Voting test shows pros, cons
 Sunday, August 18, 2002
 By A.J. RENNER Repository staff writer

 A study by the University of Maryland demonstrates the potential and
 some drawbacks of Diebold's AccuVote-TS system.

 University professors Benjamin Bederson, director of the
 Human-Computer Interaction Lab, and Paul Herrnson, director of the
 Center for American Politics and Citizenship, conducted the study for
 four Maryland counties that bought the machines.

 The study tested how easily the machines are used. It did not test
 security, reliability, or accuracy, nor did it test voting machines
 manufactured by other companies.

 One of the machines failed early on, when it would not return the
 voter card needed to activate the ballot.

 In an interview, Herrnson conceded that it was "dumb luck," from a
 research standpoint, that one of the two machines he was testing
 broke down. Mechanically, he said, it was probably easy to fix.

 But from an elections standpoint, it was a "catastrophic error," he
 said.

 "It creates insecurity of the machine, it creates insecurity of the
 system. One voter wonders if their vote was counted," he said, while
 others wonder how well the other machines worked.

 "The product that they received were some of the prototypes that we
 put together," not an election-ready machine, said Mark Radke,
 director of voting industry for Diebold Election Systems.

 Ohio counties using electronic machines anticipate such
 failures. They either assign a minimum of two machines per precinct,
 or have extra machines, to be assigned if a precinct's sole voting
 device fails.

 In the Maryland study, 80 percent of the public involved in the study
 said the touchscreen machine was easy to use, and 86 percent said
 they were comfortable using it.

 However, testing by experts and the general public revealed concerns,
 including:

 * Voters didn't know where to place the card. When they did find the
 slot, they expected it to act as an ATM, sucking it into the machine.

 * Ballot instructions were unclear, the display was not easy for
 everybody to read, there was no help button, and the layout could be
 improved.

 * The machine doesn't warn voters that they have tried to overvote.

 * The machine alerts voters when they have not chosen any candidate
 in a race, but not when they've chosen only one candidate in a race
 where they may pick more than one.

 * The audio system, used primarily by the blind, was difficult to use
   and the sound quality was poor.

 Diebold is partnering with the National Federation for the Blind to
 improve its audio system, said Radke.

 Radke pointed out some of the problems, such as unclear ballot
 language, would be addressed by local boards of elections. He also
 points out that the machine won't let voters overvote.

 "I think that the bottom line is that it's a pretty good system, and
 it probably will work pretty well," said Bederson. "Most users were
 quite pleased and thought it worked fine."

 His concern is the users who "self-select" away from computers. "Even
 though I use a computer hours and hours and hours a day, there are
 still ... a significant number of registered voters in this country
 who probably have not used a computer in their lives."

 Many of the problems could be improved with voter outreach. 

 Diebold is supplying more than 19,000 AccuVote terminals to 159
 Georgia counties. Their contract includes training and outreach. A
 mock ballot that allows voters to run through three races is
 available online.

 Any new system will require an outreach. "We did a pretty big voter
 education program in the city of Youngstown: Block Watch, council
 meetings, ward meetings, churches, civic groups," said Tom McCabe,
 deputy director of the Mahoning County Board of Elections. Youngstown
 used a system made by Election Systems & Software; the entire county
 will use the new system in November.

 Herrnson and Bederson suggest placing a practice machine in each
 polling location. It's a good suggestion, Radke said, and Diebold has
 done it in an issues election.

 But it would be cost-prohibitive for some districts: Stark County
 alone would have to buy 220 extra machines.

 The idea could work in primaries that don't involve the entire
 county, and perhaps in off-year elections where the turnout is
 expected to be low, Radke said.

 The University of Maryland study should not be used to condemn the
 AccuVote, or electronic voting in general, said Herrnson.

 "Our goal was to say ... here are some ideas or some flaws, here are
 some ideas that we can use to think about."

 Read the University of Maryland study online at:

 
http://www.washingtonpost.com/ac2/wp-dyn/A3785-2002May10?language=printer

 Vote on a mock-up of Diebold's AccuVote-TS at:

 http://www.georgiacounts.com/online_demo/index.html