-- Brad Friedman, The BRAD BLOG
What's going on in Monroe County, Arkansas?
We've been looking at their May 18 "Super-ish Tuesday" election night numbers on the AR Secretary of State's website (Monroe County doesn't have its own public election results website) since the night of the election, and the posted results can only be described as going from "impossible" on the day after the election, to possible but still entirely inexplicable.
At least six different election officials at both the county and state levels remain unable to tell The BRAD BLOG how any of it could have happened, even though thousands of votes appear to have simply disappeared in the final certified results, and the state is set to hold its primary run-off election next Tuesday...
The original tip-off to concerns about Monroe County's results came when on May 19th, the day after the election, the state's SoS website showed a total of 3,393 out of the county's 5,252 registered voters had cast ballots --- a rather impressive 64.60% turnout! But not "impossible."
Here's the screenshot of those numbers, as taken from the AR SoS website on May 19th...
Of the voters who voted in the election, according to the same website on the same night, a total of exactly 600 voters had cast votes for one of the three candidates in the Democratic U.S. Senate Primary, and 2,793 voters had cast votes for one of the eight candidates in the Republican U.S. Senate Primary.
That comes to a total of 3,393 votes recorded as cast in one of the two Senate primary elections. That means there were zero undervotes in either of the Senate contests, since the "total votes" cast that night, as reported by the SoS on May 19th, as seen above, was also 3,393. So, it appeared, everyone who cast a ballot in the May 18th election in Monroe County voted for either a Republican or Democratic Senate candidate. That seemed highly unlikely at first glance, as a "normal" undervote rate might be around 2%. But so far, the numbers weren't "impossible."
But as we looked closer, the totals --- recorded on mostly ES&S touch-screen systems, with a few optically-scanned ballots added in, and all tallied electronically on systems made by the notorious ES&S voting machine company --- became literally impossible, and eventually, just plain ridiculous.
Here are screenshots for the numbers mentioned above, all grabbed from the AR SoS website on May 19th:
There is something else curious about the numbers seen in the screenshots above. Notice all of the precincts, in both the R and D Senate races, where the exact same number of votes were cast for each candidate.
For example, in the Dem results, there are four different precincts where 4 voted for Morrison, 9 voted for Lincoln and 7 voted for Halter. Not "impossible," but curious. On the Republican side, some of the very same precincts also had identical numbers for each of the eight candidates, and a few more had nearly identical numbers. A few of the precincts also reported what appeared to be the exact same percentages for each candidate as seen in the precincts with duplicated numbers, but where the number of votes is simply doubled. All still not "impossible," but certainly getting much more improbable.
And then the numbers finally became plain "impossible" when examining the number of votes cast in the primaries for the Congressional District 1 U.S. House Race that also took place in Monroe County on May 18th.
In that CD1 race, there were 1,860 Democratic and 318 Republican votes cast, for a total of 2,178. That certainly represents a more reasonable undervote rate, but how can it be possible that there were 1,860 votes cast in the Dem CD1 race, when there were only 600 votes cast in the Dem Senate race, yet the "total votes" were all supposedly accounted for, according to the SoS's turnout numbers, in both the Dem and Republican Senate races?
Here are the screenshots from the CD1 race on May 19th:
So how can the above be explained? The state's Election Director was finally able to explain that part of the mystery. But we'll give you that explanation in a minute, because it tipped us off to a troubling concern which nobody, it seems, is able to explain even to this day.
Results Drastically Changed
The election numbers have radically changed in Monroe County since the May 18th election. At least as reported on the SoS website, and as confirmed by local officials.
It's not all that unusual for the unofficial numbers to move a bit following election day, as absentee and provisional ballots are counted and added in to the totals, and as precinct numbers are double-checked for accuracy in the post-election canvass. It is, however, unusual, for vote totals to get a great deal smaller rather than larger in the days following the election. And that's what seems to have happened in Monroe County --- radically so.
Somehow, more than a thousand votes disappeared entirely, as the election results in the Dem and GOP Senate primaries have almost entirely changed.
On May 19th, voter turnout was 3,393:
But by May 21st, turnout dropped by 1,234 to just 2,159, or 41.11% --- still one of the highest turnout rates in the state:
More disturbing, however, are the complete changes in vote totals for all candidates in both the Democratic and Republican Senate Primary races.
Dem Senate Primary results on May 19th (as previously posted above):
Dem Senate Primary results on May 25th:
A total of 1,465 votes seem to have suddenly showed up in the Dem Senate race! And then there are the disappearing votes in the Republican race...
