Joint Standing Committee on Electoral Matters: Conduct of the 2013 federal election and related matters
Joint Standing Committee on Electoral Matters - 29/07/2014 - Conduct of the 2013 federal election and matters related thereto
GREEN, Mr Phillip, Electoral Commissioner, ACT Electoral Commission
Senator RHIANNON: Regarding the counting, in your opening remarks I think you said that you use open-source software.
Mr Green : Yes.
Senator RHIANNON: Could you run us through why you made that decision, what form it takes and whether you had to further develop it?
Mr Green : Way back in 2000 we went out to tender for people to put forward proposals for an electronic voting system for the ACT. We had not at that point decided to make it open source, but one of the tenders put to us a proposal that it should be open source. We looked at the benefits and the risks of using open-source software but decided that it was more important from a transparency point of view that the software should be open source and should be able to be accessed by anyone because we thought that it was important that the system we had was not a black-box solution.
One of the ways in which we feel we can justify not having a paper trail is that we have an open-source electronic trail where people can look at the source code that is used to create the various disks that we use in the system. If it ever came to a court situation, we would be able to go to the court and say, 'Here is the software that was used to create the various disks that were used to establish the computer system; here are the disks themselves; here are the servers that were used at the polling places,' so that the whole thing would be open to scrutiny. So, while you do not have a paper trail to explore to work out from start to finish what has happened in the election, you have an electronic trail where people with the right skills are able to go through and determine that, effectively, what has gone in is what has gone out. That is really the thinking behind why we went open source.
Senator RHIANNON: So we all understand the detail, because I do not have all the technical terms at my fingertips, you are saying that, in going to open source, the decision was made so it was fully open and every stage could be understood. Was that the basis of your decision?
Mr Green : Yes.
Senator RHIANNON: Does that mean down to the source code and the way the program works?
Mr Green : Yes. My understanding, and I am not an IT specialist, is that the information on the website is the code that customises the standard Linux operating system that then creates our system. So, if you are able to plug the code on our website into the standard Linux operating system and build the system using standard operating systems, you will end up with our system. It is not like putting the entire system in one finished product on the web; it is giving you all the building blocks that you would need to create our system.
Senator RHIANNON: When you say 'on the web', do you mean it is publicly available? Somebody could go there and replicate the system and do the count themselves—is that what you mean?
Mr Green : If people had all the right computer skills to do it. It would not be a straightforward thing, but it is something that is possible. There are people at the ANU who have done that and, I think, people in the Labor Party have also looked at our code.
Mr Green : People in the Labor Party in the ACT, as I understand it, have looked at our code; and people at the ANU found some bugs in our code, which we fixed, because we made it open source.
Senator RHIANNON: You have made that decision and so it is open and available to the public. When you made that decision, at the beginning when you decided to go to electronic counting I imagine you looked at other systems around the world. When you were looking at other systems, did you come across many other systems that you were impressed with that used open-source software?
Mr Green : We did not do an exhaustive search of what was available at the time. There really were not a lot of systems out there that used the preferential type of voting system that we have here. Most of the electronic voting systems I am aware of overseas are first-past-the-post.
CHAIR: First-past-the-post, and then the US—
Mr Green : That is right. I was most familiar with the experience of the US and that was not a good example to follow. The fact that most, if not all, of the US systems that I was familiar with were black-box solutions where they were not open source was, to my mind, a reason to go down the open-source route.
Senator RHIANNON: You saw that as a risk?
Mr Green : Yes, I saw the black-box solution as being a risk and that the open source ameliorated that risk.
Senator RHIANNON: Did you come across any systems overseas using open-source software to do their count?
Mr Green : Not single transferable vote counts like ours, no.
Senator RHIANNON: So what you have done might be quite groundbreaking.
Mr Green : I am not aware of anyone doing what we did in the way we have done it before we did it, no.
Mr PASIN: Sorry, Mr Green, I came in halfway through your evidence, so I just need to clarify some things before I ask a question. The electronic voting is available for polling places, pre-poll and also on polling day in the ACT—is that right?
Mr Green : For the last two elections we have only had it available at our prepoll centres—at the last election there were six prepoll centres—but they were also open on election day as polling places.
Mr PASIN: Given that, in answer to the chair's line of questioning, you indicated that part of the reason you have not encountered some of the reasons that you might is because of the small number. Can you envisage problems if your system were rolled out widely, bearing in mind that my electorate of Lyne has 104 polling places?
