[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Election status



On Wed, Apr 03, 2002 at 07:49:37PM -0600, Manoj Srivastava wrote:
> >>"Andrew" == Andrew Pimlott <ota-4@andrew.pimlott.net> writes:
> 
>  Andrew> How about:
> 
>  Andrew>  - When you vote, you additionally generate a random id and submit it
>  Andrew>    with the vote.
>  Andrew>  - In the vote list, the secretary publishes the id next to the vote.
> 
>  Andrew> You can still verify your vote, but you have no way to prove that you
>  Andrew> chose a particular id, so you can't convince anyone that a particular
>  Andrew> vote is yours.
> 
> 	This is in no way better than the scheme we have coded and
>  working right now. If someone can force you to give up your token,
>  they can force you to divulge your random id; and if the id is next
>  to the vote, you are sunk (The trick is, of course, that I'll get
>  your ID from you before the vote tally sheet is published, so you
>  can't fake it). 

I grant you that it is susceptible to someone who "gets to" you
before the vote.  This seems very hard to defend: the enemy can just
insist that you send him your signed vote, and let him submit it.
To beat this, you would have to be able to revoke the coerced vote
in a way that makes the enemy think the vote he sent was counted,
but makes you certain that yours was counted and his was not.  Too
hard for me.

I think my scheme has the (slight?) advantage that it is not
susceptible to someone who "gets to" you after the vote.  The
existing scheme allows you to prove (willingly) your vote to someone
you meet after the vote.  And, it allows an enemy who gets to you
after the vote to coerce you to reveal your vote--unless you can
convince him that you have destroyed and forgotten your confirmation
message.

(BTW, I'm not suggesting you change the scheme.  Just exploring
ideas.)

> 	In one way it is worse: What if 50 people choose Mickey Flood
>  as their randomg ID?

Obviously, the server rejects duplicate id's (and forces the voter
to resubmit).  Ok, there is a slight problem: if the secretary is
crooked, and two people submit the same id and the same vote, he can
forge a vote.  But if people are told to choose their id's randomly,
the chance can be made negligible.

Andrew


-- 
To UNSUBSCRIBE, email to debian-vote-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org



Reply to: