Talk:Instant-runoff voting

From Wikipedia, the free encyclopedia

This is the talk page for discussing improvements to the Instant-runoff voting article.
Article policies
Peer review Instant-runoff voting has had a peer review by Wikipedia editors which is now archived. It may contain ideas you can use to improve this article.
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.
B This article has been rated as B-Class on the assessment scale.
Archive
Archives


Contents

[edit] Circular sourcing

I wonder how many newspapers and websites get their background information about IRV from Wikipedia. It seems like there could be potential for circular sourcing, in which we get our facts from them, and they get their facts from us. That could be especially the case on an issue like IRV that most people have never heard of until a proposal to implement it is raised. The reporter googles it, Wikipedia is one of the first links in the search results, and so they find our content and puts those facts in their article without saying where they got it from. People's websites often do the same thing. Accordingly, I think we should generally avoid citing newspapers and stick to academic papers and similar sources whenever possible. For instance, for the statement that Fiji uses IRV, I will try to find a link to the Fiji laws on that subject or another Fiji governmental site.

Improving the sourcing I think will go a long way toward resolving the chronic edit wars over this article. Every statement needs to be backed up by a citation from an unbiased source. We should try to avoid using organizations like FairVote, that have a pro- or anti-IRV agenda, as sources unless it's to say, "According to FairVote..." and provide their opinion. Facts, on the other hand, would be better linked to less biased sources. Anything that can't be sourced needs to be taken out.

It will take some effort to find and cite sources for everything, but the result will be a much better article which will be closer to meeting FA standards. Ron Duvall (talk) 03:56, 16 February 2008 (UTC)

Providing a source for "everything" is not standard for ANY encyclopedia, nor for Wikipedia. One might say, "how about only items that someone disputes?" Even that is inappropriate ...If an article on Tigers states that they are animals related to domestic cats, and somebody edits that out saying he disputes the factualness (god created all animals and all are unrelated other than by god's action), does that mean Wikpedia should provide a "neutral" source for the fact (zoologists are biased according to the editor who removed the fact because they all advocate the belief that Tigers are related to cats). The IRV article already has far more sources and citations than most typical Wikipedia articles I've looked at (maybe there is an actual citations per paragraph measuring tool?)
Tbouricius (talk) 00:01, 19 February 2008 (UTC)
See Wikipedia:You don't need to cite that the sky is blue; its counterpart, Wikipedia:You do need to cite that the sky is blue; and Wikipedia:Common knowledge. You are correct that most encyclopedias do not have a footnote for everything, and this encyclopedia has been pretty lax about it. However, we're getting to a point where quality is being stressed a lot more. This article is on a subject that most people aren't all that familiar with (they may have never even heard of it before coming here) and sourcing enhances its credibility. Given the edit warring that has gone on, it seems like the best course of action. The other thing is that references help students and others find additional sources (they generally are not allowed to cite Wikipedia in papers). So, there are benefits to citing even certain facts that are unlikely to be disputed. Moreover, it does happen sometimes that something plausible but incorrect will make its way into an article and stay there, and sourcing helps avoid that. Ron Duvall (talk) 00:16, 19 February 2008 (UTC)

[edit] Irish constitution and "proportional representation" in single-winner elections

I was looking for the Irish law establishing IRV in presidential elections. Article 12(2.3) states, "The voting shall be by secret ballot and on the system of proportional representation by means of the single transferable vote." (See http://www.servat.unibe.ch/icl/ei00000_.html ) I had never seen the phrase "proportional representation" used in reference to single-winner elections. Accordingly, I looked it up and found:

“ The President is usually simply said to be elected by STV or, incorrectly, by 'proportional representation'. While the constitution itself states that the President is elected under the system of 'proportional representation by means of the single transferable vote' (Article 12.2.3) this is arguably technically incorrect, because the term proportional representation can only meaningfully be applied to an election in which more than a single candidate is returned. ”

See http://www.irelandinformationguide.com/President_of_Ireland . How interesting! How did they possibly miss that when they were writing the constitution? I would love to take a look at the legislative history on that. Ron Duvall (talk) 04:12, 16 February 2008 (UTC)

I've actually read about this before. Whenever the Irish were writing their constitution, they heard about the proportional representation method from the British. Naturally, none of them knew anything about the party lists that had taken off all around the globe. STV WAS proportional representation to them, and that was it. Whenever they were talking about single-winner elections, and they mentioned proportional representation, they were talking about counting it with STV. IRV, alternative vote, and preference voting as names for the system in single winner elections had yet to enter the political syntax, so it's easier to understand it if you just take it that the Irish considered PR synonymous with STV.74.251.39.20 (talk) 07:16, 26 April 2008 (UTC)

[edit] Nomenclature: IRV vs. AV vs. STV, etc.

Should we call this article instant-runoff voting or change it to something else (e.g. alternative vote or even single transferable vote in single winner elections)? If we're basing it on which is more commonly used, we should look at the international norm rather than focusing on the U.S. (similarly to how we have an article named cannabis (drug) because that is more commonly-used term internationally, even though marijuana is the more familiar term to Americans.)

Anyway, "instant runoff" and "alternative vote" seem about equal in popularity. It appears that "instant runoff" has about 877 unique google hits. "Alternative vote" has about 833. Both of those seem kinda low; I wonder if I'm doing something wrong here. See:

(Presumably, we won't be using "ranked choice voting," since as far as I know, that is mostly limited to San Francisco and is more a description of how the ballot looks to the voter than of an actual voting method.)

So, which term is more descriptive – "instant-runoff" or "alternative vote"? "Instant-runoff" can, of course, be misleading if the results of the second round are not tabulated instantly (e.g. if ballots have to be run through the machine a second time to get the second round votes.) What is the origin of the term "alternative vote"? Is it because you are choosing an alternate in case your favorite candidate doesn't win, or because this voting system is an alternative to plurality?

Something like "single transferable vote in single winner elections," while descriptive and accurate, is quite a mouthful. Nonetheless, single-winner STV is often referred to as simply STV (e.g. in the Irish constitution) so it's a viable possibility. And it makes the connection between these two variations of STV immediately apparent. Ron Duvall (talk) 04:38, 16 February 2008 (UTC)

Wikipedia:Manual_of_Style#Retaining_the_existing_variety seems to suggest that if they are about the same popularity, then the existing term should be retained. SBPrakash (talk) 08:17, 16 February 2008 (UTC)
There's no "right" answer here, as different nations highlight different names. No one in Australia calls it "the alternative vote", for example, while people in Ireland tend to (inacurately) call it "proportional representation," as voter rank candidates in order of preference as they do in their system for parliamentary elections. "Preferential voting" is one possible alternative, but there are different kinds of ranked-choice, preferential voting methods, while "instant runoff voting" is generally accepted as one particular kind of approach. I think changing the name at this point would be problematic. RRichie (talk) 17:33, 16 February 2008 (UTC)
I would say that instant-runoff voting should be a disambiguation page with links to alternative voting, contingent vote, preferential voting, and single transferable vote. Markus Schulze 17:41, 16 February 2008 (UTC)
I don't have an answer, but I just started reading a book called Democracy in Divided Societies: Electoral Engineering for Conflict Management (2001), and wrote down the definitions used there:
  • Plurality-Majority systems:
  • Semi-proportional systems:
    • Single nontransferable vote (SNTV) - single vote, no runoff
    • Parallel (or mixed) systems - PR party list (Single-member districts & elected by proportional of party vote)
  • Proportional representation (PR) systems
    • Single transferable vote (STV) - rank vote, winner threshold, surplus transfer, bottom-up elimination
    • Party list PR systems - single vote for party, PR assignment
It says AV,SV, and STV are prefrential systems because voters can offer a rank-order of candidates.
It says there are different nomenclature by region: AV is called preferential vote in Australia (where ranking is manditory and Papua New Guinea also called preferential vote or optional preferential vote OPV since ranking isn't required. AV is called Instant Runoff voting IRV in the U.S. There's also the Sri Lankan supplementary vote.
Well, if I read this book first, I'd happily follow the terminology, and I don't mind Markus's suggest to make IRV a disambiguation page, and move the contents of this page to Alternative vote which now directs here. I don't think STV belongs here since it is more than a runoff process with surplus transfers.
It would be good to work the three articles Alternative vote, Supplementary vote, Sri Lankan supplementary vote down to TWO AV/SV. Well, previously I assumed the SV mean both a top-two runoff AND limited to two-rankings, which I thought was bad, but if SV doesn't imply a specific ranking I'm happier with it. Tom Ruen (talk) 23:06, 16 February 2008 (UTC)
WELL, THEN there is Contingent vote article here which is implied as the SV (from the book). I don't know where this name came from. So I'd say merging (Contingent vote, Supplementary vote, Sri Lankan supplementary vote) is good.
Back to a disambiguation page here, what else is called an instant runoff' besides AV/SV? I admit that's a wider open question from my knowledge, and MAYBE STV should be linked even as an incorrect application of the term. Tom Ruen (talk) 23:17, 16 February 2008 (UTC)
What do you mean by that last part, "MAYBE STV should be linked even as an incorrect application of the term"? That it is incorrect to refer to IRV as STV, or that it is incorrect to refer to STV as IRV? I agree with the latter, but not the former, because IRV is just one kind of STV. Ron Duvall (talk) 02:21, 17 February 2008 (UTC)
Well, for instance there's also a statement now in the article:
Multiseat variations of the IRV elimination process have sometimes been labeled as instant runoff voting although they should be more accurately called preferential bloc voting, since like bloc voting, multiple votes are counted per ballot at the same time.
At least preferential bloc voting is simulating a runoff process, while STV would never be performed as a multiround runoff process, although I suppose it could be done if ballots were not anonymous. Anyway, if this became a disambiguation page I'd accept linking to these multiwinner methods that use preference ballots even if they should not be called IRV. Tom Ruen (talk) 03:53, 17 February 2008 (UTC)

Which do you think is a more commonly-used system in organizations, preferential bloc voting or STV? If the latter, I think that a recommendation should be made that in the 11th edition of RONR, they include a description of STV rather than preferential bloc. P. 411 says "Preferential voting has many variations. One method is described here by way of illustration," which is somewhat of a disclaimer, basically saying they're not recommending it over other variations. Nonetheless, the authors of RONR have said themselves that the book is only to include rules that are already established by custom/practice in parliamentary law (i.e. they can't just include something they think would be cool). So, if preferential bloc is less common than STV, then the only justification I can think of for including the former in RONR is that (1) it's slightly simpler, or (2) it's a better system. #2 is not true, and #1 is not a particularly compelling reason. Something to discuss in more detail off-wiki. Ron Duvall (talk) 05:12, 17 February 2008 (UTC)

The problem with "Instant runoff" is that it was a political creation, suggested in, I think, 1996 (the name of the individual who suggested it is elsewhere in Talk) in correspondence with Rob Richie, who took the ball and ran with it. It was a political tactic, designed to capitalize on familiarity with top-two runoff, widely used where majorities are required. However, for quite a few reasons, it's quite distinct from top-two runoff and generally produces, when an election goes to runoff in top-2 or to elimination rounds in IRV, different results about one-third of the time, by my estimation. Before 1996, it was "preferential vote," which, of course, refers to the ballot, and is a name which was also used for Bucklin voting, historically. Since any preferential method, in general, can replace runoffs, at least to some degree (they all involve some kinds of "alternative vote" -- and even Approval does this, with the alternatives being equal to each other, but preferred over all others -- ) they all could legitimately claim the name of "instant runoff," but most notably, Bucklin, which was used exactly for that. However, since the name Instant Runoff was promoted by FairVote for specific political purpose, I have previously argued here that it is POV in itself. It implies familiarity, and equivalence, and this has explicitly been used to sell it. I has a lot of returns precisely because of this promotional activity, but, historically, this is not what it was called, it's a neologism, really.(This edit was from Abd (talk) 16:54, 17 February 2008 (UTC), continued below interspersal by User:RRichie.)

Abd, I didn't realize you were a mind-reader or someone who had reviewed all the minutes of our board meetings. If you had such magical powers, however, you would find that you are quite wrong. We always have supported proportional representation voting methods and still do. FairVote always has supported instant runoff voting and still does -- indeed our long-time chair John Anderson had an oped in the New York Times touting IRV (which we then called "majority preferential voting") in the New York Times in August 1992, just a few weeks after our founding convention. We weren't deceptive then and we aren't deceptive now. Your antipathy and lack of balance in this kind of thing is simply sad. RRichie (talk) 20:00, 17 February 2008 (UTC)
I'm not a mind-reader but I have a friend who is.
Richie's "you are quite wrong" is like a shotgun: I said many things. Were all of them wrong? What we know about FairVote is that it began as an organization promoting Proportional Representation, it grew out of a conference which was explicitly about that. Anderson's editorial may have been important, to be sure, but that doesn't change what was the primary thrust, for it is well-known that a major problem, looming larger than the occasional damage from the spoiler effect, is the exclusion of large segments of society, often even the majority, from any representation at all, because of single-winner elections for representatives. Singe-winner is the big problem, and the reason why resources were brought to a focus through the predecessor organization to the Center for Voting and Democracy. Wasn't it called the "Center for Proportional Representation." Correct me if I am wrong, my memory does get fuzzy. Further, this topic here is about the name. The name, Instant-runoff voting, did not exist until FairVote promoted it, it was, precisely, a neologism. Richie's contradiction of my comments, based on pure -- and largely irrelevant -- debate tactics, is thus the kind of smokescreen that I might expect from someone with a WP:COI, which is why he is totally welcome to comment here, but not to actually edit the article if there is even the smallest shred of controversy over it. He is, essentially, paid to hold opinions like those he expresses. We want to hear them, because they will become an element in what is NPOV and balanced, but they may not prevail, or we will have, as we had, a POV article.--Abd (talk) 16:22, 28 March 2008 (UTC)

(Continued edit by Abd (talk):)

The vast majority of single-winner STV elections that have been held have not been called "instant runoff voting," and the vast majority of current applications aren't called that, and giving that name to the article on the method is essentially promotion of the method. I've raised this before, but without consensus, would not move the article.

Given the prominence in the United States, which is sufficient for a separate article, I'd propose a main article on Preferential voting which there already is. And then an article on Instant-runoff voting, which covers recent U.S. history. Thus the article History and Usage of Instant Runoff Voting in the U.S. and the redundant article recently created on Implementation would be merged and confined to history under the name of IRV. Before 1996, all history in the U.S. was under a different name, and there weren't any actual elections for public office until, I think, 2004, -- and that was Ranked Choice Voting, which is rank-limited STV, which could be used for single or multiple winner, same ballot, but when there are many candidates, as in San Francisco, the ballot limitations force rather large ballot exhaustion unless voters vote strategically.

But I think the encyclopedic principle is pretty clear: the subject itself is not known, except in recent political campaigns, as "Instant Runoff Voting." The number of google returns, in fact, reflect heavy recent political activity in the U.S. When a method is in common usage, and not locally controversial, it's not going to generate a lot of hits!

Robert's Rules calls the general method "Preferential voting," but that does, in fact, refer only to the ballot form. The analysis they give is STV. STV can be analyzed in a number of ways, but, with single-winner, they all reduce to the same method. STV-PR (Proportional Representation) is in common use, but STV single-winner is not nearly as widely used, if I'm correct. There is probably a reason for that: when the method is single-winner, its vulnerability to strategy and its behavior with compromise candidates become problematic; with multi-winner, the problems are reduced to only the last candidate, generally (this may depend on details with which I am not familiar), and because IRV does, in fact, usually produce the same result as Plurality, the problems are relatively minor in the overall picture.

There really is only one good reason to pursue IRV in the U.S.: as a step to STV. Problem is, to do this, as a political reality, it's necessary to conceal the problems, to win public support, so the strategy chosen by FairVote (the original goal being PR) involves heavy misrepresentation of the method and its problems. Apparently that was considered acceptable to those making that decision, and this is where I (in later review) parted company with them. I do not believe that the ends justify the means, and, in particular, I reject deception and spin as political tactics. It's actually much more of a problem than Plurality voting. We need less of it, not more.--Abd (talk) 16:54, 17 February 2008 (UTC)

I've typically looked at IRV as being, aside from its greater complexity, a better system than plurality (although not as good as PR-STV). But you have said that in some cases, IRV would produce worse results than plurality. Can you cite a plausible scenario in which that could happen? Thanks, Ron Duvall (talk) 17:52, 17 February 2008 (UTC)

Well, if you have Plurality, voters understand the situation they are facing, and they vote strategically. With IRV, they may, instead, vote sincerely; with a better preferential voting method, this can be relatively safe; with IRV -- under some circumstances -- it can be a bad idea. Center Squeeze is a name for the problem, and this is Robert's Rules' criticism of it, though that criticism is muted: they merely point out the problem. That is, there are three parties: call them Left, Center, and Right. Left and Right parties each have more support, individually, than Center. But in pairwise elections against Left and Right, Center will win by 2:1. Any Condorcet method will choose Center, and Approval is likely to do so as well, with far, far less complexity than IRV. IRV can choose Left or Right, depending, in spite of the fact that almost two-thirds of the voters actually voted against the winner (in an extreme situation).

IRV is always associated with two-party systems, and Center Squeeze is unlikely in such an environment. Because IRV does solve the first-order spoiler effect, third parties not looking far into the future commonly support it; as do major party supporters thinking only of spoilers, and seeking those second-choice votes. But Warren Smith has called IRV "political suicide" for third parties, and some leaders have begun to realize this. IRV is fine if you stay in your proper place: as a minor, practically irrelevant party. But start to approach parity, the results could be very bad. Has this actually happened? Not with IRV, but with top-two runoff which can do the same thing. Top-two runoff, though, at least leaves the final choice in the hands of the electorate, whereas IRV elects based on the initial ballots. (the above was written by Abd (talk) 19:18, 17 February 2008 (UTC), [[User:RRichie interspersed the following paragraph:)

We know this is your point-of-view, Abd, but is not factual. In the very first mayoral elections with IRV in Burlington (VT) in 2006 and London (UK) in 2000, it played a central role in the election of third party/independent candidates -- whose candidacies almost certainly would not have happened under plurality rules. It also played a big role in the election of third party candidate Mary Robinson to the presidency of Ireland in 1990, after she finished second on the first count -- and again was liberated to run a strong race due to the presence of IRV. Sure, IRV is still a winner-take-all system, and in a nation like Australia with strong identification with the major parties, it doesn't lead to hardly an third parties winning. But they can contest every race and you won't find any third party leaders there who would trade it in for plurality voting, and no on there is talking about runoffs. RRichie (talk) 20:00, 17 February 2008 (UTC)

(continuing edit by Abd (talk):)

There are proper applications for IRV, but that's beyond the scope of what I'm writing today.

What I really had in mind was the behavior of IRV compared to Plurality and Top-two runoff. From the election experience so far in the U.S., I notice that IRV is producing the same results as would be produced by Plurality (i.e., just looking at the first round votes). Definitely, the first round in IRV isn't the same as Plurality, though what we are seeing is that all the other votes are moot, and, in fact, the top two aren't changing from first round to final result. Now, with prior elections in San Francisco and what I looked at in North Carolina (Cary), one out of three elections that went to actual runoff reversed the first round result. That's not happening with IRV. The difference, as a preliminary analysis, is significant: IRV is producing different results than top-two runoff, and the difference is significant. It is not merely replacing expensive runoff elections, it is changing the results. There are two possible explanations: perhaps having a single election is more fair. However, against this there is Cary, which has the opposite schedule from San Francisco. In SF, the "primary" is in the general election, and a runoff is special if needed. In Cary, the primaries (as with standard primaries for other offices in other situations) are special, and the runoff is with the regular election. This was, in fact, proposed in SF as an alternative to IRV. However, my own analysis considers that special runoffs are probably a good measure of preference strength, and thus probably improve outcomes overall. This has not, to my knowledge, been formally studied. But that the results are different is blatantly obvious from the statistics.

If it is desirable to have a majority of voters actually approve an outcome, jurisdictions have majority requirements and runoffs to make it happen. Then comes the IRV salesman who tells them they can have this without the extra expense. And the San Francisco ballot information actually said that "the winner will still be required to get a majority of the votes." Which was directly wrong, a misrepresentation of the actual text of the proposition. And in nearly all IRV elections that have gone to actual runoffs, the winner did not gain a majority of votes cast. (because of the way the votes have been reported, in two cases the counting terminated before all eliminations had been made, so we don't know the exact final votes; however, with reasonable assumptions, one of these did have a majority winner and the other did not, and I think there was one other runoff election where there was a majority found. All the rest, not.

I predict that, as these results become known, IRV as a reform is dead. There is a much simpler reform that may not be a perfect election method, but it is free. Just count all the votes, stop discarding overvotes. This method is very old, but was rediscovered about thirty or so years ago, and it has been promoted by election theorists and political scientists as one of the best methods from certain perspectives: Approval voting. Just count all the votes. Why didn't we do this from the start? I've encountered no early consideration of why overvotes are discarded, Robert's Rules instructs the clerk to do it because "they are errors, so the true intention cannot be known." That's circular!

