5.1
In recent years there has been an increase in electronically assisted devices and processes designed to make elections easier for voters, faster to count and more accurate. Despite these achievements, the use of technology by the Australian Electoral Commission has not kept up with public expectations nor totally eliminated integrity and security related problems.
5.2
This chapter provides an overview of technological developments, how they have effected Australian federal elections up to now, and what can realistically be expected and introduced for the next election.
5.3
While the Committee does not believe that electronic and/or online voting can securely be used in Australian elections, there are other technological developments that could significantly improve elections, and the voting experience, namely:
widespread rollout of electronic certified lists; and
scanning of ballot papers.
5.4
Notwithstanding concerns about electronic voting discussed below, there may be uses for this technology to assist, in particular, blind and low vision voters.
Electronic voting
5.5
Despite public enthusiasm for electronic voting, there are a number of serious problems with regard to electronic voting – particularly in relation to cost, security and verification of results.
5.6
In particular, as technology becomes more sophisticated, as does the capacity to interfere with it, further raising security concerns about how a ‘hack-proof’ electronic election would be delivered. This would not only undermine the integrity of the process, but also electoral outcomes.
5.7
As part of its review into the 2013 election, this Committee’s predecessor investigated the issue of electronic voting in some detail. The report reviewed the Australian experience as well as trials conducted internationally in countries such as Ireland, Brazil and Estonia. The Committee concluded:
… irrespective of one’s philosophical view about electronic voting, that there can be no widespread introduction of electronic voting in the near term without massive costs and unacceptable security risks.
Any use of technology in association with the electoral process must have the principle of the sanctity of the ballot at its core, including upholding the right to a secret ballot and ensuring transparency in the counting process.
5.8
Despite advances in technology, the same issues regarding security, integrity and cost continue to bedevil further progress on nation-wide electronic voting. As part of this inquiry into the 2016 election, which relied on scanned ballot papers for the count, Dr Vanessa Teague (private capacity) commented:
I have done a lot of research on trying to understand how the advantages of computers could be used while still providing the kind of evidence trail that scrutineers and the public could observe to show us all after the election that it got the answer that the voters actually chose. It is actually really hard to do a good job of scrutinising a computer, because a computer can print up on the screen a very comforting message saying that it has helpfully recorded the vote that you asked it to, but in fact the actual internals of what the computer is doing could be wrong. There could be an accidental configuration error or a software bug, or there could be a deliberate attempt at fraud from either the outside or the inside. So it is a real engineering challenge to design a system that allows verifiable evidence of the right election outcome if the election involves a significant use of computers.
5.9
Mr Antony Green, the ABC’s election analyst, was also not enthusiastic about an immediate move to electronic voting. Mr Green commented in his submission that:
At all times so-called technological determinism should be avoided. Just because a new technology exists does not mean old methods should be abandoned. The use of pencils and papers may seem old fashioned compared to computers, but the old methods come with traditional audit trails and the comfort of physical ballot to ensure the certainty of the result.
Any new technology needs to create new methods of ensuring trust in the process. The problems of the 2016 Australian Census will no doubt raise questions about any shift away from traditional methods of conducting elections.
There must be cost-benefit applied to where the technology is introduced first.
5.10
The Committee concurs with its predecessor’s assessment that as it stands the technology is not sufficiently mature for an election to be conducted through a full scale electronic voting process. However, the Committee remains interested in technological developments which may eventually result in a convenient and secure method of allowing votes to be cast electronically and will continue to consider new technological developments.
5.11
Other than electronic voting, there are several other electronically-assisted voting processes which may be more secure and cost-effective to implement for use in limited circumstances.
Limited electronic voting
5.12
The Australian Information Industry Association (AIIA) notes that there have been examples of eVoting in Australia as far back as 2001. For example, in New South Wales and Victoria, a limited form of electronic voting has been introduced – iVote and vVote respectively. In both cases, the systems are designed to assist people who may find attending a standard voting booth difficult – particularly those individuals who have vision-impairment, a physical impairment or difficulties with the English language.
iVote
5.13
Vision Australia was very positive about the NSW iVote system, recommending to this Committee that a similar system be introduced nationally. Blind Citizens Australia also observed that: ‘the iVote service was very well received by the blind and vision impaired community.’
