Voting Machine Foul-Ups Delay Local Count
By JAKOB SCHILLER (03-05-04)

Berkeley voters ran into a number of glitches Tuesday when the machines that clear voter cards after they are used malfunctioned throughout the day, forcing several precincts to move to paper ballots which quickly ran out and had to be re-supplied by the county.

According Elaine Ginnold, the Assistant Registrar of Voters for Alameda County, the county had trained poll workers to run the machines using a particular screen interface which somehow changed when the machines were set up at polls. As a result, the poll workers were unable to work the machines and had to call in for help. An estimated 200 precincts around the county were affected.

Ginnold speculated that the problems happened because the machines’ batteries and/or memory cards were shaken lose during transport to the polling stations.

The slowdowns forced by the machines, along with problems encountered at City Hall when Berkeley and Albany tried to submit the results to the county over secure phone lines, left Berkeley behind the rest of the county in submitting the vote tallies.

According to Councilmember Kriss Worthington, by 9 p.m. when Alameda county had counted 33 percent of their totals, Berkeley had only counted 2 percent. By 11:49 p.m., when Alameda was at 89 percent, Berkeley was still at 38 percent. Alameda county was also slower than other counties, according to Worthington. San Francisco county was finished when Alameda’s count was only half done.

The card encoders that caused Berkeley’s tardiness were controversial even before the vote, according to Jim March, one of the plaintiffs who recently filed a suit against Diebold, the manufacturer of the card encoders and the touch screen voting machines. He said the machines were never properly certified before they were bought by the county.

Ginnold said the machines did get a conditional certification but are pending further review.

Other problems encountered throughout the day included voting machine freeze-up. That forced poll workers, who were not trained to re-boot the machines, to call in county workers, again delaying voters.

Leshaun Yopack, a employee for the county confirmed the freezes but said every instance they responded to was quickly fixed by wiping out initial votes and then giving voters the opportunity to re-cast their votes.

Around the country, as well as in Berkeley, critics of the new electronic technology also used the day to enroll as poll workers to get a first-hand glimpse of other problems they say could potentially plague the system. Avi Rubin, a professor at John Hopkins university and co-author of the report that analyzed the software used by Diebold to run the machines, enrolled as an election judge in Baltimore County to help him examine firsthand several of his concerns.

In particular, opponents are worried about the technology being vulnerable to hackers who could tamper with the voting machine, the systems used to submit the votes to the county and the machines used by the county to tally the votes.

Here in Berkeley, Judy Bertelsen, a Berkeley physician who has attended several of the state-wide meetings concerning the new electronic machines, did the same as Rubin. She said she found several security concerns, prompting her to write a letter to Brad Clark, the Alameda County Registrar of Voters.

She said the voting machines were left unattended at her polling place the night before with only thin plastic strips locking them shut. One machine was not locked at all. During her training before the election, she also criticized the county’s decision to use the same password1111—to run all the machines.

“The minimal [security procedures] they run past us are insulting,” said Bertelsen. “Why have a password if everyone knows it?”

At Bertelsen’s polling place, paper results of the totals were not posted, contrary to new security s issued in a directive from Kevin Shelley, the California Secretary of State. In the directive, Shelley required that all polling places print out a paper record of the ballots cast by each machine and post them.

“Some members of the public and the media have indicated concern that once the results of the vote leave the polling place citizens have no ability to check on whether the results from that polling place are accurately conveyed to the central counting facility. Therefore, a copy of the results from each voting unit that is capable of printing out a tabulation of the results shall be posted for public inspection for at least 24 hours outside each polling place," wrote Shelley.

Bertelsen’s precinct printed out paper sheets that were sent to the accumulation site along with the voter cards that contained the tallies. “It would be such a minor thing to print a second one,” she said.

According to Ginnold from the Alameda County Registrar’s Office, Shelley’s directive came too late for the county to include the extra safety measures in their training sessions for poll workers.

On a statewide level, the malfunctioning voter card encoders also added to arguments of touch-screen opponents such as March, who say the cards and card encoder machines fall outside the current security checks that are required only for the machines used to actually cast the votes. According March, the card encoders are excluded from the security checks that only verify the information sent between the voting machines at the poll places and the county.

He said malicious code or instructions could be placed on the cards from the card encoders that would continually tamper with votes. If a machine sent a switched vote, and the county recorded the vote, the paper audit trail would not be able to detect any switch because the too votes would still match.

“If there is fun and games on these cards then all the downstream process gets funky,” said March. “This card presents a hack problem that is above the paper audit trail that posting results will solve.”