It’s sometimes hard NOT to be a conspiracy theorist. As someone who is a supporter of touchscreen technology and its uses (and voting machines would fall into the excellent use category, if done properly), the problems with touchscreen voting machines (manufactured by Diebold, which already called the election for McCain back in February), are cringeworthy. Basic, basic errors that could have, should have been detected months and months ago. Errors both from a technical and a usability standpoint. And this with reportedly 1/3 of Americans voting using touchscreen machines.
To whit, from the Charleston Gazzette:
Shelba Ketchum, a 69-year-old nurse retired from Thomas Memorial Hospital, described what happened Friday at the Putnam County Courthouse in Winfield.
“I pushed buttons and they all came up Republican,” she said. “I hit Obama and it switched to McCain. I am really concerned about that. If McCain wins, there was something wrong with the machines.
“I asked them for a printout of my votes,” Ketchum said. “But they said it was in the machine and I could not get it. I did not feel right when I left the courthouse. My son felt the same way.
Wood said some voters might not realize that touch-screen voting machines may take a few seconds to record their choices.
“The reaction time [on the machines] may be different. And when you hit the screen a second time, it cancels your vote,” Wood said. “When you get in a hurry, if you go to fast and hit it again, you can cancel what you just did.”
Wow. Just wow. Not only is the Blame the User mentality on full display here, but why would the machines take a few seconds to record choices? Is the latency that bad, and if so, why? My touchscreen ATM machine (also by Diebold), doesn’t seem to have that issue. And if tapping the button a second time cancels the vote, well, that is something that could have been fixed in design. If you wanted to change your vote after you’d pressed the button, why not change it by selecting another option instead of a second tap to cancel. Or (wait for it) have a cancel button perhaps? The design choice to not give people a print out of their vote (a receipt in other words) was also a terrible, terrible idea.
But wait, there’s more. From the Decatur County Chronicle:
Franklin Boroughs says he intended to vote for Republican but rather the computer had checked the Democratic candidate instead. Wanda and Barney Blasingim similarly said they tried to vote for McCain but the machine switched the vote.
“I noticed the problem immediately,” Wanda said Monday. She says she touched the “button” for McCain a second time and the problem was corrected. Her husband said he asked for assistance from election workers and was told the error sometimes occurs when a person’s finger touches close to the line of the box the candidate’s name is in.
While Blasingim maintains that his finger was not on the line, Election Commissioner Rick Box said the trouble may be that when a person is standing in front of the machine, it may appear their finger is poised over one button but it is actually closer to the button above. “The way the machine is set up, when you are standing in front of it and seeing it at a certain angle, it looks like you are touching the middle (of the button) when you are actually touching the line above it,” Box said.
Box and fellow Election Commissioner Grafton Dodd tested the machines on Monday. Dodd could not be reached for comment but Box said he found the area of the screen where the buttons for President are located are extremely close together. He blames the problem in part on poor design by software programmers, and adds that there may be sensitivity issues with the screen itself.
In this case, clearly the touch targets are too small and/or else the buttons have far too much overlap. Again, a simple fix if only someone had simply designed it well and tested it with users.
My home state of Maryland has tried to help by putting up an online demo of touchscreen voting, but my guess is the people who need the most aid probably aren’t online. And the problems people are having are ones that cannot (and should not) require training to fix. You shouldn’t have to learn how to vote. (Or at least not very much.)
Even as I’m writing this, the Election Assistance Commission is suspending use of the machines for failing to comply with national standards.
Of course, some of these problems have been well known for months. Back in January, the New York Times Magazine asked, Can You Count on Voting Machines? One example from the article:
In 2005, the state of California complained that the [touchscreen voting] machines [Diebold AccuVote-TSX] were crashing. In tests, Diebold determined that when voters tapped the final “cast vote” button, the machine would crash every few hundred ballots. They finally intuited the problem: their voting software runs on top of Windows CE, and if a voter accidentally dragged his finger downward while touching “cast vote” on the screen, Windows CE interpreted this as a “drag and drop” command. The programmers hadn’t anticipated that Windows CE would do this, so they hadn’t programmed a way for the machine to cope with it. The machine just crashed.
The sad answer to the Times’ question seems to be, for now anyway, no.
What can be done? Well, it’s too late now for this election, unfortunately. The problems are unlikely technical. Touchscreen technology seems to be stable and secure enough for banking, retail, and airlines. The problem is mainly in the design. The UPA has a Voting and Usability Project and it would be nice to see interaction designers being involved as well. It’s a solvable problem; it’s just too bad it wasn’t solved before now. It’s the kind of bad execution that can ruin a perfectly good solution for years to come.