This is not the place to debate the relative merits of Approval vs. IRV, that will happen elsewhere. But, as a background, to understand the importance of such seeming trivia as what "majority" means, it's important. And you asked. --Abd (talk) 19:18, 17 February 2008 (UTC)

RONR notes (p. 402): "If he votes for too many candidates for a given office, however, that particular section of the ballot is illegal, because it is not possible for the tellers to determine for whom the member desired to vote." It sounds like they're not even contemplating the possibility of approval voting as something a society would intentionally want to use. Ron Duvall (talk) 19:34, 17 February 2008 (UTC)

That's correct. The instructions to the tellers are based on an assumption that multiple votes are not allowed. To my knowledge, this is not discussed. RONR is designed for deliberative bodies, and it disapproves of elections where the winner does not gain a majority, and likewise elections using any kind of candidate elimination. That's the background. In this repeated balloting environment, Approval Voting isn't necessary. But when the preferred voting of RONR is translated to a single-ballot, plurality victory allowed application (which does require a special bylaw, since plurality victory is contrary to basic Robert's Rules), all bets are off. RONR does suggest that, if it is determined that a single ballot must determine a winner, preferential voting be used and is superior to Plurality. There are many kinds of preferential voting, and, at one time, the most common in the U.S. probably would have been Bucklin, though I think there were various STV methods in use as well. Bucklin is like IRV in that it proceeds in rounds, but it collapses the rounds as needed, becoming like Approval, in order to seek a majority winner. I presume that the editors of RONR are aware of all this, but they are constrained to only describe what is in actual use; at the time of editing of the latest edition, they may not have been aware of Approval Voting applications, or they might have suggested that as well. But for them, this is all a detail, and a disliked one at that. Deliberative process is superior, from a democratic point of view, to any single-ballot election method. There are, however, methods which do avoid the difficulties and inefficiency o multiple balloting, but which are clearly democratic, such as having a parliamentary system, elected by proportional representation of one kind or another, with some being more democratic than others. Generally, the more elected members in a district, the better the representation, if something like STV is used: I've proposed whole-jurisdiction districts with Asset voting, such that the electors created can assign votes by precinct to create seats, which then gives a combination of district-based elections (overlapping districts, to be sure) with almost perfect and total representation of every voter. Every vote counts.) --Abd (talk) 21:02, 17 February 2008 (UTC)

Abd is back. Looks like we're be archiving another round of talk pages in a couple days... If only we could "open a new section" on every topic Abd has opened, we'd have a dozen new sections here!
Should I assume there's consensus on merging (Contingent vote, Supplementary vote, Sri Lankan supplementary vote)?
My main question is I don't know where Contingent vote came from?
Okay, my book quoted above says It was first used in Queensland in 1892, where it was known as the "contingent vote". That's good enough for me. Tom Ruen (talk) 21:10, 17 February 2008 (UTC)

As to consensus on Merge, merge to what? Not to "Instant runoff voting," I think. As to being back, I was a bit distracted elsewhere; besides two small children who somehow imagine that I should be paying attention to them, and who know quite well how to make sure that I do, see Requests_for_adminship/Abd_2, and, remarkably, [1], [2], and the whole sordid history of sock puppets et al with this article, some of which is covered in the Talk page for the RfA, first reference above, and all of which is duly noted for my future guidance. Now, as I narrowly escaped being drafted into the janitorial corps here, where were we?--Abd (talk) 00:35, 18 February 2008 (UTC)

They want to merge them to contingent vote. Ron Duvall (talk) 00:46, 18 February 2008 (UTC)

As to what the article on IRV should be called, I assume Wikipedia should track actual common usage....so that depends on whether the English language Wikipedia is read mainly by Americans or internationally, I suppose. Within the U.S. probably 90% of Americans who are aware of it at all, know of it as "instant runoff voting," perhaps 8% know it as "Ranked choice voting," and at most 2% know it as Alternative Vote or other name. Tbouricius (talk) 00:19, 19 February 2008 (UTC)

Something has been done with Supplementary vote and Sri Lankan supplementary vote which really shouldn't have been done. Based on discussion here, redirects were put into place. Discussion should, at least, be mentioned on the pages involved, I think a Merge tag should be used; there should be broader input on what article should be primary. I notice that there was recent Talk for Supplementary vote, of interest, which can get lost in this shuffle. Ron, do you know proper procedure for this? Is there a tradition to copy Talk over? I'm considering reverting the redirects based on this.
As to bulk of usage, current usage is not the whole story. The method has been known by the same names in the U.S. as in the rest of the world, until 1996 and the work of FairVote. The name is a political device, intended to suggest familiarity, which is deceptive. It *looks* like runoff voting, but actually produces different results. (It's debatable, to be sure, which results are better, but it is different, that's clear.) In San Francisco, voters were told the following, in the voter information pamphlet (this was all here in Talk, but was moved by User:Tomruen to the Implementations article Talk):
Proposition A is a Charter amendment that would require the City to use an instant run-off voting method that would eliminate separate run-off elections. A winner would still have to receive more than 50% of the vote.
The previously law required the winner to get a majority of the vote. Did Proposition A do that? No. Jurisdictions have runoff elections because they want to guarantee that the winner does, in fact, receive a majority vote as traditionally defined, which is that a majority of valid ballots contain a vote for the winner. If they were content with Plurality winners, they would not have runoffs. But by naming the method "Instant Runoff Voting," the promoters of this method easily deceived voters into approving something that was quite different. When there is an actual top-two runoff, voters have an opportunity to reconsider the two candidates; some may change their votes. Top-two runoff, in the elections I examined, shifted the result from the primary in one-third of the elections. IRV did not reverse the position of the top two in any of the runoffs. While the meaning of this is as yet unclear, what is clear is that the two methods are producing different results. Strategically, top-two and IRV are, for most voters, quite similar as to first-rank votes. Turkey-raising is probably a fairly rare tactic which can, of course, backfire, I really doubt that it has much effect in real elections.
The point: the very name "Instant-runoff voting" is POV, a tactic in a political campaign in the United States. Lumping all the various methods promoted underneath the one name also has a political purpose: it allows a movement to claim "momentum," with all the various activities being considered as one, and creates a confusion in the minds of voters as to what they might be supporting in an initiative, a confusion that could easily favor approving the initiative.
Ranked Choice Voting is actually much closer to the traditional names, because it refers to the ballot, not to the elimination method. (And the name "Instant Runoff Voting" refers, not to the method, but to an alleged effect.) As I've mentioned many times, Bucklin voting deserves the name "Instant runoff voting" just as much as single-winner STV.
As an example of confusion over names, it is one thing to take a majority-required election and allow certain absentee voters to cast contingent votes (as is being done in Lousiana, for example), and quite another to collapse it all into one election for all voters (as in San Francisco. The value of those contingent votes, to those for whom voting in a second election could be extraordinarily inconvenient (or even impossible), is clear, and, for the majority of voters, there remains the classic opportunity to consider how to vote in the new light. While I certainly might prefer other reforms that might do a better job, I favor those reforms involving absentee voters; and this, indeed, is a classic usage of contingent vote. This reform, indeed, creates little additional counting cost, one of the other problems with full-on IRV.
--Abd (talk) 01:30, 19 February 2008 (UTC)

[edit] plurality spoiler vs. IRC center squeze

My vague memory is that 11% of plurality elections in practice are affected by the spoiler effect, but IRV suffers from center squeze a lot less; something like 3 or 1%. What's the correct figure? —Preceding unsigned comment added by 76.225.158.109 (talk) 09:12, 18 February 2008 (UTC)

Well, that may be based on certain narrow analyses. There is a paucity of data. Center squeeze is a peculiar phenomenon of IRV (or plurality, of course), whereas other preferential voting methods which also handle the spoiler effect don't show it.
Another point is that most elections involved are being held in two-party environments (or the equivalent, i.e., there are two coalitions of cooperating parties). In that environment, a spoiler effect is sometimes seen in plurality, but the data to determine center-squeeze isn't available, quite often. Center-squeeze requires some unusual circumstances: a polarized electorate, more or less balanced, and a centrist party that would be, broadly, the second choice of most voters, if not the first. This isn't going to be common, but when it happens, one could have up to two thirds of the voters seriously upset with the election result. Revolutions happen in environments like that. It could literally be a killer result.
With the spoiler effect, on the other hand, the situation is typically two major parties, more or less balanced, and a third party candidate sucks away some, usually small, percentage of the vote from one of the candidates, thus effectively electing the candidate those voters preferred least. Another name for this is vote-splitting. Under these circumstances, though, the support of the winner and the support of the losing major candidate are relatively close, and, while there may be some outrage, it would be unusual that it would rise to the level of civil disobedience or the like. Center squeeze? All bets are off.
In any case, this could be why Robert's Rules cautions about center-squeeze in its discussion of preferential voting, and ascribes it to the specific method described, single-winner STV, not to preferential voting in general. (Bucklin, for example, doesn't have that problem. Indeed, it's not clear what problems it did have, other than working as designed, which, I'm sure, was disliked by some with power.)

--Abd (talk) 01:44, 19 February 2008 (UTC)

What is the number from monte carlo studies? —Preceding unsigned comment added by 32.155.14.161 (talk) 22:50, 21 February 2008 (UTC)

[edit] "Exhaustive ballot" election

The article says, "The term 'instant runoff' is used because IRV is said to simulate a series of run-off elections tallied in rounds, as in an exhaustive ballot election." Most of the references I see to "exhaustive ballot" are Wikipedia mirrors, and "exhaustive balloting" brings up mostly references to the circa 1998 Australia constitutional convention, which picked between competing reforms; see http://www.suncoast.com.au/republic/convention/proc/hansard_1002_9.html . In reference to Australia, it is also mentioned at http://www.aph.gov.au/Senate/pubs/odgers/chap1108.htm . It was also used to pick Papua New Guinea's new governor-general: http://www.pacificmagazine.net/issue/2004/06/01/regional-briefs-1 . Exhaustive balloting sometimes refers to the technique described in Robert's Rules (see p. 426-427, where it is described, although it is not termed "exhaustive balloting"). I think sometimes people use it to mean that you had to take a heck of a lot of successive ballots to elect someone. See, for instance, Japan from War to Peace: The Coaldrake Records 1939-1956, page 457, which describes a process that took 28 ballots to elect someone. Supposedly, the leaders of National and Labour in New Zealand are elected by exhaustive balloting; see http://keithrankin.com/rf_shorts_1998_06juna.html . See also http://www.national-renewal.org.au/ConCon/proc/hansard_1002_10.shtml for some interesting comments on exhaustive balloting.

Ah, here are some good refs: http://www.securevote.com.au/gloss_of_terms.html and http://www.publications.parliament.uk/pa/cm200001/cmselect/cmproced/40/4005.htm and http://news.bbc.co.uk/1/hi/uk_politics/1171438.stm . I'll include one of those. Some of these other refs may go the exhaustive ballot article, which only has dead links at present. Ron Duvall (talk) 00:58, 19 February 2008 (UTC)

Good work, Ron. Repeated balloting without candidate eliminations (but with voluntary drop-outs and sometimes new nominations allowed) is what Robert's Rules prefers. This is generally considered impractical with broad public elections, but is, in fact, the norm in parliamentary procedure. Under standard procedure, no action -- and this includes elections -- can be taken without the consent of a majority voting.
"Exhaustive ballot" refers -- at least according to the Wikipedia article -- to repeated balloting with one elimination per round. For that to take 28 rounds would be extraordinary (29 candidates?), I think what was happening there was repeated balloting, not exhaustive ballot.
As to public elections, top-two runoff pretty much guarantees a majority vote: note that, because write-in votes are allowed in most runoff elections (sometimes with special rules), so "candidate elimination" doesn't totally exist, what is done is actually "ballot listing elimination," which is usually of the same effect, but the difference is a safety valve against a true problem with, say, center squeeze in top-two runoff. Generally, if I'm correct, the majority requirement is waived in the runoff election, so write-ins cannot prevent a winner from appearing. To be fully democratic, it should be possible for the election to fail due to eligible write-ins. That's the whole point: in a democracy, no vote is effective unless approved by a majority of those voting. But compromises are made. IRV happens to make a much bigger compromise than top-two runoff.
IRV with a majority requirement, and a runoff if the majority is not found, would be, theoretically, a better method. But in the elections we've been seeing, there would almost always be a runoff. However, note that in San Francisco, there was sufficient data to count the election as Bucklin, and Bucklin almost always found a majority. (IRV is missing the votes where the runner-up had lower-rank votes for the winner.) Indeed, there was only one election, as I recall, where Bucklin analysis did not show a majority for the IRV winner (who was also the Plurality winner from the first round, by the way). So if Bucklin were used, and the majority requirement kept, there would have been one runoff out of 33 elections. Isn't that interesting? (That election had a huge number of candidates, and only three ranks.) :--Abd (talk) 02:01, 19 February 2008 (UTC)
So what do you infer from all that? Ron Duvall (talk) 02:22, 19 February 2008 (UTC)
Well, I infer that, based on what I've seen so far, IRV is failing to accomplish the goal of having runoff elections: majority winners. I totally did not expect to see such massive failure to find a majority winner. So, with communities that are using top-two runoff, IRV is being sold as a method of gaining the same results, but without the extra election. Saving the cost of the extra election is a motivation that could justify the expensive of implementing IRV, which can be considerable. However, it looks like the same results could be obtained simply by dropping the majority requirement and using Plurality. The San Francisco results somewhat warp the matter: San Francisco has elections with a huge number of candidates on the ballot. 3-rank Ranked Choice Voting is, quite simply, inadequate, but, as well, even if there were more ranks, expecting voters to rank a large number of candidates is laying a task upon them that many wouldn't be able to handle; the result would be truncated ballots or donkey-voting (as in Australia, where complete ranking is required -- but with far fewer candidates!) However, Bucklin would, with three ranks -- as were actually used -- handle large numbers of candidates fairly well, at least with the Duluth version, because it allowed multiple approvals in the third ranks. Bucklin, as well, is quite a bit more efficient in finding all the votes for candidates. This very characteristic causes it to fail the Later-no-harm criterion which is touted as a feature of IRV (but which is possibly incompatible with a majority requirement for election).
Bottom line, IRV is not performing as represented when it was sold to various electorates, at least not with San Francisco, and results elsewhere, with far fewer candidates, are also showing majority failure. Really, this is, for me, a totally unexpected result, I thought it would perform better. Given that there is now some organized opposition to IRV, unlike the case in San Francisco in 2002, and these results are stark, I think the IRV movement is now pretty shaky; from sheer momentum, it may progress a little further.... but most of the decent arguments for IRV would also apply to Bucklin, Bucklin has been used before in the U.S., it solves the spoiler effect, and unlike Approval (the best bang-for-the-buck reform, infinite benefit-cost ratio), it does allow a favorite to be specified. Bucklin is really easy to implement, it's precinct-summable, no votes are wasted; if we must have single-winner, Bucklin is looking pretty good to me. And then we could turn our attention to multiple winner and probably STV, where, presto! I and, I suspect, most election experts could find common cause with FairVote. (I may think that there are better proportional representation methods than STV, but the difference isn't great. I'd *love* to see Asset Voting, but I'm not going to hold my breath.)
For the purposes of the article, most of this is dicta, background. It establishes the significance of facts that might be in the article. Facts and arguments are different; facts are -- except for balance -- NPOV, arguments are inherently POV, though they may be reported in NPOV fashion; notability then becomes the issue. Someone with a POV critical of IRV, as is true for me, will want to see the underlying facts be in the article, and someone with opposite POV may want them out. Standard policy suggests that relevant facts, properly sourced, should remain in articles, and that imbalance be addressed by balancing them, not be deleting them. For quite some time, the standard practice here was to delete what might "confuse" readers. I.e., might cause them to question what the editors wanted them to think....
--Abd (talk) 02:51, 19 February 2008 (UTC)
I prefer to see detailed stuff whose sheer volume would tend to overwhelm or throw an article off balance split off into other articles, with a summary and link left on the main article. Instant-runoff voting controversies, for instance, has been very successful. Ron Duvall (talk) 02:56, 19 February 2008 (UTC)
Summary style would suggest that subarticle examine various aspects of an issue in detail, with a balanced presentation in the main article summarizing the subarticles. It's tricky; sometimes subarticles are used by POV editors to remove inconvenient facts from an article, so that readers may not notice them. This, essentially, is the forking that is disapproved. What I've come to is that subarticles should be well-developed, and then taken back into the main article in summary style, and that what is mentioned in summary should reflect editorial consensus, it should not omit what would be important to any notable POV. --Abd (talk) 03:24, 19 February 2008 (UTC)
P.S. Only 95% of the contents of Instant-runoff voting controversies suck, down from 99.5% - real progress since that 95% is no longer in this article! Tom Ruen (talk) 03:56, 19 February 2008 (UTC)

[edit] The Summability phantom again

User:McCart42 added this to the section Instant-runoff_voting#Changes_for_counting:

It is important to note that hand counts over multiple districts are considerably more difficult. For each round of elimination required, a recount of all the ballots must occur in every district. This occurs due to instant runoff voting's lack of summability - unlike most other methods, a tally of votes cannot be compiled in O(n^k) where n is the number of candidates on the ballot. For plurality, approval voting, and other cardinal methods, k=1; for Condorcet methods and most other ranked choice methods, k=2; but for instant runoff voting, since the order of each ballot is significant and not just the relative positions of each candidate, the complexity of the sum is actually O(n!). [3]

I'm not 100% against this point being made, if it is being made in a way that an average reader has a chance of understanding, AND that it is written in a way that acknowledges options rather than just whining about an issue. I mean WTF about factorial complexity!? There's no reason for this approach. There's at most N rounds of counting. It's a POV argument to say otherwise.

What I want to know (and what readers want to know) is "what is DONE in current implementations", and since NO ONE does factorial counting, why bitch about it?

There's a number of approaches to the multidistrict counting problem:

  1. Do a "first round" count by district, and physically transport ballots to a central location for recounts as necessary.
  2. Do a "first round" count by district, and electronically transmit all ballot information to a central location for recounts as necessary.
  3. Do a "first round" count by district, and recount by district as needed.

I really see no reason to complain about summability. IRV is not SUPPOSED to use lower rankings at all, unless needed. It is supposed to be multiple rounds of counting. HECK, it might even add a little suspense to be told the "first round" tally and wait a day or two for the runoff results!

AS WELL I AGAIN must object to the use of another wiki, which was copied from an old Wikipedia article that has since been deleted! [4], now only exists as a copy at User:Abd/Summability_criterion. If there's references to be found, bring them up.

