submitted by the California Internet Task Force set up by the Secretary of State for California in his January 2000 Report (A Report on the Feasibility of Internet Voting) and the respective appendix (Technical Committee Recommendations). In this report, among other aspects, it was recommended that in the foreseeable future voting via the Internet should be seen as merely a supplement to the tradition method and, if implemented, this should be done in a phased manner to allow for the detection of problems before they occur on a large scale in a real vote. It noted that the potential for electronic attacks through a virus or Trojan Horse software poses a serious threat to Internet voting such that in order to minimise this single operating systems and Web Browsers should be used and the voters should have to take several precautionary measures before voting.

In the Presidential elections of 7 November 2000, according to date from the report from Caltech and MIT mentioned below, 34.4% of the voters used perforated card voting systems, 27.5% voted using electronic optical reader systems with paper votes, 17.8% used mechanical systems with levers, 10.7% voted directly into electronic ballot boxes, 8.1% used mixed systems and less than 1.3% voted in the traditional paper manner.

In these elections, the contest between George W. Bush and Al Gore was hotly disputed and there were problems with the counting of votes, in particular in Florida, which caused great controversy. The results obtained by the two candidates were so close that there was a difference between them of 5 votes out of 537 in the votes cast (in favour of Bush) and the difference of votes in this state, for which the opinion polls originally had called a victory for Gore, ended up giving it to Bush by little more than 2,000 votes, when the number of delegates elected in Florida was 25. In the total number of overall votes the difference was only 543,816 votes (but in favour of Gore).

In the intense debate which surrounded this event, particular doubts were expressed over votes which had been automatically counted, which had mainly been perforated card votes.

On 15 December 2000, when the results of the Presidential Elections were still being debated, Caltech – California Institute of Technology and MIT – Massachusetts Institute of Technology jointly started a project (Caltech-MIT Voting Technology Project) with the aim of developing knowledge to avoid problems occurring which might compromise elections. In July 2001, six months after the project began, the two universities published the report Voting: What is, What Could Be. Data obtained during the project showed that 4 to 6 million votes had been lost in the election and the reliability analysis of voting technologies and election systems showed that the number of votes lost could be substantially reduced by immediately taking two steps:

• Improving voting technologies. Substituting perforated cards and mechanical lever voting machines with electronic optical readers. It was estimated that 1.5 million lost votes could be recovered through this step.
• Improving electoral register systems. Improving database management with links to the polling boards and the electoral register databases and using provisional votes for voters who turn out to vote but are not identified as being on the electoral register databases. It was estimated that another 3 million lost votes could be recovered through this step.

The report also argued that in the mid and long term new technologies will be developed which could substantially improve the voting systems in elections and the report made the following recommendations:

• Use of a new modular voting architecture to provide more security and a rapid improvement in the availability of interfaces for the user.
• Significant investment by the Federal Government in R&D and voting technologies and suitable procedures to test machines used.
• The setting up by the Federal Government of an independent agency to supervise these tests and collate and distribute information with regard to equipment performance and cost.

In October 2001 the US Government Accountability Office (GAO) published the report Elections – A Framework for Evaluating Reform Proposals for Congress which proposed four criteria to be considered in the process of approving legislative reform:

• The appropriate role for the Federal Government in electoral reform, given the balance established between the constitutional authority of Congress and laws and traditions which give autonomy to local jurisdictions to manage elections.
• The balance between accessibility and integrity.
• The coordination and integration of people, processes and technology.
• The possibility of bearing costs and sustaining reforms, identifying resources and ensuring continual ongoing monitoring and re-evaluation.

The main legal reaction to the problems which were noted in the 2000 Presidential Elections was the approval of HAVA – Help America Vote Act  in October 2002, which essentially had the aims of:

• Substituting perforated card voting.
• Setting up single electoral register databases at state level.
• Improving the process of identifying voters at polling stations and for the purposes of postal voting.
• Generalising the possibility of provisional voting by voters who turn up at a polling station and who are not on the electoral register, with decisions regarding these votes to be taken after subsequent checking.
• Improving access to voting for citizens with special needs.
• Establishing principles for voting systems auditing.
• Setting up the EAC – Election Assistance Commission to support state and local administrations in organising Federal Elections, establishing technical and administrative standards with the support of NIST – National Institute of Standards and Technology, spreading good practices and encouraging the use of improved voting technology, and managing the awarding of funds for R&D projects in the area of improving voting systems, including taking into account the annual recommendations concerning research areas made by NIST.

Thus, HAVA managed to enact practically all the main recommendations of the Caltech and MIT project mentioned above. In fact, the only one of the main recommendations mentioned above which was not considered was the adoption of a new modular architecture for the voting technology, but this would to a large extent be covered by the proposals concerning new Voluntary Voting System Guidelines Recommendations to the Election Assistance Commission which were presented by the TGDC – Technical Guidelines Development Committee operating under the remit of the EAC and which were presented on 31 August 2007.

