Sometime via the 'Net?
By Scott Bradner
Network World, 02/09/04

I wrote about Internet-based voting after the chad-filled fiasco of the last presidential election. The column was not all that sanguine about the prospects, and events of the last few weeks have reinforced my skepticism.

About three years ago my column "Next time via the 'Net?" explored a few of the issues with Internet-based voting. Even though the column missed the most important conflict - the requirement for reliable authentication of the voters conflicts with the requirement for anonymity - it was still quite negative and concluded "there are no panaceas here - we can look forward to this kind of fun for years to come."?
?

Two different electronic voting-related stories have peaked in the past few weeks. The first, which never seems to end, concerns the poor design and implementation of the Diebold Elections Systems electronic voting machines. Yet another report on the systems was published last month. This time it was from a "red team" of hackers, commissioned by the Maryland Department of Legislative Services, about how easy it was to hack the Diebold equipment. The report also detailed some of the quite stupid (and that is the most positive way to put it) features of the Diebold systems, such as using the same key to the easily picked lock on all the Diebold voting machines in the state. At least now some of the state elections officials are beginning to temper their blind enthusiasm for the devices. The concern is not confined to Diebold systems, as reports have surfaced of glitches with the Sequoia Voting Systems equipment used in some California districts in the gubernatorial recall election.

The second story concerns the Department of Defense plan to use an Internet-based voting system in the 2004 presidential election for about 100,000 U.S. citizens overseas. A bipartisan group of party organizers asked Defense Secretary Donald Rumsfeld to halt use of the $22 million system known as the Secure Electronic Registration and Voting Experiment (SERVE) after a team of computer security experts concluded that it was fatally flawed.

Of course, Pentagon officials who have been overseeing the project dismissed the concerns and professed full confidence in SERVE. Watching these stories one might wonder if it's ever going to be possible to have electronic voting systems we can trust, with or without the Internet component. I expect it is possible, but the experiences to date do not indicate that it will be anytime soon.

We can get a lot closer, at least with electronic voting machines, by taking some simple steps, such as having the machines print out verification ballots (see verifiedvoting.org for more details and a petition you can sign if you are worried). But as long as officials are more concerned with justifying their decisions than in fair elections and as long as the companies producing the systems try to mimic security with secrecy instead of producing open verifiable implementations, we all will be at risk.

Disclaimer: There's no question that we will have believable electronic voting "soon" as long as we use "soon" in the context of an organization far older than the country. But that's my opinion, not Harvard's.

Bradner is a consultant with Harvard University's University Information Systems. Reach him at sob@sobco.com.