5.14
Despite the enthusiasm of disability advocates, the iVote system is not adequately secure. Dr Vanessa Teague commented:
… my colleague Alex Halderman and I did some security analysis of the New South Wales iVote internet voting system during the run of that election. We found there was a serious security hole in that particular system and, using the practice version of the server, we showed that it was possible for a malicious party on the internet to use a vulnerability in the system to take over the person's voting session on their web browser, expose how the person intended to vote, and change the vote before it got sent back in to the Electoral Commission.
5.15
The NSW Electoral Commission engaged Mr Roger Wilkins AO to undertake an inquiry on iVote and telephone voting system, in response to the NSW Government’s response to the NSW Parliament’s Joint Standing Committee on Electoral Matters report on the 2015 State election.
5.16
The terms of reference of the report are:
whether the security of the iVote system is appropriate and sufficient;
whether the transparency and provisions for auditing the iVote system are appropriate;
whether adequate opportunity for scrutineering of the iVote system is provided to candidates and political parties; and
what improvements to the iVote system would be appropriate before its use at the 2019 State General Election.
5.17
The report was tabled in late November 2018. It would be of benefit for the AEC to closely examine this report and advise both the Committee and the Special Minister of State about the feasibility of extending this system to federal elections, in line with Recommendation 24.
vVote
5.18
The Victorian Electoral Commission (VEC) was an early adopter of electronic voting, deploying such systems in 2006 and 2010. Changes in electoral legislation to allow for electronic voting had the intent of better accessibility for blind, partially sighted, and motor impaired voters through customised computer voting interfaces. The electoral legislation was amended in 2010 to extend access to voters speaking languages other than English and any voters out of state and overseas, enabling a faster return of votes into the tallying process.
5.19
Although voters were generally satisfied with the usability of the system, the Victorian Auditor-General noted the that vVote has not been widely used:
EAV [electronically assisted voting] has been available on a limited basis in Victoria since the 2006 state election. In the 2014 election, VEC implemented vVote, its modified EAV system for voters with motor impairment, voters with low or no vision, and voters with limited English proficiency, as well as overseas voters at the London early voting centre. vVote was available at 25 early voting centres—including the accessibility super centres—but not on election day. The system was not well utilised, with only 1,121 electronic votes processed. The vast majority of these—87 per cent—were cast overseas in London... VEC [Victorian Electoral Commission] made the decision to pare back the number of voting centres offering EAV between 2010 and 2014.
5.20
Both iVote and vVote proved very useful for certain groups and individuals. There are aspects of these systems which can be used to further develop electronic voting systems in Australia. But their limited utilisation, and the security concerns, mean that they cannot in their current form be considered as templates for a broader national system of comprehensive electronic voting.
Web-loading and e-mail submission
5.21
Web-loading and email submission is used by some electoral authorities so that people in remote or overseas locations can lodge their ballots. While this is not intended as a replacement for all voters in a general election, it allows ballot papers to be transferred electronically for those voters who cannot attend a regular polling place or complete a postal ballot.
5.22
Mr Antony Green observed that both Tasmania and New Zealand have developed methods to permit overseas voters to vote using less expensive methods than internet voting. He argued that e-mail and so-called ‘web loading’ could be considered as options for overseas voting in federal elections.
5.23
In both cases, voters are required to apply for the option of e-mailed or faxed versions of ballot papers. These need to be accompanied by verification documents to match the original applications.
5.24
In New Zealand, the material is then scanned after completion and uploaded to their Electoral Commission’s website, while in Tasmania the material is then e-mailed to the Tasmanian Electoral Commission.
5.25
Mr Green emphasises that these methods mean that a voter’s ballot paper may not be as secret as the usual voting methods. However, both Commissions seek to protect the voters’ privacy.
5.26
The Committee received no expert evidence on whether this would be a feasible option for federal elections, however, it may be worth considering, particularly for overseas voters such as defence personnel on deployment.
