4 Out of 12 Ballot Selections Printed Out Incorrectly, Myriad Other Laws Violated
ES&S InkaVOTE Plus System Fails Miserably, Mars What Should Have Been a Simple Voting Experience During L.A. County's Very Low Turnout State Primary Election...
By Brad Friedman on 6/4/2008, 8:04pm PT  

-- Brad Friedman, The BRAD BLOG

As a fairly well-known Election Integrity journalist who has personally covered, for years, the myriad election woes of thousands (if not millions) of voters around the country who have tried to bring seemingly endless stories of votes flipped on e-voting systems to the attention of officials, these stories always continue to be remarkable to me, even if not to many others in the rest of the mainstream media.

It's even more troubling when one realizes that so little ever seems to be done in light of so many of these horror stories, as those very same failed systems are still deployed across the nation, with little or no modification to correct the mountains of documented problems even now, as we head towards an election likely to be of historic proportions this November.

What follows is yet another one of those stories, where a voter had vote selections flipped by the electronic voting system, such that candidates were chosen other than the ones intended to be selected by the voter, though no fault of his own.

But this time, the voter is me.

Though I've covered so many of these stories, it was nonetheless remarkable to see it happen before ones very eyes, as occurred yesterday when I voted here in Los Angeles during our very low turnout California state Primary election.

The ES&S electronic voting system that I used to try to vote on yesterday, ended up flipping a total of 4 out of the 12 contests and initiatives for which I had attempted to vote.

Right before my very eyes, the computer-printed ballot produced by the voting system I was using, incorrectly filled in bubbles for four of the races I was voting in. Had I not been incredibly careful, after the ballot was printed out, to painstakingly compare what was printed to what I actually voted for, I'd have never known my votes were being given to candidates I did not vote for.

Had I been a blind voter --- as the system I was using is largely intended for use by the disabled --- I would have cast my ballot without having a clue that a full 40% of the votes I'd tried to cast for various California Superior Court judges were flipped to other candidates...

After speaking late last night about the problem to Dean Logan, the current acting Registrar-Recorder for Los Angeles County (the country's largest voting jurisdiction) and officials from the CA Sec. of State's office, I can report the failed e-voting machine in question is now being quarantined for testing to try and determine what happened in this, just the latest in a mounting string of failures by voting systems made by ES&S, the country's largest supplier of voting equipment.

While the machine is being sequestered for examination --- and one wonders if the same swift action would have been promised to someone not as well known to both the Registrar's and Sec. of State's office --- we'll call that point the "good news" for the moment.

In addition to the stunning error-rate of the failed ES&S InkaVote Plus voting machine I had tried to use, my day at the polls yesterday additionally revealed an amazing number of apparent violations of federal law, the California state Election Code, and a number of local, Los Angeles County provisions.

Yesterday's election was the final one this year, prior to the very same election equipment being scheduled for use this November as we barrel towards what will undoubtedly be an unprecedented turnout for the General Election.

Here's what happened to me and my ballot, and what I witnessed yesterday at the polls...

TROUBLE FROM THE START

Given last February's now-infamous Super Tuesday "Double Bubble" ballot debacle when the knowingly-bad ballot design of former L.A. County Registrar Conny McCormack (who quit just a month prior to the election) resulted in some 50,000 uncounted paper ballots on the ES&S InkaVote system, reports of trouble, so far, from the polls out here yesterday have been few. No doubt, that comes as good news to McCormack's hand-picked successor, acting Registrar Logan, who had initially balked at counting those perfectly-countable, but uncounted "Double Bubble" ballots from the Democratic Primary election earlier this year. Eventually, he counted most of those ballots, while choosing to leave some 12,000 or more voters disenfranchised in that race, despite the vast majority of those ultimately-uncounted ballots being perfectly-countable, as we reported in great detail at the time.

His early response, so far, to the problem I had yesterday, has been decidedly much better, as he called me back within five minutes of my having called his office last night at around 9:30pm while they were still tallying returns after the 8pm poll close. Hopefully that follow through will continue so that we can determine definitively what went wrong when I voted yesterday.

