Articles below are reproduced under the Fair Use exception of 17 USC § 107 for noncommercial, nonprofit, and educational use.
| EJF Home | Where To Find Help | Join the EJF | Comments? | Get EJF newsletter |
| Vote Fraud and Election Issues Book | Table of Contents | Site Map | Index |
| Chapter 10 Voting Problems In The 2004 Elections |
| Next Election Problems In Florida 2004 |
| Back Electronic Voting A Mess In March 2004 California Primary |
[EJF comments in Courier font]
Discrepancies found in numbers of Gaston County votes, voters
Gaston County, North Carolina investigates election tally errors
Gaston County elections director resigns
North Carolina's ballot blues by Joyce McCloy
Winston-Salem We've got a problem
Our key decision-makers are ignoring the seriousness of the problem
Election problems in Craven County due to a software glitch
More information surfaces on Carteret County problem
© 2004 North Carolina News Wire
November 18, 2004 (AP) Gastonia The number of recorded votes and voters don't match in more than half of the precincts in Gaston County, more evidence of Election Day problems.
The number of ballots cast and the number of voters signed in by poll workers were different in 29 of the county's 46 precincts, according to a review done by The Charlotte Observer. The discrepancies raise the possibility that some people could have voted twice, while others might have cast ballots that were not counted.
"What you' re sharing with me is highly irritating," said Gaston Elections Director Sandra Page, who said she had not known of the problems. [Typical response of election officials.]
She blamed the discrepancies on precinct workers who failed to follow instructions. "We' re sitting here victims of what happens out there, and we get the blame for it," she said. [Kill the messenger and definitely don't accept responsibility.]
At H.H. Beam Elementary School in south Gastonia, poll workers signed in 42 more people than the number of ballots recorded on the electronic voting machines. [Notice that absolutely no consideration is given to the possibility the voting machines may have malfunctioned as they have in numerous other locations. Had to the poll workers!]
At nearby Ashbrook High School, the problem was reversed: The precinct collected 81 more ballots than the number of names recorded by poll workers. [In numerous other locations electronic voting machines have reported thousands more votes than were actually cast. But here it was the poll workers at fault.]
Countywide, in precincts where names exceed ballots, there were about 100 more names than ballots. In precincts where ballots exceeded names, there were about 400 more ballots than names. A total of almost 64,000 ballots were cast.
Gary Bartlett, director of the state elections board, said Thursday that board employees would travel to Gastonia on Monday to examine the situation.
"Whenever something is not reconciled, it's troublesome," he said.
The county failed to include more than 13,000 votes nearly one in five of those cast in its unofficial election results. Some early votes were recorded improperly because of errors by poll workers. And a recount Tuesday of Gaston's optical scan ballots produced 75 additional votes in a pair of statewide races.
Page said she remained convinced that the county's reported election results are accurate. [Bury one's head in the sand and hope the problem will go away.] But she acknowledged that it will not be possible to know the name of every person who voted on Election Day.
When voters reach the front of the line at a Gaston County precinct, they sign next to their names on the voter roll, and a poll worker records it again in a second book used to track the total number of voters.
The number of recorded names exceeded the number of ballots cast in 15 precincts, the Observer found. In 10 of those precincts, the difference was five names or fewer.
Poll workers at some of those precincts reported that a small number of people left without voting after their names were recorded. And Page said some ballots also might not have been recorded because people pulled their voting cards too quickly from the voting machine.
Some of the largest differences may reflect only a failure by poll workers to record the names of people who voted curbside because of a disability. Those names can be retrieved from signed affidavits.
But many of the differences cannot be explained by curbside voting, Page acknowledged.
The Gaston Board of Elections said it was concerned by the findings.
Board member Richard Jackson said he would ask Page for a report documenting the number of voters and ballots cast in each precinct.
© 2004 by Binyamin Appelbaum, Charlotte Observer
November 16, 2004, Gastonia Gaston Elections Director Sandra Page said for the first time Monday that her office is investigating why more than 13,000 votes were excluded from the county's unofficial election results.
Page said the investigation so far pointed to an interrupted download as the likely cause of the exclusion of about 12,000 early votes. She said human error by poll workers probably resulted in the omission of 1,200 votes from a Dallas precinct.
