E-Voting machines fail Michigan tests and others can be mis-calibrated

Steven Hodson

Over 97,000 e-voting machines across 20 states are manufactured by ES&S (Elections Systems Software) and as a quick search can show they are no strangers to bad press about their business. As we get closer to the election and the using of these machines that are now a critical part of the American election process we have a couple fresh reports from Kim Zetter at the Threat Level blog concerning questions about the machines. In the first report Kim passes along some information about a report that the machines being used in Michigan can be maliciously calibrated to favour any specific candidate.

"If one candidate has a check box in one place and a different candidate has it in a different place, you can set it up so that if you press on one candidate it gets recorded for another candidate," said Matt Blaze, a computer scientist at the University of Pennsylvania who led one of three teams that co-wrote the report (.pdf) last year. "But if you press on the other candidate, it gets recorded correctly for that candidate. You can make it work perfectly normally in most of the screen, but have it behave the way you want in small parts of it."
Voters have recently complained in a number of states about vote-flipping occurring on touchscreen voting machines made by ES&S. Voters in West Virginia and Texas complained that when they tried to vote for Democratic presidential candidate Barack Obama, their ES&S machine registered a vote for Republican presidential candidate John McCain instead. Voters in Tennessee complained of the opposite occurring -- that when they tried to vote for McCain, their ES&S machine registered a vote for Obama instead.

Apparently ES&S hasn't returned any phones made to them on the issue at this point in time but then later in the day Kim Zetter comes back with another report on the ES&S machines and how in Michigan the machines were flunking out on the tests being run on them. It would seem that the machines can't seem to come up with the same voter count tallies in consecutive tests. Oakland County Clerk Ruth Johnson is quoted as saying

"The same ballots, run through the same machines, yielded different results each time," she wrote. "This begs the question -- on Election Day, will the record number of ballots going through the remaining tabulators leave even more build-up on the sensors, affecting machines that tested just fine initially? Could this additional build-up on voting tabulators that have not had any preventative maintenance skew vote totals? My understanding is that the problem could occur and election workers would have no inkling that ballots are being misread."