Category: National

Madison: AutoMark – Off The Mark for People with Disabilities

The last thing people with poor eyesight and other disabilities need is a machine that is unreliable and can’t see straight itself.

The last thing people with poor eyesight and other disabilities need is a machine that is unreliable and can’t see straight itself:  <read and view video>

Since 2006, federal law has required that every polling place have voting equipment accessible to people with disabilities. Madison and other local governments use the ES&S AutoMARK. It’s equipped with headphones so voters can hear the choices, and it lets those with limited vision view a magnified ballot, which it then marks. The ballot is printed out and can be reviewed, like any other, before it’s turned in.

Using the ES&S AutoMARK last week at his polling place at Spring Harbor Elementary on Madison’s west side, Shultz noticed the alignment was off. So when he tried to make a given choice, the machine would register a vote for the opposing candidate.

Shultz found this so jarring he filmed it with his cell phone camera. One clip shows his finger touching the oval that says “Kathleen Falk”; that causes the oval for “Nancy Mistele” to fill in. Another clip shows how his effort to select Shirley Abrahamson became a vote for Randy Koschnick.

“It was that way for every single one,” says Shultz, whose field of study includes man/machine interfaces. “To select the candidates you wanted, you had to push right below them.”

Shultz worries that people with visual impairments might not see these different choices being made. He adds that “some simple changes” in the software program, like putting spaces between the choices, could fix this problem.

Adam Gallagher, Madison’s deputy city clerk, says the AutoMARK machines are calibrated before they’re delivered to the city’s 80 polling places. Sometimes the alignment can be off, but this is easily corrected, and the poll workers “know how to do that.” The workers are urged to mark their own ballots with these machines — “anybody can use them” — to test them out.

But Shultz believes he was the first person to use the machine at his polling place that day. The AutoMARK jammed trying to print his ballot, something he’s had happen before. So he ended up voting the ordinary way, making his own marks on a ballot.

In Connecticut we have the IVS system for voters with  disabilities, which has proven expensive, unpopular with election officials, and used infrequently by voters.   The Secretary of the State’s Office is considering replacing it and presumably the AutoMark would be a contender.  We see no reason to reject it based on a single reported problem and potential problems, yet it is worth emphasizing that we need to pick something that works well for everyone involved before we make another huge investment.

Phoenix: Ballots Missing – Integrity and Confidence Also Missing

“If we are correct,” Risner wrote in his Monday letter to the Attorney General’s office, concerning the absence of as many as 19,000 paper ballots, as estimated by observers of the counting in Phoenix, “the question arises as to what happened to those ballots.” The latest mystery adds still more fuel to the already high-stakes, long-sought hand-count, and raises new questions in the nearly three-years long

Update 4/22/2009: Attorney General says Original Result ‘Affirmed”, but questions remain <BradBlog>

However, when a reporter asked about that quote from our report last week, during today’s press conference, Goddard admitted they did not examine the tapes.

“The two things we wanted to hear, we didn’t,” J.T. Waldron, a documentary filmmaker who has been covering the years-long struggle for transparency in the RTA election, told us after the conference. “First, they admit they didn’t look at the poll tapes. Second, they admit that they didn’t do any forensic examination of the ballots” to see if they were legitimate. Critics have pointed out that Pima County has, since the 2006 election, purchased a ballot-on-demand printer and could have printed out ballots to match any ‘fixed’ ’06 ballot counts. The originals were printed by an offset printer, rather than ink-jet, as used in the ballot-on-demand system, and a forensic examination of a sample of ballots, with a microscope, might have revealed any such tampering.

“He said he didn’t have enough evidence to justify a forensic examination of the ballots,” Waldron noted.

Sad,  that with a bit more transparency, we might have confidence in the assessment.

**********
Update 4/16/2009:  AP Article also quoted in BradBlog follow-up:

What may be of even greater concern is the voting machines in Pima County, which includes Tucson, are similar to those used in 12 of Arizona’s 15 counties and in hundreds of jurisdictions across the country.
If it turns out the election was rigged by manipulating Diebold Elections Systems (now Premier Election Solutions Inc.) computer election programs, as some fear, it will show weaknesses in electronic balloting that could endanger the democratic process.

