Talk:Instant-runoff voting controversies

From Wikipedia, the free encyclopedia

This article was nominated for deletion on 18 October 2007. The result of the discussion was no consensus.

	This article is within the scope of WikiProject Elections and Referenda, an ongoing effort to improve the quality of, expand upon and create new articles relating to elections, electoral reform and other aspects of democratic decisionmaking. For more information, visit our project page.
Stub	This article has been rated as stub-Class on the assessment scale.

Contents 1 Rationale for this page's creation 2 General ranting about voting systems, and the people who argue about them 3 Reorganization of page begun 4 "Instant runoff would end spoiler effect in elections." 4.1 Anti: "If it ain't broke don't fix it" or "Plurality Voting is Good Enough" 5 Thanks to Tbouricius and Tomruen for a very helpful series of edits. 6 About the one-person one-vote section 7 Bad to less bad? 8 faulty logic 9 Pro arguments. 10 Do not remove visualizations! 11 Visualizations of voting methods: a proposal to add to all relevant articles 12 wiki.electorama.com/wiki 13 IRV critics in Minnesota 14 Not an ad for approval voting 14.1 Request for third opinion 15 POV and disputed-section tags 15.1 taking out the disputed tag from this section 16 Wattles v. Upjohn (1920) 16.1 Pro: IRV is easy to understand 17 Segregated sections 18 About the use of advocacy groups as sources for this article. 19 edits to this article by a block-evading sock of User:Fredrick day 20 Do the CON arguments in this article apply to "any voting method"? 20.1 With IRV, voting for a candidate can cause the candidate to lose 20.2 IRV suffers from the spoiler effect 20.3 IRV violates the one person one vote mandate 20.4 "if it ain't broke don't fix it" or "plurality voting is good enough" 20.5 IRV will help preserve the two-party system 20.6 Summary 21 About debate tag added to article 22 Yet another merge proposal

[edit] Rationale for this page's creation

User:Green Giant wrote[1]: "This is not a separate concept - it should be a seciton in the Instant-runoff voting article - otherwise it risks becoming a content-fork"

I disagree; see Opposition to the Vietnam War, Criticism of Wikipedia, Capital punishment debate, Controversies regarding Jehovah's Witnesses, etc. There is plenty of precedent for this type of article. Given the edit-warring at instant-runoff voting, I think it will be helpful to provide a separate article where the arguments against IRV can be fleshed out in as much detail as desired (at the IRV page, they would have to summarized and balanced with Pro arguments as well as other IRV-related content). Captain Zyrain 13:02, 15 October 2007 (UTC)

While I agree there is precedent for this sort of article, but edit-warring is not a reason to create this article - it would be heading in the direction of a content-fork. The place to flesh out the arguments is alongside the pro's so that readers can see both sides of the argument. However as you say it might become useful for the IRV opponents but I would remind you that Wikipedia is not a soapbox. Therefore I withdraw my request for a merger but if there is cause to do so later, I will request a merger again. Thank you for your civil behaviour, it is much appreciated. Green Giant 13:20, 15 October 2007 (UTC)

Since all this has come up again in another AfD, Green Giant correctly wrote: The place to flesh out the arguments is alongside the pro's so that readers can see both sides of the argument. That is correct. This article was designed to do that. The plan that was developed and more or less followed, with some raggedness in implementation, was that Pro and Con arguments were listed, and then *each argument* was examined. So pro arguments, for example, are undeniably true, they are not, at least not directly, disputed. Similarly some Con arguments. Classification into Pro and Con allows seeing, in the subject headers, a summary of the arguments made, but then each section looks at each argument, should attribute it, showing that it is a notable argument (not a straw man, which actually did exist as a Con argument in the article for some time, or from some isolated fanatic), and then "flesh it out," making the overall report of the argument NPOV. The article is *not* a place where some conclusion is drawn; rather, the goal is that each argument is seen in the light of what can be verifiably known. With sufficient discussion, usually, the proponents and opponents of IRV can come to agreement on this. What is far more difficult is to agree upon is how to *summarize* arguments as would generally need to be done in the main article. We tried to do that, putting the material in the main article and the proponents of IRV generally said "too much detail, confusing," and opponents said "this detail is necessary or the sound bite of the argument creates POV imbalance." So the plan was to create this article; if we can come to consensus on it (and for much of it, we did), then we at least have some common ground. There will still be, I'd predict, dispute over relative importance. In fact, there is a present dispute, as this is written, over an issue that has been the subject of some level of edit warring in the past: should the discussion in Robert's Rules of Order of "preferential voting" be mentioned in the introduction. Most specifically, should the easily interpetable as positive part of the discussion be reported, while leaving out the negative. I think it obvious, but the Executive Director of FairVote seems to disagree. I wonder why? Could it be that, over the last few years, this issue has been a huge propaganda coup for them, with their report and position being repeated, verbatim, in many different places around the internet, by government officials, etc. Some editors, looking at this debate, obviously think, "why are they arguing over this small point?" There is a reason. Pro-IRV editors have risked being blocked (and have been blocked) edit warring over this detail. It is very, very important to them to have that brief -- and misleading -- summary in the introduction. This particular debate has not been brought to this article, though it is possible it could. The report of what Robert's Rules actually does say is properly in the IRV article. Only the *implications* are argument, and it generally is not notable (i.e., the debate over it) --Abd (talk) 18:17, 15 May 2008 (UTC)

The opening paragraph is confusing. I don't at all understand the Bucklin voting reference at all in regards to IRV or any other methods listed. It somehow seems to imply to me that IRV is related to Bucklin. Tom Ruen 18:26, 15 October 2007 (UTC)

Well, I just threw something together to get an article started. I thought someone had said that Bucklin was the same as IRV, but I guess that's not the case. Captain Zyrain 19:25, 15 October 2007 (UTC)

I think this is unmerited as a separate article- in particular, it's not any sort of unified objection to IRV. It's really more of an endorsement of several (wildly different) systems, and as such should be listed as Pros for those systems, instead of cons of IRV. Specifically, objections 1 and 4 complain about the complexity of IRV, while objections 3 and 5 complain that IRV isn't complex enough, advocating Condorcet methods or range voting, both of which are more complex than IRV. Objection #2 is advocating non-instant runoff elections, missing the entire point of any ranked ballot. Paladinwannabe2 19:12, 15 October 2007 (UTC)

A "Controversies" article should describe the notable controversies that exist over its topic. It should present all notable arguments, including ones that are blatantly spurious. The format that developed for this article has potential wider application, and, if I had more time, I'd be working on it more. Essentially, instead of listing Pro and Con arguments raw, with no comment, we can organized, in connection with each argument, the elements of NPOV analysis, so that readers can form their own conclusions. It's very important that these argument discussions be inclusive. The argument being discussed is an exception to the requirement that material be NPOV; strictly speaking, the fact being reported -- and which should be sourced -- is that the argument is being made, not that it is true or accurate. Thus it should also be attributed, though if an argument is being widely made, we might simply admit it without a source in the text. The discussion, then, would present fact relevant to the argument, and, in some cases, counter-arguments (which, again, should be flagged as attributed statements, not as fact). The point is to present the information that any good and unbiased advisor would present to a client who needs to understand something. With this topic and with many others, the world is full of misleading arguments, arguments which sound good at first sight and which fall apart when examined closely. This article should *not* be a hit piece on Instant-runoff voting, nor should it be a promo piece.

Now, as to the specific points made above: if a jurisdiction is considering implementing IRV, the choice between the status quo and IRV is an artificially isolated one, if other possible reforms are excluded. Changes cost money, cause confusion, etc., so the number of changes should be limited, and a change which is simple and easy may be more desirable, other things being equal, to one which is complex and expensive to implement. So the characteristics of other systems become relevant. FairVote is quite aware of this, which is obviously why they have, of late, put a fair amount of effort into developing and promulgating arguments about other systems in an apparent effort to make them look less desirable. It's politics. You are promoting a candidate, part of the strategy is, for some, to make the other candidates look bad. As an example, look at the Bucklin voting article. Bucklin is actually a solution to the "runoff" problem, quite like IRV. I've called it "Instant Runoff Approval." That is, it is ranked Approval voting, and proceeds in rounds, just like IRV. So if a candidate gets a majority in the first round, no further counting need be done, just as with IRV. The difference is that with each round, votes are added in, not substituted, so Bucklin, in a contested election, does not suffer from Center Squeeze, it behaves more like Approval. Thus it "fails" the Majority criterion, which FairVote, on its pages about Bucklin, equates to Majority Rule. That's political polemic. Aside from problems in the definition of the Majority Criterion, if we assume that Bucklin fails, it only fails under conditions which might never arise in real elections, no examples are alleged from the fairly extensive history of Bucklin in the U.S. And then FairVote claims that Bucklin was repealed because of reasons that do not apparently come from the historical record, but which come from FairVote as arguments against Approval. The real reason Bucklin was repealed, most likely, is that it was *working*. The famous case Brown v. Smallwood shows that. FairVote reports that case, essentially extracting their own argument from it, and ignoring the substance, which, to me, would pretty clearly apply to IRV as well as Approval.

My point is that other methods *are* relevant here, as to alternatives to IRV. Similarly, Controversies articles, as they come to exist, around other methods, would have similar reference to IRV as an alternative. The key is whether or not the arguments are notable. One possibility: a page that brings together all the arguments over all the methods. That's different from the Voting system page, which does, I think have some arguments and quite possibly should not... --Abd 14:09, 15 November 2007 (UTC)

[edit] General ranting about voting systems, and the people who argue about them

Ideally, of course, there would be a simple voting system that is as good as a Condorcet method. Range voting or Approval voting may be such a system, depending on how you define 'simple' and 'good'. (Range voting, for instance, is simple for the people counting, but the most complex for the voter.) All known Condorcet methods are more complex than IRV- and if the average voter doesn't understand the underlying mathematics, how can he trust his vote?

Approximate ordering of a few systems based on 4 criteria

Simplicity for voter (debated): Plurality>(IRV\Approval)>(most other ranked ballots)>Range voting

Simplicity for vote counting: Plurality>Approval>Range voting=Borda>IRV>(most other ranked ballots)

Quality of results (very hotly debated*): (Condorcet methods\Range Voting)>(IRV\Borda\Approval)>Plurality

Fewest problems with tactical voting (also hotly debated): Approval>IRV=Condorcet>Range Voting>Borda>Plurality

If your goal is the best results, Condorcet or Range is the way to go. If your goal is good results voters can understand, Approval or IRV is the way to go. If your goal is simplest fair voting possible, Plurality is the way to go. People shouldn't be arguing about voting systems, people should be arguing about what the most important goals of voting systems are. For instance, some systems are more likely to elect moderates than others- whether this is a bug or a feature seems to depend on how much you like moderates. Paladinwannabe2 19:12, 15 October 2007 (UTC)

[edit] Reorganization of page begun

I've started to reorganize this page; unfortunately, it's a bigger job than I could do in one session. I think the plan is reasonably clear. What I've done is to take an argument and then discuss it. I'm a critic of IRV, so I may well be better at criticizing pro-IRV arguments than I am at supporting them, but I've attempted to acknowledge what is true about the arguments.

As an article about controversies, it is not necessary that arguments presented here be true; but they should be sourced. This is an example where the standards of evidence are reduced; it is sufficient, for an argument to be an argument actually being made, that ... someone is actually making it! While it is conceivable that we could accept the minimal standard that someone makes it here, that is, in my view, not adequate. I'm not making up arguments; rather, I'm attempting to present arguments that I have actually encountered being made by proponents and critics. Any argument found on the FairVote site or the Center for Range Voting site would, for example, qualify.

I only presented Pro arguments, which naturally bring up *some* con arguments, but other Con arguments are not rooted in refuting a Pro argument. For example, simulations have shown that average Bayesian Regret is relatively high with Instant Runoff Voting compared to a number of different reforms (Bayesian Regret is pretty much the inverse of, in this case, average Voter Satisfaction with the outcome.) So this would be a Con argument that would be presented as such, with, then, the discussion pointing out the controversy over Bayesian regret as a standard for measuring election method performance, the alleged bias of the researcher, etc. Another related argument would be election performance as shown by Yee diagrams, which, by the way, make pretty pictures.... When someone attempted to add a Yee diagram reference to the IRV article, it was nixed by the edit cabal that was functioning at the time on the basis that the publication was not peer-reviewed; however, there is, in fact, fairly wide acceptance among election methods students of Yee diagrams as a tool for comparing election methods. Yee diagrams would be appropriate for this article, because they *are* being used as arguments against IRV, even though they, in themselves, are merely a graphical display of mathematical results, they aren't arguments.

