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

[Kevin Venzke]

Hi Juho,

--- En date de : Dim 11.1.09, Juho Laatu <juho4880 at yahoo.co.uk> a écrit :
> If there is a set of voters that form a
> majority and they all prefer all candidates
> of set A to all candidates of set B then
> candidates of set B should not win.
> 
> This helps A (as requested) by at least
> eliminating some of the candidates from
> competing with A.
> 
> This criterion may also eliminate all
> candidates. In such situations the rule of
> course will not apply.
> 
> I haven't really thought what implications
> there are. Any comments?

I would say you're close to inventing again either MDD or beatpath
GMC / CDTT. All you've essentially said is that if A has a majority
over B, B can't win. Because, each candidate could make up their own set.
Having multiple candidates in a set doesn't make any difference.

Under MDD the candidates of set B cannot be elected unless all candidates
can be placed in a set B. This is inherently not cloneproof.

Under beatpath GMC / CDTT the candidates of set B cannot be elected
unless they have a majority-strength beatpath to all the candidates of 
set A. However, lacking this, the candidates of set B can also be
disqualified when the candidates of set A merely have a majority-strength
beatpath to the candidates of set B.

A few years ago I considered a set where it wouldn't be enough for set A
to merely have a beatpath to set B, in order to disqualify those 
candidates. But from what I remember, there were monotonicity problems.
I guess there are probably clone problems also.

Kevin Venzke