Catherine J. Gatian – 4/15/05

This page was last updated on April 16, 2005.

Paper trail is vital in elections; Catherine J. Gatian; Beaver County Times; April 15, 2005.

My point in this critique is not to defend the Beaver County electronic voting system, though I’ve used it in every election since I moved back to Center Township and I’ve had no problems. It even took my write-in vote for myself and the vote appeared in the official count on the county web site. For the previous 19 years I voted using punch cards. My point is to show some folks appear to be making statements not currently supported by evidence.

To be honest, I believe Ms. Gatian is merely passing along “information” she got from Ms. Sheila Green at the meeting noted in the letter. I visited Ms. Green’s web site on this topic and I believe she’s a little extreme. I agree a paper trail is desirable, but Ms. Green has turned what probably started out as a good cause into personal attacks on persons who don’t agree 100% with her recommendations.

Here’s one gotcha for Ms. Green. Back on February 23^rd, Ms. Green issued a “press release” stating that the person who provided the technical input for the Department of State’s report “(1) hasn’t yet been proven to be capable of assessing computer security in a sufficient manner, (2) wasn’t the choice of the people, and (3) will want to pretend that he hadn’t been mistaken in approving the defective system previously.” Unless she wants to be legitimately portrayed as a hypocrite, she must reject the report’s findings. As you can expect, though, Ms. Green supported the DOS report because she got the result – at least temporarily – she desired.

Should the DOS have decertified Beaver County’s voting system? I don’t have enough verifiable information to make that call. I find the timing troublesome, though. Decertifying a system the county used for seven years only 5½ weeks before an election seems to be a knee-jerk reaction. It’s possible switching systems in only 5½ weeks will introduce more short-term problems than alleged for the existing system. Some fairly simple short-term precautions could have addressed the security issues mentioned in the DOS report. This would have allowed the county to use the system for at least the May primary. After the primary, the DOS and the counties using the UniLect system could then have decided whether the short-term precautions were appropriate for the long term.

Below is a detailed critique of the subject letter.

“On April 6, Progressive Democrats of Beaver County sponsored a meeting at the Beaver County Courthouse regarding voting issues. The speakers were Sheila Green, whose legal challenge forced the state to re-examine the voting machines in Beaver County, and Mary Beth Kuznik, who was a coordinator of the Ohio recount.”

[RWC] A little nitpicking on my part. Remember, in this context, “progressive” means “liberal.” Talk about a redundant name for a group! It’s the equivalent of naming the group “Socialist Socialists of Beaver County.”

“The use of the UniLect Patriot Direct Recording Electronic voting system in Beaver County in the past contentious election caused 5.25 percent of the presidential votes to be lost.”

[RWC] I believe Ms. Gatian is jumping to a conclusion. It’s true there was an undervote for president, but there’s been no proof votes were lost or a voting system defect was the cause. All we know is the percent of undervotes for the Beaver County voting system was higher than the state average. An undervote is when fewer votes were cast for a particular office than the total number of votes cast.

Don’t get me wrong; it’s possible some undervotes could be the result of some unknown “bug” instead of voter intent or error. After all, none but the most trivial systems are “bug free.”

The bottom line here is that we need to understand what is causing “residual votes” before we draw conclusions. Unfortunately, it appears folks like Ms. Gatian and her mentors want to bypass this process.

“Computers are wonderful tools for tabulating data, but they have shortcomings. A computer can be physically interrupted by a magnet or power surge. The UniLect voting system provides no paper trail. There is no way to perform a recount.”

[RWC] Yes “a computer can be physically interrupted by a magnet or power surge,” but this can be easily addressed if it isn’t already. My own home PC has both surge protection and a battery backup.

While it’s true a vote-by-vote recount cannot be performed, I believe a machine-by-machine (precinct-by-precinct?) recount can be conducted because each machine – I think it’s one per precinct – prints multiple copies of a totals report for each candidate.

Ms. Gatian failed to note lever-type voting machines – in use since the 1800s – don’t provide a paper trail either. They and direct electronic recording (DRE) systems are similar in this regard.

“The hardware and software of the UniLect system are proprietary. Election officials cannot know or watch how the machines are tested because it is a secret. The results of testing are also secret. Why would the test results be a secret? The only reason I can think of is the machines do not record an accurate count.”

[RWC] It should be no surprise the Unilect system utilizes proprietary components, especially the software. Nearly every piece of software in existence is proprietary. That said, vendors routinely allow “high profile” customers to perform code reviews to help ensure there are no security-related “bugs,” “backdoors,” et cetera. I speak from experience.

