Maryland e-voting controversy continues in presidential race
By William Welsh for Washington Technology. 03 November 2004

A voter advocacy group monitoring the use of electronic voting machines in Maryland reports a number of software glitches occurred during yesterday?s presidential election, but state election officials said the allegations were baseless.

The software running on the touch-screen machines used across the state failed to record some votes correctly, jumped to other pages on the ballot without being prompted by the voter and inadvertently omitted some political races, according to TrueVoteMD, a nonpartisan citizens? group focused on protecting voting integrity.

?We have received hundreds of calls from across the state,? said Bob Ferraro, the group?s co-director, said Tuesday afternoon.

The group set up a voter hotline and deployed 600 poll watchers throughout Maryland to monitor its new touch screen voting machines manufactured by Diebold Inc. of North Canton, Ohio.

Yet officials at the Maryland State Board of Elections said they had received no reports of any major problems. The only problem the board reported yesterday was a failure to have a piece of equipment capable of encoding voter access cards available at one precinct, said Pamela Woodside, the election board?s chief information officer. That was attributed to human error, not equipment failure, she said.

Maryland was one of four states in the nation that switched all of its counties to direct recording electronic, or DRE, equipment following the 2000 presidential election. The other states are Delaware, Georgia and Nevada.

Of those, only Nevada uses a so-called paper audit trail that provides a paper back up if a recount is necessary. TrueVoteMD has tried unsuccessfully to get the paper audit trail implemented for Maryland?s touch screen machines. Touch screens are a type of DRE system.

About 16,000 touch-screen machines were used in about 1,600 precincts throughout Maryland yesterday, the elections board said. The only jurisdiction not using Diebold?s touch screens was the city of Baltimore.

Ferraro offered several anecdotal stories of touch-screen software mishaps encountered by voters. A woman in Baltimore County pushed her ion for president and senator repeatedly, but couldn?t get the machine to register her choice properly. A man in Montgomery County said the machine skipped right past the presidential and senate races. A woman in Montgomery County tried to make her ion for the county school board, but the machine advanced to the next screen after she had chosen only half of the candidates.

Software advancing to the next screen on the electronic ballot before a voter has completed casting his or her vote was a new occurrence in the presidential election, Ferraro said. Maryland used the touch screens in primary elections earlier this year. The advancing is ?a serious problem,? he said.

The group plans to compile its findings in a report to be made public about a week after the election, Ferraro said. It does not plan to contest the results of the election because it is a nonpartisan organization. However, it does plan to proceed with a lawsuit next year to force Maryland to use a paper audit trail, he said.

In the past, the elections board has insisted that problems of this nature are a result of voter error and not equipment failure, Ferraro but the ongoing problems are evidence of systemic problems, Ferraro said.

?A system that has so many people making mistakes is a poorly designed system,? he said.

Maryland elections officials had about 500 technicians deployed throughout the state to monitor the performance of the touch screens, Woodside said. While most of them were Diebold employees, they also included volunteers from colleges and universities as well as government employees drawn from county IT departments and other agencies, she said.

Woodside refuted each one of the alleged software glitches and equipment malfunctions in turn. She said software that was allegedly jumping ahead to the next screen was a result of human error and not performance error. For example, the elections board learned of an instance where one voter was leaning with her purse on the machine and inadvertently activated that part of the screen that advances to the next page.

As for reports that some electronic ballots were incomplete, those charges were ?impossible,? Woodside said. ?We validated that database time and time again. We checked it before deploying the equipment,? she said.

The election board also was running a parallel monitor at its command center throughout election day on which it cast trial votes without a problem, she said.

Woodside said that the wrong candidate?s name showing up might indicate a calibration problem, but the elections board received no reports of such problems.

?If something like that were to happen, we would shut the machine down, and it wouldn?t be used,? she said.