The concerns expressed at the time concerning DRE - Direct Recording Electronic equipment can be found in the report by the task force set up in February 2003 by the Secretary of State for California which was published on 1 July 2003: Secretary of State’s Ad Hoc Touch Screen Task Force Report, which contained many recommendations regarding:

• Norms and test procedures at the Federal, State and Local levels.
• Setting up of a technical supervision commission made up of technical experts able to improve present tests and standards for software upgrading, give advice during the certification process and review questions relating to hardware and software.
• Random sampling of machines on Election Day to check that each system records the votes correctly.
• Setting up of penalties for local jurisdictions which use non-certified systems.
• System protection by keeping local networks physically isolated from external networks during the voting period.
• Vendors not being given authorisation to carry out logical tests and setting up the voting system on Election day, nor distributing the object or source code to Federal, State or Local entities.
• Use of a written message on screen transmitted to audio equipment informing the voter that the vote has been recorded so as to minimise situations where the vote is unintentionally not recorded.
• Internal security norms and procedures for programmers and technical staff and the development of voting systems to be complied with by vendors, and the imposition of strict criminal sentences if harmful code is found before, during or after certification.
• The demand that all voting systems by 2010 include a means of electronically verifying certified votes. Three members of the task force strongly argued that a voter verified paper audit trail (VVPAT) should be required by January 2007.

On 23 July 2003 a Technical Report was published by the John Hopkins University Information Security Institute (JHISI) written by Kohno, Stubblefield, Rubin and Wallach (Analysis of an Electronic Voting System) which deeply questioned the security aspects of electronic voting with the Diebold electronic voting machine, used in many polling stations in the USA. This report had many repercussions since it identified, amongst other aspects, the following security flaws: the possibility of voters being able to cast multiple votes without it being possible to detect this; the possibility of a voter carrying out certain administrative functions within the machine; possible threats from the staff providing election support or even just cleaning the polling station rooms. Besides arguing that there should be suitable quality control for software development and that use of proprietary software where the source code cannot be inspected is not secure, they proposed the systematic adoption of systems which enabled voter verified paper audit trails (VVPAT). On 30 July the Diebold company contested these claims in a document entitled Checks and balances in elections equipment and procedures prevent alleged fraud scenarios, which in turn merited a reply from the authors of the JHISI report: Response to Diebold's Technical Analysis.

A report from the Congressional Research Service written by Eric A. Fischer (Election Reform and Electronic Voting Systems (DREs): Analysis of Security Issues) was published on 4 November 2003 which used the concerns expressed by the aforementioned JHSI report to consider vulnerability issues concerning elections based on electronic machines recording votes.

The debate over the reliability of remote electronic voting led to the cancellation in the Spring of 2004 of the SERVE – Secure Electronic Registration and Voting Experiment, experiment, which was designed to enable members of the armed forces stationed abroad to vote in the 2004 Presidential Elections. The cancellation stemmed from an evaluation report which considered the Internet and the Personal Computer to be resources which were not secure enough for voting, with four members of the evaluation group (Jefferson, Simons, Rubin, Wagner) deciding to publish their evaluation report (A Security Analysis of the Secure Electronic Registration and Voting Experiment, SERVE) on 21 April 2004 where they concluded:

“The vulnerabilities described cannot be resolved through project modifications or correction of flaws. They are fundamental to the architecture of the Internet and the hardware of the personal computers currently existing. They cannot be eliminated in the foreseeable future without radical and unpredictable technological development.”

On 16 July 2004 Nevada became the first state to decide that all electronic voting would have to have voter verified paper audit trails (VVPAT).

In 2005 the Center for Democracy and Election Management at the American University, in association with the  James A. Baker III Institute for Public Policy at Rice University, the Carter Center and Electionline.org, organised a Commission on Federal Election Reform, also known as the Carter-Baker Commission as it was co-chaired by former President Jimmy Carter and James Baker III, the former Secretary of State in President George Bush’s administration. The operations of the Commission were financed by the Carnegie Corporation of New York, the Ford Foundation, the John S. and James L. Knight Foundation and the Omidyar Network, with additional financial support from the Pew Charitable Trusts for Electionline.org. The aim of the Commission was to study the USA electoral process and make recommendations concerning  its improvement, at a time when the trust of American citizens in their electoral system had fallen to the point where in an opinion poll carried out by the New York Times the evening before the 2004 November elections, only one third of the persons interviewed stated that they had considerable trust that their votes would be counted correctly and 29% stated that they were worried with problems existing with the counting of votes. The Report of the Carter-Baker Commission (Building Confidence in U.S. Elections – Report of the Commission on Federal Election Reform), published in September 2005, gave 87 recommendations divided into 5 groups:

(1) A universal up-to-date publicly available electoral register for the whole country.
(2) A uniform identification system implemented in such a way as to increase, rather than impede, participation.
(3) Measures to improve voting integrity and voter access to the voting procedure.
(4) Voter Verified Paper Audit Trails (VVPAT) and added security for voting systems.
(5) Electoral institutions which are impartial, professional and independent.

