[EM] "Beatpath GMC" compliance a mistaken standard?

[Kevin Venzke]

Hi Paul,

Regarding mutual majority:
The problem is that the B>A voters cannot be counted as solidly committed
to {A}. They can only be counted to {B} and {A,B}. The additional A
bullet voters can only be counted to {A}. C was excluded in scenario 1
because {A,B} possessed a majority. The new A voters increase the
requirement for a majority but don't increase the strength of {A,B}.
And {A} alone is not strong enough.

It's certainly possible to criticize that the B>A voters should be
allowed to help {A} somehow.


Regarding minimal defense (and I apologize for confusing the issue if
I did so, by bringing up a second criterion):

--- En date de : Sam 10.1.09, Paul Kislanko <kislanko at airmail.net> a écrit :
>> A criterion more similar to what you have in mind, and
>> which I consider
>> more essential and effective than mutual majority, is this
>> rendition of
>> minimal defense:
>> 
>> "If a majority of the voters vote for X and don't
>> vote for Y, then Y must 
>> not win."
>> 
>> Although, the effect of that criterion is that {A,B} are
>> the possible
>> winners in both scenarios.
>
> I am still not understanding. In the second scenario only A
> has a majority
> of voters' support. So how does B get included in the
> second scenario? 

A's majority support serves to disqualify C, but can't disqualify B,
because too much of A's support is also B's support. There's no majority
that votes for a common candidate and doesn't vote for B.


A criterion which said: "If any candidate receives votes from a majority
of the voters, then the winner must be one of these candidates,"
would be controversial because in a scenario like this:

49 A>B
3 B
48 C

This hypothetical criterion would require that B be elected, when many
of us would rather say that A should win this election, because A can
defeat the other candidates pairwise. Also, if B wins, then the A voters
will feel that it wasn't safe to vote for B.

Kevin Venzke