Princeton Report Rips N.J. E-voting Machines as Easily Hackable

Vendor challenges research, says its machines are safe and reliable

With eight days to go before the presidential election, a report has been released by Princeton University and other groups that sharply criticizes the e-voting machines used in New Jersey and elsewhere as unreliable and potentially prone to hacking.

The 158-page report, which was ordered by a New Jersey judge as part of an ongoing four-year legal fight over the machines, says the e-voting machines can be "easily hacked" in about seven minutes by anyone with basic computer knowledge. Such hacking activity could enable fraudulent firmware to steal votes from one candidate and give them to another, the report said.

The controversy involves the Sequoia AVC Advantage 9.00H direct-recording electronic (DRE) touch-screen voting machines made by Oakland, Calif.-based Sequoia Voting Systems.

The report comes amid news stories in at least three states -- West Virginia, Texas and Tennessee -- where voters have told local election officials that they believe the e-voting machines they used tried to "flip" their votes to other candidates.

The AVC machines can be hacked by installing fraudulent software contained in a replacement chip that can be installed on the main circuit board, according to the report. Such a part replacement is very difficult to detect, it noted.

Andrew Appel, a Princeton University computer science professor who is one of the authors of the report, said that such security vulnerabilities cause doubts about the accuracy and reliability of the machines.

The plaintiffs, a group of public interest organizations, argue in their lawsuit against the state of New Jersey that the machines should be discarded because they can't meet state election law requirements for security and accuracy. State officials who back the machines argue that the machines are adequate for the job.

The lawsuit is expected to go to trial in January, but in the meantime, the court allowed the Princeton report to be released to the public.

The report gives details on how the machines could be manipulated by someone who wanted to change the results of the election, and it strongly criticizes the designs and security of the devices.

At the same time, Appel said that while such a scenario is possible, "it doesn't mean that somebody is dishonest enough to do it."

"Even so, it's an unpleasant place to be in to have to use these machines that are so hackable," Appel said. "Early next week, I'm going to have to go out and cast my vote on one of these machines."

The problem, according to the report, is that there are many opportunities in the storage, distribution and deployment of the DRE machines where an unauthorized person could manipulate them and not be detected.

"Somebody could have hacked it at any time" during those stages before an election, Appel said.

Michelle Shafer, a spokeswoman for Sequoia Voting Systems, said in an e-mailed response that the company emphatically denies the conclusions of the Princeton report.

In a 19-page response posted on Sequoia's Web site, the company argues that the researchers who contributed to the report removed factory security hardware from the tested machine before they performed their analyses. The Sequoia response also says that an operator panel cover was not in place when the testing was conducted, which would have made a potential attack "far less likely to succeed before they are stopped, or at minimum, detected."

The Sequoia response also harshly criticizes the researchers as having an "inflammatory tone" about the company's DRE machines, while "editorializing on the wonders of paper ballots and optical scanning" as an alternative and more trusted method of voting. The company said the Princeton report includes "numerous factual errors and cases of intellectual dishonesty."

Appel defended the report's conclusions.

"There's no perfect technology [for e-voting], but I think the consensus of computer scientists is that precinct-counted, optically scanned paper ballots is the best method" in terms of reliability and accurate recount auditing, Appel said. "The voter fills out a paper ballot that is scanned and counted in their precinct. You have your numbers right at the close of the polls, with two independent records -- the computerized numbers and the pile of paper ballots in the sealed ballot box."


Copyright © 2008 IDG Communications, Inc.

Microsoft's very bad year for security: A timeline