GOP Senate Primary results on May 19th (as previously posted above):
GOP Senate Primary results on May 25th:
2,699 Republican votes have now disappeared! There were reportedly 2,793 Republican voters on May 19th, but just 94 as of May 25th! Neat trick!
To top it all off, even after all of these mystery adjustments, the "total votes" cast in Monroe County as of May 21st, as shown above, is said to be 2,159 --- so, despite the radical change in results, the county's 0% undervote rate stayed intact. Every single voter who cast a vote in either the Republican or Democratic Primary election on May 18th voted in the Senate election, at least according to the Arkansas Sec. of State.
The "good" news? In the CD1 race, while the number of Republican voters has dropped by 229 (to 89 total) and the number of Democratic votes has increased by 60 (to 1,920 total), at least there are not now more Democratic votes than physically possible, as with the county's original numbers on May 19th.
So What The Hell Is Going On?
It's taken a week or so of going from one election official to another to to unwind the mess, and only some of the anomalies detailed above were finally explained.
In trying to sort it all out, The BRAD BLOG spoke to a clerk at the Secretary of State's office, the State's Director of Elections and, on the local level, the Monroe County Clerk, Election Commissioner and Elections Coordinator. None of them were able to explain the most troubling aspect of the numbers.
State Election Director Carter Hawkins was able to explain the reason for the 0% undervote issue, in both the May 19th numbers, as well as the later, officially certified numbers of May 25th.
He told us that the way Arkansas calculates turnout is by adding up votes cast in the largest race (or, in the case of a primary election like the one on May 18th, the largest race in each party's election), discounting all of the overvotes and undervotes, and using that number as "voter turnout."
"We are the only state that does that," Hawkins said, acknowledging the misleading nature of discounting voters who may have either chosen not to vote, or didn't have their vote counted --- for any number of reasons --- in the top race. Doing it that way certainly makes it difficult for citizen-oversight and spotting problem areas where there may be an unusually high number of undervotes, due to machine malfunction or foul play.
"We are working to revise the way we calculate turnout," he told us. "That's one of my goals for 2011."
As to the impossible May 19th results and the subsequent, staggering change to the final numbers as certified by the county, nobody seems to be able to explain them.
"It could be data entry errors," Pam Ratliff, a clerk at the Sec. of State's office guessed when we first contacted them. "The only thing I can tell you, I didn't do that, so I don't know," she said before promising to have Hawkins call us back the next day.
Hawkins couldn't explain it either. Though there had been a problem in Monroe on Election Night, when optically-scanned absentee and provisional ballots were mistakenly combined with the ES&S iVotronic touch-screen votes twice, that issue was discovered and corrected fairly quickly. And, in any case, there were not nearly enough such paper ballots cast to explain thousands of extra votes on May 19th in the GOP Senate primary, for example, or Republican candidate Coleman shown as receiving 306 votes the day after the election, but ending up with just 1 vote, county-wide, after the results were double-checked.
"I cannot explain the difference from 306 to 1," Hawkins conceded. "Sounds strange. But in the last 8 or so years, I've noticed stranger things happening. You'd definitely have to research that with the county's people."
By the time we spoke with Hawkins, we had already spoken to the county's election officials. They were neither aware of the inexplicable numbers, nor had any explanation for them.
"You need talk to the Election Commission. I just use the numbers that they give me," Monroe's friendly County Clerk Janet E. Tweedle told us. In Monroe, she says, the Clerk is largely just the recorder of final results. "The Election Commission brings my office the numbers, and we record, but our Election Commission and Election Coordinator handles our elections."
She also noted that "ES&S actually programs our machines" after the folks at the Election Commission prepare the ballots and send the information to the private voting machine company which services most of the state of Arkansas --- as well as most of the U.S. as the nation's largest election vendor.
As the state has had notorious disasters with ES&S over the years (a small sampling of which were summarized in our "Super-ish Tuesday" Backgrounder article, and otherwise chronicled over the years on these pages), we asked if she'd heard of any problems with the systems on election night that might have led to the strange numbers first published.
"We've always had problems on election night," with ES&S's machines, she said, before pointing me to Election Commissioner Robert Serio and Election Coordinator Leon Lance for more information about what could have gone so wrong.
This is Lance's first year on the job, and in the two conversations we had with him, he repeatedly made it very clear that he takes the job seriously.