Online and Interactive Voice Response (IVR) systems
5.27
In 2017, the Australian Bureau of Statistics (ABS) conducted a national marriage equality survey on behalf of the Australian Government. To provide as inclusive a service as possible, the ABS reported that it utilised online and Interactive Voice Response (IVR) system options. These were developed in partnership with Amazon Web Services (AWS) using contemporary cloud computing technology to securely receive anonymous responses.
5.28
The ABS reported that online and IVR systems were available for:
people in aged care facilities;
people with a disability, injury or illness;
people unable to access mail or living in a remote area with less frequent mail service;
people experiencing homelessness; and
people in an institution with no mail access.
5.29
These groups could request a Secure Access Code (a unique 16 digit code) that could be used to provide an anonymous survey response online or through the IVR system. In total 34 447 eligible Australians responded to the survey through the online (33 889) and the IVR systems (558).
5.30
The survey demonstrated that practical electronic alternatives currently exist which can assist certain demographics lodge their vote. It must also be noted that only a little over 34 000 voters used these alternatives out of the over 16 million eligible voters in Australia. At the time of writing this report, no analysis was available on the security of these services and their appropriateness for use in a general election.
Scanning of ballot papers
5.31
As discussed in Chapter 2, the 2016 federal election saw the Australian Electoral Commission (AEC) scan millions of Senate ballot papers and record voter preferences for those papers electronically for the first time.
5.32
The AEC observed that ‘this was not a small job’ and reported that in just over three months they developed, tested, certified and operationalised a new method to count and distribute Senate preferences. The semi-automated process, using scanning and image recognition technology to capture preferences, was developed with a contractor – Fuji Xerox Document Management Services.
5.33
The AEC reported that Senate ballot papers were scanned to capture an image of everything contained on the ballot paper, except the watermark. Preferences were captured using optical character recognition and verified by a human operator. The AEC reported that over 800 staff scanned and verified preferences for 631 candidates. Counting and distributing preferences required scanning of 14.4 million ballot papers and entry into the count system by a human operator of 101.5 million preferences.
5.34
The AEC reported that all ballot papers were passed to a second human operator for full blind entry of all preferences on the ballot paper and comparison with scanned and verified data. Once verified, a digital record was generated representing the preferences on the ballot paper. Any discrepancy was directed to the AEC for adjudication and resolution.
ANAO investigation and report of AEC procurement
5.35
In January 2018, the Australian National Audit Office (ANAO) published its report on the Australian Electoral Commision’s (AEC) procurement of services for the conduct of the 2016 federal election. The ANAO were critical of a number of issues surrounding the introduction of scanned ballots for the Senate vote. The ANAO concluded:
In delivering the 2016 federal election the AEC established and managed contracts for the transportation of ballot papers and, in a short timeframe, for a Senate scanning system. Insufficient emphasis was given by the AEC to open and effective competition in its procurement processes as a means of demonstrably achieving value for money. Its contract and risk management was also not consistently to an appropriate standard.
The AEC has not demonstrably achieved value for money in its procurement of Senate scanning services. It has not used competitive pressure to drive value nor given due consideration to cost in its procurement decision-making. The AEC sought to encourage competition amongst transport providers but at times struggled to achieve value for money. It would have benefited from additional logistics expertise and transport industry knowledge when establishing and managing transport arrangements.
Most contracts with suppliers contained comprehensive security requirements that appropriately reflected the AEC’s ballot paper handling policy. The AEC was generally satisfied that the requirements were implemented.
The AEC addressed risks to the security and integrity of ballot paper data through the design and testing of the Senate scanning system. The AEC accepted IT security risk above its usual tolerance. Insufficient attention was paid to ensuring the AEC could identify whether the system had been compromised.
The Senate scanning and transport suppliers delivered the services as contracted. The AEC had limited insight into whether its contractual and procedural risk treatments were effective. Going forward, the AEC needs to be better able to verify and demonstrate the integrity of its electoral data.
5.36
The ANAO made four recommendations to improve the AEC’s processes and performance. The AEC agreed to two recommendations in full, and agreed with qualifications to the other two.