The problem at my polling place was immediately apparent when I arrived to vote late in the day, only to find that, once again, the ES&S InkaVote Plus voting system for my precinct was down when I arrived. My polling place features two different precincts combined into a single room, each "precinct" being run at a separate table.

Last February, the InkaVote Plus system was also "out of order" when I went to vote, and the pollworkers there were all afraid to touch it for fear of being electrocuted due to exposed wires hanging out of it. I'd planned to write that matter up for The BRAD BLOG back then, until my attention was drawn to the astounding "Double Bubble" fiasco.

The InkaVote Plus voting system --- which is an assistive audio device, meant to meet the federal Help America Vote Act (HAVA) requirements to allow disabled voters to vote "in a private and independent manner" --- was broken yet again this time at my "precinct". Additionally, the optical-scanner, used with the InkaVote's regular, hand-marked paper ballots used by most voters, was also down, as it had been the entire day. Poll workers told me they had called the county, but as also happened last February, and even though turnout yesterday was far lower, nobody came out to service or replace the system the entire day.

Fortunately, the InkaVote system is a paper-based system, so voters can vote whether the scanner is up and working or not. Though they won't receive warnings about overvotes and such, at least they can cast a ballot on such systems, and hope that they later will be counted (and perhaps even accurately, but who really knows?) at the county election headquarters after the polls close.

The InkaVote Plus part of the system allows disabled voters (well, actually, only blind voters, read on) to make their selections and move through the ballot with buttons, as they listen to the ballot read to them via headphones. At the end of the process, a paper ballot is printed out with the voters' selections --- theoretically, the actual ones they voted for --- and that ballot is then run through the InkaVote's optical scanner and counted by the same means as the regular, hand-marked paper ballots used by everyone else.

Fortunately again --- given that the assistive device for my "precinct" hadn't worked the entire day --- as in February, there was not a single voter who asked to use either it, or the one that was (theoretically) working in the other "precinct" located inside the same polling place room.

HAVA allocated 4.8 billion federal tax dollars to encourage the "upgrading" of polling places, usually with electronic voting systems, in large part under the claim that disabled voters required such electronic devices to vote "in a private and independent manner" under the new law. And yet, in two elections in a row, not a single voter had used either of the two devices at my large, metropolitan polling place. As it turns out, given my experience in trying to use it, that's a very good thing.

Logan would tell me later that night, when I called him to tell him what had gone so terribly wrong, that while he didn't have definitive numbers at-hand, "the number of voters who used [the 'accessible' voting devices] last February, was very low."

I had told the poll workers that I wanted to try to vote on the InkaVote Plus device in order to learn how --- and if --- it worked. At first they thought I was kidding, but when they learned I wasn't, most of them became interested in helping me to do so, since they had never seen it in action. As it was the end of the day, and there were very few other voters who came in while I was there, most of the poll workers, including a very helpful precinct captain, oversaw the process as I tried to make it work.

Because my own "precinct's" InkaVote Plus was down, I was told, I'd have to vote on a provisional ballot at the other "precinct" across the room, if I wanted to vote on the assistive device. So I did.

ALL HANDS ON DECK

I filled out the information on the provisional ballot envelope and they sat me down to vote on the assistive device. After five minutes, or so, of several workers trying to figure out how to tell the system that it should begin my audio ballot, I was finally off and "voting".

Though it was a laborious process to listen to the entire ballot as it was read, trying to use it to create my ballot was generally easy enough for a fairly technically proficient guy like myself, even if it took about 20 minutes before finally finishing the rather short ballot, on which I had even skipped a number of races, before the device finally printed out my selections (incorrectly).

