County Registrar tells BRAD BLOG how e-vote and human error led to big gain for GOP candidate during canvass of tight race...
By Brad Friedman on 11/20/2013, 6:05am PT  

Sometimes it's a good idea to get a full explanation before these things become fodder in a contentious partisan legal election contest. So that's what we've tried to do. Happily, the General Registrar of Bedford County, VA was more than willing to help.

Last week, and the week before, The BRAD BLOG devoted quite a bit of coverage to the incredibly close Attorney General's race in Virginia. As of last week, the Democratic candidate Mark Herring was certified by the state's local voting jurisdictions (counties and cities), as the "winner" over Republican Mark Obenshain by just 164 votes out of more than 2.2 million cast.

The contest is, for now, in the hands of the State Board of Elections which will issue its own official official certification of results on November 25th, after which the candidate declared the "loser" is almost certain to ask for a "recount" and potentially file an election contest thereafter, depending on the outcome. (See the last section of this article for an explanation as to why we put quotes around the word "recount", especially in Virginia.)

During the week-long roller coaster canvass by jurisdictions across Virginia following the November 5th election, there were a number of minor adjustments to local tallies as county and city election officials checked and double-checked results printed by touch-screen and paper ballot optical-scan tabulation computers from Election Night and then adjudicated provisional ballots for tally and inclusion in the final results. While most of the adjustments made during the week following the election were relatively small, each was of great significance in a race this tight.

But there were three rather large changes to the results during the post-election canvass process --- two were in the Democratic strongholds of Fairfax County and the city of Richmond, and one was in heavily Republican Bedford County. All of the large tabulation adjustments were said to have been caused by various combinations of computer tabulator and human error.

Two of them, the ones which resulted in about 1,300 votes beginning picked up by the Democrat over his Republican rival in both Fairfax and Richmond, were covered and explained in the media in some detail. (See our coverage here of the discovery of the thousands of "missing" Fairfax votes and the eventual explanation the next day. The Richmond additions are described here.) The adjustments made during the canvass by heavily-Republican Bedford County, however, which resulted in a net pick-up of about 500 votes for Obenshain, received considerably less public explanation.

Given that all of these matters may be revisited once again in a "recount" --- and the inevitable legal challenges to go with it --- we thought it might be good to get the explanation for Obenshain's big vote pickup in Bedford County on the written record. Bedford's General Registrar Barbara Gunter was kind enough to reply to our queries last week on that point, offering her explanation for the known details of the computer and/or human errors that led to more than 700 votes in the AG's race being initially misreported on Election Night in Bedford County...

Error at Precinct 702

In a series of emails (all CC'd to the three members of the Bedford County Electoral Board), Gunter helped to clarify what went wrong in the two different precincts. It took some clarification and follow-ups to help us understand the problems correctly, but by the end, we're fairly sure we're clear on what seems to have happened in each. We've posted all of those notes back and forth below, in case they become useful for the record down the road or for #VAAG election geeks today.

The first problem in Bedford, known early in the process, resulted in the addition of 135 votes that were not included in unofficial Election Night results. It was caused by an electronic voting machine that malfunctioned after the close of polls at Precinct 702 at Liberty High School. In that case, the machine in question failed to print a results tape after at the end of Election Night. When the problem came to light, the machine was "sealed by the Officers of Election" (poll workers) at the precinct, according to Gunter, and brought back to election headquarters. The next day, "all Officers from the precinct joined our voting equipment vendor's technician in the Registrar's office to unseal and retrieve the results" on Wednesday morning, prior to the canvass.

"Results," she said, were "retrieved by removing the memory cartridge from the machine and generating a pdf of the results tape. The Officers then included that machine's results in with their completed Statement of Results and submitted it for review at canvass."

There is still no explanation for the technical failure of the touch-screen machine (#18628, for the record) that kept it from printing out results at the end of Election Night at Precinct 702 as it is supposed to. "After this election is certified and all challenges are settled, the technician will come back and inspect and test the machine before use in another election," Gunter said.