Tom Ruen (talk) 01:08, 27 February 2008 (UTC)

I agree that it is inappropriate for this Wikipedia article to wade into irrelevant math formulas. No jurisdiction using ranked ballots reports sums of all possible or voted permutations from polling places, so this hyporthetical number of possible permutations is irrelevant. It would suffice to say the reason IRV can take longer to tally (as stated in the prior paragraph) is that district totals cannot simply be summed as with plurality voting and some other methods. 71.161.192.19 (talk) 02:15, 27 February 2008 (UTC)
  • The reason district totals cannot simply be summed with IRV as they can with every other method (I cannot find another which fails summability) is that the number of possible ballot permutations is O(n!) in the number of candidates. The counting is not O(n!), rather the final summed information is O(n!). This is why for most IRV elections, you see incomplete data being presented as results in terms of "number of first preferences", etc. By publishing incomplete data, you are making it impossible for outside observers to verify that IRV chose the correct winner, or how close the counting actually was. It is true by definition that if a Condorcet winner exists, and IRV does not choose that candidate, a majority of voters would have preferred the Condorcet winner over the winner of the IRV election. Regarding summability again: for Condorcet methods, you could publish a pairwise matrix of n^2, or n x n, size, which would reveal all the necessary information about the ballots. - McCart42 (talk) 06:38, 27 February 2008 (UTC)
The total information needed is n! if you like hypersparse data of 99.999% zeros among 20 candidates! BUT all you're saying is that it wouldn't be done with more than a 3-4 candidates. Instead you can transmit total ballot information (n*m elements), for n candidates and m voters, say 20 million numbers for 20 candidates, 1 million voters - that's much bigger than n^2,(400 for 20 candidates) but far short of factorial information. Summability is a nice property if you can get it, but it's just a mathematical property, nothing to do with any democratic principle. It isn't needed to verify the vote, and it might even be undemocratic since lower preferences are promised to be left uncounted in IRV, so some might argue there should be no lower preference tallies at all in IRV. It is conditional data that is intended only to be counted under conditions specified. You can recount IRV's (n-1) rounds as many times as you like. ANYWAY, my point is that "summability" isn't something that needs mentioning in a "changes for counting" section. Just give the facts - lower preferences are not needed to be counted unless later rounds demand it. There's three approaches to this in districts. End of story. No need to talk about summability there. (Incidentally, why has no one reintroduced a summability article in wikipedia if it is so important?) Tom Ruen (talk) 07:23, 27 February 2008 (UTC)
At one point I found reliable source for the summability criterion, I don't have time to check it out now. If the article is in my userspace there might be a ref there, but I think I found it with google search. I'd introduce the article if I had time, it's on my list of things to do. By the way, our old friend Yellowbeard is back, the one who snuck the AfDs, including this one, under everyone's radar, he's become a pure SPA now, only one purpose: damage Abd and his friend. Special:Contributions/Yellowbeard. Nice friends you have, FairVote editors. Here is, as this is written: his latest edit: [5]. How did he respond to this message? [6]. —Preceding unsigned comment added by Abd (talk • contribs) 08:05, 27 February 2008 (UTC)
  • By withholding complete ballot data from the public, governments running IRV elections are ensuring that the mistakes made by the method cannot be revealed. In elections where full information is available, these mistakes are obvious [7]. Lower preferences are discarded by IRV but that doesn't mean that the voters would consider them meaningless. They wouldn't have filled out the ballot beyond their top two if they didn't put some meaning behind their vote. - McCart42 (talk) 23:44, 27 February 2008 (UTC)
    • I'm confused - what's a mistake made by the method? You mean when a Condorcet winner isn't selected? Tom Ruen (talk) 23:57, 27 February 2008 (UTC)
That's okay, Tom, confusion is the beginning of knowledge. It would mean something that would cause the electorate to regret that IRV was the election method, at least a majority of them. If the data were available, in such a situation, the majority might, indeed, seek to change the method. In what McCart42 described, it would be the loss of a Condorcet winner. But there is a bigger one. Imagine a three-candidate election, A B and C, where every voter who does not rank B in first place ranks B in second place, never in third place. And the electorate is evenly divided between all three candidates as to first place choice, almost exactly. However, by a small margin, B is has the lowest vote count of the three in first rank. This scenario is not a terribly uncommon one when an election is outside the two-party system. Just imagine Left, Center, Right. Not at all difficult to imagine that this could be roughly an even split, with one third (plus, say) voting A>B>C, one third (minus) voting B>A>C and B>C>A, and one third C>B>A. If the method is IRV, B is eliminated in the first round, and only the first round votes are reported (often), so it remains invisible that B was preferred to either of A or C by a margin of 2:1. This is a case where two-thirds of the electorate would consider that the method made a "mistake." And the C supporters could be regretting their sincere vote, for if they had changed their votes to B>C>A, B would have won by a two-thirds margin. I.e., if even a few of them had followed Favorite Betrayal, just enough to make B the second place candidate in the first round rather than the third place candidate, they would have gained a far better outcome.--Abd (talk) 04:56, 28 February 2008 (UTC)
ummmm, I think I'll play dumb since I'm tired. At least I'm confirming this has nothing to do with a section called Changes_for_counting. Tom Ruen (talk) 05:01, 28 February 2008 (UTC)
  • No, it definitely does. You can count all the IRV ballots in polynomial time, fine, but, after counting, you can't present the results in polynomial complexity without removing information by truncating preferences. And that's a problem - maybe not for 3 candidates (which is why San Francisco uses rank-3 IRV), but for any more candidates it breaks down - and the whole point of non-plurality methods is that they are supposed to handle a more diverse slate of candidates. IRV cannot. Every other summable method can. - McCart42 (talk) 15:24, 29 February 2008 (UTC)
  • (I'm considering the "show the results" as part of the counting procedure, since what good is counting if you can't represent all the information you were given?) - McCart42 (talk) 15:28, 29 February 2008 (UTC)
  • I have no idea what you're talking about, not playing dumb. I guess I am dumb. Tom Ruen (talk) 19:29, 29 February 2008 (UTC)

I'd like to make this clear so everyone can understand it. Restate it in your own words if it will make it simpler. When counting ballots, you must store the result after each ballot is counted. Let's call this result storage the "tally", for simplicity.

So let's say we have a 4-candidate election, with candidates named A, B, C, and D. Under plurality or approval, this tally is a 1-dimensional array 4 entries long, and each entry is the total votes for each candidate. That array is what shows up on the news reports after an election, and it shows that candidate A got more votes than the other 3 candidates, so A won. Cool.

Under any Condorcet method, the tally is a 2-dimensional array, 4x4, where each cell represents a preference of the row candidate over the column candidate. You can name the rows and columns ABCD rather than 1234. Let's say the first ballot is A>B>C, with D unranked. Then row A gets +1 in columns B, C, and D. Row B gets +1 in columns C and D. Row C only gets a +1 in column D, and row D is still all 0s. So this is a little more complex, and maybe the anchors on the news reports need to use greenscreens like on the weather reports and point at the 2D array to reveal just how a particular candidate won, but that is the inherent complexity in a well-designed preferential voting system.

Now under IRV, not every preference is counted. You start by counting the first preferences, and maybe you show a 1D array for the first count, and if there's no majority then you go plurality-style and discard the candidate with the fewest 1st-rank votes and show another 1D array for the 2nd count, and so on until you have majority (assuming everyone fully ranks their ballots, otherwise you may not get a majority). But the point is that in counting under IRV, every ballot's permutation of preferences is significant, because you're not actually going to count all of the preferences. So when you store the ballot information to transfer it to a central location, or when a local news program displays the results of their area's election returns, neither the series of 1D arrays nor the 2D array is an accurate picture of who won the election. More importantly, there are real problems with having to do a recount across all districts in the case of a close election. If Florida is a mess, so is the rest of the country, under preferential methods. Approval voting is consistent, while no preferential methods are. So you can clearly see where this is an issue for changes for counting -- and could drive up the cost dramatically in close elections. - McCart42 (talk) 22:44, 12 March 2008 (UTC)

[edit] Terminology

There is an incorrect entry comparing the Iowa Caucus method of voting to IRV:

"Observers described the Democratic Party caucuses in Iowa as like "instant runoff voting" because backers of minor candidates could move to a viable candidate after the first tally at a caucus.[10]"

If anything, this entry furthers the misunderstanding of what IRV is. In the Iowa Caucus, you have the chance to change your mind on who your second choice is after you see how the first choice votes are awarded. But with IRV, you make all your choices at one time and you can't change them.

While it may be true that observers thought that the Iowa Caucus was like IRV, they were wrong, and their opinion should not be in an article about IRV unless a strong statement is made in the article to explain the major differences between IRV and the Caucus. Then we could theoretically spend pages on how other systems seem like IRV but really arent. Why do it? I would like to get a second opinion on removing this language please. --Ask10questions (talk) 04:39, 11 March 2008 (UTC)


[edit] Archive 6

I moved content before February 13 to a new Talk:Instant-runoff_voting/archive6. Tom Ruen (talk) 02:19, 28 February 2008 (UTC)

[edit] WikiProject Parliamentary Procedure has been formed

I just wanted to let everyone know about the formation of WikiProject Parliamentary Procedure. We hope to cover all the major motions and parliamentary procedure terms. There will probably be a certain amount of overlap with Wikipedia:WikiProject Voting systems and Wikipedia:WikiProject Elections and Referenda. As mentioned at Wikipedia_talk:WikiProject_Parliamentary_Procedure#Voting_methods, voting-related topics are being classified under the headings of:

There is a bit of overlap between voting system and voting basis, of course. Obuibo Mbstpo (talk) 19:42, 8 March 2008 (UTC)

[edit] delete nomination for Instant-runoff voting controversies

I have nominated the "instant-runoff voting controversies" article for deletion here:Wikipedia:Articles for deletion/Instant-runoff voting controversies (2nd nomination). Please review the article and weigh in. QuirkyAndSuch (talk) 13:19, 13 May 2008 (UTC)

Note that WP:FORK describes the formation of a detailed article (even merely about "criticism,") under conditions which were met here, and particularly when it is a result of a consensus of editors, which was the case for the nominated article. The "controversies" article is not, as the nominator claimed, a "POV fork." It needs work, to be sure, a good deal of material was added by various writers based on common knowledge, and sources remain to be supplied. For example, I'm a critic of IRV, but when facts were stated in the article that I know to be true (and verifiable), I don't generally remove them; to do so, while useful for demanding sourcing, actually can unbalance an article, thus violating NPOV. But it is, of course, anyone's right to demand sources; but the respectful way to do it, when there is suspicion that the text is true, and sometimes even when one suspects it is not, is to place a [citation needed] tag, and only remove if too much time elapses with no source provided. Best, of course, is to find and cite a source. --Abd (talk) 03:09, 15 May 2008 (UTC)

[edit] About IRV and Robert's Rules of Order

The nomination of the Controversies article brought my attention back to this article, and I realized that the reference to Rober'ts Rules of Order was really improper in the introduction. That reference is clearly an attempt to promote the method, it is not sufficiently important to justify placement so prominently, and if the reference is made as simple as it needs to be to be in the introduction, the simplification turns it into an apparent recommendation of "IRV." To properly present what RRONR says about preferential voting requires much more text than is appropriate for the introduction. Much of the push to keep it in the introduction came from User:MilesAgain, who was subsequently found to be a banned editor, who had long participated in maintaining a POV slant to this article. So I took it out, and I bolstered the section in the article describing what RRONR has to say about the topic.

In particular, in previous discussions, I had a hard row to hoe in claiming that RRONR required a majority for election, a true majority, i.e., a majority of ballots cast, not of ballots which contain votes for the candidates remaining in the last round. The latter is a mathematical certainty, and it clearly defeats requirements of election by majority. IRV advocates claimed my position was preposterous or pedantic; however, I realized that, right under our noses, was text in RRONR that confirmed that a majority is required. It is suggesting that members be "thoroughly instructed," and, in particular, it says that "Sometimes, for instance, voters decline to indicate a second or other choice, mistakenly believing that such a course increases the chances of their first choice. In fact, it may prevent any candidate from receiving a majority and require the voting to be repeated." This is a clear reference to what I had assumed on parliamentary principles; the problem had been was that the text on how the counting is done and the winner chosen was possibly ambiguous.

This majority requirement, in fact, changes certain characteristics of the method. The editors of RRONR actually erred in what I just wrote. When a majority is required, IRV may no longer meet Later No Harm, depending on what process ensues. Adding a second choice with majority-required IRV can hurt your first choice, and here is how: if you don't add a second choice, perhaps the election will fail and your favorite will go on to prevail in the repeated balloting. RRONR does not favor any sort of elimination with repeated balloting, but bylaws could provide for it. The Vermont constitution does require a true majority for the election of the governor, and if there is no majority found, the election goes to the legislature, to be decided by secret ballot between the top three candidates. Adding a second choice vote could cause that candidate to gain a majority, where otherwise the candidate could have a chance to win in the legislature. The legislation proposed by our own User:Tbouricius in the Vermont legislature included a ballot instruction that reproduced the RRONR error. (I'm not aware of anyone else describing this, this may be my own discovery, so I'm not attempting to put it in the article, but it is clearly true and can be verified by anyone. But, of course, if someone finds a source....)--Abd (talk) 03:33, 15 May 2008 (UTC)

Abd -- Here is a simple question for you. If an organization is following Robert's Rules of Order and is holding a vote-by-mail election where they've decided to get it done with one mailing (as would be the case in an overwhelming number of such vote-by-mail elections), what system do yhou think they will use if they follow Robert's Rules of Order? And as a related point, why do you think so many more private organizations use instant runoff voting than Bucklin? Just might it be because they are following Robert's Rules of Order? RRichie (talk) 14:48, 15 May 2008 (UTC)

Nice argument, Rob, but ... RRONR is explicit. RRONR does not allow preferential voting unless a bylaw authorizes it, and bylaws can authorize just about anything. RRONR does not recommend preferential voting -- of any kind -- except as a lesser evil. RRONR mentions that there are many other forms of preferential voting than what you call IRV. Then, if it *is* sequential elimination, the majority required, unless the bylaws have otherwise provided, is a majority of ballots cast, not a last round majority, and RRONR is explicit about that. By the way, the APSA "adoption" is probably going as well. It is practically trivia, since it has not only never been used, it isn't even close to being used. You want APSA mentioned because it would seem that political scientist would use the best method, right? But, if so, then the best method is to have a nominating committee that makes a single nomination which is ratified at the annual meeting. That's what they actually do. They not only don't elect by IRV, they don't elect by mail ballot either, they elect deliberatively from what I think is an appointed body, though I'm not sure. Bucklin I mentioned because, first of all, when it was used in the U.S., it was called "preferential voting," which is why I sourced Brown v. Smallwood for the name, it was convenient. It has actually been used. In any case, *don't* edit the article itself where it's controversial. You stated it was controversial, so you should have known not to do that. Don't make that mistake again. What I put in was sourced, and clear from the source.

And what I took out was a simplified and therefore POV, and unnecessary, mention of the RRONR discussion of preferential voting, in the introduction, where controversy and spin should be rigorously avoided unless explicitly stated as such *and* necessary. --Abd (talk) 15:25, 15 May 2008 (UTC)

[edit] Concerns about Abd making controversial edits without discussion / Request for others to get involved

Abd is quick to say I have a conflict of interest because I am an advocate of instant runoff voting and work at a non-profit that promotes the idea. It's true, I run FairVote. It's also true I worked here and advocated instant runoff voting for a long time without any compensation. I see myself as a well-informed advocate who is quite capable of working with others to create an objective article.

Just as I am an advocate, Abd is a well-known zealous opponent of instant runoff voting. True, he doesn't get paid. But his level of objectivity is no different than mine.

Objectively, many of the edits Abd has made on this article are outside the conventional way that all established encyclopedias talk about this subject. For instance, in political science literature, it is not even remotely controversial to call instant runoff voting a majority voting system -- yet Abd vigorously contests that. It is not controversial that nearly all private organizations would interpret Robert's Rules of Order on this subject to mean what this instant runoff voting Wikipedia article has said on the subject - -yet Abd vigorously asserts that Robert's Rules isn't suggesting IRV for vote-by-mail elections in the face of all evidence to the contrary.

For something like this subject of Robert's Rules, which has gone back and forth here for months entirely at Abd's instigation, I don't think it's appropriate for Abd to jump in and without warning make major changes. Certainly it's revealing to to me that he inserts for no clear reason mention of a very different system he prefers, Bucklin Voting.

I sure wish there was a Wikipedia "super-editor" who could take a look at this subject, look at how other well-established resource material writes about it, and sees if the article is consistent with that resource material. In the meantime, I think Abd should restrain himself and not make such controversial edits before other clearly objective editors (if they're out there) weigh in. RRichie (talk) 15:29, 15 May 2008 (UTC)

Just to make it clear. RRichie is financial COI, for starters. I have opinions, as do many editors. Richie's major work is promoting IRV and many of the common arguments found for it were his work. He introduced the very name "Instant runoff voting," which was, politically, a brilliant move. And quite deceptive. IRV is not a major focus for me; I happen to have become knowledgeable about it and about the political spin involved. So while I have a POV, so do most knowledgeable editors, I am not COI, I do not have a formal conflict of interest. Now, there are two changes made here:
I removed the mention of Robert's Rules from the Introduction. I've explained why at great length before, and we had an RfC on it, and IRV advocates kept returning it. The Introduction is a political prize. IRV advocates know that getting an important point into the introduction is putting it where far more people will read it, and, if it creates a positive impression, where a more detailed examination would deflate that, they really want it in the intro. This is an example. Robert's Rules of Order would greatly prefer that repeated balloting be used, and they still expect that repeated balloting will be used if no majority is found, even if the sequential elimination method Richie calls "Instant runoff voting" is being employed. What they recommend, if mail balloting is required, is "preferential voting," quite specifically not "Instant runoff voting." Preferential ballots can find a majority more thoroughly than Plurality, which is why they say it is fairer. But IRV is sold as finding a majority without runoffs, and the method, as implemented here in actuality, terminates without finding a majority, except by the *new definition* of a "majority of last-round ballots" obtained by simply eliminating from the basis for the majority, all ballots not containing a vote for one of the top two. Legally, they would never be able to get away with calling this a majority when law requires a majority, so, in San Francisco, part of what the Ranked Choice Voting proposition did was to strike that provision. At the same time as the voters were being told, in the voter information pamphlet, that a winner "would still be required to attain a majority."
This difference is crucial. In spite of being sold as a way to "obtain majority winners," there are very few examples where it has done this. In what I recall as 23 IRV elections in the U.S., since 2004, that went to the "instant runoff rounds," only maybe two found a majority. IRV isn't finding majorities. Runoffs find majorities, because there is a real ballot with real candidates, and the voters see a clear choice to make, and when they care, they turn out. Before IRV, top-two runoff was reversing the preference order for the top two, when the runoff took place. Now, after the "instant runoff," there are no examples of preference reversal, or is there even any change in which candidate is the runner-up. IRV results are, without exception (recently) the same as Plurality results. At huge expense.
Robert's Rules "preferential voting," as described in the book, would be quite different. In all these elections, in spite of the effort to avoid a runoff by using preferential voting, the effort failed and a runoff would be held. Would it be top-two? No, not if the Rules were followed. It would be a whole new election. However, top-two would satisfy the Robert's Rules requirement of a majority, in spite of all the specious arguments that they have and quite possibly will introduce. Robert's Rules is very strong in opposition to election by Plurality; but all organizations are independent, so Robert's Rules simply says that election by Plurality requires a bylaw for it to be legitimate. IRV, as it is named and as it is being implemented, is election by Plurality that, under certain apparently rare circumstances, has some better results.
The other change was that I added some references and some more language to the description of what is in Robert's Rules. I'll address this in my response to User:Tbouricius.
I can accept Abd's moving the mention of Robert's Rules of Order Newly Revised (RONR) out of the introduction, and limiting it to the section I added months ago on non-governmental use of IRV. However, his attempt to insert his favored "Bucklin" method into that section when RONR clearly does NOT describe it, but instead describes IRV is disingenuous. His attempt to say that RONR is not describing IRV, but only a "similar" method is also disingenuous. The IRV method can be, and IS, applied with a majority threshold defined as either a majority of non-exhausted ballots, or as a majority of ballots cast. If an office must be filled, and there is no provision for a new election, many implementations of IRV stop the count once a candidate has a majority of non-exhausted ballots (since the winner is known), although continuing the elimination process might have also shown the candidate to have a majority of all ballots originally cast. The distinction is irrelevant in most applications of IRV.
Tbouricius (talk) 17:15, 15 May 2008 (UTC)
For the benefit of newcomers to this discussion, there are reams of Talk on this, it is not out-of-the-blue as Richie claims above, but it's merely been that my attention was elsewhere, and the current AfD for the Controversies article brought my attention back here. And I found something that I somehow had missed before. I'd pointed out the majority election problem, and found Terrill's introduced IRV legislation in Vermont that acknowledged that the method can fail to find a majority, but the claim was still made that the method described did not require a majority. I based my opinion that it *did* require a majority simply on the basis of a general understanding of parliamentary law, and Robert's Rules general insistence that no decision should be made in a democratic organization without the consent of a majority voting. What I noticed, now, was that Robert's Rules clearly assumes a majority requirement. What it is describing is sequential elimination with a majority requirement.
But sequential elimination with a majority requirement is not Instant Runoff Voting, which, in most implementations (all?) has been sold as replacing runoffs by finding a majority without a runoff. And Robert's Rules is not assuming a top-two runoff, unless other candidates voluntarily withdraw. It's proposing a whole new election. What they are proposing is the use of a preferential ballot, they say that there are many forms, and then, for reasons I've described elsewhere, what they describe is sequential elimination. With a majority requirement. Given that IRV is not finding majorities, in actual elections, it would be avoiding less than ten percent of runoffs that would have taken place without it. IRV is avoiding runoffs by simply dropping the majority requirement.
Note that I'm not trying to put all this into the article. I can't. It's original research, unfortunately, as far as I know, though it is easily verifiable. What I'm doing, however, is insisting on accuracy in the article, an accuracy that I know is necessary as a result of this research. Bucklin voting was mentioned because (1) RRONR says that there are other forms of "preferential voting" and that it is only showing one form, and (2) Bucklin is an example, one that was known in the U.S. as "preferential voting." So, in fact, if we want to be very hard-nosed, which I don't, it could be claimed that RRONR doesn't mention IRV at all, to claim otherwise is synthetic. Are there any reliable sources that state that what Robert's Rules describes is "IRV"? I've certainly seen it said a lot, but not in reliable source.
Tell you what. We take out "resembles IRV" and simply put in what it says. RRONR doesn't say that the method resembles IRV. I simply don't want the article to say "Robert's Rules of Order describes IRV," (Much less what it used to say, when we started, that "Robert's Rules of Order recommends IRV." Find a way to link it with IRV without implying identity, and I'll be fine with it. I thought what I wrote was inoffensive. What's wrong with it? It's single transferable vote, as a method, used for single-winner. But what Robert's Rules is describing, they obviously assume from what I sourced, a majority is required, and thus it isn't "instant runoff voting," which, from clear usage, is a plurality method, that is, it will elect without having found a vote from a majority of voters voting and casting legitimate ballots in the election. Given that what is actually being sold as "Instant runoff voting" is two or three-rank preferential ballot, it's even worse, for there will be far more spoiled ballots. Consider San Francisco, with over twenty candidates on the ballot. But San Franciscans obviously thought that finding a majority was very important, so they were willing to bear with the cost and inconvenience of runoffs. Here come the salesmen who tell them that they can do it in a single ballot, and what is really weird, nobody, not even strong opponents, noticed the bait and switch. Goes to show. --Abd (talk) 00:05, 16 May 2008 (UTC)

Unless someone can explain to me why the Fijian House of Representatives is more important than Robert's Rules, I will restore the paragraph. Smithers Ltd. (talk) 17:24, 15 May 2008 (UTC)