Starting in July 2005, three successive reports were published by the NGO Black Box Voting which also identified security failures in the Diebold voting machines: “Hursti Report” – 1st part (The Black Box Report – Security Alert: Critical Security Issues with Diebold Optical Scan Design, July 4, 2005), “Hursti Report” – 2nd part (Diebold TSx Evaluation : Critical Security Issues with Diebold TSx, May 11,2006, “Hursti Report” – supplement (Diebold TSx Evaluation – Supplemental report, additional observations, May 22, 2006). The concern with voting security for these machines was reinforced by the security problems detected by researchers from the Center for Information Technology Policy, Princeton University which were published in the article Security Analysis of the Diebold AccuVote-TS Voting Machine, on 13 September 2006, and through the video available at http://citp.princeton.edu/voting/ which shows how a voting machine can be altered in less than one minute so as to provide different voting results. Verification tests are not able to detect these, as is the case with viruses used for such purposes to contaminate other voting machines without human intervention.

September 2005 also saw the publication by the US Government Accountability Office (GAO) of the report prepared for Congress Federal Efforts to Improve Security and Reliability of Electronic Voting Systems Are Under Way, but Key Activities Need to Be Completed which was a response to HAVA – the Help America Vote Act of 2002 where reports of problems with the electronic voting system had arisen, in response to requests from Congress to:

(1) Determine the main concerns over the security and reliability of electronic voting systems.
(2) Identify relevant practices to ensure the security and reliability of these systems.
(3) Describe the actions taken or planned to improve their security and reliability.

In this report, the GAO emphasised several examples of vulnerability and problems with electronic voting systems (possibility of modifying votes cast, files specifying votes and audit logs; protecting administrator functions with passwords which are weak or easily discernible; locking systems and bare unprotected electricity fuses; local administrations which configure systems incorrectly leading to problems on Election Day; voting systems which had operational failures during the elections; installation of voting systems by uncertified suppliers) and recommended that the EAC – Election Assistance Commission, set up with HAVA in 2002 to help and improve state and local administration of Federal elections and expand the use of electronic voting systems, define specific tasks and the respective deadlines to improve standards, test capacities and support the management of national voting systems.

In July 2005 the Brennen Center for Justice from New York University Law School set up a task force to carry out a systematic analysis of the security vulnerabilities of the three electronic voting systems most commonly purchased by authorities. The methodology, analysis and text were extensively revised by the NIST – National Institute of Standards and Technology. The report of this task force, known as the Brennen Report, was published in 2006 with the title The Machinery of Democracy: Protecting Elections in an Electronic World. The main conclusions of the Brennen Report were:

• The three voting systems have significant security and reliability vulnerabilities which put the integrity of national, state and local  elections in real danger.
• Few jurisdictions had carried out measures which would have made it more difficult for less complicated attacks to have been possible.
• The machines most vulnerable to attacks are those with wireless components. Just the states of New York and Minnesota had banned wireless components from their voting machines.
• Creating a voter verified paper audit trail for auditing purposes is only useful if routine auditing is actually set up. Of the 26 states which required voter verified paper audit trails, only 12 demanded regular auditing.
• There should be a realistic testing of a random sample of voting machines on Election Day.
• Clear and efficient procedures should be institutionalised to deal with evidence of fraud or error.

The Election Data Services, Inc. published a  map of voting equipment used in each USA county in November 2006. In terms of the percentage of electors covered by each of the systems, the situation was as follows: 48.9% of systems used optical readers of paper ballots, 38.3% direct electronic voting systems, 6.8% systems involving mechanical voting with levers, 5.5% with mixed systems, 0.2% with perforated cards and finally 0.2% using paper only. Almost one third of the registered voters used new equipment for voting, which had chiefly been acquired following the effort to modernise the technology as envisaged in the HAVA – Help America Vote Act of 2002. As intended, perforated cards had been almost completely eliminated (falling from 26% in 2000 to 0.8% in 2006) and the systems using levers had reduced significantly (from 18% in 2000 to 6.8% in 2006).

