Voting machine tester departs, leaving questions

E-Mails Reveal Confusion

By Chris Bragg
THE COLORADO STATESMAN

On Dec. 20, 2007, John Gardner, the state’s voting systems expert, sent an e-mail to subordinate Department of State equipment testers entitled “Potential lame workaround.”

In it, Gardner proposed to fix a problem with the Sequoia electronic voting machine that had resulted in Secretary of State Mike Coffman’s decertification of the machine just three days earlier. The complex procedure proposed by Gardner involved interchanging two printing devices on the machine in order to conduct a “print test.” Gardner was having trouble performing the test.

“Doesn’t that sound like a crazy idea? But would it work?” Gardner wrote.

The next day, Gardner reported good news to Coffman. He had solved the “print test” problem.

But it wasn’t by exchanging the printing devices. Rather, a Sequoia representative had told Gardner that directions on how to perform the test were in the machine’s instruction manual.

In a Dec. 23 e-mail, Coffman asked Gardner, “Then Sequoia should have been certified from the beginning?”

At a public hearing five days earlier, Coffman had suggested fixing the problem with the Sequoia “print test” by passing a law allowing a federally uncertified software patch onto the machine — a fairly unconventional move.

Now there was no problem?

Coffman wrote another e-mail to Gardner two hours later.

“I don’t understand how you could have missed this during the testing process,” Coffman chided.

Less than six months later, on Friday, June 6, Gardner put in his last day at the secretary of state’s office. Gardner held the post through two tumultuous election cycles, and the reasons for his departure are unclear. However, it doesn’t appear he left under a cloud, or directly because of tensions with Coffman.

He apparently was offered his old job back as the information technology director for the El Paso County Clerk’s office, a position he had held from 2001 to mid-2005.

Department of State spokesman Richard Coolidge said he couldn’t comment on the reasons for Gardner’s departure because it’s a personnel matter. Gardner did not answer an e-mail seeking comment.

Gardner’s departure from the Department of State was welcome news to voting-integrity activists, who consider him a well-meaning civil servant who was in way over his head.

“I think it’s good news for him, and it’s good news for Colorado,” said Paul Hultin, the plaintiff’s attorney in the 2006 court case Conroy v. Dennis, in which opponents of electronic voting attacked the state’s certification of voting machines under then-Secretary of State Gigi Dennis.

Hultin had focused his case on Gardner’s allegedly inadequate IT background, competency and job skills. In his ruling on Conroy v. Dennis, a Denver district judge gave a negative review to the methods Gardner used testing the state’s voting equipment for the 2006 elections and ordered all the equipment retested for 2008.

And, although Gardner is gone, his work testing the state’s voting equipment is likely to continue to have repercussions on the elections in November.

Voting activists are particularly concerned that, six months after tests were completed, the Department of State has yet to release the documentation of Gardner’s work for public scrutiny.

In an interview with The Statesman last October, the former Chief Information Office at the Department of State, who had retired in March 2007, expressed concern about whether Gardner, a former colleague, was up to the task of testing the state’s voting equipment.

“He’s gotten a ton of info. But can he read it? I think on some points, vendors are having to point out what’s in the documentation,” Brian Balay said. “Mr. Coffman just does not know that he needs a little more horsepower in there.”

“He was an ‘IT power user,’ ” Balay continued. “That means he’s pretty good on a PC, that he can use a spreadsheet, word processor, et cetera. He does not have an IT background.”

Although the paper records of Gardner’s testing are unavailable, e-mails between Coffman and Gardner following the Dec. 17 decertification decision offer considerable insight into that later effort.

Those e-mails also, perhaps, illuminate the approach Gardner and Coffman took during the earlier effort of testing equipment — an experimental approach county clerks saw as better suited to a laboratory than an actual election.

Gardner and another member of his testing staff indicated they felt pressure from Coffman to hastily fix perceived problems that led to the decertification. In the saga that would play out over the next two months, Gardner and his testing staff dutifully, if with some initial reluctance, tried to respond. Coffman and Gardner even came up with one solution involving industrial-strength Velcro.

