Ill-informed "experts" are not helping the debate on e-voting, write Annrai O'Toole and Brendan Tangney
In all the talk about e-voting there has been precious little rational, data-driven analysis.
Computer-based information systems are used in every facet of modern life. Why? Because they provide more accurate and cost-effective methods of information processing. This central point seems to have been lost.
We're not particularly passionate about electronic-voting and not entirely sure why the Government wants to introduce it. We have no commercial or political interest in the proposed system. But we have been exasperated by the debate and three issues issues in particular:
• So-called "experts" making ill-informed arguments about computing systems.
• The complete failure to use data in comparing the current system and the proposed alternative.
• The sometimes deliberate confusion of two specific objections - the accuracy of systems and their vulnerability to malicious attack.
Computers are used in aircraft, cars and navigation systems because they are less prone to errors than humans. Yet, to listen to some "experts" going on about the unreliability of computing machines is like being back in times when rail was being introduced and they thought it wouldn't be safe for humans to travel faster than 20 m.p.h..
If computer error or failure was the cause of road deaths or aircraft crashes there would be an almighty hue and cry, and rightly so. Yet computers do their job silently, effectively and with fewer errors than humans, every day.
More frustrating than these Luddite views is the complete absence of data about how error-prone the current manual system actually is.
It is full of errors. People make mistakes counting. Boxes have gone missing - at least temporarily. Other errors have occurred. But no-one knows exactly what the not-insignificant error rate is. But there were more than 20,000 spoiled votes in the 2002 general election and it is estimated that at least 95 per cent of these were spoiled by accident (i.e. someone puts three number one preferences on the ballot paper, etc). So we know there is a minimum error rate in the current system of about 1 per cent every election. The real error rate is likely to be greater.
Data from the Department of the Environment gives us some idea of what the error rate in the computer system will be. With 6,500 machines operating during 16 hours in a polling day, the probability that an error can occur is 0.08 (i.e. there will be one observable error every 12.5 elections nationwide).
The key here is observable error rate - when these errors occur the error will be spotted by the user, the machine reset, and the vote then correctly captured.
As such the computerised system is at a minimum 12.5 times more accurate, less error prone and therefore more democratic that the manual system. With this huge improvement in vote accuracy we can think of no earthly reason, nor is there any data offered, as to why a simultaneous paper trail (or voter verifiable audit trail) is needed.
The second critique of electronic voting is based on the argument that there are sinister forces (shades of the X-Files) that will deliberately tamper with the system to change the outcome of the election.
So how easy would it be to manipulate or "hack" the computerised system?
Is there a way in which any voting system, either manual or computerised, can made 100 per cent safe from unwanted interference? No there isn't. If someone wanted to corrupt the manual vote then, given enough resources, they would find a way. We are not suggesting that this has ever been done; we are merely pointing out that it could be done.
However, the difficulty in corrupting the Nedap machine is at least as difficult as corrupting the manual system, if not more so. The Nedap machine is not similar to a PC. It runs specialised software and hardware that is not accessible to anyone outside of the manufacturers. Equally, the manufacturers take an exhaustive set of precautions to make sure that none of the software or data in the machine can be tampered with. To corrupt the computerised system would firstly require detailed technical knowledge of the Nedap machine (and therefore collusion with the machines' manufacturers) and access to the Nedap machines, presumably by the collusion of the returning officers.
A "big ask".
If we have reservations it is about the counting software. The problem is not its technical complexity - we could write a program to perform the entire Irish general election count using Microsoft Excel in a day. Our concerns are about the transparency, both perceived and actual, of the software itself. We believe the source code should be freely available. Indeed, for a very modest sum the Government could fund an open-source project to have the universities write a piece of code to count the votes.
Unfortunately this debate is also symptomatic of a wider problem, our inability to hold rational discussions about the relationship between technology and society. Social attitudes oscillate between naïve technological determinism on the one hand, exhibited in the headlong rush to parachute computers and the Internet into every classroom in the country, while on the other hand much of the e-voting debate is Ludditte in nature.
As IT professionals we believe electronic voting is good for democracy; the data show that it is a less error-prone and fairer system.
In fact the electronic tally data provided from now on should be published on the web so every voter and every school can see what happened in their community. Analysing voter trends should not be limited to a high priesthood of tallymen. Electronic voting could yet make experts of us all!
Annrai O'Toole is the CEO of Cape Clear Software. He was also founder of IONA Technologies. Brendan Tangney is a senior lecturer in the department of computer science in Trinity College Dublin.