Election Day voters in Bedford County use 100% unverifiable Direct Recording Electronic (DRE, usually touch-screen) voting machines made by Sequoia Voting Systems, a company with a very troubled past and now owned by the Canadian firm Dominion Voting. It is strictly impossible to know if any vote cast on those DRE systems was accurately recorded as per any voter's intent. A "recount" will yield no more information about the accuracy of votes cast on them than we already have, by and large, as the process for "recounting" DREs, according to the VA election code [PDF], consists of "open[ing] the envelopes with the printouts and read[ing] the results from the printouts. If the printout is not clear, or on the request of the court, the recount officials shall rerun the print out from the machine or examine the counters as appropriate."

Given the extraordinary limitations and unverifiability of DRE voting systems --- similar ones are used in much of the state --- the results reported on the printouts at the end of Election Night are all that officials will have to go on during a "recount". So, establishing what is actually on those Election Night tapes --- whether they are accurate or not to how people voted is unknown --- is important to the "recount" process.

Error at Precinct 404

The second and larger problem, the one that resulted in a surprise 597 votes being added to the totals late in the canvass process, occurred at Precinct 404 in the St. Paul Baptist Church. That problem, said to have been discovered "in the process of keying the official results" into the State Board of Elections website, according to Gunter, occurred when "the Registrar noted a large discrepancy in the unofficial results" reported telephonically by precinct workers on Election Night and the "Statement of Results" signed and submitted by those poll workers thereafter.

"She then pulled her copy of the Statement of Results and compared the machine report tapes to both the official and unofficial results," says Gunter. The difference was 597 votes.

What happened? On Election Night, she says, after the polls are closed, a cartridge is placed into each machine at each polling place and an "Official Consolidation Report" is then created for each precinct. "That cartridge retrieves the results from machine one, then is placed in the auxiliary port of machine two to retrieve its results, etc. until the cartridge has been placed in every machine within the precinct. On the final machine, the operator selects a command to tally the results for all machines and produce a Consolidated Results Report tape that presents the accumulated totals from all machines and lists the ID number of all machines from which those totals were collected," Gunter explained.

That process, she says, occurred just as it is supposed to. But when the precinct workers called in the unofficial results to county headquarters they inadvertently read the results off of a single machine's tape, rather than the full "Consolidated Results Report".

"The consolidate tape is the one that should have been picked up to call the unofficial results to our office," Gunter says. "In error, they picked up the single machine totals tape from machine 18627 and reported those results to the office" instead.

"When completing their Statement of Results, they used the consolidated results tape and entered the correct set of numbers on their Statement of Results," she says. "The correct set of numbers is what the Electoral Board saw during their canvass," before noting the discrepancy with the unofficial results called in on Election Night.

When that discrepancy was discovered and announced, the Republican Obenshain picked up about 500 votes in what had been (until the other corrections in other counties eventually reversed it), his lead over the Democrat Herring.

"The important thing is that every vote was accounted for and properly reported in the official results," Gunter said in her initial email to The BRAD BLOG. "This experience attests to the importance of the canvass of election results. It accomplished its intended purpose, which is to double-check all numbers reported and to verify the accuracy of results."

* * *

The emails back and forth between The BRAD BLOG and Barbara Gunter, General Registrar of Bedford County, VA (cc'd to all three members of the Bedford County Electoral Board), follow below for the historic record. They have been very lightly edited for typos, readibility, and to remove some personal information...

From: Barbara Gunter
Sent: Tuesday, November 12, 2013 5:20 AM
To: Brad Friedman
Cc: Carol Lee; Charlie Walker; Charlie Walker; Janeta Dudley
Subject: RE: MEDIA REQUEST FOR COMMENT

Mr. Friedman,

Bedford County uses the Sequoia Edge touch-screen voting machines in all of our voting precincts.

On Election Night the Election Officials working each precinct phoned in their unofficial election results which were promptly posted to the State Board of Elections website. On Wednesday afternoon, the Electoral Board began their canvass of the election. This is a process in which the Statement of Results prepared by each precinct is compared to the result tapes printed from each machine used in the precinct. The purpose of canvass is to verify that numbers were reported correctly. After the canvass is completed, which was Friday afternoon around 1 p.m., the Official Results were then keyed into the State Board of Elections website. In the process of keying the official results, the Registrar noted a large discrepancy in the unofficial results reported from precinct 404 - St. Paul Baptist Church. She then pulled her copy of the Statement of Results and compared the machine report tapes to both the official and unofficial results. The unofficial results reported on Election Day were an exact match to the results tape from Machine 18627 rather than the "Official Consolidation Report" tape, which is a summary of totals from all machines used in the precinct. The results the Officers of Election reported on their Statement of Results were reported correctly from the "Official Consolidation Report" tape and are the numbers verified by the Electoral Board during canvass. The difference in votes reported on Election Day and after the canvass was 597 votes.

