Email from Acting Registrar Dean Logan Says Poll Workers Entered Incorrect Ballot Codes into E-Voting System
By Brad Friedman on 6/12/2008, 9:01pm PT  

Last week I reported in great detail that four out of the twelve votes I'd cast on my ballot on L.A. County's e-voting system, during California's state primary on June 3rd, were flipped to candidates that I hadn't voted for. The ES&S InkaVote Plus system printed my ballot incorrectly.

I was voting on the audio ballot voting system meant primarily for blind and visually-impaired voters. Had I not been able to inspect my ballot visually after it had printed --- for example, had I been blind --- I would never have known that I was about to cast votes for candidates that I had not selected.

Because the "disabled-accessible" e-voting system at my own precinct was not working, again, for the second election in a row, I was told I'd have to cast a provisional ballot on the "disabled-accessible" e-voting system at the other "precinct," in the same polling place, in the same room, as my own "precinct."

The machine had been down all day, and nobody had come out to replace or repair it, even though the county had been called early in the day and told about the problem. Last election, on Super Tuesday, the machine was also broken, and none of the poll workers would touch it, due to exposed wires.

This time, despite asking to vote in a specific party's open primary, I was given a non-partisan audio ballot to compound the fiasco when I tried to vote provisionally on the other "precinct's" e-voting system that was up and running, even though not a single voter had used it in either this, or the previous election.

The voting went smoothly enough --- even if it took 20 minutes to move through the short audio ballot, shorter than usual because it was a state primary, and because I wasn't allowed to vote in partisan races for some reason --- at least until it misprinted my selections, which I had confirmed as correct while actually moving through the audio selections.

Luckily, after noticing that the e-voting system had printed my ballot incorrectly, I was allowed to VOID that ballot, and this time chose to vote on a regular, hand-marked paper ballot instead. My hand filled in the ballot far more accurately than the computer did.

That night, I alerted officials at the CA Sec. of State's office and L.A. County's acting Registrar of Voters, Dean Logan, about the problem, and the machines and ballots in question were isolated for testing. Whether someone who wasn't a loud, muckracking journalist, specializing in issues of election integrity, would have had that kind of quick service, I can't tell you.

Logan promised the machine would be tested, and the SoS office promised to follow the testing closely.

Earlier this week, Logan sent me an update via email, with photos, on what the investigation had so far found. The problem, Logan wrote, is being attributed to the "human error" of the poll workers who, he says, entered incorrect information into the e-voting system when setting up my ballot for me to vote.

Logan's complete email, detailing the findings of the cursory investigation to date, follows below, along with my email in response to it. As you'll see, the all-too de rigueur "human error" excuse does not sit well with me...

From: Dean Logan []
Sent: Monday, June 09, 2008 1:14 PM
To: Brad Friedman
Cc: Evan L. Goldberg; Lowell Finley
Subject: RE: Report on flipped InkaVote Plus votes

Mr. Friedman --

Brief update: As indicated last week, we have identified and secured the InkaVote Plus devices and other voting materials from your polling place at our Election Operations Center (photos attached).

Based on your report and photographs, our staff has conducted an initial assessment and review of what occurred. Our preliminary assessment is that what occurred was the result of human error in selecting and entering the ballot code and party selection when setting up the Audio Ballot Booth for your use. As a result you were presented a ballot with a ballot group designation that did not match the voting location/precinct you were in. This would account for a different rotation of candidate names from the rotation listed on your sample ballot and in the voting booths at your poll site. Using the photographs in your report, we were able to identify the incorrect ballot code printed on the VOIDED audio ballot slip and to match your selections to the rotation associated with the incorrect ballot code that was used. The incorrect ballot code and party identification also likely explain the reason you were not presented the partisan contests you requested as provisional ballot status should not affect your ability to cast a partisan ballot.

Had your voted audio ballot been inserted into the Precinct Ballot Reader (PBR) after it printed from the Audio Ballot Booth, it would have been rejected based on the incorrect ballot group identification and further action should have been taken to ensure the correct ballot was issued. In your case, however, since you were voting provisionally, procedures would have called for your ballot to be returned in a provisional ballot envelope for confirmation of voter eligibility. Once verification occurred, the audio ballot slip would have been duplicated to the correct ballot group format for the precinct in which you are registered to vote. Properly conducted, that process would have resulted in your selections being recorded and counted as you intended.

While preliminary indications are that this was not a system failure - rather, it appears to have been a process and training deficiency - it is clearly an issue that must be addressed. The complexity of the situation is further complicated by the variables and multiple party ballots associated with our Primary. Many of these factors will not apply in the November General Election. Your experience shows, however, that they do need to be improved for future primaries in which the InkaVote Plus system is used. Once our review and investigation is complete, we will demonstrate what occurred and show how the instructions were intended to be carried out.

This email is intended as an update only. A detailed report of these findings and an explanation of their impact will be prepared as part of our continuing review and investigation into the issues raised in your report. Additional review and analysis needs to be completed to ensure we have been thorough and complete in our investigation and findings. We will keep you informed of our progress.

I hope this information is helpful.

-- Dean
Acting Registrar-Recorder/County Clerk
Los Angeles County

My response to Logan's letter follow below...

