by Eric Reiss - 30 October 2004

 

Did we replace one usability problem with another?

After the hotly contested 2000 election results, the State of Florida hopes next Tuesday's election won't leave anything hanging - including chads. That's why 15 of Florida's 67 counties have purchased touch-screen voting systems. The question is, have these machines done away with a problem or just created some new ones?

For those of you who want a recap of the 2000 debacle, check out Dan Brinklin's superb blog.

Basic mistrust of the technology
A lot's been written about these new devices, which will be used by almost 30% of America's voters on November 2. But most of the discussion has been pretty negative. In 2003, Professor Avi Rubin at Johns Hopkins University examined the source code for one of the most popular models, the Diebold AccuVote-TS, and pointed out how easy it was to hack the system. This is one of the reasons the State of Ohio dumped the Diebold system in favor of the old-fashioned punch cards that gave Florida so much trouble back in 2000. Remarkably, Ohio is home to Diebold Manufacturing.

It's frightening to learn that according to a recent Time Magazine poll, of the 1,059 American voters interviewed, only 50% could claim they were "not worried at all" their vote would be counted accurately. Good grief! Even Afghanis seem to have more faith in their electorial process (finally some proof of headway in the War on Terrorism).

Despite these concerns, it's important to remember that Florida's inability to count votes was far outweighed by the inability of Florida's voters to cast them. If you recall, hundreds of Gore supporters in Palm Beach County appear to have voted for the ultra-conservative Pat Buchanen.

User instructions
As a voter in Miami-Dade County, I finally received my registration card with mere days to go before the 2004 presidential election. Along with it, I found a set of instructions for using the new iVotronic machines from ES&S. Here they are:

I found Step 2 particularly intriguing because it includes a non-standard operation: canceling the first choice in order to make a new choice. This is odd since you won't find a lot of other applications that force you to manually cancel a choice - in our electronic age, a new choice usually replaces the old one automatically. Like when you use a radio button or drop-down.

It's easy to envision a frustrated voter stabbing vainly at the name of an alternate candidate and not understanding why the machine won't react. Will these voters assume their vote has already been recorded and eventually leave the booth without ever advancing to the Review screen? There's some evidence to support this. In a hotly contested January 2003 election in Florida's infamous Palm Beach County, only 12 votes separated the winner from the loser. But here's the kicker - despite the fancy touch-screen voting machines from Sequoia, 134 voters left the booth without actually casting a vote!

Outdated workflow?
One reason for this odd workflow could be a worry that a voter's hand could inadvertently brush another choice, causing the ballot to change. However, you'd expect the Review screen in Step 3 to provide a measure of security. Also, I have yet to meet a touch screen that didn?t need a pretty determined jab to make it react.

 

But I think there's another reason - a method for canceling a choice has always been an integral part of designing paper ballots (see the instructions for the Florida absentee ballot at the right). Instead of changing the workflow to reflect the new technology, the old workflow was forced into the new interface. Like holders for buggy whips on early automobiles.

The Disability Rights Education & Defense Fund, a California-based advocacy, lists cancellation as one of the 15 functions it deems necessary in a touch-screen system. In most respects, this is a pretty savvy group. But I think they slipped up on this one.

Taking user ability for granted
Another potential problem is the fact that many voters simply don't have a lot of computer experience. Keep in mind that 17.6% of all Floridians are over 65 years old. That's over 40% higher than the national average.

Although most people may do alright at their local ATM, don't expect these folks to understand the concept of scalable fonts or other features that have become second-nature to the word-processing crowd. Nevertheless, scalable fonts for the vision impaired is always prominent on the list of usability virtues.

Miami seems to have advanced from being the only officially bi-lingual city in the nation to the first tri-lingual area - everything official is now printed in English, Spanish, and Cajun (for the benefit of former Haitians). Choice of languages always presents a usability challenge - particularly if you choose the wrong language and have to backtrack. Is it fair to assume inexperienced users understand back buttons or iconographic arrows?

The State of Maryland commissioned researchers from the University of Maryland to do a usability study of the Diebold AccuVote-TS. And apparently, the State of Florida didn't read it. Bad move.

To paraphrase astronaut Jim Lovell, "Tallahassee, we have a problem." Let's hope I'm proven wrong a few days from now.

Comments or questions?

Back to "Articles" page