Sunday, February 26, 2006

One Hundred Thousand Alleged Florida Voting Machine Errors in 2004 Presidential Race

La Times Article is Here
The internal logs of at least 40 Sequoia touch-screen voting machines reveal that votes were time and date-stamped as cast two weeks before the election, sometimes in the middle of the night.

Black Box Voting successfully sued former Palm Beach County (FL) Supervisor of Elections Theresa LePore to get the audit records for the 2004 presidential election.

After investing over $7,000 and waiting nine months for the records, Black Box Voting discovered that the voting machine logs contained approximately 100,000 errors. According to voting machine assignment logs, Palm Beach County used 4,313 machines in the Nov. 2004 election. During election day, 1,475 voting system calibrations were performed while the polls were open, providing documentation to substantiate reports from citizens indicating the wrong candidate was selected when they tried to vote.

Another disturbing find was several dozen voting machines with votes for the Nov. 2, 2004 election cast on dates like Oct. 16, 15, 19, 13, 25, 28 2004 and one tape dated in 2010. These machines did not contain any votes date-stamped on Nov. 2, 2004.

You can find the complete set of raw voting machine event logs for Palm Beach County here:
Note that some items were not provided to us and are ommitted from the logs.

The logs rule out the possibility that these were Logic & Accuracy (L&A) test results, and verified that these results did appear in the final totals. In addition to the date discrepancies, most had incorrect polling times, with votes appearing throughout the wee hours of the night. These machines were L&A tested, and the L&A test activities appeared in the logs with the correct date and time.

According to the voting machine assignment log, these machines were not assigned to early voting locations. The number of votes on each machine also corresponds with the numbers typical of polling place machines rather than early voting.

Many of these machines showed unexplained log activity after the L&A test but before Election Day. In addition, many more machines without date anomalies showed this log activity, which revealed someone powering up the machine, opening the program, then powering it down again. In one instance, the date discrepancy appeared when someone accessed the machine two minutes after the L&A test was completed.

Voting machines are computers, and computers have batteries that can cause date and time discrepancies, but it does not appear that these particular discrepancies could have been caused by battery problems.

The evidence indicates that someone accessed the computers after the L&A and before the election, and that this access caused a change in the machine's reporting functions, at least for date and time. Such access would take a high degree of inside access. It is not known whether any other changes were introduced into the voting machines at this time. As learned in the Hursti experiments, it is possible for an insider to access the machines and leave no trace, but sometimes a hasty or clumsy access (such as forgetting to enter a correct date/time value when altering a record) will leave telltale tracks.

For another example of time discrepancies, see the
Volusia County poll tapes

Approximately 4,000 votes were cast on these machines. The vote pattern and activity pattern appears to be identical to typical patterns found on Election Day -- All votes on the discrepant machines were spread over a 12-hour period, the length of time the Florida polls are open.

A member of the Palm Beach County electronic voting technical committee asked for the names of the technicians for Palm Beach who had access to the machines during that time, but the IT person, Jeff Darter, remained silent and never answered the question.

The Palm Beach County Supervisor of Elections, Arthur Anderson, said that his staff had looked into the problem and that the votes were normal, it's just that the dates somehow changed.

Other anomalies (Anomaly info)

"Card Stuck" error: Occured at least 70,000 times.

The logs show that these cards were placed in the machine (which normally "swallows" the card like old-fashioned ATM machines, holding the card inside until the voting activites are complete, then ejecting it). The logs show that the card was authenticated, indicating that the machine believed the card was valid and had retrieved the appropriate ballot. Just before the vote was cast, the "card stuck" error appeared.

According to Michelle Shafer, who is now the spokesperson for Sequoia Voting Systems, a card stuck error stuck error appears "any time an activation card makes contact with the activator in the electronic voting unit and comes back out. This happens for the following reasons:

• A voter does not push the card all the way in so it comes back out
• A voter inserts the card again after having already used it to vote once...
• A voter inserts the card backwards
• The card actually gets stuck in the machine (not typical)

Previously, a Sequioa rep attributed the card stuck error to jiggling the card while it is inserted, however that doesn't seem to hold up since it would take a pair of tweezers and considerable manual dexterity to jiggle it.