The Gaston elections office has faced mounting criticism for its initial failure to count the votes and because almost a week passed after the November 2, 2004, election before the errors were corrected. The errors were caught before the county submitted its official results to the state, and they did not change the outcome of any local race.
Monday was Page's first day at work since last Tuesday she has been home sick and she said she moved quickly to find out what went wrong.
Page said she has asked the company that manufactured Gaston's vote-counting machines, Diebold Election Systems of McKinney, Texas, to review the operation of its equipment. A spokesman for Diebold confirmed that the company is doing so.
The county pays a technician from Diebold to operate its systems on Election Day. That person was in charge of transferring early votes from electronic storage to the counting computer. Diebold believes the transmission was interrupted, said spokesman David Bear. [Note that the election is run by an unnamed "technician" of whom nothing is known and over whom citizens have no control. This "technician" could be Osama bin Laden or one of his henchmen for all we know!]
"It's understood that it was an interruption, and now the question is why didn't we catch it," Bear said.
Page said she was also planning to speak with poll workers at the Dallas precinct about the possibility that they transmitted their votes incorrectly to the elections office on Election Day.
The laptop computer used to transmit results from the Dallas Civic Center recorded only a single vote from each of the precinct's voting machines, Page said.
The votes are transferred from the machines to the laptop by disk. Page believes the error occurred because poll workers removed the disks from the laptop too quickly.
"These aren't computer people," she said.
December 15, 2004 Gaston County Elections Director Sandra Page is resigning effective February 15, 2005, following an all-day, closed-door meeting Wednesday of the county Board of Elections.
And in a surprise move, the chairman of the elections board, Tony Branch, announced his resignation effective immediately. He said he felt he bore some of the responsibility for the public's loss of confidence in the Gaston election process.
Page, who with her lawyer was present for Wednesday's meeting, remains on medical leave for unspecified problems including chronic sinusitis. She will be using banked vacation and sick time, and is not expected to return to work before her resignation is effective.
Computer scientist Dr. David L. Dill of Stanford University
AP Newswire, November 13, 2004
November 26, 2004 North Carolina's election problems will not be that easily remedied. This year's disaster shows that many election workers are in over their heads.
Problems with voting machines, central tabulators using outdated and secret software, registration confusion, poll worker training, provisional ballots and absentee ballots are not easily remedied.
Add to all this the lack of a voter-verified paper ballot and you have no disaster recovery plan.
This is the case with more than 40 counties using touchscreen or "dial a vote" machines. The security of their votes depends on the software, source code and hardware of the voting machines. Election workers' ability, or lack thereof, to operate and troubleshoot the machines can affect the security of the votes as well.
Lost: 4,500 votes in Carteret County paper ballots verified by voters and retained by the election officials would have saved these votes.
Omitted: an entire precinct of 1,209 votes in Gaston County.
Missing: 12,000 more votes in Gaston County not reported. The election director hired a voting machine technician to upload the county vote totals and did not oversee the process.
Bamboozled: Guilford County bought vote-tabulating software that used outdated technology and with insufficient vote storage. As a result, Guilford County's public vote totals for president were off by 22,000 votes.
More votes than cast: Craven County reported 11,283 more votes for president than cast, voting with the same software as in Guilford County.
The State Board of Elections has relied on the advice of voting machine salesmen and turned a deaf ear to the good advice and warnings of computer scientists.
Voting machine salesmen gain access to some election officials via a private organization called the Election Center. This organization's mission is to educate and inform election officials, yet it admits to accepting money from voting-machine companies. The Election Center hosts conferences for election officials at which salesmen provide parties, prizes and even a dinner cruise on the Potomac. North Carolina's director of elections, Gary Bartlett, sits on the board of directors of the center.
Continued computer breakdowns and miscounts prove the need for a voter-verified paper ballot. This is not a receipt but a paper printout of the ballot, to be verified by the voter and kept by the election officials in case of recount, audit or computer breakdown.
The State Board of Elections can do the right thing by consulting computer scientists to recommend real requirements for our voting systems. It should also allow sufficient time for a thorough review by outside experts, to ensure that North Carolina's voting system is the most secure and trustworthy in America.