If we can bring this article up to speed, we can then remove the Pro and Con sections from the Instant Runoff Voting article. One of the problems with Pro and Con bullets is that misleading arguments can easily be communicated that way. I attempted to reduce the problem with warning language in the IRV article, but that was, of course, taken out.

That language was intended to avoid repeating over and over again with each claim "Proponents claim that ....", thus making the bullet points simpler and easier to understand, while still framing it all as arguments, not facts. Abd 04:39, 16 October 2007 (UTC)

[edit] "Instant runoff would end spoiler effect in elections."

http://www.fairvote.org/?page=1165&articlemode=showspecific&showarticle=831

Some editors -- in this case anonymous, it would be appreciated if editors identify themselves -- may be confused by the normal Wikipedia practice of being conservative with regard to claims. This is an article about a political controversy, not directly about pure facts (i.e., objective fact other than the fact of an argument being made by someone in some significant way). Political activists don't use weasel words unless they are attempting to put argument into Wikipedia!

We need to either discover existing policy about how to present controversy, or develop new policy. The round outlines are fairly clear to me.

In any case, this article is not about Instant Runoff Voting, but about the *controversy* over it. That controversy includes political polemic and puffery on all sides. I've never seen the argument that "Instant Runoff Voting will eat American babies," but this is sometimes the tone from some critics....

We would not include, here, arguments that are: (1) made by an isolated writer with no evidence that this writer is influential. (2) straw men set up by those who actually oppose the supposed implication of the argument.

My intention in listing Pro arguments what that these be arguments actually being made, in public, and thus verifiable even though I have not yet provided sources, by proponents of Instant Runoff Voting. There are isolated proponents who make preposterous arguments; I would not include these. But the argument here is actually being made, and not just by anyone, but by Rob Richie, Executive Director of FairVote, in a prominent newspaper.

Is the argument true? Of course not, not if we insist on the literal and absolute meaning. It will not "end" the spoiler effect, but, as I wrote, it is not controversial that it will reduce it. In other words, the *substance* of Richie's claim is relatively true, as is commonly accepted in political puffery. When we get to details, that is perhaps another matter. Abd 13:22, 16 October 2007 (UTC)

This is a rather bizarre article - Why not rename it to: "Why all argument points in favor of IRV are wrong"? Tom Ruen 02:11, 18 October 2007 (UTC)

It seems to be making some progress. Captain Zyrain 02:56, 18 October 2007 (UTC)

(unindent). I can understand why Tom might be upset by the inclusion here of a discussion of Pro arguments, which necessarily includes argument intended to refute, as well as confirmation -- but I'm a little irritated that Mr. Ruen seems to have not noticed the confirmation part! -- but I simply did not get to the other side, the Con arguments to be similarly treated. I left the original article below to facilitate the next editor's work. Sloppy, yes, but I don't think that will last long! Anyone who knows better than I how to proceed, please "Make it so!" But my goal, here, as with the other articles I'm working on, is to, where necessary, attract knowledgeable editors, including editors with various POVs, and then seek to find a consensus. If we can, the articles will be truly accomplishments worthy of recognition.

Tom's sarcastic insertion of the heading, "Reasons for support and why they're wrong" was then used as a claim in the pending Wikipedia:Articles_for_deletion/Controversies_regarding_instant-runoff_voting that the article was mere political polemic and POV. It's a bit of a sad commentary on some of those who quickly respond to RfDs that an editor, noting this heading, got it backwards! It's a common Wikipedia problem: editors -- and administrators -- who make snap judgments, and get enough of these together in one place, it's an apparent consensus, sometimes short-circuiting the process under WP:SNOW.

My thanks to Terrill Bouricius for removing the sarcastic header, and for his work on this article. Indeed, I also thank Tom Ruen for his contributions other than that sarcastic heading. I am reviewing both of these, because I know that both these editors are very involved with promoting Instant Runoff Voting, but this is in no way an objection to their participation. We need them.

If we do our work well here, the main Instant Runoff Voting article may be one step closer to featured article status; I'd think that those who favor IRV would like to see this! My own work, on the other hand, given the context, will be to attempt to ensure that the article is balanced when it gets this recognition. In the attempt to create this balance in the presence of what I see as strong imbalance in one direction, I may certainly go to far in the other direction, and thus we all need each other to restrain each other. This is how, in my view, Wikipedia articles on even highly controversial subjects can manage (and have managed) to become efficiently and enjoyably informative as well as balanced. Abd 15:59, 18 October 2007 (UTC)

[edit] Anti: "If it ain't broke don't fix it" or "Plurality Voting is Good Enough"

Some point to the fact that most elections in the U.S. use plurality voting, and voters seem to accept plurality winners as legitimate. The fact that some revered leaders, such as Abraham Lincoln did not receive a majority of the vote is sometimes mentioned. Others argue that a voting method that allows a candidate to win who is the LEAST favorite choice of a majority of voters (as is the case with plurality voting) is a disaster waiting to happen, and that majority rule should be a bedrock of our democracy. Plurality voting also increases the likelihood of "spoiler" scenarios, with all that that entails.

That's the argument my delegate used in reference to IRV. He wrote, "I'll take a look at it. But my first impression is that it may deprive voter of time for another look at the two remaining candidates. Sometimes run-offs sharpen the debate. Frankly, I think our system runs well now. Do you have an example of where it hasn't worked well in Virginia?" I would say the "if it ain't broke..." argument could also have been used to oppose the 17th Amendment to the US Constitution, since appointment of Senators by legislatures produced such legendary senators as Henry Clay, who greatly influenced Abraham Lincoln. Captain Zyrain 15:49, 18 October 2007 (UTC)

[edit] Thanks to Tbouricius and Tomruen for a very helpful series of edits.

I mention above that I might be casting a jaundiced eye at this series of edits, but it looks to me like both editors were really working well to attain NPOV statements of the arguments, I couldn't find one single problem insert or removal. In some cases, additional argument might need to be added to be fully balanced, but nothing seemed incomplete. If, however, someone who really believes the "one person, one vote" con argument shows up, we might see some more cogent argument for that. In fact, it seems to be that I've come across some argumentation on this point that was more cogent than what was presented. (Because IRV does, in some implementations, completely discount ballots containing only votes for eliminated candidates, there is a bit of loss of one-person, one-vote in that. I didn't make up this argument!) But this is fine for now!

We are indeed making progress. My wonder now is that User:Tomruen felt it necessary to complain with that sarcastic header! (I'm mainly jealous for lack of time to play. Tom Ruen 18:06, 18 October 2007 (UTC))

Now, we need sources for the arguments. In some cases, there are facts in there which we can assert without explicit attribution, because we don't disagree about them, but, it may ultimately be necessary to source them even if we agree on them. This article is going to be viewed, I'm fairly sure, with hostile eyes by some.

There was one edit that weakened a "refutation" (I'm going to change that language, it isn't really the best): the election in San Francisco to which I had referred had one winner with, as I recall, an explicit vote of well under 40%, and I just put that figure in as conservative. We can fix this when we get sources. I was just writing first; if I had to put together sources for everything I know, it would take me forever. One source at a time, for me. But others can and will find sources. I was not intending to make any controversial statements in this article, and, it seems, I mostly succeeded, because two strongly pro-IRV editors removed almost none of it, simply rewording it to improve my language in most cases. Again, thanks! Abd 16:20, 18 October 2007 (UTC)

[edit] About the one-person one-vote section

Abd, I removed your disclaimer about MPV in Ann Arbor, as it is factually incorrect. I am pasting the relevant charter language from the 1975 charter here FYI.

(9) During each transfer of votes a separate subdivision shall be made of the non-transferable votes, such non-transferable votes to be excluded from the calculation of the majority.

(12) The Board of Canvassers shall record the number of votes credited to each candidate for Mayor at the end of every count. Such record shall also include (A) the number of non-transferable votes excluded from the calculation of the majority, and (B) the number of votes necessary to form a majority to be elected Mayor.

You can argue whether they should have named it Majority Preferential Voting (as I know you reject this definition of majority), but the system was IRV as discussed in the U.S. in every respect. However, even if they DID maintain the majority threshold and not discout exhausted ballots, it would STILL be IRV as discussed in the article as used in most of Australia...As with our old RONR debate...IRV CAN and DOES exist with both of these majority definitions in use.

Tbouricius 21:33, 18 October 2007 (UTC)

First of all, my thanks to Tbouricius for correcting my error. The problem was that I did look at the source, the judicial decision as quoted on the FairVote site, and, quite like we see in Brown v. Smallwood, the decision sets up a condition and then violates it (without, by the way, clearly explaining why). I did not read far enough. It appears to me that IRV is, in fact, vulnerable to legal challenge where it redefines "majority" through disregarding exhausted ballots, if provisions of law requiring a majority vote are not supplanted by implementation of the method. The definition of "majority" in question is not mine, it is that of Robert's Rules. My point about this majority issue is that IRV behaves differently as a method and has different implications if it can elect a winner who did not receive a majority of votes cast; in spite of the Ann Arbor decision, which I'm going to strongly suspect was politically motivated, IRV does not magically create a "majority," and what was properly offensive about the Ann Arbor situation is that the court decided that it did, and thus disregarded a requirement in law that is rooted in a fundamental principle of democracy: decision by majority vote. No majority vote, no decision. We make compromises with election methods, and clearly in many places it has been decided that completing the election is more important than waiting for a majority vote, but, in standard parliamentary procedure, the default is that a majority is required for election, and this majority cannot be obtained simply by discarding ballots containing votes for the wrong candidates. If the Bylaws permit election by plurality, so be it.

But this question is important politically because IRV is sold as "guarantees a majority winner." It does not, except through redefining "majority" in a way unacceptable under parliamentary procedure. IRV can and does elect a plurality winner, because of exhausted ballots, and not only are exhausted ballots practically forced in some places due to ballot restrictions, but even where full ranking is allowed, it is offensive to force it, and not ranking a candidate is equivalent to bottom-ranking equal, i.e., voting against the candidate and all others not ranked.

This possible legal challenge to IRV does not apply, if the method is the one described in Robert's Rules, or if the law does not require a majority vote. Tbouricius, in Vermont, proposed legislation that would have no problem in this respect, as an example, for the Vermont constitution requires, with the election involved, a majority of ballots cast, or the election goes to the Assembly for choice among the top three. It is reasonable to assume that the top three would refer to the situation after all transfers have been made to reduce the set of remaining candidates to three.

As an aside to Tbouricius, it seems that, in the light of this, the ballot instructions specified in his legislation were incorrect. A vote for a second preference *can* harm a first preference, if the second preference vote results in the first preference not being included in the top three. Abd 04:19, 19 October 2007 (UTC)

[edit] Bad to less bad?