‘No pressure on me’

Gardner had joined the secretary of state’s office in July 2005 as an assistant tester. His boss quit after three months, throwing Gardner into the job of writing rules for testing the state’s electronic voting equipment, then certifying that equipment.

To the consternation of voting activists and some county clerks, despite the judge’s criticism of Gardner’s work in his ruling on Conroy v. Dennis, when Coffman came into office in January 2007, he kept Gardner as the state’s voting systems specialist.

On Dec. 17, 2007, Coffman decertified at least some equipment from 52 of Colorado’s 64 counties. E-mails written in the days before the decertification indicate most of the office’s focus had been on finishing the testing and scheduling a date to announce certification results after a six-month delay.

A Dec. 14 memo written by department legal adviser Stephanie Cegielski laid out post-certification options: holding paper-ballot elections or mail-ballot elections and using electronic voting machines only for disabled voters, or doing a post-election audit on all decertified equipment.

Coffman, however, would quickly propose an entirely new course.

The first inkling of the secretary of state’s plans came at a Capitol hearing on Dec. 18, one day after the certification decision. At that hearing, Coffman laid out a series of fixes to problems found during the testing, many of which he said he’d come up with the night before in a brainstorming session with Paul Craft and Glenn Newkirk — nationally known private voting consultants who were in town to audit Gardner’s testing. (And who have reputations, at least among election integrity activists, as boosters for the electronic-voting-machine industry. They gave Gardner’s testing a glowing review.)

Coffman believed the series of fixes to what he called “tipping points” — the big problems that had resulted in each decertification — would require changes in the law, and he asked the Legislature to pass laws narrowly tailored to address each specific fix.

In addition to asking for a law allowing him to put a software patch on the not-actually-broken Sequoia machine, Coffman’s legislative wish list included suggestions for equally quirky solutions. For a problem with the Hart optical scanner, Coffman proposed a law allowing him to import testing data from California — a solution soon scrapped as unworkable.

Coffman directed Gardner to start coming up with procedures to fix the voting equipment. Gardner was less than thrilled. He had thought the work was supposed to be done largely on Dec. 17, the date of the certification decision.

“Can you tell there’s a rush to have these events take place? No pressure on me ... I’m supposed to be on vacation,” Gardner wrote in a Dec. 20 e-mail to his testing team.

On the same day, Gardner wrote an e-mail to Newkirk indicating Coffman was the one pressuring him.

“I’ve been avoiding him … hoping to get through the last two days before vacation,” Gardner wrote.

Gardner then referred to a story broken that day by the Rocky Mountain News, revealing connections among Coffman, the electronic voting vendor Premier Election Solutions and a political consulting firm in Denver.

Gardner wrote in his e-mail to Newkirk that he thought the story might affect Coffman’s future certification decisions because “you don’t want too many bad stories out there when you’re running for a congressional seat.”

Coolidge, the Department of State spokesman, also declined comment on those e-mails.

Within days, solutions were found to problems that had resulted in the decertification of the Sequoia electronic voting machine (the one in the instructional manual), as well as with the ES&S optical scanner (counties that used the scanner could come to the office and demonstrate that it counted 10,000 ballots correctly.)

Over the next week, Coffman and Gardner exchanged e-mails in which Coffman pushed quick solutions to fix the rest of the equipment.

The most pressing need was to fix the decertified Hart InterCivic optical scan machines. The machines are used to count paper ballots in 47 of Colorado’s 64 counties, and the decertification enraged county clerks in the affected counties. Gardner had found the Hart scanner counted certain extraneous marks on ballots as votes.

In a Christmas Day e-mail to Gardner, Coffman proposed to fix the problem by having poll workers check ballots for such marks before scanning them. Coffman indicated the fix might not be practical, but it would give him some breathing room.

“This, no doubt, would be a very inefficient solution … but it may give comfort to the smaller counties,” Coffman wrote.