Mr Green : There are quite a few issues to scaling up an electronic voting system like ours because our system is based on hardware that is now fairly old fashioned. It is standard PC equipment joined together by wires. It takes technicians to put it together. When you only have four, five or six places to put together, that is something that is fairly easy to achieve. If you are doing it in hundreds of polling places then you need hundreds of teams of people to go out there and do it, so that would be quite a complex issue.
Mr PASIN: On the question of identifying people who have voted: you indicated that here in the ACT it is done electronically. Is that across all polling places?
Mr Green : Yes.
Mr PASIN: But it requires internet connectivity?
Mr Green : We use the 3G wireless telephone network.
Mr PASIN: The system could not operate, could it, in areas that do not enjoy internet connectivity?
Mr Green : No, and even in the ACT we have found some polling places were not getting a good enough signal to actually get a connection happening.
Mr PASIN: I can certainly think of dozens of polling places in my electorate where you would not get a signal.
CHAIR: Senator Rhiannon.
Senator RHIANNON: I am now thinking of the Senate voting system. Would there be any usefulness in the system you are using that could apply to the Senate counting system?
Mr Green : On the counting side of things I think the Senate experience at the last election would get a lot more benefit out of our scanning system than it would out of our electronic voting system. I think our scanning system would go a long way towards addressing some of the issues that were apparent at the last Senate election in terms of getting an accurate count of handwritten ballot papers. It is quite obvious that a scanning system that is using computers to verify that the preferences on ballot papers have been correctly recorded is far superior to a hand count of ballot papers even when looking at just single first preference above the line. A scanning system is going to give you a much more accurate count than a hand count will.
Mr GRAY: What do you base that on? It sounds counterintuitive.
Mr Green : I am saying a scanning system will give you a much more accurate count than a hand count will every time. If you look at the recount figures that are available on the AEC website, which simply lists those polling place total numbers that were counted in the first count and compared it to the total number of ballot papers counted in the second count, you see they mad miscounts in every division in Western Australia and miscounts in more than half of the polling places, and we are just talking counting first preference votes above the line—single-ticket votes. Hand counting and hand sorting using humans alone is an error-prone thing. This is what we found in 1998. If you look at the result of the recount in Western Australia, you can see that hand counting even a single first preference on a ballot paper is something that human beings are not very good at. Computers are very good at it.
Senator RHIANNON: How easily adaptable is your system? You have worked with the AEC, so it sounds like you would have good knowledge of their counting system. How adaptable is your system to what we have in the Senate now? From what I am hearing, because it is based on scanning, it could be fairly easily adaptable. Is that a reasonable assumption to start with?
Mr Green : I think it certainly is adaptable. Whether it is easily adaptable will depend on the availability of scanning machines. They are fairly rare machines, so you would need to make sure you had enough of them to do a Senate election. The other really big issue would be the size of the ballot paper. If it is a very large ballot paper, it might not go through the machine.
CHAIR: And it is very large at the moment. The maximum printable width—as you know, we have had another report on that. To enlarge on Senator Rhiannon's question: the other point you made earlier is the transport and security of the ballots before they are scanned.
Mr Green : Every electoral event everywhere in the world has that issue.
CHAIR: But the point you made earlier is that you have perhaps the lowest risk on that front just because of the unique nature of the electorate here in the ACT compared to electorates like Mr Pasin's, which is how large?
Mr PASIN: Sixty-four thousand square kilometres.
CHAIR: And how many polling booths?
Mr PASIN: One hundred and four.
Mr Green : A thought that I presented at the Senate voting workshop a few months ago did not get a lot of support, but I will throw it on the record anyway. Over 95 per cent of people voting for the Senate vote above the line. An option that would certainly make scanning and counting them an awful lot easier is to have two different Senate ballot papers. Have an above the line ballot paper and a below the line ballot paper and offer people the choice. That would make your scanning process much easier. It would also make the voters' job a lot easier. It would enable you to put the ballot papers in a column like the House of Representatives ballot paper so that it would not be quite as daunting as the current sideways ballot paper.
CHAIR: How would that work in practice? They would have to make a choice of whether they were going to vote above or below the line—
Senator RHIANNON: As they came in.
CHAIR: In most cases, they looked at the candidates below the line and made that judgement.
Mr Green : Yes, you would have to have a process where people were offered the choice.