Technology and the Voting Process
V. Experience with Voting Technologies
A. Canadian experience
1. Televoting in Canada
The Canadian experience with electronic voting options has been limited largely to municipal elections and provincial political party leadership selection. At the provincial and federal electoral levels a number of factors have served to moderate the administrative demand for shifts to electronic voting: the infrequency of electoral events, the relatively small population densities throughout most of Canada, and the fact that only one electoral contest is held per event.
2. Provincial political party leadership conventions
Several provincial political parties have used televoting (voting by telephone) to choose their leaders. The reason given was generally the attempt to increase participation to reflect the democratic norms of party memberships. Participation generally increased but participation rates in such contests are also dependent upon the relative competitive position of the party and the openness of the process in terms of rules adopted.
Concerns with televoting to date have included the ability of the telephone network to handle the peak period volume of calls, the technical capability to trace votes through PIN matching techniques, PIN-based fraud, the reduced sense of political community and possible coercion and intimidation.
Canadian political parties have utilized televoting on four occasions for the selection of party leaders – the Liberal Parties of British Columbia, Nova Scotia and Alberta and the Saskatchewan Progressive Conservative Party. Below we provide information on the Nova Scotia Liberal Party Leadership contest, one of the earliest examples of televoting in Canada.
A study of the Nova Scotia Liberal Party Leadership Contest conducted by Ian Stewart, Agar Adamson and Bruce Beaton from Acadia University suggests that the use of televoting for the contest not only altered the size but composition of the electorate. As the authors note, "the number of eligible voters in 1992 was almost four times what it had been six years previously?(and) it is not implausible that certain types of individuals, who would have been unlikely to participate had the Liberals held a traditional leadership convention, were effectively enfranchised by the new procedures. Specifically, the convenience of being able to vote at home should have encouraged the participation of disproportionately more people from the furthest reaches of Nova Scotia."Footnote 15
Stewart, Adamson and Beaton draw three conclusions from the Nova Scotia leadership experience:
(1) that the process was remarkably open – if voters paid the requisite fees they could vote – raising concerns over tactical voting as in American states which feature cross-over primaries (where individuals can be registered to vote for leadership contests in more than one party);
(2) that there are legitimate concerns over televoting security – individuals could vote as many times as they had PINs; and
(3) that although televoting does provide ease of access, 56 percent of party members did not vote.Footnote 16
This third conclusion can be explained as the result of a number of factors from the registration fee to member apathy. Nevertheless, it is an important reminder that electronic voting options should not be pursued as a solution to voter apathy and disillusionment but rather as a means of improving accessibility and providing voters with more options for casting their decision.
While the televote ended successfully, it had to overcome a number of hurdles including a complete systems failure on the first attempt as a result of system capacity issues.
After the initial failed attempt at the Nova Scotia Liberal Party Leadership Contest, KPMG was engaged by the Liberal Party of Nova Scotia and the supplier – MT&T Technologies to (1) determine whether the system could be structured to provide integrity, completeness and confidentiality and if so, (2) conduct a review and issue an opinion on the adequacy of control procedures, and finally (3) audit and express an opinion on the vote results.
KPMG confirmed that an appropriate system could be structured, and subsequently issued opinions on the adequacy of control procedures and the contest results. Subsequently, KPMG updated the opinions for each use of the televoting system and issued an opinion on each of the vote results. One of the subsequent leadership contests was reported to have had technical problems – which in our view were directly related to a lack of consistent application of predefined procedures outside of the technical components of the televoting system.
The result of Canada's four provincial leadership contests are consistent in that they all eventually worked. However, all of these contests required significantly more planning in advance of the contest than originally anticipated by some of the organizers. Generally, those contests that were well planned and included comprehensive procedures and guidelines were executed with very few problems. Inadequate planning, testing and failure to consistently apply necessary procedures resulted in difficulties for two of the contests.
The use of televoting and in fact most technologies requires a consistent and well structured process, with well pre-defined and described deviations and their resolutions.
