Features

net.wars: Bushed

Presidential elections have always been followed up by acres of political analysis. We now realise that 2000 permanently ushered in a new era, in which the political analysis is accompanied by an almost equally lengthy technical analysis.

We live in increasing and justifiable uncertainty that the technology we are using is delivering anything like reliable results. In the 2000 election, the problem was that the numbers in Florida (not the only such state but the most obvious) were within the margin for error. This time round, there seem to have been at least two such states – Ohio and New Mexico. This seems to me the key issue: probably no system we ever devise will be perfectly accurate down to a single vote, so there will always be a percentage difference below which the count is unreliable.

Slashdot has been collecting reports, of course, but the central clearinghouse is the Election Incident Reporting System site, where more than 31,000 incidents were logged and counted. The raw numbers don't tell you anything, of course. Many of the logged incidents are minor-league stuff, like people who have been issued with absentee ballots wanting to vote in person instead, or people unable to find their polling places. What matters, and will take much more lengthy analysis, is the pattern of these incidents.

Of course, a few Incidents have already made their own noise. You've probably already seen the reports of things like the voting machines whose memory cartridges began counting backwards once they'd reached a certain number of votes, the 800-voter Ohio precinct that recorded nearly 4,000 votes for Bush, or the Florida counties where high percentages of registered Democrats nonetheless translated into low percentages of votes for Kerry.

Was the vote hacked? Bev Harris thinks so, claiming that all optically scanned votes are ultimately tabulated on a single, central Windows PC and that the Democratic Florida counties voting Republican all used optically scanned ballots. And what about those exit polls?

The New York Times has just published a rebuttal of some of these claims. The apparently Democratic counties in Florida voting Republican, for example, is an old and well-known phenomenon. ("Can't we throw them out?" a Democrat said to me plaintively last week. "They betrayed us.") In a separate article,the paper also concludes that voting technology worked mostly as expected. A joint Caltech/MIT project has concluded in a report just released that there is no evidence from the exit polls to indicate that electronic voting machines were used to steal the 2004 election and award it to Bush. Essentially, they argue that the variance between the exit polls and the actual vote tallies were, again, within the margin for error when you examine them at the state and local level, where elections are actually conducted.

In many ways, all of this is good news. Due to the efforts of groups like the National Committee for Voting Integrity and the ACLU, and individuals like Rebecca Mercuri, Avi Rubin, and Peter G. Neumann, the technology is under increasing scrutiny.

Also good news is the increasing drum beat in favour of voter verifiability using a paper trail. That is in fact one of the benefits of optical scan: the voter fills out the ballot and can check it over thoroughly, then inserts it into a sealed box via a reader which tabulates it. If there is any question, the box can be opened and all the votes recounted. And it's easy enough to test such a system by recording some votes and comparing the machine count with the hand count.

Does the election result make me happy? Regular readers probably know the answer is no. I never came round to believing that Bush won the 2000 election, and like a lot of Americans, I dislike his policies. But if we are to believe there was fraud in 2004, there is going to have to be more evidence than that 49 percent of the US didn't like the result. Everyone I know is exhausted, confused, and alienated: how could the country have done this?

But the answer is not, this time, so far as we can prove, to blame the technology. Yes, we need better voting technology. We need it to be accountable, voter verifiable, open to independent scrutiny (many voting system vendors demand that election officials sign non-disclosure agreements, something that is wholly inappropriate), testable, and accessible. But, in the immortal words of Utah Phillips, "If God had meant us to vote, He'd have given us candidates."

That does not mean we can sit back and stop worrying about voting systems: election fraud is as old as elections themselves. With so much at stake, as the technology matures, we can expect methods for subverting it to become better developed and more subtle. The scary thing is not gross fraud that is easily detectable, even if it's successful (see for example Greg Palast's analysis of the 2000 cleansing of the Florida electoral rolls). The scary thing is the fraud that's too fine to detect. Did that happen this time? Only your conspiracy theorist knows for sure.