Re: [EM] Largest-Remainder
Juho: In your explanation of "minimizing violated pinions", you described your own interpretation of proportionality. If you're interested in proportionality, the matter of optimally equal representation per voter, I've told you where to read about it. If you're interested in minimizing violated opinions, then PR is not for you. If I were doing PR, I'd be glad to instruct you in proportionality. And, if so, I'd be glad to walk you through the subject and and explain to you your specific errors. But, as I said, I no longer do PR. As it is, I can only refer you to the sources that I cited. Read them conscientiously, and don't cling to your assumptions. It's arrogant to believe that you're more qualified in a subject, when you don't want to take the time to study it. Don't cheat yourself in that way. That's my best advice for you. Now, forgive me if I say a few things that I've already said. I'm saying it with other things that I haven't said. For allocating seats to parties: I used to prefer Sainte-Lague (as recently as earlier in this discussion), because I felt that if you want PR at all, then you might as well get the best pure proportionality. Also, I guess I was influenced by others, who believed that the Repubs & Dems would remain the big 2, and wanted the small parties to get their fair share. I bought into that feeling. But I have more confidence in the voters than that. That's why I prefer single-seat districts, or at least the use of a good single-winner method instead of PR. I believe that if some of the small parties are better, then they won't remain small for long, when Approval is in use. But, as I said, I have nothing against PR--I just don't consider it necessary at all. But of course PR would be a lot better than Plurality or Runoff. I'm convinced that if the voters can support what they really like, then there will be all the improvement we could ask for. As I said, PR isn't viable here anyway, because it's a drastically different notion of representation and govt--whereas a better single-winner method is nothing other than a better way of doing what we already do--electing candidates to single-member districts. Because I don't consider PR necessary anyway, For allocating seats to parties, I now prefer d'Hondt, with its balance between proportionality and majoritarianism. (and, for STV, the Droop quota, for the same reason). Someone who doesn't consider PR necessary at all has no reason to insist on the purest all-out un-compromised Plurality of Sainte-Lague. But it just depends on what you want. I have no criticism of Sainte-Lague, or for the goal of optimally equal representation per person that requires Sainte-Lague. In fact, as I said, any of the PR systems and methods would be fine. Allocating seats to districts: That's a whole other ballgame. For that application, there's really no room for debate. Living in a some particular district is very different from preferring an unpopular party. If the party that you like best is unpopular, then that's the way it goes. But, regardless of which district you reside in, you have an obvious right to equal representation. To the greatest extent possible, you have a right to as much representation as anyone else. That's why Sainte-Lague is the way to allocate seats to districts. When allocating seats to districts, the question is: Do you or do you not want as much representation as the other people? If so, then the answer is Sainte-Lague. But, if you're happy with avoidably less representation than other people have, that's your business, and it isn't for me to tell you what you should want. Mike Ossipoff Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Declaration's policy on single-mark ballots (was Re: Do any of you have any thoughts about California's top-two primary?)
On 11.6.2012, at 0.46, Jameson Quinn wrote: > > > 2012/6/10 Juho Laatu > It is easy to fill the ballot in VPR. It is one step more difficult to check > the preferences of the candidates and decide whom to vote. If one goes one > step further in this simplification path, one might end at tree voting. We > could have a candidate that belongs to the free rifle group of the green > group of the socialist party. That's close to open lists but allows voters to > clearly position themselves to the level of a full binary tree, provides > proportionality also within the parties, allows voters to see easily what > each candidate intends to stand for, and is quite strategy free. Voters may > vote a green socialist or a socialist green, depending on which criterion is > more important to them. One can say that trees are policy oriented > (candidates rank themes) while VPR is person oriented (candidates rank > candidates). > > Trees show promise for eliminating voting paradoxes by limiting voter freedom. Yes. Voter freedom is limited to candidate given (theme) rankings. There are no voter specified rankings that could "jump between different branches". > However, you'd have to actually develop a system for building the trees. Just > assuming that they exist doesn't cut it; that hides hairy strategy and > coordination problems for the candidates, factions, and parties. Probably parties will decide which candidates they will take on their lists (in all methods). Even if that is the case, candidates could be allowed to freely position themselves in the tree within the party specific branch. Any group of (already qualified) candidates could in this approach establish a new "theme"/subgroup under the already existing ones. Alternatively parties would have full control of allocation of candidates in the tree wihin the party branch. In this case parties could plan the optimal tree structure for them. Candidates could be forced to groups that the party considers to be strategically best. Also in VPR parties could try to force their candidates to give certain kind of rankings (e.g. rank own party candidates first, or rank party favourites first). Maybe one should ban or discourage branches that have no political meaning (e.g. branches "A" and "B" that contain candidates with similar political opinions). Only clear political messages would be allowed, e.g. "green", "pro nuclear power". Whatever the rules are, the system should not be allowed to degrade to a mechanism where the party tries to dictate which candidates will be elected and which ones not. The approaches might be different in different societies. One robust approach would be to allow some officials/court decide which branches are proper political branches and which ones are not, and must therefore be flattened/combined. One could also have rules that encourage parties to use a good tree stratcture (e.g. rules that allow more candidates in "good" trees). > > Also, the way trees work is by privileging certain dimensions of a candidate > over others. How is one branch, opinion or "dimension" different from another? Do you mean that parties would build the trees and make them favour some candidates? > One set of dimensions which is almost sure to get short shrift is quality – > that is, intelligence, honesty, hard work, you name it. > > In other words, I'd be interested in reading about a system built from the > ground up around trees, but I don't think it's a good idea to vaguely > speculate that VPR would be even more perfect if we just sprinkled magic tree > dust on it. Ground up? Does that refer to candidate driven decisions instead of party leadership driven decisions? Juho > > Jameson > > Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Declaration's policy on single-mark ballots (was Re: Do any of you have any thoughts about California's top-two primary?)