5.37
In its defence, the AEC argued that it had achieved a great deal in a short time. Electoral Commissioner Mr Tom Rogers wrote:
The AEC’s Senate scanning solution was developed and implemented in less than 12 weeks, and then operated through the election to deliver the election of 76 Senators to the Australian Parliament. On any reasonably measure, the solution was an impressive accomplishment which functioned as intended. It has subsequently received an innovation award at the 2016 Australian Information Industry Association iAwards ACT.
5.38
The Committee notes the ANAO report, its finding and its recommendations. The 2016 federal election was the first occasion that Senate ballot paper scanning was introduced and only a short time before the election itself. ‘Teething problems’ can be expected and, given the circumstances, the AEC performed some remarkable work. The ANAO report shows, however, that there is room for improvement, and the Committee will follow the progress of the ANAO recommendations as part of its AEC oversight inquiry.
Scanning House of Representatives ballots
5.39
Given the apparently successful outcome of the new Senate ballot paper scanning procedure, it has been suggested that a similar procedure could be introduced to count House of Representatives ballot papers.
5.40
There is logic in this suggestion; however, it would be beneficial if the Senate scanning system was further developed before adopting the system for the House. As the ANAO report and the scrutineer evidence outlined in Chapter 2 indicated, there are still some deficiencies that need to be remedied.
5.41
House of Representative ballots are much smaller and easier to count. The multitude of candidates and the Senate’s proportional voting system mean that Senate papers are large and can be cumbersome. Introducing a scanned counting system for the Senate gives a greater benefit for the cost compared to a similar system for the lower house.
5.42
The Committee’s Third Interim Report briefly reviewed this question and recommended that the Commonwealth Electoral Act 1918 be amended to allow for an electronically assisted counting process. This would permit the AEC to pilot the scanning and electronic counting of House of Representatives ballot papers at the next federal election. The Committee continues to support this recommendation.
5.43
Despite implementing the ANAO recommendations, as noted above, evidence to this inquiry raised vulnerabilities with the Senate scanning process and the lack of ability to scrutineer accurately. More work needs to be undertaken to improve this process prior to implementing electronic counting for the House of Representatives ballot.
Electronic Certified Lists
5.44
The Australian Electoral Commission (AEC) reported that during the 2016 federal election 1 544 Electronic Certified Lists (ECLs) were deployed for issuing ordinary and declaration votes by polling teams and voting centres. This followed an evaluation of an ECL pilot at the 2013 federal election, which used about half the number of ECLs. The AEC reported that the pilot found ECLs were most useful in pre-poll voting centres (PPVCs) and mobile teams.
5.45
According to the AEC, ECLs enabled elector identification and mark-off and real time update of a central copy of the certified list where network connectivity was present. This reduced the risks of polling official error and multiple voting, and enabled more efficient search for electors, including by location.
5.46
Electronic polling place management systems in place in the ACT and the Northern Territory have reduced multiple voting incidents by up to 100 per cent and 80 per cent respectively.
5.47
At selected issuing points, ECLs are able to print House of Representatives ballot papers. According to the AEC, where printing was used it virtually eliminated the risk of providing the incorrect House of Representatives ballot paper to an elector. This is a risk in the ‘superbooths’ and for declaration vote issuing, where votes are issued for multiple divisions. The AEC reported that this also simplified ballot paper reconciliation at the close of polls.
5.48
The current application used in ECLs allows for some monitoring of queue times, ballot paper issue and inventory. The AEC reported that due to resource restrictions, the number of ECLs deployed to polling locations is negligible compared to the total number of polling locations.
5.49
Rollout of ECLs to all polling booths will potentially alleviate many problems that arise at polling booths. For example, as discussed earlier, during the 2016 election, there were 148 370 more Senate ballot papers lodged and accepted than for the House of Representatives. This is largely due to confusion about the voter’s address and which House of Representatives division they are in when applying for a declaration vote. An elector may be given the wrong House ballot paper even though the correct Senate ballot has been issued resulting in, what is called by the AEC, ‘partially admitted declaration votes’.
5.50
This effectively disenfranchises voters and, with the trend towards very close elections, has the potential to affect the outcome in close seats. The AEC is aware of these problems and indicated that:
Although a national rollout of the current Electronic Certified List (ECL) system is not feasible in time for the next federal election within current resource constraints, the AEC is planning an extended deployment of ECLs to all pre-poll voting centres and mobile teams. As a result, the AEC will more than double the number of ECLs for the next federal election.
