• IT SUPPORT OFFICER (MATERNITY...
• Head of Business Intelligence -...
• Senior Project Manager - Multi...
• Project Manager Opportunities
• Chief Information Officer

Electronic voting and open source

Mike Bowern, Information Age

14/12/2005 09:37:06

Australia's electoral systems have been in the news recently. In October, Senator Abetz, the Special Minister of State, suggested, amongst other things, that we should amend the rules about voter identification, voter registration and the disclosure of political donations, and introduce some trials of electronic voting.

At about the same time, it was suggested by another senator that we should abolish compulsory voting. This article will consider some aspects of electronic voting systems and the need for trust in them.

Australians should be proud, but not complacent, of their electoral systems, covering the three levels of government: federal, states/territories, and local. These systems have evolved, starting with some world-first initiatives by some of the colonies in the mid 19th century. Significant reforms such as the secret ballot, adult male franchise, and payment of elected representatives from the public purse, were introduced in this period. After federation we continued this development, for example with votes for women in 1902, electoral administration by independent authorities, preferential voting, and compulsory voting from 1924.

Another step in this evolution was in 2001, when the ACT conducted a trial of electronic voting in 10 per cent of the polling places. In these places voters had the option to use a system called eVACS to cast their vote for the Legislative Assembly instead of using a paper ballot. A description by Dr Clive Boughton of the development of this system appeared in Information Age in December 2003; and a full report of the review of the Electronic Voting and Counting System can be found on the ACT Electoral Commission's Web site (www.elections.act.gov.au).

Many people are aware of the problems that confronted the USA elections in November 2000, and they often make a joke about "hanging chads" when these problems are mentioned. This joking may be from a sense of superiority, because these types of problems are unlikely to happen in Australia, partly because we do not use punched cards, but more importantly because our election systems are very robust and reliable.

However, these problems in the USA were more than just with outdated technology, since there were also issues with electronic voting machines and with the conduct and administration of the elections. Actually the punch card problems were more to do with a lack of maintenance than with the age of the equipment.

There was some disquiet about the use of electronic voting systems, especially direct recording equipment (DRE), which typically do not provide a paper record of each vote cast. Some people believe that some form of hard copy of the vote is necessary to enable the voter to verify his/her selection before the vote is committed to the system. The voter does not keep this paper record, but it is stored for auditing purposes and to support a recount if an election result is disputed.

During 2003 and 2004 there were a large number of protest activities by concerned citizens, and several events where academics discussed solutions to these election problems. One of these events of particular note was a symposium held at the Kennedy School of Government, Harvard, in June 2004, which produced a report called "Electronic Voting Best Practices".

This report identifies a number of required practices which address problems associated with electronic voting and other aspects of the US electoral systems. These other recommended practices include: • the establishment of independent authorities for election administration; • transparent processes throughout, especially the acquisition process for voting equipment; • trained and motivated poll workers; • well-defined election auditing processes; and • the use of a paper record for vote verification and audit purposes.

The report also recommended that the underlying technology and software of the voting systems should be in the public domain, or at least available for inspection by the security community.

If the practices identified in the Harvard report are compared with what happens in Australia, we come out very well. We have had independent electoral officials for about 100 years, our polling place officials and support staff are well trained, and our electoral processes are well publicised and transparent.

Also, I believe that compulsory voting benefits our electoral processes. If the law requires you to vote, then the electoral system must make it easy for you to do so. Therefore we have pre-poll periods, postal voting, mobile voting in some hospitals, all polling stations stay open for the stipulated time, and there are always enough ballot papers to go round.

On these last points, there are reports that in the USA 2004 election, more than several polling places opened late and/or closed early, and some ran out of ballot papers.

Some benefits of electronic voting systems Neil Postman, Professor of Communications at New York University, and now sadly deceased, used to pose the following question when trying to assess the usefulness of a piece of technology: "What is the problem to which this is the solution?" Perhaps this could be a good question to ask about a digital camera built into a mobile phone.

What is the problem to which electronic voting systems are the solution?

The USA has been using some form of automation to support voting for over 50 years, including mechanical and electromechanical devices using punched cards; optical mark-reading technology; and now electronic voting machines. The main problems that these technologies address are the need to capture and store the votes of a large voting population, and count them to produce the election result in a reasonable time.

Australia, with its relatively small population, has never seen the need to use mechanical or electronic technology for its voting systems, although there is now widespread use of ICT for electoral roll maintenance and election administration. Another reason for not using this "old" technology is that since 1918 Australia has used preferential voting rather than first-past-the-post, and the preferential method is more difficult to mechanise, for example by using punched card or optical mark recognition technology. This technology would be more difficult for voters to use, and more prone to error than a paper ballot.

So, what was the problem in the ACT to which electronic voting was the solution?

The ACT has a complex electoral system using the Hare-Clark method of proportional representation and preferential voting. This system is prone to inaccuracies when manual counting is used. Also, a full manual count is expensive because it takes about 10 days for several people to conduct and scrutinise, and it is doubly expensive if a close result requires a recount. The recount is likely to give a different result, but still not necessarily the correct one. These problems of inaccuracy and expense were solved by the electronic voting and counting system.

The electronic system includes components for voting, counting and data entry. The data entry component is used to enter and verify batches of ballot papers, to provide electronic records of paper votes.

