Did electronic voting pass the test?
Paper trails, anyone?
At about the time that Senator John Kerry had accepted defeat and phoned President Bush to congratulate him, stories were circulating on the Internet claiming that the electronic voting machines in Florida and Ohio and some other states might have been rigged for a Bush victory.
The claim stems from the fact that exit polls were indicating a marginal Kerry victory in those key states, but his apparent exit poll advantage was not reflected in the total vote count. This indeed was the shape of the story if you sat through the election night telethon. At first it looked as though Kerry was doing well, but as the night wore on a Bush victory became more and more likely.
So what are we to think of the claim? Despite the "conspiracy theory", there is good reason to believe that it was a genuine Bush victory. First of all, the final outcome reflected the fact that Bush held a small lead in the opinion polls right up to election day. Although all of the individual polls were subject to a margin of error greater than Bush's lead, the aggregation of the polls was still slightly in favour of Bush (and this reduces the statistical error margin).
The pollsters had been plagued by suggestions that they were not properly accounting for the youth vote and most, if not all of them, examined, re-examined and adjusted their weighting parameters in an attempt to account for the expected high youth vote for Kerry. The pollsters have a big self-interest in not being too far wrong.
The indications, on election night itself, were that the level of disenfranchisement through technology failure, long lines of voting and voters being turned away from the polls for lack of proper credentials, was much lower than in 2000 and, although there may have been one or two areas where there were problems, there is no reason to believe that the election was skewed by such incidents.
Another straw in the wind was the gambling money - which has historically provided a reasonable guide to an election's outcome. While it is illegal for most American's to place bets over the Internet (on anything), many of them do. Throughout the whole campaign the betting odds were in Bush's favour - in effect predicting a Bush victory simply by the weight of money that was gambling on that outcome. The figures for the total bets placed (on Betfair one of the leading sites for such bets) was $4.2m on Bush and $1.2m on Kerry.
Finally, the results from Florida and Ohio, which were only marginally in Bush's favour were not particularly out of line with the voting in the US as a whole. As it worked out, these results seemed to reflect the mood of America.
So what are we to think of the electronic voting "conspiracy theory"? Here too there are reasons to pause for thought. The companies that supply the machines (Diebold Election Systems, Election Systems & Software, Hart InterCivic, and Sequoia Voting Systems) would destroy their own business if it were ever discovered that the technology was compromised. Would they take the risk? I personally doubt it, especially as it would involve bringing more than one or two people into the "conspiracy", any one of whom could go public on what was going down.
Also, bending the software to affect the result in a very subtle way (and get it right) is probably very difficult to achieve. The margin for failure is high and the whole scheme is very risky.
There is however legitimate cause for concern in the simple fact that many of the electronic voting machines that were deployed did not have audit trails that validated the figures they gave. If there were any kind of malfunction in any of these, there was simply no way to validate the figures. The justification for complete transparency and validation of voting technology is not only desirable but necessary. Indeed if ever there was a case for the open sourcing of program code then this is it.
One hopes that by the time the next major elections in the US come round, there will be paper audit trails on every voting machine deployed.
Copyright © 2004, IT-Analysis.com