Australasian Science Magazine

A single-use swipe card gives Canberrans access to the electronic ballot for the ACT elections. Credit: Elections ACT

By Stephen Luntz

Electronic voting has been in place for more than a decade, so why are we still using pencil and paper for this year’s federal election?

Despite all the technological advances of recent years, this year’s federal election will be conducted in a manner barely changed from a century before.

While far more people now vote at prepolling centres, the act of voting is still done with pencil and paper – something almost archaic to most of our usual practices. Many may wonder whether the electoral commission is stuck in some pre-computing Dark Age, but there are good reasons why voting continues to be done in the time-honoured manner while the rest of our lives shifts online.

Ancient Athenians conducted some elections by dropping black or white pebbles in a bucket. The system allowed both secrecy and for the vote to be checked and recounted. It took two millennia for the system to be improved.

The idea for the modern system of secret ballots began in England, but was first implemented in Australia with the Tasmanian electoral act of 1856, quickly followed by Victoria and South Australia. The initiative spread to other democracies, and for a time was known worldwide as the “Australian” or “Victorian ballot”.

While voters might promise their vote to the local landowner or bully, in the privacy of the polling booth their conscience was their own. By making rigging hard and cutting down on corruption and intimidation, the Australian ballot was a key feature of Australia’s progress.

Advocates of the introduction of paperless voting have the challenge of finding ways to maintain the security of these gains.

Electronic Voting in the ACT

Australia’s first large example of electronic voting was for the Australian Capital Territory Legislative Assembly. Since 2001, pre-poll centres and sometimes a limited number of polling booths have provided voters with the option of casting their ballot by computer.

In this system, voters are marked off in the usual manner and offered the choice of a paper ballot or barcode. Those that choose the barcode enter a private booth with a computer and swipe their barcode to begin. They then vote using an online ballot before being offered a chance to undo their preferences and swiping the barcode again to confirm their vote. Computers are connected to a central hub over a secure system, allowing for safe storage of the vote.

Rohan Spence of the ACT Electoral Commission says: “There are many advantages to using electronic voting at an ACT Legislative Assembly election”. For example, while deliberate informal votes are allowed, the system the ACT uses prevents accidental informal votes through the incorrect use of ticks and crosses, or the repetition of numbers.

Spence also notes that electronic voting makes it easier to provide instructions in languages other than English. However, probably the most important advantage is the ability to calculate preferences immediately for those votes cast electronically. The ACT’s complex form of proportional representation means that election outcomes can be hard to predict from primary votes alone, and the ability to see where the voters directed their preferences can make results clear on election night, rather than requiring weeks of suspense.

When introduced this was widely seen as a step to the future. Twelve years later, however, it has not been adopted elsewhere and indeed is now offered only at voting centres that are open prior to polling day, rather than at a limited number of additional locations on the day as was the case at the first election.

As more and more voters cast their vote at prepolling centres, the proportion of ACT ballots cast electronically has risen. “In 2012 there were 59,200 electronic votes cast, which equates to 82.4% of votes cast at electronic polling centres,” says Spence. However, he says there is “considerable logistical effort in setting up and configuring” electronic voting for a polling centre for just one day.

While Spence anticipates that more prepolling centres may be established at future elections, he says electronic voting is “unlikely to be expanded into ordinary election day polling places at future ACT elections”.

If the costs aren’t viable for Canberra’s established system to run electronic voting at most booths, other jurisdictions are unlikely to rush to adopt the same system. Tasmania aside, the rest of Australia does not have the type of proportional voting that makes preference allocations so hard to predict on polling night.

Why Not Online?

Online voting could potentially be cheap, easy for the disabled and those living remotely, and a boon for anyone too lazy to leave the house or with an aversion to long queues. While it is increasingly common for voluntary associations to use it, state and federal governments have not been so keen. Concerns have been raised about people being bullied by family members as to how to vote, or that there could be a revival of bribery since anyone voting online could prove how they voted.

The same problems exist with universal postal voting, as is now used by all local councils in Tasmania and South Australia, and many in Victoria. The latest Melbourne City Council election saw a police raid on a candidate over 442 suspect voter applications.

While no one has been charged over the apparently fictitious registrations, remote voting lends itself to fraud far more easily than elections where attendance at a polling place is required. Online voting would make the process easier still. Evidence that 30 people were registered in a single one-bedroom flat for the Melbourne City Council election has led to a review of the Melbourne City Council’s voting system, and may have wider ramifications.

Logically, the larger the prize the more likely it is that someone will try to rig the election. It is unsurprising that the first evidence of large-scale voter fraud in an Australian election for many decades came in a universal postal election where the Lord Mayor of Melbourne was being chosen, rather than in a suburban council. A federal election without the requirement for most voters to turn up in person would be a more attractive target still.

