Update Nov. 20, 2009. "The issue was a bug in the Dominion source code that caused the machine to hang while creating ballot images for certain vote combinations in multiple candidate elections." The bug was discovered during pre-election testing, and ballot configuration files were modified in some machines, freeing up enough memory to prevent the hang. But not all machines with the problem were identified, so the malfunction occurred on election day in those machines. Story

Update 11/6/09. Sequoia explained that there was an "as yet unknown problem in the connection to the county server", and when the transmission didn't go through clearly, the cartridges went into "review" mode. Story Archive

Update 6/5/09. The county is planning a special mock election in hopes of figuring out what keeps going wrong in reporting results from real elections. Story Archive

Update 11/12/08. Officials are concerned that irregularities may indicate larger problems in the District's election software. For example: In Single-Member-District 6B11, which houses the District of Columbia Jail, the final unofficial election results showed no precinct reporting and 5 undervotes. Story Archive

When two machines malfunctioned at Conklin Hall near the Rutgers-Newark campus, voters were turned away and asked to come back later after the machines were replaced. Story Archive

Update 10/03/08. State Representative Mary Brandenburg requested a hand count of her race, which is on the crease on some ballots. Both Sequoia and the elections office acknowledged that if the crease could cause the machine to count a vote in that race even if the voter did not vote, but the state declined to allow a hand count. The County Board said the crease could be used for a challenge. Story Archive Story2 Archive2

After a closed door investigation, the DC Board of Elections and Ethics "determined that one defective cartridge caused vote totals to be duplicated into multiple races on the summary report." They called it a "vote-reading error." Many questions remain. Sequoia is investigating. (DC uses Optech 3P Eagle scanners and Edge e-voting machines.) Story Archive Story2 Archive2 Q&A to a Board Member Q&A Archive

Update 9/12/08. "Industry specialists" don't believe a single defective cartridge could add thousands of write-in votes to multiple races across party lines. Sequoia suggests, instead, that it was static discharge or human error. Q&A to a Board Member Q&A Archive

Update 9/17/08. The cause of thousands of phantom write-in votes remains unexplained. The "defective cartridge" theory appears to have been wrong. Dan Murphy, elections board spokesman said that workers ran the same cartridges through the same program but arrived at three different results. The last set, he said, was correct, and that the rationale for that decision would be explained in their report. A hand-count of paper ballots from four randomly chosen precincts showed that none of them matched the machine tallies. "In Precinct 21, which is in Ward 2, the results for 19 of 70 candidates that appeared on the ballot were off by one or two votes." Sequoia has been subpoenaed for a hearing to be held Oct. 3. Story Archive

Update 10/2/08. Four (of 144) precincts were hand-counted. Discrepancies between the optical scanner count and the manual count occurred in multiple races in all four. Report pdf

Update 9/4/08. After counting all ballots by hand, the county found 2,700 of the 3,400 missing ballots. The county intends to do a "more formal" count. Archive

Update 9/4/08. A welcome report that 2,700 of the 3,478 missing ballots had been found on Thursday was replaced with the grim news that workers had recovered only 957 ballots. Outcomes of other contests are now being questioned. Story Archive Story2 Archive2 Story3 Archive3

Update 9/5/08. The search for the missing ballots continues, with county employees searched polling locations, and elections staffers continuing to inspect their records. A check of voter registration logs from the polls is under way. So far voter totals look to be closer to the original ballot count after last Tuesday's election. A machine count of the ballots Thursday night (Sept 4) turned up nearly 3,000 fewer ballots than the original vote total after the election-- almost 500 fewer than a hand count earlier in the day. Canvassing board appears to have given up. "We can't count any more," said County Commissioner McCarty.
Wennet (hand-count winner) is pressing to certify the hand count. Abramson (election-night winner) is pressing to certify the original results. Story Archive Story2 Archive2.
Many more stories on this debacle. Update 9/12/08 AM. County officials believe they have found the missing ballots "in the county's vote-tabulating center near West Palm Beach." Story Archive

Update 9/12/08 PM. New troubles. Now there are more ballots than were counted on election night. 110 ballots recorded on a cartridge from one election day precinct weren't included in the intial count or the recount. And 139 more were found in the elections office. So, the county found a total of 249 more than were counted in either count. Story Archive

Update 10/2/08. Wennet's team did a machine recount of 262 previously rejected ballots. When the 262 ballots were fed through two tabulators used during the recounts, they rejected different numbers of ballots. All of them should have been rejected. "On the first two tests of 160 ballots, the machines accepted three ballots as good votes. On tests on 102 more ballots that should have been rejected, the machines first accepted 13 ballots as good votes and then 90 on a second run." Story Archive Story Archive

Update: 10/7/08Excellent summary of events with new information about the scanners' inconsistencies discovered during testing.