Re: [EM] Why I think IRV isn't a serious alternative KD
At 04:31 AM 12/21/2008, Kristofer Munsterhjelm wrote: In any case, it may be possible to have one of the LNHs and be monotonic and have mutual majority. I'm not sure, but perhaps (doesn't one of DAC or DSC do this?). If so, it would be possible to see (at least) whether people strategize in the direction of early truncation by looking at methods that fail LNHarm but pass LNHelp; that is, Bucklin. Was bullet voting pervasive under Bucklin? In some contexts, yes. However, we see upwards of 30% or so usage of additional preferences in the municipal elections I've looked at. I consider that high. Bullet voting occurs for reasons other than LNH concern. As Lewis Carroll pointed out, it's simply how many people will vote, representing their best knowledge, they may not have sufficient knowledge to intelligently rank or rate the rest of the candidates. Further, if they have strong preference for their favorite over all others, they may not care to vote for any of the others, not wanting to contribute to the victory of any of them. Voting is a moral action, and choosing the lesser of two evils isn't always the best thing to do. Sometimes the best action is to reject both evils, and that's what a bullet vote for the best candidate could be doing. In other words, Nader supporters in 2000, if they really believed that Gore and Bush were Tweedledum and Tweedledee, might not have added an additional ranked choice for Gore even if the method had allowed it, and LNH has nothing to do with that. We don't know, unless we do some serious ballot analysis -- the necessary information is available from a few elections now -- how many IRV voters truncate, because we don't know the lower preference expressions from those who did vote for a frontrunner. My guess is that the numbers are quite similar to what I've seen with Bucklin historically and what I'd expect from Bucklin today. We can stil get some idea of how easily voters would strategize by looking at Bucklin, though; or for that matter, at ranked voting methods that fail both LNHs. Schulze's used in some technical associations (Debian, Wikimedia), and, although I don't have raw voting data, they seem to be mostly honest. The Wikimedia election had no Condorcet cycles down to the sixth place, for instance. What I've seen from Bucklin, there is a very extensive analysis of the Cleveland election of 1915, I think it was, is that voters who didn't want to vote for a candidate didn't. Truncation, at least in Bucklin, is not insincere! All things considered, the numbers of additional preference votes are actually higher than I'd have expected. FairVote claims additional preference votes on the order of 11% in a series of Alabama party primary elections, and that majority failure was universal. I'm not sure what to make of that, beyond a possibility that most primary voters simply knew who their favorite was and trusted that the plurality favorite would be good enough. In nonpartisan elections, it seems, regardless of theory, the first preference leader wins the election, exceptions have to be pretty rare. (None so far in the U.S. with well over thirty such elections.) 11% additional preference will flip some elections, and apparently it did. Indeed, some of the opposition to Bucklin seems to have come from parties and candidates who lost elections due to additional preference votes, considering that this somehow violated their basic right to win if they get the most first preference votes. Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Why I think IRV isn't a serious alternative KD
Kristofer, Woodall's DAC and DSC and Bucklin and Woodall's similar QLTD all meet mono-raise and Mutual Majority (aka Majority for Solid Coalitions). DSC meets LNHarm and the rest meet LNHelp. Chris Benham Kristofer Munsterhjelm wrote (Sun.Dec.21): In any case, it may be possible to have one of the LNHs and be monotonic and have mutual majority. I'm not sure, but perhaps (doesn't one of DAC or DSC do this?). If so, it would be possible to see (at least) whether people strategize in the direction of early truncation by looking at methods that fail LNHarm but pass LNHelp; that is, Bucklin. Stay connected to the people that matter most with a smarter inbox. Take a look http://au.docs.yahoo.com/mail/smarterinbox Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Why I think IRV isn't a serious alternative KD
James Gilmour wrote: Kevin Venzke > Sent: Tuesday, December 16, 2008 1:49 PM The reason I believe LNHarm is more valuable than monotonicity is that when a method fails LNHarm, the voter is more likely to realize in what insincere way to vote differently, in order to compensate. When a method fails monotonicity, a voter will rarely know to do anything differently because of it. LNH is important to ordinary electors, as I have explained in a recent post, at least where the voting system is susceptible to LNH effects. If the vote counting method is not LNH-compliant, electors are likely to vote strategically in an attempt to avoid or mitigate the effects of LNH-failure or to try to gain some real or imagined advantage from its effects. Monotonicity, or more specifically, the lack of monotonicity, is of no importance whatsoever in public elections because neither candidates nor voters can exploit it. It would be "nice" if the vote counting system were monotonic, but we cannot have monotonicity AND some of the other criteria we consider desirable. For example, monotonicity and later-no-harm are incompatible in IRV and STV-PR. Of the two, LNH is important - non-monotonicity is irrelevant. We