In precinct 702 - Liberty High School, the Officers were unable to retrieve results from Machine 18628. A machine technician had to retrieve the results from that machine. The results reported by the Officers on Election Day did not include the results from Machine 18628 but they were included on the Statement of Results those officers submitted to the Electoral Board for use in the canvass. The difference in votes reported on Election Day and after the canvass was 135 votes.

Comparison of unofficial results with official results has not been a part of the canvass process for us in the past; however, this has been a learning experience and will probably result in that becoming a step we add in future canvass procedures so that erroneous unofficial results can be identified and corrected more promptly on the public website for viewing election results. The important thing is that every vote was accounted for and properly reported in the official results. This experience attests to the importance of the canvass of election results. It accomplished its intended purpose, which is to double-check all numbers reported and to verify the accuracy of results.

Barbara Gunter
General Registrar
Bedford County

From: Brad Friedman
Sent: Tuesday, November 12, 2013 3:34 PM
To: Barbara Gunter
Cc: 'Carol Lee'; 'Charlie Walker'; 'Charlie Walker'; 'Janeta Dudley'
Subject: RE: MEDIA REQUEST FOR COMMENT

Barbara -

Very much appreciate the detailed response, as well as your diligence in checking, double-checking, correcting, etc.

Just so I'm sure to understand your explanation, allow me to confirm some of those understandings.

On the Precinct 404 (St. Paul Baptist Church) matter: Each machine at the precinct printed out its own results tape on Election Night. But the "Official Consolidation Report", which is presumably run courtesy of the Edge machines daisy chained together, seems to have left off the results from Machine 18627. The "missing" results from those machines, though reported correctly on Election Night after phone in, were subsequently changed during the canvass to match the (incorrect) "Official Consolidation Report"? And that inaccuracy was subsequently noticed and then corrected (after checking against the Election Night result tape of Machine 18627) during the final process of keying in the final official results to the SBE website.

Do I understand all of those points correctly? And, if so, do we yet have a cause or explanation for the "Official Consolidation Report" having dropped (or not included) an entire machine when it was created on Election Night?

On the Precinct 702 (Liberty High School) matter: A few questions here. You say "A machine technician had to retrieve the results from that machine." Presumably that was a Dominion/Sequoia tech? If so, how and when were they retrieved? Directly from the machine or from the memory card at a different time and place? Presumably it was after they had phoned in their totals, but also before the precinct turned in their "Statement of Results", so I'm guessing the results were pulled (somehow) either the night of the Election at the precinct, or the next morning before the precinct turned in their official "Statement of Results"?

Thanks again for helping me to understand all of this clearly, so I can report it to folks clearly as well, if possible. I've got a pretty savvy group of e-voting experts who follow my reporting, so I know they'll especially appreciate the clarity.

Brad

From: Barbara Gunter
Sent: Tuesday, November 12, 2013 1:16 PM
To: Brad Friedman
Cc: Carol Lee; Charlie Walker; Charlie Walker; Janeta Dudley
Subject: RE: MEDIA REQUEST FOR COMMENT

Brad,

Thank you for taking the time to understand exactly what happened so that it is reported correctly.

Please understand that machines that are daisy chained together share only electrical current, they never share any data. Each machine printed a results tape of the totals on that specific machine at the time that polls were closed. In precinct 404, there was no failure on any machine in the precinct … we have the results tape from each machine. The Election Officials then put a cartridge into the auxiliary port of each machine to create a consolidated cartridge. That cartridge retrieves the results from machine one, then is placed in the auxiliary port of machine two to retrieve its results, etc. until the cartridge has been placed in every machine within the precinct. On the final machine, the operator selects a command to tally the results for all machines and produce a Consolidated Results Report tape that presents the accumulated totals from all machines and lists the ID number of all machines from which those totals were collected. The consolidate tape is the one that should have been picked up to call the unofficial results to our office. In error, they picked up the single machine totals tape from machine 18627 and reported those results to the office. When completing their Statement of Results, they used the consolidated results tape and entered the correct set of numbers on their Statement of Results. The correct set of numbers is what the Electoral Board saw during their canvass.

