The NSW iVote project – Internet based voting in Australia

By Sam Campbell.

In March 2015 the state of New South Wales, known locally as NSW, in Australia held the largest (well – at the time anyway!) binding government Internet delivered election in the world with over 280,000 votes using their iVote system.  An election channel that was extremely well received by voters (recording an astounding 97 % satisfaction rate), proved an increase in the online voting channel of 500 % and was an historic step forward in the use of innovative technology for democracy.

Australia has seen a number of trials and implementations of electronic voting projects, from those based on attendance based solutions to the trial on a specialised network for the collection of votes from Australian soldiers in the field.

The 2015 iVote project was an extension to the prior election in 2011, the first Internet delivered election in Australia.  So far NSW is the only Australian state that has delivered ballots to remote and disabled voters over the Internet.

Context

The physical manifestation of a voting system is tightly linked to the political and social environment within which that voting system exists. First past the post, proportional representation, compulsory or non-compulsory voting, plural voting and other aspects are all factors reflected in the system that will continue to evolve as the political environment evolves.  Within this environment built on the participation (compulsory or non-compulsory), the counting (FPP or PR), the funding rules and overall election management there sits the question of the actual collection of the votes.

In Australia the electoral system is compulsory, enforced through the fining of eligible voters who fail to vote – think of it as a speeding ticket.  It’s been compulsory in NSW since 1928 which is 26 years after women were granted the vote.  What this shows is a progression over time of increasing the voting franchise to the voters of NSW, in a similar way to other states of Australia and indeed the federal government within a largely comparable timeframe.

A court found in 2008  that the NSW Government should provide a vote in a ‘secure and private manner’ to a blind voter in a case brought by Darren Fittler.  As a consequence of that finding it was determined by the Electoral Commission that the best way to meet the electoral need of the visually impaired voter was to make available an online voting system – and in so deciding the iVote system became a reality.  Behind this decision was the observation that only 10% of visually impaired people in NSW can actually read braille, thus eliminating some of the mechanical options that might otherwise have been available.

This combination of compulsory voting requirements, the penetration of Braille, budget constraints and a healthy dose of what I’d best describe as ‘political will’ created an environment that successfully incubated and fostered the desire to implement an online voting solution.

Risk Management

The running of the 2015 iVote project was an exercise in risk management. Looking at Internet based electronic voting through the prism of risk management allows the electoral commission to look at the project in much the same manner as they look at the more traditional vote collection methods – poll based, postal and so on.  The risk management approach (risk identification through to management plans) provides a framework that can be understood within the business environment of the electoral commission.

To those who work with the electoral process, you might hear them mutter under their breath that there is no difference in the above paragraph to running a paper based election.  They’d be right.  Overlay the specifics of the domain knowledge of the collection of electronic ballots and you’re there.

And this is the meat of an Internet voting project.  The naysayers will say that the risk cannot be eliminated.  With a thorough understanding of the risks, and the mitigation processes for those risks such as would be provided by a secure electronic voting provider such as Scytl, these risks become quantifiable, and then manageable.  Without going into it – no, we’re not just talking about a drop of https with some SHA1 for good measure – Scytl goes a lot further than that as can be seen from our extensive published papers and patent portfolio.

Who could use online voting?

In order to be an eligible voter for the NSW iVote system, a voter had to meet the following criteria:

• The visually impaired
• Those with a disability
• Those who live in remote locations
• Those who were going to be outside NSW on Election Day.

What did NSW get for the investment?

A review of statistics made available at the completion of the project brings to light some of the returns that the state of NSW can measure as a result of the use of the iVote system.

• The iVote system returned approximately 5% of the vote.
• The iVote system was taken up largely by those who were outside the state on polling day (91% of votes taken).
• 97% of users reported being fairly or very satisfied with their use of the iVote system

Data analysis would show that these voters have come largely from two cohorts of voters in prior elections – those who postal voted in the past, and those who failed to vote in the past – therefore increasing the franchise of the overall voting system.

So in high level terms NSW returned the following for the states investment:

• An increased franchise
• A system accepted remarkably well by voters
• A system understandable and acceptable to the stakeholders
• A sustainable cost profile.

To quote the NSW Electoral Commission CIO “You can give away money in a government program and not get satisfaction levels that high” (Technology in Government Conference, Canberra, Australia, 5th August 2015).

iVote does increase Participation

In electoral circles there has long been the view that electronic voting has no impact on participation, however in my view that question has been hard to truly answer in the past – how do you truly measure engagement?  Would people truly vote if they had another voting channel, or are you simply moving voters from channel A to channel B?  It’s hard to collect statistics between elections as the context of each election is always different.  I’d say even more so in areas of non-compulsory voting – how does a statistician with the data remove all the other effects to truly isolate the participation?

As the NSW Electoral Commissioner said at the time “The world has changed since the last century, and peoples time on a Saturday is precious.  iVote in 2011 was successful – in 2015 it went gangbusters!”

According to the NSW report into the conduct of the 2015 state general election, the use of the iVote system has increased participation, with one in ten users reporting that without the system they would not have cast a vote.  This is clear evidence that the system does in fact enfranchise voters in the NSW context.

In closing

This has provided commentary about just a few areas of the iVote project. Not touched on at all is the running of the project, the garnering of information about what works in an interface depending on your expected user base, the activities of activists who will work against an electronic voting project for their own motivations and, well, one of the most vital pieces of the puzzle – how security is actually managed in an online voting system – which as was mentioned earlier, more than just https.  These are all interesting topics to those in the electoral field, and also those in the IT field, but they’ll be left for another day.

Sam Campbell is the Operations Director – Asia Pacific for Scytl.

Do you think the UK should introduce an online voting option for elections?  Take a moment to sign and share our e-petition.