[EM] STV - the transferrable part is OK (fair), the sequential round elimination is not

[Juho]

If one really wants a two-party system and doesn't want voters to  
change that fact then one could ban third parties and accept only two.  
That would solve the spoiler problem :-). From this point of view e.g.  
the US system is not really intended to be a two-party system but just  
a system (target state unspecified) that has some problems with third  
parties. On the other hand the option of third parties could be left  
in the rules intentionally. The voters are given a chance to change  
one of the two parties to some third party if they want that so much  
that despite of the associated spoiler problems they will eventually  
give the third party enough votes to beat one of the leading parties.  
Actually two-party systems need not be based on two parties only  
nation wide. In principle each district could have its own two parties  
that are independent of what the two parties are in other districts.  
There is however some tendency to end up with two or small number of  
parties nation wide.

Juho


On Nov 3, 2009, at 9:22 AM, Kristofer Munsterhjelm wrote:

> Juho wrote:
>> On Nov 2, 2009, at 4:50 PM, Jonathan Lundell wrote:
>>> Is this acceptable PR? I hope your answer is "of course not" (if  
>>> it isn't, we can have that discussion).
>> I note that a two-party system can be seen as one style of  
>> democracy that may be chosen intentionally. But if the target is to  
>> have PR then such single-seat FPTP systems are of course not good  
>> at all.
>
> If the people truly want a two-party rule, then using STV (or some  
> other party neutral PR method) can't hurt - they'll have that two- 
> party rule if they want, and can at any moment escape from it if  
> they change their minds. See Malta.
>
> In addition, if the method is any good "between the hard limits"  
> specified by the DPC or analogous proportionality criterion, then  
> there will be competition between the candidates inside of the  
> party. STV is IRV between the hard limits, so one may doubt how good  
> it is at this, but in reality, it does at least provide some measure  
> of that; my clustering methods are much more Condorcet and so  
> presumably would provide greater such competition. My  
> proportionality simulator shows it to be much better than STV, but  
> I've discovered that said simulator also has a significant small- 
> party bias, so I'm taking the results with some salt until I can get  
> proper correlation going.