I suppose a strategy of good leaders is to set a bad example to inspire others with higher standards to clean things up. Well, I'm caught. Hopeless to do anything but near-full rewrites of two sections. At least I believe my content is sound, even if wording and inability to step back from technical details that obscures perspective means I'm also leaving lots of room for clean up. Tom Ruen 23:31, 23 October 2007 (UTC)

You can always put up some cleanup tags to alert editors of the need for cleanup. See Wikipedia:Template messages/Cleanup. StrengthOfNations 06:43, 15 November 2007 (UTC)

[edit] faulty logic

Wouldn't it be more accurate (in the Con Arguments at the end of the page) to say "changes to vote counting procedures or voting equipment could make it more expensive to count votes." The current statement "is more expensive to count than Plurality voting or Approval Voting, requiring changes to vote counting procedures or voting equipment" is actually a fallacy. The contention relies on the existence of vote counting procedures and equipment for an existing plurality or approval system. Small point, but still.maxsch 00:22, 25 October 2007 (UTC)

Agreed by me - bad writing everywhere! Also somehow it ought to make it more clear if it is "upgrade" costs vs "counting" costs. I read it to mean "Support for rank ballots may require upgrading equipment and software." Tom Ruen 00:35, 25 October 2007 (UTC)

• Failure of the summability criterion means that it becomes mathematically intractable to transmit (a representation of) the complete ballots as the number of candidates increases. Most other ranked choice voting methods meet summability and thus do not have this problem. Under IRV, there must be secure lines to every polling place in the event of a runoff or all the ballots must be transferred to a central counting place - this has been the case in countries and areas that have passed IRV. - McCart42 01:18, 15 November 2007 (UTC)

• Who is demanding summability by candidates or permutations of ranked candidate preferences? The ballots need not be physically tranported for centralized counting. Each ballot can be transmitted digitally - maximum information = #_ballots*#_election*#_rankings. (Example 10,000 ballots in a precinct, 10 elections, 6 rankings means up to 600,000 numbers to transmit.) Tom Ruen 01:46, 15 November 2007 (UTC)

• I'm not positive your calculation is correct, since there are n! possible combinations of rankings for a single slate of n candidates - even more if you don't require voters to rank every candidate, which is usually the case. But the point is that for a Condorcet-satisfying ranked-choice method, those ballots could be summed up into a set of 10 6x6 matrices; 360 numbers would be transmitted, regardless of how many people vote. And this information is easily publishable. Summability criterion was previously an article, but it has been deleted. There is nevertheless more information available at Practical implications. But this is just one facet of the issues with IRV. The visualizations here are more revealing. - McCart42 06:53, 15 November 2007 (UTC)

Also, see the Summability criterion article at another wiki. It was deleted here for some reason but as you can see there is a good deal of information pertinent to the practicality of various methods there. The issue is not whether or not it can be done - it is that transmitting large non-publishable, non-verifiable quantities of data as part of an election method should be considered insecure and a potential vector for fraud. - McCart42 07:02, 15 November 2007 (UTC)

For the Summability Criterion, see the Wikipedia:Articles_for_deletion/Favorite_betrayal_criterion page. A whole series of criteria were deleted in one fell swoop. There seems to have been practically no participation in the deletion discussion by the Voting Methods Project editors. I've seen quite a few pages deleted recently; there are some editors who scan for what they think are useless, non-notable pages, they propose deletion, and if not enough people are watching the pages, poof! the articles and all the work done on them is gone. What happens is that people come here, interested in election methods, do some work, but are not "full-time Wikipedia editors." So they don't check their Watchlists regularly. Now, we could get these pages back, any admin can view them and retrieve them. But I've already got plenty to do. Summability is important for a number of reasons. If precincts are summable, then manual counting at the precinct level will do just fine. IRV manages to be used with manual counting because of a fairly cumbersome back-and-forth process of dropping centrally followed by recounting locally and resubmission. With local counting, any possible fraud, to have a major impact, must be distributed and thus more detectable. If centralized counting is necessary, then audits by checking an individual precinct become less effective at detecting fraud. But it's still possible; there are many voting security experts, however, who consider summability to be important.

Forget the factorials and matrices. Say candidates are numbered 1=apple, 2=banana, 3=carrot,... 100=zebra, then my ballot has up to 100 numbers, value 1..100, or one byte per ranking. So one ballot can be transmitted as 100 bytes of information. If you have 10,000 voters, that's 1 million bytes of information. Having centralizedd FULL copies of ballots seems BETTER than only transmitting candidate counts apple=2654, banana=2345... since recounting means going back to the original ballots which at that point might be tampered with. However full-ballot transfer means any tampering will be detectable. Each ballot can also have ad id number, so tampering could be traced back to a single ballot - no hanging chads or late-markings will be left undetected if a hand-recount is done. And once the centralized count is done and elimination order is determined, every precinct could do its own HAND-RECOUNT to compare to the centralized counts for that precinct, so again no need to move ballots, easy to detect tampering if the central numbers have been altered. Tom Ruen 08:39, 15 November 2007 (UTC)

Remember, this is a Controversies page. That an argument is silly or misplaced is not reason to exclude it. The question is whether or not the argument is actually being advanced in the public arena (not merely here). Tom Ruen is arguing for full-ballot transmission. I've elsewhere argued for more than that, I have argued for full-ballot *imaging*. That's more data per ballot, to be sure, but not more data than can easily be handled with old technology. (A fax machine image of a ballot would be just fine, so no new equipment would be necessary. Only difference: the fax is received by a computer, and the image is either manually recognized or machine recognized, both are trivial. But at the same time, the images are *public.* In Florida, ballots are clearly public record, and may by law be photographed; in some other states the matter is murkier. But with full imaging of ballots, done immediately and *redundantly* at the point of opening the ballot box, most security issues go away. It's quite low-tech, actually. And it need not be done at public expense. Any reasonable digital camera could handle the public images, all that needs to be done is allow election observers, who already can *see* the ballots, to photograph them. But this is irrelevant here. And even though it is irrelevant, creating the full-ballot data sets as Ruen describes is not simple; it means that there is a lot more work for the election workers. In real IRV elections, as I've mentioned, with manual counting, they only count the first-preference votes, then, later, additional preferences only if needed. Note that I was just looking at some voting software results for an IRV election, and, even there, all that was reported was the minimum necessary to determine the winner. Note, as well, that IRV was passed in some jurisdictions, as is noted (triumphantly?) in the Instant-runoff voting article, "pending implementation." Why is it pending? (For three years in one case.) Let me put it this way: if IRV were summable, they wouldn't be waiting for advanced voting equipment and software. --Abd 15:19, 15 November 2007 (UTC)

• Abd nails it on the head. Yes, you can do all of these things to make IRV possible, and yes, modern communication protocols and computers make it trivial to implement. But there is still a cost associated with it which is non-trivial. IRV requires a secure, verifiable communication channel to transmit the relatively large amount of data that is completely unnecessary in any other method. Even using Tom's numbers, this data is orders of magnitude more than other methods. That much is certain. - McCart42 16:25, 15 November 2007 (UTC)

• Something to think about is recounts. Hand recounts are required by law in some states when the computed results are within a set percentage. In a hand recount of a summable method like approval voting or a Condorcet method, the poll workers take each ballot and add its contents to a sum, either +1 for each of the candidates approved by the ballot (AV), or +1 for each pairwise victory (Condorcet). The number of "sums" is determined by the classes of the summability criterion, or, if you like, can be determined for each method independently by simple math as CRGreathouse has done. Now how would a hand recount be done under IRV? This isn't a rhetorical question, I'm honestly curious. I suppose it depends on whether each precinct is allowed to do their own "runoffs" or whether this is done at the state level, but I suspect it involves a great deal more coordination (and thus expense). - McCart42 23:48, 15 November 2007 (UTC)

After thinking about this a bit more, I suppose in a hand recount under IRV you would actually have to look up the record of the ballot, requiring an ID in the system for each ballot so as to verify the ordering of that particular ballot. I don't believe this type of comparison to the exact ballot is necessary under any other method, and that fact surprises me about IRV. - McCart42 23:51, 15 November 2007 (UTC)

[edit] Pro arguments.

Earlier today, I rewrote two of the Pro argument sections, in one case to expand and in the other to replace a convoluted argument about "Condorcet" that I have never seen from pro-IRV advocates with one which is actually being promoted, the Core Support argument. An anonymous editor reverted that, and also added slightly more coherent argument that was still poorly written and still not, to me, actually intelligible. I don't have a lot of tolerance for IP editors who use reverts to insist upon the text without discussion, hence I have reverted those changes.

If any editor thinks that what I put in was irrelevant or inappropriate for the article, then it would be a courtesy to discuss the reasons here before removing material, especially sourced material (the article badly needs better sourcing). Since the Condorcet argument was poorly expressed, not sourced, and also unfamiliar -- I've been following IRV arguments for quite some time -- I don't think it is ready for this article now, and perhaps ever. But, again, if anyone disagrees, by all means, discuss it here.

However, this article is not a place to develop novel theories about election methods. It is about actual controversies, not original research, and is only unsourced at this point where arguments are familiar or very clear and the work to source them has not been done yet.

The diffs for my reverts are ^[1] and ^[2] --Abd (talk) 04:25, 25 November 2007 (UTC)

[edit] Do not remove visualizations!

http://zesty.ca/voting/sim/ has some useful visualizations demonstrating the myriad problems with instant runoff voting. These are probably the best illustration of voting system results in practice. Users have removed these links because they say they do not understand them - this is unhelpful. Please do not remove material just because it doesn't make sense to you. - McCart42 (talk) 22:01, 4 December 2007 (UTC)

Sorry, that was me. To me they were lacking in explanation, so were unhelpful. I found the text explanation (which was in the reference I added) much more helpful. But if others think that the pictures genuinely explain the problem then sure, they can stay in. Peter Ballard (talk) 23:06, 4 December 2007 (UTC)

Not wanting to be a bad guy since I think the visualizations are cool, and I appreciate all thoughtful sources myself, but Wikipedia:Verifiability#Self-published_sources_.28online_and_paper.29. Tom Ruen (talk) 23:49, 4 December 2007 (UTC)

The author of the link in question was contacted for a published review for the state of California on electronic voting machines. [2] The visualizations are notable, and the author's credibility is established by publications in the very similar field. - McCart42 (talk) 13:48, 5 December 2007 (UTC)

A common tactic of the sock puppets I've dealt with has been to assert Original Research (which is similar to "self-published") to keep inconvenient material out of articles, or to get articles deleted. It works, lots of editors buy it, and administrators too. It's an error. Original Research cannot be used as reliable source. However, if the research is *notable*, information about it can be included in articles, with proper attribution and balance. It cannot be asserted as fact *unless* it can be verified by any user. Now, it's an interesting question. Yee diagrams can be verified, but it might require some math or some effort. I do have a solution to the problem, which is a community one. The community does include plenty of people who can handle the math and the complexity, such as it is, of Yee diagrams. (They aren't *that* complicated....) For starters, if Yee diagrams are notable, as I agree they are, they can have an article of their own. That article can describe what they are, can link to them, it can even present them; it must, of course, note any controversy about them, it must be balanced. Claims of fact must be attributed unless reliable source is found. But, in fact, there isn't any real controversy about them, that I've seen. Just claims that it's original research, self-published, all that, which is essentially wikilawyering and a form of POV-pushing when it is used to repress *notable* information. I'm pretty sure that the self-published problem will disappear fairly quickly, however.

There is a technique I've used for myself to verify difficult-to-verify facts. It isn't perfect, but it might be more accurate than your average traditional encyclopedia, and it can uncover lots of obscure facts -- or verify the absence of them. In a community where there are many people with an attachment to a topic, and the means to verify facts, and someone comes in and asserts some obscure and inconvenient fact, the more unpopular the better, and *some* in this community clearly have access to resources to confirm or deny this fact, and it's considered important, and nobody comes up with source material or other actual evidence that the fact is wrong, they just attack the presenter as being deluded or badly motivated, it's right. I've never seen it fail. Takes some time, though. If Yee diagrams are faked, it would be easy to detect, and there are people with motive and the ability to detect it. Maybe even Tom Ruen. --Abd (talk) 05:18, 6 December 2007 (UTC)

[edit] Visualizations of voting methods: a proposal to add to all relevant articles

I have found multiple independent sources for the visualizations mentioned earlier, and while they are self-published they are very important to this issue. Ka-Ping Yee has participated in a published study for the state of California on electronic voting machines and has some authority on this topic - however, in this case the simulations are so basic that anyone can verify the results themselves; I would encourage both authors to publish their sources if they haven't. (EDIT: source is available - anyone can see the methods of calculation of the data)

• Brian Olson
• Ka-Ping Yee

It is time we started to include these valuable resources in the articles on voting methods so that readers can understand the real-world consequences of each system. - McCart42 (talk) 03:45, 6 December 2007 (UTC)

It it curious to me why the analysis excludes two round system or a top-two primary election, which have the same chaotic effects as IRV due to forced elimination. Is the purpose of this analysis to ban candidate elimination? If so it's a bigger fish to fry than just comparing to IRV. Tom Ruen (talk) 04:30, 6 December 2007 (UTC)

The analysis also excludes Bucklin. Clearly it's a conspiracy that goes back to that corrupt Minnesota court. Seriously, candidate elimination is a problem. Perhaps it might be worth reading Robert's Rules on the topic.... Robert's Rules strongly opposes candidate elimination, if it is possible to avoid it. Hey, the Instant-runoff voting article has some quotes about that, if the socks and FairVote editors haven't taken it out again.... (I think they have mostly agreed to leave it in, but .... it took quite a struggle to allow exact quotation from a thoroughly reliable source, and I even bought the book so I'd quote it exactly without relying upon FairVote's copy. Every description was shot down and reverted out, and still there is an effort to get it out of the article on the basis that it isn't "clear" enough, too many words. But when it was said with fewer words, as accurately as I could do it, it was "POV-pushing." Right. It dented their POV, unfortunately, and if I am working for The Truth (TM), anything which offends my POV, er, The Truth (TM), must be POV with fanatic, partisan intent. WP:ABF On alternate days I think that some of these editors are "getting it," but, then, something else comes up.