Joyce McCloy is coordinator of the North Carolina Coalition for Verified Voting.
© 2004 Sue Book, New Bern Sun Journal
November 05, 2004 A systems software glitch in Craven County's electronic voting equipment is being blamed for a vote miscount that, when corrected, changed the outcome of at least one race in Tuesday's election.
Then, in the rush to make right the miscalculation that swelled the number of votes for president here by 11,283 more votes than the total number cast, a human mistake further delayed accurate totals for the 40,534 who voted.
The glitch occurred Tuesday night as absentee ballot totals for one-stop early voting at three Craven County locations and ballots mailed-in were being entered, said Tiffany Miller, Craven County Board of Elections director.
The Elections Systems and Software equipment had downloaded voting information from nine of the county's 26 precincts and as the absentee ballots were added, the precinct totals were added a second time. Precincts affected were Havelock East, Havelock West, River Bend, Cove City, Ernul, Fort Totten, Grover C. Fields, Glenburnie and West New Bern.
An override, like those occurring when one attempts to save a computer file that already exists, is supposed to prevent double counting, but did not function correctly, Miller said.
"I have redone every (personal electronic ballot) completely and am adding the absentees," she said early Thursday. "Every precinct was redone."
The second set of incorrect numbers came when the total for one of the batches of absentee ballots was not included in the first manual recount.
"That's why we have a week before the votes are official, so if we do find problems we can get them straight before the votes are certified," said Miller, who was in her office before 8 a.m. Thursday, hand-crunching the numbers retrieved from the voting machines.
[Spend millions on electronic voting machines and then the numbers have to be added by hand anyway. Progress!]
New numbers put incumbent Commissioner Leon Staton in front of Republican challenger Tony Michalek for the Craven County Board of Commissioners District 5 seat. It does not appear that any other local races will be affected as a result of the malfunction, even with the results of about 1,000 provisional votes still to be entered Nov. 9.
Joined by her staff when the Craven County Administration Building opened, Miller already had been discussing the problem and remedies with Craven County Board of Elections Chairman Gloria Stanley, Electronic Systems and Software local representative Owen Andrews and North Carolina State Board of Elections Director Gary Bartlett.
Other than the cranking sound of numbers adding and an occasional ring of the phone, the office was dead quiet as they scrutinized the emerging tapes and hand-posted them on notebooks to be added by hand.
Craven County Commissioner Renee Sisk, a Republican who would be in the new majority on the board with Michalek's election, came by to assess the situation, followed by Craven County Manager Harold Blizzard, then Craven County Republican Party Chairman Steve Tyson, who both came in to inquire about the problem and possible solution. Then a national Republican political consultant who had watched the returns and had some concern that other precincts were involved in the problem entered the office.
Andrews arrived and made contact with the ES&S home office in Omaha, Neb.
"What she is dong now is the failsafe way to make sure we get it right," said Andrews, of Owen G. Dunn in New Bern.
"When a voter casts the vote, it stays in the memory of the machine, which has redundancy as a safeguard," he said.
While there is no paper ballot, Andrews said "(Miller) has a paper trail. She can print as many paper vote tallies out of the machine as she'd like." [But what if the DRE is misprogrammed, or fails to register the ballot at all?]
That information remains in the voting machine until the election is decided and it is deliberately [by a hacker perhaps?] removed.
Andrews will work with the manufacturer [ES&S], Miller and the elections board to correct the problem to ensure it will not happen again, but said "it really has nothing to do with the integrity of the vote as cast or counted." [If they keep their heads buried in the sand the problem will go away.]
"We will produce an honest outcome by the time we canvass Tuesday," said Craven County Board of Elections member Walter Leake, who was one of the last to stop by the elections office Thursday. "The poll books will match the machine." [One would have a great deal more confidence if we could be assured the poll books matched the ballots as they were cast rather than matching whatever is in a secret machine.]
Election difficulties also were reported in a number of other North Carolina counties, including nearby Carteret, where 4,530 early votes were irretrievably lost.
Sue Book can be reached at (252) 638-8101 ext. 262 or sue_book@link.freedom.com.
