by Rhona Mahony. In three weeks, Americans will elect a new President, They’ll also elect new Senators, Congressional representatives, and many state and local officials. Voters in six U.S. states, though, will vote on “direct-recording” electronic (DRE) machines that produce no paper print-out that can be used to double check the accuracy of the machine. Voters in 29 other states may get a paper print-out but, like those in the paperless states, will have no way of knowing how error-prone or easy to manipulate their DRE voting machine is. (See VerifiedVoting.org.) Independent tests of voting machines–done outside the closed labs of the manufacturers–have not been encouraging. Last year, Debra Bowen, California’s Secretary of State, asked computer scientists from the University of California, Berkeley, to help her staff do a “Top-to-Bottom Review” of many of the voting machines that we have been using in California. The result? Ms. Bowen’s team found that machines from Diebold (now Premier Election), Hart InterCivic, and Sequoia were so inaccurate or so insecure or both, that they have “decertified” the machines. If you don’t live in California, you may find yourself looking at one of those duds on November 4.
Better Security with Transparent Software
Should we throw our votes into a black hole? Should we let vote-stealers snicker at us? There may be a better way.
Computer engineer Alan Dechert and his colleagues are offering a system that they call “Open Voting.” It prints out a paper ballot that the voter can read over herself. The ballot has a bar code on it that the polling station’s bar-code reader can count quickly. Third and best of all, the software that runs the system is not secret, like the software running the machines sold by Diebold, Hart Intercivic, and Sequoia. Dechert and his colleagues have published much of it at SourceForge.net. That’s why the software is called “open source”; the source code–the program that the programmer wrote–is open to inspection. Any and all of the tens of thousands of people around the world who learned the Python programming language in high school, college, or later can check over the program for mistakes and security weaknesses. Go ahead; take a look at it. As crytographer Bruce Schneier has written, anybody can invent a system that he himself can’t crack. You don’t know whether other people can crack it until you give them a chance. By publishing the software program and inviting comments and cracking attempts, the writers get to improve the program. This world-wide, collaborative improvement is a virtue of open source software. For full credit, though, Dechert and his buddies will have to publish the full version of the program.
The Open Voting system has one more advantage; it is cheap. Like most authors of open source software, Dechert, et al., are not selling the program. They are giving it away for free. Moreover, no special, proprietary machinery is necessary to run it. It runs on off-the-shelf touch screens, bar-code readers, computers, and printers that many different manufacturers sell. Your county can shop around for the best deals it can get on those machines, on-line and at the local shopping center.
Is It as Easy as They Say?
I had a chance to vote on an OVC machine in August, 2008, at the LinuxWorld convention in San Francisco. The OVC people had set up a little polling station to give people a chance to see the system in action. When I went in, a poll worker handed me an empty manila folder. I walked into the polling booth and found a touch screen and a Hewlett Packard printer. The touch screen listed my choices in big print. I pressed on my choices. The printer chugged and printed out an 8.5 x 11 inch sheet of paper. That was my ballot. I read over my choices. Yup, I disapprove of Digital Rights Management and approve of Barack Obama. I slipped my ballot into the manila folder.
At the exit of the polling station, I fed my ballot into the bar-code reader. If people later suspect mistakes or mischief, they can refeed the ballots through the bar-code reader, or a different bar-code reader, or read them with their own eyeballs and count them with their own hands.
Barriers to Adoption
Why aren’t we voting with free, transparent software on cheap machines that give us paper ballots? First, ignorance. Many states and counties don’t know that the option exists. Second, money. Some states require that a voting system pass tests before it gets adopted. That’s a wise policy. Unfortunately, many states also require the proponents of the voting system to pay a fee for the test. Who pays the fee to test free software that runs on off-the-shelf hardware sold by many competing companies? Dechert, et al., won’t make any money from sales of the software or of the hardware. They have no future revenue stream, no venture capitalist support. They are proposing that states should charge only a low fee, or no fee at all, for tests of open source systems. Getting states to make this change, though, will take time. How many elections will take place in that time?
Los Angeles Leads the Way
Debra Bowen, California’s Secretary of State, has asked officials in Los Angeles County to consider adopting the Open Voting Consortium’s voting system. The fellow in charge is Dean Logan. I have exchanged email with his assistant, Paul Drugan.
He confirmed for me on September 19, 2008, that, “You are correct that we met with representatives of the [Open Voting] project, who gave Mr. Logan and his staff a system demonstration. Currently, we are simply reviewing possible systems and are not in a decision making mode at this time.”
He has declined to describe for me their decision-making criteria or schedule. If you live in Los Angeles County and plan to vote some time, feel free to follow up. The email address of L.A.’s Registrar-Recorder/County Clerk is: voterinfoATrrcc.lacounty.gov.
Running an election for Los Angeles County would be a challenging and revealing test for the OVC. LA has over four million voters. The biggest election that the OVC project has run so far is the one that I voted in. It had 816 voters. Should smaller counties also be running live tests? Would your county be interested?
Larry Magid‘s Digital Crossroads article, “Panel calls for open source software on voting machines,” on September 29, 2008
Scientific American, October 2008, “Voting Machines: Competing Candidates,” by Mark Fishetti, pp. 100-101.