--Abd (talk) 05:31, 6 December 2007 (UTC)

• You can't include every method ever proposed; any comparison is going to be by its nature limited in some way. However, if you would like to generate the plot for two round system or a top-two primary election, feel free - the source is available and you can publish your results when you've simulated. These graphs need to be added to these articles to demonstrate what lack of monotonicity actually means so people understand how bad IRV is in practice. - McCart42 (talk) 05:55, 6 December 2007 (UTC)

The decision of whether or not to include links to visualization sources is an editorial one. These visualizations aren't being used to prove monotonicity failure, they are being used to make it easier to understand. As the text in the article says, there are multiple independent sources for these, I'm aware of another, some of the people involved with the Center for Range Voting have done them. They are directly verifiable, if someone wishes to go to the trouble. There is now secondary source regarding them and perhaps they should have their own article now. --Abd (talk) 14:24, 21 May 2008 (UTC)

[edit] wiki.electorama.com/wiki

• It seems rather strange to link to [[3]] which was BASED off an older version of Wikipedia's IRV article! Tom Ruen (talk) 20:57, 12 December 2007 (UTC)

[edit] IRV critics in Minnesota

For juicey senseless quotes against IRV, great website of opinions and disinformation to quote! Yay! http://www.mnvoters.org/IRV.htm Of course the Minnesota Voters Alliance may yet only exist as a single person, but that's democracy right? No wait, there's at least five people, two with PhDs! http://www.mnvoters.org/About%20Us.htm Seriously, I do agree with their position that eliminating the primary is a bad idea, that reducing the field is an important quality of primaries. [Tom Ruen wrote this, comment added by Abd]

Yes. Some of the arguments there are, shall we say, weak. However, they are worth looking at.

Part of the problem is that FairVote has made a lot of arguments available that are just as misleading. There definitely is a problem in Minnesota for any kind of alternative vote, contrary to the propaganda. While FairVote is contending that Brown v. Smallwood only applies to Bucklin because of the Approval-like aspect of Bucklin, that analysis seems based on only a small part of the decision. It appears that BvS is negating any kind of alternative vote, and there has been recent legal opinion to that effect.

What will happen when IRV hits the fan in Minnesota? It's not clear to me. The just outcome would be that Brown v. Smallwood is reversed, period, which would open the door for IRV *and* Approval or Bucklin in Minnesota. The reasoning in BvS is clearly defective and was contrary to prevailing legal opinion at the time, I don't think there is any particular peculiarity in the Minnesota Constitution that prohibits either method. The argument of "impairment" is phony from start to finish. However, there are two other possibilities:

(1) Brown v. Smallwood is confirmed. Tough luck, Minneapolis. I'm hoping that voting reform activists, including those opposed to IRV, will help prevent this outcome. While my opinion is that IRV is defective in many ways, it is not more defective than what we have as a standard, and it is an improvement in some ways, albeit an expensive one. The decision should be up to the voters, and IRV does not violate one-person, one-vote, not in any clear way that should make it unconstitutional. (There are, in some implementations, problems with majority failure, but that's another issue, and plurality is worse in this respect.)

(2)Brown v. Smallwood is overturned, but only with respect to the alternative vote with sequential elimination, approval-style simultaneous comparison of votes for multiple candidates remains illegal. Terrible outcome, and likewise not supported by legal opinion anywhere. Except in opinions issued by FairVote activists.

--Abd (talk) 19:56, 24 December 2007 (UTC)

[edit] Not an ad for approval voting

I deleted references to approval voting and condorcet methods which are not used where IRV is being considered. Then I made a simple change to correct punctuation. Both edits were reverted, with a request for discussion on the punctuation change. The substantive reversion's edit summary was, "Not an ad for IRV or Plurality, either." Indeed, and my edit added absolutely nothing about either one of those systems, which says to me that the edit was not even read before being reverted.

Here is the discussion: Using Wikipedia for advocacy of your favorite method which is not actually used for any political elections is called "POV-pushing" and is against the rules. If you can't bother to read what you are reverting, then you might consider whether you are emotionally involved as described in WP:COI. I am reverting the reversions. MilesAgain (talk) 09:59, 26 December 2007 (UTC)

This is old stew. First of all, these are articles about voting methods. Voting methods are not only of political application, so the "used for political elections" is utterly irrelevant. Approval is used for elections, both currently and in the past. Secondly, controversy over IRV includes consideration of alternative reforms. It is part of a political agenda (one actually being pursued politically at this time) to assert that the only relevant comparison is between IRV and Plurality, when, in fact, there *is* ongoing consideration of other reforms. I've previously, here or in the IRV article/Talk, shown sources for this, but, without giving the URL here, it's tedious to keep repeating this, the Colorado Voter Choice Task Force, created to advise the Colorado legislature on voting reform, is actively considering multiple methods, including Approval, and they have received testimony in person from Rob Richie of FairVote, Warren Smith of the Center for Range Voting, and an Approval advocate whose name I don't recall.

It is, indeed, "POV pushing" to claim that NPOV reporting of comparison of IRV with alternative reforms being considered or proposed in lieu of IRV is "POV pushing," particularly unseemly for a sock, and it is, here, threatening edit warring, even more unseemly. If Miles were not an acknowledged sock, I might use dispute resolution, but he is welcome to propose and initiate such. And, yes, I read what I was reverting. The punctuation change was to the same section as contained the approval deletion, so the first reversion (punctuation) was necessary to enable the substantial reversion (restoring extensive language removed by MilesAgain without discussion), otherwise reversion was not possible. Assuming good faith, I will not claim that MilesAgain made the punctuation change to make reversion of his substantial edit more difficult, but that he complains about the reversion of what became a moot punctuation change does worry me. He's got a laundry list of objections, here. [[WP:COI} does not apply to me. Having a POV does not establish COI, and if I've made any specific edits that are POV-pushing, as distinct from establishing or restoring balance or introducing neutral commentary or sourced material, by all means, object specifically. Editing many sections of this article based on an invented personal standard, no. Don't be surprised if such is reverted. Make edits one at a time, showing that the material removed is POV or POV pushing, we can consider that.

As to the claim that his edit said nothing about those systems (IRV and Plurality), he is correct. The article, however, contains claims about IRV in comparison with Plurality. Is the article an "ad" for Plurality? It's an advertising trick to show a "neutral" comparison of a product with an inferior one. But in a neutral discussion or explanation, what would really take place is a comparison of the product with all reasonable alternatives, not just with a selected one. I've again reverted the change.

--Abd (talk) 22:00, 26 December 2007 (UTC)

I certainly don't want to get much involved here, but could note at least that the Minnesota League of Women Voters created a 2004 study guide for its members to compare various voting methods proposed for political elections (PDF [4]) - including plurality, Instant runoff voting, Borda count, Approval and Condorcet method. A valiant attempt at a neutral presentation, even as three of the methods fail the one person, one vote political standard. The presentation could be considered skewed towards IRV (with input from FairVoteMN), and since it was the only political reform under serious discussion and helped promote interest in the study. I don't have a quick link of results, except that member feedback from the meetings encouraged the LWVMN to endorse IRV. The study guide does have some quote from politicians, republicans specifically, who are content by their advantage within Plurality and advocate no change at all, including being against the principle of 50% majority rule as unimportant. As to IRV, I've read opinions it is slanted in favor of IRV, but overall I was very impressed by it - I couldn't improve much as a comprehensive document introducing a vast topic. Tom Ruen (talk) 22:15, 26 December 2007 (UTC)

Yes. I am, of course, familiar with the document. It's certainly not complete or accurate in every respect, but a good introduction. Might be useful to examine here where it's got problems. A *lot* of discussion of IRV and other methods has been skewed by analysis presented by FairVote, which is, after all, an advocacy organization, not even some kind of think tank, and with no public review process. (Fair enough, by the way, just not unbiased.) An example would be, look up Bucklin voting and try to find information not presented and framed by FairVote. More exists now than used to exist, because of my research writing, but it's still inadequate. What, actually, was behind the elimination of Bucklin? The FairVote argument actually makes no sense. (One can expect in an Approval-like method that most voters in a two party system and in some other applications will not add additional Approvals, they will bullet vote. But it is the few percent that kills the spoiler effect. What does make sense is what is apparent in Brown v. Smallwood. Political bias. (Center for Range Voting functions more like a think tank, with some attempt at objective analysis, though a lot of advocacy and opinion does get mixed in. That's unfortunate, in my view. I did attempt to create a review process for the Range Voting community, but ... there was not adequate support at the time, the vote was actually evenly balanced, but with, in my opinion, too small participation, which is why I abandoned the effort. --Abd (talk) 22:51, 26 December 2007 (UTC)

I've accepted the opinion of old EM hacker, Donald Davison, said Bucklin voting was the best of the approval methods (since it can identify an initial majority opinion at least). I wish the LVW report would have considered it. I have nothing against it personally - better than Borda or Approval - voters just have to know they can truncated preferences to help a favorite and potentially lose influence, and it becomes plurality if everyone does it. NOW politically, I have still ZERO interest in the lets-get-more-votes movement. I'm sure FairVote puts-down Bucklin primarily to sepatate IRV from it, which makes total sense to me. Why advocate any respect for a poorer brother who is already muddied? Tom Ruen (talk) 05:02, 28 December 2007 (UTC)

[edit] Request for third opinion

Does anyone agree with Abd that approval, a system not actually used for political elections, should be mentioned by name six times in this article (per this revert by Abd) which is ostensibly supposed to be about IRV? MilesAgain (talk) 06:24, 28 December 2007 (UTC)

Ummmm... well, better here than Instant runoff voting. I don't really care to fight its inclusion - this entire article is largely undefendable in my mind. I agree it is confusing to read controversies points unless you know the point of view of the opinion. It gets messy to have plurality advocates seemingly on uncareful reading AGREEING with Condorcet and Approval advocates on things like lack of monotonicity and such. Ultimately the ENTIRE debate is at best newspaper published opinions, no better or worse in my mind than website opinions by EM hacks. So as far as I will say, LET ALL OPINIONS FLY, just keep track of WHOSE opinions they are, categorically Approval advocates or whatever. Tom Ruen (talk) 06:38, 28 December 2007 (UTC)

I agree with MilesAgain...Abd is creating the odd impression that the major opposition to IRV comes from advocates of Approval, which is simply not the case. If some editor were an advocate of monarchy and kept inserting arguments against IRV from a perspecitve that was in favor of famlily-based autocracy and repeatedly named monarchy as a better system of government, simply because that editor liked it, I would object to that as well. This article needs to reflect the nature of the public debate that actually exists, not the one in Abd's mind. Criticisms from the perspective of Approval should be included, but should be dwarfed by the perspective of plurality and two-round runoffs.

Tbouricius (talk) 21:20, 29 December 2007 (UTC)

The problem with that comparison is that you're setting up IRV to win, and this has been brought up several times on this discussion page alone. IRV is better than plurality in almost every measure but monotonicity, and clearly less expensive than two-round runoffs while preserving all of its benefits. It's very much a rigged comparison, whether or not it is the most common one made. I think it's important to have comparisons to several other notable voting systems, starting with approval voting and a Condorcet method or two. See Schulze method for an example of a good (if terse) comparison to the merits of other methods. After all, this article is not Instant-runoff voting but an exploration of the controversies. Why sell the article short by limiting the debate to those methods IRV beats hands-down? - McCart42 (talk) 07:27, 30 December 2007 (UTC)

It is funny to me to see the two-sided attacks on IRV. The plurality/runoff (no-change) supporters so often falsely claim it unfairly rewards third parties in the way that Condorcet actually does, while (single-winner) IRV is fact a minimalist reform of very little consequence, asking for rankings and barely making proper use of them. It is difficult to me to see IRV as a feel-good reform of little consequence attacked by EVERY SIDE - being unwanted by some for doing too little and unwanted by others for fear of it doing too much. It just seems to me to make for a muddled article trying to express ALL-SIDES. As a person USED to Plurality/two-round primaries, first hearing about IRV, I want to know what people are saying as a reform FROM what we have TO IRV. I guess I must agree with Tbouricius, the controversies of IRV of importance are those related to CHANGE, not alternatives that offer greater reform that will be resisted even more. I suppose we need two articles IRV controversies from plurality supporters, and IRV controversies from Election Method Theorists. I seriously can't unify these arguments in a way that I think is informative to readers trying to see what the debate is about. (Of course smarter people than me can do wonders I can't imagine, so again, I'm more frustrated than opposing arguments.) Tom Ruen (talk) 04:14, 31 December 2007 (UTC)

(unindent)IRV is indeed a minimalist reform, probably, *in effect*, but it's sold as more than that, yet what it probably does accomplish can be done with far less cost and arguable better performance. Articles, to be NPOV, should indeed express all notable sides. Not all sides. Ruen seems to have utterly misunderstood what this article is. It is not an anti-IRV article. It is not a POV fork, and POV fork is exactly what he has, perhaps sarcastically, suggested. Rather, there *are* controversies over IRV, that is, controversial claims advanced by IRV supporters and controversial claims advanced by critics. This article examines those, and, for some level of organization, it examines *arguments,* categorizing them into Pro and Con arguments, according to whatever side advances the argument most prominently. Essentially, this article is exploring what people say about IRV, including experts. We will be seeing more here from the literature, of which there is a lot, I've been reading quite a bit on monotonicity failure, which many experts consider the most serious problem with IRV. I don't. I consider the complexity, the expense, and center squeeze to be the biggies, and the monopolization of the election method reform movement.

