[EM] Strategies for RRV/RSV and BR for multi-member constituencies

[Kristofer Munsterhjelm]

Gervase Lam wrote:
>> Date: Thu, 05 Feb 2009 16:05:31 +0100
>> From: Kristofer Munsterhjelm
>> Subject: Re: [EM] Strategies for RRV/RSV and BR for	multi-member
>> 	constituencies
> 
>> I imagine one could make PAV variants for any of the "denominator" 
>> methods (D'Hondt, Sainte-Lagu?, Imperiali, proportions, etc). A 
>> Huntington-Hill variant would go like this: 1/0+ + 1/sqrt(2) + 1/sqrt(6) 
>> + 1/sqrt(12) + ....
> 
> See the following Election Methods posts to find something that you may
> need to be wary of:
> 
> <http://lists.electorama.com/pipermail/election-methods-electorama.com/2002-February/007367.html>
> 
> <http://lists.electorama.com/pipermail/election-methods-electorama.com/2002-February/007382.html>
> 
> <http://lists.electorama.com/pipermail/election-methods-electorama.com/2002-February/007391.html>
> 
> <http://lists.electorama.com/pipermail/election-methods-electorama.com/2002-February/007429.html>
> 
> You may want to follow the complete thread using the Election Methods
> archive.

Of course sequential PAV isn't PAV. I didn't know about the increased 
manipulability of Webster-based methods, though, but I suppose it boils 
down to whether one wants a "fair" method (like Webster/Sainte-Laguë), 
or one that can resist strategy better.

To my knowledge, Warren showed that D'Hondt is the only divisor method 
that doesn't have such splitting incentive. Adams works the opposite 
way: parties never have an incentive to combine; and it is unique (among 
divisor methods) in having this property.

Perhaps QPQ based on Webster would also have this susceptibility 
problem, but my program doesn't test that, so it may give it a better 
score than would "really" be warranted