[EM] IRV vs Plurality

[Juho]

On Jan 27, 2010, at 7:02 AM, Abd ul-Rahman Lomax wrote:

> At 08:10 PM 1/26/2010, Juho wrote:
>
>> The scenario that you described requires some goodwill among the
>> voters.
>
> That's correct. We seem to imagine that better voting systems will  
> produce better results even if people continue to lack goodwill and  
> cooperative spirit. It's a fantasy. There are structural changes  
> that will encourage the seeking of consensus, but voting methods are  
> only a tiny part of that.

I think also the strong tradition of trying to develop methods that  
have as good performance as possible even when voters are competitive  
and strategic is a good and healthy tradition. An ideal method might  
achieve even better results when voters want to cooperate without  
sacrificing the performance with strategic votes. Or if the  
environment is non-competitive then one could use also methods that  
rely on the required sincerity of all voters.


>> In typical
>> political environments good poll information including approvals and
>> ratings is thus a positive thing, but it may still be necessary to
>> assume that strong competition is not uncommon in the actual election
>> and prepare for that.
>
> Yes. I do suggest Bucklin. Most voters will bullet vote, it's very  
> likely, but, then, use it as a primary in a runoff system, which  
> provides a very specific meaning to the lowest approved rank: I  
> prefer the election of this candidate to holding a runoff. It's an  
> absolute, sincere vote that is strategically maximal! Because that  
> is exactly the effect it has, monotonically.

Bucklin + runoff might indeed work better than e.g. plain Approval  
(and plain Plurality of course) (assuming that the increased  
complexity is not seen as a big problem).

Juho