(As a small aside...Somewhat maddeningly, during the 2 or 3 minute preamble audio explanation of how the system worked, the instructions said I could hit "the up arrow button" to increase the volume "at any time". So I did. That resulted in the entire preamble instructions starting over again from the beginning. This time, as I listened to the complete instructions, I learned, as the very last instruction during the introduction, that adjusting the volume button would result in the instructions starting over from scratch. Nice planning, ES&S! Had there been a bus load of blind folks who showed up to vote at this polling place, I can only imagine the disaster there might have been as the line to vote would have been nearly endless.)

Luckily, unlike other voters who might need to use the device in order to vote "in a private and independent manner", I have use of my arms. So I was able to push the buttons on the machine to make my selections and move through the ballot as instructed. Had I not, I would have had to have asked for help from someone. While I could have told my helper which buttons to push for me without them easily being able to figure out who or what I was voting for --- they wouldn't have been able to hear the audio as I heard it on the headphones, and thus, wouldn't have known which button pushes were selecting which candidates --- I would not have been able to vote "independently" as HAVA's Sec. 301 mandates.

When I asked Logan about the fact that, unlike other "accessible" voting devices, the ones we use in the country's largest voting jurisdiction do not seem to allow such folks to vote as per federal law, he told me that "jurisdictions all over the country are dealing with that, in terms of accessibility."

"The post-HAVA environment has focused heavily on visual issues, blind voters," he explained. "I think it's widely recognized that the accessible devices out there are not equipped for every possible disability. I don't know that there's a certified accessible machine out there that deals with all of those possibilities."

But when I told him I understood there were voting devices for the disabled with "sip and puff" capabilities for voters with dexterity and other motor skills limitations, he said it was his understanding that "there are prototypes for those features, but to my knowledge they're not widely available on voting systems at this time."

In fact, a number of theoretically-"accessible" voting devices have such features, including the AutoMark, also distributed by ES&S, which are certified for use in California. It was Logan's predecessor, McCormack, however, who selected L.A. County's current "accessible" voting system. So, for the moment, we'll not hold the fact that L.A. uses systems that don't meet the federal standards against him.

John Gideon, of VotersUnite.org (who is also a frequent Guest Blogger at The BRAD BLOG), was surprised to hear that Logan was unaware of the currently available devices. "Unless Dean Logan slept through the sales speeches from the AutoMark vendors, at conferences held by NASED, NASS, and the Election Center, he certainly knows that the AutoMark system, and others, have binary switches that allow for voters without use of their arms to vote independently," he told me this afternoon.

Stephen Heller of VelvetRevolution.us, the Election Integrity watchdog coalition that I helped to co-found (Heller has also been an occasional BRAD BLOG guest contributor), pointed more towards this document [PDF] at the National Disability Rights Network which notes that "three voting systems on the market --- AccuPoll, AutoMARK, and eSlate machines --- have a dual switch input option. A dual switch input on a machine allows voters with dexterity disabilities who use technology, like sip and puff, to mark their choices on a ballot independently."

The eSlate system is made by Hart InterCivic, and used in several California counties. The AccuPoll is no longer being marketed, Ellen Theisen of VotersUnite told me. She added that Sequoia also has a "sip and puff" feature on one of its systems, though she wrote that there have reportedly been problems with that portion of their "accessible" e-voting machine.

But back to the errors that I would discover after I finally finished the audio process and my ballot was printed out.

I had brought a "cheat sheet" to the polls with me, that I had printed out before going to vote, which included the candidates and initiatives I'd intended to vote for. As I had been forced to vote provisionally, I presume, I was not allowed the option to vote for any of the partisan races, even though it was an Open Primary and I'd told them I wanted to vote in a specific party's primary. I forgot to mention that to Logan to see why I wasn't allowed to vote for all of the races on the ballot for that party.

Nonetheless, I was able to vote in the 10 California Superior Court races on the ballot that I had planned to vote for, as well as for the 2 state initiatives that were on the ballot.

Since I was working from my own pre-printed "cheat sheet", I was careful to select the candidates in the judge contests that I knew I wanted. Each candidate's name was repeated back to me correctly by the audio system just after I selected it as my vote.