3. The experiences of Canadian municipalities
A number of Canadian municipalities are using electronic vote counting systems. The reasons include: new complexities in electoral processes, increases in the number of electoral contests per event, difficulties in recruiting and training polling workers, onerous manual counting processes, decreasing operating budgets, and controversy over rejection of ballots.
It is possible that some of these same factors will lead municipalities to explore electronic voting options as well.
For example, the City of North York held a telephone referendum on the Metro-Toronto amalgamation. The referendum administrators found that a telephone referendum could be successfully conducted in a municipality containing 400,000 voters, that it could be conducted in a number of languages, that confidentiality of the vote can be guaranteed, that the vast majority of the public can use the system, and that a telephone referendum can be conducted for about half the cost of a regular election.Footnote 17
The North York experience identified three possible challenges to future use of telephone referendum and voting. The use of this technology requires some form of unique identifier to qualify a voter. The use of a 10 digit PIN was selected – and such an identifier was distributed to qualified voters by mail. Some qualified voters living in multi residential complexes (apartments) reported that they failed to receive their PINs. There was some speculation that PINs were stolen and inappropriately used.
A second problem arose from the volume of callers which exceeded the capacity of the telephone system as it was configured. Some experience was gained regarding the time of calls, the type of telephone used by callers and the duration of calls. This experience subsequently was used to better tune the system.
Finally, some businesses raised the concern that their access to the telephone system supporting their businesses (telephone orders, credit card authorizations, etc.) could be compromised by the system load of a televote.
4. Electronic voting at the provincial level
A move from manually counted ballots to the use of electronic voting or vote tabulation is generally not seen as feasible by provincial governments at this time. There is not yet a clear need to implement new systems from the current electoral demands placed on voters (approximately once every four years) in a "first-past-the-post" system with uncomplicated ballots (unlike longer, more complicated municipal ballots). These make the electoral process at both the provincial and federal levels relatively simple.
In addition to this lack of clear need, some provincial administrators profess skepticism as to the accuracy and security of electronic voting meansFootnote 18. Another key obstacle for provincial administrators is the magnitude of the up-front development and capital costs which are associated with such a venture. This underscores the importance of partnerships between provincial electoral agencies and Elections Canada.
5. Canadian universities
In the Summer of 1997, the University of Calgary Student Union was approached by a Calgary-based firm called Voting Systems International (VSI) which was interested in finding a venue to test its voting technology. In the Fall of 1997, VSI carried out student government by-elections for free using its touch screen terminal voting technology. Based on the success of the by-election, VSI carried out the general student government election in February 1998, for a nominal fee of $3,000. VSI had an opportunity to test its technology, while the University dramatically decreased the number of spoiled ballots and increased the speed of counting ballots. The use of the technology did not increase voter participation; however, a random survey of voters indicated the ease with which people used the system.
VSI's electronic voting system uses touch screen terminals supported by laptop computers which store voting results. Votes cast using the terminals are transmitted to the laptops using low frequency radio waves. The results can either be stored electronically on each individual laptop and then downloaded periodically, or transmitted in real time (or periodically) to a central computer.
In the University of Calgary situation, the process commenced with potential electors presenting themselves to the polling clerk, telling the clerk their student number and presenting some form of picture identification. The student number was used to activate the elector's ballot, a ballot which would be different depending on the elector's course of study and faculty (electors voted for council executive as well as faculty representatives). The ballot was then transmitted to one of a number of touch screen terminals set up behind cardboard partitions much like traditional polling booths. The system prompted electors for their responses and allowed electors to change their ballot before submitting it. When the ballot was submitted, it was transmitted back to the laptop and the result stored electronically. At the time of submission, the identity of the elector was separated from the ballot.
