[EM] PR favoring racial minorities

[Kristofer Munsterhjelm]

Raph Frank wrote:
> On 8/18/08, Kristofer Munsterhjelm <km-elmet at broadpark.no> wrote:
>>  That sounds like MMP. I think MMP can work if done right (with STV instead
>> of FPTP as base, and reweighting to avoid lista civetta). Using party list
>> here is probably better than the party-neutral version where you'd rank
>> representatives for local, regional, and national levels, and then it keeps
>> the reweighting at each stage; simply because there would be an immense
>> number of candidates at the national level, and ranking them all would be
>> Herculean.
> 
> It is different to MMP as it only counts
> votes which don't elect someone locally.
> 
> In the MMP situation, it would be like having
> a rule that only votes which don't vote for the
> local winner count for working out proportionality
> at the national level.
> 
> (Though this is inexact as in single seaters, the
> 'quota' would be nearly 100% of the local
> votes cast, so nobody would be elected locally).

For STV, that's probably the right thing to do. For FPTP, it's less 
clear, since the lesser-of-two evils type strategy shows up there as 
well, and so the constituency votes are much more likely to be strategic.

The decoy list strategy appears because it's possible to vote for a 
different national party and regional party (constituency candidate), 
which leads to an overhang that can be exploited to turn top-up into 
parallel MMP.

A simple countermeasure is to weight the vote pairs in a way that if you 
got what you wanted in the constituency part, your say is diminished in 
the list part. One way to do that is to say that if you helped elect a 
candidate from party A, constituency-wise, and you chose party B for the 
list part, and B has an overhang, then count a fraction of your vote 
towards A instead. Schulze's STV-MMP does that, but since STV gives 
detailed information about the extent to which one's vote mattered 
(weight after all transfers, or multiplier after the diminuation by all 
retain factors in Meek), one could also just reweight the list vote 
power proportional to how much one's constituency vote *didn't* matter.

For a single seat election, STV would reduce to IRV. For 1000 ballots, 
the Droop quota would be (1000/(1+1)) + 1 = 1000 * 0.5 + 1 = 501, which 
seems right, considering a majority is 50% + 1.

> One possible issue is that voters would
> just vote for their party's local candidates
> and then for their party's national list, i.e.
> it would greatyly decrease the importance
> of cross party transfers.
> 
> This would eliminate the need for partys
> to be civil to each other as it reduces the
> need to avoid alienating the supporters of
> other parties.
> 
> There would be little need for voting pacts
> and the like.

If the voters vote a straight party ticket, wouldn't the method reduce 
to party list PR? Well, a strange sort of party list PR that allocates 
some of the seats according to a quota rule and others according to a 
divisor method, but party list PR nonetheless.