An Election Assessment Mission from the Office for Democratic Institutions and Human Rights of the OSCE – Organization for Security and Cooperation in Europewas present at the interim elections for Congress which were held in November 2006 following an invitation from the Government, and it produced the report (The United States of America Mid-Term Congressional Elections 7 November 2006 - OSCE/ODIHR Election Assessment Mission Report) which contains a set of recommendations, including the following:

• Reinforced measures to ensure transparency and confidence in voting integrity, such as access to software source code, independent tests, supplying a voter verified paper audit trail, and multiple auditing mechanisms, Out of the 37 states which used direct electronic voting machines, 22 required a voter verified paper audit trail, and 15 did not require this.
• Reinforced security measures for the storage and transportation of electronic voting equipment to ensure it is not tampered with. In certain cases where it was not possible for the equipment to be stored in a public facility on days prior to the elections, the equipment remained in the houses of election support staff.
• Components which have wireless communication and use of external communication networks should be excluded.
• Introduction of the need to use independent security access codes for different individuals which are independent of each other for certain operations involving the management of voting equipment.
• Compilation of a single electoral register for the country, in order to avoid multiple voter registration. Legislation required that each state should have a single electoral register by 1 January 2006, but 12 states did not comply with this deadline.
• Credible verification of voter identity in the polling stations using common criteria. Less than half of the states required that all voters showed an identification card, despite the fact that in states where the presentation of an identity card is required, provisional voting is not allowed until identity has been confirmed. Three states required the showing of an identification card showing a photograph, but allowed normal voting if the voter presented a written declaration of responsibility.
• Reconsideration of the possibility of voting by fax, which is offered in limited circumstances, as this may compromise the secrecy of the vote.
• Revision of voting by postal vote so as to ensure that security procedures carried out in polling stations also apply to as great an extent as possible to postal votes. This type of voting raised serious concerns: there is no way of ensuring that voters cast their vote in secrecy, and the possibility of the vote being shown makes it more vulnerable to coercion; the identification mechanism is weakened; counting is held back because votes may arrive several weeks after Election Day; votes may arrive in an unreadable condition so that they cannot be read by optical reader machines and in these cases the election support staff have to write out a vote in conformity with that received, which also raises a number of issues.
• Specification of common standards for testing, certification, use and data protection, and ownership of electoral data records and voting, as well as equipment.

On 18 May 2007 the Department of Defense (DoD) of the USA published a new report on voting via the Internet for members of the armed forces abroad (Department of Defense: Expanding the Use of Electronic Voting Technology for UOCAVA Citizens) which ignored most of the objections which had been made in the 2004 evaluation, which led Jefferson, Simons and Rubin to publish a commentary on the new DoD report (A comment on the May 2007 DoD report on Voting Technologies for UOCAVA Citizens).

On 31 August 2007 the TGDC – Technical Guidelines Development Committee, set up under the scope of HAVA – the Help America Vote Act of 2002 and researchers from NIST – the National Institute of Standards and Technology sent the document Voluntary Voting System Guidelines Recommendations to the Election Assistance Commission which they had prepared for a future voting system standard, written with the aim of addressing issues relating to the next generation of electronic voting equipment and a complete revision of the Voluntary Voting System Guidelines of 2005 (Vol. 1 – Voting System Performance Guidelines and Vol. 2 – National Certification Testing Guidelines), the first which were produced and developed by the EAC – Election Assistance Commission,under the scope of HAVA.

CANADA (remote electronic voting and at polling stations)

In Ontario the city of St. Catharines has used optical readers and electronic counting of paper ballots since 1988. Windsor used electronic ballot boxes with a touch screen in 2002 and 2003, in the period before Election Day. Elections were held between 5 and 10 November 2003 which involved remote voting in 12 municipalities via the Internet or via telephone, without any voting using paper ballots.

Around 100 thousand voters registered to vote. 55% of the voters cast their vote, clearly more than have usually done so in previous municipal elections where the figure is normally between 25% and 30%. That State set up the Secretariat for Democratic Renovation after those elections to work on proposals to reform the electoral process in Ontario, including making the Internet a voting option. Markham used remote voting via the Internet in 2003 and 2006, Kingston used electronic ballot boxes with a touch screen before Election Day in 2006. Ottawa used machines to optically read paper ballots in 2006. Prescott also experimented with voting via the Internet.

In Quebec the first uses of electronic voting in local elections date back to 1996 and have been steadily increasingly since then. In 2001 around 2.3 million voters (42% of the electorate) voted electronically in voting machines or electronic ballot boxes with optical readers for paper ballots, both of these being used in polling stations  In November 2005 more than 50% of the voters in Quebec cast their vote electronically. In these elections there were numerous problems which led the Director-General of Elections to seek an evaluation of the electronic voting processes and suspend the use of electronic voting in elections and referenda. The evaluation report and annex was published on 24 October 2006, the day on which the Director-General of Elections recommended the continuance of the moratorium regarding the use of electronic voting and stated that these systems presented important risks associated with the absence or lack of formal protection and security measures, that they had defects and broke down when in action or working in a network, that the way in which the electronic ballot boxes and optical reader voting terminals were tested, installed and administered was also worrying, and in general noted a lack of knowledge of the voting system components on the part of the technical and administrative staff responsible for them, including gaps in the technical abilities of the latter and the suppliers of the services.