The current ECL deployment plan for the next federal election using the AEC's current ECL system includes approximately 3,900 ECL devices and 2,300 printers for use in all:
interstate voting centres, and
static super booths (large polling booths that can issue ordinary votes for all divisions within a state/territory).
‘ECL Lite’
5.51
The Australian Electoral Commission (AEC) testified that they are exploring options, including a possible Electronic Certified List (ECL) application for mobile phones. Using this app, declaration-issuing officers at polling booths or perhaps even voters themselves could check which House of Representatives division a voter is enrolled in:
one of the things we are exploring at the moment is: could we provide the ECL look-up in a different way? To come to your point—could people bring their own device?—that is one of the things that we're looking at: can we, using the technology, secure that and make it still usable on a range of devices…
5.52
Although the AEC expressed a concern about ensuring security of the information provided, the AEC was open to providing new services through improved technology to minimise anomalies such as roll divergence.
5.53
Accordingly, the AEC will be seeking to expand its ECL capability through the development of ‘ECL Lite’:
The AEC is developing a new ECL Lite solution which could involve a limited pilot of fewer than 1,000 devices enabling look-up and mark-off capabilities on a range of lower-cost devices. Unlike the existing ECL solution, the searchable certified list will be accessed remotely via a mobile network connection. If implemented in time, ECL Lite will be piloted at the next federal election.
The ECL Lite solution will allow a declaration vote issuing officer to confirm the enrolled address of the voter to ensure the correct House of Representatives ballot paper is issued. This capability could reduce the number of partially admitted declaration votes. For example, at the 2016 federal election, 11.2 per cent of declaration votes (provisional, absent and pre-poll) that were issued without using an ECL were partially admitted, while only 0.19 per cent of declaration votes that were issued using an ECL were partially admitted.
5.54
The Committee is encouraged by the AEC’s recognition of these types of voting anomalies and their innovation through technology in trying to solve them. The Committee, as part of its Third Interim Report on AEC modernisation, already recommended that:
the Australian Electoral Commission extend the deployment of electronic certified lists at the next federal election to ensure all polling places (including all absentee voting points) and mobile teams are equipped with at least one electronic certified list, or as a minimum an electronic roll lookup facility.
5.55
Although this recommendation remains extant, the Committee also encourages the further development of a convenient, accessible and secure mobile phone application though which declaration-issuing at polling booths and perhaps eventually voters themselves can check details, and ensure electors are issued the correct ballot papers.
5.56
The Committee recommends that a national rollout of Electronic Certified Lists and/or ‘ECL Lite’ be fully funded and implemented prior to the 2019 federal election.
Conclusion
5.57
The Committee observes with interest innovation by electoral agencies and software developers with regard to improving the election process. The Committee believes there is a huge potential in making elections faster, more accurate, convenient and efficient while not undermining the security and integrity of the vote.
5.58
Nonetheless, the Committee remains of the view that large scale electronic or internet voting is not achievable at this time, though it encourages future iterations of this Committee to review the question regularly.
5.59
The Committee acknowledges the innovative approach taken by the Australian Electoral Commission and state electoral commissions in designing and trialling their own electronic voting capabilities – such as Electronically Certified Lists and iVote – which can be used to serve niche demographics within the electorate. The Committee is particularly interested in innovation that will assist voters who have physical or mobility impairments, or ensure that declaration ballots are issued with greater accuracy.
5.60
The Committee also acknowledges the AEC’s achievement of establishing a functional Senate ballot paper scanning system in a short time – one which, despite some teething issues, proved itself to be generally effective in the 2016 poll. The Committee encourages the refinement of the Senate ballot counting system and also encourages development and trial of a similar system for House of Representatives ballot papers. At the same time, the Committee is mindful of the Australian National Audit Office report and the steps that need to be taken by the AEC.
5.61
Finally, the Committee notes that the AEC is exploring options for the greater sharing of electoral materials through the Electoral Council of Australia and New Zealand (ECANZ). The Committee strongly encourages the sharing of electronic resources such as ECL tablets to be part of this sharing of materials.