© 2004 Charlotte Observer
November 10, 2004 Double and triple counting votes. Not counting votes. Losing votes. Election Day is gone, but the election's still not over in several North Carolina counties, including Mecklenburg. Troubling Election Day flubs have kept the vote counting going.
The biggie was in Carteret County, where more than 4,500 early ballots were lost because the maker of the county's voting machines did not update the computer software. The company said the machines would hold 10,000 votes; in fact, they were programmed to hold just over 3,000. The mistake literally disfranchised voters and could cause not only a new election in Carteret, but maybe even a new statewide vote in state races where those lost votes might determine the outcome.
A whopping mistake discovered Tuesday in Gaston County could negate the need for such a revote, though. During their canvass of last week's results, Gaston officials found about 12,000 votes that had not been counted. They were mostly early and absentee votes that officials said were in their computer system but not released from the machines when other votes were tallied. The number of uncounted votes is larger than the margin of victory for state commissioner of agriculture and superintendent of public instruction and could change the outcome of those races.
These mistakes don't inspire confidence in the voting process. Neither does what happened in Mecklenburg County. Elections Director Michael Dickerson said human error caused ballots from at least seven machines used for early voting to be counted twice, and seven others not to be counted at all. [The EJF is skeptical of the "human error" excuse. Electronic voting machines are supposed to prevent such errors and would with proper programming.] The resulting error, spotted by county Republicans, left two county commissioner seats unresolved for a week. Challenges are still likely even with the official tally.
Election night results are always unofficial. Audits and verifying of vote counts are standard parts of the process. Errors are caught at that time that usually won't change race outcomes.
This year's errors and problems could provide valuable lessons for elections officials. Better election preparation, training and staffing could have helped. The factors officials say contributed to problems high voter turnout, heavy early voting, the state's long ballot, a state law prohibiting elections officials from counting early votes or absentee ballots before 2 PM on Election Day were known or could have been anticipated well before the election. And in reviewing what went wrong, officials described situations where workers seemed to be juggling too many tasks. More workers at key times or better trained workers could help get the vote right the first time.
Early voting clearly has altered North Carolina elections. The security of votes should not be compromised, but the heavy early turnout this year underscores the need for sensible changes in our state's elections process. A state law prohibiting opening voting machines earlier than 2 PM. on Election Day appears to be a hindrance that lawmakers should reconsider. Even if future early voter turnout declined, the extra time still would be valuable in making election night less chaotic.
Mr. Dickerson is right. Human error is inevitable. Some problems could not have been anticipated or avoided. But some could. Those we can prevent or fix, we should.
November 18, 2004 More information on the early-voting tabulation problem in Carteret County, North Carolina using the Unilect Patriot voting system has surfaced after some additional testing ("Warning light came on, state tests reveal"). You'll recall that the central problem here was that the system continued to allow recording of votes after its memory was full which resulted in 4,438 out of 7,536 early ballots to be lost.
It turns out that the Patriot system's central controller (it has a central controller and a group of daisy-chained voting terminals) displayed an error message, "Voter Log Full," until the controller was reset for the next voter. However, the display continued to increment the number of ballots cast. Poll workers are not experts so I'm sure that they took the incrementing of the number of ballots cast to be evidence that votes were still being recorded. Even technical experts would admit that a message like "Voter Log Full" doesn't sound critical on its face; it sounds like some audit log that records when ballots are cast is full, not that the machine is no longer recording ballots. [Note this is sophomoric programming problem that testing should have found long before this system was used in an actual election. Where was the Independent Testing Authority that we hear touted so often for voting machines?]
This is a great illustration of the dangers with paperless DRE voting, or, at least, voting without robust auditability. If this had been an error with an optical scan system, there would still be paper records that could be recounted. What should have the Patriot system have done? It arguably should have not allowed a single vote to be cast once full, and should not have allowed poll workers to override the error message.
| EJF Home | Where To Find Help | Join the EJF | Comments? | Get EJF newsletter |
| Vote Fraud and Election Issues Book | Table of Contents | Site Map | Index |
| Chapter 10 Voting Problems In The 2004 Elections |
| Next Election Problems In Florida 2004 |
| Back Electronic Voting A Mess In March 2004 California Primary |