In elections in 2004 Alberta had also used electronic ballot boxes with touch screens before Election day.

BRAZIL (electronic voting at polling stations)

The setting up of a central database containing the electoral register for Brazil was started in 1987 by the Supreme Electoral Court of Brazil. Between 1993 and 1994 a network was set up linking all the regional electoral courts and allowing regular updating of the national voter database.

By 1996 the number of electors had reached more than 101 million and electronic voting system use was started in Brazil, with the aim of reducing the time taken to calculate the results of elections, which was in the order of several days, as well as reducing fiscal fraud, and facilitating voting for the illiterate.

The system used is one of electronic ballot boxes in polling stations (see the photograph below) in which each ballot box has a screen where each voting alternative appears alongside a several-figured number and an image of the candidate and the vote is recorded by each voter on a numeric keyboard, with different numbers previously attributed to different voting choices before the electoral campaign (during the electoral campaign the candidates also publicise their identification numbers for voting purpose), after registering their identity in the electronic ballot box by the polling station board. After polls have closed the polling station Chairperson introduces his/her identification code in the ballot box to obtain a paper record using an attached printer of the total votes cast for each candidate and the total of spoilt votes. These results are put up and an encrypted recording on diskettes or CDs is made, and preparation is made to transmit the results to the corresponding electoral tribunals. In different versions of the system and its development this transmission was carried out in various ways: transportation of diskettes or CDs, or transmission through VPN by satellite or landline. The machines can use 110 or 220 V mains power and can also run off batteries with an initial battery life of 3 hours which extended to 12 hours in 2002.

In 1996 around 32% of the votes were cast in approximately 77,000 electronic ballot boxes manufactured by Unisys in 57 locations.

In 1998 around 58% of the votes were cast in around 167,000 electronic ballot boxes manufactured by Diebold-Procomp and Unisys in 537 locations.

In 2000 100% of the voting was carried out using electronic ballot boxes. The number of voters totalled more than 109 million persons and more than 400 thousand electronic ballot boxes were used and the results were announced in less than 24 hours.

Elections in 2002, 2004, and 2006 also had 100% of the votes being cast using electronic ballot boxes, with the number of electors in 2006 being 125 million, requiring the use of more than 426 thousand electronic ballot boxes.

Photography of a Brazilian Electronic Ballot Box (2008):

Criticisms have been levelled at the voting system, including the following:

• In 1985 the Electoral Supreme Court decided to remove the photograph from the voter’s electoral identity card, which weakened the identity check of the voters. This was only corrected in November 2005.
• In 1999 the Senate received draft legislation which envisaged the machine having to print out a ballot paper as a form of validation by the voter; the obligation to carry out a random audit of at least 3% of the electronic ballot boxes after the election; the impossibility of recording the identity of the voter on the same machine, and the use of open and free software. However, in 2001 the Electoral Supreme Court made several suggestions to Congress to alter the draft legislation: 1) delay the printing of a paper ballot as a means for the voter to check the vote until 2004, along with the requirement to audit the electronic ballot boxes. 2) establish a sample of machines to be audited before the election. 3) allow the recording of the voter's identification on the machine where he/she votes. 4) enable the use of software which is not free and the code of which has not been submitted for external audit. In 2003 the Supreme Electoral Tribunal suggested to Congress that it pass legislation to cease printing paper ballots for verification purposes and also take away the requirement to audit 3% of the electronic ballot boxes, which had previously been envisaged for 2004, using the argument that paper trail tests carried out in the elections on 2 October 2002 had shown the reliability of the system and eliminating printers from the machines would save 100 million dollars and make voting considerably quicker.
• There is the conviction that almost 90% of the electronic ballot boxes used in the elections in Brazil have security flaws which enable voting results to be altered, with the integrity check tests alluded to, given that 375 thousand of the 426 thousand electronic ballot boxes used in the October 2006 elections were manufactured by the Diebold company, under the brand name Diebold-Procomp, and given the security flaws identified in the reports published by the NGO Black Box Voting, regarding the voting machines used in the USA elections, specifically refer to Diebold voting machines: “Hursti Report” – 1st part (The Black Box Report – Security Alert: Critical Security Issues with Diebold Optical Scan Design, July 4, 2005), “Hursti Report” – 2nd part (Diebold TSx Evaluation : Critical Security Issues with Diebold TSx, May 11,2006, “Hursti Report” – supplement (Diebold TSx Evaluation – Supplemental report, additional observations, May 22, 2006). The concern with voting security for these machines was reinforced by the security problems detected by researchers from the Center for Information Technology Policy, Princeton University which were published in the article Security Analysis of the Diebold AccuVote-TS Voting Machine, on 13 September 2006, and through the video available at http://citp.princeton.edu/voting/ which shows how a voting machine can be altered in less than one minute so as to provide different voting results. Verification tests are not able to detect these, as is the case with viruses used for such purposes to contaminate other voting machines without human intervention.

