Dave Eckhardt's Voting Machine Page

April, 2012

In late February, the Pennsylvania Department of State posted on their web site "test protocols" for re-examining voting systems currently certified by the Secretary of the Commonwealth and currently used in Pennsylvania elections, to determine whether they should continue to be certified.

When I read the "test protocol" document for the ES&S iVotronic, it seemed to me that some important sources of information were being overlooked, so I wrote this report: VoteAllegheny Analysis of the Pro V&V iVotronic Re-examination Test Protocol. Copies were sent to various people at the Department of State and to the Secretary's contract examiner, but no response has yet been received.

November, 2011

In summer/fall 2011 I participated in an investigation of the primary election held in Venango County, Pennsylvania in May of 2011. The Venango County Board of Elections had various concerns and had received a variety of complaints from voters.

This investigation was variously influenced and complicated by the Venango County Commissioners, Venango County's voting-system vendor (ES&S), and the Pennsylvania Department of State. The investigation did not run to an orderly conclusion and not all results were made public, but some items did enter the public record. These are local copies of selected items (more are available on the VotePA web site).

December, 2008

Here is a report I wrote on Allegheny County's iVotronic firmware verification process. This process has some rough edges, but in the case of the vast majority of electronic voting machines used in the United States, nobody has ever checked to see whether they are running the correct firmware. In many jurisdictions, there is no cross-check on vote totals reported by electronic voting machines (because there is no voter-verified paper ballot and not even any kind of paper "audit trail").

September, 2008

Here are some notes I prepared on Speaking to County Council .

October, 2007

Here are some notes I prepared on observing Logic & Accuracy Testing of voting machines.

August, 2007

The Allegheny County Citizens' Election System Advisory Panel issued our Initial Report to County Council.

January

On Tuesday, January 31st, 2006, I made a statement to the Allegheny County Board of Elections. It was written on a deadline to address a particular go/no-go decision on Diebold voting machines and hence doesn't represent my full thinking on the topic of electronic voting machines.

Some links relevant to my statement:

California
- CA Secretary of State's Staff Report on the Investigation of Diebold Election Systems, Inc.
- Attorney General Lockyer Announces $2.6 Million Settlement with Diebold in Electronic Voting Lawsuit
Montgomery County, MD's 2004 Presidential General Election Review--Lessons Learned
Greensboro, NC News-Record article Diebold pulls out of contract competition (this link may be unstable)

Press coverage:

Wednesday, February 1st
- Pittsburgh Post-Gazette: New voting machines a crazy idea
- Pittsburgh Post-Gazette: Allegheny County remains undecided about new voting machines
- KQV Radio: interview with Joe Fenn, 15:30-15:40 (I don't have a link for this)
- In response to the second Post-Gazette article, I submitted a letter to the editor clarifying that what I'd said in the meeting wasn't just my opinion. That LTE didn't appear because it was superseded by the op-ed piece below.
Sunday, February 5th
- Pittsburgh Post-Gazette op-ed with Danny Sleator: Forum: Computers and voting -- a dangerous mix
- Pittsburgh Post-Gazette article: Voting machine firm's past questioned

February

I attended, I believe, two more meetings in February and March, but did not prepare formal statements.

March

On Tuesday, March 21st, 2006, I made a statement to the Allegheny County Council. In theory this was in support of a Council resolution to move $3 million from the general voting-machine-purchase budget to a separate line item dedicated to paper-trail upgrades for DRE machines. However, the Sequoia AVC Advantage machines currently slated for deployment don't have any sort of printer attachment, let alone one with federal or state certification, so this was to some extent a symbolic gesture. And most of the VotePA testimony was actually focused on explaining to the Members why we oppose the Advantage.

My statement referred to the Verified Voting Foundation's study Are Voter-Verified Paper Ballots Cost-Effective?.

On Monday, March 27th, 2006, I was quoted in a Pittsburgh Post-Gazette article, Professor to try to hack voting machines.

April

On Tuesday, April 4th, 2006, I made a statement to the Allegheny County Council. Formally I was supporting a Council resolution, by Bill Robinson, to create a citizen board to advise the county on voting-machine issues. I used the occasion to argue that immediate intervention by Council is necessary to steer us away from likely disaster. I was able to get through my statement in just under three minutes, so that annoying alarm didn't go off.