In any case, what the plan of this article has come to be, so far, is to take an argument, which is necessarily POV, and examine it in context with fact. Properly, the facts asserted must be sourced according to WP:RS which can be tricky, but some of that can be avoided by keeping with attributed argument, which is inferior. This has been misunderstood by a few editors coming across this article. What are these "opinions," clearly POV, doing on Wikipedia? Well, real people really have those opinions, and it's verifiable that they do. The opinion is the fact, not the content. Then we look at the relationship between the opinion, contrary opinions, and what we can reliably state as fact.

My own habit is not to challenge statements from other editors on the basis that they are not reliably sourced *unless* I think that they are false or inaccurate, particularly if the error leads to spin. Some editors, particularly POV-pushers, will insist upon reliable source for matters that are common knowledge among those in a field; theoretically one should be able to find it, but sometimes it's not so easy. And, writing in articles, I may write from what I believe can be shown from sources, but don't necessarily put in the sources immediately; it's a lot of work! Lots and lots of articles start that way, someone knows something about something and just writes about it; it's a stub, and these stubs can become good articles with work.

It can get pretty complicated, indeed. I know that's frustrating to Tom, who would like it all to be nice and simple, but I just wrote a report on the San Francisco 2000 elections and put it in Talk for Instant-runoff voting. I've never seen the issue addressed that the report brought up: Top-two runoff (in the event of majority failure) in San Francisco, apparently, combined with nonpartisan Supervisor elections, was causing the field of candidates to expand drastically, to the point where one election had seventeen candidates on the ballot, and most had many. Now, looking at these elections, and realizing that the use of top-two or IRV will cause voters to vote more sincerely (that's a good thing in one way, but if they avoid making the necessary compromises, it can be a problem), I see that Plurality probably would have been producing, in all or nearly all cases, the same results. In other words, SF could have avoided the runoff problem by simply moving to Plurality. However, they wanted to preserve the tradition of requiring a majority for victory. So they implemented IRV, *and are getting elections by plurality.* What they implemented was Ranked Choice Voting and I think there are three ranks. With seventeen candidates, what does that do? Yes, I imagine you can guess. And that is what it is doing. --Abd (talk) 06:43, 31 December 2007 (UTC)

[edit] POV and disputed-section tags

I agree with Tom Ruen above that this entire article is indefensible. It's a Wikipedia:POV fork, to begin with. For example, I just looked at the "Con: IRV violates the one person one vote mandate" section, and saw that in fact there were no sources at all saying that anyone has actually said in a reliable source that IRV violates one-person-one-vote. All that is there is a mere analogy to the court case involving Bucklin voting. Wikipedia does not allow for that kind of original research.

Accordingly, I have added a {{POV}} tag to the top of the article, and a {{disputed-section}} tag to the "Con: IRV violates the one person one vote mandate" section. MilesAgain (talk) 06:58, 28 December 2007 (UTC)

Well, I wrote that section, to be sure. It's true that, as it is currently sourced, it would seem to be original research, except for the fact that there is official legal opinion on record, reliable source, to the effect that IRV does violate the precedent established by Brown v. Smallwood. I think it might be sourced currently in the IRV article or in Talk for that. I should get that source, but, hey, if you want to be helpful, you could find it yourself. Google "Brown v. Smallwood" and you will find an opinion issued for Minneapolis when it was considering IRV. Minneapolis decided to ignore the opinion.

By the way, I think they were right to set it aside *if* they were willing to spend the legal costs to fight it. I think that Brown v. Smallwood should be set aside, it was a poor decision from the beginning.

--Abd (talk) 05:05, 29 December 2007 (UTC)

Okay, do you mean this? It says, "the Minnesota Supreme Court found preferential voting (a voting method similar to IVR and Borda Count) unconstitutional in the case of Brown v. Smallwood, 153 N.W. 953 (Minn. 1915)." But isn't the "similar" preferential system involved Bucklin voting? Similar in ballot marking, but not much else, and not equivalent to STV, which makes a big difference, wouldn't you agree? MilesAgain (talk) 05:14, 29 December 2007 (UTC)

I mean that source, yes, but I do suggest reading it in its entirety. The attorney concludes "the City of Minneapolis appears to be precluded from adopting a preferential voting system generally unless such a system is provided for by the Minnesota Constitution pursuant to Brown v. Smallwood, 153 N.W. 953, 957 (Minn. 1915)." I think he's right. If Brown v. Smallwood stands, IRV in Minneapolis is dead, at least according to the reasoning in the decision itself. There seem to be quite a few people who haven't read this case but are quite ready to sit in certainty about it. The Court in Brown v. Smallwood appears to have concluded that any kind of alternative vote is unconstitutional in Minnesota. The reasoning is, actually, perverse, so trying to understand it logically is a bit of a problem.--Abd (talk) 06:19, 31 December 2007 (UTC)

Well I did read the whole thing and here's the excerpt from Brown v. Smallwood that it says the concerns are based on:

The preferential system directly diminishes the right of an elector to give an effective vote for the candidate of his choice. If he votes for him once, his power to help him is exhausted. If he votes for other candidates he may harm his [first] choice, but cannot help him.

As is so often pointed out, since IRV violates monotonicity, voting for other candidates can help one's first choice, so "preferential voting" in the ruling can not be IRV. Is there any evidence that the ruling does not use "preferential" merely to mean Bucklin? MilesAgain (talk) 08:34, 31 December 2007 (UTC)

(unindent) First of all, read Brown v. Smallwood if you want to know what was in the decision. This attorney, I'd guess, did that, and did not report every aspect of Smallwood relevant to his determination. Secondly, when you quote a source, and a passage is brief, don't excerpt the point that supports what you are claiming and neglect the rest. Here is the rest of that quotation from Smallwood.

The preferential system directly diminishes the right of an elector to give an effective vote for the candidate of his choice. If he votes for him once, his power to help him is exhausted. If he votes for other candidates he may harm his [first] choice, but cannot help him. Another elector may vote for three candidates opposed to him. The mathematical possibilities of the application of the system to different situations are infinite.

That later-no-harm quality that is alleged for IRV is actually false, *if* a majority is required for election, and full ranking is not mandatory, or write-ins are allowed (write-ins will not be ranked by other voters, generally).* That point will come out, I suspect, in any deep consideration of this issue. However, that language is from the original decision in Smallwood, which went on with other arguments, and the reconsideration decision was even more specific; the Court was, by my non-lawyer analysis, denying the constitutionality of any preferential voting system where the voter expresses more than one vote, alternative or not. What MilesAgain is doing, and continuing to do after it's been pointed out, is to *interpret* the source. That source, the PDF cited, is an expert, retained to issue precisely that opinion, and his conclusion was the opposite of what MilesAgain asserts. And, yes, there is legal opinion on the other side, *but issued by attorneys with a conflict of interest, advocates for IRV.* So, for starters, as to "evidence," there is reliable source that it *does* apply to IRV as well as Bucklin, and no legal authority in Minnesota nor reliable source to the contrary. Does it *actually* apply to IRV? That's *controversial*, and that should be obvious. The point has been that this is an issue over which there is disagreement. --Abd (talk) 22:55, 31 December 2007 (UTC)

The quoted paragraph seems to me to clearly apply to Bucklin voting, although equally applies to Bloc voting too (while it has never been argued as unconstitutional to my knowledge), but NOT IRV by any understanding by me. I really don't follow what you're saying at all. Tom Ruen (talk) 00:42, 1 January 2008 (UTC)

In my opinion, the following sentences in Brown vs. Smallwood are interesting: "It was never thought that with four candidates one elector could vote for the candidate of his choice, and another elector could vote for three candidates against him" (page 498). "We do right in upholding the right of the citizen to cast a vote for the candidate of his choice unimpaired by second or additional choice votes cast by other voters" (page 508). I interpret the last sentence as saying that the winner of an election must always be the candidate with a plurality of first choice votes. Also J. Hallam interprets this sentence in this manner. He writes in his minority report: "None of these men had a majority or even a plurality of first choice votes. If the majority opinion in this case is right none of them had a semblance of a right to the office" (page 503). Markus Schulze 22:28, 9 January 2008 (UTC)

In Talk:Bucklin voting#Unconstitutional Why?, I post a more detailed analysis of Brown v. Smallwood, which includes links to the actual decision. The page 508 quote above is telling. The decision is ruling out any sort of alternative vote, explicitly. FairVote has been, for some years, claiming that Brown v. Smallwood was only about the alleged "problem" with Bucklin that your second choice votes could hurt your first choice, i.e., that it would not apply to IRV because IRV allegedly does not have this characteristic. However, it does, *if* a true majority -- i.e., a majority of valid ballots cast -- is required for election. But that was not really the major concern of the court in Brown v. Smallwood. It's worth reading the whole decision, though it can certainly be frustrating for one who knows election method theory. The court, at one point, cites another case approvingly: "Our system of government is based upon the doctrine that the majority rules. This does not mean a majority of marks, but a majority of persons possessing the necessary qualifications, and the number of such persons is ascertained by the means of an election." Then, remarkably, the court proceeds, immediately, to consider the number of marks rather than the expressed position of the voters. Bucklin, like other preferential voting systems, attempts to assemble a majority supporting a particular candidate. It simply does it in a different manner than IRV. The dissenting judge appears to have had his head on straight, but he was outvoted. I'm sure the politics of that court would make an interesting story.

Tom, the decision in Brown v. Smallwood is quite clear, if you read the whole decision and not the little excerpt that is some commonly reproduced by [FairVote]]. You've been had. Indeed, that quote makes it seem Brown v. Smallwood is about later votes harming earlier ones. That comment *might* have formed some basis for a decision, but it is not what the court followed, it might as well be dicta. Note the concern in what is quoted above about additional votes from *other* voters, and this language is in the very brief justification of its decision issued by the court in response to a request for reconsideration. Brown v. Smallwood has always been considered to prohibit "preferential voting," and that conclusion is correct; the conclusion of FairVote-affiliated attorneys that it does not apply to IRV is quite innovative. It is the very use of alternate votes that the court found offensive, explicitly. As I noted yesterday, a Michigan court cited Brown v. Smallwood to outlaw "the Hare system" of proportional representation. Again, that is an explicit legal conclusion that Brown v. Smallwood would have applied to IRV (the "Hare system" used for single-winner elections) as well as to Bucklin.

Tom, you have elsewhere expressed the opinion that Bucklin is an interesting system. You're right. It's simple, it's easy to implement, and it is really a staged Approval election, allowing the one thing that people commonly think is missing when considering Approval: the ability to specify a Favorite. Markus, I'm not pretending that Bucklin is an ideal method, merely that it's quite good for being so simple. I'd make it allowable to vote for more than one in all the ranks. It might be enough to have only two ranks. In some jurisdictions, there might be complications from ballots not indicating a favorite, so, in those places, it would be more standard Bucklin: the first rank would not allow overvotes. But I wouldn't toss those ballots, they would merely be blank votes for any issue being determined by current single-vote plurality votes, other than the actual determination of winner. (Like future ballot rights for the party, public campaign funding, etc.)

