Palm Beach County's election woes continue

Recounted votes don't match up with ballots cast

Two weeks after Palm Beach County's Aug. 26 primary, election officials still don't know why the number of paper ballots they've recounted aren't matching up with the number of ballots that were cast in the primary.

After multiple recounts using optical-scanning machines and manual hand counts, the numbers aren't in agreement, said Robert Weiner, an administrative aide to Arthur Anderson, the county's supervisor of elections. Last week, Weiner said he believed the discrepancy was probably the result of boxes of ballots that weren't reprocessed as part of the recounts, but the controversy continues.

Election workers have been continuing the recounts around the clock since the election to try to resolve the discrepancies, Weiner said, but even a state-mandated random audit of one voting precinct came out wrong. In the random audit of one precinct, 144 ballots were counted but the vote tally was one higher, or 145 votes.

"Does that mean we have a fault in our electronic equipment? Does that mean we have a fault in the counting, such as a human error?" Weiner asked. The next step, he said, "is to look to see whether there was something improperly tabulated by computers or humans.

"We have to figure out why [this happened] and sometimes there is no answer," he said. "In this case, there appears to be no direct answer."

The recounts of the past two weeks were conducted because of a race for a judgeship that was separated by 17 votes, triggering a mandated recount to check the results.

The problems continued when the recounts found varying total numbers of ballots. On election night, the county reported that 102,523 paper ballots were filled out by voters, but the ensuing hand recounts tallied up only 98,775 ballots or about 101,000 ballots.

"So unfortunately, this is a mess and we come off looking very poorly, where in fact we are a group of tremendously dedicated professionals," Weiner said.

This is the first time that the county is using the paper-based, optical-scanning balloting system from Sequoia Voting Systems in Oakland, Calif. Previously, the county used touch-screen, or direct recording electronic (DRE) voting machines. The DREs have been largely replaced by the optically scanned paper ballots under a July 1 state mandate to move to a more reliable system that maintains a paper record of ballots cast.

"You have to remember that this is brand new equipment ... that we have not used before [and] was thrust on us by the legislature ... at a time when national elections are pending," Weiner said. "This is now the equipment that we have, and this is the situation that we are in."

Weiner said he believes the problem relates to the new tabulating machines, which scan and count the paper ballots. "I do not feel that the machine counts on the night of the 26th were accurate," he said. "And why not? That's a good question."

One possibility is that some ballots were accidentally run through the counting machines twice. Another possibility is that one voting precinct suffered a power outage on election night and that ballots were run there by accident before and again after the outage.

"I don't know if that's accurate or not," he said. "Every yield has had a number that's significantly less. There is an issue being presented somewhere that we're not yet able to identify. The process itself is sound. The ability of the people to perform is sound."

"We're doing everything we can to satisfy the voters, to satisfy the world in general" about the results. "We can't identify anything right now [that's causing the problems], but it doesn't mean we stop. We will continue to work to figure out what happened."

Representatives from Sequoia have also been working on the problems, he said.

Michelle Shafer, a Sequoia spokeswoman, said the vendor maintains that the problems are not with the machines, but are related to the election administration and recounts themselves in the county.

"There doesn't appear to be any equipment problems at all," she said. "We believe this is completely an administrative and ballot counting issue, and it does not relate to the performance of the ballot count machines in Palm Beach County."

Sequoia "had folks there last week and we worked with the county and the [state] secretary of state's office and [the machines are] fine," she said. "They're tried and tested machines. Elections come down to the people, the process and the technology, and all of them have to work together for elections to go smoothly."

Avi Rubin, a professor of computer science and an e-voting activist at Johns Hopkins University in Baltimore, said such recounts can be confounding.

Recounts don't always mean that every number is going to be in agreement because of the large number of ballots cast and the problems inherent in humans recounting them manually, he said.

"In a recount, the goal isn't to try to get the ballots to match up to the machine count, but it's to provide an independent count," Rubin said. "Hand counts of a lot of ballots are prone to error" within a small and acceptable margin of error.

What officials should do, he said, is to "try running them through the machine again. A lot of ballots are designed such that if you are manually counting for just one particular race, it's going to be very difficult because that one race might be [placed physically] somewhere in the middle of the ballot. People [doing the recounting] are looking through a ballot that may have a whole lot of things on it. For a specific race, that's a difficult process."

If that doesn't help, he said, officials should consider an audit of some of the scanning and tabulating machines by running a known quantity of ballots through them and comparing them to the results to see if they give the same answers as the hand counts.

"Try spot-checking the machines that way," he said. "Because one thing that you want to eliminate is the possibility that the machines are not counting right."

Proponents of the DREs could hold up these paper ballot recounting difficulties as evidence that DREs are a better system, but that would be a false conclusion, Rubin said.

"I think what we're having here is the lesser of two evils," Rubin said. "We have a problem here that's solvable. We actually have the ballots, and we have this difficult problem of counting them. With DREs, you don't even have ballots" to be able to recount them.

Copyright © 2008 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon