Over the past few years, security researchers have also reported various flaws in e-voting systems that they have claimed make the systems easy to compromise.
One of the most sensational was a report by researchers at Princeton University that showed how attackers could install vote-stealing code in an electronic voting machine in less than a minute.
"The problem when you are dealing with pure software is that you really have no way to verify if the software has been operating correctly," Lipari said. "We know software can be hacked and that it has flaws and that programmers make mistakes."
"We see it happening every year, and there is no way to verify the results you are getting from the software. You've got to trust it," he said.
Since the Help America Vote Act (HAVA) was passed in 2002, most states have moved to systems that have some sort of a verifiable paper record, Lipari said. Some systems require voters to mark ballots and then scan them into an optical reading device, while others are direct-recording systems that generate paper copies.
"We have seen a lot of progress over the last four or five years," Smith said. "But we still have nearly a quarter of the voters using some kind of paperless voting system."
Merle King, executive director of the Center for Election Systems, a cooperative venture between Georgia's secretary of state's office and Kennesaw State University, refuted the notion that paperless systems are inherently unreliable.
Most arguments for paper-based systems are based on the assumption that credible auditing has to always be centered on paper, he said.
"I think the notion that electronic systems are not auditable would come as a shock to every accounting firm, every auditing firm, the federal government, the airline industry and all who have paperless systems," King said. "The notion that paper equals auditability is old-fashioned at best and ill-informed overall."
Most election fraud has historically happened with paper ballots, which even today is more prevalent with mail-in absentee ballots, he said.
Likewise, concerns about the security of e-voting systems are somewhat misplaced, King argued. In many cases where researchers have broken into voting systems, the models that were used to simulate hacking have been divorced from how the systems are used in an actual election, he said.
The real question should not be about just how well-protected e-voting systems are, but rather how easy it is to detect tampering, King said.
Sign up for Computerworld eNewsletters.