Of course, the all-time-winner of the greatest-bang-for-the-buck terminally-simplest electoral reform has to be Count All the Votes. I.e., Approval (using a standard Plurality ballot). What ever made us think we'd do better by *not* counting all the votes? --Abd (talk) 02:55, 10 January 2008 (UTC)

I don't see any reason to maintain the POV tag on the section of the article about one-person, one-vote. User:MilesAgain is no longer with us; as I had suspected, he was a sock of James Salsman, aka User:Nrcprm2026, long involved in disruptive or POV-pushing editing of the Instant-runoff voting article (as User:BenB4 and User:Acct4 and quite possibly others. (Salsman has been an IRV activist in California.) However, before taking it out, I'm asking the opinion of other editors. The objection above of MilesAgain was spurious, there is official legal opinion that IRV specifically is contrary to the precedent of Brown v. Smallwood, and Brown v. Smallwood can be cited directly as to alternative votes. If there is a POV problem with the argument, it can be fixed. As to someone advancing the argument, MilesAgain's demand for "reliable source" is not necessarily appropriate for a Controversies article; rather, a source should be sufficiently notable and the text should show attribution. These are arguments being described, not underlying facts. In this article, we also provide relevant underlying facts, and those require reliable source as usual. ==Abd (talk) 04:21, 5 February 2008 (UTC)

I'm not capable (patience) enough to follow extended talk page discussions. Email me if you'd like a response from be. I'll be glad to take up the cause to keep the POV tag out of stubbornness. I don't have the time or patience to support improvement or reject the current contents. Tom Ruen (talk) 05:31, 5 February 2008 (UTC)

The section LOOKS like original research to me, an argument made by you (a wiki writer) rather than by a source somewhere. I'm not a lawyer. You're not a lawyer. I'm tired by the whole thing. Quote lawyers and leave out the commentary, and maybe it won't need the tag. Tom Ruen (talk) 05:37, 5 February 2008 (UTC)

[edit] taking out the disputed tag from this section

I'm not claimng that the section is perfect, for sure, it still contains some disputable statements that are not explicit from the sources, but I also wouldn't object to those being taken out. If someone takes something out and I think it should be there, it's up to me to provide sourcing. Meanwhile, for some reason, though we did discuss sources above, they didn't make it into the article. I've fixed that, and found another, for this section. And there is more that can be found. In particular, some of the Minnesota sources go into other arguments about IRV, including the monotonicity failure, etc. I took out the tag, which was originally placed by a sock puppet of a banned user, User:MilesAgain. --Abd (talk) 23:32, 20 May 2008 (UTC)

[edit] Wattles v. Upjohn (1920)

[5] links to an article in the Columbia Law Review (1921) that refers to the application of Brown v. Smallwood to the "Hare system". I.e., STV, the method used in IRV. This is a quote:

In Wattles v. Upjohn (1920) 179 N. W. 335, the Supreme Court of Michigan recently decided that a clause in the charter of the city of Kalamazoo providing for the Hare system of minority or proportional representation in the election of members of its governing body was unconstitutional. This conclusion was reached by an adherence to the reasoning announced in the leading cases of Maynard v. Board of Canvassers and Brown v. Smallwood, sanctioning a very questionable use of judicial power and serving to perpetuate by purely formalistic reasoning the evils of the existing method of popular representation.

I happen to agree with the opinion of the author of this article, Brown v. Smallwood was indeed a questionable use of judicial power, for a number of reasons. But the relevant point here is that, apparently, Wattles v. Upjohn explicitly applied the Brown v. Smallwood precedent (though not binding in Michigan) to the Hare system, i.e., STV. This is, in my opinion, correct. That is, Brown v. Smallwood was about *any* system of alternative or preferential voting. However, I'm looking for more information about that case. --Abd (talk) 03:15, 10 January 2008 (UTC)

I found another reference to Wattles v. Upjohn, conveniently provided by FairVote. Preferential voting ("Ware system") was used in Ann Arbor, Michigan, for a short time until rescinded by public initiative. It was challenged, and confirmed as lawful by the Michigan Supreme Court.

Plaintiff cited Wattles Ex Rel Johnson v Upjohn, 211 Mich 514, 179 NW 335 as authority for its claim that Preferential Voting is unconstitutional. While Wattles was decided under the 1908 Michigan Constitution, the crux of the matter is that the facts in the present case are clearly distinguishable from Wattles. In Wattles, the Court was dealing with a multiple office situation involving proportional representation. The Preferential System employed was the "Hare" System, which is clearly different from the "Ware" or "M.P.V." System used in Ann Arbor.

Which is a bit odd, in fact, since the Ware system is the Hare system, only applied to single-winner. (When multiple winners are being elected through the Hare system Single transferable vote, the election of the last winner is essentially an Instant-runoff voting election using the remaining votes.) If the Hare system is unconstitutional, one would expect, most likely, that the Ware system would be also. Go figure. What I suspect here is that this court preferred to allow preferential voting, single winner, and simply did not want to re-examine the question of proportional representation. I have some suspicion that the plaintiff was astonished by this decision, he probably thought it was open and shut. Democrats on this Supreme Court? Note that I *do* agree that these methods are basically constitutional. Unfortunately, with these courts, it can sometimes depend on whose Gore is being axed.--Abd (talk) 03:42, 10 January 2008 (UTC)

[edit] Pro: IRV is easy to understand

I removed this incompehensible POV trash-talked section. Tom Ruen (talk) 20:15, 5 February 2008 (UTC)

Proponents argue that IRV is similar to a runoff, and therefore easy to understand. A Condorcet method may produce better results, but proponents of IRV feel that Condorcet methods add too much complexity to what should be a simple and transparent voting process. However, "easy to understand" may be deceptive; that is, an easy understanding may be incorrect. For example, IRV is not monotonic, which means that raising the rank of a candidate can cause that candidate to lose. While proponents of IRV argue that such situations would be rare, it is not easy to understand. Similarly, tactical voting in Condorcet methods is not easy to understand without study.

[edit] Segregated sections

Currently the article is just a couple pro and con lists. Wikipedia policy states that this sort of split should be avoided. For example, there can be a spoiler effect section that addresses all the arguments in a neutral manner. --Explodicle (talk) 17:42, 13 May 2008 (UTC)

It might seem so. Actually, "fixing the spoiler effect" is a pro argument, and what is currently called a "spoiler effect" in the con section is actually what's known to voting systems experts as "center squeeze." I've sometimes called it a spoiler effect on steroids, but, in fact, the term "spoiler affect" is really inappropriate for it. So what is two separate pieces now would still be two separate pieces in a better-written article. The article is 'not' just a "couple of lists." What is ignored by this comment is that, yes, arguments are listed, but then each argument is examined in detail. These examinations should be NPOV, with proper sourcing, normal reliable source. Categorization by argument is not the only way to approach this, but, in fact, it is simpler and clearer than what seems to be implied by Explodicle. The arguments presented in the article are really being made, and that should be established with reliable attribution. The difficulty is in the grey area of what is and what is not notable argument, but, as a bottom line, everything in the article should be clearly verifiable. Note that to state that there is an argument that "IRV fixes the spoiler effect" is verified by verifiable attribution, which does *not* establish or show that IRV actually fixes the spoiler effect. Often attribution, for well-known arguments, has been skipped, which is why we all agree, I think, that the article needs work! (Plus POV material has crept into the article. The overall plan of the article is not POV.) --Abd (talk) 03:38, 14 May 2008 (UTC)

Facts about IRV will typically belong in the IRV article. This article is about controversies, i.e., about the arguments that are being made; these arguments are then placed in juxtaposition with sourced fact or attributed opinion. If an argument is being made, *who* is making it. If an argument states an alleged fact, what NPOV examination of the fact can we include?

If there are any sections presently in the article that are POV, that should be fixed. Each section, each argument presented, should be presented in an NPOV fashion. The "For" section is *not* a section arguing for IRV. Rather it is a list of arguments presented for IRV, out there (not made up for this article), attributed and sourced -- if not, they should be -- together with relevant facts, also sourced and balanced.

Arguments, as arguments, can generally be classified into two categories: For and Against. If it isn't one of those, it is more likely to be, not an argument, but a neutral view and may belong in the main article or in the section on related argument(s) (as what we may say about the argument.)

Note that there may be an argument made on both sides. An example from the IRV article, and I forget whether or not it's here, is "IRV favors a two-party system." This argument is made by both sides, and both sides include some people who consider it true and some consider it false. First of all, who is making the argument. If there is notable argument, it will be attributed and sourced. This might, as an example, fall into some of the exceptions, a notable blog may be a reliable source for an argument made by a notable author. The exact boundaries are difficult to specify, which is why I'd rather see the decisions made by a consensus of editors, always keeping in mind the goal: a balanced and informative article, with every fact in the article verifiable. (When it is said that So-and-so claims that "blah, blah," the verifiable fact is that So-and-so said it, not that "blah blah" is the truth.). Finding balance and finding classical reliable source in this can be, sometimes, contradictory goals, but both are necessary for NPOV.

This article has suffered from lack of attention for some time, and, as with many articles, parts of it were written by editors who knew the subject and who wrote what they knew well without providing reliable source. Many of the arguments on the page are well-known. Anything which is simply original research should be removed. Sometimes, however, original research can be verified. And the topic is sufficiently complex that I don't expect that an editor who does not take the time to become familiar with the issues is likely to get it right based on simply applying rules. Usually, though, with sufficient effort and patience, it's possible to satisfy all sides and goals, except for those which are coming from someone with a true axe to grind.

Note, as well, that true sock puppets have been involved with the IRV articles, or at least users who have been blocked and were blocked at the time of editing; James Salsman socks were heavily involved with the IRV article: User:BenB4, and User:Acct4. As far as this article is concerned, the first AfD for this Controversies article was from User:Yellowbeard, who may or may not be a sock, but who was a clearly experienced editor who created an SPA used for AfDing articles on topics related to Instant runoff voting, typically articles on other methods or, as an example, Bayesian regret, which was clearly notable with academic reliable source existing. Because of low participation, many of these AfDs succeeded. Then another James Salsman sock was involved with this article: User:MilesAgain. I knew that MilesAgain was a sock, it was obvious, and suspected James Salsman as a matter of course, but took no action against him beyond warning him regarding edit warring, and he never pushed it. He was busted based on an anonymous tip to Thatcher, I think it was, leading to checkuser, and I have no idea who was behind that.

The implication of the current AfD nominator that the creator of the article was a sock puppet is false. That user registered in 2005, and retired in 2007, and was never charged with sock puppetry except in error, once, and the associated block was immediately retracted. He later came back, and used several accounts serially, spiking one and never using it again when the next was created. He's been accused of sock puppetry, often, later, but the only "sock" issue for him, in fact, has been the creation of new accounts for block evasion; he was not blocked for any issue related to this article and, in fact, he worked in close cooperation with User:Newyorkbrad on parliamentary procedure articles, and his work was greatly appreciated.

Anyway, this is all background. The article was created as a proposed solution, accepted by consensus, to reduce edit warring on the main article, Instant-runoff voting. The creators had no intention of it becoming a POV fork, but rather a way of exploring the controversies in detail, a detail not appropriate for the main article; too often, an attempt to reduce a controversy to the bare essentials, particularly with arguments that have been designed as sound bites, can make it very difficult to find NPOV. Those who have helped to create and improve this article have included both supporters and critics of the use of instant-runoff voting in elections, and the sentiment for deleting it as a POV fork does not seem to be coming from any of the editors of the main article. There have been several forks created, most of them quite appropriate, though some of them doubtless could be better merged.

There may be, and probably should be, information in the article on the working out of the controversy. This is NPOV reporting on the debate, not what I've described above, which is about the *arguments*. But that isn't the work that I've done with this article, it merely is material that could be in the article that is more traditionally "fact." --Abd (talk) 21:34, 13 May 2008 (UTC)

Ok, whatever. --Explodicle (talk) 14:02, 14 May 2008 (UTC)

[edit] About the use of advocacy groups as sources for this article.

This article is about a controversy, and in particular about arguments being presented in the political arena in favor of or opposed to the implementation of Instant-runoff voting. Generally, advocacy groups present arguments on their own web sites, and, if the organization is notable in its field, may be reliable source for a statement that the argument is being made. In putting together this article, as well as in the main article, advocacy organizations have not been -- or should not have been -- used as sources for fact, other than the fact of an argument being made. Originally, in the IRV article, the Pro and Con section consisted only of arguments stated in few words, without sources at all, like "Fixes the spoiler effect," or "Reduces negative campaigning," and without any examination of these claims based on what is available from reliable source, or comparison with counterclaims. Some of that has crept into this article, and it will take some work to remove this or properly source it.

There are charges in the current AfD that this article is a POV fork, i.e., a fork of the article designed to express a particular POV. However, which POV would this be? I've just started going over the article and find some FairVote argument inserted, for example, referenced to FairVote, presented as fact and not attributed. I.e., instead of the original plan that arguments were presented (and attributed, which could be to notable advocacy sites or sources), but the discussion of the arguments would be strictly NPOV, with reliable source required, POV sourcing for *fact* has been inserted. Is this a pro-IRV POV fork or an anti-IRV POV fork. Actually, it is neither, but there are pieces which have, to some extent, become one or the other.

