Loomio
Fri 7 Apr 2017 6:16PM

Other Multi Winner Elections Options, PR and otherwise.

SW Sara Wolf Public Seen by 372

So far almost all talk of multi winner options has focused on reweighed versions like STV and RRV.

These don't meet equality criteria any way you slice it and are complex enough that the result accuracy is nebulous and hard to judge. They both seem quite accurate and like they may well elect a fair proportional representation of the population but more than that is a lengthy conversation on some other threads.

Let's talk about multi winner by district, and other non-reweighted systems here!

SW

Sara Wolf Fri 7 Apr 2017 6:23PM

I've was home sick last week and a girl can only watch so much Netflix. I've been thinking about how to measure accuracy and who should win in Score PR systems and it sparked something. Maybe the system should award seats based on a quota? So this led to a new idea for another Proportional Score Runoff Voting system and am really curious what you think. It's a LOT easier to do the math than RRV or SRV-PR and seems to have similar results. I give 3 different systems with increasing complexity for comparison. I think #1 or 2 might be good for multi winner primaries and I think the 3rd might be the sweet spot for general elections. Check it out:

Examples given are for 4 winner races with 8 Candidates.
A and B are Green. C and D are Democrat. E and F are Republican and G and H are Libertarians. There are 400 voters. There are 170 Democrats. 140 Republicans. 60 Greens and 30 Libertarians. All voter give their party 5s, their second favorite party 4s and their 3rd favorite 1s. Quotas given are for rounding. Examples are very simplified and rounded to the beginning of the Alphabet.

SAMPLE ELECTION
60 Greens- A:5 B:5 C:4 D:4 E:0 F:0 G:1 H:1
170 Dems- A:4 B:4 C:5 D:5 E:0 F:0 G:1 H:1
140 Reps- A:0 B:0 C:1 D:1 E:5 F:5 G:4 H:4

30 Liber- A:1 B:1 C:0 D:0 E:4 F:4 G:5 H:5

1. Simple Score Multi-Winner:
Score all Candidates 0-5. Add up all the scores and the top 4 win.
Winners: C-Dem, D-Dem, E-Rep, F-Rep.

2. Multi-Winner Score Runoff (not SRV-PR):
Score all Candidates 0-5. The top 2 go to a runoff where the one preferred by the most voters wins. Then the next top 2 go to a runoff and so on until enough winners are chosen.
Winners: Dem, Dem, Rep, Rep.

3. "Equal Representation Voting- Partisan plan"
Score all Candidates 0-5. Add up the total number of scores given to all candidates and divide by the number of seats. This is the target number or quota for 1 seat. Any candidates that got that number get a seat.
Round 1 Round 2

Green- A:5 B:5 C:4 D:4 E:0 F:0 G:1 H:1 = 1200 = 1200*
Dem- A:4 B:4 C:5 D:5 E:0 F:0 G:1 H:1 = 3400* = 1400*
Rep- A:0 B:0 C:1 D:1 E:5 F:5 G:4 H:4 = 2800* = 800
Liber- A:1 B:1 C:0 D:0 E:4 F:4 G:5 H:5 = 600 = 600
Total- = 8000 = 4000
Quota- = 2000 = 1000

Round 1: Voters gave 8000 points total so the quota for each seat is 2000. Off the bat Democrats and Republicans each get a seat.

Round 2: Delete points already spent to get a seat. (Delete last rounds quota from each Count points remaining points for all Candidates. Any with over 1000 get a seat. Democrats and Greens get a seat. Continue rounds till all seats are filled.
Winners: Dem, Rep, Dem, Green.

4. "Equal Representation Voting- Non Partisan Plan"
I tried but couldn't figure it out. :(

KEY IDEA: Who gets to run? Suggestion: Each party could be allowed to run twice as many candidates as their voter registration shows. So if 28% of voters are registered Democrat then the Democrats can run enough candidates to fill 56% of the seats. If there are 8 candidates total then they could run 4 of them. This allows there to be twice as many candidates in the election as there are available seats.

AZ

Adam Zielinski Fri 7 Apr 2017 10:38PM

How about Mixed Member Proportional Representation? It is basically a hybrid voting method. And you could substitute SRV for the district winner half of it instead of plurality.

https://www.youtube.com/watch?v=QT0I-sdoSXU

https://en.wikipedia.org/wiki/Mixed-member_proportional_representation

https://www.youtube.com/watch?v=8Uk44aykGg4

SW

Sara Wolf Sat 8 Apr 2017 6:52AM

I could see MMP being okay for the House of Reps but I have to say that after reading and watching all of that I still only get the gist. Maybe because there are a few variations? Is MMP ever used for just local elections? It seems like usually parliament. It'd be good to see a simple write-up that clearly explains the mechanics and pros and cons. How do we measure it's accuracy?

Is the outcome better than my #3 system above which can be explained in a few sentences? Equal Representation Voting? I just made it up and I'm super curious what other people would think. Feedback!

Like I said, I'm not a huge fan of Party based PR anyways because currently I feel like the parties are really unrepresentative of the people. Part of that is that our current system has made it that way on purpose but old habit's die hard and I feel like they are sort of divisive in nature...

I'm sure a great option is possible, we just have to find it.

CS

Clay Shentrup Sat 8 Apr 2017 7:44PM

How about Mixed Member Proportional Representation?

Well, it's a rather archaic system with two major flaws:

1) It uses Plurality Voting to pick the local MP's.
2) It uses party lists, which discriminate against independents.

