July 13, 2004. Computer Ate My Vote. National Day of Action.
Speech by Ellen Theisen, presented at the event in Everett, Washington.

If we bought a car that had never undergone any worthwhile testing to see if it worked well, we would never think of trusting it to drive us cross-country. Yet, even though everyone now agrees that the testing process for these paperless electronic voting machines is completely inadequate, election officials across the county - and in Snohomish County - are going to trust them with this crucial November election.

Less than 3 weeks ago, a United States House Science subcommittee held a hearing focused on the testing and certification of voting machines. The purpose was to discuss what is currently being done, and how much can be improved before the November election.

In addition to three members of the House Science committee, those present included an array of experts on the topic:

Representative Rush Holt.
A Vice President of one of the three Independent Testing Labs approved by the National Association of State Election Directors (NASED).
The acting director of the National Institute of Standards and Technology (NIST).
The chair of the NASED Voting Systems Board, which selects the testing labs.
Michael Shamos who was the statutory examiner of electronic voting systems for Pennsylvania for 20 years.

Shamos' testimony was the clearest,* and nobody at the hearing argued with him. On the contrary, it was clear that they all agreed. Here's what they agreed:

The testing and certification process is not just broken, it's virtually nonexistent.

The ITA process is dysfunctional, as can be seen by the fact that all the broken and buggy, malfunctioning systems have been qualified by ITAs.

The standards are totally inadequate. All the machines currently in the field have been tested by 1990 standards. That's standards that were developed before Windows. But all the systems use Windows. So these Windows systems were qualified using pre-Windows standards. Yes, dysfunctional.

But there's more. The vendors contract with the testers, and there isn't any government oversight at all. The testers report their findings to the vendors - not to the public, not to the election officials, but to the vendors.

And there's still more. According to the President of MicroVote, the vendors provide the test plans for the testers, and the testers don't test outside those plans. And there is nothing in the standards that says the machines have to work.

So we have obsolete standards, applied to the vendors' equipment by testers who are being paid by the vendors, doing test plans supplied by the vendors, reporting the results to the vendors and only to the vendors. At this point, "dysfunctional" seems too mild a word.

But at least the testing authorities test for viruses (you know, malicious software). WRONG. It's up to the vendors to ensure that their secret software doesn't contain any malicious code.

Well, you might think it's better at the state level. But according to Shamos, state certification procedures are even less adequate, and he should know. He was in charge of them in Pennsylvania for 20 years. He says that since NASED set up the national qualification process, states just rely on it. Yes, they rely on a dysfunctional process to ensure the voting rights of their citizens.

Then Shamos dropped yet another bomb. The testing done at the local level is the worst of all. And, of course, everyone in the hearing is nodding in agreement. And they're right. The ballot programming, which maps your touches on the screen or marks on a ballot to the vote counts in each candidate's column, is done at the county level. It's new and unique for every election, and no independent tester ever examines it. Errors in the ballot programming cause votes to be miscounted or counted for the wrong candidate. And the testing for the ballot programming varies from minimal to none at all.

Worse, lots and lots of counties have the vendors do the ballot programming. And lots of those counties get the test ballots from the same vendor. And lots of these elections have been disastrous - precisely because the ballot programming was done wrong. Why wasn't it caught in pre-election testing? Well, whatever the reason, it wasn't.

You're lucky in Snohomish County. The county does the ballot programming, and they are very careful about it because they know how important it is. And the county tests the DREs much more rigorously than many other DRE counties. But they don't enter 50, or 30, or even 10 ballots of every ballot style and make sure the results are what was expected, even though this type of testing is done on optical scan machines. Ballot programming errors could easily slip through the pre-election testing unnoticed. And then, the results of the election could be wrong and there would be no way to ever find out.

Okay, the last bomb from the hearing. Representative Ehlers asked Shamos, what could be done before the November election to improve the testing and certification process. Shamos said, "Nothing." Specifically, he said, "I do not believe that Congress can act meaningfully in the 130 days that remain before the 2004 election. Even if it could, the states would be powerless to comply in so short a time."

So now we know. In this historic election, all across the country, we'll be using computerized voting machines (optical scan as well as paperless) that have gone through a non-existent, dysfunctional testing and certification process. These machines have shown themselves to be poorly designed and poorly constructed in election after election after election. It's become clear that they are constructed from hardware that isn't up to the job, software that has never been examined by anyone other than someone the vendor paid to examine it. They count our specific votes with ballot programs that are barely tested and never audited.

Many election officials feel pressure to "get with the 21st century." In the hearing 3 weeks ago, Representative Gilchrist mentioned that people resisted electricity when it was new and implied that our resistance to paperless voting is just resistance to newness. We know that's not true. And what's the problem with providing a backup? Why risk so much, when we have seen the evidence against these machines unfold faster and faster before our very eyes?

When my great grandparents wired their home for that new-fangled invention - electricity - they kept their kerosene lamps ... just in case.

When my grandparents bought their Model-T Ford, they didn't sell their horses.

When I got my first dishwasher, I kept my dish drainer for a while.

And when I got my first computer, I kept my typewriter close at hand. In fact, there were quite a few times I was glad I had it.

Running an election, without paper, on voting machines that have never been properly tested and which have proven themselves to be unreliable time after time is like jumping out of an airplane with a parachute you know might not open - and not bothering to wear a backup parachute.

Notes: * Testimony of Michael I. Shamos Before the Environment, Technology, and Standards Subcommittee of the U.S. House of Representatives' Committee on Science. June 24, 2004.

Those present at the hearing.
Representative Vernon J. Ehlers, Michigan
Representative Mark Udall, Colorado.
Representative Wayne Gilchrist, Maryland.
Representative Rush Holt, New Jersey.
Carolyn Coggins, Vice President, SysTest Labs (Independent Testing Authority).
Dr. Hratch G. Semerjian, Acting Director, National Institute of Standards and Technology.
Michael Shamos, faculty member of School of Computer Science at Carnegie Mellon University in Pittsburgh, formerly the statutory examiner of electronic voting systems for the Secretary of the Commonwealth of Pennsylvania.
Thomas R. Wilkey, Former Executive Director, New York State Board Of Elections, Chair NASED Voting Systems Board.