By Michael Richardson
It happened at Harvard. Stanford University computer scientist David Dill was at Harvard's computer resource center talking about electronic voting machines. Dill, one of the nation's foremost "paper trail" voting machine advocates, is the founder of a lobbying group called Verified Voting. About ten minutes into his Power Point presentation to the assembled Harvard intelligentsia, Dill's laptop computer crashed leaving him without a script. The irony was unmistakable.
Dill then departed from his prepared remarks explaining, "I know so much I can't organize a talk." The next hour was devoted to a Q&A session that rambled in a self-contradictory trajectory revealing more about Dill than electronic voting machines.
Before the "glitch", Professor Dill was in full reformer mode and sounded pretty good. Dill explained he had spent his two decades at Stanford, "trying to check software correctness, but it's not something we can do."
The three big "unsolvable" problems with electronic voting identified by Dill were error, security, and making sure the system is running what you think it is running. "We can't prove correctness....We don't know how to make systems secure....Why do we even trust the hardware?" Dill warned, "It is wrong to hand control of elections to private companies."
Dill declared that any voting technology should be at least as trustworthy as hand-counted paper ballots, which he characterized as the "gold standard" for voting. We should "give up" on audits and instead "empower each voter to check their vote."
"We have made a mistake by focusing on technology; instead we should focus on procedure."
Then the questions started and Dill lost his way. After advocating for precinct optical scanners or printers on touch screen machines, Dill admitted that optical scanners do not always count the ballots correctly, "A careful hand-count is more accurate than optical scan." Touch screen printers were open to "nefarious individuals that could cause the paper record to be unreliable." Dill also admitted that self-deleting malicious code would not be detectable.
The Harvard computer experts in the audience got Dill to admit the push for electronic voting machines came from marketing by the vendors; that there were problems with machine certification standards leaving a "gaping hole"; that there was no way of testing for viruses; that the Election Assistance Commission is "highly politicized" and incapable of the tasks it is presented; and that, "What we have now are a bunch of bad voting systems."
Dill acknowledged that "vote-flipping" happened all over the country in the 2006 election and that it is an "insidious" phenomenon without explanation. Dill said that 1% percent audits are "frighteningly bad" in anything but a statewide race and that an "over-qualified janitor at an electronic voting machine vendor could rig an election."
After admitting that a "careful hand-count" is the most accurate and cheapest way to count votes and that optical scanners could be rigged and don't always accurately record the ballot entries of voters, Dill then advocated precinct-based optical scanners as his solution to the problem of election fraud.
After his confusing, contradictory talk, Dill was asked about his support for H.B. 550, a "paper trail" electronic voting machine bill pending before Congress. "That is my public position, although the bill is being rewritten and I don't know where I stand."
[Permission granted to reprint]