Fine. I'll take out the Fijian House of Representatives.... since this apparent imbalance was the objection, no problem with that, eh? The issue is not importance, but POV balance. I have no attachment either way about the Fijian thing, but it seems to me that it does no harm. The RRONR reference *does* do harm. It's actually imbalanced, placed in the introduction. The editors who have just restored it are both major IRV activists, and they definitely want it there, because it gives the impression that RRONR approves of "IRV." The truth behind that is a lot more complex. It's already in the article, in detail, except where the same editors have attempted to remove details, so the mention in the introduction is purely political spin, attempting to call down for the readers the aura of authority of Robert's Rules. Which, quite clearly, *disapproves* of IRV except under certain narrow circumstances, and which does not recommend sequential elimination over other kinds of preferential voting. But as a manual of actual practice, not of prescription, and with the only *common* form of preferential voting being used being sequential elimination, that is the only one they were at liberty to describe. Then they warned about it.--Abd (talk) 23:05, 15 May 2008 (UTC)
You are trying to push the Bucklin method in this article, but you think your judgment is objective enough to suggest removing Robert's Rules, the most important parliamentary manual in the world, from the introduction? And when challenged you threaten to take more from the lead to which you previously did not object? I can not imagine a less convincing argument. I have removed references to Bucklin voting from the article, but I will replace it if you can name a single current use of it. Smithers Ltd. (talk) 04:13, 16 May 2008 (UTC)
The "threat" wasn't serious; the mention of Fiji was harmless at worst, and I said that, and that this was used to blame starts to ... stink. IRV is not used very much, compared to other systems, so every little application is notable! The issue isn't, mostly, about notability. The issue is accuracy. What is in an introduction should be impeccable and not controversial. FairVote has broadly publicized the idea that Robert's Rules of Order "recommends" IRV. Does it? Actually, Roberts Rules recommends and describes something critically different from what FairVote promotes, in almost every application. It might seem like a detail, until one realizes the implications and the fact that substituting IRV for top-two runoff, unless a majority remains to be required, not only can change results, it is changing results. Roberts Rules *never* recommends election by a mere plurality, and the method that they describe (they still don't like it, obviously) still requires a majority. So you cannot claim that Robert's Rules supports a campaign to replace top-two runoff -- which does, at least, require a majority -- with an IRV that is satisfied by plurality. Yet that is what has been implied by every mention of Robert's Rules in the introduction, except a few that were unacceptable to the IRV advocates. In the past, language which incorporated enough of what RRONR actually says has been proposed, and always rejected as too complicated for the introduction. But there is no necessity for this being in the introduction at all. It's notable, so it's in the article, in the necessary detail. Hence, contrary to what Smithers Ltd. has claimed, Tbouricius stated above he was willing to accept the removal. --Abd (talk) 04:58, 17 May 2008 (UTC)
Bucklin is relevant and is mentioned because (1) current use is important but not controlling. This article is about history and about potential use as well. (2) Robert's Rules of Order mentions preferential voting and that there are various forms of it. It gives single transferable vote, while still retaining a majority requirement, as an example. It does not mention "instant runoff voting." IRV as implemented does not match exactly what Robert's Rules describes, because all known public election applications of IRV -- except for one in Vermont that failed to pass the legislature, and I'd have to look at the current legislation there, it was just vetoed if I'm correct -- are plurality methods: they can elect without obtaining a majority of valid ballots cast. So, obvious question that I'd think some readers would ask: what are other forms of preferential voting? Given that there is a form which was used in the U.S., and which does have an article of its own and is clearly notable -- so the "current use issue" is a big red herring -- I've mentioned it. Just like the "IRV" we know here -- it's a neologism, by the way, created for political purpose -- Bucklin voting violates a principle of Robert's Rules, one that IRV may be considered to satisfy (Bucklin has more than one simultaneously active vote, as with any Condorcet method) -- but does not break one that IRV does break: (1) Bucklin is more efficient at finding a majority; this can be seen by Bucklin-analyzing a recent set of San Francisco elections which did provide enough data (most don't), and (2) Bucklin does not create a false majority by eliminating candidates and exhausting ballots. Am I promoting Bucklin in this article? Mention, when relevant, is not promotion. IRV proponents, however, may think it is. Bucklin does have another obvious advantage, though. Because it is precinct-summable, it is far easier to count. Just count all the votes and add them up. Further, this article is named "instant runoff-voting," and the name is justified because the counting proceeds in rounds kinda like a runoff. Bucklin counting proceeds in rounds also. If "instant-runoff voting" were a generic term, it would apply to Bucklin. Current debate over the legality of IRV in Minnesota is based on a court decision involving Bucklin (So far, official legal advice, I think it was the City Attorney of Minneapolis was that Brown v. Smallwood applies to IRV, and contrary opinion has been issued by attornies affiliated with FairVote.) Put all of this together, an article here with this topic that does not at least mention Bucklin is POV-imbalanced. So ... there will be a few more edits to the article, I expect!
By the way, there is extensive argument in the current AfD that the Controversies article should be merged with this one. If so, or if that article is straight deleted, I'd expect to see more about other methods here, since much of the controversy is related to comparison of IRV with other possible reforms. That article was created as a way of avoiding contentious debate here, and it worked. Consensus of editors, folks.--Abd (talk) 20:56, 16 May 2008 (UTC)

You certainly have a lot to say. Abd.

Factually mentioning that IRV is used extensively in NGO elections seems important to mention in the opening, and referencing that it is detailed as an option in Robert's Rules (which is simply fact) is perhaps a way to move forward. See those edits. RRichie (talk) 12:37, 17 May 2008 (UTC)

The problem isn't notability, though the term "many" is POV and subjective and, in fact, unsourced. Yes, I have a lot to say, the question isn't reducible to the sound bites Richie has crafted. Though I've got one: Count All the Votes. IRV doesn't. Bucklin does, so does Approval and any Condorcet method. Is IRV "detailed as an option" in Robert's Rules? Fine. If an organization simply adopts what Robert's Rules described, they have a much better method than what Richie has promoted and succeeded in getting implemented around the U.S. The method still requires a majority. Because Richie promoted the name "Instant Runoff Voting" and sold the method as replacing runoffs -- which the RRONR method does not, it merely makes them less likely -- he's kind of stuck. What Robert's Rules describes isn't what he's selling. Close, but no cigar. And dropping the majority requirement is not a detail to the authors of RRONR. It's so important that they assumed it, which is why they didn't make it crystal clear, and why the IRV people were able to argue, as Richie et al did, that the "majority" in the IRV description was a final round majority, not a true majority. They assumed, in their turn, that RRONR had their motive: replacing runoffs. It didn't. IRV satisfies the Majority Criterion, but so does Plurality, so when Bouricius has noted that IRV is considered a majoritarian method, it's correct. But so is Plurality. And when RRONR, thus, says that "preferential voting" is better than election by a Plurality, it's assuming that the election fails if there is no majority. As they say, the devil is in the details.
In other words, it is not a "fact" that "IRV" is detailed as an option by Robert's Rules. It is an interpretation, and a misleading one unless properly qualified. What Robert's Rules described and what Richie's organization has promoted are different. In a way that is actually crucial. But because this is easily overlooked, and it has been overlooked, for surely the opponents of IRV in San Francisco would have used this and they did not, Richie has been able to score a propaganda coup by having this claim in the introduction of this article, and it's been copied all over the internet. Sorry, but it's time the lollipop is taken away. It's not good for you. If Robert's Rules is going to be mentioned in the introduction, the mention must be complete.
We've been around this carousel before. I'd take it out, they would put it back. I'd add language, a few words, to make it accurate, to mention that what Robert's Rules "describes" requires a majority, and they would take that out, always trying to leave behind the impression of a recommendation even though they did allow the removal of that word. You want an example of POV-pushing, there it is. *Who* pushed for this? Well, most strongly, Richie himself, I think originally as an IP editor freely violating edit warring rules until he was blocked. (Look at article history, starting when my edits started to hit this article, and all this has been detailed in article Talk.) Then there were the socks, BenB4, Acct4, and MilesAgain, all of them shown ultimately to be socks of Nrcprm2026, also known as James Salsman, a banned editor and known, outside, to be an IRV activist. Now there is Smithers_Ltd., who is an SPA, apparently. A high percentage of SPAs turn out to be socks, particularly if they dive into edit warring as he did. And then there are *two* legitimate editors that are also arguably WP:COI. Tbouricius is clearly COI, he is a published co-author of a paper with Richie and has been associated with FairVote, officially, but I considered his presence here important and, when he was blocked for acting like an SPA and apparently cooperating with the socks, and a possible "meat puppet," I intervened to help him be unblocked, and he has proven to be capable of compromise and is not the tendentious promoter that the previous editors were. I continue to welcome his work with this article. Mr. Bouricus has consented to the removal of the mention of RRONR from the Introduction, above. And finally, User:Tomruen is less obviously COI. He has described himself as associated with FairVote, but as a volunteer, and on the main topic here, the inclusion of RRONR in the introduction, he's gone back and forth and is clearly a real Wikipedian. I think he also would allow its removal.
And who is on the other side? Well, myself, of course, and I'm known as a critic of IRV, but I'm not formally associated with the Center for Range Voting as has been claimed, I'm more distant from them than Tom Ruen is from FairVote, plus there is nearly every other passer-by, people who have commented in the RfC. I'm not arguing here that my position has consensus, such a determination is debatable. But the position that the mention of RRONR belongs in the introduction has no consensus, if socks and SPAs and COI editors are excluded, and introductions should be rigorously neutral and a matter of consensus. The whole article should be a matter of consensus, in fact, and that is what we are working on, most of us. I actually agree that the mention of preferential voting in Robert's Rules is quite notable, but the *whole* mention is notable, not what is POV-extracted from it. RRONR, in fact, gives one of the most notable criticisms of STV single-winner: IRV's vulnerability to center squeeze, which is related to its failure to satisfy the Condorcet criterion. I'd be quite happy, as a critic of IRV, who has come to the conclusion that it is a step backwards from top-two runoff, to have a balanced mention in the introduction, but to do so in summary style will be, I suspect, quite difficult. We tried before, and I'll try again, here.--Abd (talk) 14:02, 17 May 2008 (UTC)

[edit] reverts to article

I made several reverts to yesterday's edits of the article. This is an explanation.

(1) Terrill Bouricius removed specific references from the Robert's Rules section, and restored language that increases the perception that Robert's Rules is "recommending" -- the original language used in the article and in FairVote publicity elsewhere -- Instant Runoff Voting. I reverted it. In fact, it recommends "Preferential voting," and notes that there are many forms. Mention of at least one would seem useful to the readers. I'm not attached to it being Bucklin, but Bucklin is particularly similar to IRV in that it uses the same three-slot ballot as many implementations of IRV, and the counting proceeds in rounds. Plus it was actually used in the U.S., so it is historically relevant. It was actually called "preferential voting." And that was sourced. Don't remove sourced material unless it is clearly *irrelevant*. At the same time as FairVote-affiliated editors have claimed I was being pedantic or insisting on too much detail, some have been willing to edit war over seemingly inconsequential wording. I must infer that it is not inconsequential, that there is a desired impression that depends on exact details of wording. In the presence of this possibility, I insist on accuracy and faithfulness to source. Which includes being complete. Is what I inserted incorrect? Does it distort the source? The mention of Bucklin isn't crucial to this, but it would seem appropriate to explain the source. Perhaps, quite simply, we can link to Preferential voting? I fact, I think that better, I've made it so.

(2) Smithers Ltd. restored the mention of Robert's Rules of Order in the introduction. Introductory text is very sensitive, and should reflect the highest possible degree of consensus among the editors. We've been over and over this before, and there has been RfC on it, and no consensus was ever obtained that it should be in the introduction. And those arguing for it being in the introduction included, I think, Tbouricius (a consultant to FairVote and published author on the subject, a legislator who introduced IRV legislation in Vermont), and User:MilesAgain, one more sock from the drawer than has been a problem here for a long time. He's been blocked. That text, in my opinion, is utterly inappropriate for the introduction; it's impossible to state it in sufficient detail that it becomes NPOV without making it overlong and thus imbalanced. So I reverted this as well.

(3) Likewise, Smithers Ltd. removed the entire section on Bucklin voting, a subsection of "similar systems," giving the edit summary: (delete Bucklin method text, because it has been inserted here, as far as I can tell, only for the purposes of promoting it by someone who has been promoting it over months of talk page history). The alleged motives of editors may be of interest, but it is quite possible that a good edit is made by someone who has some ulterior motive. Removing text can never be justified solely based on speculation as to motives, nor even on proven motive; removal must be based on Wikipedia guidelines or policy. Removing only Bucklin, with no discussion of whether or not it is a "similar system," is, in fact, POV. The question would be notability, and Bucklin is quite notable because of its historical usage and its current relevance because of the legal precedent of Brown v. Smallwood. (Bucklin was outlawed in Minnesota (under the name "preferential voting," and there is legal opinion (from higher courts and from a City Attorney asked for his opinion in Minneapolis) that this opinion applied to sequential elimination as well (which was also called "preferential voting.)" FairVote attornies have opined the opposite. This alone shows that IRV and Bucklin are considered, in some ways, to be related. They are also different, and the FairVote opinion was based on such a difference, rooted in a single sentence in the court's decision. The rest of that decision, in my opinion, makes it clear that what the court was considering unconstitutional was, indeed, preferential voting *of all kinds,* carried out with a single ballot. My opinion. But pretty well grounded and actually quite explicit if you read the whole decision without FairVote tinted glasses. I reverted. I'm not thrilled at an SPA, newly registered, who edits this article as his first edit, being quite so bold, given the history. Special:Contributions/Smithers Ltd. He is welcome, of course, until proven otherwise. Please do understand that I edit as if ArbComm is looking over my shoulder. --Abd (talk) 21:34, 16 May 2008 (UTC)

Given your history with Bucklin, which you try to push in this article, over and over again, why do you think ArbComm or anyone else could possibly approve? Wikipedia is WP:NOT a soapbox. The Roberts Rules reference is so obviously fit for the introduction -- is there anyone at all besides you who wants to take it out? It's been there for months and months, with lots of people refining it, and one person, you, Abd, deleting it. It clearly has the consensus you say it should have and you clearly should be prohibited from editing an article that you have repeatedly used to try to advance a different method of voting that nobody uses. Smithers Ltd. (talk) 03:13, 17 May 2008 (UTC)
There was an RfC on the topic of RR in the introduction.[8], and another at [9]. Maybe Smithers Ltd. should read them. By the way, the most contentious arguments in the RfCs, and the RfCs themselves, came from User:MilesAgain, later blocked as a sock puppet continuing the same work as User:BenB4 and User:Acct4. If you look at who has written as you have, Smithers Ltd., you may see that you are in some bad company. As to removing something that many worked on, Smithers Ltd. removed the whole subsection on Bucklin from [[Instant-runoff voting#Similar systems|Similar systems]}that had been worked on by many editors, and that represented some level of consensus. Editors who worked on this section were User:Tbouricius, who just removed the Bucklin mention from the detailed section on RRONR with the comment that Bucklin "has its own section," which becomes untrue with Smithers Ltd.'s removal, and who had worked on that section with [10], . User:Tomruen, who has been a FairVote activist, likewise worked on the section with [11][12], and User:RRichie himself did the same with [13], and there is more, but perhaps you get my point. Bouricius summed it up pretty clearly, acknowledging that Bucklin is indeed a "similar system," with [[14]]. I created the mention of Bucklin as a "similar system" with [15], and I added language to it with [16], in September 2007. And what I wrote was pure fact, well-known, except for my calling it "Instant Runoff Approval Voting," which, though true, was certainly a neologism. That was, in fact, one of my first edits to the IRV article. And it, overall, was not controversial. Details have been and may continue to be. So, the pot claims that the kettle quacks. I'm reverting Smithers Ltd.'s radically unjustified and massive edit; given that his arguments hold no water at all. (1) It's been discussed, and it is clearly controversial to have the mention of Robert's Rules in the introduction, and (2) Bucklin is acknowledged as a similar system and therefore belongs in that section. The argument that it is not in use is totally irrelevant; there may have been more Bucklin elections in the U.S. than IRV elections, and Wikipedia is not just about current affairs. And even if no elections had been held, but there was reliable source about the method, it would belong. --Abd (talk) 04:35, 17 May 2008 (UTC)
Please answer the question: Does anyone agree with you that Robert's Rules should not be in the lead? If not, why do you call including it controversial? Since you have been using this article to push your activist agenda of another voting method, which, you admit, is not used by any adult government agency, notable business, and has very few mentions in the literature, are there any reasons that the community should not restrict you from editing this article? If so, please state them. Smithers Ltd. (talk) 23:27, 17 May 2008 (UTC)
I didn't answer directly, I referred to the answer. The question, however, isn't whether or not RRONR can be in the lead. The question was always whether or not particular summarized text can be there. I'm fine with the way it is now. Thanks for moving it up, Smithers, prominence as it is is fine with me, as long as it's not misleading. Now, as to those who supported removing it -- or expanding it to make it accurate -- there would be meamemg, [17]. The !votes in the RfC raised at the end of December, 2007, on this issue were two comments that it's a problem, two comments that it's okay, and one abstention. But one comment that it was okay was MilesAgain, an abusive sock puppet and the other was Tbouricius, with a fairly strong COI. So even if we count the sock puppet and Bouricius, it was clearly controversial. In the second RfC at the end of January, User:Rrius saw it as a problem in the introduction.[18].
What voting method am I pushing using this article? Is there any inappropriate mention of other systems still in this article? I consider top-two runoff better than IRV and IRV slightly better than plurality but much more expensive. (IRV is probably less expensive than actual runoffs, though that is actually controversial and may not be true. San Francisco would be an odd case because it had extraordinary numbers of candidates, and therefore lots of runoffs, but they'd have gotten exactly the same election results with Plurality, quite likely, all the extra counting did not shift candidate positions at all (among the top two). It is quite likely that they are getting worse results with IRV than with top-two. Why? Top-two runoff produces "comebacks" about one-third of the time, looking at San Francisco and Cary, NC. I.e., the runner-up in the primary wins the runoff. IRV is producing *no* comebacks. And the candidate in second place *remains* in second place, that position is not being usurped by third place. What this means to me is that, in some cases, a candidate who would have lost the runoff is winning the election. FairVote has arguments why this is a good thing, but they are essentially made-up and don't consider one important fact: the allegedly lower turnout in runoffs (they are not always lower, it depends) just might be because some voters think both candidates good or both of them bad. Turnout in comeback runoffs, according to FairVote, was higher than usual for runoffs. Quite likely, the additional much more focused campaigning makes a difference too. You will notice that critics of IRV claim that IRV preserves two-party systems, and that just may be because it generally produces results quite similar to Plurality. There are occasional exceptions. (Ann Arbor, MI, where preferential voting was used once in the seventies.)--Abd (talk) 03:48, 18 May 2008 (UTC)
Just so Abd's statements aren't seen as realities, IRV has "comebacks" when a majority support the second-place finisher over the first-place finisher. This certainly happens in major races -- examples include the 1991 presidential race in Ireland, where Mary Robinson became the first woman president after finishing well behind on the first count. Similarly, in Ann Arbor's IRV election in 1975, a candidate won who trailed 49% to 40% on the first count, but was the second choice of nearly every third party voter in that election. It happens in Australia in every election. It has come closer to happening in some of the more recent U.S. uses of IRV; it will happen when there's been a case of vote-splitting that IRV heals.
Abd supports runoffs over IRV, but small percentages of voters do -- indeed EVERY single demographic of voters in San Francisco (as measured by ideology, party,age, race, neighborhood, etc) supported IRV over runoffs in the University of San Francisco exit polls after the first IRV election -- consistent with other exit poll surveys and the various ballot measures with IRV defeating runoffs. Traditional delayed runoffs lso exacerbate campaign costs and, at least in certain elections like the Board of Supervisor races in San Francisco, came with huge drops in turnout in the second round. They also create opportunities for special interest money to pour in -- in San Francisco, independent expenditures soared in runoffs, generally spent in negative ads against candidates. That's a certain kind of "deliberation" that a lot of voters didn't like.
And note that in San Francisco, the first-choice candidate in the first round of runoffs overwhelmingly won in he runoff in the old system, just has been happening with IRV.
Finally, much of what concerns Abd is the elimination of "compromise" candidates. Two-round runoffs are more likely to do that than sequential IRV -- a good example being the French presidential race in 2002.
RRichie (talk) 13:56, 18 May 2008 (UTC)

(unindent) Richie wrote: Just so Abd's statements aren't seen as realities, IRV has "comebacks" when a majority support the second-place finisher over the first-place finisher. Sure. It can have such. I wrote that it wasn't having such, in all the IRV elections in the U.S. held since San Francisco adopted RCV in 2002, first used in 2004. Considering only the elections where "instant runoff" took place, there were 9 such elections through 2007. (I think that above I mis-remembered this as 23 elections, that may be the total of IRV elections in that period.) I previously reported in Talk for this article that results from a set of prior top-two runoff elections (held in San Francisco and Cary) where I found one out of three "comebacks." FairVote reports, in federal top-two elections,[19], 29.17% (28/104). Pretty close. With nine elections, we'd expect to see a comeback, in fact, we'd expect to see two or three or four. Not one is significant. In any case, I'm not claiming proof here. I'm claiming, instead, that it is quite reasonable to surmise that IRV is not simply reproducing the results of top-two runoff without the expense and trouble. It is changing the results to those more consistent with Plurality. Absolutely, under certain special conditions, IRV may produce a better result than Plurality. But it is not doing what it was sold to do: find majorities, and, in the name of "finding majority winners without the expense of runoffs" it is abandoning the majority election concept.