2012/6/10 Juho Laatu > It is easy to fill the ballot in VPR. It is one step more difficult to > check the preferences of the candidates and decide whom to vote. If one > goes one step further in this simplification path, one might end at tree > voting. We could have a candidate that belongs to the free rifle group of > the green group of the socialist party. That's close to open lists but > allows voters to clearly position themselves to the level of a full binary > tree, provides proportionality also within the parties, allows voters to > see easily what each candidate intends to stand for, and is quite strategy > free. Voters may vote a green socialist or a socialist green, depending on > which criterion is more important to them. One can say that trees are > policy oriented (candidates rank themes) while VPR is person oriented > (candidates rank candidates). > Trees show promise for eliminating voting paradoxes by limiting voter freedom. However, you'd have to actually develop a system for building the trees. Just assuming that they exist doesn't cut it; that hides hairy strategy and coordination problems for the candidates, factions, and parties. Also, the way trees work is by privileging certain dimensions of a candidate over others. One set of dimensions which is almost sure to get short shrift is quality – that is, intelligence, honesty, hard work, you name it. In other words, I'd be interested in reading about a system built from the ground up around trees, but I don't think it's a good idea to vaguely speculate that VPR would be even more perfect if we just sprinkled magic tree dust on it. Jameson Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Declaration's policy on single-mark ballots (was Re: Do any of you have any thoughts about California's top-two primary?)
It is easy to fill the ballot in VPR. It is one step more difficult to check the preferences of the candidates and decide whom to vote. If one goes one step further in this simplification path, one might end at tree voting. We could have a candidate that belongs to the free rifle group of the green group of the socialist party. That's close to open lists but allows voters to clearly position themselves to the level of a full binary tree, provides proportionality also within the parties, allows voters to see easily what each candidate intends to stand for, and is quite strategy free. Voters may vote a green socialist or a socialist green, depending on which criterion is more important to them. One can say that trees are policy oriented (candidates rank themes) while VPR is person oriented (candidates rank candidates). Juho On 10.6.2012, at 20.22, Steve Eppley wrote: > It's a bad idea for the Declaration to denounce all single-mark ballot > methods, because one of them--Vote for a Published Ranking (VPR)--has > desirable properties that distinguish it from the others. (One can also make > an argument that VPR is better than many voting methods that require more > complicated ballots.) > > VPR: > Two weeks before election day, each candidate publishes a top-to-bottom > ordering > of all the candidates. (Any candidate who fails to meet the deadline will > be treated > as if s/he'd ranked him/herself on top and all others tied for bottom.) > > On election day, each voter simply selects one candidate. > > Then each vote is treated as if it were the ordering published by its > selected > candidate. These orderings are tallied by a good preference order > algorithm > to determine the winner. > > Some interesting variations: > 1. Give each candidate the opportunity to withdraw after the vote totals are > published; withdrawn candidates will be dropped to the bottom of each > ordering before the orderings are tallied. With this option, tallying > algorithms such as plurality rule & instant runoff would become nearly as > good as condorcet algorithms because withdrawal would mitigate their > vote-splitting problem. (Borda would still be terrible due to its clones > problem.) Also, withdrawal would be useful in presidential elections--with > VPR and other voting methods--to help candidates avoid fragmenting the > Electoral College. > 2. Technology permitting, allow each voter to select an ordering published by > a candidate or by a non-government organization (NGO). Some example NGOs: > the New York Times, the Sierra Club, the National Rifle Association... > 3. Technology permitting, let each voter modify the ordering published by her > selected candidate, before submitting it as her vote. > > Obviously, being a "single-mark" method, VPR maximizes simplicity. Yet it > can be expected to handle the vote-splitting problem well. It ought to > typically allow each voter to vote for her sincere favorite, assuming her > favorite publishes an ordering the voter considers reasonable. (Or > strategically reasonable. If an election has a strategy problem, the voter's > favorite can handle it by publishing a strategic ordering, or by withdrawing > if