In any case, there is quite a bit of material in the article which is not properly sourced. I'm familiar with the topic, and much of it *can* be sourced, it's simply that sources weren't originally added, or, I suspect, some sources have been removed where an editor believed the source was improper, which, with an article like this, in some cases, may have been incorrect. The nominator for the AfD seemed to think that the Center for Range Voting couldn't be used as a source for anything, and that is not true. Generally, it cannot be used as a source for fact, but sometimes might be appropriate as a source for argument, or, in one case, hosts a readily available copy of the Minnesota decision, Brown v. Smallwood, which is quite relevant and not so easily obtained elsewhere. (the URL, if used, would be to the Brown v. Smallwood page, not to Center for Range Voting discussion of it, as to reliable source, though argument from that source *might* be cited as argument, if notable. --Abd (talk) 03:28, 14 May 2008 (UTC)

[edit] edits to this article by a block-evading sock of User:Fredrick day

Edits to Wikipedia, from the IP range of 87.112-87.115, when connected in some way with me or certain other editors, have always been edits from User:Fredrick day, blocked for incivility. He follows my contributions, and, I assume, when he saw my comments in the AfD for this article, he jumped in. I'm generally more free with reverts of IP edits, particularly when they come from such an identifiable source. Some admins have removed edits from Fd on sight, as have I, and the only time it caused any fuss was when the edits were, essentially, a trap: he removed allegedly unsourced material from articles on porn stars. The nature of the material would cause someone who didn't research it to think that it was improper for a biography of a living person....

Now, as to the edits. I've just reverted his removal of sourcing for arguments to the Center for Range Voting (http://rangevoting.org). This is certainly a notable organization critical of instant-runoff voting. Note that the nominator of the AfD himself referred to it as a prominent critical organization. There used to be an article on it, but back in 2006, the same editor who previously nominated this article ("controversies"...), who was an SPA dedicated to AfDing articles relevant to election methods, showing blatant sock characteristics, though the master hasn't been identified, successfully AfD'd it, as one of his first edits. It probably was not sufficiently notable then. It is now, see Gaming the Vote, by William Poundstone.

CRV has collected and presents many of the arguments against IRV; it is, itself, prima facie evidence that the arguments are being advanced, but it is better if other sources be used if they can be found, as they generally can. Those writing this article referenced the Center for convenience. Further, with election methods, many facts are well-known, they are like reporting that A plus B equals C, when anyone can verify it by learning how the method works and looking at sample elections, so a lot of the basic facts about IRV and other methods are tricky to source. It can be done, and should. --Abd (talk) 15:06, 15 May 2008 (UTC)

User:Fredrick day, using known IP (a range with very rare edits not from him, see the comment of Sarcastic Idealist in the AfD), is continuing to disrupt the process here. As usual, he is attempting to set up a situation where a responding user will then call down the wrath of heaven upon himself. He's done it with edits to porn articles, and now he's found this convenient foil. Commonly, the source he is removing would indeed be unreliable. However, there are other considerations, and that is why editorial decisions are made by a consensus of editors, not by rigid application of rules. Indeed, there is no rule about sources, there is a guideline, WP:RS, and there are two superior *policies*: WP:V and WP:NPOV. This is an article about controversy, which means arguments. To show that an argument is being advanced, a notable source might be used even if it could not be used, for a fact other than the fact of an argument existing, due to, say, bias. As an example, FairVote is an advocacy organization, and it cannot properly be used for facts. However, it's notable, and that an argument is given on the FairVote web site is evidence that it is not a straw man argument, it is being truly advanced by real people. So there may be, in this article, citations to FairVote attributing and showing the notability of an argument.

Now, this is my opinion, there may be other opinions. How do we determine, for an article, where to bend the guidelines and where not? We do *not*, at least not initially, go to some reference work, the Guidelines, and see what we can make of the letter of the law. That's called "wikilawyering" and is disapproved. We decide by *consensus*. Where we cannot so decide, *then* we go through the whole dispute resolution process. But *all* of this is for legitimate editors. Blocked and banned editors may not participate, particularly when they were blocked for disruption of the process by which editors find consensus!

Generally, I treat edits by legitimate editors (even IP editors who don't show clear bias) very differently from edits by a blocked and disruptive sock. I tend to limit myself to 1RR when a legitimate editor is involved. But with socks and abusive IP editors, I am not so restrained. When I was cutting my Wikipedia baby teeth, I pushed 3RR with a sock editor working in concert with an abusive IP. A new sock was created which, as its only action, reported me to AN/3RR. The result, they were all blocked. I was utterly and totally vindicated. Fredrick day is a vicious vandal, and, while he was relatively uncivil when editing under the registered account, he was truly poisonous when editing from the IP range we have just seen. He managed, at least once and possibly more than that, in inciting a wikiriot that left one long-time editor indef blocked. He knows how to throw mud of just the right consistency that some of it may stick if people aren't being careful. He could succeed in getting me blocked for, as an example, reverting him here.

However, my goal is the welfare of the project, and I've often written, only half-jokingly, that if you've never been blocked, you aren't working hard enough for the project. Blocked editors should not be allowed to interfere in the process of finding consensus, and this includes a decision of what sources are proper for what text. I did not add the sources he has removed, but I see a purpose to them, and would want to discuss the matter with the editor involved, or with a legitimate editor who wants to remove them, in an attempt to find consensus. That's how Wikipedia worked, that's how it still works, mostly, and that's how it should work, and, yes, I'll lay my account on the line if necessary. I don't think it is, but .... you never can tell, there may be a couple of administrators who'd love to find a reason to block me. It would be pretty disruptive, though, I expect. I'm standing, here, for a pretty solid principle. So ...

Legitimate editors: if you think that anything Fredrick day removed from this article should stay removed, by all means, remove it again. It will not be put back in without discussion or other legitimate process. I do not edit war, except with socks and, in one case alone, a clear COI editor, bound by guidelines to avoid editing the article in any controversial way.

By the way, one of the claims currently being made is that I'm COI on this topic. See WP:COI. Sometimes ArbComm decides that an editor is sufficiently involved with a topic that they cannot maintain balance, and this is *like* COI, but it is simply called a topic ban. I am under no topic ban, nor have I faced any RfC or been blocked or otherwise sanctioned. The COI editor I mention makes his living with the topic, it's blatant financial COI, not merely a possibly biased opinion. What would be true is that, having studied the subject, and with no intrinsic conflict, I have opinions. A Point of View. But my belief is that we find NPOV, not by outlawing points of view, but by including them, by finding consensus among those who see things, each in their own way. I've been quoted by the Center for Range Voting, and they sometimes ask me to comment. That's the extent of it, I'm an independent writer, and not even that strong a proponent, politically, of Range Voting. --Abd (talk) 22:55, 15 May 2008 (UTC)

[edit] Do the CON arguments in this article apply to "any voting method"?

In the current AfD, it was written, It should be noted that most of the "CON" arguments apply to any voting method, so, even if they were sourced, they shouldn't apply to this article or the parent article. — Arthur Rubin (talk) 22:23, 18 May 2008 (UTC)

Let's look at the list: (starting subsections below) --Abd (talk) 21:47, 20 May 2008 (UTC)

[edit] With IRV, voting for a candidate can cause the candidate to lose

This is failure of the monotonicity criterion. For starters, see monotonicity criterion for a list of methods which satisfy and those which fail. IRV is unusual in failing. --Abd (talk) 21:47, 20 May 2008 (UTC)

Non-montonicity occurs in any system other than simple weighted voting. I seem to recall a theorem to that effect. What distinguishes IRV with non-I RV? — Arthur Rubin (talk) 03:21, 21 May 2008 (UTC)

No. Plurality is, for example, monotonic. You cannot, in Plurality, cause a candidate to lose by voting for the candidate. Look at the article I referenced. Not as a source for our article, of course, but it's enough for this purpose here. I could find reliable source on monotonicity, for sure. One problem at a time! I think Mr. Rubin has confused monotonicity with Arrow's theorem. Monotonicity is one of the properties used in Arrow's theorem. Arrow's theorem, for starters, doesn't apply to all voting systems, and, relevant to here, it certainly does not say that no system is monotonic; rather that monotonicity may be incompatible with certain other traits of fully-ranked systems that are also considered desirable. Non-instant runoff voting isn't a system covered by Arrow's theorem, to begin with, it's a two-stage system and as such isn't deterministic from the original ballots. However, if the runoff system is based on plurality elections in sequence, majority required, it's monotonic, and considers only first preference at each stage. Raising a candidate to first preference (only vote you cast!) can't hurt that candidate, period. See the article on monotonicity, it gives an example for IRV.

Mr. Rubin, because he mentions "weighted voting," may be referring to Range voting. Raising your rating of a candidate in Range cannot cause another candidate to lose, except to the candidate you raised the rating for. In Range, the vote for any candidate is completely independent of the vote for another. Same with Approval voting (Approval is really the simplest form of Range voting.) But this is a different property than monotonicity, though certainly Range and Approval are monotonic. In Range/Approval, you never cause a worse election outcome by giving your maximum vote to your favorite. That's not true of plurality (and, indeed, not of -- most? or all? -- ranked methods). Monotonicity only deals with the effect of your ranking of a candidate on the election of that candidate, not on the overall outcome. Montonicity failure means that raising the rank of a candidate can cause that candidate to lose. --Abd (talk) 04:06, 21 May 2008 (UTC)

The section on nonmonotonicity could be clearer. It's a property that is counter-intuitive, many people see this and go "Huh? How could that be?" And this is contributing to the opinion that this article is a hit piece against IRV. But, in fact, there is problematic material in both directions in this article. It is a work in progress, gradually being tightened up with reliable source and editorial consensus. There is a good explanation in the specific article on the Monotonicity criterion. There is disagreement over how important this criterion is; majority opinion among experts seems to be that the criterion that should go (given that ranked methods can't maintain all the ostensibly desirable properties, by Arrow's theorem) isn't monotonicity, but later-no-harm. If you think about it, later-no-harm is an anti-compromise principle. If you acknowledge that your second choice might be acceptable, in a negotiation with your neighbors, you could cause your favorite to lose. Later-no-harm is violated by any method which attempts to discover, from the votes, these compromises. Hence Robert's Rules of Order criticizes this method (strictly, it is describing STV with a majority election requirement, not exactly "IRV," but the problem is the same) for failure to find a compromise winner due to low "core support." To take this to an extreme, if every voter would be happy with Joe, but every voter has a personal favorite that isn't Joe, IRV *cannot* pick Joe. Even if every voter ranks Joe second. No method can force compromise, but some make it much easier to (1) express your true favorite and (2) still indicate reasonable compromises. And necessarily, then, the method can thus fail later-no-harm. (Later-no-harm is the property that IRV satisfies: unless a method requires a majority, which most IRV implementations don't, your adding a candidate at a lower rank can't hurt your favorite's chances, because that vote won't be counted until your favorite has been eliminated. But this conceals your lower ranked choices, which is why the method can fail to find that compromise.) There is, by the way, more material on monotonicity, from peer-reviewed journals, in the External Links section of the article. Some writers apparently, adding content, didn't specifically source it, but just, perhaps, dumped the full source at the end. --Abd (talk) 12:55, 21 May 2008 (UTC)

[edit] IRV suffers from the spoiler effect

This one is stated in a bit of a confused and confusing manner, and, as with much of the article, it was laid out to be a skeleton to be cleaned up with sources. The sources abound. The "spoiler effect" is not the correct word, that is synthetic. What should be here is what is known as "center squeeze." Robert's Rules calls it a failure to find a "compromise winner." IRV does ordinarily solve the spoiler effect, which refers specifically to a situation where there is a minor party with no chance of winning, itself. The problem happens under IRV when a third party rises to parity. IRV becomes very erratic, and can elect, under those conditions, a candidate who is opposed by two-thirds of the electorate, with just three candidates in the race. With more candidates, running neck and neck, it can be even worse. This is, of course, a rare situation. And apparently IRV acts to help prevent it from happening, by effectively suppressing third parties. At least that is how one line of argument goes. The point, though, is that center squeeze doesn't happen with any Condorcet method, nor does it happen with Range voting, Approval voting, or Bucklin voting. --Abd (talk) 21:47, 20 May 2008 (UTC)

An editor, without discussion here, removed a link added to the article to aid in reader understanding.[6] That wasn't inserted as a source, but because an article exists which covers the topic. However, the statement about Condorcet methods was incorrect. I'll remove it, but restore the link and see if I can find, also, a source to put in both articles. It's well-known, by the way. This isn't controversial; the debate over it centers over how commonly it occurs, not over it being possible.--Abd (talk) 14:28, 21 May 2008 (UTC)