Richie introduces, as he usually does, a host of irrelevant arguments, that is, irrelevant to the point in front of us. Popularity isn't an issue here. The issue is how IRV behaves, in actual elections, compared to what it is replacing. Richie refers to two elections where there was a comeback. One of them was in Ann Arbor, a very special case: a strong third party that controlled the City Council for a short time. For some strange reason, that was missing from this article until I added it.

Now, something that is a new realization for me; as with much of this, it's been right in front of us. IRV was used in the U.S. before, see IRV implementations in United States#1912 Florida, Indiana, Maryland, and Minnesota: for party primaries. What was it replaced by? According to that article section, by, among other reforms, top-two runoff. Unlike IRV, top-two runoff can really be said to guarantee a majority winner. Apparently that goal has been considered, historically, worth the expense. As has been described elsewhere, if there are three equally liked candidates (in first preference), IRV eliminates a candidate, and roughly one-third of the time, this would be the true favorite in pairwise analysis. If the party members care about the difference, they could be expected, in a real runoff, to choose that candidate. And in real runoffs, we do see a comeback one time out of three. Because this candidate does in fact have broader support in the party, the candidate could be expected to, on average, go on to win the ultimate election more often than not.

But the section claims that IRV was abandoned because of a series of elections where the additional preferences did not make a difference. That was not a coincidence. From what we have seen recently, that is quite normal, and it appears that it happens more often with IRV than with top-two runoff. If top-two runoff produced no changes, it would quickly be abandoned because of the expense. But one time out of three it does change the result.

Richie then turns to the "compromise candidates" issue. While this is certainly related, I certainly would not claim that top-two runoff can't show this problem as well. Indeed, top-two and IRV share the particular problem now mentioned. Both methods can eliminate the wrong candidate, basically. However, when this happens with top-two, the electorate becomes focused, and, at least, makes sure that it picks the best of the two remaining. IRV doesn't necessarily do that. In Australia, they avoid part of the problem by forcing all voters to vote a full preference list, and I would expect, there, some reversals to happen because of donkey voting (just going down the list and marking preferences in sequence) which, though it's ameliorated by Robson rotation, still introduces noise; more significantly, in the U.S. marking a ballot for a candidate indicates support of some kind. When votes are coerced, it may indicate choosing between absolutely terrible, I'm leaving the country, and even worse, I'm buying my ticket if first results show this candidate as coming close to winning. Or it may indicate simple ignorance, that name sounded nicer to me.

Most of this is dicta. The present issues are (1) What does Robert's Rules describe? And what has previously been overlooked or suppressed from this article is that Robert's Rules, first of all, still requires a majority. Sure, in *one* proposed implementation in the U.S., the majority requirement wasn't dropped with IRV. And that was clearly because they *couldn't* drop it without a constitutional amendment. Here, Richie et al argued that, no, Robert's Rules did not require a majority, a majority of final round ballots would suffice, and that all this was not important, a trivial detail. Get it out of here. We have to consider how the Wikipedia article was being used. The mention of Robert's Rules, here, and on the FairVote web site, was being used to sell "IRV" as being recommended by Robert's Rules, but as a replacement for top-two runoff, with no majority requirement. I.e., Robert's Rules good name was being used as if it were a recommendation of IRV, when a careful reading shows quite the opposite, once the details are considered. For a time we settled on language that "Robert's rules describes...." but that remains misleading unless the critical difference is noted: what Robert's Rules describes is different from every proposed IRV implementation in the U.S., and if the Robert's Rules form were actually implemented, runoffs would, for the most part, not be avoided. Just look at the elections. Out of the nine runoff elections described above, I think it is seven that would still have gone to runoff.

And then comes other methods. Bucklin, for example, would have avoided several of the runoffs that would still have been required by IRV. It also fixes the spoiler effect. Approval does the latter, but Bucklin preserves the full Majority criterion compliance that most think important. (Approval technically fails the criterion, according to most reports -- it depends on some subtleties of definition -- but that failure, in real elections, would be very rare.) Bucklin is far, far easier to count than IRV. This isn't an attempt to promote Bucklin, it's simply a comparison, pointing out that there are methods that Robert's Rules would more solidly recommend. Bucklin is far more likely to find that compromise winner, it does not only if strategic voting conceals the lower preferences. In which case there would be a real runoff! And the compromise candidate would almost certainly be in the top two. This is because, unlike IRV, Bucklin does not conceal votes. The same characteristic also makes Bucklin find more majority winners. So if one wants to reduce runoffs, Bucklin would be an obvious choice, better than IRV, it *actually* does more of what IRV claims to do. So why isn't it being considered? In a word, FairVote had a different strategy, it's really trying to pave the way for proportional representation using STV, which was its original goal. A noble goal, in fact. If FairVote, which used to be called the Center for Voting and Democracy, hadn't become a pure political action machine, dedicated to a clear goal, not responsive to broader expert opinion, using the standard political tricks of carefully-designed spin, I'd actually be on their side. But I can't abide by deceptive spin. Especially on Wikipedia. And Richie worked long and hard to maintain it here. Don't know the history? I'll explain it if asked. --Abd (talk) 17:27, 18 May 2008 (UTC)


No, Abd, the reason Bucklin isn't used anywhere is not because of the magical powers of FairVote, but it's because it violates the "later no harm" criterion. For people who understand politics, violating that criteria is a huge problem. Whenever you use a system where indicating a preference or giving points to a lesser choice can help defeat your higher choice, you have a system that is unlikely to last in competitive elections -- which indeed has been the case, with Dartmouth the latest to likely toss approval voting among the very few places trying it in recent years despite some zealous advocacy and high-profile press for it. It's also going to be very hard to persuade politically savvy people like elected officials and charter commissions to do it because they understand the psychological implications of that problem.

If you think Robert's Rules is really more likely to support Bucklin over IRV, why don't they give Bucklin as an example? They don't. They describe IRV. And they describe IRV in a way that clearly generates winners in one count -- just like wth IRV. It would be a lot easier for Robert's Rules to describe Bucklin, but they don't. It's no accident. For those new to this debate, see the whole text posted here: http://www.fairvote.org/irv/robertsrules.htm

The Robert's Rules issue has had probably zero impact on whether IRV has been placed on the ballot in all the places doing it in recent years. I oppose you on this because I think you're simply wrong. The language to me TRANSPARENTLY is saying IRV is a good system to use when repeated ballot isn't possible. And you say otherwise. I think any objective reader would agree with me. So it's rather irritating to see you keep after this with such persistence. Some others here like Terry Bouricius have more patience with you and try to achieve compromise, but at the end of the day, let me repeat that I think any objective reader of the Robert's Rules section would agree with me and disagree with you.

So... at the very least I'm going to oppose you on your trying to edit the intro of the IRV article to suggest the Robert's Rules thinks that Bucklin is better than IRV. That's not at unsupported by the text of Robert's Rules. It is VERY clear about it being repeated ballot -- with new votes based on seeing how people did in the first round -- that is the key to being able to move toward a compromise voting. RRichie (talk) 01:10, 19 May 2008 (UTC)

There is no evidence that Bucklin was abandoned because of "later no harm." Propaganda to that effect is generally traceable to FairVote. Brown v. Smallwood makes an offhand comment, dicta actually, that can be interpreted that way, but the core of BvS was that the voter was allowed alternative votes. Period. Brown v. Smallwood, if you actually read the whole decision and the rehearing, clearly would have also outlawed IRV on exactly the same basis. Legal opinion was sought by the Minneapolis city government on this, and it confirms what I'm saying. However, FairVote supplied legal opinion from affiliated lawyers and apparently swayed the City Council, hence the matter is presumably going to be heard in Minnesota courts. Bucklin was abandoned elsewhere, according to some reports (it's hard to find any sources on this that aren't FairVote) because few voters were using the additional ranks. But IRV was likewise abandoned for the same reason, in the same environment: primaries. Anyway, much of this is repetitive. Richie has made the same claims before, and repeats them now as if they were new arguments. Does it really matter? Am I insisting on Bucklin having something to do with Robert's Rules? That was taken out and, since it wasn't crucial to anything, it was merely a mention of an alternative form of preferential voting, and Robert's Rules states there are many, one that was actually used in the U.S. So why didn't Roberts Rules mention another method? There is an obvious reason: they had space for one, and they are not a prescriptive manual, leading the field with new ideas, they are a descriptive manual, a compendium of actual practice, and the method they described is actually in use. However, since they dislike any election or choice not by a majority vote, they were able to describe a majority form of IRV. They still don't like IRV, or they wouldn't take pains to note the problem with Center Squeeze (which is really about Condorcet failure). However, if you must use IRV, by all means, still require a majority! And this, then, explains their comment that "preference voting" is more fair than election by plurality. As IRV is being implemented, sometimes helped along by the alleged recommendation of Roberts Rules, it is being shown that, generally, IRV elects by plurality. And plurality only sometimes elects by plurality! More often, a majority is found.... IRV, in the Robert's Rules form, reduces runoffs but does not eliminate them, and so it really cannot be called "instant runoff voting." It is, rather, "preferential voting," a method of finding majorities more deeply than vote-for-one plurality. And there are other methods that do a better job of this, that don't suffer from Center Squeeze. I can't say in the article that Robert's Rules would prefer one of these, because there is no reliable source for it. But it's pretty obvious. And just like I can't put that in the article, they can't put it in the manual. Similar problem. Not based on significant actual practice.

There have been a number of editors read the debate we had and agree with me. The *only* editors who have come to a different conclusion (if I except the most recent editor whose position is a bit mysterious, he seems to have imagined I was promoting Bucklin here and was otherwise unfamiliar with the history) have been FairVote editors. Sure, if you read Robert's Rules, especially as framed by FairVote, it's very easy to come to the conclusion Richie claims is obvious. I held the same opinion myself for about a year. I was a bit upset with the editors of RRONR, why did they "recommend" such a bad method as IRV? But then I read it more carefully. Any parliamentarian who implements what they have written will require a majority. And it turns out this is obvious from the text. I actually missed it even after I understood the problem and had come to the conclusion they were requiring a majority, but I thought they had been vague about it. I now wonder how in the world I missed it. It's blatant. And FairVote editors argued again and again that a majority wasn't being required.

It's because they are thinking about "instant runoff voting." For them, the whole point is avoiding runoffs, doing it in one election. But just as the public in many places has considered an extra election worthwhile to avoid election by plurality, so does Robert's Rules. If governments can do it, so can private organizations. Quite simply, it is not that expensive. Richie harps on "repeated balloting." Yes, RRONR prefers that, as do I. This is not runoff elections, this is repeated elections. No eliminations, period. And they know this can go on and on. But usually it does not. Preferential voting does not intrinsically discard the majority requirement, but it will, presumably, make repeated balloting less necessary. And repeated balloting is exactly what they continue to think standard process, they do not refer to any elimination if there is majority failure, they state that the election will have to be repeated. Repeated balloting. Being described as "fairer than Plurality." Plurality is what organizations often implement in order to avoid runoffs. They are essentially saying, instead of going to Plurality, which violates basic democratic principles, become more efficient at finding a majority when there are more than two candidates, use a form of preferential voting. Here's one. But they absolutely are not going to recommend that election by Plurality be done, and IRV without a majority requirement is a plurality method.

In any case, let me quote from Roberts Rules as to my claim that it requires the election to be repeated if there is majority failure. It was in a section where they are talking about the education of voters, and they make the claim about later-no-harm (which is actually an error if there is a majority requirement, but that's not controlling here, it's rare that it would have this effect), saying, "Sometimes, for instance, voters decline to indicate a second or other choice, mistakenly believing that such a course increases the chances of their first choice. In fact, it may prevent any candidate from receiving a majority and require the voting to be repeated." This makes it very clear that a true majority is being required, otherwise, there would be no such effect. I believed that already to be true, from the description of the process, which does not describe excluding any ballots from the basis for a majority, and elsewhere in the book the standard basis for majority is described. If it were being changed, they'd have been explicit. But this was synthetic and absolutely, the FairVote editors, some of whom have been fine wikilawyers (particularly the socks), wouldn't have allowed it.

"Preferential voting" (any kind) will tend to avoid runoffs. IRV as a form is actually not at all the best at performing this service. Bucklin analysis of San Francisco elections -- sufficient data exists for some of them -- shows majorities found in more of the elections, this is because Bucklin counts all the votes, it doesn't eliminate any. However, there are only three ways to eliminate runoffs: (1) limit the number of candidates to two (this is what some runoffs do). (2) force voters to rank all candidates (Australia). (3) eliminate the majority requirement. While RCV in San Francisco was sold as "candidates will still be required to gain a majority of votes," the Proposition actually struck the majority requirement from the election code. If it "still required a majority," why did they strike that requirement? I'm sitting here, still, scratching my head wondering why the opponents of RCV in San Francisco managed to overlook that, but they did. Apparently, it's pretty easy to overlook. That's changing, Rob. You can't put the cat back in the bag.

As to the claim that Robert's Rules wasn't of any influence, well, I've learned that I can't take Richie's word for anything, unfortunately. Here is what I know: the claim is repeated all over the internet, using the language used by FairVote and here. It is repeated by city clerks in official city web sites. However, to me, the real clue is that FairVote editors have been willing to edit war over what to others must look like an insane level of attention to exact wording, consistently removing more accurate language from Robert's Rules as "too confusing," "too much detail." And putting disputed text back into the lead. Richie claims that his only interest here is accuracy? Yet there is plenty of sloppy language elsewhere in the article. This was very important to him. Why? It's obvious.

As to Bucklin in the introduction, there is no mention of Bucklin in the introduction, nor have I any plans to put one there. I added a mention of Bucklin to the detailed discussion deeper in the article, several days ago, simply as an example of preferential voting that doesn't show the problem RRONR mentions for the STV method. That was reverted out by someone who thought I was trying to promote Bucklin by it, which was not the case. If RRONR mentions that there are many other forms of preferential voting, I simply thought it relevant to mention one that has actually been used for public elections. It's arguable, though, that since this section is about NGOs, and I'm not aware of any application of Bucklin other than for public elections, maybe a Condorcet method would be better, since such is in actual use. --Abd (talk) 03:15, 19 May 2008 (UTC)

By the way, here is why Richie says: The language to me TRANSPARENTLY is saying IRV is a good system to use when repeated ballot isn't possible. And you say otherwise. The following is the language from RRONR that leads him to think it so obvious:

While it is more complicated than other methods of voting in common use and is not a substitute for the normal procedure of repeated balloting until a majority is obtained, preferential voting is especially useful and fair in an election by mail if it is impractical to take more than one ballot. In such cases it makes possible a more representative result than under a rule that a plurality shall elect. It can be used only if expressly authorized in the bylaws.

First of all, this doesn't mention IRV. It mentions "preferential voting." So it is making the statement about a class of methods, not just one, and it gives one "by way of illustration," which is the STV method, which, as you know, can require a majority or not. Now, what it says that will easily be taken as implying that the method will complete in one ballot is this: "... if it is impractical to take more than one ballot." This paragraph, indeed, can certainly be read as terminating the election with one ballot. Yet the later language clearly recognizes that the election can fail. The only way I find that I can interpret it all together is that "impractical" is not the same as "impossible." If, indeed, it is truly impossible, the organization can decide, in the bylaws, that a preferential voting plurality shall stand as a victory. That would be better than a simple vote-for-one plurality. But what they actually describe requires a majority, and not a "last-round" majority, as FairVote activists argued was meant by "majority." So I conclude that the best interpretation is the consistent one, that doesn't make one paragraph or another into a mistake. They are not suggesting that the majority requirement be abandoned, they are suggesting a middle ground: make it less likely that you will encounter majority failure. Use preferential voting. And repeat the election if those voters refuse to rank additional choices. After all, if they prefer to avoid voting again, they can rank all the candidates, that is essentially a vote against a repeat. --Abd (talk) 03:33, 19 May 2008 (UTC)

[edit] Edit warring by COI editor

I'm not filing any formal complaint at this time, nor am I adding a warning to User talk:RRichie, but edit warring is a blockable offense for a COI editor. RRichie clearly fits WP:COI, he is the paid Executive Director of FairVote, Rob Richie. According to WP:COI he should avoid making any edits which are reasonably expected to be controversial; he may, instead, suggest such edits in Talk, which he may generally edit as any other editor. With [20], I took out the following text in the "Terminology" section:

Observers described the Democratic Party caucuses in Iowa as like "instant runoff voting" because backers of minor candidates could move to a viable candidate after the first tally at a caucus.[1]

My edit summary was: "This really isn't relevant to instant runoff voting, and the source didn't say exactly what was claimed."

Richie reverted it[21] with "(Undid revision 213291056 by Abd (talk)sure it's relevant to instant runoff voting -- which is why term used in this context)"

This is what the source says:

"For the Democrats, the caucuses are not about giving raw popularity numbers, they are about forcing people to make compromises and play politics," he said. "In a way, they accomplish the kinds of things that instant runoff voting accomplishes, but do so through social mechanisms instead of mathematical computation."
Instant runoff voting, which exists in some cites and states, ranks candidates. If there's no clear winner with more than 50 percent, then candidates with the fewest votes get taken out of the mix. When that happens, their supporters' second choice gets counted for one of the remaining candidates until someone has a majority. Iowa's Democratic caucuses are like that. If a candidate does not get 15 percent of the vote, that candidate's people must vote for someone else. That compromise -- that part of the process -- is what is valuable for the rest of the country.

When I wrote that the source did not support what the text said, I was referring to two things. First of all, the source did not say that the caucuses were "like instant runoff voting." It said that "in a way, they accomplish the kinds of things that instant runoff voting accomplishes, but through social mechanisms instead of mathematical computation." Saying that a hammer accomplishes the same goal as a screwdriver (fastening things) isn't the same as saying "a hammer is like a screwdriver."

The second problem is that this was not "observers," this was one person, Doug Jones. He's a professor of computer science, his specialty is verified voting. I'm sure Richie has met him. He appears to be a supporter of instant runoff voting, which is somewhat unusual in his field.

Saying that a process which is much more like top-two runoff -- in common use in the U.S. for some primaries -- is like instant runoff voting is pretty naive. The Iowas caucuses are not top-two runoff, they are more like what Robert's Rules recommends. Repeated balloting. No absolute eliminations based on first choices. What are being elected are delegates, but the "voting" is the formation of affinity groups, very flexible. It is a proportional representation system, and much more like Asset Voting than IRV. But the term "like" is pretty fluid. I could say that I'm like a tuba because we both have weight. The Iowa caucus is a deliberative system for choosing delegates, radically unlike IRV.

But, hey, if a non-COI editor thinks that the Iowa caucuses are like IRV, or that there is notable opinion there to report, I could certainly consider standing aside. But there was another editor who objected to this text as frivolous, and I noticed it the other day and finally got around to taking it out. I was actually a bit surprised that Richie was so bold as to revert me on this one. --Abd (talk) 01:46, 19 May 2008 (UTC)

And more, I see now. User:RRichie took out the Roberts Rules text from the lead, http://en.wikipedia.org/w/index.php?title=Instant-runoff_voting&diff=213372575&oldid=213370065, that had been accepted by another user and moved to a more prominent position, with the summary: (I suggest better to take out Robert's Rules until some other editor can say whether Abd Lomax is misreading the Robert's Rules text, as discussed in talk).

I find this fascinating and utterly unsurprising. I had been taking the mention of Robert's Rules out of the lead for a long time, claiming that, if sufficiently detailed, it was too much text for the intro, and if brief, it was misleading, and IRV advocates had kept putting it back. So I did what I'd done before. Okay, if you want it in the intro, I'll make it clear and accurate, and, this time, it was pretty brief. So now Richie wants to take it out until "some other editor" can confirm it. But what I've said in Talk isn't actually relevant to whether or not the text is accurate. If it is accurate -- and it simply says what is later in the article, so accuracy has not actually been challenged -- then whether or not I've misread the text isn't relevant.

Here is what I think is really going on. Having Robert's Rules be prominent as "recommending" IRV has been a major help to Richie's organization. I was just this evening talking with a political activist and she mentioned it. It's a big deal for them, and it was far more useful in the introduction than deeper in the article, because the introduction more commonly leaves an impression. But, in fact, Robert's Rules, when we look at the detail, is saying something that, if it becomes widely known, could be practically fatal to Richie's campaign. After fighting for a long time to keep a "compromise" text in the introduction, once it's made even more accurate, without the problem of being too much text, it's gotta go. Consider this:

IRV is being sold to places that considered it worthwhile to go to the expense of an extra election in order to gain a majority winner. FairVote activists have gone to these places and have claimed that IRV will find a majority winner, without the expense. And Robert's Rules has been used as part of that sales effort. IRV has significant setup expenses, but, if, indeed, runoffs could be avoided, money would be saved overall. (Maybe, I've seen contradictory data.) But the procedure Robert's Rules describes does not eliminate the possibility of runoffs. In fact, if the majority requirement had not been eliminated in San Francisco, Burlington, and Cary, i.e., if they did what Robert's Rules actually describes -- which reduces runoffs but does not eliminate them -- we'd have seen a reduction in runoffs from nine to seven. Not worth the setup expense. Plus the problem that IRV is almost certainly producing different results, no "comebacks." IRV is not finding majorities, *usually*. Robert's Rules insists on them. Definitely, the IRV campaign has depended on people not being aware of this problem. Bringing it to prominence in the lead to the article is a serious difficulty for him. Hence, in spite of being warned before about edit warring or even making edits reasonably likely to be controversial, he's done what he's done. The stakes for him are large. Richie was blocked before, as an IP editor cooperating with sock puppets to control this article. He could be blocked again.