We would emphasize that we should not “fear, it will show weaknesses in electronic balloting that could endanger the democratic process”.  We should welcome the information so we can take action.  And even if it turns out this particular election was not rigged and had no errors, we should be worried that the real risks and dangers to democracy are ignored.

**********Original Post

Two stories on the recount of a 2006 election in Arizona currently underway.  AZCentral has a good background article, Vote probe raises fears over ‘fix’ <read>

Under the scrutiny of criminal investigators, election workers in Phoenix have spent the past week in a painstaking recount of 120,821 ballots that were cast three years ago for a Pima County transit tax.

The primary objective is to determine whether someone rigged the election by tampering with the optical-scan polling machines in Pima County, transforming “no” votes into “yes” votes…

And, no matter what happens with the recount, state Attorney General Terry Goddard said he is convinced the equipment jeopardizes election integrity.

“These (Premier) systems are very, very bad,” Goddard said. “(They) are not state of the art in terms of security. They are not state of the art in terms of transparency.”

Chris Riggall, a spokesman for Premier, said his company uses the most advanced technology available and urges clients to establish security protocols. Although there are internal safety features to prevent and detect tampering, Riggall added, those must be supported by external controls over election workers who handle the equipment.

“You can manipulate any voting system devised by man,” he said, noting that even paper ballots are subject to fraud…

During an initial investigation, the Attorney General’s Office hired an independent company to analyze the machines. Experts at iBeta Software Quality Assurance found a number of “irregularities” and determined that the Premier system had “fundamental security flaws.”

They reported that a trained technician could have altered the vote and removed all evidence of the crime. However, they concluded that human error was a more likely culprit than fraud because evidence of tampering had not been erased.

The Attorney General’s Office dropped the case in 2007, but the Democratic and Libertarian parties continued private investigations and filed lawsuits for election record…

Former Pima County employee Zbigniew Osmolski filed an affidavit in July alleging that he was in a Tucson lounge when Crane admitted that he had “fixed the RTA election on the instructions of his bosses” and was fearful of indictment…

Risner noted that, according to sworn courtroom testimony, Crane routinely took voting data home during elections and had his office computer connected to the GEMS system. He also claimed that Crane purchased a hacking device before the election, one with “no other purpose than to illegally alter the programming of precinct voting machines.”

Now for the latest disturbing part from Brad Friedman, ‘Thousands of Ballots’ May be ‘Missing’ from AZ’s Criminal Investigation Ballot Count in Phoenix<read>

“If we are correct,” Risner wrote in his Monday letter to the Attorney General’s office, concerning the absence of as many as 19,000 paper ballots, as estimated by observers of the counting in Phoenix, “the question arises as to what happened to those ballots.” The latest mystery adds still more fuel to the already high-stakes, long-sought hand-count, and raises new questions in the nearly three-years long investigation into the 2006 election results.

If the ballots are indeed missing, did they ever actually exist? Was the Diebold electronic ballot box stuffed? Have ballots been surreptitiously removed by someone for some reason?

As we have said there is no reason to be confident that referendums are not subject to skull duggery: FAQ: Why Would Anyone Steal A Referendum?

Court Affirms Minnesota Recount and Election Fair

After seven weeks of trial, the factual record is devoid of any allegations of fraud, tampering, or security breeches on Election Day, during the recount process, or during the election contest.

We are not comfortable that a statewide or even a Congressonal District recount in Connecticut would result is a thorough or as credible a recount as we have seen and admire in Minnesota.

We highly recommend the 2nd video, if you are willing to invest the 87 minutes.

The full 68 page ruling <read>

The Court DECIDES, DECLARES AND ADJUDGES that Contestee Al Franken is the party to the contest who received the highest number of votes legally cast…and is therefore entitled to receive the certificate of election…Contestants’ [Coleman] Notice of Contest is dismissed with prejudice;…costs of the contest must be paid by Contestants

Pretty clear.  And some details:

The overwhelming weight of the evidence indicates that the November 4, 2008 election was conducted fairly, impartially and accurately…There is no evidence of a systemic problem of disenfranchisement in the state’s election system, including absentee balloting procedures…After seven weeks of trial, the factual record is devoid of any allegations of fraud, tampering, or security breeches on Election Day, during the recount process, or during the election contest.