"Accountability, integrity, transparency --- no manipulation in any way shape or form. It is what it is in Monroe County," said Lance, who says he was recruited for the job because of his background in electronics. "We're really busting our tails here to make sure the public knows the elections are as honest as they can be."
He didn't know anything about the inexplicable May 19th numbers either, but he was quick, like his boss Serio, to take responsibility for anything that may have caused them. "I'm a first timer here. If something went wrong it was probably my fault."
He was happy to offer details on the process the county uses to tabulate votes, the initial problem that led to absentee and provisional ballots being counted twice, a snag they ran into prior to the election while trying to upload test results from the pre-election "Logic & Accuracy" tests to the Sec. of State's test election results server (the county, like many others in the state, uploads results to the SoS via the Internet), and the thirteen and a half hours he says he and Serio spent double-checking results the day after the election, as well as the additional "seven hours and two complete recounts" they performed before certifying the final results.
As Serio would later offer, some 70% of the county's votes are cast on unverifiable ES&S iVotronic touch-screen systems. So a "recount" on those systems largely involves pressing the button again for another computerized report of the same information. It was combining the paper ballot results (optically-scanned at election headquarters on ES&S m650 machines) with the touch-screen tallies on the ES&S Election Reporting Manager (ERM), that seems to have caused the problem with the double-counting of absentee and provisionals. But that was not what caused the thousands of extra reported "phantom" votes on May 19th.
"I'd have to look at that information," Lance responded, in reply to the insane numbers. He was, however, confident in the results which were eventually certified --- the numbers as seen in the May 25th reports above.
"Honesty and integrity are absolutely paramount. If my name is going be on it, it's got to be 100%," he noted again. "Before we certify the election, I have 99.9999% accuracy, that it is what it is."
He also confessed that he was "not a big fan of ES&S either," adding, "but as far as those [May 19th] numbers go, I'm sorry, but I can't clarify."
Neither could veteran Election Commissioner Serio. In his 32nd year at the Commission, he too was quick to take responsibility for whatever may have gone wrong. "Yeah, probably my error," he said, initially presuming our inquiry concerned the absentee/provisional issue.
He noted the problems they initially had uploading test results to the state's website, and the absentee/provisional problem, but other than that, he had no idea what might have caused the "impossible" numbers. He has no love for ES&S either, and told of problems he's had with both the company and its machines over the years.
Lisa Burks, the founder and former National Coordinator for the Arkansas Headquarters of Verified Voting, wasn't surprised when we told her of the strange numbers in Monroe, and the reactions from county officials and the Sec. of State's office.
State Election Director "Carter Hawkins is telling you accurately. He has seen stranger. Everyone in the SOS office has, and most AR county election officials have," she explained in an email.
She recounted her own bizarre election results when she ran for state-wide federal office in 2004: "Both mine and my opponent's totals went up, then down, then up again nearly 1000 points. I tell folks when they ask me how many votes I received that it depends upon what count they want me to go by. My vote count started at around 46,000 some odd, went up to 48,000 some odd and ended up being certified at 47,000 some odd votes."
"Every election I am contacted by either a candidate or an election worker about some bizarre counts. The Secretary of State's office position has been to pass one on to the county to inquire to them, the county will pass you on to the Sec. of State's office."
That's almost identical to what happened to us throughout our investigation.
"The SOS office, nearly every election, says 'AR election went smoothly, no problems with machines,'" Burks added.
So What Now?
While all of the officials we spoke with said they intended to look further into the matter, nobody had any real knowledge as to what may have gone on inside the computers of the private company which records and tabulates virtually every vote across "The Natural State."
Throughout our discussions with county officials Serio and Lance, it quickly became apparent that the problems in Arkansas --- at least in Monroe --- are likely to get much worse before they get better, as the county has plans to buck the trend in the rest of the country, which is moving away from 100% unverifiable e-voting systems, back to paper ballots.
"We didn't have very many paper ballots over all," on May 18th, Serio said. "We push them to vote on the 'iVos,'" ES&S' notorious, unverifiable touch-screen voting machines.
He says that beginning next Tuesday, when AR holds its primary run-off elections --- including the hotly contested Democratic Senate primary run-off between incumbent Sen. Blanche Lincoln and her challenger Lt. Gov. Bill Halter --- they're "gonna use all iVos. We're not gonna use paper anymore at all."