Notice the irony of this. I was being condemned for being a POV-pusher for taking the Robert's Rules mention out of the introduction. I certainly was not the one insisting on it being there. However, after being reverted for taking it out, I simply edited it to make it more accurate. Nothing was in there that was not already in the article, deeper in, sourced. I think he just hoisted himself on his own petard. It's hard to get other editors to look at the article, we've tried in the past. But the argument was made, pretty strongly, by the editor who was just accusing me of POV-pushing that Robert's Rules was very important. Maybe it is about time that what Robert's Rules actually says be given more prominence. I certainly did not plan it this way! --Abd ([[User talk:Abd|talk]]) 02:17, 19 May 2008 (UTC)

Abd. I'm an utterly confident that any objective reader of the Robert's Rules section would disagree with you. Its basic logic is: 1) We suggest repeated balloting. 2) We recognize that sometimes this isn't possible, as in vote-by-mail elections. 3) In such cases, we suggest preferential voting. 4) As an example of preferential voting, we describe IRV (certainly not Bucklin, which would have been much easier to describe).
Note: Robert's Rules is not saying to invalidate such vote-by0amil elections if a candidate who wins a majority in the final round of the IRV count does not have a majority of the initial vote. It indeed is saying to try preferential voting when you're not going to do repeated balloting.
I think it's quite clear what is being said here. Again, I would suggest any objective reader would come to thee same conclusion that dozens (who knows how many) of private organizations have concluded in adopting IRV, probably sometimes after reading Robert's Rules: that IRV is a sensible system in such situations. Seems like we should find such a reader and see what they think instead of having you insert highly questionable interpretations in there (one, suggesting that Robert's Rules would say you should invalid certain vote-by-mail IRV elections and two, implying that they think other preferential voting systems are better).
On Iowa -- note that "instant runoff" comes off in relation to the IOwa caucuses in various sources. See:
greglist.blogspot.com/2007/02/its-official-howard-dean-gets-credit.html
Just seems reasonable to explain that connection, but no big deal.

RRichie (talk) 12:11, 19 May 2008 (UTC)

Let's keep it simple: suggesting that Robert's Rules would say you should invalid certain vote-by-mail IRV elections It's not "would say," it's "does say." Period.

implying that they think other preferential voting systems are better).That is, in my view, implied by what they actually say, but I wouldn't dare try to put that in the article. A suggestion: while we might mention facts and arguments peripherally relevant, let's not waste time arguing over what doesn't impact article text. We are here to build an encyclopedia, not to debate what grek is best if you want to have a blantook. If RR, being notable and a reliable secondary source, says that greks are best, fine, we can put it in. But if, reading RR, and looking at other evidence, one of us concludes that greks would be considered best by RR, or thinks that any reasonable person would so conclude, that's not enough for article content. At least not content attributed and sourced by RR. --Abd (talk) 17:17, 19 May 2008 (UTC)

[edit] Request for objective editors: Please review summary of Robert' Rules.

An advocate of other voting methods who clearly wants Robert's Rules to support these other voting methods, Abd Lomax wants the top part of the instant runoff article to read as follows

<<The sequential elimination technique used by IRV is detailed as an option in Robert's Rules of Order[2], though Robert's Rules' version requires a majority vote or the election fails, unlike most IRV variations in use. Robert's Rules also notes that this form of preferential voting, because of the eliminations, can fail to find a compromise winner (preferred by a majority over all other candidates).>>

Here are parts of this sentence that I think are not supported by what Robert's Rules actually says;

"Requires a majority vote or the election fails, unlike most IRV variations in use" -- Robert's Rules suggests you do preferential voting when repeated balloting is not possible. That suggests to me Robert's Rules isn't suggesting "failing" an election if the majority winner in the final IRV found does not have a majority of the initial majority vote, but rather electing someone in one round in a way that is more representative than plurality voting.

Notes that this form of preferential voting, because of the eliminations, can fail to find a compromise winner (preferred by a majority over all other candidates) -- Robert's Rules suggests repeated balloting. It isn't suggesting another form of preferential voting as better than IRV. If it were, it seemingly ould have described this system rathe than iRV. When I changed the reason to "because as opposed to repeated balloting", Abd reverted to his spin, which is one that he thinks points to other kinds of preferential ballot. But repeated ballot can get to a compromise winner not because of voting algorithms, but because of voter reacting to the information in the earlier rounds of voting and changing their votes to compromise on someone.

Here's the full excerpt of Robert's Rules, with key sections bolded. I believe a fair reading of this section would lead one to say that a single vote-by-mail vote with IRV is being suggested here if you can't do repeated ballot. It also EXPLICITLY is counseling against Bucklin voting, as it says that voter education is important because "Sometimes, for instance, voters decline to indicate a second or other choice, mistakenly believing that such a course increases the chances of their first choice." This sentence would not be true of Bucklin, as in Bucklin withholding a second choice does in fact help your first choice -- its fatal flaw. Note that there is no mention of "election failing" when they describe who wins the count and no mention of other methods as preferable to IRV except for repeated ballot. (I suspect it's more likely that the reference to "many variations" is to the different subtle ways IRV can be done, such as "top two IRV" and so on.)

Robert's Rules of Order Newly Revised In Chapter XIII §45. 10th edition, 2000, pp. 411-414

§45 VOTING PROCEDURE

Preferential Voting: The term preferential voting refers to any of a number of voting methods by which, on a single ballot when there are more than two possible choices, the second or less-preferred choices of voters can be taken into account if no candidate or proposition attains a majority. While it is more complicated than other methods of voting in common use and is not a substitute for the normal procedure of repeated balloting until a majority is obtained, preferential voting is especially useful and fair in an election by mail if it is impractical to take more than one ballot. In such cases it makes possible a more representative result than under a rule that a plurality shall elect. It can be used only if expressly authorized in the bylaws.

Preferential voting has many variations. One method is described here by way of illustration. On the preferential ballot—for each office to be filled or multiple-choice question to be decided—the voter is asked to indicate the order in which he prefers all the candidates or propositions, placing the numeral 1 beside his first preference, the numeral 2 beside his second preference, and so on for every possible choice. In counting the votes for a given office or question, the ballots are arranged in piles according to the indicated first preferences—one pile for each candidate or proposition. The number of ballots in each pile is then recorded for the tellers’ report. These piles remain identified with the names of the same candidates or propositions throughout the counting procedure until all but one are eliminated as described below. If more than half of the ballots show one candidate or proposition indicated as first choice, that choice has a majority in the ordinary sense and the candidate is elected or the proposition is decided upon. But if there is no such majority, candidates or propositions are eliminated one by one, beginning with the least popular, until one prevails, as follows: The ballots in the thinnest pile—that is, those containing the name designated as first choice by the fewest number of voters—are redistributed into the other piles according to the names marked as second choice on these ballots. The number of ballots in each remaining pile after this distribution is again recorded. If more than half of the ballots are now in one pile, that candidate or proposition is elected or decided upon. If not, the next least popular candidate or proposition is similarly eliminated, by taking the thinnest remaining pile and redistributing its ballots according to their second choices into the other piles, except that, if the name eliminated in the last distribution is indicated as second choice on a ballot, that ballot is placed according to its third choice. Again the number of ballots in each existing pile is recorded, and, if necessary, the process is repeated—by redistributing each time the ballots in the thinnest remaining pile, according to the marked second choice or most-preferred choice among those not yet eliminated—until one pile contains more than half of the ballots, the result being thereby determined. The tellers’ report consists of a table listing all candidates or propositions, with the number of ballots that were in each pile after each successive distribution.

If a ballot having one or more names not marked with any numeral comes up for placement at any stage of the counting and all of its marked names have been eliminated, it should not be placed in any pile, but should be set aside. If at any point two or more candidates or propositions are tied for the least popular position, the ballots in their piles are redistributed in a single step, all of the tied names being treated as eliminated. In the event of a tie in the winning position—which would imply that the elimination process is continued until the ballots are reduced to two or more equal piles—the election should be resolved in favor of the candidate or proposition that was strongest in terms of first choices (by referring to the record of the first distribution).

If more than one person is to be elected to the same type of office—for example, if three members of a board are to be chosen—the voters can indicate their order of preference among the names in a single fist of candidates, just as if only one was to be elected. The counting procedure is the same as described above, except that it is continued until all but the necessary number of candidates have been eliminated (that is, in the example, all but three).

When this or any other system of preferential voting is to be used, the voting and counting procedure must be precisely established in advance and should be prescribed in detail in the bylaws of the organization. The members must be thoroughly instructed as to how to mark the ballot, and should have sufficient understanding of the counting process to enable them to have confidence in the method. Sometimes, for instance, voters decline to indicate a second or other choice, mistakenly believing that such a course increases the chances of their first choice. In fact, it may prevent any candidate from receiving a majority and require the voting to be repeated. The persons selected as tellers must perform their work with particular care. —Preceding unsigned comment added by RRichie (talk • contribs) 13:29, 19 May 2008 (UTC)

It would be great if we can get some other editors involved. We've tried in the past. There is a formal RfC process, and we have used it, and have attracted, as I recall (and it is referenced above), two editors. Who both agreed with my position. But the more the merrier. As long as they actually read the sources and arguments, or, at least, read the comments if they make edits based on first impressions.
Richie is a professional spin-master. It is his job to take sow's ears and make silk purses, beautiful ones. He does it well. In this case, certain apparent ambiguities in Robert's Rules, read by someone not keeping in mind basic parliamentary principles, but wanting to eliminate inconvenient repeated balloting, can lead to the conclusion he wants. But if one reads the whole Robert's Rules section, it's quite clear. What they are describing is not election by plurality, it is not IRV as commonly implemented. They use a different name; that ought to raise some caution. Let's look at the details. Richie has conveniently, above, quoted the relevant section, as I have previously. One thing is left out, the background. What is a "majority of ballots?" RRONR, p. 387:
... [basic requirement] it means more than half of the votes cast by persons legally entitled to vote, excluding blanks or abstentions, at a regular or properly called meeting at which a quorum is present.
... A plurality vote is the largest number of votes to be given any candidate or proposition when three or more choices are possible; the candidate or proposition receiving the largest number of votes has a plurality. A plurality that is not a majority never chooses a proposition or elects anyone to office except by virtue of a special rule previously adopted. If such a rule is to apply to the election of officers, it must be prescribed in the bylaws. A rule that a plurality shall elect is unlikely to be in the best interests of the average organization. In an international or national society where the election is conducted by mail ballot, a plurality is sometimes allowed to elect officers, with a view to avoiding the delay and extra expense that would result form additional balloting under these circumstances. A better method in such cases is for the bylaws to prescribe some form of preferential voting (see pp. 411-14.)
It is quite easy to see why someone not keeping mind the utter importance of obtaining a majority for any decision, and particularly not being aware of the propensity of IRV to elect by plurality when there are lots of candidates (something which has apparently escaped the attention of even experts in the field), to think that this is recommending IRV. But does pp. 411-14 describe "IRV"? The very name implies a substitution of some kind of preferential voting or the like to avoid runoffs.
But RRONR describes a form of IRV that still requires a majority. From the context, provided above, it might seem otherwise. But if an organization were to simply provide in its bylaws that "the form of preferential voting described in RRONR, pp 411-14, shall be used," would a majority still be required? The answer is actually obvious, once one looks at the whole text: yes. It would still be required. Richie has quoted it, but he has avoided directly confronting the clear meaning. Instead he uses indirect arguments, appeal to "common sense" or the "average reader," and the like. Here is what he keeps avoiding:
The members must be thoroughly instructed as to how to mark the ballot, and should have sufficient understanding of the counting process to enable them to have confidence in the method. Sometimes, for instance, voters decline to indicate a second or other choice, mistakenly believing that such a course increases the chances of their first choice. In fact, it may prevent any candidate from receiving a majority and require the voting to be repeated. [my emphasis]
What is called truncating, not indicating additional preferences, can result in an exhausted ballot. Standard "IRV", which is a political term actually invented by a friend of Mr. Richie and successfully promoted in place of the traditional names that don't mention runoffs, considers these ballots the same as if they were spoiled. In Australia, they are literally spoiled, they don't get counted at all. Clearly, though, from the above quote, Robert's Rules expects them to be counted. It's explicit. When we look at the counting procedure, it never describes excluding these exhausted ballots from the basis for a majority. It would have been very simple for them to do so. Given the importance that is given by RRONR to a true majority, cited just above by me, if they were making an exception, they would have stated it, instead of only indirectly implying it from the recommendation to use preferential voting when it is desired to avoid a runoff. Preferential voting can avoid runoffs, especially when there are only three candidates, and one of them is minor, and the supporters of that candidate realize it, so hedge their bets with a second preference. So, even if a majority is still required, the comment that preferential voting is preferable to election by plurality still stands. It hasn't been contradicted, except in Richie's mind, and, rather easily, obviously, also in the minds of non-parliamentarians who are more concerned about expense than about fundamentals of democracy, or who simply don't realize the implications.
It is crystal clear, once one looks carefully at the source. It's not ambiguous. I will not address the deceptive application of this paragraph to Bucklin, which isn't a question here. That section wasn't written about Bucklin, and it is quite a stretch to connect it. People's impressions of "IRV" aren't colored by an impression that it is like Bucklin, most people have never heard of it, RRONR wouldn't be cautioning against almost totally non-existent opinions. Clearly, though, it is not escapable, try as Richie might: What they describe incorporates the fact that if voters vote as described, a majority may not be found, and require the voting to be repeated.
They could have written, instead, and thus the voter won't have a say in determining the winner. But, instead, they described the possible consequence of truncation. I happen to disagree with the RRONR interpretation of truncation. It is not necessarily voter ignorance, a voter may consider a repeated ballot to be desirable. If it is a true repeated ballot, perhaps another better candidate will be nominated in time to get on the ballot. If it is a reduced candidate set, as with top-two runoff, a common compromise, the voter may consider that additional time and opportunity for debate will shift the result to their favored candidate. If they add a second preference, that candidate may win in the first round; if they do not, their favored candidate may win the runoff. Therefore the claim that IRV satisfies later-no-harm only applies to it as a plurality method, if voters are allowed to vote freely. Australia avoids this problem by simply requiring full ranking. Rank or else we will toss your ballot. Richie knows that doesn't have a snowball's chance in hell in the U.S.....
I have only focused on the majority requirement part. There is more to be said about the criticism given of sequential elimination. One miserable little question at a time! Does the method described in RRONR require a majority vote? I.e., a majority of all valid ballots cast? And the answer is not controversial, if we look at the text. If we look at what we want, then, obviously, questions are raised. Just because a fact has been relatively obscure does not mean that it is not notable and not worthy of prominence. Now, if you can find reliable source that claims that RRONR doesn't require a majority, we'd have a horse race. We'd have to note it. Got any other than your own web pages and those that have picked up on the propaganda? Casual mentions and dicta don't count. Explicitly, reliable source, perhaps written by a parliamentarian and published, that gives such an opinion.
Otherwise, Robert's Rules of Order is a secondary source, considered quite reliable. It is a compendium of actual practice, put together by parliamentarians, experts in the field. (The primary source is the actual procedures used by organizations, typically incorporated in their bylaws and standing rules.) --Abd (talk) 15:26, 19 May 2008 (UTC)
Robert's Rules also says to consider the most recent statement to prevail in conflicts. You consider the majority requirement earlier in the text as having more priority than the subsequent statement that sequential elimination is "especially useful and fair .... more representative." Smithers Ltd. (talk) 05:59, 20 May 2008 (UTC)
Very funny. No end to the arguments one can invent. No, I consider the majority requirement, repeated in the text, and which is described as a fundamental part of democratic process, decision by a majority of those voting, to prevail over what is actually a vague statement, susceptible to multiple interpretations. What Robert's Rules describes is not election by Plurality, ever. The process described is consistent with that, and probably would be read that way by a parliamentarian, even if it were not for the later comment that cinches it. Therefore it is quite easy to understand all the statements consistently, which is the *actual* legal principle. "Latest statement" only applies to direct conflict, or to clarifying statements issued after an ambiguous one. This is all one piece, one publication. Besides, in fact, the reference to a majority and to Condorcet failure comes in the last part of the section. Further, the text doesn't say that "sequential elimination" or "IRV" is "especially useful and fair etc." It says that about "preferential voting," in general. Then it describes sequential elimination "by way of example." Why that example and not a method without the problems it then describes? Because the better methods aren't in sufficient use. Sequential elimination is.
My opinion is that the RRONR editors were aware of the problems with IRV, but they were constrained by being a manual of practice (kind of like our guidelines), based on actual practice and deriving from that, not a manual of theory to be imposed. No method can guarantee a majority without redefining what a vote is (i.e, in Australia, it's required that all voters rank all candidates, which turns the ballot of a voter who equally detests two candidates into a vote for one of them. And this is known as a problem there, there is reliable source for that (which, by the way, should be in the Controversies article, I don't think it is, and if that article's AfD closes with Merge and Redirect, which is the current drift, this could mean that all the arguments, if they are notable, will be coming back here....). --Abd (talk) 13:08, 20 May 2008 (UTC)
I redid the summary, using quotes, but the version before SL's edit is simpler and says basically the same thing. We ran into this before; remember, we've been around this cycle before. I'd put in a brief summary. The summary would be taken out as not accurate. So I'd put in exact quotes. That would be taken out as too much detail. So I'd take it all out, the detail could be in the fuller explanation. And a cherry-picked version would then appear back in the lead, starting the cycle over again. Let me say this straight. What Robert's Rules describes can't be said to be "IRV" without qualification. The name was invented to describe an unconditional runoff replacement, though, where forced, IRV advocates have allowed the name to be used where there is further process in the presence of majority failure. But it was like pulling teeth to get any admission that there was, in fact, any majority failure at all, it was insisted that the last round majority satisfied majority requirements. Which the legal experts knew wasn't true. Politically, having the majority failure possibility in people's faces was intolerable to FairVote, hence the vigorous efforts to (1) keep it out of the article in the first place, and (2) if it is in the article, at least don't let it be in the lead!
Robert's rules says that STV "makes possible" a more representative result. It never says that a runoff won't be necessary. In theory, IRV reduces the risk of majority failure, and this becomes a certainty if all voters rank all candidates. (Which is equivalent to truly voting No on only one candidate.) So those who want to avoid runoffs will certainly be attracted by the method. But if they want to *force* election without a runoff, they are implementing a Plurality rule, something that the RR procedure isn't. It still fails if no majority is found. This is true for all preferential voting methods, though some are more efficient at finding majority winners (hence my occasional mention of Bucklin, which is moot at this point. Condorcet methods would also do it.) In fact, it turns out that, in real elections in the U.S., IRV is, roughly 70% of the time, failing to show a majority winner. I don't think anyone expected that, at least I didn't, and I've never seen that as a criticism of IRV until very recently. I think. --Abd (talk) 13:43, 20 May 2008 (UTC)

[edit] One sentence on Robert's Rules in the intro

(unindent) It isn't obvious which discussion heading I should put thoughts about RONR, so I am sticking them here at the bottom of the last section. I propose that we limit the mention of RONR in the introduction to a short neutral undeniably true sentence, and leave all the details, quotes (and edit battles) to the section on non-governmental use. Here is my proposal for the introduction. "IRV is one example of "preferential voting" as presented in Robert's Rules of Order, Newly Revised (discussed below in this article)." Tbouricius (talk) 18:40, 20 May 2008 (UTC)

