On Sept. 22, Carnegie Mellon University was the epicenter of the nation's electronic voting debate, as one its most prominent critics, David Dill, spoke at the invitation of Computer Professionals for Social Responsibility. But CMU's resident computer voting supporter, Michael Shamos, still thinks paperless voting is the wave of the future.
Since 2000, county election officials have been hustling to avoid Florida's infamy. Out with the old, in with the new. Old, of course, means the disgraced punch-card machines -- which will be used in 11 Pennsylvania counties this year, including Butler and Washington. In many cases, new means electronic touchscreen voting machines, also known as Direct Recording Electronic machines (DREs). Greene and Beaver counties will use these, as well as six other counties statewide, including Philadelphia.
This November, Allegheny County will use the old lever machines, which count votes mechanically, like an odometer. It'll be the last hurrah for these old gals, at least in a federal election. By 2006, says Pennsylvania Department of State Spokesman Brian McDonald, the counties still using lever machines or punch cards will have to buy new machines -- possibly DREs -- to comply with the federal Help America Vote Act of 2002.
Though DREs seem to eliminate human screw-ups like impregnating a chad, as well as Tammany Hall-style election tampering, critics like Dr. David Dill, a computer scientist from Stanford University who founded VerifiedVoting.org, worry that the average voter won't believe his vote is being counted correctly.
An expert in software bugs, Dill turned to politics when a conspiracy theory about electronic voting machines -- that a certain manufacturer contributed to Republican candidates and so might want to steal an election digitally for them -- "couldn't be disproved," even though he found no evidence supporting the theory.
"It's not sufficient for elections to be accurate," Dill says. "We have to know they're accurate. Democracy rests on people accepting the results of the election." Suspicion can undermine public confidence -- even if it's wrong.
DREs require a leap of faith -- their vote-counting and re-counting are both electronic, and thus not observable by voters. A flaw that changed only one percent of the votes might not be obvious -- and it could decide a close election.
Dill prefers the less expensive technology of optically scanned paper ballots, which voters fill out like standardized tests. They should be scanned at each precinct, so a voter can create an electronic record and have a chance to correct an "unscannable" ballot. Though computers count the votes, the fundamental records remain paper ballots.
But Dr. Michael Shamos, Carnegie Mellon-based computer scientist and longtime researcher in electronic voting, believes the recent alarm over the security and integrity of electronic voting machines is overblown.
"Based on a purely hypothetical look at voting machines," Shamos says, "these people who have very little expertise with the conduct of elections have decided that these machines have security flaws that are so severe the machines should not be used." Electronic voting machines in various forms have been used for 25 years, he says -- though never before as widespread as today -- and there's been no known fraud. If fraud were possible we'd have seen evidence by now, Shamos says. As for bugs, good advance testing can flag them.
Electronic voting's greatest advantage, Shamos says, is that it eliminates history's number-one source of voter fraud: human beings. "We have a long and well-known catalog of techniques for tampering with the integrity of a paper trail [in elections]. And paper-fiddling didn't stop in 1870. The idea that in 200,000 precincts in this country with 1.4 million pollworkers, that somebody is going to be able to insure the integrity of these paper ballots is a pipe dream."