Actually, this whole section is thoroughly confused, and it led me into that as well. There is the Condorcet criterion, which IRV fails, and that's related to center squeeze. By definition Condorcet methods satisfy the Condorcet criterion. There is Favorite Betrayal, which is a separate issue, and which should have a separate Con section. I'll try to fix the current section today. For starters, wrong heading. Robert's Rules of Order calls it a failure to find a compromise winner, as I wrote above, so the section, renamed, will probably use that language. I'm going to take the whole section out, pending.--Abd (talk) 14:40, 21 May 2008 (UTC)

[edit] IRV violates the one person one vote mandate

This *charge* is made against every voting system reform, but it's only true for a few. The exact meaning of the "one person one vote mandate" isn't clear, which explains the ambiguities. However, IRV counts some votes and doesn't count others, and some have argued that this is a violation of 1p1v. IRV advocates have argued that Bucklin violates this mandate, and that is one interpretation of Brown v. Smallwood, though one rejected by independent legal authorities I've seen, and by analogy this could apply to all preferential voting systems. Nobody, though, has raised this argument against Plurality, nor against top-two runoff. --Abd (talk) 21:47, 20 May 2008 (UTC)

[edit] "if it ain't broke don't fix it" or "plurality voting is good enough"

This argument obviously does not apply to plurality, but also not to top-two runoff; it's a general status-quo argument, and is actually seen. Here is an interesting fact, expect to see some reliable source on this soon. (The data is easy to find and confirm, but the conclusions haven't been published yet, to my knowledge, though they are blatantly obvious.) In all 9 IRV elections in the U.S. which went to runoff rounds, the frontrunner in the first round won the election. And the runner-up was still the runner-up. No exceptions. However, in one out of three top-two runoff elections held in the same places prior to IRV, the runner-up in the primary won the runoff. We *could* conclude from this that IRV is usually reproducing the results of Plurality and not normally coming close to changing them, so ... if it ain't broke, don't fix it. Plurality is good enough, would be the argument. And it is the argument. I don't claim that, by the way. Plurality is broken, but if it were very expensive to fix it, and IRV is very expensive, plus unreliable under certain conditions, I'd say, indeed, stick with Plurality. It usually works. Every one of the 23 elections that have taken place with IRV since 2004, Plurality would almost certainly have produced the same result. But, of course, this is debatable. The point here is that, no, this argument does not apply to all voting systems. There is one reform that is blindingly simple that it's a wonder we didn't implement it long ago. Just Count All the Votes. Which is a good idea in any case. --Abd (talk) 21:47, 20 May 2008 (UTC)

[edit] IRV will help preserve the two-party system

This is, again, obviously not an argument applicable to all voting systems. Here is source for it: australianpolitics.com This is not an advocacy site. It takes a neutral point of view (at least on this topic, I haven't read the rest of it). It gives, as an advantage of "the preferential system" -- which is we call "instant runoff voting" here, "It promotes a strong two-party system, ensuring stability in the parliamentary process." And as a disadvantage it gives "It promotes a two-party system to the detriment of minor parties and independents."

Yet we see the opposite arguments also. When IRV was thrashed in Ann Arbor, after one election where the Democrat won due to the single IRV election that was held, an argument made was that IRV would destroy the two-party system. And some third parties in the U.S. similarly argue that IRV will help them. But the general opinion among election experts is that IRV does, in fact, stabilize a two-paty system. The Republicans who ran the campaign against Ann Arbor IRV probably did not understand this, though I'm pretty sure their real motive was to win some more elections. And there are a few prominent supporters of third parties who have apparently realized that IRV isn't so great for them after all.

Does this argument apply to all voting systems? Probably not. The biggest problem facing third parties, though, is single-winner elections. The ultimate goal of FairVote is proportional representation, and this is where FairVote and most election experts do agree. The STV method, when used for proportional representation, especially with large districts, does allow third parties more effective participation. The Range Voting people generally think that Range Voting, while it obviously won't give minor parties election wins, will free the public to accurately express their support for third parties without risking a spoiler effect. And there is some evidence for this, whether or not it could be put in this article is another matter. --Abd (talk) 21:47, 20 May 2008 (UTC)

[edit] Summary

None of the arguments apply to "all voting systems." --Abd (talk) 21:47, 20 May 2008 (UTC)

[edit] About debate tag added to article

A debate tag was added to the article that says, "This article or section is written in the style of a debate rather than an encyclopedic summary." Now, this is an article about public debate and controversy. Reporting this controversy is reporting a debate. Currently, the article plan is this:

(1) List the arguments being made, pro and con.

(2) For each argument, source it as an argument actually being made. This can be to notable advocacy sites not otherwise considered reliable source, but clearly reliable as to arguments actually being made. (Not as to fact other than the fact of an argument not being a straw man or not notable.)

(3) Explain the argument and bring in what relevant facts can be reliably sourced.

Each individual argument section should be NPOV. And the overall article should be NPOV about the debate, the controversy.

Then, when this article is cleaned up, tightened up, and properly sourced, a summary can be taken back into Instant-runoff voting using summary style.

I understand that some think that the article reads like a debate. That's because, I'd suggest, it is about a debate! But because each argument section is (or should be) NPOV, it is already true that "pro and con" arguments are brought together. The problem may be, in some cases, that the section has not become complete. Or it is complete, perhaps with well-known facts (well-known among those familiar with the topic), but has not yet been sourced. Thus, generally, ^{[citation needed]} tags are quite appropriate, where something is not sourced. I don't personally take things out merely because of lack of sourcing, when I believe that sources can be found, and this is true for arguments and facts on all sides. I'll take out something that I think is *wrong* or *misleading* and can't be easily fixed. Ultimately, if a citation tag has been there for sufficient time, editors, again properly, may remove the unsourced text.

They can also do this at any time. Unsourced text may be removed, but I do suggest trying to understand it first! Maybe even try to find a source! --Abd (talk) 14:57, 21 May 2008 (UTC)

The article contains points (pro and con), rebuttals, and at least one surrebuttal. If this article is to exist anywhere in any form on Wikipedia, arguments about the same aspect (pro and con) should be combined, rather than split the way they are now. I still think it should be merged (as suggested below) into a general aspects of choice of voting systems article, but, even if restricted to IRV, the "pro"s and "con"s should be combined by aspect. — Arthur Rubin (talk) 19:05, 21 May 2008 (UTC)

I don't recognize the actual article in what Mr. Rubin has described. Yes, there are arguments, pro and con, a list of them, as section headers. But the sections are not "rebuttals," which implies POV. Rather, each section is, in intention at least, a neutral explanation of the argument, with attribution of the argument, and what is known about the facts asserted or implied in the argument. This should be NPOV. If a section states that IRV is not monotonic, that is not an argument, that is a fact. However, we do place a simple statement like that as a section header, and it becomes an implied argument. The complete argument would be something like "IRV is a poor method because it is monotonic." But, briefly, the section head is currently, "Voting for a candidate with IRV can cause the candidate to lose." Even though the real definition of monotonicity is a little more complicated, that described behavior is a consequence. (In reality, the word applies to harming a candidate by ranking the candidate higher; in some simple systems, like Plurality, there are only two ranks: a vote for or a blank, which can be, effectively, a vote against. So "vote for" is being used here for "elevating in rank." I don't think there is anyone familiar with the topic who disputes that.

Generally, an aspect (pro or con) is asserted by one side or another of the debate. There are, however, a couple of aspects that are asserted by both sides, taking different implications, there is one, about two-party systems, that really exists in all four forms (I.e., IRV will do A, that's good, IRV will do A, that's bad, IRV will not do A, that's good, and IRV will not do A, that's bad. This drove some editors crazy when it was in the main article!)

Controversy tends to be delineated, reasonably well and most of the time, by the arguments made by the various parties, and, through them, the underlying realities. For example, it is claimed that IRV will reduce negative campaigning, this is a very common argument. Nobody is claiming that reducing negative campaigning is a bad thing, and opponents of IRV don't bring up the negative campaigning argument except to rebut the claim. Therefore there is an easy categorization: reducing negative campaigning is a pro argument. Now, what can we say about this claim that is based -- or potentially based, as a matter of article process -- on reliable source? Well, we can first of all do what we should do with all the arguments: attribute them. These are *arguments,* if we were merely talking about the "properties" of IRV, it belongs in the main article; it is only when a property or alleged property is used as an argument for or against using IRV that it becomes relevant here. Anyway, what should be done, if it's not done yet, it should be, is (1) attribute the claim to some notable source; in this case, a letter to the editor of a newspaper *might* be sufficient, particularly if the argument is well-known; better would be an editorial or more extensive opinion piece, and we also make an exception here and attribute arguments to, say, FairVote, as a notable organization advocating IRV, and Center for Range Voting as a notable organization that generates critical content. The exact choices and standards are up to editorial consensus.

Then we can look for studies or reports regarding IRV and negative campaigning. There is one in particular, that has been cited, a report on the situation in San Francisco, published in an on-line (edited and reputable) newspaper there, which essentially claims that IRV does not reduce negative campaigning. Is this a "rebuttal"? You could take it that way, but, in fact, if you can find any notable facts to insert on the issue, that's equally welcome. The on-line newspaper piece wasn't exactly an opinion piece, it was reportage, and similar material should be put in if it will increase our understanding of the issue. The section on the argument is not organized by assertion/rebuttal, except that the assertion (the "argument") leads. The examination is then carried out it the best way possible to make the significance of the arguments clear, in an NPOV fashion. In this case, it might be mentioned that there is no study showing that IRV reduces negative campaigning. This kind of negative claim can be difficult to source; on the other hand, it's also easy to refute if it is wrong, and I've seen these kinds of negatives many times. We may be able to find reliable source for it, though, of course, it will always be, in a way, an opinion, because negatives are not actually objects of knowledge. So we say it as -- or should say it as -- "no study is known...." (meaning, the editors could not find one.) The first time there is a study at all close to the topic, that text comes out, and the study goes in. What might be of interest here is that a claim is being made, without any reference to studies. It is purely theoretical, based on the idea that candidates will seek support from the followers of other candidates. I think I've read some sources on this, with time, I'll look.

I'm not at all attached to keeping anything in the article that can't be verified. On the other hand, I tend to extend to other editors the courtesy of not removing something they put in that I think is true, from my general knowledge, even if it hasn't yet been sourced. I don't demand that in return. If someone takes out something that I put in because it is unsourced, then it is my responsibility to source it if nobody else does. We have the right to demand sources for anything, but my opinion is that the project benefits, we have a better encyclopedia faster, if we confine removal to stuff we suspect isn't true and cannot be verified. Our general consensus then serves as a diffuse "source." And I'm glad it is actually that way for many articles. But, obviously, in the long run, everything beyond the trivial must be sourced, in particular anything potentially controversial must be sourced. But the key policy is WP:V. And, in the meantime, there are always cn tags, to set a time limit for something being unsourced. --Abd (talk) 22:25, 21 May 2008 (UTC)

[edit] Yet another merge proposal

As I noted in the AfD, much of the "debate" (if adequately sourced) could be rewritten without much difficulty to discuss relative "pros and cons" of various (single-winner) voting systems, and placed in a general article about choices of voting systems. — Arthur Rubin (talk) 19:01, 21 May 2008 (UTC)

I don't think so, really. For example, there is a fair reform movement that focuses on Range voting. The ideas and principles behind that are quite different from Instant Runoff Voting. While there is some debate that might be notable on Approval vs. IRV, most of the controversy over IRV is really specific to IRV vs Plurality and IRV vs. top-two runoff, and then, there is a little bit of relevance of other methods to this article. There is an article, Voting systems which is quite good. And just about totally unsourced. Doesn't touch controversy. Lots of articles in the voting systems area are like this, written by knowledgeable people, trying to be clear and neutral, but not providing sources for what they consider well-known. And then there was Yellowbeard, who showed up and started AfDing anything inconvenient to the IRV agenda, it was blatant, once discovered. He showed up and voted in this AfD, first edit in months.

You know what's ironic here? If I were POV-pushing, I'd be arguing for Delete or Merge and Redirect, as many have in the AfD. Why? Well, that would provide reason to put back all the sourced stuff from this article, some of which is damaging to the IRV campaign, into the main article. In the AfD, there seems to be some conclusion coming up that this article was started to "stop edit warring." Sure. Consensus stops edit warring. Ah, well, you can lead a horse to water. The goal was to take summarized material from here back to the main article, not to keep all criticism here. This isn't an article for "criticism," it is for "arguments," which includes arguments advocating as well as arguments criticizing. This approach is one known for finding consensus, outside wikipedia. Just come to agreement on what the issues and arguments are, first, before trying to resolve them and figure out which ones are most important, etc. --Abd (talk) 23:16, 21 May 2008 (UTC)