When Ms. Gatian wrote, “Election officials cannot know or watch how the machines are tested because it is a secret,” to which “election officials” does she refer? I used to be in the software business and I have a hard time believing the vendor wouldn’t permit the customer to understand the testing process and view the results. Reputable vendors tend to be proud of their quality assurance processes.

Ms. Gatian failed to note Pennsylvania has its own test for verifying a voting system’s accuracy. Ms. Gatian also failed to note the system correctly counted all votes during this test, even during the most recent test cited in the Secretary of the Commonwealth’s report of April 7, 2005.¹

“Concerns were validated on April 7 when the secretary of the commonwealth decertified the UniLect Patriot voting system.”

[RWC] Intentionally or not, this is a misleading statement. Though the report cited concerns, it noted the system correctly counted votes during the tests. I believe Ms. Gatian wanted us to believe her allegations about an inaccurate count were validated.

As a reminder, this is the same office that certified the UniLect system in the late 1990s.

“One solution to this problem would be using a paper ballot the voter can review before leaving the booth. A study by MIT reported that paper ballots are still more reliable than machines. A more expensive electronic solution would provide the voter with a printed bar code that could be used to verify the ballot on line.”

[RWC] This sounds as if Ms. Gatian wants to turn the electronic voting machine into little more than a fancy “pencil” that simply prints a paper ballot with our choices but stores no counts electronically. Vote counts would then need to be tallied manually or via scanning. Using Beaver County as an example, in addition to the current hardware we’d need to buy as many printers as we have voting booths (to ensure vote secrecy) and some number of ballot scanners to count the votes. I could be wrong but I believe the cost of this approach would make the electronic voting machine extinct.

Here are some more points Ms. Gatian failed to note from The Caltech/MIT Voting Technology Project.² In the 2000 elections, paper ballots represented only about 1.3% of all votes cast. In the 2000 vote, residual votes for DRE systems were actually lower than for punch card ballots. Residual votes for DRE systems fall almost exactly in the middle of the range between optical scanning systems – which had the least residual votes – and punch cards. The reliability difference between the best and worst systems was only 1.5% of votes cast.

“All citizens should be confident that their votes are counted. This is a national issue, and there is active resistance to paper verification. Contact your county commissioners and demand paper verification.”

[RWC] Don’t get me wrong; I don’t resist paper verification. I think it’s a fine idea. That said, I do wonder why it suddenly became such a big deal. Remember, mechanical voting machines – dating back to the 1800s – provide less of a paper trail than today’s DRE systems yet I can recall no paper trail pushes for that system. In any case, I don’t believe we should “throw out the baby with the bath water.” We should fix any meaningful flaws that affect an accurate vote count.

If the “only a paper ballot will do” folks really do have an altruistic objective – a secure election, I believe they could be making the classic mistake of assuming only their solution – a paper ballot – will ensure an accurate vote count. I used to be involved in software development. Too many customers want to tell you how to solve their problem instead of allowing you to develop alternative approaches before you and they select the optimal solution. As an example, think of a customer who doesn’t just want you to build an airplane, the plane must propel itself with flapping wings.

That said, I can’t help but feel there is some other agenda behind the “only a paper ballot will do” movement, especially in the case of Ms. Green. After all, Ms. Green alleges she has “decades of experience with management of computer systems and teach[s] college courses on the subject.” Anyone knowledgeable of computer systems and available hardware and software security techniques will tell you that electronic voting systems can be made secure – if they aren’t already – without neutering them.

There are a couple of other things that make me suspicious. First is the rhetoric Ms. Green uses on her web page and in “press releases.” If you disagree with her, you are corrupt, “incompetent,” “ignorant,” and/or “lying.”

Second, consider the following response from an “only a paper ballot will do” proponent to a DRE that allegedly does exactly what Ms. Green allegedly wants. “But, they will still be a Trojan Horse. We learned in Ohio that simply denying precincts sufficient machines, resulted in voter suppression.” This is from a web site Ms. Green cites as a reference. At the top of its page, the site opines that electronic voting machines are perhaps “why the Republicans control the White House and both houses of Congress.” Don’t you just love conspiracy theories? For some people, I guess it’s better than facing the truth.

1. Reexamination Results of UniLect Corporation’s Patriot Direct Recording System; Commonwealth of Pennsylvania - Department of State; April 7, 2005.

2. Residual Votes Attributable to Technology; The Caltech/MIT Voting Technology Project; March 30, 2001.