Gardner wrote back, however, that the proposed solution was not “reasonable or practical.” (Two months later, after the idea was panned by county clerks, Coffman himself would term the idea “ineffectual and unnecessary.”)

Coffman then asked Gardner whether testing had yet begun on a different version of the Hart optical scanner that had been certified in California. It apparently did not have the same problems as the version tested in Colorado, and Coffman believed it could be used here as a substitute.

Gardner, who by that time was finally on vacation, forwarded Coffman’s question about whether testing had begun to a subordinate tester, Michael Chadwell.

In his response to Gardner, Chadwell said he was in “shocked disbelief at the assumption of the SoS that [led] to the premise of your question.”

“NO! OF COURSE I AM NOT TESTING!”’ Chadwell wrote. “I saw this as a top-down, start from the beginning exercise that did NOT fall under the model of ‘quick-and-dirty’ path testing.”

Gardner, responding to Chadwell, tried to buoy his co-worker’s spirits, noting that it somehow had miraculously snowed in his tropical vacation destination that night, and implied that a similar miracle could be pulled off in fixing the state’s voting equipment.

“I can see from the press releases that we have the full blessing of the secretary of state to certify machines,” Gardner wrote. “If they can make it snow when it’s 70 degrees, we can get through this.”

“Inefficient, inelegant and narrow”

In a series of e-mails, Gardner and a Hart representative would discuss new tests to be performed on the more advanced version of the Hart optical scanner. (In the e-mails, the Hart representative expressed concern about whether Gardner would use special “Hart pens” to mark ballots during the tests. Gardner assured that he had used the special pens in previous testing, and would continue to do so.)

It eventually turned out, however, that the new Hart scanner suffered from the same glitch as the one previously tested in Colorado, and that idea also was scrapped.

If using “special pens” to mark ballots sounds like the conditions of a lab, rather than an actual election, county clerks also noted a similar theme when Coffman and Gardner proposed a fix for the ES&S electronic voting machine.

The problem was the machine could be shut down if someone stuck a magnet into its “PEB slot.” In a Jan. 10 e-mail to his testing team, Gardner wrote, “The Secretary and I just discussed the possibility of gluing or Velcro-ing a plastic box to the PEB slot that would prevent our magnet from getting close enough to trigger the events.”

If followed to its logical conclusion, the plan would have called for county clerks to place a plastic cover with a padlock over the slot and to attach the cover to the machine with industrial-strength Velcro.

At a February hearing where that potential fix was formally presented, Jefferson County Clerk Pam Anderson mocked it, calling the fix an “inefficient, inelegant and narrow option that screams to the voter that it must have been created in a lab.”

That fix also was discarded.

And there was yet another problem. A videotape from Gardner’s tests had been misplaced that showed how a magnet would shut down the machine.

“Everything else can be put on hold until we either — re-create the stack dump or find the video. Ugh,” Gardner wrote in his Jan. 10 e-mail.

On Jan. 11, ES&S vice president Steve Pearson would start badgering Gardner for the videotape, calling his cell phone, work phone and e-mailing him. And whether or not the tape existed — and by extension, whether the defect was actual — would become an ongoing point of contention between the secretary of state’s office and Mesa County, which uses the ES&S machines.

If Mesa County officials had been privy to a Dec. 17 e-mail written by Department of State Chief of Staff Jacque Ponder only a couple hours before the certification decision, they probably would have been even more suspicious.

“Since Ohio just released its study on electronic voting machines, stating a Treo and a magnet would be all the tools you need to tamper with the working of the electronic voting machines, would you like to include the same statement about ES&S in our [press] release?” Ponder wrote to Coffman.

To resolve the issue, Gardner’s staff performed the magnet test at a Feb. 21 hearing before Mesa County officials, showing that the problem was, in fact, real.

