|
lacked the cryptography necessary to protect key information from hackers. As a result, they said, a hacker could vote as many times as he wanted or tamper with the machines to divert votes cast for one candidate to someone else. Diebold said the scientists had inspected an old version of the software and defended their machines as safe from tampering or error.
Election officials can ask the machines to recount the votes, but if a software bug or a hacker changed the results before they were counted, the machine would spit out the same incorrect tally the second time around. It's like asking a bank to recalculate your monthly statement without having evidence that all your transactions were properly recorded.
Computer scientists say the answer is simple: allow the machines to print paper receipts that could be treated like ballots in a recount.
"We wouldn't trust the software if we wrote it ourselves," said Eugene H. Spafford, a Purdue University professor and executive director of the Center for Education and Research in Information Assurance and Security. "Having paper or some other kind of permanent record gives us the ability to independently perform a check on these machines."
Nevada is the only state that has mandated printers for its touch-screen systems by November. California and Ohio have required them to be in place by 2006, and lawmakers in 21 other states are considering similar action, according to Electionline.org, a Washington group that follows election reform.
Though some people call them "receipts," they're not like the transaction records produced by ATMs. Voters can't take them home, or even touch them.
Here's how it works: After the voter chooses candidates by touching the screen, a printed record of the choices appears under clear plastic or glass so he or she can verify that the computer recorded the selections correctly. The machine will still tabulate the votes electronically. But if a malfunction is later discovered, or if the vote is close enough to warrant a recount, election officials can pull out the paper for a hand tally.
Joined by former presidential candidate Howard Dean, members of Computer Professionals for Social Responsibility, the Assn. for Computing Machinery and hundreds of other advocates have rallied across the country in support of printers, which would cost $500 to $1,000 to add to each machine.
Engineers are also working on a hybrid: a touch screen whose purpose is not to count votes but simply to print ballots that would be read by traditional optical scanning machines, which tally votes by counting ink or pencil markings.
But some advocates of electronic voting say printers are an unnecessary expense and could be troublesome to the poll workers who would deal with paper jams and other malfunctions. The four congressmen who wrote the Help America Vote Act, a 2002 law that gave states nearly $3.9 billion to upgrade voting systems, said in a note to colleagues that printers "would essentially take the most advanced generations of election technologies and systems … and reduce them to little more than ballot printers."
Some computer scientists insist that adding expensive printers to existing machines is better than doing nothing. Even better, they say, would be to set the high-tech systems aside and stick with optical scanners.
Studies of past elections concluded that optical scanners were more accurate than touch screens. One survey by the Caltech/MIT Voting Technology Project found that optical scanners had a 1.6% error rate in U.S. presidential elections from 1988 to 2000, compared with a 2.2% error rate for electronic voting machines and 2.6% for the most common punch-card system.
"People often look for a quick technological fix to a complex social problem," said Will Doherty, executive director of the Verified Voting Foundation, a nonprofit started by a Stanford University computer science professor. "I don't believe those machines are the wave of the future."
In Florida, state officials found that electronic voting machines were three times as likely to record "undervotes," or ballots cast that didn't register a vote, compared with optical scanners during the 2002 gubernatorial election. The South Florida Sun-Sentinel found that electronic voting machines were six times as likely to record undervotes in the presidential primary in March.
|
|