I was then allowed to choose from "the candidates" I wanted for the two ballot initiatives. Yes, the audio voice described my "YES" or "NO" options for the initiatives as "candidates". Hopefully the others who used the system were not confused by that inaccurate description.

Finally, after nearly twenty minutes had gone by, and the polls were just about to close at 8pm, I was done. Given the late hour, and the fact that I had had the option to confirm each "candidate" selection, by name, after I had selected it while moving through the ballot --- each one was read back to me correctly, as I had selected it each time --- I chose not to listen to the entire ballot read back to me in the summary confirmation that was offered to me. Instead, I simply chose to go ahead, confirm the ballot, and have it printed out so that I could cast it via the optical-scanner.

The machine seemed to print fine --- except that 4 of my 12 ballot selections were printed incorrectly!

ONE BALLOT, FOUR FLIPPED VOTES

Had I been blind, of course, I'd not have been able to check printed ballot to ensure that it had been printed as per my intent before putting that ballot into the optical-scanner. It hadn't.

My selections on the two ballot initiatives were correctly printed. But 4 of the 12 candidates selected in the Superior Court races were for candidates that I hadn't voted for! I was able to check the printed-out ballot against the proper bubble numbers for each candidate, as seen in the regular, non-"accessible" voting booklet.

The error rate, of course, would seem to far exceed HAVA's federal mandatory requirements for accuracy on such e-voting systems, as detailed by Theisen, of VotersUnite.org, in another recent guest blog posted here at The BRAD BLOG.

For the record, these are the CA Superior Court judge races that were marked incorrectly on my ballot by the ES&S InkaVote Plus system...

  • Office No. 72: I chose to vote for Harvey A. Silberman (bubble #117), ES&S InkaVote Plus selected Serena Raquel Murillo (bubble #118) for me instead.
  • Office No. 84: I chose to vote for Lori-Ann C. Jones (bubble #135), ES&S InkaVote Plus marked Pat Connolly (bubble #132) for me instead.
  • Office No. 95: I chose to vote for Patricia D. Nieto (bubble #141), ES&S InkaVote Plus printed my choice as Lance E. Winters (bubble #142) for me instead.
  • Office No. 125: I chose to vote for James N. Bianco (bubble #159), ES&S InkaVote Plus filled-in Bill Johnson (bubble #158) for me instead.

I was stunned. As were the poll workers who were almost all eagerly following my adventure. I checked, and double-checked against the ballot booklet to make certain I was reading the print out correctly and that the bubbles filled in on the ballot were indeed printed incorrectly in all four of those races.

Though I was able to document much what happened with cell phone video and photographs, I was kicking myself that I hadn't videotaped myself voting the entire ballot, and repeated out-loud, for the camera, who I was selecting in each case as I went through it. No doubt, as usually happens for even non-Election Integrity journalists who encounter such polling place failures, I had little doubt that ES&S --- as is the norm for most of America's horrible Voting Machine companies --- would try and describe the problem I encountered as "voter error".

Luckily, since I was the only voter to have tried to vote on the device that day, and since I have the personal "privilege" of knowing both Logan and many of the officials in the Secretary of State's office, I was able to contact them directly last night to explain what happened, and ask that they immediately quarantine that particular machine for testing.

Logan says that doing so, in any similar situation, is both "helpful" and what he recommends all voters do.

One of the SoS officials emailed me back quickly late last night to say that they had spoken to Logan about the matter immediately, and that they were assured the machine would be isolated and tested. The official also added that the SoS office would participate in the testing to find out what went wrong.

Logan himself called me back quickly after I initially called him, late on an election night as they were still tallying votes (this was much appreciated, since he failed to call me back at all during the "Double Bubble" fiasco, a point which I confronted him about when we met face-to-face some weeks afterwards at a state hearing on the matter.)

After I explained what had happened, the acting Registrar promised that he would "immediately get word out so that we can isolate the machine to see what happened."

