[EM] Proportional representation method where additional voters do not change result

[Kristofer Munsterhjelm]

On 08/17/2012 07:30 PM, Clinton Mead wrote:
> Is there a proportional representation method such that, given N
> candidates, adding N! votes to the set of votes, each one of those N!
> votes being one of the possible sequences possible, does not change the
> result?
>
> (i.e. adding a whole lot of votes equally for all candidates doesn't
> change the result)?

I've been kept busy by real life, but I should have time to reply to this.

SNTV, if that counts as proportional.

Divisor methods will probably fail, e.g. something like:

10 seats, 4 parties, support is
45
32
8
1

with Webster gives 5, 4, 1, and 0 seats respectively. Now add 4!/4 = 3! 
= 6 to each:

51
38
14
7

gives 5,3,1,1.

Quota methods *might* work, but I'm unsure. The general idea is that if 
you add a constant to every ordering, no given coalition will decrease 
in its number of effective quotas' worth. Perhaps it comes down to 
tiebreakers - i.e. which DPC-permitted coalition the method picks -  or 
maybe I'm wrong.