Following the event, the system is capable of printing off a number of reports including the results and whether or not an elector voted more than once by visiting different polling stations. Because there was more than one polling station, and the system was not connected in real time, it was possible for an elector to vote more than once. If election administrators had found that someone has voted more than once, the university could have enforced a number of penalties of varying severity. VSI notes that this problem can be mitigated by connecting the polling stations and the real time transmission of who has voted to all polling stations.
B. International experience
Our survey of international experiences has led us to the conclusion that electronic voting is still in its infancy.
While there are numerous examples of jurisdictions which employ various technologies for the purposes of vote tabulation, examples of organizations (e.g., political parties, broadcast companies) which employ technologies like the telephone for the casting of decisions or opinions, and evidence that various jurisdictions have researched electronic voting, there are very few examples of jurisdictions which are planning to employ technology in the act of casting a vote and even fewer which have successfully done so. There are fewer still examples where the use of technology from remote locations (i.e., from locations other than the polling station like home or office) is planned. However, there are also a number of firms which are developing and marketing voting technologies or products which support the feasibility of electronic voting such as encryption software and verification tools.
Below we present some interesting cases from our research.
For mayoral and county representative elections in 1996, 33 million Brazilian voters made use of laptop computers in polling stations to record their votes electronically. Electronic voting involved 52 of Brazil's largest cities including the 26 county capitals and 26 other cities with voting lists of more than 200,000 electors.
The Brazilian Federal Superior Electoral Tribunal, which oversees election matters, introduced electronic voting as a means of reducing fraud and increasing the speed of the vote count (from 30 days to approximately six hours).
The electronic voting system was built by Unisys Brazil. Despite security and integrity concerns around electronic voting, electoral administrators view the system as far better than the conventional paper-based balloting once marked by long waits to vote, frequent recounts, and ballot box fraud.
The system runs off small computers, like laptops, specially fitted with numerical keypads and three large keys in different colours (all keys have Braille for blind voters). Each candidate in the election has a code number, which the voter enters into the computer after registering with the supervisor at the polling station. A picture of the selected candidate appears on screen, complete with details regarding party affiliation. The voter then uses the coloured keys to confirm the choice, correct the choice, or decide not to make a choice.
Each computer stores details of the votes recorded on a disk using double encryption. The computer does not store the identity of voters. At the end of polling day, the disks are sealed and transported to local consolidation centres for computation. The results are then transmitted to the Federal Electoral Court in Brasilia.
The Belgian electoral system is one of proportional representation.
To increase the speed of the count and the ease with which voters cast their decisions, since 1985, a small firm from San Diego, California called dZine has been developing a number of prototypes for the Belgian government which employ light pen technology and standard personal computers (employing standard software packages such as Windows). Electors cast their decisions by using the light pen (much like one would use their finger with a touch screen kiosk) to mark their selections on the computer screen.
In 1985, the government contracted dZine for a pilot project of 60 machines. Since then, the government has undertaken a long and detailed testing process and has tendered for electronic voting systems for several regions of the country using the light pen-standard personal computer design. The technology is used in both local and regional elections.
The "light pen" voting process unfolds as follows. Electors present themselves to the polling clerk at the polling station at which point their eligibility to vote is confirmed. Once confirmed, the elector is given a credit card-size card and directed to one of many terminals located in the polling station. Electors cast their decisions on the screen using the light pen, and then have a choice of either sending their decisions to the system or changing their vote. At the same time as the elector sends their decision to the central electronic repository, the elector also sweeps the card they were given at registration through the personal computer's card sweeping device. This action stores the decision on the card as well. Then, the voter deposits this card in the traditional ballot box. Sending the decision electronically to the central repository facilitates the speed of the count while storing the information on the card which is then submitted in the traditional ballot box provides a back-up means of verifying the count.
The Centre for Information Law and Policy (CILP), a joint effort between the Villanova University School of Law and the Illinois Institute of Technology's Chicago-Kent College of Law, recently formed a "Project Bosnia" team which worked toward the establishment of an Internet-based voting system in Bosnia.
