Schulze, SEC, & many candidates

Blake Cretney bcretney at
Fri Oct 16 12:10:16 PDT 1998

On Mon, 12 Oct 1998 19:46:28   Mike Ositoff wrote:
>I left something out: As I & Markus have said, Schulze's method,
>with votes-against, doesn't create a situation, when nothing
>is known about other voters, where a voter indifferent between
>2 candidates can rank one over the other without risk to his
>favorite, in general.
>Sure, as Blake pointed out, that situation can exist for Schulze(VA)
>when there are just 3 candidates. But when there are sufficiently

Actually, I pointed out the opposite.  My mail of  "Re: Does 
VA Schulze violate SEC?" (Tue, 06 Oct 1998 15:27:44) gives an 
example where random-filling back-fires for 3 candidates.

>many candidates, and you insincerely order 2 candidates between
>whom you're indifferent, you might be strengthening the beat-path
>that will make your favorite lose to someone else by beat-paths.

I have NEVER said that random-filling cannot back-fire.  That isn't
my point at all.  My SEC criterion has nothing to do with a no 
back-fire situation.  

Consider a voter who prefers A > B > C.  We would hope that such a
voter would vote sincerely.  However, his vote of B > C, could
back-fire, and prevent A from being elected.  This is true of
any Condorcet method.

So, why would a rational voter vote sincerely?  Would the voter be
forced to weigh their desire to get A elected versus their desire to 
defeat C?  Actually, no.  Because even though the B > C ranking CAN
defeat A, it could also elect A.  Unless the voter has knowledge
of how everyone else is voting, the rational vote is A > B > C.

Under VA, the rational voter uses the random-fill strategy.  Of
course it can back-fire, but so can any rankings beside the first
and the last.  Should I be paralyzed by fear of a possible back-fire
into not voting a full ranking?

An interesting problem with VA, which I have previously not methioned,
is that it requires the kind of weighing of interests I mentioned above.
For example, if I sincerely rank A > B > C, I could vote this way.
However, I can increase the likelihood C will be defeated by voting
A = B > C.  In other words, to have the most effective vote against
C, I must not express my preference between A and B.  Of course, this 
sort of weighing of interests is one of the most obvious characterstics
of Approval, but it seems out of place in a ranked method.

-----== Sent via Deja News, The Discussion Network ==-----  Easy access to 50,000+ discussion forums

More information about the Election-Methods mailing list