Mark Ritchie
Mark Ritchie

For more information, we would highly recommend the video of Secretary of State Mark Ritchie’s talk to the National Association of Secretaries of State <watch>. Or a longer talk about the recount, the length of time allowed for recounts, and how a national recount could be handled. <watch> This took place at the 2009 Legislative Conference, a national meeting of county election officials.

We highly recommend the 2nd video, if you are willing to invest the 87 minutes.  It covers the Recount, Election Day Registration, Early Voting, and the National Popular Vote.

Our Opinion: The Minnesota process was thorough and fair, the actually counting of ballots was accomplished in a few days, yet the counting of disputed ballots and reviewing absentee ballots took the bulk of the 40+ days.  There needs to be a way of streamlining that part of the process with more planning ahead.  Consider the problems, pointed out in the 2nd vidoe by SOS Ritchie, that could occur with a statewide primary to be followed closely by an election, when the primary requires a recount! In the first video SOS Ritchie gives great credit to the planning and preparation for potential recounts, something we see as sorely lacking in Connecticut — We are not comfortable that a statewide or even a Congressonal District recount in Connecticut would result is a thorough or as credible a recount as we have seen and admire in Minnesota.

CIA Agent: Electronic Voting Risky

“I follow the vote. And wherever the vote becomes an electron and touches a computer, that’s an opportunity for a malicious actor potentially to . . . make bad things happen.”

We agree with the agent that electronic voting can be compromised, but some details in the testimony are questionable.

Update 5/7/2009 Boston Progressive Examiner: Electronic voting machines in U.S. at risk from foreign hackers attacking military computers <read>

The U.S. Election Assistance Commission should be paying attention to what has been happening at the Department of Defense. America is under cyber attack each day with thousands of attacks on defense websites. As computer technology spreads in election offices around the country the risk of foreign hacking of American elections grows.

Update 4/8/2009 Wall Street Journal: Electricity Grid in U.S. Penetrated By Spies <read>

Last year, a senior Central Intelligence Agency official, Tom Donahue, told a meeting of utility company representatives in New Orleans that a cyberattack had taken out power equipment in multiple regions outside the U.S. The outage was followed with extortion demands, he said…

The sophistication of the U.S. intrusions — which extend beyond electric to other key infrastructure systems — suggests that China and Russia are mainly responsible, according to intelligence officials and cybersecurity specialists. While terrorist groups could develop the ability to penetrate U.S. infrastructure, they don’t appear to have yet mounted attacks, these officials say.

*************************Original post:

McClatchy:  Most electronic voting isn’t secure, CIA expert says<read>

“You heard the old adage ‘follow the money,’ ” Stigall said, according to a transcript of his hour-long presentation that McClatchy obtained. “I follow the vote. And wherever the vote becomes an electron and touches a computer, that’s an opportunity for a malicious actor potentially to . . . make bad things happen.”

Stigall said that voting equipment connected to the Internet could be hacked, and machines that weren’t connected could be compromised wirelessly. Eleven U.S. states have banned or limited wireless capability in voting equipment, but Stigall said that election officials didn’t always know it when wireless cards were embedded in their machines.

While Stigall said that he wasn’t speaking for the CIA and wouldn’t address U.S. voting systems, his presentation appeared to undercut calls by some U.S. politicians to shift to Internet balloting, at least for military personnel and other American citizens living overseas. Stigall said that most Web-based ballot systems had proved to be insecure.

We agree with the agent that electronic voting can be compromised, but some details in the testimony are questionable.

Appearing last month before a U.S. Election Assistance Commission field hearing in Orlando, Fla., a CIA cybersecurity expert suggested that Venezuelan President Hugo Chavez and his allies fixed a 2004 election recount, an assertion that could further roil U.S. relations with the Latin leader.

Both a Princeton/Johns Hopkins study and the Carter Center have studied the Venezuelan election and refute some of these contentions in that particular case.

PRINCETON, N.J. — An analysis of polling data from the Aug. 15 referendum in Venezuela to recall President Hugo Chávez indicates that certain forms of computer fraud were unlikely to have occurred during the electronic voting process, according to a study by computer science researchers from Johns Hopkins and Princeton universities.