The Dayton Accords, which brought an end to the fighting in Bosnia, allow all Bosnian voters to vote in the elections of their home region. This includes those voters who had been displaced from their home regions during the war. In an effort to alleviate the tension that would result from voters returning to land occupied by former enemies on election day, the CILP team began working on a system that would allow voters to cast ballots using the Internet. As a preliminary trial, Villanova Law School developed an Internet-based voting system that the school used for internal student elections. Although the Bosnian Internet-based voting system has not advanced much beyond the planning stages, the CILP team attracted the attention of the International Foundation for Election Systems (IFES) which was working with the Government of Costa Rica on an Internet-based voting project.
4. Costa Rica
IFES had been working with the Government of Costa Rica on the possibility of Internet-based elections. In the Summer and Fall of 1997, members of the CILP team went to Costa Rica to meet with government officials about the possibility of providing assistance to this end. The Costa Rican Supreme Electoral Tribunal agreed to the idea of planning and testing an Internet-based election system.
One of the key challenges for the Costa Rican Supreme Electoral Tribunal is that only about 65 percent of qualified Costa Rican citizens go to the polls. In part, this is because citizens typically do not register to vote any place other than where they originally registered when they reached the age of 18. At election time many eligible voters must decide whether to travel in order to vote or not vote at all. As a consequence, political parties spend millions of dollars each election year to bus voters to polling stations or to supply them with gas money to get to wherever they happen to be registered leaving the Costa Rican system unduly exposed to inducements and coercion.
The CILP team offered to address this key challenge as follows.
- The team would set up five to ten test polling stations where constituents could vote both manually and electronically. Training would be completed on-site for voters who needed it.
- Citizens would be able to vote at their local public school, and would not have to travel to the place where they were registered to vote. Locally registered voters would use the manual process, while those who were registered elsewhere would vote electronically.
- Each "electronic" voter would present an identification number to someone at a registration desk, and the registration clerk would enter that voter's name into a computer at each polling station. The Internet would then be used to verify that that voter has not actually voted before, and also to deliver the appropriate ballot for that voter.
- The electronic interface would show photographs of the various candidates and their party flags, and be supported by either a touch screen or a light pen for voters to choose the candidate of their choice.
- Votes would be transmitted electronically, over the Internet, to a central server run by Project Costa Rica and AT&T on behalf of the Electoral Tribunal.
The benefits of such an approach were to include reduction of voter fraud and political coercion, increased voter participation, and improved electoral administration efficiency. The CILP team also proposed the possibility of making registration functions available over the Internet.
However, Internet-based voting is yet to be introduced in Costa Rica because of political opposition (the government did not feel comfortable about the level of system security) and also because of the country's political culture (election day is seen as a national holiday, a day on which people travel to their home cities, towns or villages to visit friends and family).
5. State of Florida
In October of 1997, the State of Florida announced an initiative to provide voting via the Internet for 1998 state elections. Internet voting is to be offered as an option to overseas military and civilian voters as a means of alleviating the time constraints of the current mail-based, absentee voting system and in so doing allow for greater participation.
Internet voting will be similar to Florida's current absentee voting system. An overseas voter will request an authorization to vote via the Internet. Upon receipt of this authorization, the voter will complete a virtual ballot on his/her computer terminal. The vote will be registered instantly with the supervisor of elections, without the present fear of loss in overseas mail.
To date, these enabling provisions have not been tested in an electoral event in Florida.
Return to source of Footnote 15 Stewart, Ian, Adamson, Agar and Bruce Beaton; Pressing the Right Buttons: The Nova Scotia Liberals and Tele-Democracy, in Stewart, Ian; Roasting Chestnuts: The Mythology of Maritime Political Culture, UBC Press, Vancouver, 1994.
Return to source of Footnote 17 The televote cost $2.20 per vote cast, or $1.11 per eligible voter.
Return to source of Footnote 18 Scholtz, Christa, Electronic Voting: Preliminary Research and Recommendations; July 23, 1997.