My statement referred to the March 21st primary disaster in the Chicago area and the March 29th fiasco during the PA certification examination of the AVC Advantage.

On Tuesday, April 11th, the Post-Gazette somewhat belatedly weighed in on the side of verifiability: Editorial: Ballot blues / The county is rushing on new voting machines. Better late than never!

On Tuesday, April 18th, 2006, I yet again made a statement to the Allegheny County Council, formally in support of a resolution calling on Congress to delay the HAVA implementation deadline. Right now the Elections Division is ramping up to deploy iVotronics in the four weeks (!!) remaining before our primary.

The bulk of my statement was related to two issues. First, County Executive Dan Onorato's office has been issuing a form-letter response to questions about the iVotronic touch-screen DRE's versus optical scan. The operative part (clipped from a April 17th response to a constituent query) is:

Please be aware that optical scan voting machines were not selected for a number of reasons. First of all, optical scan machines typically require frequent and expensive maintenance. Secondly, many senior citizens found optical scan machines difficult and cumbersome, and finally, optical scan machines are not ideal for visually impaired voters. In addition, any extraneous marks or incorrectly completed bubbles can cause problems and confusion when tallying the vote.

Of course, the Allegheny County VotePA group has been bombarding the Board of Elections and County Council with specific detailed information on these topics for months (as just one example, many counties who actually used touch-screen DRE's in 2004 elections found them to have very high maintenance costs and have switched to precinct-counted optical scan as a result). Since February I have been asking the BoE to provide a detailed cost comparison for their favorite touch-screen DRE versus a precinct optical scanner. If such a comparison exists and is favorable to the DRE this should be easy, right? And what does it mean if not?

Second, the County Executive's office says ES&S is hard at work creating and certifying a sip-and-puff interface to make the iVotronic accessible to people with disabilities sometime soon (of course, we could have had this already with either AccuPoll or AutoMark). Meanwhile, there is still talk about some sort of paper trail someday maybe. It seems as if ES&S and the executive branch of the county pay more attention to ordinances requiring things than to any amount of citizen input. So I urged Council to pass a law requiring voter-verified paper voting records.

Press coverage:

Thursday, April 20th
- Pittsburgh Tribune-Review: E-voting under fire at the wire
Friday, April 21st
- KQV Radio: interview, 06:30-06:35 (I don't have a link for this)

On Tuesday, April 25th I received my poll-worker training. There were some omissions.

They told us about headphones (for blind voters) but we never saw them, let alone saw them in use. That is, we never received instructions on how to reconfigure the machine for a blind voter or how to instruct a blind voter.
We received no instruction on removing the electronics pack from the machine body to assist voters in wheelchairs. Since then I have been told that this is an iVotronic feature, but we heard no hint of it at the training session.

May

We've received some measure of vindication with respect to Diebold--apparently there are (despite piles of federal and state certifications of the machines!) serious security flaws in their voting machines.

Inside Bay Area, May 10th, 2006: Voting glitch said to be 'dangerous'
"It defies reason that anyone who works with security would tolerate this design," [said Douglas Jones, a University of Iowa computer scientist and veteran voting-system examiner for the state of Iowa].
Cleveland Plain Dealer, Thursday, May 11th, 2006: Diebold voting machine spurs security concerns
"It's worse than a hole," said [Michael Shamos, a computer science professor at Carnegie Mellon University], who has been briefed on the vulnerability of the Diebold machines. "It's a deliberate feature that was added by Diebold that we all believe is unwise."
Newsweek, Monday, May 29th, 2006: Will Your Vote Count in 2006?
"If Diebold had set out to build a system as insecure as they possibly could, this would be it," says Avi Rubin, a Johns Hopkins University computer-science professor and elections-security expert.
Techdirt, Wednesday, May 10th, 2006: Major Diebold Voting Machine Security Hole Discovered... Again

What next? A nationwide consensus on voter-verifiable voting machines?

Well, there was an election on the 16th, and I'll have to back-fill my report on that and what happened soon thereafter.

June

On Thursday, June 1st, our citizen volunteer voting oversight team had a press conference, expertly coordinated and led by Celeste Taylor of People For the American Way, on the steps of the City/County building. Here are temporary links to the VotePA-Allegheny contributions to the press kit: our press release and our report on iVotronic irregularities.