Jennifer McCoy directed the Carter Center’s observer mission in Venezuela and is a Latin America expert at Georgia State University in Atlanta:

In conclusion, the vote itself was secret and free, but the CNE’s lack of openness, last-minute changes and internal divisions harmed public confidence in that vital institution both before and after the vote. Divisive rhetoric and intimidating tactics from Chavistas, and the opposition’s still-unsubstantiated claims of fraud, have exacerbated Venezuelans’ cynicism toward elections. It will take a huge effort by both sides to restore trust in this fundamental democratic right before next month’s election for governors and mayors.

Ada Lovelace Day – Appreciating Marla Ludwig, “The Backpackin’ Granny”

Over 1000 blogs are celebrating Ada Lovelace day today: Ada Lovelace Day is an international day of blogging to draw attention to women excelling in technology…We want you to tell the world about these unsung heroines.

Nobody makes a bigger mistake than (s)he who did nothing
because (s)he could only do a little

Over 1000 blogs are celebrating Ada Lovelace day today:

Ada Lovelace Day is an international day of blogging to draw attention to women excelling in technology…We want you to tell the world about these unsung heroines…We’re not just interested in hardcore ninja programmers, but any woman who creates, invents, or uses any technology in an innovative way. Feel free to interpret it as widely as you like.  http://findingada.com/

Here at CTVotersCount we are recognizing Marla Ludwig a long time friend and former co-worker.  Marla is an unassuming data storage support expert with a day job to keep the “plumbing” working and avoiding data disasters for a very large insurance company.  That would be enough, yet Marla is also known to the rest of the world as “The Backpackin’ Granny”.  She spends her vacations transforming a small village in Ghana.  She has built a library, brought water purification, and is now building a school.

Nobody makes a bigger mistake than (s)he who did nothing
because (s)he could only do a little

PURPOSE: Bright Star Vision is a secular grassroots organization whose purpose is to create a world in which children are free from harm and disease; a world in which they are educated confident and motivated; a world in which all children have clean water, food in their stomachs, shoes on their feet, smiles on their faces, and joy in their hearts.

Bright Star Vision is a grassroots support organization that originated in 2005 when a woman named Marla Ludwig decided to visit the African Country of Ghana for a vacation with no more than just her backpack earning her the nickname “Backpackin’ Granny” . When asked by a friend whether she would consider bringing some books or supplies to some of the underprivileged there, she initially overlooked the idea. But upon further evaluation, she stopped wondering “Why?” and began to think “Why Not” This action led to the outcome of her single-handedly sponsoring a village in need in the Volta Region of Ghana.

Bright Star Vision is the result of one woman’s vision. The name for the organization was inspired by Marla’s dream of a bright star shining over the Dalive Village, as well as the children’s enthusiasm and eagerness to read, learn, and do well in school upon being provided with books, school resources, and student sponsorships.

For more information: http://www.brightstarvision.net/index.html

Another Take On ATM’s vs. Voting Machines

Security firm Sophos reported this week that it received three samples of a trojan that was customized to run on Diebold-manufactured cash machines in Russia…

CTVotersCount.org Myth #8 – If we can trust our money to ATMs we can trust our votes to computers. <10 myths> <also>

Perhaps ATM’s are not as safe as we sometimes think.

Today a story shows that ATM’s are vulnerable.  SCMagazineUS has the story: ATM malware appears, Diebold issues security update <read>

Security firm Sophos reported this week that it received three samples of a trojan that was customized to run on Diebold-manufactured cash machines in Russia, said Graham Cluley, Sophos’ senior security consultant. The malware was able to read card numbers and PINs — then when the attacker returned to the ATM, he inserted a specially crafted card that told the machine to issue him a receipt containing the stolen information.

“Basically [the malware] would be spewing out the identity information,” Cluley told SCMagazineUS.com on Wednesday. “It’s a really cunning scheme. You need to know how to talk to the ATM. It was working with the Diebold DLL (dynamic-linked library). It knew what API (application programming interface) calls to make, which is information, I suspect, not normally in the public domain.”

Diebold this week disclosed that it issued a security update in January for its ATMs running a Windows-based operating system to address the problem. Diebold told its customers in a letter that a number of its machines in Russia were infected — but the company did not reveal specifics on the attacks.