From: Brad Friedman
Sent: Monday, June 09, 2008 1:14 PM
To: Dean Logan
Cc: Evan L. Goldberg; Lowell Finley
Subject: RE: Report on flipped InkaVote Plus votes

Dean -

The information provided in your update/response, including photos, is very much appreciated. I look forward to the final determination and confirmation of what went wrong in the L.A. County voting system when I voted during the recent primary. It's especially troubling, of course, because, according to your description of the problem, it would never have been noticed by a vision-impaired voter for whom the system I voted on was generally supposed to serve.

The only reason I was able to notice the failure was because I was able to visually (and very closely) examine the computer-printed ballot to determine it had mis-printed my intended votes. If I understand your explanation, then it seems my incorrectly printed ballot --- since I was forced to vote provisionally --- would not have been noticed by anyone to have included four candidates for whom I did not vote.

While you suggest the failure, attributed to "human error", would have been caught had the ballot gone through the normal precinct-based op-scan procedure, mine would not have been noticed to have been incorrect because I was forced to vote provisionally, despite appearing accurately on the voter rolls and despite my own "precinct" being across the room from where I was forced to vote.

If your description of the problem is accurate, then it sounds like poll workers at any of L.A. County's more than 4000 precincts would have the ability to enter in codes for virtually any precinct (and/or ballot rotation) when they set up a voter to begin voting on the ES&S audio ballot system. If that is the case, it raises a number of serious questions and concerns.

  1. For a start, it seems like this is a recipe for fraud. Particularly when the voter in question would usually be a visually-impaired orblind voter and would have no way to know that a poll worker has set them up to vote in a way that the audio tells them they are voting one way, but the computer-printed ballot results in different selections, unknown to the voter.
  2. Why was I forced to vote provisionally, simply because the machine at my "precinct" --- across the room, in the same polling place --- was not in service all day? Since it was in the same room (polling place), and the federal Help America Vote Act (HAVA) only requires a single "disabled-accessible" voting device in each polling place, and since the machine could have had a code entered to match my real "precinct" (across the room), why should I, or any other voter have to vote provisionally in such a case? As you know, provisional ballots are not counted at the same rate as "normal" ballots, and as further information from your note mentions, such ballots are then (incredibly) "duplicated" by hand, by some unknown poll worker at county headquarters before they are ever even counted.
  3. Since this was the second election in a row where the "disabled-accessible" ES&S voting device at my precinct was out of service the entire day, and since nobody used any of those machines at my polling place in either of those two elections other than me, wouldn't it be a better use of tax payer money to have a single machine serving all precincts at a single polling place? While you would still be following the letter of the law, the second machine could then be used for a backup at the polling place, since obviously these things break down frequently (or never work at all from the moment they were set up, as was the case in my own precinct in two elections in a row)? Doing so would help to ensure that disabled voters would have a better chance of being able to cast their vote without being forced to do so provisionally.
  4. I am exceedingly disturbed to hear that --- had my provisional ballot gone through as it was supposed to, and had "verification" of my eligibility to vote occurred --- someone at election headquarters, who I do not know, and who I could not oversee to assure they were doing it correctly, was going to "duplicate [my ballot] to the correct ballot group for the precinct in which [I am] registered to vote."I hope you'll forgive my dubiousness that whether "Properly conducted" or not, and despite your assurance that "that process would have resulted in your selections being recorded and counted as you intended," there is absolutely no way for either me or you to know that as a fact. Nothing personal, but no, I do not trust a poll worker to "duplicate" my ballot properly by hand and I should not have to do so. Such a system is rife with the possibility for "human error" and/or fraud and, frankly, I take it as a personal affront that someone would "duplicate" my ballot before it is ever counted. I hope you will take a very serious look at that appalling practice prior to November, as it seems to me to be an outrageous and offensive practice, on a number of levels.

I look forward to your thoughts on the above, and I thank you again for the seriousness with which you have approached this unfortunate and disturbing matter. I also look forward to participating in the actual testing of your theory as to what went wrong as I requested previously.

On a final note, you should know that the poll workers at my polling place were exceedingly hard working, helpful and very engaged in the process. I'm troubled that, once again, it seems this failure may ultimately --- and inappropriately, in my opinion --- be attributed to the "human error" of the poll workers.

Time and again, in virtually every one of the thousands of instances of voting system failures similar to the one I encountered, and which occur with frightening regularity across the country, such failures are attributed to 'lack of' or 'poor training of poll workers.'You described it as a "process and training deficiency."

As I have pointed out time and again --- including in the testimony [PDF] I was recently invited to give to the U.S. Election Assistance Commission (EAC) --- we need to implement simple, transparent voting systems that do not require patriotic, over-worked and under-paid poll workers to also be rocket scientists. We need to stop blaming them for the failures of the systems we have forced them to use. I would suggest that it's not the dedicated poll workers who are under-trained, but rather the voting systems, such as the one we use in Los Angeles County, that are overly-complicated.

If there is a "deficiency" anywhere, might I suggest we look firstat the voting machine companies --- in this case, ES&S --- who have provided us with voting systems that fail miserably, time and again, to meet the simple, specific, mission-critical purpose for which they were supposedly designed, at the cost of billions of L.A. County, California and federal tax-payer dollars.