As to precinct 702, the voting machine was sealed by the Officers of Election. Prior to canvass on Wednesday morning, all Officers from the precinct joined our voting equipment vendor's technician in the Registrar's office to unseal and retrieve the results. Results are retrieved by removing the memory cartridge from the machine and generating a pdf of the results tape. The Officers then included that machine's results in with their completed Statement of Results and submitted it for review at canvass.

Barbara Gunter
General Registrar
Bedford County

From: Brad Friedman
Sent: Tuesday, November 12, 2013 9:46 PM
To: Barbara Gunter
Cc: 'Carol Lee'; 'Charlie Walker'; 'Charlie Walker'; 'Janeta Dudley'
Subject: RE: MEDIA REQUEST FOR COMMENT

Hopefully, I'm getting closer to understanding this. Thank you again for taking the time to make this clear. A few more questions to clarify:

Precinct 404: So, it sounds like I misunderstood your original explanation about this precinct and Machine 18627. What actually happened is that the Election Official at the precinct gave the precincts results from Machine 18627 only instead of the combined "Official Consolidation Report" that included the sum total of ALL the machines used at the precinct that day. When ALL machines were included in the tally, another 597 votes go added to it. Do I finally have it right? (Relatedly, just out of curiosity, how many DREs are used at that particular precinct?)

Precint 702: Okay. Think I understand this one now. What error or error message occurred that kept them from printing out on Election Night?

Thanks again. This helps a great deal.

Brad

From: Barbara Gunter
Sent: Wednesday, November 13, 2013 5:53 AM
To: Brad Friedman
Cc: Carol Lee; Charlie Walker; Charlie Walker; Janeta Dudley
Subject: RE: MEDIA REQUEST FOR COMMENT

Brad,

Precinct 404 - Yes! You have it right. That particular precinct was provided with three programmed voting machines. The Election Officials set up and used two of the machines.

Precinct 702 - I don't have an error code. At the end of the day, the Officials remove a seal from the "Close Polls" switch cover and move that switch to the closed position which should then generate the machine results report. When the polls switch was moved to closed, the report did not display on the screen and provide a display option to print the required number of copies to attach to their Statement of Results. Since we did not know what the problem was, we did not "try" anything to correct the issue. The machine was sealed and the technician was contacted to retrieve the results. After this election is certified and all challenges are settled, the technician will come back and inspect and test the machine before use in another election.

Barbara Gunter
General Registrar
Bedford County

* * *

Previously related #VAAG coverage at The BRAD BLOG...

E-Voting Trouble Reported During Today's Elections in Virginia, New York, and Elsewhere [11/5/2013]

'Recount' in Virginia AG Race? Good Luck With That. (But, Perhaps You Can Help) [11/6/2013]

BREAKING: Thousands of Votes Discovered 'Unaccounted For' in Virginia AG Race [11/7/2013]

Tallying VA's 'Missing' Op-Scan Votes; And Other Surprises in the Incredibly Close VA AG Election [11/8/2013]

Down to the Provisionals: 55 Vote Margin (or less) Out of 2.2 Million Cast in Virginia AG Race [11/9/2013]

BREAKING: Democratic Candidate Takes Lead in Razor-Thin VA Attorney General Tally [11/11/2013]

Provisional Votes Tallied in Fairfax County, VA - Last Ballots Before Certification of the Razor-Thin VA AG Race [11/12/2013]

BREAKING: Democrat Mark Herring 'Wins' VA Attorney General Race by 164 Votes Before Final State Certification, Almost Certain 'Recount' [11/12/2013]

* * *
Please support The BRAD BLOG's fiercely independent, award-winning coverage of your electoral system --- now in our TENTH YEAR! --- as available from no other media outlet in the nation...