Wednesday, 19 March 2008

E-voting: should we use it?

There are multiple methods of voting and, for many years, a paper-based method has been preferred (in the UK, the United States and many other countries). However, in recent years there has been movements towards electronic methods. Several people have pointed out flaws, but there are also some great benefits. So, should we change to e-voting and what version of it should we use?

Governmental stances
The Institute for Public Policy Research published a background paper called E-voting: Policy and Practice and it revealed that the UK government has plans to implement an e-voting system as a way of increasing voter turnout. In the government paper In the Service of Democracy, there were four things listed that could help to achieve their goal:
  • Online electoral register
  • Online registration and online applications for postal votes
  • Online and text voting
  • Electronic counting and collating of election results
The United States have had e-voting systems for a number of years. In March 2002, California approved the Voting Modernization Bond Act, which allowed the purchase of modern electronic voting systems to replace their existing punch-card method.

The following shows the state's committment to this form of voting:
"In December 2003 California Secretary of State Kevin Shelley released My Vote Counts: California's Plan for Voting in the 21st Century, which outlines California's plan for complying with the Help America Vote Act (HAVA). The state expects to receive over $100 million in HAVA funds. In November 2003 the Secretary of State issued a position paper on the deployment of touch-screen voting systems in California."
The Electoral Reform Society disapprove of most of the current state of e-voting in their policy document that can be found here. However, one thing they do approve of is electronic counting of paper ballots. They feel it speeds up the whole process and if it failed, you could always do a manual count as there is a paper-baseed element to it. The IPPR document mentioned earlier also details the benefits of e-counting and goes on to say that "In India the electronic system allowed the results to be announced a matter of hours after the polls closed".

I'm glad that there is approval for electronic counting and I can understand why some people would want a paper backup. However, there really is no need for paper providing the technology is implemented properly. For example, you could have a voting machine using RAID 1, which means that if the primary disk fails, you still have the information on the second disk and you could even remove it and do the counting on another system. If you have to use paper ballots, you could always do multiple electronic counts (possibly on more than one machine) to ensure accuracy. That would reduce the amount of staff/volunteers required and therefor reduce costs.

Machine voting
The following is from the Electoral Reform Society's policy:
"To minimise the risk of fraud, voting machines should produce voter verifiable audit trails. Rather than the voter completing a ballot paper, the machine should produce a ballot paper which the voter verifies and then puts in a ballot box. Should there be a dispute over the result, the paper ballots should be regarded as the definitive votes rather than those recorded on the machines.

Additionally, there should be safeguards equivalent to those described for e-counting."
I get the impression that they would be happy happier if machines weren't used as their suggestion still goes through the same amount of paper as a non-electronic system, therefore reducing the machine to 'an extra hurdle', which could potentially slow things down.
"Following the March 2004 primary election, the performance of Diebold touch-screen systems used in some California counties came under increased scrutiny. In public hearings conducted by the Secretary of State's Voting Systems and Procedures Panel, it was confirmed not only that uncertified versions of Diebold software had been used in some counties, but that some of the software had been inadequately tested and had performed poorly, resulting in lost and miscast votes"
If you read the quote above, you can see why some people would stop trusting machine voting. However, that situation wasn't totally the fault of the machines. It was the counties at fault for not implementing approved systems.

Remote voting
I can understand why the ERS don't approve of this as networks can be hacked and if you have unsupervised locations, there's the possibility of coercion. Despite this, you could still have polling stations with electronic voting machines until the security for remote voting has been suitably improved.

In all the articles and research about e-voting, the biggest problem is security (especially in the case of remote voting). The IPPR document states that
  • ID cards and/or passwords could be stolen
  • If passwords are to be used, they would need to be short so they can be remembered, but that makes them more vulnerable
  • Biometrics could be used, but there would be a huge cost (the UK government estimates £31bn)
  • Viruses, firewall holes and network hacking
  • Voting programs are made by commercial sources. In the US there were calls to make the code 'open source' to ensure transparency, but doing that would mean voting systems could be hacked more easily
The last two problems could instantly be solved by not having remote voting until security has improved. You could just have unnetworked voting terminals and put together the totals at the end of the voting period. With biometrics, there is a long-term benefit, so the high cost might be worth it. Biometric cards would definitely be better than standard ID cards.

So, how would you improve security so that remote voting could be trusted? Well, you could use strong encryption on the database where the votes are kept. You could also use SQL stored procedures for website logins. This has been proved to protect against things such as SQL injection. There's also RAID, mirrored servers and making sure the server is in a physically secure location. Some would say that encryption can be weak, but there are also extremely strong varieties.

Paper-based systems
Dr. Rebecca Mercuri is a noted expert in this field and was involved with the decision to have a hand recount of votes in Florida in the 2000 US Presidential election. She strongly opposes any 100% electronic method (so she'll probably not be happy with the fact that 23 US states don't require paper records of votes). In this article, she mentions the problems in California. What Dr. Mercuri fails to realise is that is was at least partly the fault individual counties for not using approved versions of the Diebold voting system. She also doesn't consider the fact that a lot of security problems are caused by the machines being networked (they don't have to be). E-voting speeds up the counting process and can help people with disabilities, so there is benefits.

Disabled people
According to the IPPR background paper, privacy is increased for disabled people (this is because they can use the same systems instead of going to a separate location). The height of the machines could also be increased or decreased for those with back problems (or for people in wheelchairs). You could also have audio versions of the ballot for those who are blind. E-voting can therefore make democracy more inclusive.

In Britain there were several trials (15 in total) and the most notable ones were in Swindon and Sheffield. In both cases the voter turnout increased. In Swindon, 61% of voters in a survey felt that e-voting was better and 94% stated that they would use e-voting again in a general election. In 2002 (the Swindon trial), turnout was as high as 31.2%. This may seem low, but it's still a significant increase compared to previous years (for further details of the trials, see the E-voting: Policy and Practice document).

Usage in the student movement
Many student unions across the country have recently started to use e-voting and most seem to include remote voting in their implementation because it means people don't necessarily have to go to the campus just to vote (they may not necessarily have lectures/seminars/labs on that day). At Hull University Union, the first year of e-voting had had 1718 voters, which was a 25% increase on the 06/07 total. There has been a lot of controversy with remote voting though. The University of Essex's student union had to change the result of their presidential election because there was electoral misconduct and an unusually large number of votes coming from certain IP addresses. This could have meant that people were taking others over to a particular machine and influencing the way they vote. Coercion might have happened, but cancelling all the votes from those IP addresses could mean that some perfectly legitimate votes were made useless. They should have got the usernames and investigated those people instead.

There are (currently) a number of security issues with e-voting and many of those are linked to remote voting. This is unfortunate because remote voting allows greater flexibility. However, there are ways to improve security. E-counting and machine voting definitely have benefits and there is no reason why they cannot be used straight away (providing approved systems are implemented).

So, what do you think?

Technorati tags: E-voting, Technology


Edinburgh said...

Your quotations from the USA are extremely selective. I think you will find there are many US initiatives to revert to "paper and pencil" voting, and indeed, a number of relevant recent court judgements.