Pressure kept building from the Colorado County Clerks Association to find a solution that would get all the equipment recertified. Before the certification decision, there had been limited communications between the secretary of state’s office and county clerks. The office was afraid communications with county clerks could be construed as political pressuring by the clerks and used against the office in a lawsuit attacking the certification process. (Some clerks contend the lack of communication resulted in unwarranted decertification. They say the problems discovered by Gardner that resulted in decertification were resolved soon after Dec. 17 because of increased communications with clerks and vendors).

Now, however, the clerks lobbied freely. El Paso County Clerk Bob Balink, who uses the Premier Election Solutions system and was among the 12 clerks whose equipment was certified from the get go, reported in an e-mail to Coffman that less fortunate clerks were clamoring for answers concerning the decertified equipment.

“CCCA sentiment is forcing a Jan 31 decision,” Balink wrote.

At the same time, legislators were trying to pull together a bill that would allow the decertified equipment to be recertified. Coffman had recruited Rep. David Balmer, a Republican, as a sponsor, but Balmer was having trouble keeping Democratic Sen. Ken Gordon on board. Having a Democratic sponsor was important, since Coffman and Balmer wanted to the bill to be bipartisan and non-controversial.

Balmer said he would try to recruit Sen. Ron Tupa as a Democratic sponsor instead, and Coffman agreed.

“I’m not entirely sure if Gordon is the right choice for this bill. He wants to placate everyone, and I’m not sure if that is possible. I’m sympathetic to those who want paper ballots at polls, but they go too far,” Coffman wrote to Balmer on Jan. 7.

“Gordon seems squishy, while the one good thing about Tupa, whether you agree or disagree with him, once he makes a decision he tends to stick
with it.”

That statement would prove ironic, as Coffman himself would soon change his mind from favoring holding primarily a paper-ballot election in 2008 to advocating on behalf of mail ballots, then later to supporting whatever kind of election each individual county clerk wanted to hold. Meanwhile, Gordon wound up sponsoring the bill, House Bill 1155, which quickly passed both chambers unanimously.

In late February, Coffman recertified all the state’s electronic voting under HB 1155, and all the state’s electronic voting equipment will be available for widespread use this November.

Coffman came full circle in his overall view on which type of election to run, from initially supporting paper ballot elections — in opposition to county clerks — to supporting whatever type of elections the clerks wanted to hold. Over the same period, Coffman came to abandon the fixes he and Gardner had produced in favor of more practical, common-sense solutions proposed by the clerks.

“By the end of the week”

The unresolved question is whether there might be another Conroy-v-Dennis-type lawsuit this summer or fall.

Hultin won’t say one way or the other. But several citizen election integrity activists say it’s impossible to know whether it’s time to attack Gardner’s work again.

Six months after the tests were completed, the testing documents produced by Gardner and his staff that resulted in the initial Dec. 17 certification decision are still not available to the public. So while a Denver District Court deemed Gardner’s work in 2006 troubling enough to order a retest for 2008, the competency of Gardner’s testing for the 2008 elections remains publicly unexamined.

Before it became clear HB 1155 would pass, electronic voting vendors expressed enthusiasm about quickly distributing documents to the public. After all, immediately following the Dec. 17 certification decision, it appeared the only avenue for vendors to get their equipment recertified was to hold hearings challenging Coffman’s decision. And allowing the Department of State to release testing information to the public for scrutiny was a necessary step in getting those hearings scheduled.

The vendors wanted to redact proprietary information from the documents before the documents were released, but, according to one e-mail, that would be no problem. In a Jan. 15 e-mail, Sequoia vice president Ed Smith wrote that his company could be done with redaction on their documents “by the end of the week.”

That never occurred. Now, in late June, those redactions have yet to be completed by any of the four vendors. The secretary of state’s office and vendors reportedly have been unable to agree on what information is proprietary.

Following the passage of HB 1155 and the recertification of all the equipment, neither side seems in any particular rush to get the documents from Gardner’s testing to determine whether the state’s voting electronic voting equipment is secure, accurate and verifiable into the hands of the voting activists who may sue them.