[EM] PRfavoringracialminorities

[Kristofer Munsterhjelm]

Raph Frank wrote:
> On Wed, Aug 27, 2008 at 7:59 PM, Kristofer Munsterhjelm
> <km-elmet at broadpark.no> wrote:
>> True. I just gave it as an option for the perfectionists who aren't
>> satisfied with Webster, or for the case where the election system is so
>> complex that adding the calculation wouldn't be noticed in theg rand scheme
>> of things (and where every little bit helps).
> 
> Someone would probably call you on it :).  You would have to justify
> what the particular method is best.

Well, yes, but I meant something like that if you're switching from Meek 
(numerical solution for nonlinear systems) to an election system with a 
divisor component, then a single exponential equation might seem simple 
in comparison. Of course you would have to explain it, but the society 
would already be used to the idea that voting systems may have to be 
complex to give good results, and thus would accept it (if they accepted 
the explanation, or trusted those who accepted the explanation) more 
readily than those who were not used to that.

>> For any ranked method, you could have a "new election" option, being
>> shorthand for "I'd rather have a new election with the status quo going on
>> in the meantime, than elect any of those listed below this option". Then, in
>> the social ordering, if this option ranks first, there's a new election (and
>> all the candidates of the previous election are barred from participating in
>> the next one). If it doesn't rank first, whoever wins wins.
> 
> Well, I guess it depends on the method.  However, if it was IRV, I
> think there is a reasonable case for making the NOTA option not be
> subject to elimination.
> 
>> For multiwinner elections, you could either redo the entire election, or if
>> one of the seats go to "new election", give those who were elected prior to
>> this their seats and then elect the remaining seats anew. The latter option
>> would be very complex, however, because you'd make sure that it retains
>> proportionality. The obvious way to do so is to retain weights, but then you
>> have to match those up with the voters, and doing that while keeping the
>> secret ballot secret would be very difficult indeed.
> 
> One option is to use Asset voting for that situation.
> 
> Your vote can designate a named candidate as responsible for voting
> for you if a NOTA option wins a seat.
> 
> This might be a separate column.  You rank the candidates in one
> column and then mark one of them as your NOTA delegate.
> 

That could work. However, the NOTA list would have to be longer than the 
ordinary list, I think, because in situations where NOTA ranks first, 
that means that the candidates below NOTA are not considered good 
enough, and thus by implication would not be considered good enough to 
give one's vote (as an asset) to. The same argument, if weakened, could 
be used where all but one is below NOTA - presumably only those who gave 
the above-NOTA candidate the asset vote would consider him good enough, 
and so all those who gave below-NOTA candidates the asset would be left 
out in the cold.

>> What I found of Forest discussing summability in 2007 was this:
>> http://lists.electorama.com/pipermail/election-methods-electorama.com/2007-April/020081.html
>> . It seems to say, in essence, that since voters first k preferences are
>> what count (for some small k), you can store them and average out the rest
>> to make "standard ballots" that won't lose much from reality.
>>
>> I think this may be iffy; there are no hard rules for how much
>> proportionality you lose, and if there are more than k seats, the averaging
>> could upset things.
> 
> Hmm, this looks like a more general case.  However, I think you would
> maintain proportionality as long as you meet the Droop criteria.
> 
> If all voters from a group vote for their candidate first choice, then
> they are guaranteed a seat (assuming the group has a Droop quota).
> 
> The only effect is on lower rankings.
> 
> One issue is that if a party expects to get more than 3 seats, then
> there could be issues.  However, even then it mightn't be a major
> problem.
> 
> Abuse would require that the abusers vote 1,2,3 for the party and then
> try to mess up their 4th rank.  I think that this is likely to
> increase the number of votes received by that faction rather than
> decrease it.

I don't even think it needs to be malicious. If voters have sufficiently 
many opinions they compare the candidates on, that might cause honest 
differences in the 4th rank and below. Averaging, by necessity, throws 
away some of this data.

>>> I really don't like PR-SNTV, but it would work.
>> It works *if* all parties run what is in essence vote management:
> 
> Yes.  However, vote management strips voters of their power to choose.
>  They can't bottom rank a disliked party candidate (without the party
> losing a seat).

True, and I don't think you could make a summable SNTV DSV method.

>> The election system here in Norway is somewhat like this. In the national
>> election, you vote for a party (closed list PR). For each district,
>> candidates are allocated to the parliament according to modified
>> Sainte-Laguë. In some cases, this leads to a (slight) disproportionality, so
>> after the district seats have been filled, top-up seats ("leveling seats")
>> are used to move the assembly back to proportionality.
> 
> I guess that is sorta like what I was saying.
> 
> I would have implemented the district seats as open list election.
> 
> Ofc, it uses a MMP like top-up seats rather than Fair Majority Voting
> reversal of low margin victories to return to balance.
> 
>> However, parties with
>> less support than the threshold get no top-up seats, so they have to rely on
>> district seats alone.
> 
> Not entirely fair.

Nor are thresholds in general. I don't like information being thrown out 
like that, so if the point is to make for a more stable government, I'd 
require a slight supermajority (50% + threshold + 1) and also have the 
German solution to executive transiency - the opposition has to agree on 
the composition of the executive before replacing it. If the government 
and the opposition still gets in a mutual exclusion fight (where one 
faction supports executive A' and another B', and they always propose A' 
or B' respectively), then one could add a rule where the executive has 
to be different from a previous one unless that would exclude all 
possible executives. I don't think that's needed, though.

