In the aftermath of the Florida recount debacle of the 2000 presidential election, the U.S. Congress appropriated billions of dollars for state and local governments to buy electronic voting systems. But in the years since, a string of problematic elections has led much of the voting public to join early critics in concluding that available machines are buggy, easily subverted, and impossible to accurately audit.
So perhaps it was only a matter of time before members of the open-source movement would enter the fray, with the claim that their kind of technology can guarantee free and democratic elections. Already, two bellwether states, California and New York, have taken notice. This spring, California’s state assembly considered a bill mandating that new voting systems be based on open-source software. The bill didn’t pass, despite support from the California secretary of state, whose office certifies voting systems. But at least one major (and for now undisclosed) California city is considering open-source voting. So the issue is likely to come up again. Meanwhile, New York’s state board of elections decided late last year to waive certification fees for open-source voting systems.
The catalyst for open-source voting is the Open Voting Consortium, based in Sacramento, Calif., which demonstrated its electronic balloting software in August at LinuxWorld in San Francisco, using it to take a straw poll for the presidential election and also to determine the conference’s best-in-show award. According to OVC president Alan Dechert, the vote tallied 816 ballots over 16 polling stations, using ballots that had to be created on the fly after the best-in-show finalists had been chosen.
Dechert admits it’s a long way from a straw poll to a state—or even a town—election. But, he says, open-source machines already have much of the functionality of closed-source ones, are more trustworthy, and are one-tenth the cost. A comparable closed-source voting machine costs US $4000, he says, ”and it does its job stupidly. With ours, the hardware costs $400 and the software is free.” The greatest advantage of open-source voting software, Dechert says, is that anyone who can read computer code can inspect an OVC system to ensure its security.
In the OVC system a voter chooses candidates on a touch screen and then prints a ballot, which shows those selections in bar-code form. The printout would be stored in a ballot box until the polls close. Then, with election monitors observing, the ballots would be hand-scanned using a bar-code reader. Proprietary systems by manufacturers like Diebold, Election Systems & Software, Sequoia Voting Systems, and Hart InterCivic typically tally votes as soon as the voters make their selections. This runs contrary to the OVC’s philosophy, which in Dechert’s words is ”You cast your vote in private, but it’s counted in public.”
The problem with the OVC model, says Rebecca Mercuri, owner of computer security firm Notable Software, in Philadelphia, is that while OVC may be more secure than today’s proprietary systems, it is no more secure than electronically scanning paper ballots.
Moreover, Mercuri says, there are problems that OVC wouldn’t necessarily solve. ”In poorer precincts or in precincts where there is some deliberate disenfranchisement going on, we find that the machines aren’t quite working properly,” resulting in long lines at the polls, says Mercuri, who has been called in as an expert in a number of elections.
Earlier this year, Mercuri taught a class at the University of Pennsylvania that built an OVC system from material available on OVC’s Web site and ran a mock election. The vote-counting database was buggy, she says. And OVC’s software for designing ballots was so complex that even her Ph.D. students were scratching their heads. Of course, as with any open-source product, the solution may be to simply enlist more developers to refine the code. However, Mercuri thinks the open-source model of continual refinement might not fit with a government certification process that takes months and hundreds of thousands of dollars.
Those problems aside, if open-source voting gains in popularity, says Mercuri, it could pressure proprietary voting-machine companies to open their systems to greater scrutiny. ”The viability of a real open-source product out there in the market would now kick open the door of the vendors who are saying we’re never giving out our source code,” she says.