[EM] Smith, FPP fails Minimal Defense and Clone-Winner

[Juho]

On Mar 11, 2010, at 2:13 AM, Raph Frank wrote:

> On Wed, Mar 10, 2010 at 10:12 PM, Juho <juho4880 at yahoo.co.uk> wrote:
>> I didn't quite understand your question. The method could also be
>> non-party-list-based (like STV).
>
> It depended on what you meant by 1/N of the votes.  I was just
> wondering if you were doing national level rebalancing, like MMP or
> FMV.

Yes, I meant to include rebalancing to get national level  
proportionality. Or (following the earlier discussion) we could also  
have one nation wide constituency of size n to get fully proportional  
results.

>
>>> If both parties were roughtly 1/2 each, then most districts would  
>>> end
>>> up electing one from each party.
>>
>> Actually we have a continuum from single member constituencies to  
>> full
>> proportionality. Use of 2 seat constituencies provides a really rough
>> system, but still one step smoother than single seat constituencies.
>
> I think that 2 seats might be worse that single seaters, if the
> parties' support is similar throughout the country.
>
> The election would be decided by the districts that have a 2/3
> majority for one or other party.
>
> I had a look at the 2008 House election results, and there are a
> reasonable number of districts where one candidate got more than 2/3,
> so maybe it isn't as big an issue as I thought.  OTOH, maybe it was
> that in those districts, the minority knew that they had no chance, so
> didn't bother turning out.  Maybe the minority in that district would
> be able to manage 1/3 of the votes.
>
> I think an odd number of seats have the advantage that a majority of
> the voters in a district can assign a majority of the seats.

Ok, if we have two major parties then two seats could mean stagnation  
(one seat for each party in most elections forever).

One seat based system is at its best when there are two major parties.  
A two seat based system is maybe at its best when there are three  
parties. In that case a typical set-up in the competition would be  
"which one of the three parties will be left without a seat in this  
constituency". A typical government could be a two party coalition  
(since the largest party often does not have majority).

A constituency with two parties only (with 50%-50% support) would be a  
good place for a third party to enter. The next election could lead to  
45-45-10, and the following one to 35-32-33. The good point is that  
two seats allows third parties to grow easier and doesn't have as bad  
problems with spoilers.

Two seats could also allow four parties. The border lines get softer  
when the number of seats goes up.

(And of course regional parties could also be possible. I.e. the  
number of seats determines the number of parties primarily separately  
in each constituency but does not demand that the set of parties  
should be the same in all constituencies throughout the country.)

Juho