>> Because all votes are by party, and the same counts are used for district
>> and top-up calculations, decoy lists are impossible.
> 
> They can also be made impossible in MMP (and Fair Majority Voting),
> if your party vote is inferred from your local vote.
> 
>> One extreme is that the candidates have no party loyalty, and the other
>> extreme is when all within a single party vote as one. Most real-world
>> situations would be somewhere in between, and it'd seem that to have the
>> assumption of power proportional to number of seats hold, it should be
>> closer to the "no loyalty" extreme than to the other extreme;
> 
> I think the voting system matters a lot for this.
> 
> PR-STV combines difficulty in getting re-elected with the inability of
> parties to prevent their members from running as independents.
> 
> This gives little power to parties to discipline their members by
> threatening to kick them out of the party, while giving voters lots of
> power to remove representatives who don't represent them well.
> 
>> however, if
>> one goes too far, there's no party to speak of to have political power.
>> Also, it's advantageous to parties in power if they coordinate closely. The
>> corollary is that the opposition can survive being fragmented more than the
>> parties in power can, because the various subgroups of the opposition can
>> then compromise in differing ways, trying to get the parties in power closer
>> to their points of view on at least some of the issues.
> 
> In Ireland, the governing coalition always votes as a bloc.  They have
> effectively "signed up" to the program for government document that
> they publish at the start of each new term.

Maybe that's a reaction to the weakening of party loyalty that PR-STV gives.

>> For that method to work, it would have to make assumptions about the shape
>> of the distributions, since those aren't available. All we have is the
>> distances of the distributions, and the method would also have to make a
>> reasonable guess as to what the distance operator is. Finally, "closest
>> possible" would be defined in a way that doesn't turn into minisum (majority
>> rule).
> 
> I think PR-STV actually does something like this.
> 
> It is in effect (to an approx at least)
> 
> Assign each voter to an elected representative (same number per
> elected representative)
> 
> Find the set of representatives and voters that minimises the distance
> between the voters and their representatives
> 
> Declare elected that set of representatives.

The VQ method could do better: if it could infer that some candidates 
are close to each other, opinions wise, it could deweight them. STV does 
this to some extent by deweighting the voters (which would impact the 
other candidate if most who prefer one also prefer the other).

What STV does is group together preferences so that all of the 
grouped-together preferences pass the Droop quota. It does this by 
eliminating candidates and preserving those who have passed the quota.

 From a distance point of view, the elimination process removes some 
candidates that are closer to the voters so that their "closest 
candidate" become someone that pass the Droop quota. In theory, that 
should work, but it's not perfect (as IRV, the single-winner case, shows 
most readily, though the effect weakens with multiple winners).

It's also possible to do this evening-out by a method similar to 
DSC/DAC: Start with sets and then eliminate until you have enough Droop 
quota single candidates. PSC-CLE does that, but its performance is worse 
than STV, even when the rounding down is corrected to rounding off.

> Right, I guess it ultimately depends on the scoring function.
> 
>> But let's say there's a Condorcet method that does return scores. If the
>> scores are to make any sense, a candidate that would have ranked higher in
>> the social ordering must have a higher score than a candidate that would
>> have ranked lower.
> 
> Condorcet is fundamentally linked to rankings.
> 
> What about something like
> 
> Sn = score for candidate who is in nth place in the condorcet ranking
> 
> Sn = Sn-1 * (1 - ( ( margin of victory for (n-1)th candidate of nth
> candidate )/(votes cast) ) )
> 
> S1 = 1000 (or some constant)
> 
> So if when comparing the first and 2nd place candidate the result was
> 55% to 45%, then
> 
> S1 = 1000
> S2 = 1000 * ( 1 - ( 10 / 100 ) ) = 900
> 
> This gives scores for all the candidates based on how well they
> compared to the candidate higher than them.  Also, it ranks the
> candidates in condorcet order.
> 
> It has issues handling circular ties. Maybe all in a tie should get equal score.
> 

Another method I've mentioned before is this: Start off at any random 
candidate. Now go to any other candidate with probability proportional 
to the adjusted winning votes ( > 0 if the other candidate beat this 
one, < 0 if this candidate beat that candidate). Do so a million times. 
Count the number of times each candidate has been visited - that's your 
score, and higher is better.

>> Moreover, we see that, in the example above, for an assembly of size 2, Left
>> and Right have a Droop quota each. This means Center must not be elected.
>> But if we use a Condorcet method that returns a set of scores, then the
>> Condorcet winner, which is Center, must get more seats than either Left or
>> Right. That's a contradiction, so adapting a Condorcet method in this way
>> must fail Droop proportionality.
> 
> Right.  I think it is fundamentally different to trying to find a PR method.
> 
> Condorcet is a centerist finding method.  PR is designed to spread the
> seats evenly between all the voters.
> 
>> The same goes for any method that returns a set of scores and also gives
>> Center the highest score. Plurality doesn't.
> 
> That is a fault with plurality.
> 
> A single seat method should aim for the centre as that means that
> everyone is best represented rather than representing one side of the
> electorate.

Does that mean Range is unsuitable for party list PR for the same reason 
-- broadly, that single-winner methods are too center-focused for the 
party list transformation to be any good?