A brave new world of voting

Humanist, Jan-Feb, 2004 by Michael I. Niman

Wired magazine reports that researchers from the Johns Hopkins University Information Security Institute found "stunning flaws" in Diebold's Georgia program. In addition to geek taboos such as embedding security passwords into program source code, the Johns Hopkins analysts found flaws that could allow voters to vote multiple times of allow votes to be changed by a third party after being cast, in some cases by remote access.

Another group of analysts, working on contract for the Maryland state government, found, according to the Independent, 328 software flaws, including 26 which they deemed as putting the election "at risk of compromise."

Georgia and Nebraska haven't cornered the market in suspect elections. They seem to be arising wherever the new electronic voting machines pop up. The odd thing is that wherever an electronically administered election defies statistical predictions, it is almost always to the favor of the Republican candidate. In Alabama, for example, a seven thousand-vote tally shift threw the close gubernatorial election from the incumbent Democrat to the Republican challenger. And again, in Alabama as in Georgia, there was no recount.

Touch screen voting machines aren't inherently prone to election manipulation. Touch screen machines that generate a paper receipt, verified by the voter and stored by the machine, allow for accurate recounts. They also allow voters to examine the choice that the machine reports they made. This is important because the new machines, aside from being susceptible to tampering and malicious programming, ate also error prone. One study conducted jointly by the California and Massachusetts Institutes of Technology (Caltech and MIT) found the new touch screen machines to be more error-prone than the notorious punch card machines of Election 2000 fame. One major problem has to do with alignment. The spot on the screen with the candidate's name may not line up with the coded segments of the screen that register a vote for that candidate. Voters in many recent touch screen elections, for example, have complained of machines that flash the opponent's name when they try to vote for their preferred candidate.

Additional voting machine problems came to light with the November 4, 2003, elections. According to the Washington Post, in Fairfax County, Virginia, it took "more than 21 hours to get final election results from [the county's] new computerized machines; when all was cast and done, enough doubts existed to prompt legal action by some Republicans who lost." Ten machines from nine precincts had broken down during the day, had been removed for repairs, and then were returned to service; Republicans later asked a judge to impound them. There were also voter complaints about several tries being necessary before votes were registered. And a few precincts even went back to paper ballots. The Boston Globe focused on a Republican school board candidate in the county who lost by a small margin and then "learned that at least one of the computerized ballot boxes had a glitch that may have subtracted some of her votes."