PHILIPPINES (electronic voting at polling stations)

The first experimental piloting of electronic voting in the Philippines took place in the 1996 autonomous Muslim Mindanao elections, one year before the electoral system reform which made electronic voting official.

In the May 1998 General Elections official electronic voting was carried out in the same region in 6 polling stations with 68 machines with optical readers for paper ballots. Due to various technical problems these machines stopped being used.

Two systems of voting, both located in polling stations, were used in the August 2008 elections, also in Muslim Mindanao:  electronic voting machines with touch screens (in the province of Maguindanao) and optical readers of paper ballots (in the provinces of Basilan, Sulu, Tawi-Tawi, Shariff Kabunsuan and Lanao del Sur), with the aim of producing results which could be useful in determining which option to follow with regard to the generalising of electronic voting throughout the country in 2010. Around 85% of the 10.5 million electors cast their vote. 3,300 voting machines were used in Maguindanao and 156 automatic paper ballot counting machines with optical readers in the other 5 provinces. This electronic election cost around 14 million dollars.

INDIA

Electronic voting was introduced in India in 1982 through an experimental pilot in elections for the North Parur assembly in the state of Kerala, but the Supreme Court of India ruled this election to be illegal.

Legislation was altered in 1989 so as to allow the possibility of electronic voting.

Just as in Brazil, the main motivation to introduce electronic voting in India was to reduce the time to ascertain the results of the elections, reduce fiscal fraud and facilitate voting for the illiterate, in a country with more than 600 million voters, as well as to avoid the need to print hundreds of millions of voting ballots and the associated costs of the paper, printing, transport, storage, security and distribution.

India introduced an Electoral Identity Card containing a photograph in 1993 which more than 70% of the electorate were using by 2008 and which had become obligatory to use in future elections. In 2004 a pilot project was started to prepare electoral registers containing photographs of voters, to be published on the Internet and in public places, including post offices, prior to the elections The aims of these measures are to provide greater control of identity in polling stations and reduce the possibility of identity fraud.

Electoral registration in India only started to be digitalised in 1998, initially involving some states in the North of the country and the country as a whole was completely digitalised before 2004.

The first experiments with electronic voting were held in November 1998 in 16 local polling stations in the States of Madhya Pradesh (5), Rajasthan (5) and NCT of Delhi (6), using electronic voting machines manufactured in 1989-90 which have developed into the ones currently in use throughout India. 

In 2003 all state and intermediate level elections were held using electronic voting machines.

In the elections for the Lower House in 2004 electronic voting machines were only used in around 700 thousand voting locations, equipped with more than 1 million electronic voting machines, where 390 million of the 671 million voters cast their vote. The Elections Commission estimated that the use of electronic voting machines saved one and a half million trees from being needed to produce 8 thousand tonnes of paper, as well as reduce the number of polling stations from 770 thousand to 700 thousand, since they enabled the maximum number of voters for each polling station to rise from 1,200 to 1,500. There were 1,351 candidates from 6 national parties, 801 candidates from 36 state parties, 898 candidates from other officially recognised parties and 2,385 independent candidates. The Elections Commission employed more than 3 million people to organise the elections.

The machines are not connected to any network for security reasons. They run off a 6 V power source and have batteries which enable them to function when necessary without being connected to the electricity network. The voting machine is the size of a travelling suitcase and consists of two units: the Control Unit used by the people in the polling station and the Voting Unit. This has a panel where the list of names and candidate symbols appear and these are lined up with switches which are pressed to select the chosen candidate for office (see the figure below). The machines’ memories are designed to last and keep a record of the votes for at least ten years, if so wished, even without a power supply, which also makes it possible to replace a voting machine if it breaks down, without any votes cast being lost. The machines were designed by the Elections Commission in cooperation with two public companies and are manufactured by these companies. The microchip on which they are based is manufactured in Japan through a special order from India, and has a seal which cannot be broken until used.

Photograph of voting machine from India - Control Unit and Voting Unit (2008):

From 2004 onwards, elections have been totally carried out using electronic voting machines. There have been state legislative elections in part of the 30 States, specifically 2 States in 2005, 5 States in 2006, 7 States in 2007 and 1 State in 2008.

The counting of votes in a national election, which involves around 400 million votes and around 5,500 candidates, is carried out in 2 to 3 hours instead of the 30 to 40 hours it used to take using a conventional paper balloting system. Besides this, voting itself is quicker than using paper. Transporting the voting machines is easier than the ballot boxes for paper voting.