The combination of the data entry and electronic vote counting components ensures the accuracy of the count, and reduces the time, and hence the cost, to produce a final election result. If there is a close result from electronic counting it is pointless to conduct a manual count since the result from that count will probably be wrong.

A number of additional benefits are gained through the use of this electronic system. It provides a flexible voter interface, with instructions in English and 11 other languages, and an audio module for sight-impaired and those with poor reading skills. In the ACT in 2001, Australian blind voters were able to cast a vote in secret for the first time. In addition, the number of informal votes is reduced through the use of the electronic system, since informal voting cannot happen accidentally, as it can with a paper ballot.

In the event of a casual vacancy between elections, for example through the death or resignation of a member, the ballot can be recounted electronically with the name of the candidate who caused the vacancy omitted, giving a quicker and accurate result than manual counting, at virtually no cost, compared with the cost of a manual recount.

The use of a paper record for vote verification by the voter was mentioned earlier, and while there is a push for this in the USA, it was not considered necessary for the ACT system. Surely the purpose of an electronic system is to remove, or reduce the need for paper records.

The matter of trust in electronic voting, and open source software

From the Symposium report and other public evidence, it seems that there is little trust by the voters of the USA in their voting processes and technology, and the debate on the use of paper verification is one indication of this mistrust.

In general, Australians seem to trust their current paper-based electoral systems and processes, and one could imagine the outcry if a polling place opened late, closed early, or ran out of ballot papers.

Likewise the voters in the ACT appear to trust eVACS, and there are a number of reasons to support this trust.

• The ACT Electoral Commission followed an established process with probity for the selection of the developer, and provided regular reports, mainly through the press, on the progress of the implementation of the system. • The request for tender gave a reasonably clear picture of what was required; an Internet based solution was excluded, and all software was to be audited by an independent authority. • A Reference Group was formed with representatives from political parties, members of the Assembly, the blind community and the Proportional Representation Society. This Group was used for advice and consultation, during the development of the system. • The software development team worked closely with the ACT Electoral Commission staff, providing detailed specifications and plans. • All software, the application modules and the Debian Linux operating system, was audited by an independent group, BMM International, in Melbourne. • The source code of the application was published under the GNU GPL.

The source code used in the 2001 and 2004 Assembly elections is freely available on the ACT Electoral Commission's Web site, and this availability may be a factor in engendering trust in the system. Certainly, in 2001, some political parties used their own software specialists to conduct independent reviews of the code.

Making the source code available to the public may engender trust in the product, even though the vast majority of people would not be competent, or could be bothered, to inspect the code. However, the fact that the developer is prepared to be open with its product may be enough to engender trust for many of these people.

However, publishing the source code of the system is only one part of the process of generating trust in it. Other aspects include gaining trust in the people who developed the system, and the processes that they used. This has certainly been the case in the ACT.

Of course, publishing the source code of a product can reduce the business opportunities of the original developers, and the owners of the intellectual property, as another group can take the code under the GPL and modify it for another legislature's election. This may have already happened in the USA where the Open Vote Foundation is purported to be modifying eVACS, to add a paper copy of the ballot for verification purposes, in California.

For this reason the developers of the ACT system have decided on a limited open source model for the next versions of eVACS, which will provide sufficient openness for the code to be inspected and verified by interested parties and competent authorities, but still protect their commercial interests.

The test will be whether the voters who use the later versions will have the same trust in the software. There should be no reason why not, because most voters have not taken the opportunity to inspect the current versions of the code, although it is available for them to do so.

The next steps An article in Computerworld magazine of August 17, 2005 covers the news that the Australian Electoral Commission (AEC) is considering the possibilities of using open source software for its next generation of election applications, including the possibility of remote voting. This could be the next step in the evolution of Australian democracy.

Australia's electoral systems are governed by a number of principles to ensure free and fair elections; namely transparency, accountability, security, accuracy, equity, professionalism, secrecy and timeliness. The ACT electronic system was established using these principles and, no doubt, the AEC will follow them in the development of its new systems.

These principles need not only apply to election systems. They are appropriate to all systems which require public trust. A method of providing access to the source code of these systems would help to meet some of these principles.

There is a growing interest in the use of open source software within government agencies, at present mainly in the areas of operating systems and system software. Perhaps in the future, other government applications, for example in taxation, payment of benefits and identity management applications, could be open to some method of public scrutiny to provide further levels of public trust in these applications.

Conclusion It has been predicted by some people that, with the advent of the Internet and other aspects of ICT, there will be much greater use of this technology in our democratic processes, especially voting, regardless of the model of democracy used. Citizens could be using electronic voting systems to select a representative, as is done now.

With a more direct form of democracy, citizens could have some participation in decision making, and perhaps, even vote for some legislation or regulations. Thus the voting system will be critical to the success of any democratic model, and therefore its design and implementation methods must ensure that it is reliable, usable and trusted by all citizens.

Open systems and their development, as opposed to proprietary systems, can help to engender this trust. And in a really open democracy these ideas could apply to other government applications.

Mike Bowern MACS is a PhD candidate at the Centre for Applied Philosophy and Public Ethics, Charles Sturt University. He was previously an employee of Software Improvements, and was part of the eVACS development team in 2001.