necessary, if withdrawal is an option.) > > Also, VPR would make it easier for good candidates to win without spending a > lot of money, since they can win by persuading other candidates to rank them > over worse candidates. For example, Centrist might persuade Left to rank > Centrist over Right, and Right to rank Centrist over Left. Furthermore, an > honest centrist might persuade Left & Right to rank her over corrupt > centrists, and when she can't due to Left & Right also being corrupt, the > corrupt orderings they publish would presumably attract negative attention > during the two weeks preceding the election, reducing their votes. > > Regards, > Steve Eppley > --- > On 6/8/2012 2:20 PM, Richard Fobes wrote: >> Although this is a bit of a simplification, the "top-two" runoff form of >> voting in the U.S. consists of using single-mark ballots combined with a >> variation of instant-runoff voting. > -snip- >> The way this fits into the "Declaration of Election-Method Reform Advocates" >> is that the Declaration denounces single-mark ballots, regardless of how >> they are counted. > -snip- >> I think the easiest way to explain the concept is in the context of vote >> splitting, > Richard Fobes >> > On 6/7/2012 8:31 AM, Adrian Tawfik wrote: > -snip- > > Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Declaration's policy on single-mark ballots (was Re: Do any of you have any thoughts about California's top-two primary?)
2012/6/10 Steve Eppley > It's a bad idea for the Declaration to denounce all single-mark ballot > methods, But it doesn't. Also, I think we should start calling it the DEMRA. > because one of them--Vote for a Published Ranking (VPR)--has desirable > properties that distinguish it from the others. (One can also make an > argument that VPR is better than many voting methods that require more > complicated ballots.) > > VPR: > Two weeks before election day, each candidate publishes a > top-to-bottom ordering > of all the candidates. (Any candidate who fails to meet the deadline > will be treated > as if s/he'd ranked him/herself on top and all others tied for bottom.) > > On election day, each voter simply selects one candidate. > Then each vote is treated as if it were the ordering published by its > selected > candidate. These orderings are tallied by a good preference order > algorithm > to determine the winner. > This is very similar to SODA. It is simpler to explain (though not as much as you might think, because you still have to explain the underlying ranked system used), but I think worse in several regards. Still, I'd agree that it is a very good method overall. > Some interesting variations: > 1. Give each candidate the opportunity to withdraw after the vote totals > are published; withdrawn candidates will be dropped to the bottom of each > ordering before the orderings are tallied. With this option, tallying > algorithms such as plurality rule & instant runoff would become nearly as > good as condorcet algorithms because withdrawal would mitigate their > vote-splitting problem. (Borda would still be terrible due to its clones > problem.) Also, withdrawal would be useful in presidential elections--with > VPR and other voting methods--to help candidates avoid fragmenting the > Electoral College. > SODA effectively allows withdrawal in its default form. > 2. Technology permitting, allow each voter to select an ordering published > by a candidate or by a non-government organization (NGO). Some example > NGOs: the New York Times, the Sierra Club, the National Rifle Association... > Pretty soon, you'd have all permutations available, and it becomes just equivalent to the underlying method. Perhaps a bit simpler for a trusting voter, a bit more complicated for a non-trusting voter. > 3. Technology permitting, let each voter modify the ordering published by > her selected candidate, before submitting it as her vote. > Ouch. Suddenly the ballot design is a nightmare. Don't like this one at all. > > Obviously, being a "single-mark" method, VPR maximizes simplicity. No, SODA is simpler for voters. You can unintentionally spoil your ballot in VPR, but not in SODA. > Yet it can be expected to handle the vote-splitting problem well. Assuming the candidates are honest, yes. However, I think I see ways that rational candidate strategy could still give pathological results, in situations that SODA handles well. > It ought to typically allow each voter to vote for her sincere favorite, > assuming her favorite publishes an ordering the voter considers reasonable. > (Or strategically reasonable. If an election has a strategy problem, the > voter's favorite can handle it by publishing a strategic ordering, or by > withdrawing if necessary, if withdrawal is an option.) > Similar to SODA in this regard, and yes, this is an important strength. > > Also, VPR would make it easier for good candidates to win without spending > a lot of money, since they can win by persuading other candidates to rank > them over worse candidates. As with SODA in this sense, though SODA has a failsafe so that if this kind of thing leads to a total "surprise" winner, perhaps because of dishonest strategy, at least one other candidate will be able to back out, letting their stronger opponent win instead of the dark horse they'd said they preferred (perhaps without proper vetting). > For example, Centrist might persuade Left to rank Centrist over Right, > and Right to rank Centrist over Left. Furthermore, an honest centrist > might persuade Left & Right to rank her over corrupt centrists, and when > she can't due to Left & Right also being corrupt, the corrupt orderings > they publish would presumably attract negative attention during the two > weeks preceding the election, reducing their votes. > Again, I think this is a very good method; I just think that SODA is better. Jameson > > Regards, > Steve Eppley > --- > On 6/8/2012 2:20 PM, Richard Fobes wrote: > >> Although this is a bit of a simplification, the "top-two" runoff form of >> voting in the U.S. consists of using single-mark ballots combined with a >> variation of instant-runoff voting. >> > -snip- > >> The way this fits into the "Declaration of Election-Method Reform >> Advocates" is that the Declaration denounces single-mark ballots, >> regardless of how they are counted. >> > -snip- > >> I thi
[EM] Declaration's policy on single-mark ballots (was Re: Do any of you have any thoughts about California's top-two primary?)
It's a bad idea for the Declaration to denounce all single-mark ballot methods, because one of them--Vote for a Published Ranking (VPR)--has desirable properties that distinguish it from the others. (One can also make an argument that VPR is better than many voting methods that require more complicated ballots.) VPR: Two weeks before election day, each candidate publishes a top-to-bottom ordering of all the candidates. (Any candidate who fails to meet the deadline will be treated as if s/he'd ranked him/herself on top and all others tied for bottom.) On election day, each voter simply selects one candidate. Then each vote is treated as if it were the ordering published by its selected candidate. These orderings are tallied by a good preference order algorithm to determine the winner. Some interesting variations: 1. Give each candidate the opportunity to withdraw after the vote totals are published; withdrawn candidates will be dropped to the bottom of each ordering before the orderings are tallied. With this option, tallying algorithms such as plurality rule & instant runoff would become nearly as good as condorcet algorithms because withdrawal would mitigate their vote-splitting problem. (Borda would still be terrible due to its clones problem.) Also, withdrawal would be useful in presidential elections--with VPR and other voting methods--to help candidates avoid fragmenting the Electoral College. 2. Technology permitting, allow each voter to select an ordering published by a candidate or by a non-government organization (NGO). Some example NGOs: the New York Times, the Sierra Club, the National Rifle Association... 3. Technology permitting, let each voter modify the ordering published by her selected candidate, before submitting it as her vote. Obviously, being a "single-mark" method, VPR maximizes simplicity. Yet it can be expected to handle the vote-splitting problem well. It ought to typically allow each voter to vote for her sincere favorite, assuming her favorite publishes an ordering the voter considers reasonable. (Or strategically reasonable. If an election has a strategy problem, the voter's favorite can handle it by publishing a strategic ordering, or by withdrawing if necessary, if withdrawal is an option.) Also, VPR would make it easier for good candidates to win without spending a lot of money, since they can win by persuading other candidates to rank them over worse candidates. For example, Centrist might persuade Left to rank Centrist over Right, and Right to rank Centrist over Left. Furthermore, an honest centrist might persuade Left & Right to rank her over corrupt centrists, and when she can't due to Left & Right also being corrupt, the corrupt orderings they publish would presumably attract negative attention during the two weeks preceding the election, reducing their votes. Regards, Steve Eppley --- On 6/8/2012 2:20 PM, Richard Fobes wrote: Although this is a bit of a simplification, the "top-two" runoff form of voting in the U.S. consists of using single-mark ballots combined with a variation of instant-runoff voting. -snip- The way this fits into the "Declaration of Election-Method Reform Advocates" is that the Declaration denounces single-mark ballots, regardless of how they are counted. -snip- I think the easiest way to explain the concept is in the context of vote splitting, Richard Fobes > On 6/7/2012 8:31 AM, Adrian Tawfik wrote: -snip- Election-Methods mailing list - see http://electorama.com/em for list info
