Brent Batten: Voting machines pass early test

By BRENT BATTEN, bebatten@naplesnews.com
August 12, 2004

The bad news is, errors showed up in the test of touch-screen voting machines conducted Tuesday.

The good news is, they were of the human kind.

A random sampling of 18 machines, which tabulate votes without producing a paper receipt of each person's choices, accurately recorded and transmitted the choices input by 18 volunteers at the Supervisor of Elections Office.

The successful test didn't come as a surprise to Elections Supervisor Jennifer Edwards, who says she's had confidence in the machines all along. But she acknowledges that some people won't be comfortable with the new technology unless and until there are a few controversy-free elections in the books. "I think it has to do with change," Edwards said.

The anxious moments Tuesday occurred when Deputy Supervisor of Elections Gary Beauchamp announced that the results compiled by the voting machines did not match the scripts the elections staff had drawn up beforehand.

But the staff quickly deduced that the discrepancies occurred because the volunteers hadn't followed their scripts.

In one case, a volunteer who was supposed to input 10 different mock ballots only input nine, causing the vote tallies for candidates on the missed ballot to be one short of the script. The fact that the volunteer failed to check off any of the candidates she was supposed to vote for on one of the mock ballots in her stack tipped off the staff. When the candidates on the unchecked mock ballot all came up one vote short of the expected amount, the case was solved, Beauchamp said.

In the other instance, a volunteer who was supposed to vote five ballots instead voted six. That happened because the volunteers, inputting several ballots apiece, opened a new ballot upon completing a previous one. That can't happen on election day when a poll worker opens just one ballot for each voter. "They (volunteers) act as both voter and activator in (the test). At the polls, a worker does that," Beauchamp said.

The volunteer who opened the extra ballot apparently realized his mistake and closed the unauthorized ballot, but made it leave his screen by pressing "vote" instead of "cancel." The result was several unscripted "undervotes" appearing in the final tally, as if a voter had gone to the polls then not voted in any race.

"I'm satisfied the evidence that's been presented explains the anomalies," said County Court Judge Vince Murphy, who oversaw the testing as a member of the county canvassing board.

That doesn't mean there isn't room for improvement.

In an effort to appear fair, the elections staff scripted many races that will appear on the Aug. 31 primary ballot to end in ties in Tuesday's test. But that means there was no way to know if the machines were correctly assigning votes to the proper candidate. With all the candidates getting the same number of votes, how do we know that Candidate A didn't get Candidate B's votes, and vice versa, wondered one observer.

Oddly enough, the mistake of the volunteer who failed to complete one of her ballots helped alleviate that concern, since the expected candidates were the ones found to be one vote short of the script.

Murphy agreed that, in the future tests, it would be wise to vary the vote total for each candidate.

And everyone looks forward to the day when tests of voting machines aren't considered news.