"My workers do better with iVo's," Serio told The BRAD BLOG. "I thought people would be resistant to it. I like to have a paper record to look at if something goes wrong." He also conceded problems with the so-called paper trail printers on the touch-screens. "Nobody looks at those paper trails," which he says have jammed without voters or poll workers even noticing in the past. Nonetheless, he plans to have all voters use them for all elections in the future.
"I would say about 70 to 80% vote on the iVotronics already. I had 3 precincts that didn't have any paper ballots at all [on May 18th]. We were pushing 'em. Our workers like working with them."
But whether they like working with them or not, as The BRAD BLOG has pointed out again and again over the years, it's strictly impossible to verify that any vote has ever been recorded accurately on one of them during any election. It's also a fact that the iVotronics, and other similar Direct Recording Electronic (DRE) voting machines, have been decertified in state after state following scientific study after scientific study showing them to be easily manipulated and prone to both breakdown and error.
The iVotronics and other DREs have been decertified by states like California, Colorado, Ohio, and even Florida, where iVotronic touch-screens famously lost some 18,000 votes all together, in the 2006 special U.S. House election between Christine Jennings (D) and Vern Buchanan (R). Buchanan would be named the winner of that election by just 369 votes, despite the "lost" votes in the Democratic stronghold county. The reason for the disappearing votes has never been confirmed, even after a number of scientific investigations.
That made Lance's comments about the surprising number of undervotes he noticed in Monroe all the more troubling.
"It was amazing to me," he told us, "how many people voted in local elections, but did not vote for important stuff, Congressman, etc. There was a tremendous number of undervotes where they didn't participate in the national election, but did participate in the local contests."
Neither Serio nor Lance seemed to be aware of any of the studies and problems with iVotronics mentioned above until we called them to their attention. Both men seemed to be troubled by the information which they seemed to be hearing for the first time.
In fact, Lance told us that he thought the ES&S touch-screen systems "are federally mandated." They aren't. But apparently both lobbyists and the outgoing Democratic Sec. of State Charlie Daniels have gone out of their way to tell election officials in every Arkansas county that they are.
Election Integrity advocate Burks explained: "You must understand that in our state most election commissioners, county clerks, (even the media reported this initially) believe the e-voting machines to be 'federally mandated.' This was by design. This is part and parcel because our outgoing SOS Charlie Daniels and his staff went county to county telling them they had no choice to purchase the machines because they were 'federally mandated' under HAVA [the Help America Vote Act] in 2002 after he was elected to that office."
"Of course," she continued via email, "the information from ARs Chief Election Officer, SOS Daniels and his Chief Deputy, Janet Miller Harris, was completely inaccurate, the electronic voting machines WERE NOT federally mandated as I explained, all across this state, for YEARS. Even AR legislators thought --- and some still do think --- they are federally mandated machines because they believed what the Sec. of State's office said to them."
So the outlook remains grim for Arkansas at this hour. It is also still unknown whether the problem that occurred in Monroe --- whatever it was --- was a problem with the ES&S tabulation system, as suspected by Burks, or something else that may have happened after results left Monroe County, moved across the Internet on rural Monroe's copper wires, and then landed in the Sec. of State's election results database and website.
Was it just a computer "glitch"? Human error? Or something more insidious? While Monroe is a very small county, it's possible that a 'bad guy' could have been using the May 19th election as a test run in preparation for a future election, since most of the counties in the state --- indeed, across the country --- use a system nearly identical to Monroe's.
"That's the reason why I got involved," Lance explained in a follow-up call. "To keep the bad guys from getting their fingers involved here. That's exactly why I got involved. To make sure nobody is able to manipulate elections in any way, shape, or fashion."
But with an inability to see inside computers, Lance's hopes may be impossible to achieve.
"I wish I could tell you all that happens to our vote totals and why. No one can without reviewing the software programming codes," Burks wrote.
"One thing is completely certain," she said. "Arkansas has among the worst election problems with vote counts and totals in the nation and --- with the exception of The BRAD BLOG, VotersUnite.org and NCVoter.net --- has been widely ignored."
Arkansas' state run-off elections are scheduled for "Super-Duper Tuesday," next week on June 8th. On the same day, California, Georgia, Iowa, Maine, Montana, Nevada, New Jersey, North Dakota, South Carolina, South Dakota, and Virginia --- many states which also use ES&S equipment, as ES&S is the largest e-voting vendor in the nation --- will hold elections as well. The BRAD BLOG hopes those who follow them will take screenshots of election results websites all night, and in the days that follow --- for whatever they may be worth.