What Multiwinner Method To Push For Local Boards?
-
@jack-waugh said in What Multiwinner Method To Push For Local Boards?:
What pros and cons could be given about staying with STV vs. switching to SPAV?
I'm guessing SPAV would be easier to count. SPAV is also monotonic, unlike STV (normal STV anyway - I think Schulze STV might be monotonic but I'm not sure it's been used for anything ever and it's very complex). In STV's favour, it's more tried and tested over the years. Also SPAV can be disproportional with overlapping factions.
It also depends on what sort of ballots people prefer to use. STV is ranked. SPAV is approval voting. The age-old debate.
-
Is monotonicity equally so important for the multiwinner context as it is in the single-winner context?
-
@jack-waugh said in What Multiwinner Method To Push For Local Boards?:
For a county board of supervisors, for example, what's a nice simple multiwinner system to suggest?
Would Allocated Score (STAR_PR) work well?
-
@gregw said in What Multiwinner Method To Push For Local Boards?:
@jack-waugh said in What Multiwinner Method To Push For Local Boards?:
For a county board of supervisors, for example, what's a nice simple multiwinner system to suggest?
Would Allocated Score (STAR_PR) work well?
If you're using scores it's probably better to use a cleaner system like SPAV + KP or Phragmén + KP.
-
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
If you're using scores it's probably better to use a cleaner system like SPAV + KP or Phragmén + KP.
By cleaner do you mean better criterion adherence or less complex?
-
@gregw said in What Multiwinner Method To Push For Local Boards?:
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
If you're using scores it's probably better to use a cleaner system like SPAV + KP or Phragmén + KP.
By cleaner do you mean better criterion adherence or less complex?
Probably both. When Allocated Score was chosen to be the "STAR-PR" method, there were a few other options that worked in the same basic way. That is - they elect a candidate and then remove a quota of votes, and then go through the same process until the seats are filled.
But I would say that quota-removal is a very clunky way to go about PR. Quotas are essentially arbitrary, especially as later-elected candidates might not reach a full quota so get elected more "cheaply". I discussed the problem of quotas here as well. My conclusion was that essentially quota removal methods are just a poor man's Phragmén.
The other thing is that the difference between the methods that went up for consideration to become "STAR-PR" was all about how they dealt with scores. For approval voting, they were all essentially the same method. But scores are messy to deal with, and all these methods (including Allocated Score) had their own unsatisfactory and messy way of dealing with these scores. The KP-transformation simply cleans it up. It's a consistent way of dealing with scores that takes away the weird edge cases and discontinuities that the other methods threw up. And then you just use your base approval method once the scores have been converted to approvals. See my post here as well.
-
@jack-waugh said in What Multiwinner Method To Push For Local Boards?:
Is monotonicity equally so important for the multiwinner context as it is in the single-winner context?
Yes—it makes no sense that, if I give a candidate an extra star, we respond by deciding the candidate is "too good to win" now. It also makes honest voting impossible (because ranking A over B is no longer the same as giving A more support, so you can't give A the correct level of support without knowing everyone else's exact ballot).
-
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
If you're using scores it's probably better to use a cleaner system like SPAV + KP or Phragmén + KP.
Thank you for your analysis of quotas and the suggestion of SPAV + KP.
Concerning SPAV and SPAV + KP:
Would SPAV with Jefferson give too much favor to large parties and/or tend towards block voting?
Would Webster give too much favor to small parties?
Would a Score election using SPAV + KP make any difference regarding the possibility of block voting and the use of Jefferson or Webster? -
@gregw said in What Multiwinner Method To Push For Local Boards?:
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
If you're using scores it's probably better to use a cleaner system like SPAV + KP or Phragmén + KP.
Thank you for your analysis of quotas and the suggestion of SPAV + KP.
Concerning SPAV and SPAV + KP:
Would SPAV with Jefferson give too much favor to large parties and/or tend towards block voting?
Would Webster give too much favor to small parties?
Would a Score election using SPAV + KP make any difference regarding the possibility of block voting and the use of Jefferson or Webster?I would say that Webster gives objectively the most proportional result (wrongheaded arguments in favour of Huntington-Hill elsewhere on this forum notwithstanding) rather than favouring small parties. Jefferson does favour large parties, but it is more strategically robust. Large parties or factions can deliberately split up and force a Jefferson result under Webster voting if they can co-ordinate well enough.
-
I admit I am desperately seeking voting systems that are appealing and accurate. So I do appreciate the help I have received on this forum. When I first decided to advocate for better voting systems I had no idea how difficult the choices would be.
I think that voters will prefer Score ballots over Approval ballots, but I could be wrong. Has anyone done any polls on this?
Is there any reason to think SPAV would be more or less proportional than SPAV + KP?
-
@gregw said in What Multiwinner Method To Push For Local Boards?:
I think that voters will prefer Score ballots over Approval ballots, but I could be wrong.