Press coverage:

Pittsburgh Post-Gazette: Groups cite flaws in electronic voting
WPXI TV: News Conference Addresses Voting Problems
KDKA TV/radio: Report Highlights Problems With New Voting System
Scoop: Voting Machine Problems Reported Statewide In PA

On Tuesday, June 6th, several of us spoke at the regular County Council meeting. One common theme was our observations of the vote total adjustment process we observed on Friday, June 2nd at the Elections warehouse on the North Side.

Press coverage:

Pittsburgh Post-Gazette: Allegheny County to examine vote machine concerns
Pittsburgh Tribune-Review: Group says vote tally unreliable
Pittsburgh Tribune-Review: County voting machines under review by state
New Pittsburgh Courier: Editorial: Accuracy of vote tabulation should concern us all

During the June 12th certification meeting of the Board of Elections, I attempted to obtain answers to a list of questions about the integrity of the voting process. After various Elections Division staff and ES&S employees answered some of my questions it became clear that it would take quite some time to address them all. It was decided that, in the interest of progressing the meeting, Tim Johnson would prepare a written response to my questions. He sent me a letter to that effect dated June 15th.

On Thursday, June 29th, I responded in writing to Tim Johnson's June 15th letter. In addition to expressing my interest in obtaining the Board of Election's report, I took the opportunity to point out that the previous two weeks saw a rising tide of public sentiment in favor of increased voting-system integrity in general and voter-verifiable paper records in particular.

By the way, a fun quote from the Common Cause report:

The [Nevada Gaming Control Board] concluded that the Diebold machines represented a legitimate threat to the integrity of the election process.

July

I heard nothing from the County throughout July.

August

On August 9th I spoke with Tim Johnson by telephone. This letter summarizes my understanding of the conversation. It is very difficult to feel that appropriate progress is being made.

On August 15th, a lawsuit was filed by a group of Pennsylvania voters and poll workers alleging that the certification by the Secretary of the Commonwealth, Pedro Cortés, of paperless DRE voting machines was in violation of the Pennsylvania Election Code. Here is the legal complaint.

On Wednesday, August 23rd I received a letter from Mark Wolosik in response to my June 12th questions.

On Friday, August 25th I commented on the Commonwealth's proposed amended HAVA compliance plan.

September

On Tuesday, September 26th, several of us spoke at the regular County Council meeting (after several hours of Council action on a county-wide smoking ban). I raised the issue of software audits and presented Council with a draft ordinance on software audits.

October

On Thursday, October 5th the Pittsburgh Post-Gazette printed a letter to the editor written by several members of the CMU Computer Science faculty: Most computer security experts agree: We need a paper trail (you will need to scroll down the page).

On Thursday, October 19th, the Pittsburgh City Paper ran a thoughtful and carefully researched piece by Marty Levine entitled Machine Nation.

On Tuesday, October 31st, one week before the November election, several of us attended the "Logic & Accuracy" testing of the ES&S M650 central-count optical scanner used to scan absentee, emergency, and provisional ballots. Our report on the testing details several disturbing security weaknesses surrounding the machines designated to run the Unity election vote-tabulation software for the election and a worrisome plan to re-wire the tabulation infrastructure the day before the election... which would result in the use on Election Night of a software configuration different from the one which underwent L&A testing.

November

Press coverage before the election:

Post-Gazette, November 2nd: Voting machines, key races cast election spotlight on Pa.
Post-Gazette, November 3rd: Fear of vote fraud: The Chavez connection is the least of our worries (weirdly partisan and weirdly silent on how Allegheny County could have chosen a system with a paper trail!)

On Tuesday, November 7th we had an election, which wasn't a widespread meltdown.

