Sequoia E-Voting Machines Reporting Inaccurate Totals in New Jersey

Post-Super Tuesday Canvass Reveals Discrepancies Between Internal Printouts and Memory Cartridges on DRE Systems in Five Separate Counties

Same 'Tamperproof' Machines Recently Hacked by Princeton University Professor, Also Failed to Start Up Properly on Election Day...
Guest Blogged by John Gideon of VotersUnite.Org


The Newark Star-Ledger is reporting that New Jersey election officials have found a discrepancy in the state's Primary Election results as reported on the Direct Recording Electronic (DRE) voting machines used on Super Tuesday. Voter totals reported by the internal paper tapes on their Sequoia AVC Advantage DRE in a number of counties are failing to match up with totals found on the memory cartridges, used for both ballot definition and results storage, on the same machines, according to the report today...

As Union County Clerk Joanne Rajoppi tried to verify returns in this month's historic presidential primary, she kept coming up with errors for a handful of voting machines.

The numbers from the cartridges that print out vote tallies and the paper-tape backup within the machine didn't match. Rajoppi asked her colleagues in other counties to perform the same test, and similar problems were found in voting machines for Bergen, Gloucester, Middlesex and Ocean counties.

Problems with Sequoia's AVC Advantage systems also emerged early on the morning of Super Tuesday, forcing a 45 minute delay for the state's Governor, who was unable to cast his vote when the machines failed to boot up on Election Day. In February of last year, the same machines were hacked by a Princeton University Professor after he'd been able to purchase a number of the $8000 systems for just $86 apiece on the Internet.

The Star-Ledger today goes on to report some of the details on the latest failures now emerging in New Jersey during the state's post-election canvassing...

The errors involve voter-turnout totals. For example, both the tape and cartridge on one machine in Cranford show 225 votes were cast when Democrat and Republican votes are added.

The problem is, the tape and cartridge don't agree when the numbers are broken down by party, Rajoppi said. The tape counts 168 votes for Democratic candidates, while the cartridge shows 170. On the Republican side, the tape counts 57 votes, but the cartridge shows 55.

When Rajoppi discovered the problem, she called the vendor, Sequoia Voting Systems...

"Initially, when I called Sequoia they said it was an anomaly. And I said, 'Excuse me. It's not. It's an error,'" Rajoppi said.
Congratulations to Rajoppi for making that important distinction. These are not anomalies, glitches, snags, snafus or hiccups, they are failures by these expensive products and/or voting machine companies to perform as expected and well-paid.

Sequoia's spokeswoman, Michelle Shafer, says that the company is working with election officials to determine the cause of the problem. The Associated Press reports that a Sequoia technician has blamed the problem on a corrupted computer chip. That seems to be a stretch in logic because it would mean that chips failed in five counties that have all reported the same problem, and they all failed at the same time.

The BRAD BLOG previously reported problems on the same voting machines early on the morning of Super Tuesday in New Jersey, when the Advantage machines failed to properly start up, resulting in embarrassment when NJ's Gov. John Corzine was forced to wait 45 minutes before he was able to cast his vote. Reports of votes flipping from Hillary Clinton to Barack Obama were also reported that morning.

In February of 2007, we reportedon the same Sequoia Advantage machines having been hacked in five minutes by a professor at Princeton University, who had purchased five of the machines on the Internet from an on-line government equipment clearinghouse for $86 apiece.

The same systems were recently purchased by the state of New Jersey for $8,000 each.

"We can take a version of Sequoia's software program and modify it to do something different --- like appear to count votes, but really move them from one candidate to another," Princeton computer science professor Andrew Appel told the New Jersey Star-Ledger at the time. "And it can be programmed to do that only on Tuesdays in November, and at any other time. You can't detect it," he said.

At the time, an Election Integrity group had filed a lawsuit against the state, claiming the 10,000 AVC Advantage machines were in violation of state law since they were never properly certified as secure or reliable, as required by the state.

"There is zero documentation --- no proof whatsoever --- that any state official has ever reviewed Sequoia machines," Attorney Penny Venetis, co-director of the Rutgers Constitutional Litigation Clinic, argued at the time. "This means you cannot use them...These machines are being used to count most of the votes in the state without being tested in any way, shape or form."

"Had the machines been tested," the state Election Integrity advocates found, "they would have proved to be a hacker's dream."

Sequoia Voting Systems responded to the reports at the time by claiming their election products were "tamper proof."

Independent studies commissioned recently by the states of California, Ohio and Colorado have all found otherwise, as the testers were reportedly able to violate the security measures on Sequoia's DRE systems in a matter of seconds.

Additional reporting for this story by Brad Friedman

http://www.bradblog.com/?p=5718