"We have all kinds of evidence of machines misbehaving," says e-voting expert.
While political scientists likely will find additional evidence of voting problems in last week's presidential election, it is unlikely to overturn the outcome. But that should not dissuade government officials and others from doing all they can to improve a system full of vulnerabilities before the next election, an e-voting expert said Tuesday.
"We have all kinds of evidence of machines misbehaving," Dan Wallach, an assistant professor of computer science at Rice University in Houston, said at a computer-security conference sponsored by the Computer Security Institute. "[But] what really keeps me awake at night is the notion of undetectable tampering."
Wallach said he favors optical-scanning machines, which he said are more accurate than electronic punch-card systems. But paper receipts are essential to ensure that votes are tallied the way they were cast, he said.
If software is buggy, there is no fallback for counting votes, as seen in some precincts last week. Right now, Americans must use a "faith-based system," trusting that their votes are properly recorded somewhere, he added.
Currently, only three independent testing authorities are given the voting machines' software code, which some critics argue is counterintuitive for an open election system. The authorities release only a decision that a system meets Federal Election Commission standards.
Those standards need to be raised, Wallach said, adding that a more complex method of certification is needed and that the code needs to be more secure.
He said a vendor "absolutely" could hack into a system and described various ways that fraud could be perpetrated and concealed. Voters and election officials also could tamper with systems, he said, and logic and accuracy tests of voting machines could be faked.
Voting methods are subject largely to state laws, and systems vary by state and even county. Federal or state legislation is needed to outline e-voting "best practices," Wallach said.
He proposed several improvements, including printing optically scanned paper ballots on special paper that is physically placed in ballot boxes by voters, or using a Brazilian system of letting voters see paper ballots beneath glass but not touch them.
Some ways of checking the veracity of last week's voting were employed. For instance, several states used forms of "parallel voting," in which voting machines were randomly pulled in some precincts and tested during the day.
Wallach also recommended using special voting computers less susceptible to hacking than the ordinary computers used for tabulations.
He said there is "a lot of political pressure" from disaffected Democrats to change the system. But he added that voting is "not a partisan issue" and that many election officials are "explicitly nonpartisan."
He said additional funding might be needed to make the changes because much of the money under a 2002 federal election law already has been spent on the existing, flawed machines. But he said there are ways to engineer cheaper -- and better -- machines.
NEXT STORY: Agencies in race to smash supercomputing records