Loomio
Tue 10 Apr 2018 5:40AM

Multi-winner adaption of STAR voting

PF parker friedland Public Seen by 28

SRV-PR (the method that is most commonly brought up when disusing the proportional multi-winner version of STAR voting) is not even proportional. If you want further proof of this, check out the example Warren created where it can fail to elect a Democrat when 49% of voters give a max score to the Democrat and min stars to the Republicans: https://groups.google.com/forum/#!searchin/electionscience/proportional$20srv$3F%7Csort:date/electionscience/aOMyxkoey4Q/KepDTrY7AwAJ (the example is mentioned in Warren's first comment in the linked discussion). It can meet relaxed definitions of proportionality in which each party runs clone candidates equal to the number of seats up for grabs, however the only reason these relaxed definitions of proportionality work is because they rely on candidate cloning which is something that turns SRV-PR into RRV just like it turns STAR into score.

Obviously STAR voting advocates shouldn't advocate for a broken "proportional" voting method that isn't even proportional. But they still need a proportional multi-winner version of STAR voting method that they can mention to as a long term goal when ever the topic of proportionality comes up.

Here is a discussion about what that method should be: https://groups.google.com/forum/#!topic/electionscience/HkgoqQiGN4k

PF

Poll Created Tue 10 Apr 2018 5:52AM

Harmonic Voting + Runoff Closed Tue 3 Nov 2020 8:04AM

Harmonic voting ( http://scorevoting.net/QualityMulti.html ), except that between the outcomes with the highest quality according to harmonic voting's quality function, hold an automatic runoff between those two outcomes where everybody's vote automatically goes to the outcome out of the two that had candidates that the voter gave a higher scores on average too.

Results

Results Option % of points Voters
Agree 33.3% 1 PF
Abstain 33.3% 1 SW
Disagree 33.3% 1 AZ
Block 0.0% 0  
Undecided 0% 0  

3 of 3 people have voted (100%)

PF

parker friedland
Agree
Tue 10 Apr 2018 6:03AM

If any of the these methods could be considered the "natural" multi-winner adaption of STAR voting, it would be this one.

SW

Sara Wolf
Abstain
Tue 1 May 2018 8:58AM

I'm lost.

AZ

Adam Zielinski
Disagree
Tue 4 Jun 2019 12:56AM

Too complicated

PF

Poll Created Tue 10 Apr 2018 5:55AM

Combination of RRV type proportionality + STV type proportionality Closed Tue 3 Nov 2020 8:04AM

Duplicate each vote so you have two versions of everybody's ballot (you don't have to actually duplicate each person's ballot, but it is easier to explain it this way). We shall refer to these two parts as two different types of votes: score votes and a preference votes. One of the two ballots will be used for scores and the other will be used for the runoffs. Each round, use the score votes to pick the top 2 candidates and the preference votes to chose which of the two candidates gets elected that round. Each round, the score votes are re-weighted using the RRV re-weighting equation. However, the preference votes will not be re-weighted. Instead, a droop quota % of the preference votes that preferred the runoff winner will be exhausted (and the % of those votes chosen for exaustion will be the % that gave the highest difference in scores to each of those two candidates that made it to the runoff).

Results

Results Option % of points Voters
Agree 25.0% 1 PF
Abstain 50.0% 2 SW WW
Disagree 25.0% 1 AZ
Block 0.0% 0  
Undecided 0% 0  

4 of 4 people have voted (100%)

PF

parker friedland
Agree
Tue 10 Apr 2018 6:05AM

Not as good as harmonic voting + runoff, but it might be the best sequential (round based) multi-winner adaptation of STAR in which the STAR method is still used in each round.

SW

Sara Wolf
Abstain
Tue 1 May 2018 9:00AM

This sounds like an interesting line of questioning. It sounds like this would result in an algorithm that is twice as complex as either though. And how different are the results of RRV vs STV anyways?