I wanted to make sure that nobody from ES&S would have access to that machine between now and the time they were tested to figure out, if possible, what the hell went wrong with it.

"They have no access to that machine," Logan assured me. "They have a few support staff on site," but otherwise "the initial review will not be done with them. The machine will be isolated. That will be the protocol," he assured me.

I asked him to keep me in the loop as to what the procedure would be and when and where the testing would occur, which he promised he would do. Of course, I'd like to be present during the testing of the machine (serial no. LAC016054, as I was able to photograph the back of it) myself, so hopefully Logan will keep his word and follow up appropriately.

"A GOOD CHANCE IT WASN'T JUST THIS MACHINE"

In relating the story to Voters Unite's Gideon, he noted that it was likely that the problem might have occurred on other machines as well.

"There's a good chance that it wasn't just this machine," he mentioned this afternoon. "If those judges were on the state ballot, there's a good chance that the problem could have occurred on every machine used in the county. And that points to how well they did pre-election 'Logic and Accuracy' testing and if they even did it at all."

Indeed L&A testing, though it's supposed to occur on each voting machine prior to Election Day, often doesn't, as election officials have told me in many instances that they are simply unable to test each and every electronic machine to ensure that it's working correctly before Election Day, after the ballots have been programmed for it.

Michael Dean wrote a frightening essay several weeks ago for The BRAD BLOG, describing how Ballot Definition Files (BDF), which are created by employees --- sometimes working for the county, sometimes working for the voting machine companies, none of them necessarily voters or even American citizens --- are almost never fully tested properly for accuracy before they are implemented for use in an election.

Of course, the voting machines themselves are not properly tested in almost any jurisdiction either, much less the BDFs that are programmed to be used with them, often with hundreds of differing formats and languages in all.

Needless to say, we don't yet know what caused my ballot to be printed incorrectly yesterday (that would be a 100% error rate for ballots printed on that ES&S InkaVote Plus machine, since I was the only one to have tried to use it!) so it's impossible to know whether the audio BDF was programmed incorrectly, or whether the machine itself simply failed as ES&S' often do.

Either way, I'm lucky I was able to verify the ballot for accuracy, since I wouldn't have been able to had I actually been a blind voter. Thus, I was able to VOID that provisional ballot, and go back to my actual "precinct" (the orange table, as opposed to the green one) and vote on a normal, hand-marked paper ballot.

In an email this afternoon, Logan confirmed that the very same voting equipment "was deployed and used at 4,383 locations throughout Los Angeles County" yesterday.

RIGHT TO A SECRET BALLOT LOST

Since the scanner at my "precinct" wasn't working, as described earlier in this article, after voting on my hand-marked InkaVote ballot a poll worker examined it before I was about to place it in the ballot box (a plastic tub). He wanted to check my ballot to make sure all the dots had been inked in fully, so the votes would be properly read (in theory) by an optical-scanner back at County headquarters later that night. The task of warning of overvotes is supposed to be done by the optical-scanner which had been broken since the polls opened yesterday morning.

While his efforts to ensure the "integrity" of my selections was appreciated, in truth, I had just lost my right to a private ballot in the process of his well-intentioned examination.

But moreover, it also then occurred to me that I would have lost my right to a private ballot had I been able to successfully complete my attempt at using the "accessible" InkaVote Plus device. My ballot would have been the only ballot printed out by the machine that day, so it would have been a simple matter to tell my votes apart from everyone else's that were hand-marked and cast on heavy card stock, as opposed to thermal printer paper, at the end of the day when the ballots were removed from the ballot box by the poll workers.

The issue of secrecy with machine printed ballots has been a rather contentious one across the country for some time, as experts have found that the same problem occurs with the paper trails that are printed by Direct Recording Electronic (DRE, usually touch-screen) voting machines.