Moreover, any federal election conducted online would be a magnet for attempts to hack the voting system. Even if Australians did not wake up to discover Donald Duck declared Prime Minister, a denial of service attack that sent the system down for a substantial period of time could provoke controversy over whether the election was legitimate. Even a result out of step with opinion polls might have its validity questioned.

More subtle problems would also need to avoided. In 2004 the University of Melbourne held its student elections online. One team set up a fake website with a URL similar to the real site. They encouraged voters to vote at the fake site, using their unique password, and then harvested the passwords to cast votes at the real site.

Even if the vote was conducted validly, the secrecy of the ballot needs to be maintained. Verifying an individual’s eligibility to vote, but de-identifying the vote itself so that no log can ever be released that reveals how individuals have voted, is not a trivial problem.

The NSW Experience

In 2011 New South Wales allowed voters who were blind, disabled or would be outside the state on polling day to register for an “iVote”. Roughly 50,000, or more than 1% of the electorate, chose to do so. The system has become even more popular at subsequent by-elections, with 5.8% of votes in the Sydney by-election cast in this manner. The state government is proposing to relax the registration rules for by-elections, allowing those who are out of the electorate, as well as out of the state, to register.

The NSW Electoral Commission’s website lists multiple protections adopted to ensure iVote’s security. These include “external expert scrutiny and extensive testing… [and] sophisticated encryption and automated processes to decrypt and print the votes in a way that ensures each elector’s vote is secret”. Multiple systems in different locations are maintained to protect against power or equipment failures.

Although iVote was initially designed to making voting easier and more private for the disabled, 92% of users were outside the state on election day, according to an assessment report by Allen’s Consulting. As a result, the system had four times as many users as anticipated. Facilitating interstate and overseas voters is normally an expensive part of election administration, and iVote was found to be a cost-effective way of running this part of the election. Feedback from users was generally positive. Strangely, however, it largely failed in its initial objective, with fewer vision-impaired people taking it up than expected.

Overseas Experience

Many countries have explored electronic voting, but most have shied away. Ireland spent €52 million on electronic voting machines, only to scrap them after security testing.

Most successful examples of electronic voting have involved machines at polling places, as in the ACT example, rather than voting online, and usually only as an option.

Nevertheless, Brazil has made electronic voting machines universal for its elections, except in cases of machine failure. However, the security of the system is backed up by paper records, with the voting machines printing a paper ballot that can be checked in the event of a challenge to the outcome of the electronic count. This system combines reliability with speed, since in most cases the computers can produce the result of the election within 2 hours of the polls closing. However, the system is expensive and does little to make voting easier for the public.

Electronic voting was widely used in The Netherlands until 2006, when a group of hackers demonstrated that they had the capacity to access the machines and determine how an individual had voted. This led to a return to paper ballots. The same group of hackers demonstrated that the Dutch internet voting system was also vulnerable, derailing plans to increase online voting.

Worried about falling voter turnout, the United Kingdom ran trials of a range of alternative voting systems at selected local councils in the early years of this century. Although few technical problems were recorded, voting rates actually dropped and the trials have largely been abandoned.

In Arizona the US Democratic Party experimented with universal internet voting for the Primary that saw Al Gore selected as the party’s presidential candidate in 2000. The trial received much publicity but has not been repeated despite the growth of internet access in the intervening years.

The one widespread use of internet voting has been in Estonia, where 15% of votes in the 2009 European Parliament election were cast online. However, this is facilitated by Estonia’s national identity card, which fits into a special port in home computers and provides a proof of identity that would not be available elsewhere.

A Role for Electronics

Even if widespread electronic voting seems to be off the agenda, more limited applications are possible.

The use of electronic electoral rolls could improve, rather than hinder, ballot security. Fears are often expressed that voters can visit several polling centres and be issued votes at each. However, electoral rolls at polling booths are already compared after each election, and there is little evidence that duplicate voting occurs.

The Victorian Electoral Commission has trialled the use of electronic rolls at recent by-elections. Once a voter is crossed off the roll at one polling place, the information is transmitted to all others, reducing costs as well as preventing fraud. However, during the 2010 local government elections the electronic roll crashed, forcing polling stations to return to paper rolls. Insufficient staff had been provided at some polling centres to manage the more time-intensive paper rolls, leading to queues up to 2 hours long.

In 2007 the Australian Electoral Commission ran a trial of electronic voting for the vision impaired and for defence force personnel located overseas. Some voters appreciated being able to cast a vote without depending on others for the first time. Nevertheless, uptake was low, leaving costs per user high.

A Joint Standing Committee on Electoral Matters report found: “The combined costs of the trials was over $4 million, with an average cost per vote cast of $2,597 for the trial of electronically assisted voting for blind and low vision electors and $1,159 for the remote electronic voting trial for selected defence force personnel serving overseas. This compares to an average cost per elector at the 2007 election of $8.36.”

The trial has been discontinued and at this point there are no plans to revive it.