The costs of introducing an electronic voting system were estimated at around 300 million dollars, a cost which should be made up after a few elections in terms of the savings in paper and transportation.

Electronic electoral registration was started on an experimental basis in July 2004 with a pilot project in Mumbai and two districts in Madrasta, and was enlarged to the whole country in 2008.

Problems which have been mentioned include the fact that the machines do not supply a paper trail and the source code of the software is not available for inspection.

VENEZUELA (electronic voting at polling stations)

Venezuela passed electoral legislation in 1998 stating that elections should be entirely carried out using electronic voting, with the aim of reducing electoral fraud present in the country's elections and making counting easier and more exact.

Between 1998 and 2004 an electronic optical reader system for paper ballots was used.

In the referendum on the re-election of President Hugo Chávez, which took place in August 2004, all the voting was carried out in polling stations using new electronic voting machines.

The electronic machine has a touch screen where an image of the ballot paper on which the vote is made appears, an accessible USB port where a pre-programmed memory stick is inserted for each polling station, where votes are recorded in encrypted XML files as well as logs and other data, and also on the machine's hard discs, so as to provide another record of the election results and in the event of a technical problem voting can continue with another machine. A printer is also attached to print out paper ballots to establish a paper trail. The system uses a randomising algorithm when sorting the recorded votes, to minimise possibilities of identifying votes based on their order. The machine is powered by electricity or by a car battery and runs an adapted version of the Windows XP operating system. It was designed by the Smatmatic company in cooperation with the National Election Commission (CNE).

Once the elections have finished, the machine prints out a count of the votes with the final results and only afterwards is the machine linked to the external network through a VPN to communicate the results to the data gathering centre through landline, satellite or mobile phone, and when no links are available the results are transported on the machine’s memory stick to the nearest voting station from where they can be transmitted. The data gathering centre only accepts telephone calls from pre-determined numbers and does not accept transferences before polling stations have closed. The log files are not transmitted.

The CNE is the owner of the source code utilised and has audited it, but this code is not available for public scrutiny, neither is there any hardware nor software audit external to the CNE. In the referendum on the Presidential election in which the machine was used for the first time a paper trail audit was used for 55% of the votes.

These elections also used a separate electronic voter identification system based on fingerprint reading and comparison with a remote server.

The same electronic voting system was then used in the 2004 Governmental elections, the 2005 municipal elections and the 2005 Parliamentary elections.

In the 2005 Parliamentary elections the identification system based on reading fingerprints was not used because risks were detected regarding identifying votes through the data correlation available from the two systems.

A European Union observation mission was present during the 2005 Parliamentary elections. This mission’s report (European Union Election Observation Mission Venezuela 2005 – Final Report), which was published in March 2006, mentioned that it was not aware of the audit procedures adopted by the CNE and it was not clear if the CNE technicians had been sufficiently trained in order to perform such audits. This mission also identified other failings, including:

• Voters had problems using the machines in more than 75% of the monitored polling states and had to ask for help. In many cases help was given by people not belonging to the polling board and who were linked to parties or were members of the armed forces carrying out security duties. In some cases the person helping cast the vote. These situations raise serious doubts about the secrecy of the vote. There were also cases in which the voters did not manage to vote in the 3 minutes which was allocated to them and thus ended up not voting.
• In a very few cases voters were able to vote without their identity cards.
• The use of members of the armed forces to ensure security seemed excessive and beyond what is laid down in the law, given their presence in polling stations is only permitted when requested by the polling station boards, and their role was often not just one of providing security but also offering various types of assistance to voters.
• Country, party and other observers were present in only 6% of the polling stations visited, a percentage which rose to 19% during the counting and aggregation of the results.

Photography of an electronic voting machine from Venezuela (2005):

 

JAPAN (electronic voting at polling stations)

Various experimental pilots have taken place in local elections in Japan since 1999 using voting machines with touch screens located in polling stations. The first experiment involved 11 of the 78 polling stations in the municipality of Kawaguchi where there were 55 thousand registered voters. The system used a card with a magnetic band to enable the voter to cast their vote and actually vote on the tactile screen.

Legislation was passed in 2001 to enable electronic voting in local elections.

Other experimental piloting of this nature took place in June 2002 in Niimi, in which 19,381 voters (6% of the electorate) voted electronically in 43 polling stations, in February 2003 in Hiroshima, in July 2003 in Sabae and in Kani, where the results of the election were annulled by the Supreme Court in July 2005 due to problems with electronic voting which resulted in long periods where it was not possible to vote in all the polling stations (from 9 minutes to 1 hour and a half).

In 2005 the Government admitted that of the 12 elections involving electronic voting which had taken place until that time, only 3 had not had any problems.

PANAMA (electronic voting tests in polling stations)

In the 1999 elections, experimental piloting of electronic voting was carried out at various points in Panama.