The somewhat comforting part of this story is that Diebold issued a fix in short order for the problem – while problems in their voting machines go unaddressed for years through multiple software versions.

However, it is a reminder of the vulnerability of any computer system to which somone gains access, including voting systems.

Diebold Audit Logs Miss Critical Data

“Today’s hearing confirmed one of my worst fears,” said Kim Alexander, founder and president of the non-profit California Voter Foundation. “The audit logs have been the top selling point for vendors hawking paperless voting systems. They and the jurisdictions that have used paperless voting machines have repeatedly pointed to the audit logs as the primary security mechanism and ‘fail-safe’ for any glitch that might occur on machines. To discover that the fail-safe itself is unreliable eliminates one of the key selling points for electronic voting security.”

In Connecticut we avoid these specific problems. But we don’t avoid similar problems.

Kim Zetter at wired has on of several reports on hearings in California <read>

Summary:  “The Humboldt Election Transparency Project” discoverd 179 missing ballots in the original election accounting.  One memory card total was dropped in accumulating votes after the election.  Subsequent investigations found that there was a known (to some) problem in the code that could cause that to happen, yet, no record of the event was in the audit logs and the audit logs could be easily deleted.   Yesterday the Secretary of State held hearings as reported by  Kim Zetter:

“Today’s hearing confirmed one of my worst fears,” said Kim Alexander, founder and president of the non-profit California Voter Foundation. “The audit logs have been the top selling point for vendors hawking paperless voting systems. They and the jurisdictions that have used paperless voting machines have repeatedly pointed to the audit logs as the primary security mechanism and ‘fail-safe’ for any glitch that might occur on machines. To discover that the fail-safe itself is unreliable eliminates one of the key selling points for electronic voting security.”

Following a public records request of GEMS logs, Threat Level previously reported that the Premier/Diebold logs did not indicate when election officials in Humboldt County, California, intentionally deleted more than two dozen batches of ballots from their system during the November general election.

The finding raised questions about the integrity of elections conducted with the system, but it was unknown at the time whether the problem with the audit log existed with other versions of the GEMS software used in other counties in California and across the country. Premier/Diebold didn’t respond to phone calls seeking information at the time.

In Connecticut we avoid these specific problems as we do not use the GEMS system for election totaling votes from memory cards.  (Our vendor, LHS uses GEMS for programming the memory cards for each of our elections).  But we don’t avoid similar problems, Connecticut uses an error prone three step process of manual transcription to produce our vote totals – for the November 2008 election this system dropped and added even more votes than the number of ballots dropped in California: e.g. <here> <here> <here> <here>

********
Update 3/25:  Diebold tries to cut off Humbolt, Are the sending a message “Don’t Tread on Dieblod?”

Here is the Daily Voting News Summary which is good summary of what we have so far <read>

Yes Virginia! – No Ballots, No Problems – Trust The Memory

Close election in Fairfax County decided by reading computer memory.

Maybe it is all mostly accurate. But, without a voter verified paper record who knows? Maybe there is a lesson in here for us. Unfortunately, there is also a lesson here for those looking for ways to game the system in the future.

A close election in Virginia with electronic touch screen voting.  They have optical scanners, but they saved some paper and used their expensive touch screen machines.  One machine made an obvious error so they counted the votes in the memory log on the two machines in that precinct and declared the records in memory accurate.  Yet what about all the other machines that counted 89 votes more for one candidate out of 12,000 cast?

Stories in the Washington Post,  BradBlog and LocalTV

According to Brad and the WaPo:

the geniuses who run Fairfax County’s election decided to use only touch-screen systems in the election yesterday, despite having used both paper ballots and touch-screens in last November’s election. The WINVote “is the most widely used touch-screen voting machine in Virginia,” according to the Washington Post story in which explanations are given for why the Republican “narrowly defeated” the Democrat by 89 votes.

Officials are “not yet sure what caused the device to malfunction.”

WaPo’s earlier story — when the Republican John Cook was said to have been leading the entirely-unverifiable election by 69 votes, before the wholly-unverified and unverifiable “votes” from the failed machine were printed out, one-by-one, and then tallied by officials from the machines memory — notes that the race came down to the votes cast in the single precinct where the machine failed.

