American Prospect articles on all-mail voting in May 2006 issue

[eridani]

To send a letter to the Editors: letters {at} prospect.org.

http://www.prospect.org/web/page.ww?section=root&name=New+Ballot+Box

My response--

Your articles in the May issue on all-mail voting covered most of the critical issues raised with respect to election integrity with a couple of glaring exceptions. For one thing, there was not one single word about oversight of the actual tabulation process.

Optical scanning uses software that is proprietary, and that is checked only with functional testing (misnamed “logic and accuracy testing” by vendors specifically to bamboozle people who know nothing about computer security). The purpose of functional testing is to verify that the software carries out the tasks required in its specifications. Even the best functional testing cannot detect software which is time or event triggered. Real security testing is an entirely different thing, and is not even required for software certification.

There is an old joke about an engineer, a physicist and a mathematician each discovering a fire in their hotel room wastebaskets. The engineer uses the room fire extinguisher to put out the blaze, and empties the rest of its contents onto the wastebasket and the surrounding area as an extra safety measure. The physicist does a quick test blast of the fire extinguisher, does a series of calculations on his laptop, and then aims a precisely timed squirt at the base of the fire which is exactly enough to put it out. The mathematician looks at the fire, goes into the bathroom, runs the water, dabbles his hand in it and says “Aha! A solution exists!” and then goes back to bed.

Elections officials currently act like the mathematician, when they really need to act like the engineer. It is a good thing to have paper ballots that can be audited, but it is utterly useless to have tabulation processes that are auditable, if THEY ARE NOT ACTUALLY AUDITED. Not a single one of your articles discussed auditing.

As an analytical chemist, I am delighted to not have to analyze chromatograms that are a series of peaks on a strip chart by measuring areas of standard and sample peaks by hand and doing the standard curves by hand on graph paper. However, just this week, my computer refused to run one of my system pumps, and I had to delete and reinstall a whole new configuration. The software running that particular configuration had somehow been corrupted. The same software package does all my calculations automatically, and I have no guarantees that the occasional power failure or just generic bit rot is not corrupting my final results. For that reason, I periodically check the system by doing manual integration and calculating results directly from initial peak areas, sample weights, etc.

I flat out do not believe that vote tabulation software is any better than my chromatography software—in fact, it would just about have to be worse simply because it isn’t used anywhere near as often. Even in the absence of malicious or buggy code, bit rot can compromise the results. Where are Oregon’s checks and balances? As computer security expert David Dill has said “It is not enough that election results be accurate. We have to KNOW that they are accurate, and we don’t.”

In addition, the articles all assume that signature checking is an adequate guarantee that someone’s vote will actually be counted, when all that actually happens is that people can find out whether or not their ballots were received. None discuss what happens to ballots that optical scanners refuse to count. This can be from stray markings on the ballot in the wrong place, checkmarks or “x” being used instead of filling in the oval completely, people with shaky hands having problems keeping their pens within the oval, having the ballots partially shredded by the post office, etc. As an elections observer in 2005, I can personally testify that the scanners rejected quite a few ballots for no reason whatsoever that was visible to the naked eye. Are people with mobility impairments disenfranchised without their ever knowing about it?

This is not the case in Washington State, where voter intent is the rule even if said voters are jerks who insist on writing their comments on the state of politics and the world in general (or displaying their talent as editorial cartoonists) in the margins. The only thing that will disqualify a ballot in Washington is actually signing it. This is where poll voting is far superior to mail-in ballots. People can be sure their ballots are counted, because if they are rejected by the polling place scanner, they have the option of getting another ballot and trying again.

In Washington, what happens to unscannable mailed ballots is that election workers duplicate the ballots that are not accepted by the scanners. (Up to 2004, the ballots were enhanced by whiting out stray marks and darkening checkmarks. This practice was changed after the close gubernatorial race that year. As a scientist, I strongly approve of keeping the originals around and linking them to the duplicates with a handwritten record of ballot numbers as is now done.) However, this can be an extremely expensive and time consuming process, as extra ballots (of 1000+ different formats in King County) must be printed and logged in. Is this done in Oregon, and how much extra expense does this add to their process? If it isn’t done, how many voters are disenfranchised without their knowledge?