can't have both LNHs, mutual majority, and monotonicity (by Woodall). FPTP has LNH* (simply because later choices are ignored) and monotonicity. IRV has LNH* and mutual majority, but not monotonicity. I'd say that IRV's monotonicity problem is indeed a problem, because it's so pervasive. Just look at Yee diagrams. On the other hand, I'm not unbiased, and so I may be saying that because it's "unaesthetic". In any case, it may be possible to have one of the LNHs and be monotonic and have mutual majority. I'm not sure, but perhaps (doesn't one of DAC or DSC do this?). If so, it would be possible to see (at least) whether people strategize in the direction of early truncation by looking at methods that fail LNHarm but pass LNHelp; that is, Bucklin. Was bullet voting pervasive under Bucklin? Unfortunately, no method that passes only LNHarm has been used, so we can't do the same there (to see if there was pervasive random filling in that case). We can stil get some idea of how easily voters would strategize by looking at Bucklin, though; or for that matter, at ranked voting methods that fail both LNHs. Schulze's used in some technical associations (Debian, Wikimedia), and, although I don't have raw voting data, they seem to be mostly honest. The Wikimedia election had no Condorcet cycles down to the sixth place, for instance. James Gilmour No virus found in this outgoing message. Checked by AVG - http://www.avg.com Version: 8.0.176 / Virus Database: 270.9.19/1857 - Release Date: 19/12/2008 10:09 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] Why I think IRV isn't a serious alternative KD
Here's one explanation to why LNH might be more important to voters than monotonicity. Most voters are used to election methods where they bullet vote one of the candidates. It is a quite natural thought that if one votes multiple candidates on a single ballot then the vote that the second favourite gets may take some power away from the first favourite. This may not be based on facts, but of course we don't expect the voters to be aware of all the technical properties of the methods. The voter may be happy and trust the experts if they tell that this particular method is LNH compatible and therefore they can mark also their later preferences without losing power. Also this may be based on facts but need not be. Strictly speaking it is enough if it is more probable that later preferences will help the voter than that they will cause harm. Therefore a method that is LNH compatible in 90% of the cases may be good enough, i.e. the voter could happily mark also the later preferences in the ballot. Monotonicity is a similar but reverse case. Typical voters expect their vote to support the candidates that they vote for. They don't expect it to make the results worse from their point of view. Therefore the voters are not afraid of non-monotonicity but are happy to vote although it could be that their vote will make the results worse. If the experts would convincingly tell the voters that their vote may actually lead to a worse result, then maybe some voters would stop voting in the hope of improving the results. But also here it can be claimed that it is enough if it is more probable that the vote will improve the results than it is to make them worse. Voters may thus find LNH more important than monotonicity. But this does not mean that the method should be formally LNH compatible. It is good enough if the methods typically behave as wanted. It is easier for experts to convince the voters if they themselves are convinced of their cause, e.g. as a result of a compatibility proof that proves that the method meets some criterion 100%. Sufficient compatibility with the criteria may thus often be enough. And risks may be higher in areas where the voters don't expect risks than in areas where they expect to find them. But the fact remains that voters may fear some threats more than others, rationally or irrationally. Juho --- On Sun, 21/12/08, James Gilmour wrote: > > The reason I believe LNHarm is more valuable than > > monotonicity is that when a method fails LNHarm, the > voter is > > more likely to realize in what insincere way to vote > > differently, in order to compensate. When a method > fails > > monotonicity, a voter will rarely know to do anything > > differently because of it. > > LNH is important to ordinary electors, as I have explained > in a recent post, at least where the voting system is > susceptible to LNH > effects. If the vote counting method is not LNH-compliant, > electors are likely to vote strategically in an attempt to > avoid or > mitigate the effects of LNH-failure or to try to gain some > real or imagined advantage from its effects. > > Monotonicity, or more specifically, the lack of > monotonicity, is of no importance whatsoever in public > elections because neither > candidates nor voters can exploit it. It would be > "nice" if the vote counting system were monotonic, > but we cannot have > monotonicity AND some of the other criteria we consider > desirable. For example, monotonicity and later-no-harm are > incompatible in > IRV and STV-PR. Of the two, LNH is important - > non-monotonicity is irrelevant. > > James Gilmour > No virus found in this outgoing message. > Checked by AVG - http://www.avg.com > Version: 8.0.176 / Virus Database: 270.9.19/1857 - Release > Date: 19/12/2008 10:09 > > > > 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] Why I think IRV isn't a serious alternative KD