If you limit it to just one sentence, why do you think that will make it easier? Do you think there would still not be argument about what that sentence says? Smithers Ltd. (talk) 23:24, 20 May 2008 (UTC)
The problem is this: a one-sentence mention, no matter how you slice it, looks like a recommendation of "IRV," unless that sentence is very, very carefully constructed. The problem is that Robert's Rules does, in certain narrow circumstances, "recommend" preferential voting, or, more accurately, notes that it is fairer than a "result by plurality," but what they describe does not elect by plurality, so ... it is fairer! IRV elects by plurality, in almost all implementations, and it really is doing this, it's not merely theoretical. What preferential voting has done, in that case, that makes it attractive to RRONR, is to find a majority, it collects contingent votes. This isn't "instant runoff voting," which is a neologism, invented only a decade ago. It's preferential voting or the alternative vote, or, Sri Lankan contingent vote, all these names have been lumped under one name by FairVote. IRV is so named because it is "like a runoff," but it isn't a runoff, and it doesn't elect the same as runoffs do, and Robert's Rules points out the difference (the opportunity to base later choices on earlier results.) It's a mess. What has been considered notable by the IRV proponents, thus to be in the lead, has been an error or distortion, selective presentation of what RRONR actually says. --Abd (talk) 03:01, 21 May 2008 (UTC)
What is your source for "a decade ago"? I found an apparent mention from 1975. Smithers Ltd. (talk) 03:18, 21 May 2008 (UTC)
I've seen correspondence from the person who claims to have suggested it to Richie. I also did a search of usenet archives. The term didn't appear, I think, until 1996 when it started to sporadically appeared. Then it increased each year. There was lots of discussion of the method, as I recall, but not under that name. The article in referenced by Smithers Ltd. simply says that an official called the vote transfer process in Ann Arbor the "instant runoff" phase. It didn't catch on, apparently. But there may well be other instances, but the term doesn't come into wide use, for sure, until FairVote started promoting it. Richie might read this. We could ask him! There is some discussion of this, I think, in the Talk archives here and I think Richie made some comment there. The method isn't called "instant runoff voting" anywhere else in the world, it is purely a U.S. thing. Most elections using STV for single winner are held in Australia, where they call it the alternative vote, I think. Here, it was generally called preferential voting. So was Bucklin. Same ballot, in fact. List of preferences. Different analysis.
But I'm certainly willing to think about Bouricius' suggestion. I'm tempted to accept it, but I'd want to consider it a bit. I personally find it quite important that RRONR does nail three major criticisms of IRV in a few words: (1) failure to find compromise winner (which I think means Condorcet winner; repeated balloting, Condorcet methods, etc. do find Condorcet winners, generally). Top two runoffs can fail just like IRV, same problem, really. (2) compared to runoffs, prevents voters from making their second choice in light of the first outcome. (3) is implied, not stated, because they aren't actually talking about "IRV." What they describe still requires a majority for election, it would be a separate change, requiring explicit mention in the bylaws, to allow election by Plurality. So the implied criticism is of electing by plurality, which, it turns out, IRV often does. --Abd (talk) 03:07, 21 May 2008 (UTC)

[edit] IRV, Roberts Rules, and WP:SYN

This controversy over mentions of Robert's Rules, to me, clearly calls into question whether WP:SYN needs to be tightly enforced here. There's an awful lot of original research around here, and this is the root of the controversy. When in doubt (as always), the sources should guide content. Apparently Robert's Rules says something about IRV. That's fine, include that something, and cite it to Robert's Rules. But if you want to include "Robert's Rules" in the content, you need a third reputable source commenting on the combination of IRV and Robert's Rules. I am not convinced that Roberts Rules says anything about IRV. Can someone with access quote explicitly the meaningful part from ref 2, Robert's Rules of Order, Newly Revised, 10th edition, pp 411-414, while noting that synthesis, even if you consider it "obvious" is not OK. Also, there is a lot of excessive verbage here that bamboozles, and doesn't help. --SmokeyJoe (talk) 04:27, 21 May 2008 (UTC)

It is quoted above, and in many other places through Talk for this article. Generally, what is in the article is not synthesis, it is exact quotation, with little exception. For convenience, FairVote, which has extensively promoted the idea that "IRV" is described in RRONR, has the relevant section quoted. Is there reliable secondary source that says that Robert's Rules describes IRV?
Believe me, we have been here before. Robert's Rules describes "preferential voting." It says there are many forms. "Preferential voting" really refers to the ballot form; for example, Bucklin voting was called "preferential voting." Then it describes by way of example, the Single transferable vote method of elimination. It does not give it a specific name, but the method is, indeed, transparently, single transferable vote. "Instant runoff voting" is a neologism, not occurring with any broad usage prior to 1996 or so. (Again, this is covered above in Talk, at length). What is "instant runoff voting?" It was invented to cover single winner STV methods proposed in the U.S. I'm not sure what was being proposed at that time; it may have been general STV, which allows ranking all candidates. But, I'm sure, they ran into implementation snags, and what has been actually implemented in the U.S. is either RCV (ranked choice voting), with three ranks allowed, or Sri Lankan contingent vote, with two ranks allowed. Voting methods experts have opined that this article on IRV shouldn't be under that name, it's a political name, designed to convey an impression which, it turns out, can be at least somewhat misleading. They didn't call it IRV in San Francisco, I've seen alleged, because it was anything but "instant." Final counts were delayed for months. It's *really complicated* to count, if it goes to runoff rounds. A single mistake from one precinct (or simply in one round, if, as in San Francisco, the ballots are all taken to a central tabulating site) can require recounting all subsequent rounds. And IRV is sensitive to ties, and presents many more opportunities for them.
If we have an article about Preferential voting, all the IRV material could be there, because all IRV implementations are forms of preferential voting. But there are, as Robert's Rules notes, many other forms of preferential voting than the STV method. Instant runoff voting, I rather reluctantly agree, deserves its own article, but I think that the article should focus, in fact, on the U.S. campaign, which is how the name "instant runoff voting" arises.
Now, many people, myself included, have read the Robert's Rules description, and thought that Robert's Rules was talking about instant runoff voting. When I first started to point out, here, that what was described was not precisely IRV, I was ridiculed. Preposterous, it's obviously IRV, is what was said. (at the time, there was no mention in the article of any possible majority requirement or of majority failure as an issue, and this matches the real world fact that serious opponents of IRV, as in San Francisco, did not even present it as an argument. That Roberts Rules "recommends" "IRV" is repeated all over the net, including on the official web sites of city election officials explaining IRV. Because of the language used, in fact, much of this can be traced, probably, to FairVote and to the Wikipedia article, which used to say that.
What is obvious is that the method is that of STV, manually counted, and the only difference is that "majority," for Robert's Rules, means "majority of valid ballots," whereas for practical IRV implementations, it means "majority of ballots containing votes for the top two, i.e., those in the last round." Given that there was one proposed IRV implementation that did have that requirement (the Vermont legislation introduced by our own User:Tbouricius, it's hard to say that it is *not* "IRV" that is described, to to say that it is, without explanation of the difference, turns out to be critically misleading. IRV is being sold as a runoff replacement, to save money. It's expensive to set up, but, the argument goes, it will pay for itself if it avoids many runoffs. But, in fact, it isn't finding true majorities, in most of the elections where there is no majority found in the first round. *If* some organization simply implements preferential voting as described in RRONR, they'd have a majority-required form, and thus the possibility of runoffs. But that is not, we have seen, what gets actually proposed. What is implemented is a specific set of instructions for counting the votes, and the rules are designed to find a "majority of last-round votes," not a true majority. This is often justified by analogy to runoff voting. But RRONR specifically notes, as well, that "preferential voting" as they have described it deprives voters of the ability to based their subsequent choices on earlier results. Runoff voting, by definition, finds a majority of votes cast in a single election. Robert's Rules wants "repeated balloting," which isn't runoff, which usually means top-two runoff. Top two runoff will almost always find a true majority. (It can fail if the ballot allows write-ins.).
It is true that a very strict application of WP:SYN could blow this out of the article here. However, my own philosophy of inclusion is subtly different. To me, the core policy is WP:V, verifiability, combined with WP:CONSENSUS. All guidelines require interpretation (and even all policies). How do we interpret them? Through consensus. Wikipedia functions through the consensus of involved editors, and there are processes for resolving issues where local consensus cannot be found. But most of these processes do not involve a truly detailed examination of the issues, and my observation is that "outside intervention" often gets it wrong.
So, basic question: is what Robert's Rules describes "IRV"? Properly qualified, yes. Nobody who has spent some time with the issue disagrees (except possibly for SmokeyJoe). What is more exact is that the method described for vote analysis is the single transferable vote method with a majority-of-all-ballots requirement. Can we find reliable secondary source for this? It might be quite difficult as is not uncommonly the case with obvious things. It's so obvious, why mention it? Now, the reference to Robert's Rules being a recommendation of "IRV," we might find secondary source on that, but it is an interpretation that turns out, we can see, to be problematic unless qualified. So what we did here was to quote RRONR, and only summarize where consensus agreed that the summary was fair. There were then attempts by IRV advocates to remove some of it as "unnecessary detail," but it was not difficult to resist that. If you look over the history of this article, the mention of IRV in the lead, and later description of it, has been crucial to IRV advocates, and some were willing to edit war over it.
A strict application of WP:SYN might appear to resolve this, but at the expense of the article. As another editor new to the article pointed out, Robert's Rules is, all by itself, highly notable. Robert's Rules is a secondary source, a compilation and summary of actual practice. I prefer to seek editorial consensus, and place this above WP:SYN when the "synthesis" is very, very clear. I did, indeed, point out that it was synthetic to state that "Robert's Rules recommends IRV," because, quite simply, it isn't true, or it is drastically oversimplified in a manner that creates a POV slant. But to say that Robert's Rules of Order describes the method of sequential elimination used by IRV, but with a majority requirement, is an obvious truth, as far as anything I've seen advocated by editors here. In other words, we don't say that "Roberts Rules describes IRV." We say something else that is directly supportable by the source and by the knowledge of terms. "Majority requirement" wasn't obvious to many, though. But it is quite clear, and if someone objects to that piece of synthesis, it's enough to just quote the passages that establish it, and the reader can make their own conclusions. What has misled many is that a majority requirement is incompatible, generally, with the total avoidance of runoffs, and many have read RRONR on preferential voting through the tinted glasses of unconditionally avoiding runoffs. Which requires election by plurality, something which RRONR really doesn't like; an organization must specifically allow that in the bylaws, and they don't recommend it. They are describing preferential voting as a way of possibly avoiding runoffs by collecting additional votes as ranked preferences. If a majority can be found, runoff avoided. And then they go into the two other big problems. Real runoff elections reverse the initial plurality result one time out of three, roughly. (My study was that, FairVote has an analysis of federal primaries with runoffs that showed 29%.) IRV virtual runoffs aren't reversing first round results, it's apparently rare. (9 runoffs held since 2004, majority failure in 7, no "comebacks.") Robert's Rules mentions this as the problem that handling it all in one ballot deprives voters of the right to base later choices on earlier results; and then they also describe the "center-squeeze" problem that can cause a candidate to lose who is actually supported by a strong majority (two-thirds vote with three candidates) over the preferential voting, STV method, winner. That's a specific criticism of the STV method, not of preferential voting in general, and they say it that way.--Abd (talk) 14:14, 21 May 2008 (UTC)
Can someone offer an abridged version of this? Grrr, okay, me!
How about:
  1. RRONR says something about a procedure similar to IRV.
  2. RRONR is apparently strict about defining a majority to include exhausted ballots and thus possibly denying a winner.
  3. A given implementation of IRV may or may not(?) allow the possibility of denying a winner based on a failure due to exhausted ballots.
  4. Abd thinks IRV makes little practical difference to Plurality based on empirical evidence of IRV election results. (Possibly because of the failure to demand a true majority of ballots for a winner.)
  5. Abd thinks RRONR is NOT advocating aything like IRV except as an inferior compromise for long distance elections, best avoided, and doesn't want statements connecting RRONR and IRV without more complete qualifiers of its limitations.
SockPuppetForTomruen (talk) 22:46, 21 May 2008 (UTC)
Thanks Tom. The length and details of the verbose response tells me again that probably WP:NOR is being pushed too far. It seems to me that: “RRONR” (Roberts Rules) says nothing about IRV. I believe it even predates IRV. Therefore, any connection between RRONR and IRV needs to be sourced to a reliable source. What the article says should be guided by such reliable sources. Points 3-5 sound like original research and don’t belong, and even if this stuff can be sourced, I think it would then belong in a more specialised article. --SmokeyJoe (talk) 05:56, 22 May 2008 (UTC)

I went back though Abd’s answer, and pulled out what I think is relevent (in bold).

FairVote, which has extensively promoted the idea that "IRV" is described in RRONR,

If FairVote is a reliable source, and the information is interesting to the article, then quote FairVote. I suspect that FairVote is not an authoritative source, and so information must not be assumed to be fact.

That Roberts Rules "recommends" "IRV" is repeated all over the net, including on the official web sites of city election officials explaining IRV.

Then the following would be appropriate: It is often claimed that Roberts Rules recommends IRV (cite three examples)

So, basic question: is what Robert's Rules describes "IRV"? Properly qualified, yes. Nobody who has spent some time with the issue disagrees (except possibly for SmokeyJoe).

I do not agree or disagree with the answer. I question whether the answer is reliably sourced, and if it is I think it is not interesting enough for more than a breif mention and a cross-reference to Robert's Rules of Order.

It might be quite difficult as is not uncommonly the case with obvious things. It's so obvious, why mention it?

This is not a rare argument used in attempts to avoid WP:RS. The argument against it is that Wikipedia aims to be reliable, and it is only as reliable as its sources. Admitting the obvious is the thin end of the wedge of sloppy scholarship and original research. On analysis, every step of a thesis can be argued to be obvious. I am not opposed to the obvious, but if anything that is obvious is involved in a dispute, then it is time to enforce WP:SYN and to stick with what the sources say.

Most of the rest of Abd’s response tells me that he is very much engaged in his own OR on this and related topics, that he has well beyond a passing interest, and that maybe he would benefit from a careful adherence to WP:COI. I don’t mean to be rude, I think he cares too much. --SmokeyJoe (talk) 05:56, 22 May 2008 (UTC)

This all reminds me a little of the Woody Allen movie where he's in a movie line hearing someone talk about Marshall McLuhan and really disagreeing with him, and then is able to turn around and have McCluhan walk in and tell the person "no, you're absolutely wrong." In this case, would Robert's Rules folks be able to tell us what they mean? To me, what they are transparently saying (and as FairVote as "source" is simply putting the unedited excerpt on the web) that if you can't do repeated ballot because you're holding a vote-by-mail election or for some other logistical reason, then don't use plurality voting. Instead use "preferential voting" -- and then it describes quite precisely and accurately a sequential IRV count It also says to nail down all the specifics in your bylaws to avoid confusion and to do good voter education, including making it clear that ranking less-favored candidates won't hurt your more-favored candidates (thus clearly discounting Bucklin as an alternative).
An organization deciding how to hold an election with a copy of Robert's Rules in hand quite reasonably would set up IRV exactly as it is used in so many governmental elections. It would make a separate decision about whether to hold a new election if the final round majority winner does not have a majority of the original vote. I suspect most would go with the election results as they are because they already have decided not to do repeated ballot, but that would be a separate decision.
On terminology issues, I was intrigued to find that reference to "instant runoff" in the New York Times article that Smithers found on the 1975 IRV election in Ann Arbor -- the last IRV election in the U.S. before San Francisco in 2002. Abd is correct that in the mid=-1990s, we thought that "preferential voting" wasn't specific enough to describe one system (as indeed there are multiple variations of ranked-choice voting methods, of course) and "instant runoff voting", as suggested by a supporter of ours in California, seemed to nail it. Clearly that's what some Ann Arbor folks thought too -- a nice independent verification of it being an accurate way to understand how the system works.
Getting back to the article, to me it makes sense at this point to briefly mention Robert's Rules early on -- as indeed that helps explain the very real fact that so many private organizations use IRV, even if some might know it as "preferential voting" -- and quote directly from it later in the article with some, but not extensive commentary if disputable by different readers of the passage. (For example, I don't think it's accurate to say that Robert's Rules is suggesting all IRV elections should "fail" if the majority on the last count is not a majority of the initial count -- rather, it's saying your bylaws should be clear on that issue. And the practice of private organizations using IRV suggests that they almost always indeed do use one IRV election no matter what the final-found winner's total is compared to the first count.)

RRichie (talk) 10:49, 22 May 2008 (UTC)

Just so folks know, RRichie is Rob Richie, Executive Director of FairVote. Thanks for your comments. If one reads carefully, Rob confirms much of what I've said. He's correct, for example, that deciding to elect by plurality is a separate decision. However, this is the problem: that it is a separate decision has often been obscured by promotion of "instant runoff voting." After all, if the goal is to avoid runoffs, don't we want to make the decision always with a single ballot? The term "instant runoff voting" was brilliant. In certain respects, IRV does, in fact, represent a condensation of what happens with real runoff voting, and in *some* cases (roughly two-thirds of runoffs) it produces the same results. Now, can I source *this* summary? No. However I can source -- and have in the past -- the specific data from which anyone could make that conclusion. Yes, to answer SmokeyJoe, I'm engaged in "original research," which is what brings these facts to my attention. But I don't put the results of original research in the article. I talk about it, and that research is then verified, corrected, criticized, by others. Richie has, for example, pointed out errors in my work, at one point I misread the election data. What ends up in the article is not the product of original research, it's the data on which original research would be based. Indeed, my intention is to publish it, with conclusions, subject to peer review, and at that point those conclusions could be in the article.
We have a narrower point here. First of all, Robert's Rules is a secondary source. It is describing parliamentary practice. It never uses the term "instant runoff voting." It does say, quite clearly once one notices it, that the method they are describing can fail to find a majority, and that the election would have to be repeated in that case. They mean, in fact, "repeated." Not a runoff. However, if an organization wants to avoid having *three* or more polls, they *could* use elimination. But I'm not sure that anyone actually does that. Now, is there what would be technically RS for the statement that Robert's Rules "recommends IRV." Probably. This statement has been repeated so many times by so many people who either didn't notice the difference or who did but concealed it, that somewhere, I would not be surprised, it may have made it into casual mention in print, perhaps in a newspaper.
Yes, it's possible to wikilawyer this to attempt to exclude the fact that Robert's Rules' method is different from IRV. And, indeed, this was done before, by Richie, as an IP editor edit-warring, and as supported by sock puppets subsequently blocked. My own response is to stick with the policy: WP:V. When I insert synthetic material into the article, I do so subject to consensus that the degree of synthesis is low enough that reasonable editors agree that it is fair. Any time we summarize what a source says, instead of quoting the source directly, we are synthesizing. Unless we are quoting from a source that analyzes the source, in which case we are not synthesizing, we are quoting, and we would then source the statement from that removed source.
Note that the statements about Robert's Rules and IRV have been in the article for years, without any tertiary source quoted. Robert's Rules *is* a secondary source. Primary source would be the body of rules and procedures of specific deliberative bodies. (I have to look it up be sure, right now I'm not certain; Robert's Rules could be a tertiary source.).
I have not and do not insist on any synthetic analysis being in the article. What I have insisted on, in fact, is (1) No synthesis, interpretation of what RRONR says, unless it is balanced and we can agree that it is accurate. (2) If we cannot agree on summarization, then I've substituted quotation of the reliable source, which is RRONR itself. This, then, has frequently been removed by IRV advocates as "too much detail." And then they have put their summary back in, and the cycle has repeated.
As to my alleged COI, I was, before beginning to edit this article, a critic of FairVote argument outside and of the behavior of Rob Richie, and, previously, and several times, Mr. Richie, while he was blocked, attempted to make that point. I note that, now, it is not he who is bringing in this outside activity in, Mr. Richie is being quite civil and cooperative, even if he is predictably making arguments in a certain direction; I expect him to make those arguments and, indeed, welcome them. It is part of what we need. This is Talk, not the article. Now, it is SmokeyJoe making the suggestion. When I discovered this article and its condition (it was being maintained as a promotional piece by a cabal of editors, including Mr. Richie and the socks), I began to unravel the web and serious attempts were made to arrange for my blocking. I began to "research" the various points made in the article, and every time I did so, I found that what was being said here was different from what the sources actually showed. Often these claims were not sourced at all, but FairVote did have the quote from Robert's Rules, nicely framed so that people reading it would come to the proper conclusions. That is one of the few FairVote citations that I've continued to allow, simply because the actual quotation was accurate, and allowing people to see the source directly I consider to outweigh the problem of using a source that frames the material in an improper manner. If I wanted to wikilawyer, I could have removed that source because of the biased frame. I don't.
I was a wiki user before I got involved with Wikipedia. The concept of Wikipedia excited me, and I believe strongly in the power of community consensus, and the fundamental policies of this project. However, over the years, increasingly, there is reliance on the letter of guidelines, such as no original research and reliable source. The core policy is verifiability, however, which requires that material in Wikipedia be verifiable by any reader, with reasonable effort. And these policies should be interpreted, not by wikilawyers, but by consensus. No original research, interpreted as it seems SmokeyJoe may be interpreting it, would prevent practically everything but copying. But if original research comes up with a new way of seeing something, that can be verified easily by anyone, the sources for that research may be quoted. My research led me to notice that RRONR is explicitly describing a majority-required system. Note that when I first started saying this, the fact of it was denied, over and over, or attempts were made to cover it up with distraction, such as Richie's claim above: An organization deciding how to hold an election with a copy of Robert's Rules in hand quite reasonably would set up IRV exactly as it is used in so many governmental elections. Indeed they might. Indeed they have, in some cases. Reliable source?
Still in the article is a claim that that the American Political Science Association uses IRV. The claim was made more accurate after I challenged it. What method do they use? Do they require a majority? Well, they don't actually use IRV, they may never have used IRV, and the details aren't specified in the Constitution. It simply says "the preferential method" or something like that, and the details are left to a committee. Again, when I tried to take this out, edit warring to keep it in. Frankly, I think it isn't notable, and should be out, entirely. FairVote wanted it in for obvious reasons (you can see it if you see the various ways that FairVote activists refer to APSA, they wave it as a badge of approval, i.e., if political scientists use IRV, it must be a good thing. But that argument, if actually pursued, would surely lead to curiosity over what they actually do. They actually elect their president with a nominating committee appointed by the previous president, and contested elections are unknown. Should we adopt this method?
My interest in FairVote, and my opposition outside, has always been based on a dislike of political spin. If the electorate or the legislatures want to adopt instant runoff voting, fine. I just don't want it to happen based on shallow, deceptive arguments; but a political activist who makes his living based on success has different motives. What he wants is effective arguments, those which persuade and convince, even if they do so through spin and misrepresentation. Am I COI? If I am, anyone who takes interst in an article, and who researches the topic, and who then forms opinions regarding the debate, would be COI. I have a POV, that has been developed through study of the topic. Does that exclude me from directly editing the article? My "original research" is not a source. The sources are the sources, and what I'd put in the article is text from sources, or summarization that is agreed upon by consensus as accurate, true to source.
However, if SmokeyJoe still thinks my behavior improper, it looks to me like he knows what he can do. I trust the community consensus. More or less, it can be, increasingly, somewhat knee-jerk and increasingly relies on strict reading of guidelines; in my opinion, Wikipedia is devolving, slowly leaving behind the original principles that made it great, increasingly becoming vulnerable to sustained and sophisticated pressure from special interests. But ArbComm decisions are generally more considered. No guarantees exist. And it is a huge hassle. --Abd (talk) 12:59, 22 May 2008 (UTC)
No, I did not mean to suggest that Abd’s behaviour is in any way improper. The reference to WP:COI was made because I believe it contains friendly and useful advice for people who know more about a subject than you might get from a quick reading of the source material. --SmokeyJoe (talk) 21:49, 22 May 2008 (UTC)
Thanks for clarifying. I can understand why someone would mention WP:COI, and, if I actually publish some research I would become, as involving it, COI, which means I could not take strong positions on content, enforced by reverts or controversial edits. I couldn't add a reference to my research to the article, but I would, as can User:Tbouricius and User:RRichie, make suggestions here in Talk. Power on Wikipedia is a funny thing. The cabal opposed my RfA, though with little effect: with not nearly enough edits under my belt, I still got 50% of the votes. But being an administrator would have given me no ability to push my POV and, in fact, it becomes more dangerous to even try. I care about Wikipedia more than I care about my POV, which is why I was willing to accept the nomination in the first place. Not because I wanted it, I didn't. --Abd (talk) 21:59, 22 May 2008 (UTC)

[edit] How to respond to AfD that resulted in Merge/Redirect.