Warren Smith proposed this MMP-like system which is a hybrid of Score Voting and Asset Voting, and seems clearly superior.

http://scorevoting.net/ScoreAssetHybrid.html

AZ

Adam Zielinski Wed 12 Apr 2017 1:07AM

I really, really, do not like Score Asset voting at all, because of this part:

"6. Candidates now negotiate. Any candidate may give any other some or all of their juice."

No f-ing way. So the voters vote, and then nothing is decided until candidates who are still unelected to anything negotiate who among them is actually going to get a seat? What could possible go wrong, except everything. This is basically a worst case scenario and the worst proposed reform I have ever heard of.

CS

Clay Shentrup Fri 21 Apr 2017 4:29PM

Adam,

Before shooting down a complex policy proposal because one part of it sounds alarming, consider first that its pros and cons have been extensively considered.

http://scorevoting.net/Asset.html

Negotiation is part and parcel of the political process. It is what already happens every time Congress votes on anything. Your representative is negotiating on your behalf.

Suppose you like Bernie Sanders, and he and Clinton have between them enough votes for a seat. So maybe he doesn't have the votes for a seat, but he says to Clinton, "I will give you enough of my surplus votes to get this seat instead of that other more conservative Democrat over there. But, I demand that you work on these four critical issues that my constituents demand."

In this sense it is "perfectly proportional" in that a candidate conveys her power proportionally to her support, even if she doesn't have enough votes for a whole seat. It's approximately like being able to win "half a seat" or "a third of a seat".

Now you can go and advocate for PR systems that don't have this negotiation process. But they all are radically more complicated, and have worse "rounding error" (are less accurate in their proportionality), and many of them (like MMP) have party lists, which inherently disadvantage independents and concentrate power in the hands of party bosses who make their internal decisions...

negotiating in smoke-filled rooms.

AZ

Adam Zielinski Sat 8 Apr 2017 12:03PM

I'm not really understanding what the problem is with STV or RRV in regards to equality or accuracy. The thing I like about those systems is how when a candidate gets an overvote, then over votes go to those voters' second choice.

Those systems include dividing up the vote by the number of seats to get the target or threshold required to get a seat. Please don't call this a quota. Quota is a loaded term with bad connotations, at least for me.

So your #3 sounds kind of ok but because you want to shy away from the transferable vote or reweighting, then you are not using any overvotes to help elect second choice candidates. So your system ends up electing fewer third party members and is more major party and status quo friendly.

The reweighting and transferring of votes is what makes RRW and STV more accurate and representative in my opinion.

I'm sorry but I didn't participate much in other threads about equality and accuracy. Maybe I should go back and read them. I guess I'm not understanding what the problem or concern is or why you think existing PR methods are lacking in some way.

AZ

Adam Zielinski Sat 8 Apr 2017 1:12PM

Here's the videos on STV and RRV

STV:
https://youtu.be/l8XOZJkozfI

RRV:
https://youtu.be/kaZB84uipFk

SW

Sara Wolf Tue 11 Apr 2017 10:16AM

RE: "I'm not really understanding what the problem is with STV or RRV in regards to equality or accuracy."

I'm not saying that RRV, STV or SRV-PR aren't accurate. They seem accurate to me at a glance. My point is that I don't get how that can be measured and the systems can be compared. I think we need to work on how to measure the systems if we want to choose and push one of these systems and then justify that choice to voters.

For PR when we say accuracy it's not clear what exactly we mean. In a simplification- are we trying to say that each candidate needs a full 1/4 of the votes to deserve 1 out of 4 seats? Or is it better to round to the nearest seat and round up if some voters don't have enough to get a seat but if we still want them represented somehow? How much rounding is fair and good? Does rounding up improve VSE? How does this translate to systems were voters can show nuanced support for multiple candidates?

I'm not sure what the correct answers to these questions are. I'd like to know what the exact goals of each system are in a measurable form, and then look at how well each one does at accomplishing it's stated "best set of winners." Each could be measured to it's own standard and we could compare standards and look at what criteria is most important to us.

AZ

Adam Zielinski Wed 12 Apr 2017 1:21AM

"My point is that I don't get how that can be measured and the systems can be compared. I think we need to work on how to measure the systems if we want to choose and push one of these systems and then justify that choice to voters."

I don't disagree with this as a goal but I guess I'm not as concerned about it. It's more in the nice to have category rather than in the critical category. The main difference between the two is scoring vs ranking. So the question is do you want to score or rank candidates. We are saying we want to score candidates because it allows voters to express more nuanced opinions and leads to better results and a minimized spoiler effect in the single winner SRV election. RRV is just the PR version of SRV voting and similar to STV, but with scoring rather than ranking.

I suppose we could run some simulated votes and see how they might come out different under various scenarios.

SW

Sara Wolf Mon 24 Apr 2017 7:31PM

For me Scoring vs Ranking isn't the key concern or choice. I'd like it to be consistent so voters can get used to their ballot. I prefer scoring not for its own sake but because SRV offers a voting system that lets us vote our conscience, that has accurate results, and that gives everyone an equal vote!

If a ranking system can offer honest voting, accurate results and equal voting power then I'd support it, especially if its fairly expressive and simple.

The main problem I see in the ranking systems like IRV and STV that are being proposed is that the results only look at the ballots of some voters at any given time. The final results are using a threshold that only looks at the finalists so it's not necessarily still accurate if we were to look at all the ballots and voters all at once. This can give a confusing and false sense of accuracy that may or may not hold up in STV. I haven't seen a response yet that adequately addresses this concern.

I think for ranking to hold up to this higher standard for accuracy and accountability it should probably be based on Ranked Pairs or Borda Count, not IRV.