AZ

Adam Zielinski
Disagree
Tue 4 Jun 2019 12:54AM

Too complicated

PF

Poll Created Tue 10 Apr 2018 5:58AM

RRV but elect the last winner via STAR voting Closed Wed 4 Nov 2020 6:03AM

RRV but elect the last winner via STAR voting

Results

Results Option % of points Voters
Agree 66.7% 2 AZ PF
Abstain 33.3% 1 SW
Disagree 0.0% 0  
Block 0.0% 0  
Undecided 0% 0  

3 of 3 people have voted (100%)

PF

parker friedland
Agree
Tue 10 Apr 2018 6:08AM

It's technically a multi-winner adaption of STAR voting. It may resemble RRV a lot more then STAR, but it doesn't have any of the flaws SRV-"PR" has.

PF

parker friedland
Agree
Thu 12 Apr 2018 8:28AM

It's defiantly the simplest multi-winner adaption of STAR voting. And the last round of RRV generally matters the most.

SW

Sara Wolf
Abstain
Tue 1 May 2018 8:55AM

We've discussed RRV with an added reweighted runoff for the last seat only. Is that what you mean here?

AZ

Adam Zielinski
Agree
Tue 4 Jun 2019 12:53AM

Easiest to explain and comprehend and tabulate. True PR with a STAR twist at the end.

PF

Poll Created Tue 10 Apr 2018 6:01AM

Current SRV-"PR" method Closed Fri 20 Nov 2020 8:04AM

Elect the STAR winner each time while using RRV's re-weighting equation each round

Results

Results Option % of points Voters
Agree 0.0% 0  
Abstain 33.3% 1 SW
Disagree 66.7% 2 AZ PF
Block 0.0% 0  
Undecided 0% 0  

3 of 3 people have voted (100%)

PF

parker friedland
Disagree
Tue 10 Apr 2018 6:08AM

Nope.

SW

Sara Wolf
Abstain
Tue 1 May 2018 8:53AM

To clarify. STAR-PR is the same as SRV-PR. It just got renamed. If you want to compare variations it would help me if you describe details. ie. runoff every round or certain rounds. Reweighting every round or certain rounds...

AZ

Adam Zielinski
Disagree
Tue 4 Jun 2019 12:52AM

Not really PR; overly complicated

PF

Poll Created Wed 11 Apr 2018 8:55AM

MSPAV + STAR voting Closed Tue 3 Nov 2020 8:04AM

Results

Results Option % of points Voters
Agree 0.0% 0  
Abstain 75.0% 3 SW WW PF
Disagree 25.0% 1 AZ
Block 0.0% 0  
Undecided 0% 0  

4 of 4 people have voted (100%)

PF

parker friedland
Abstain
Wed 11 Apr 2018 8:55AM

?

SW

Sara Wolf
Abstain
Tue 1 May 2018 8:51AM

What are you proposing?

AZ

Adam Zielinski
Disagree
Tue 4 Jun 2019 12:48AM

Too complicated

PF

parker friedland Wed 11 Apr 2018 9:00AM

Are we all in agreement that the method with the most agrees minus disagrees (net approval voting) should become STAR-PR?

PF

parker friedland Wed 11 Apr 2018 9:18AM

Anybody who disagrees with the idea that the method with the most agrees minus disagrees should become the new STAR-PR can effectively disagree with that idea by giving the broken SRV-"PR" method an approve and every other option a disapprove. Thus, there is no need for another poll on whether we should have an election on the future of STAR voting. While some might argue that equal vote's STAR voting poll should be used to decide STAR-PR's own fate, I like this multiple proposal idea better because people can constantly submitted new multi-winner versions of STAR voting without having to create a new STAR poll every time a new idea is proposed. Also, the ability to explain your votes and possibly persuade others to change their own votes is a nice feature as well.

PF

Poll Created Thu 12 Apr 2018 8:26AM

RRV with smothe transition from score voting to STAR voting Closed Tue 3 Nov 2020 8:04AM

Normal SRV-"PR", except that each round, the number of ballots the candidate with the second highest score needs to be preferred on to beat the candidate with the highest score gets greater and greater. In the last round, the candidate with the 2nd highest score would only need to be preferred on 1/2 of ballots to win (The last time I forgot to mention that ballots that are re-weighted would still count less to whether one candidate is preferred to another), on the the 2nd to last round, they would need to be preferred on 2/3rds of ballots, on the 3rd to last round, 3/4th of ballots, on the forth to last round, 4/5th's of ballots, on the 5th to last round, 5/6th of ballots, etc.

