Security of Voting Systems

Ronald Rivest, Viterbi Professor of Computer Science, M.I.T.

This evening I attended a lecture at Carleton College, in Northfield: 'Security of Voting Systems,' by Ronald Rivest. About 100 people attended; it appeared to be a cross-section of techies and politicos, with emphasis on the former.

Rivest gave a PowerPoint presentation providing some introductory information on the history of voting and the intersection between technology and voting. He explained the many attributes of an ideal voting system, one which would be secure and easy for the unlettered and disabled to use without being cumbersome for others. You want a system that prevents people from voting more than once, you want machines that work when needed. You want integrity--by which he means that one's ballot can't be changed after being cast.

Rivest announces that we don't want a system which issues a receipt to the voter. Issuing a receipt to the voter [showing the votes cast, along with a unique number which the voter might later verify online to confirm her vote has been accurately counted]...seems like a great solution, no? But Rivest says this would benefit vote-buying and exacerbate voter intimidation [by spouses, significant others, employers, etc.]. Rivest spends a good deal of the lecture explaining an alternative system called Scantegrity which somehow allows online confirmation without benefiting ballot sales. (No, I don't quite get it.)

The professor notes that in Oregon, voting is done by mail. Rivest dismisses the Oregonian solution, saying that it enables vote-buying and intimidation. But the MIT professor doesn't provide evidence of any such problem, nor does he attempt to persuade us that Beaver Staters ought to dislike their voting system--which is in fact wildly popular there.

After the talk, I speak briefly with Rivest, making my point: It seems you're making this huge effort to frustrate vote-buying...but I'm not convinced that we really have a vote-buying problem here. Rivest amicably corrects me: The US has a long and rich history of vote-buying. Fair enough, I say, but it's just not a problem in USA-v2008. Rivest recommends I read Steal This Vote and I'll learn how serious this problem is, even now.

Someone mentions that anyone desiring a receipt--in order to get paid by a vote buyer--could simply use his cell phone camera to record his optical scan ballot, prior to submitting it. Rivest then says he wants cell phones banned in voting booths--clearly an unworkable suggestion. He hedges a bit.

I still think the Oregonians have a pretty good solution--that the problem of vote-buying and voter intimidation can be discouraged through a public awareness campaign and voter education. But if we want to stick with the system we now have, voters should be allowed to receive a web-verifiable paper receipt of their ballot. Vote-buying just isn't a big enough problem now, in this country, to justify doing otherwise.