🚀 go-pugleaf

>Thanks for the pointer, Ken.  His scheme looks good.  Here's a sketch:
>
>  The election takes place between voters and k candidates, at least
>  one of whom is assumed honest.  Each voter communicates with all of

One voter or one candidate?  I don't know of any election for an actual
government elected office (student council doesn't count) in which I 
would trust any of the candidates with tallying the election results.

>  the candidates in order to register and vote.  The voter obtains
>  an "eligibility token", prior to election day, signed by all the

In other words, the candidates get to decide who votes (in this case it
appears that one blackball by one candidate denies a voter the right to
vote), by refusing to communicate with them.  The White Supremacy Party
would find it easy to not communicate with certain neighborhoods.

>  candidates.  On election day, the voter submits a vote to each
>  candidate using a zero-knowledge protocol so that they can verify
>  the validity of the vote but not its contents.  To tally results,
>  the candidates must sign and publish their sub-tallies, and (I think)
>  only then can the results be computed.  Correctness and privacy is
>  assured as long as one of the candidates is honest.

This also doesn't seem to address problems with voters double-registering,
voters pretending to be other people who don't usually vote in addition to
themselves, voters using the token of other voters who died between 
registration and the election, and external denial-of-service attacks on 
the voting system.

>One concern I have with Iverson's method is that the protocol stalls
>whenever any of the candidates is unresponsive.  Of course, "candidate"
>may be a misnomer -- perhaps the parties involved should be mutually
>independent watchdog agencies.

Better.

					Gordon L. Burditt
					sneaky.lonestar.org!gordon