Maybe a better question is which is more likely to go over with politicians who could change it. Not all States have initiative and referendum.
Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].
-
@jack-waugh said in What Multiwinner Method To Push For Local Boards?:
Maybe a better question is which is more likely to go over with politicians who could change it. Not all States have initiative and referendum.
Good point. Even in ballot initiative states we will be talking to politicians. Part of the plan is to "Free the Politicians".
-
@gregw said in What Multiwinner Method To Push For Local Boards?:
I admit I am desperately seeking voting systems that are appealing and accurate. So I do appreciate the help I have received on this forum. When I first decided to advocate for better voting systems I had no idea how difficult the choices would be.
I think that voters will prefer Score ballots over Approval ballots, but I could be wrong. Has anyone done any polls on this?
Is there any reason to think SPAV would be more or less proportional than SPAV + KP?
Well, SPAV is purely approval whereas SPAV + KP is scores, so which ends up being more proportional might depend on exactly how you define proportional and also how people vote in practice. There's always been the question with score voting of whether some voters will lose out by casting a more honest ballot but losing out strategically.
-
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
Well, SPAV is purely approval whereas SPAV + KP is scores, so which ends up being more proportional might depend on exactly how you define proportional and also how people vote in practice.
How do I define proportional?
When electing representatives from a multi-member district the goal is to, as accurately as possible, reflect voter support for each party.
When electing representatives from a party list the idea is to find n candidates who will be the best elected officeholders.
When choosing candidates for a party list the idea is to find n candidates with the greatest appeal and strong party loyalty.
When choosing the best four or five candidates to advance from a blanket primary to a single-winner general election we want four or five strong, politically diverse candidates for voters to choose from. A choose one, or cumulative vote might be better than a proportional vote.
When electing a local board, having never participated in a local board, I haven't the foggiest.
Perhaps we need to persuade the world to try more voting systems, so we can answer three questions; which systems work the best, which systems voters prefer, and which systems politicians prefer. Voting equipment manufacturers will bitch.
At this point in time people seem to prefer voting systems that do not work well.
-
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
Well, SPAV is purely approval whereas SPAV + KP is scores, so which ends up being more proportional might depend on exactly how you define proportional and also how people vote in practice. There's always been the question with score voting of whether some voters will lose out by casting a more honest ballot but losing out strategically.
Thus my question in another thread, about whether Harmonic voting might lose the stable winner set properties of PAV. The stable winner set seems like it could provide some very strong strategy-resistance properties, similar to Condorcet in single-winner elections.
-
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
There's always been the question with score voting of whether some voters will lose out by casting a more honest ballot but losing out strategically.
When the voting system is Choose-one Plurality (bad COP), do they lose out by casting a more honest ballot but losing out tactically?
Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].
-
I can see that Cardinal systems may be more be more tempting to strategists, but folks can rank insincerely as well.
One reason IRV is less tempting to strategists is the capricious nature of IRV tabulations makes strategy backfires are more likely.
Group strategy is difficult to pull off, especially if you want to be discreet.
Individual strategy is the larger issue. For individuals the most common strategy is Score or IRV is to give your favorite a top score or rank and your feared rival a bottom score or rank. For many people this would be a sincere vote, leading to the fear that people who view the world in black and white may have more voting effect, particularly in Score.
Your suggestions from other threads of 100, 99, 1, 0 ballots and 100,99, 90, 50, 10, 1, 0 ballots could help. The instructions: score the best candidate 5, the worst 0, the others in comparison, ties are ok, also helps.
STAR is another attempt to ward off strategy. STAR has fails a lot of criterion because it has a utilitarian component and a majority component, yet it does very well in Jonathan Quinn's satisfaction simulations. Maybe there is some value halfway passing the mutually exclusive criterion. I believe it would take a good number of real world close three-way elections to find out.
I am rooting for the good folks in Eugene, OR. The election for their STAR ballot initiative concludes May 21st. I am perturbed by FUD dump by Unite America in the election campaign.
-
@lime said in What Multiwinner Method To Push For Local Boards?:
@toby-pereira said in What Multiwinner Method To Push For Local Boards?:
Well, SPAV is purely approval whereas SPAV + KP is scores, so which ends up being more proportional might depend on exactly how you define proportional and also how people vote in practice. There's always been the question with score voting of whether some voters will lose out by casting a more honest ballot but losing out strategically.
Thus my question in another thread, about whether Harmonic voting might lose the stable winner set properties of PAV. The stable winner set seems like it could provide some very strong strategy-resistance properties, similar to Condorcet in single-winner elections.
I don't think its strategy resistance is as strong as it would be with ranked ballots. With approval voting, you still have to decide whether to approve candidates you don't like as much because you think they've got a better chance of being elected.
Schulze STV uses ranked ballots and reduces to the Schulze Condorcet method in the single-winner case. It's probably more strategically robust than an approval-based method that satisfies core stability.
