Punchscan

[mmcirvin]

Via Lindsay Beyerstein: Punchscan is interesting. They've come up with a way to give voters a receipt they can check against a website to ensure that their votes were counted properly, without being able to use that receipt to prove how they voted to anyone else. The randomly-generated ballot separates into two pieces, and both are necessary to tell how you voted, but the one you keep can be used for verification.

(The requirement of nonverifiability to others is why current voting systems don't give you a receipt: if you could prove how you voted, that could be used to pay or coerce people to vote a certain way.)

Despite their assurances, I do wonder about usability issues; they've given some thought to disabled access, but these ballots do seem slightly more complex than the simple optical-scan system we use in most elections here, though not anywhere near as bad as Palm Beach-style punch-card ballots.

Comments

[avitzur.livejournal.com]

Rivest proposed a method to achieve the same goals of anonymity, privacy, and auditability without complex crypto. Schneirer describes it here (http://www.schneier.com/blog/archives/2006/10/new_voting_prot.html). Rivest's paper is here (http://theory.csail.mit.edu/~rivest/Rivest-TheThreeBallotVotingSystem.pdf) (pdf).

[mmcirvin.livejournal.com]

The Rivest system looks much harder for the voter to use than the Punchscan one; I'd think that it would create a problem with erroneous ballots far greater than the usual troubles with untabulated votes.