Brad had the same reaction that I did to a quote in the local TV story:

Voters are mixed. “I think the electronic equipment these days is pretty good,” said Fairfax County resident Julie Stewart. “But paper would be fine if they’ve got a lot of money and they want to spend the time doing it,” said Fairfax County resident Richard Carlson.

…Dear Richard: Paper elections are cheaper, more accurate, and take no more “time” to tally than touch-screen elections. And at the end of the day, it’s possible to know who actually won them.

Maybe it is all mostly accurate.  We have no reason to assume the result is inaccurate.  But, without a voter verified paper record who knows?  Maybe there is a lesson in here for us.  Unfortunately, there is also a lesson here for those looking for ways to game the system in the future.

Of Levers, WPE, and The National Popular Vote

How certain can we be that George Bush won the popular vote in 2004?
How certain can we be that Al Gore won the popular vote in 2000?
How many votes would be added or subtracted if the reported popular vote was close and a process like Minnesota’s were used for a nationwide recount of the paper ballots?

We have made several posts lately about the relative merits of touch screens (DREs), vs. optical scanners vs. lever machines.  We have also warned of the risks in going to the National Popular Vote given the current state by state variations in voting integrity.

We ask you to contemplate:

  • How certain can we be that George Bush won the popular vote in 2004?
  • How certain can we be that Al Gore won the popular vote in 2000?
  • How might the totals change if every error or fraud in every state could contribute to the popular vote and it would decide the President?
  • How many votes would be added or subtracted if all the absentee ballots and provisional ballots were counted in every state and the results for every district and precinct reviewed to make sure each state’s totals were accumulated correctly?
  • How many votes would be added or subtracted if the reported popular vote was close and a process like Minnesota’s were used for a nationwide recount of the paper ballots?
    (answers below)

Within Precinct Error (WPM) is a measure of the difference between the exit polls and the actual result.  We have heard a lot about the discrepancies between the exit polls in 2004 and the election results that claim that Kerry was consistently higher in the polls than the results and that it points to fraud <e.g.>.  Countered by strong arguments from statisticians that there are other explinations. <e.g.>

A story today reminds us that we cannot trust the fictional national popular vote number. <read>

The 12% NY WPE cut Kerry’s vote margin (and increased Bush’s) by about 900,000. Kerry also won late paper ballot votes (absentees, provisionals, etc.) by the same 64%.  Just a coincidence, Lever bots will surely say. But Past is Prologue. Obama won 71% of NY late votes and just 63% of the recorded vote. Gore won 66% of late votes and 60% officially. Mechanical Levers had the highest average WPE (11%) of all voting machines. Paper ballots had 2%, optical scanners and unverifiable touch-screens 7%. At least there is a paper trail with optical scanners,

New York is just one state:  No paper – No means of proving/refuting – No confidence

Our answers to questions above:

  • How certain can we be that George Bush won the popular vote in 2004?
    Not certain enough to feel confident.
  • How certain can we be that Al Gore won the popular vote in 2000?
    Given the errors in 2000 accounting in FL and fraud found in other elections its quite likely that the magnitude of error exceeded the margin nationwide.
  • How might the totals change if every error or fraud in every state could contribute to the popular vote and it would decide the President?
    The incentive for fraud would apply to every district, precinct, and state, not just a few close states.  There would be more for advocates to monitor.
  • How many votes would be added or subtracted if all the absentee ballots and provisional ballots were counted in every state and the results for every district and precinctreviewed to make sure each state’s totals were accumulated correctly?
    Lots.
  • How many votes would be added or subtracted if the reported popular vote was close and a process like Minnesota’s were used for a nationwide recount of the paper ballots?
    None.  This is a trick question.  Each state only recounts its on votes based on a close margin in its own totals.  Even in a very close national popular vote, none or very few states would have a recount.  Also many states, like New York, do not have  paper ballots to count.

Or Could We Have Been Even Better Off With Levers?

I read HAVA. It clearly does not ban levers. I recently discovered what has helped fuel this misinformed opinion in part: it is the discredited position of the discredited U.S. Election Assistance Commission (EAC)

But there is more to the story.