First, the party line:

Post-Gazette, November 8th, 2006: Problems with new voting machines scattered, minor
Scattered problems with voting machines were reported in Allegheny County and across the state yesterday, but elections officials said there were no serious malfunctions with the new touch-screen systems.
Twenty electronic voting machines were removed from polling places across Allegheny County yesterday morning because of technical glitches, and a polling site in Monroeville resorted to paper ballots for 45 minutes, elections officials said.
In some precincts, the balky machines were replaced with backups. In others, poll workers made do with the machines available.
[...]
In Allegheny County, some machines were not "zeroing out," or clearing the slate to start tallying votes, Mr. Wolosik said. The zero-vote count serves as a baseline to ensure no votes are stored on the machines before official voting begins.
Between 7 and 8 a.m. the county's election call center received about 150 calls. By late morning the calls had trickled down to 27 per hour midday, Mr. Wolosik said. He considers the new machines a success.
[...]
In Westmoreland County, a handful of polling places were late in opening because of programming errors in the machines, said Ted Kopas, chief of staff to county commission Chairman Tom Bayla. The problem was solved and all polling places were open by mid-morning.
Post-Gazette, November 27th, 2006: Pa. lacks paper backup to voting electronically
Allegheny County deployed more than 4,600 iVotronics on Nov. 7, experiencing minor problems. Some poll workers had trouble printing the "zero-tape," a piece of paper that shows the machines had no votes at the beginning of the day. Technical glitches forced a site in Monroeville to use paper ballots for 45 minutes.
"The difficulties that we had were minimal," said Kevin Evanto, a spokesman for county Chief Executive Dan Onorato. "We were very pleased with the performance of our ES&S machines."

That all sounds very encouraging, but it's far from the whole story.

At some polling places, it took long enough to start up the iVotronics that voters left or were turned away. For example, we have a reliable report of voters leaving the Vann School polling place (Pittsburgh, 5th ward, 4th district) when neither iVotronics nor emergency paper ballots were available.
Reliable information indicates that poll workers in Upper St. Clair (4th ward, 3rd district) were unable to collect and print vote totals at the end of the night, that the County's voting-machine mechanic was unable to access the vote data, and that all vote data for the polling place departed in the mechanic's car, outside the oversight of the poll workers.
A programming error meant that many or all iVotronics in Westmoreland County thought the election was over before it began, with the result that poll workers needed to manually reassure each iVotronic it should keep running, once per voter. The iVotronic user interface makes it easy to accidentally shut the machine down instead, and this happened in some locations.
A team of citizen observers at the Allegheny County tabulation site observed a variety of computer security vulnerabilities, iVotronic data-extraction errors, and the use of untested equipment and uncertified software (see December 4th below).
Sarasota County, Florida, which also uses iVotronic voting machines, experienced a massive and statistically implausible undervote in a congressional race (Sarasota elections office braces for 'intense' recount, State to audit flawed Sarasota County vote). Until we know what happened there it is hard to be confident that it won't happen here.

How many problems of what nature and severity were there? All we know at present is that the situation was worse than the impression provided by media accounts of government pronouncements. Hey, here's a radical idea: the Elections Division could tell the public what really happened, and the public could then decide whether or not it was smooth enough!

After the dust settled from the election the Post-Gazette issued a call for upgraded integrity: Help the voter: It's time to address electronic voting problems (November 29th).

December

The integrity issue is heating up fast. The more people look at paperless DRE's, the less widely they are loved.

December 2, 2006: Feds fear voting security breaches (better late than never!)
December 3, 2006: Forum: In praise of paper (by noted computer security expert and Brennan Center report co-author, Bruce Schneier)

On Monday, December 4th the Allegheny County Board of Elections met to certify the November 7th general election, and a variety of election integrity activists spoke.

Supplemental Material

This is a 60 Minutes piece on what we're facing.

Here's a good piece which points out that many DRE's can't be meaningfully "recounted": Does Every Vote Count?, Rebeca Chapa, Express-News, 4/09/2006.

Here's a speech by Oregon Secretary of State Bill Bradbury, given to the Oregon Association of Counties on November 17, 2005 (approximately when Allegheny County had its voting-machine fair). Observe how far ahead of PA on the curve OR was--they already understood that public confidence matters as much as the druthers of county election officials (!!!) and had set up a financial plan for funding automatic post-election audits (of their optical-scan system, already certified and selected, including a built-in voter-verifiable paper record). Oregonians got genuine state-level leadership. Why didn't we?

Here's a Rolling Stone interview with Reverend DeForest Soaries, the Bush-appointed head of the federal "Election Assistance Commission". Judge for yourself how well the federal government did at safeguarding your right to vote.