As to putting the card in backwards or upside down, the message that normally appears is probably the "invalid card insertion" message. Because of the high number of these errors, and because no reports were produced indicating that any voters had reported the card popping out while they were trying to vote, Black Box Voting recommended to Palm Beach that testing should be done to replicate the error, making sure that the explanation holds water and that there is no adverse impact on the vote.

A member of the Committee asked whether a testing day could be set up, but Jeff Darter again sat silent, and despite some prodding, no such testing appears to be on the horizon.

AC Power Off Incidents Any of us who use computers know that it is not a good idea to yank the power from your machine while you are entering mission-critical data, especially without a backup. (The Palm Beach voting machines lack voter verified paper trails.)

Dozens of voting machines were turned off during the middle of the election while the polls were open. Machine # 6359 in precinct 1036 was powered down 128 times during the election.

Other power-related issues included "Main Battery not charging" and "backup battery too low".

"Unknown event" messages

A handful of machines showed "unknown event" messages, apparently of different kinds. This is an interesting error message, since the FEC guidelines frown on undefined exceptions. What is the point of having an error message if you don't reveal anything about what the error is?

Machine number 5875 in Precinct 1077 showed two different "unknown errors," listing them as "unknown error 219" and "unknown error 220."

auto-act election info bad and "auto-act write ver fail" messages also show up in the logs, with the "election info bad" message appearing hundreds of times.

Card encryption bad and Card read fail errors also appeared, with the encryption error message the more frequent of the two.

Polls closed and results report messages would be expected to appear on every voting machine at the end of the voting cycle, but these revealed problems with poll worker training and procedures at the administrative/training level. Some logs reported one report printed, some two, three, four or five, and several not only had no results tape printed but showed no closing of the polls. (Closing the polls tells the voting machine not to accept any more votes).

Simulation not sim task was a message that offered no ready explanation, and another that left us wondering was the "Maint Official AT Report" error. Call a maintenance official? Maintain an official AT report?

SyErr 23: RC/AT Verify
and Sys Err 31: Vote Not Rec 1 imply a system error of some type, at least one of which would affect the vote.

EEPROM failure
Now this is a message you don't want to see on a voting machine. It happened a couple dozen times. It is somewhat akin to seeing a "hard disk failure" message on your computer -- not a good thing at all if you are in the process of entering critical time-sensitive data.

The logs indicate that poll workers used significantly different operating procedures from one place to another. One of the least desirable actions some poll workers were taking was to perform multiple calibrations on the machines during the day, every few hours.

Hundreds of records were simply missing, not provided at all, making it impossible to complete a formal audit.

After meeting with the authorities to determine protocols about releasing the detailed report, Black Box Voting plans to publish a detail report giving full log details on the 40 machines accessed by an insider.

Sequoia machines - locations

Sequoia touch-screens are also used in Pinellas County (FL), Riverside, San Bernardino and Santa Clara countis (CA), New Mexico, New Jersey, and formerly in Snohomish County (WA).

A sampling of Palm Beach precincts with votes appearing on wrong date/time

Precinct 3066 machine #8438 counted Oct. 15
Precinct 3068 machine #8490 Counted Oct. 28
Precinct 3086 machine #8316 counted Oct 14
Precinct 2132 machine #7441 counted Oct. 15
Precinct 6006 machine #7914 counted Oct. 14
Precinct 6018 machine #7877 counted Oct. 14
Precinct 4068 machine #8997 counted Oct. 16
Precinct 5142 machine #9724 counted Oct 18
Precinct 2072 machine #6848 counted Oct. 15
Precinct 4140 machine #9289 counted Oct. 17
Precinct 4084 machine #8101 counted Oct. 17