Kevin Venzke > Sent: Tuesday, December 16, 2008 1:49 PM > The reason I believe LNHarm is more valuable than > monotonicity is that when a method fails LNHarm, the voter is > more likely to realize in what insincere way to vote > differently, in order to compensate. When a method fails > monotonicity, a voter will rarely know to do anything > differently because of it. LNH is important to ordinary electors, as I have explained in a recent post, at least where the voting system is susceptible to LNH effects. If the vote counting method is not LNH-compliant, electors are likely to vote strategically in an attempt to avoid or mitigate the effects of LNH-failure or to try to gain some real or imagined advantage from its effects. Monotonicity, or more specifically, the lack of monotonicity, is of no importance whatsoever in public elections because neither candidates nor voters can exploit it. It would be "nice" if the vote counting system were monotonic, but we cannot have monotonicity AND some of the other criteria we consider desirable. For example, monotonicity and later-no-harm are incompatible in IRV and STV-PR. Of the two, LNH is important - non-monotonicity is irrelevant. James Gilmour No virus found in this outgoing message. Checked by AVG - http://www.avg.com Version: 8.0.176 / Virus Database: 270.9.19/1857 - Release Date: 19/12/2008 10:09 Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Why I think IRV isn't a serious alternative KD
At 08:49 AM 12/16/2008, Kevin Venzke wrote: Thus, *all things being equal* (which must be kept in mind if it's IRV that is on your mind), I would expect that failing LNHarm will provoke more insincerity (and thus destroy more information) than failing monotonicity. Highly speculative. Bucklin probably experiences about the same level of bullet voting due to LNH fears as IRV, not much more, because the "harm" only happens when a majority isn't found in the first round. And it isn't really a "harm," that's an unfortunate aspect of the name. What the second vote harms is not exactly the favorite, but the first vote for the favorite, and only in the specific pairwise election against the additionally approved candidate. It is backing up and saying, okay, if my candidate isn't getting a majority from favorite votes, this additional candidate is also acceptable to me. And, sure, that can cause the additional candidate to win. If that candidate is also approved by more people than my favorite. (Multiple majorities are rare, BTW, I've never heard of one happening in a Bucklin election, and I doubt that it ever happened. So, if the second vote "harms the favorite," it's because lower-ranked candidate got a majority, and the favorite didn't. IRV can pretend that there is no harm from the second vote for two reasons: it eliminates the first candidate before using the second vote. The *voter* hasn't harmed the first candidate, because the *method* has. Requiring that the favorite be eliminated before a second rank choice is considered cuts two ways. It prevents this alleged "harm," but it also prevents "help." I.e., second rank votes coming from other candidates; those are exactly the votes which would allow IRV to find a "compromise winner," as Robert's Rules notes. In other words, Center Squeeze is a direct consequence of LNH compliance by IRV. LNH is also incompatible with actual runoff voting when there is majority failure, unless the runoff is simply the top two IRV candidates. In Vermont, the governor must be elected by a majority. If there is majority failure, the election goes to the legislature, which votes by secret ballot (Plurality election, if I'm correct) from among the top three candidates. The IRV legislation introduced by Terrill Bouricius had a ballot instruction included: voting for a lower ranked candidate can't hurt your favorite. It wasn't true. Your lower ranked vote could cause the election to complete, whereas without it, there would be further process, which your candidate could win. Even Robert's Rules of Order gets this one wrong. Interesting, eh? Top three. A Condorcet winner is almost certainly in there! IRV has other issues that can lead to a different conclusion, but that isn't what I was discussing. Kevin Venzke 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] Why I think IRV isn't a serious alternative KD
Hi Kathy, You are responding to me, not Abd ul-Rahman Lomax. --- En date de : Mar 16.12.08, Kathy Dopp a écrit : > > Hi, > > > > --- En date de?: Dim 14.12.08, Abd ul-Rahman Lomax > a ?crit?: > >> > >> That's not very generous. I can > think of > >> a couple of defenses. One would > >> > >> be to point out that it is > necessitated by > >> the other criteria that IRV > >> > >> satisfies. All things being equal, I > consider > >> LNHarm more desirable than > >> > >> monotonicity, for instance. > > Abd ul, > > That is about the strangest position I've seen you take > on any subject > because it is equivalent to saying that it is more > important for a > voting method not to hurt my lower choice candidates than > my first > choice candidates. The reason I believe LNHarm is more valuable than monotonicity is that when a method fails LNHarm, the voter is more likely to realize in what insincere way to vote differently, in order to compensate. When a method fails monotonicity, a voter will rarely know to do anything differently because of it. Thus, *all things being equal* (which must be kept in mind if it's IRV that is on your mind), I would expect that failing LNHarm will provoke more insincerity (and thus destroy more information) than failing monotonicity. IRV has other issues that can lead to a different conclusion, but that isn't what I was discussing. Kevin Venzke Election-Methods mailing list - see http://electorama.com/em for list info