AUSTRALIA (electronic voting at polling stations)

In Australian the Australian Capital Territory (ACT)  has had the greatest experience of electronic voting in elections for Parliamentary Bodies. The first parliamentary elections in which it was possible to vote electronically took place in October 2001, in which 8.3% of the votes were submitted electronically in four locations in the two weeks prior to and in eight locations on the actual Election Day. The votes were authenticated by the voter with a card containing a bar code with an electronic signature which was randomly assigned by the Chairperson of the Polling Board. There was no transmission of votes through a network, such that voters had to vote at their own polling station. Votes cast on paper were digitalised by two independent operators so that electronic counting of all votes could be carried out.

The system was also used in elections on the 16 October 2004, in which 13.4% of the votes cast were electronic. The Report of the Electoral Commission of 27 June 2005 (Electronic Voting and Counting System – Review) recommended that the same system be used in the October 2008 elections, although with the future possibility of scanning the paper ballots for automatic digitalisation. The Electoral Commission maintained its view that it was not appropriate to use Internet voting for Parliamentary Elections in the near future, given that it was not considered a secure medium for this purpose, and that electronic voting should remain only in the controlled environment of polling stations.

The Australian Federal Elections of 24 November 2007 offered the possibility of electronic voting only to persons with sight impairments or members of the Armed Forces abroad. The votes had to be cast at a predetermined balloting post and the voters had to enrol for electronic voting in advance, in September 2007.

The Final Evaluation Report of the Australian Electoral Commission (AEC) for the Armed Forces was issued in March 2008. The Defence limited remote electronic voting to those serving in Iraq, Afghanistan, Timor Lorsae and the Solomon Islands since it was only possible in these territories to access the Australian Defence’s secure communication network. Voters who took advantage of the remote electronic voting comprised 60% of the potential electorate. Of the voters casting their votes electronically, 55% mentioned a lack of privacy when exercising their right to vote. The costs per vote were high ($521, when compared with the average cost of a conventional vote of $9.49).

The Final Evaluation Report of the AEC for voters with sight impairment was also issued in March 2008. These voters were able to use voting machines in 29 voting locations, in the two weeks prior to Election Day. The votes were printed on to a bar-coded system to then be decodified and added to the other votes for counting purposes. It is estimated that the voters who used electronic voting made up 4.2% of the voters with sight impairment in the areas of the 29 voting locations equipped for this purpose and around 0.5% of all the voters with sight impairment problems. The costs per vote were high ($2,597).

PARAGUAY (electronic voting at polling stations)

The first experimental piloting of electronic voting in Paraguay took place in the municipal elections of November 2001, after the ruling of the Supreme Electoral Court of Justice (TSJE) in the same year. This experiment was made possible through a cooperation agreement between the TSJE, the Organisation of American States (OAS) and the Supreme Electoral Court of Brazil, which guaranteed technical assistance to the project and loaned 178 electronic ballot boxes, of which 119 were used in polling stations covering around 34 thousand voters.

In the Presidential and Parliamentary Elections of April 2003, once more within the cooperation framework with the OAS and Brazil, Paraguay became the second Latin American country to officially make a binding electronic voting system available, covering more than 50% of the electorate and involving  around 3,800 electronic ballot boxes.

All the elections which have followed have used electronic ballot boxes loaned by Brazil and which have covered the majority of voters. From 2005 Paraguay was able to make use of more than 17,000 loaned electronic ballot boxes.

However, after more than four years of voting mainly carried out electronically, distrust in the security of electronic ballot boxes had grown to such an extent that the TSJE took the decision in February 2008 that electronic voting would not be used in the April 2008 Presidential Elections and that all votes cast would use paper ballots.

COSTA RICA (electronic voting tests in polling stations)

On 1 December 2002 experimental piloting of electronic voting was carried out in the local elections of Costa Rica.

GUATEMALA (electronic voting tests in polling stations)

In December 2002 experimental piloting of electronic voting was carried out in the local elections of Guatemala in 16 polling stations. The system was based on voting machines with a screen and numeric keyboard, with audio instructions. In one of the polling stations the process was suspended.

ARGENTINA (remote electronic voting and at polling stations)

There have been various experimental pilots with electronic voting machines in polling stations in Argentina from 2003 onwards, in the city of Ushuaia with 105 electronic voting machines, and at various places in the province of Buenos Aires with electronic ballot boxes lent to it by Brazil. The electronic voting programme for Buenos Aires has gradually been extended since then.

In December 2005 the province of Mendoza carried out experimental piloting of remote electronic voting.

ECUADOR (electronic voting at polling stations)

The Supreme Electoral Court of Ecuador decided to start using electronic voting machines in polling stations in the local elections of October 2004, with 0.8% of the polling booths using electronic ballot boxes given to it by Brazil. It designed a plan for the steady increase in electronic voting which has never undergone this planned development.