by Ari Armstrong
The following article originally appeared at Boulder Weekly on October 14, 2004.
Al Kolwicz is a royal pain in the ass. He's also often right about voting systems. We all remember with nausea the Year of the Dangling Chad. Yet 2004 might produce even more ambiguous election results. Getting rid of punch-card systems is great, but not if they're replaced by "direct record electronic" (DRE) voting. Todd Engdahl of the Denver Post (citing the New York Times) claims a third of American voters will use DRE machines this year, including Colorado voters in Arapahoe, Denver and Jefferson counties.
The basic problem with DRE voting is that it's impossible to verify. The hardware or software could malfunction. Saboteurs or hackers could undetectably alter the results. A digital voting machine that spits out a duplicate paper receipt is worthless, if the vote is still counted from the digital source. How can we know what the paper says matches what was recorded inside the machine or counted later?
Boulder County officials are to be commended for going with a system that provides more substantial records. Yet, to Kolwicz, not even Boulder's system provides verifiable voting. Is Kolwicz Boulder's Cassandra or Don Quixote?
Before the primary vote, county officials conducted a "logic and accuracy" test of the voting machines. Kolwicz, the Republicans' official representative (before he was replaced), marked ballots with a check, "x," or circle rather than fill in the box completely. Tom Halicki, Boulder's election manager, said his office was fine with that.
However, according to Halicki, Kolwicz also duplicated some of the test ballots, marked out bar codes, put food on some of them, and wanted to run ballots in a certain order on certain counters. Halicki said Kolwicz "had a fundamental misunderstanding" of the purpose of the test.
A ballot with food on it -- a risk only for absentees -- doesn't get put through the counting machines. Instead, it is assigned to a duplication team, consisting of party appointees, and reproduced on a new ballot.
Halicki said some testing was done for bar-code problems. The counting machines are able to report certain sorts of potential errors, which prompts a review by a team of people.
Kolwicz was ejected from one of these testing meetings. Then, the sheriff's office referred criminal charges to the district attorney, who wisely declined to prosecute. Halicki said the basis for the charges was Kolwicz's obstruction of the testing process. Halicki also suggested Kolwicz ought not to have duplicated test ballots, even though he did so for purposes of the test. But even if Kolwicz made some inappropriate suggestions for testing the machines, that hardly justifies criminal charges, which smacks of petty retribution.
After the primary vote, the canvass board met to review it. Kolwicz was again the Republican representative.
Halicki said that, prior to the meeting, he would have bet money that Kolwicz wouldn't sign off on the canvass report. Indeed, Kolwicz refused to do so, and the Secretary of State's office finally told the board to proceed with their report absent Kolwicz's signature. (This is according to both Halicki and County Clerk Linda Salas.) Kolwicz said the canvass board didn't have the authority to submit a report without his signature, and he filed a complaint with the Secretary of State. Kolwicz claimed the board was unable to verify either the number of ballots printed or the number cast. Halicki said both of these numbers are available.
Another activist, Sheila Horton, said she tried to record a canvass board meeting, but she was prevented from doing so. "There should not have been any reason that it was not acceptable to record it," she said.
Kolwicz complained Boulder voting is not anonymous because a bar code accompanies each ballot. Halicki countered that this doesn't reduce anonymity, and the bar code is necessary for identifying and verifying ballots the machines have trouble counting.
Another problem with Boulder's system, Kolwicz alleged, is that there's no way to make sure the final vote tally is correct. Halicki said that issue was addressed in the logic and accuracy test.
I asked Halicki what would happen if I believed the tally was compromised and I wanted to personally recount all the ballots. He said I could not have access to the digital reproductions of the ballots because of contractual obligations with suppliers, nor could I look at the actual ballots. (The primary ballots are "locked away," he said.) However, he added, if I filed a legal challenge, a court might allow me to see both the digital files and the original paper ballots. Knowing those ballots exist, somewhere, even if they're hard to get to, makes me feel a lot better about the process.
I asked Halicki why all the ballots can't simply be posted on the Internet. What's more verifiable than that?
Kolwicz described his ideal system. You go into a booth and vote by touch-screen. The blind can talk to the machines. Then, the machine spits out a completed paper ballot. The voter reviews the ballot and casts it. And -- this is the most important part -- the vote is tallied from the paper ballots.
On related matters, I agree with Kolwicz that mandatory mail-in elections are a bad idea because they reduce anonymity and increase the risk of undue influence. Also, voter fraud is a perpetual problem that can be solved only with robust prosecution.
If you think Kolwicz is paranoid, remember that just because you're paranoid doesn't mean they're not out to get you.