BradBlog reports that the Election Assistance Commission (EAC) may have been wrong when it issued an advisory indicating that lever machines were banned by the Help America Vote Act (HAVA) <read>

The EAC Lied, Lever Voting Machines (Almost) Died

Exclusive: Discredited federal E-voting oversight commission issued an incorrect 2005 ‘legal advisory’ helping to keep NY on a collision course with democracy
But it’s not too late to save the last transparent electoral system in the United States…

So what is driving New York State to stick with a law that so many in New York believe to be such a bad idea? As a New Yorker who has been talking to many election commissioners, legislators and citizens, I was surprised to learn how many people believe the “Help America Vote Act” (HAVA) actually banned lever machines.

I read HAVA. It clearly does not ban levers. I recently discovered what has helped fuel this misinformed opinion in part: it is the discredited position of the discredited U.S. Election Assistance Commission (EAC), as detailed in a newly-unearthed document prepared for the state of Pennsylvania, at their request, in regard to the legality of lever voting machines.

We remember that Connecticut and New York were “late” in becoming HAVA compliant.  Connecticut rushed to comply almost making the expensive and risky mistake of purchasing uncertified “touch-screen” from Danaher,  with the Secretary of the State, thankfully, changing course to optical scanners.  New York chose defy the feds and stick with levers until a suitable alternative could be found.

But there is more to the story. Here is a quote in the article, not clearly attributed, purporting to describe lever machines:

* For those who don’t know, here’s how lever machines work:
Voter pulls lever for candidate of her choice; gears increment a mechanical counter by one and only one vote — only for the desired candidate. No vote switching or overvoting is possible! (Some machines increment the counters as the big lever is pulled, but unlike software, either method of operation can be observed and thoroughly tested before and after each election and both have been completely disclosed in the machines’ patents.) Rinse and repeat for the entire ballot, which takes less than a minute for most voters. Change or correct your votes as many times as you like – not just three. When you’re done, just pull the big lever that casts the ballot, locks in all your votes, opens the privacy curtain, and repositions the candidate levers for the next voter, leaving the locked immutable mechanical counters as the durable record of all the votes cast on the machine — until after the election is certified. On election night, a permanent paper record of the vote tallies on each machine is produced by the machine, and/or by bi-partisan teams of poll workers, before the machine is moved and the poll workers are permitted to leave.

Not so fast.  Looking at comment #3 below the article, David Jefferson, makes the case against levers:

Without commenting on the rest of this posting let me say that the italic comment at the bottom that describes lever machines somewhat mischaracterizes them.

1) Although it is true that the mechanics of a lever machine are vastly simpler than software, and can be understood by careful observation (if the back of the machine is open) by a mechanically inclined person, it is easy to tamper with the gear mechanism so that it miscounts votes, either by failing to increment one time out of 10 or by failing to carry into the next decimal place. This almost always causing an undercount for particular candidate(s), and not an overcount. If such a problem occurs, it is unlikely to be discovered very quickly, since undercounts never lead to any outright inconsistency with the counts of voters or any other data. And whether the problem is detected or not, there is no possibility of recovery, because there is no redundancy at all, let alone anything you would call an audit trail.

2) The same is true if the machine has been misconfigured (equivalent to having a bad election definition file). There my be no recovery.

3) It is true that you can change your (tentative) vote as many times as you want with lever machines. But you can also do that with DREs. The max number of three spoiled ballots is only a limitation of paper ballots, and then only because of an arguably obtuse law–not for any fundamental reason.

4) The counters are not any more “immutable” than any other volatile memory medium. If I remember correctly, a single key turn allows all counts to be zeroed, with no record of the time the occurred, or who did, it or anything else. Arguably, the paper record of the counts is just as durable as the counts that are stored mechanically.

5) A lever machine does not accumulate a “record of all the votes cast”. It records only counts of the votes cast, which is vastly less information than any other voting system. There is absolutely no redundancy in this information, as there is with all other forms of voting, which is why it is impossible to do a meaningful audit that corresponds in any way to the audits that are possible with paper ballots or VVPAT.

We tend to agree with David Jefferson about the attributes of lever machines.  Yet, given all that we know now – the cost of optical scanners — the risks without sufficient, reliable audits — the stories we have heard from registrars about problems with lever machines covered over in the backrooms in Connecticut — it is a close call.  But with sufficient audits, a stronger chain-of-custody for ballots, and manual recounts we still would favor optical scanners.