In California, after her landmark, first-of-its kind "Top-to-Bottom Review" of all of the state's electronic voting systems, Sec. of State Debra Bowen decertified most of the state's DRE systems, other than for one-per-polling-place use as needed, in order to marginally meet HAVA's accessibility mandate. When she issued her findings, she initially required that if one person voted on such a machine, poll workers would be required to ensure that several other voters used it as well, so nobody's right to a private ballot would be violated.

It was then pointed out, following the announcement of that new requirement, that it would be quite difficult, if not wholly illegal, to force voters to vote on a DRE if they did not want to, when others were allowed to vote with a paper ballot. In my case, since I was voting just before the polls were to close last night, it would have been largely impossible to find several other voters to vote on the same device, in any case.

"I can't say for sure. I'd have to look at the documentation," Logan told me, when I asked if the InkaVote Plus certification documents required more than one voter to vote on the machine, if someone had used it to create a ballot.

"My recollection is that that became a recommendation [as opposed to a requirement] that poll workers should try to do that, because there was a lot of discussion of whether poll workers could actually require someone to do that," Logan said, though he was unable to tell me for sure what the rules on that were, according to the state's recent certification documents for the InkaVote Plus system.

"I don't believe that language is in the InkaVote Plus recertifcation document, but I'll have to check that," he added.

I checked it for him today. Though the certification document [PDF] reports that "expert reviewers demonstrated that the physical and technological security mechanisms...for the InkaVote Plus Precinct Ballot Counter Voting System were inadequate to ensure accuracy and integrity of the election results," and described "serious design flaws" along with myriad "vulnerabilities" and opportunity for malicious "attacks that could potentially be used to modify any of the information stored in the election results database," it's rather amazing that SoS Bowen ending up recertifying the system for use at all.

That the system was allowed to be used is perhaps even more amazing, given that ES&S spent months withholding it from review entirely by the Secretary of State's independent team of testers. It was finally recertified, at the last minute, just prior to last February's Super Tuesday Primary, largely when it was too late to move to another system in such a large county.

Nonetheless, no specific mandates (or recommendations) are listed in the recert doc requiring (or recommending) that more than one voter use the system, if anybody decides to use the "accessible" portion of the system. It is, however, required that "No poll worker or other person may record the time at which or the order in which voters vote in a polling place."

That requirement, presumably, is meant in order to make it difficult to track who voted when, so that it becomes more difficult to discover how a specific person had voted. That my ballot would have been the only one on at the entire precinct printed on thermal paper, however, means that no matter when I voted, it would have been a simple matter to determine who and what I had voted for.

The recertification also required that a number of "Use Procedures" be created by the voting machine vendors and county officials, though I was unable to find those documents at the SoS website, so can't report whether or not such procedures spoke to the particular privacy issue in question on the InkaVote Plus system.

Of course the AutoMark system, the electronic ballot-marking device which does offer the "sip and puff" functionality I mentioned earlier, and does, as I understand it, use a regular card stock paper ballot which might not be easily noticed as different from the others cast, is also marketed and sold in California by ES&S.

Unfortunately, Bowen was recently outraged to discover that the company had secretly and illegally swapped uncertified AutoMark systems in for the certified versions in several counties. The company now faces millions of dollars in fines from the state and possible decertification. Again.

Last November, Bowen filed a $15 million lawsuit against the company for that unauthorized use of voting hardware which was "never submitted to, or reviewed by, the Secretary of State." The suit was filed after a four-month investigation by her office revealed that they had "repeatedly violated state law."

"ES&S ignored the law over and over and over again, and it got caught" said Bowen, in the press release announcing the suit late last year.

What will or won't "get caught" in light of the extraordinary failures found during my own personal attempting at voting on the ES&S InkaVote Plus voting system in Los Angeles, remains to be seen.

UPDATE 6/12/08: 'Human error' (though not mine) blamed for my votes being flipped, upon initial investigation, according to the County Registrar. Details here...

Prices now slashed in The BRAD BLOG's 2008 Election Integrity Fund Drive! Please support our continuing coverage of your election system, as found nowhere else. Click here for a number of cool new collector's edition Premium products now available starting at just $5!