> From: "Trei, Peter" <[EMAIL PROTECTED]>
> Date: Tue, 30 May 2000 09:33:33 -0400
> > There are a number of results in the crypto literature on receipt-free
> > voting, most recently (that I'm aware of) one presented by Kazue Sako
> > at last month's Eurocrypt 2000. Receipt-freeness means that voters
> > cannot demonstrate to third parties how they voted, and thus addresses
> > the bribery and coercion issue.
> >
> This is nonsense. If the person whose vote is being coerced has the
> coercer looking over their shoulder as they cast it, no receipt is needed
> to convince the coercer that their demand has been met.
My point (which I guess my example didn't adequately convey) was that
even looking over the voter's shoulder the coercer may not be able to
tell what the vote is, because it depends on a bit in the voter's head
that s/he can undetectably lie about, and has no way to convince the
coercer what it is.
> > A completely different tack is to allow voters to cast as many ballots
> > as they like and count only the last one. This effectively defends
> > against buying and forcing of votes because the voter can always vote
> > again. (I gather that corporate proxy voting works this way.)
> >
> This is more workable, as it increases the work factor for the coercer:
> he/she/it has to ensure that the last vote cast was cast the way
> demanded. I don't regard it as sufficient however - the greater
> complexity opens the way for error.
I think it might work reasonably well in practice, because ensuring
that the last vote cast was cast the way demanded seems infeasible on
any significant scale. I agree though that the additional complexity
could lead to error.
By the way, my parenthetical remark was intended to point out that
multiple voting exists in practice, and not to imply that corporate
proxy voting is immune to buying or forcing. Corporate elections lack
some properties that political elections (allegedly) possess:
1. They are generally not secret ballot--it may be possible to verify
how specific shares are voted, so changing a vote may be detectable.
2. They are not "one person, one vote", so it may be advantageous for
a would-be coercer to single out a large shareholder and not only
coerce his/her vote, but also prevent him/her from voting again.
Of course buying votes in corporate elections is trivial (by buying
shares), and this is generally considered a Good Thing rather than a
Bad Thing, at least by those who approve of the corporation as an
institution.
Ray