Results

Results Option % of points Voters
Agree 0.0% 0  
Abstain 75.0% 3 SW WW PF
Disagree 25.0% 1 AZ
Block 0.0% 0  
Undecided 0% 0  

4 of 4 people have voted (100%)

PF

parker friedland
Abstain
Thu 12 Apr 2018 8:26AM

Not sure if it is proportional

PF

parker friedland
Abstain
Thu 12 Apr 2018 8:28AM

Not sure if it is proportional.

SW

Sara Wolf
Abstain
Tue 1 May 2018 8:50AM

I don't get what you are saying here.

AZ

Adam Zielinski
Disagree
Tue 4 Jun 2019 12:47AM

Too complicated

PF

Poll Created Thu 12 Apr 2018 8:35AM

Broader definition of STAR-PR Closed Mon 2 Nov 2020 8:03AM

Why don't we just refer to STAR-PR as a broader category of proportional voting methods where each voting method that satisfies Warren's Strong PR criterion while naturally reducing to STAR voting in the single winner case is it's own version of STAR-PR.

Results

Results Option % of points Voters
Agree 40.0% 2 WW PF
Abstain 20.0% 1 SW
Disagree 40.0% 2 AZ JB
Block 0.0% 0  
Undecided 0% 0  

5 of 5 people have voted (100%)

PF

parker friedland
Agree
Thu 12 Apr 2018 8:36AM

Supporters of different proportional STAR voting methods could all unite under the same name.

SW

Sara Wolf
Abstain
Tue 24 Apr 2018 7:02AM

I would define STAR-PR as the 5 star ballot proportional voting method that best meets the 5 pillars for a just voting system. Accuracy, Equality, Honesty, Simplicity, Expressiveness. More refining and testing are needed before it's a done deal.

JB

Jonathan Bright
Disagree
Sun 10 Mar 2019 3:48PM

"STAR-PR" should refer to a specific (recommended) system

AZ

Adam Zielinski
Disagree
Tue 4 Jun 2019 12:57AM

We should decide on something specific

CS

Clay Shentrup Thu 12 Apr 2018 11:14PM

My view is that RRV makes the runoff component of STAR kind of superfluous, so at most I think it should only exist in the first round. Doing it the last round adds significantly more complexity because you reweight the actual head-to-head majority, which seems weird. Doing it every round is VERY complex compared to straight RRV.

That said, I still think STAR-PR as currently envisioned is superior to most other ideas.

PF

parker friedland Fri 13 Apr 2018 12:55AM

I originally decided on having STAR on the first round, but then I realized that by doing it on the last round, voter's can't manipulate how their ballots are re-weighted when electing (which is something they can do when you have it during the first round), however the ballots are never re-weighted again after the last round, which makes that kind of manipulation impossible.

SW

Sara Wolf Tue 1 May 2018 8:41AM

Parker. I'd like to chime in on that CES thread but it won't let me comment. Do I need to be added or invited or something.
https://groups.google.com/forum/#!topic/electionscience/HkgoqQiGN4k
For the record the PDX chapter is not set on the exact algorithm for STAR-PR and we are 100% focused on the single winner ballot initiatives. That said, we are confident that the specific proposal outlined would be a huge improvement to our current at large voting system used by city council and other groups here and we want to plant seeds for STAR-PR. I'd like it to be a lot simpler personally and I also think that for voter education its important that the ballot itself match the STAR single winner ballot.

We're currently in the process of doing more research into how variations perform and are very interested in simulations to compare how the different systems compare in real life. Edge case scenarios and pass/fail criteria are interesting and relevant, but we're really interested in the bigger picture.

I'm not a mathematician so if you'd like to convince me of something here you might need to explain the point in English. Thanks for caring about this and weighing in!

PS. for effectiveness sake Loomio really works best if we have broad agreement of a question and understand the options here in the thread before we make proposals.