While AfDs are supposed to be closed based on arguments, not on votes, the Afd for Instant-runoff voting controversies closed with Merge and Redirect, which was a majority position; however, the closing admin appears to have considered the article a POV fork, even though nobody answered multiple challenges to this in the discussion. So the closing admin then deleted the article content, replacing it with a redirect to this article, saying "someone else can Merge." Merge and Redirect somewhat implies that Merge comes first, and the deletion of content that accompanies Redirect is then done. Redirect hides the original content. If you want to see it, follow the link above to the original article. It will bring you back to the IRV article. But at the top, you will see "redirected from ...." Follow that link and you will be back at the original article page. If you look at history for that page, you can then pick an earlier version and read it.

Since the link to the Controversies article now produced a double-redirect, it was removed and then the one removing it tagged the section on controversies as unreferenced. Of course, all the references are in the removed content.

Now, in my view, the closing was incorrect. What do we want to do? There is Deletion Review to reconsider an AfD. However, my opinion is that a Merge/Redirect result from an AfD is advisory only, it is not a fixed conclusion. I suggest that we go to DRV only if ordinary editorial process fails to find consensus only because outside editors (i.e., those not actually involved with editing this article or the other one) interfere based on the AfD result.

Given that the redirected article was all relevant to this article, and that problems with that article were purely a matter of the editorial process being incomplete (there was plenty of material that was unsourced as yet, but which was put in because the author considered it sourceable; this is quite common with voting systems articles), there is one immediate possible reponse, however, and that is bringing the entire content from the other article into the section of this article on Controversies. Then we could work on it here. And if we find that this is too much detail for this article (that's what I argued in the AfD, and it is what was argued before here, which is why the controversies article was started in the first place), we can then reopen the controversies article and put stuff back there.

In my opinion, an AfD does not substitute for editorial consensus on articles. AfD was designed for deletion decisions, and Merge/Redirect is, I believe, merely a suggestion, it is really a form of Keep: i.e., the content is notable and it was the opinion of the closing admin (based or not based on the arguments) that the content belonged in the main article. No additional right to remove content is created by a Merge/Redirect decision that did not exist before.

So, accordingly, I'm going to pull all the content from the other article into this one. Some of that content is unsourced. If you think it incorrect, remove it. If you think it possibly correct, but you want to see sources, tag it. And if you can, provide sources for what is unsourced. I started doing that with the article while it was under AfD (which is always a problematic thing, I've done it many times only to see the whole article deleted, waste of time). I provided sources for arguments for and against IRV.

So: we can work on this content while it is here, and, if it is too much detail, we can re-open the other article. I'm not going to do that as the Lone Ranger. If someone reverses what has been done -- any editor can do that -- I'll support it. But I'm not going to edit war with an administrator, especially. A single revert isn't an edit war, and we can go to DRV if it all becomes problematic. But I'm starting with accepting the AfD result, which quite specifically, suggested bringing the content here, and since nothing that belonged there is intrinsically not belonging here, I'm doing just that. I'd have preferred to fix problems there, and then bring summary back here, but .... wishes aren't horses, and, Tom, your weak arguments in the AfD did not help. The Controversies article allowed detail examination of arguments; now all the arguments will be in one place, here, even more crammed together. Damn straight it is confusing, look at the 2-party argument. As I noted elsewhere, all four combinations actually exist and can be found in reliable source: IRV helps preserve the 2-party system, and that's good. Ditto, and that's bad. And it will destroy the 2-party system, and that's good, and, ditto, that's bad. Different people make those arguments. And it is not up to us to decide who is right. We, properly, report on the controversy, provide the reader with verifiable information, and it is up to the reader to decide. --Abd (talk) 15:24, 23 May 2008 (UTC)

[edit] Special Request: Please do not mass-remove content.

Given that the Controversies article, though problematic, was relatively stable, please do not mass remove the content from it. Please don't take out more than one section at a time, if you take out anything. That way, we can edit separate sections and find consensus on one issue at a time. The AfD conclusion, generally, was that there was useful content, and definitely some of it was properly sourced, so a mass removal, for example, of all that content would be removing sourced material, which shouldn't be done without strong reasons.

On the other hand, if there is a desire to reorganize the content, the only practical way to do that would be to work on a page set up for that, and, in the meantime, the content would stand here. By all means, if you think unsourced content is wrong and can't be fixed, take it out. But, please, do it one detail or small section at a time, otherwise we could end up with unnecessary edit wars. --Abd (talk) 15:43, 23 May 2008 (UTC)

In the same spirit, is it wise to mass-insert material as you have done today? There is lots of original research, etc, as you of course know as its primary author.
RRichie (talk) 19:41, 23 May 2008 (UTC)
The mass removal was merely correcting an inappropriate mass insertion. There is a huge difference between an article on controversies of IRV and a main article about IRV. You can't just dump it all into the main article (I didn't check, but you may have doubled the article size by doing that). This is totally out of whack and inappropriate.
By way of analogy...There is an article on evolution. There is also an article on creationism. There is another article on the controversy between the two (I think "the politics of creationism," or something). All such articles are appropriate for Wikipedia. What is INAPPROPRIATE would be sticking all of the material about the controversy between evolution and creationism into the article on evolution, making that article largely about that controversy, rather than about evolution itself.
So Abd, if you want to cut the controversy material down so that it is comparable in length to the other sections in the article (for example, the section on strategic voting), we could then haggle over the exact wording of that section. But it needs to be pared down to an appropriate size before you try to insert it again.
Tbouricius (talk) 20:24, 23 May 2008 (UTC)
No, I'm not planning on "cutting it down" like that. If notable sourced material is too much to find in this article, the guidelines indicate creating another article. I'd be happy to once again cooperate in that, but for now, we are stuck with this. By all means, if a section is inappropriate in your opinion, take it out if it is not sourced or discuss it if it is. If you only take out one section at a time, I'm not planning on pulling out the COI card. But if you mass revert or remove, I will. We can continue to cooperate, as we have before, or we can let this turn into another mess. The opinion at the Afd, from too many editors who don't know beans about this topic, was that there wasn't enough "good" material in the controversies article that was proper to unbalance the IRV article. What was actually happening, I believe, was that they were mistaking unsourced for unverifiable. Where did all that unsourced stuff come from? Well, I put some in, I've written about that before, it's part of the editorial process; but much of it came from Mr. Bouricius and others.
Richie's claim above that I was the author of most of what was in the other article is not necessarily true. I didn't make the bulk of the edits, that's for sure. I'll come back with an analysis. During the AfD, I left alone plenty of material I disagree with, and tried to focus on adding sources. I added sources for some pro-IRV arguments, the pro-IRV editors hadn't bothered. --Abd (talk) 02:10, 24 May 2008 (UTC)
One more point: I requested that mass removal not be done *before* Mr. Bouricius did his mass removal. From his comment above, he may not have been aware of that. --Abd (talk) 02:28, 24 May 2008 (UTC)

There is so much selective use of information and questionable interpretation of information in this new section of the article, it's hard to know where to start in editing it - -and it's massive and out of scale with the rest of it. Abd seems to have endless time for it, but I'm not sure if others do. RRichie (talk) 12:10, 24 May 2008 (UTC)

[edit] Controversies merge draft

DRAFT WORK FOR MERGE IS HERE:

I'm not accepting the use of this fork. First of all, I already edited the other article to fit it in, i.e., rewrote the intro paragraph and changed the section headers to one level deeper. Since the other article wasn't getting sufficient attention, and it was a live article at top level in mainspace, linked from this article, I see no reason to believe that this fork will get better attention. I argued long and hard, with little support, that the other article should stay in place, precisely because it was going to cause problems brought back here. Unless someone challenges the AfD, that content is coming here, and it should have come here before the redirection, in fact, leaving all edit history in one place. So now we will have *three* edit histories?

No. I'm not touching it. I'm working on *this* article. Except for what should be noncontroversial edits to the intro, nothing I brought in here was out of the ordinary. If it was good there, it's good here, and that seems to have been the opinion in the AfD. If it wasn't good there, it's not good here. What was not good there shouldn't be here, but the way to deal with that is to edit it out, not assume that an entire subarticle's worth of material, en masse, does not belong. If you don't like this, and I don't particularly like it, then challenge the AfD. I'll join in that.

I'm bringing the content back in, and I'll be working on it *here*. Meanwhile, as to this being too much material, and I'd agree, read what should be done about it at WP:SUMMARY. We already did that, and we can do it again. In my opinion, the AfD does not override editorial consensus; if we need to go to Deletion Review -- and we might not have to --, then I'll support it. But I'm aware of nothing in the article that is clearly outside the legitimate. Definitely, when we are done, some material will be gone, but more will also be added. While I was trying to do some cleanup during the AfD, I found *lots* of sources, including sources for plenty that is currently unsourced.

I'm going to remind Mr. Bouricius that he is COI; his unblock agreement made reference to it, and he really should not have taken out that whole mass of material regardless of the content. This is not an ordinary insertion of mass material, this is a Merge, the specific decision of the AfD, and I really wonder, if someone is going to object to it en masse, why that person didn't object when it was in its own separate article.

I'm not planning to edit war to keep in any particular piece, though I will stand firm against the removal of sourced material. What I asked for above was that, if you were going to remove something, remove no more than a section at a time. That, then, makes working on that section much easier. If there is anything worse than mass insertion of material, partly sourced, it is mass removal of such.

If this amount of material unbalances the article, then there is a clear solution: a subarticle. We *could* just undo the redirect, but if one of us does that without first finding consensus here, I'm quite sure it would just be reverted back. Unless we go to DRV. We might get a different result from DRV, but probably only if those who realize what this will do to the IRV article get off their duffs and are active with the review. Tom, with his idiosyncratic theory about the conflicting arguments, wasn't helpful, as I wrote above. Again, above, he writes that he created the working copy, but he wasn't planning on actually working on it.

I'm going to emphasize this: Do not remove sourced material, discuss it or edit it to make it more accurate. All this material is public debate over IRV. It's relevant here.

(Whatever is not verifiable should come out immediately. So take it out, if that's what you think! But plenty in the section is sourced. So don't take that out! There is a reason why reverts are limited. It is far too easy to press Undo when you don't like a whole section; it takes work to come to consensus, to add sources, to selectively delete, etc.)

Much of the Merge thinking in the AfD mentioned that the controversy had been taken out of this article, which was true. The plan was to find consensus in the other article, then bring back summary of it to this article. It would then, indeed, be relatively brief here, and fleshed out there. But that didn't happen, and the idea of the separate article was quite what was rejected by the AfD. The AfD was flawed and the closure was flawed, but, unless someone challenges it, it stands at least for a while. --Abd (talk) 01:54, 24 May 2008 (UTC)

The IRV article is already TOO long, and I see no reason not to keep the debated merge in a draft copy rather than cluttering up the IRV talk. But make trouble as you like. The content was crap there, is crap here, whatever sources might support a fraction of it. SockPuppetForTomruen (talk) 03:56, 24 May 2008 (UTC)
I moved this draft to userspace. --Anticipation of a New Lover's Arrival, The 09:56, 24 May 2008 (UTC)

[edit] Who did write the old Controversies article, anyway?

It's hard to tell without a tool that studies what content came from where. I think such exists, but I did a quick study with a spreadsheet. I simply looked at the byte count shown with each edit. The numbers below show the net byte count (sum of all edit differentials) for all editors of the Controversies article from creation to the placement of the AfD tag.

Abd 9,659
Tbouricius 7,934
Tomruen 2,391
Captain Zyrain 1,680
McCart42 1,557
210.50.205.106 822
76.68.39.224 821
Jackms 671
Wilfred Day 637
Peter Ballard 392
Paladinwannabe2 315
Yellowbeard 243
58.178.192.225 175
Jackms 171
SmackBot 130
Xaosflux 117
Green Giant 34
71.202.63.202 17
70.170.103.198 10
207.172.80.2 -1
ClueBot -10
Retired username -243
StrengthOfNations -1,347
MilesAgain -2,149

I find this pretty interesting, actually. Look at the net negatives. "Retired username" is an admin who requested disappearance, out of respect for that I won't say his name. He apparently removed the AfD tag that Yellowbeard had placed (and that was Yellowbeard's only edit, I think. Yellowbeard only did AfDs. Appeared out of retirement to vote in this latest one, having been one of the few editors to be notified by the nominator this time. Tom Ruen and Paladinwannabe2 were the only others, and the latter hadn't edited in a long time. I saw the AfD from Tom Ruen's talk. Srength of Nations popped in and made some odd edits in various places, I've suspect that he was another sock of ... the puppet master for MilesAgain, who was James Salsman. Seeing who was net negative is a bit of a sign....

Now, as to positive contributions. Yes, I'm the largest single contributor. However, Tbouricius is close behind, and then Tomruen. Together they had a few more net bytes than I. There was little edit warring on this article, so .... Tom, exactly where did all that terrible content come from? If we just look at users with net positive contributions, I contributed 35% of their total.

Look, if we want to, and can agree on it, I think we could reverse the AfD result. But I haven't seen anyone else here really willing to push for that, so ... we are stuck with putting what is good from the controversies article here. And if it wasn't good, generally, why the Sam Hill didn't anyone object before? The same people who edited that content, other than me, the largest contributors, are now saying that the material doesn't belong here, except maybe a little. Absolutely, there is a fair amount of poorly-written and unsourced material there, including plenty that I certainly didn't write. But most of it is a skeleton for stuff that *does* belong in. I think that by the time we find the sources and clean it up, it could easily be longer than it is now. I took out the whole section on center squeeze during the AfD because it was confused. Or I was. :-) Given that center squeeze is one of the aspects of preferential voting that Robert's Rules takes pains to note as a problem, it really should be in this section. --Abd (talk) 05:21, 24 May 2008 (UTC)

Abd notes that I was the second largest contributor to the controversies article. But I would nevver have suggested that even one tenth of that material was appropriate for the main article. It is totally out of scale. My analogy about making the article on Evolution 40% about the conflict with Creationism...its not appropriate for the main article of an encyclopedia, regardless of how many sources and footnotes.Tbouricius (talk) 13:42, 24 May 2008 (UTC)
While I can certainly agree in some ways with Mr. Bouricius, the problem is that the AfD decided the content should be merged. So now I'm going to argue for that. Please understand that I considered the ideal solution to be a separate article on the public debate with brief summary brought back here. That's what WP:SUMMARY would suggest, and that is what I argued for with very little support. However, I'm now suspecting that there may have been some reason behind the madness of the AfD. An instinct that was being expressed.
This is not Evolution vs. Creationism. The name "Instant runoff voting" is a politically-coined neologism. It did not exist fifteen years ago (the "analogy" was made before, but not in any widespread way, for sure, and it is not how voting systems people thought of "single transferable vote" or the "alternative vote." Why are people coming to look at this article? I'd suggest that the number one reason would not be abstract interest in the voting system, it would be a result of the debate. They've seen some of it, and they will come here to learn more. In fact, much of the material in this article should be moved to Preferential voting. And then the article on Instant run-off voting would be entirely about the meaning of the words (it is being used to cover a number of different systems with varying characteristics, which causes certain errors to be made) and the debate over implementation.
So, for the time being, I'm not accepting the argument that the debate is a minor aspect of this article. It actually should be the core of it. That this article and preferential voting also exist as descriptions of the method,, is far more a violation of WP:FORK than whatever problem was seen with Instant-runoff voting controversies.
It's been a constant problem writing the article that sources would refer to "preferential voting" and would then simply be assumed to be referring to "IRV." Maybe, maybe not. There is a salient example of the error of the assumption of identity. Preferential voting satisfies later-no-harm, if the result terminates with a winner even when no majority is found. Require a majority, and later-no-harm does not apply. You can harm your favorite by adding a second choice. (Your second rank vote may cause that candidate to gain a majority, removing the possibility of a "comeback" in a runoff. Comebacks in Runoff voting seem to occur in roughly 30% of runoffs. So, absolutely, there can be some strategic value in "plumping" as it's called in Australia, if a runoff would take place if there is majority failure. (ACE notes that plumping is reasonably common where allowed, NSW and Queensland, I think. But runoffs don't exist there, if I'm correct.) Since it had already been concluded that Robert's Rules' "preferential voting" was "IRV," no attention was paid to details, including the very specific mention of election failure due to voters not fully ranking.
In any case, Mr. Bouricius did a single massive edit to the article. Reviewing it briefly on a handheld, some of that seemed okay, and some possibly not, but," given how easy it is to make changes with smaller edits, I again request that, unless there is a good reason and it is done carefully, massive revisions with a single edit be avoided. They are, for one, more likely to be reverted, just on that basis, even when they include some good stuff. (My merge from the other article was an exception, having to do with efficiency of process. That was mostly standing content, it had been there with the presumed consent (see WP:CONSENSUS of editors. Even though clearly needed a lot of work, i.e., cleanup and sourcing.) *Especially* when making edits which could be controversial, what is likely to be uncontroversial should be made as separate edits. In fact, this will make it easier to find consensus, because gradually what we can all accept will appear.
I'll repeat this: 'I'm not accepting WP:WEIGHT, at this time, as an argument for exclusion of sourced and notable material.' It is far too easy an excuse for maintaining POV imbalance. There are other solutions to genuine weight problems that do not involve disappearing good content. --Abd (talk) 15:08, 24 May 2008 (UTC)

[edit] section on runoffs and IRV

With this edit I reverted the previous edit of User:Tbouricius. The text I restored read:

In comparison, a two round system (with a majority requirement) will always allow interested voters to participate in the runoff round, and the result satisfies requirements of election by majority. However, depending on election timing, it is argued, there may be a significant drop-off in voter turnout in the second round and it is possible that the nominal "majority" winner of the second round may receive fewer votes than the loser received in the first round of voting.[2]
This drop-off did not occur in runoff elections in Cary, NC, where runoffs were held with the general election in November, the primary being held a month before,[citation needed] and this kind of shift of election timing was proposed as an alternative in the debate over Ranked Choice Voting in San Francisco.[3]

This had been substituted for it:

In comparison, a two round system (with a majority requirement) requires interested voters to participate again in the runoff round, with the majority threshold ignoring the number of votes cast in the first round. There is almost always a significant drop-off in voter turnout [4] in either the first or second round in a two-round runoff system, and it is possible that the nominal "majority" winner of the second round may receive fewer votes than the loser received in the first round of voting.

First of all, we should be clear about something: a section (or article) on controversy does not free us from the responsibility to express the topic in an NPOV fashion. We can quote FairVote for advocacy arguments, but not for facts, and the section Bouricius inserted presented what is really FairVote propaganda as if it were simple fact. In the section I had written, believing that it would be roundly accepted, I wrote, about runoff drop-off, that "it is argued" that there may be a significant drop-off, and the comment about "fewer votes than the loser" -- which is a manufactured argument, the kind that makes sense until one really looks closely at it -- was included in that, because I know FairVote makes this argument. That this is a common FairVote talking point when comparing IRV with top-two runoff, even though parliamentary authorities place no weight on the argument at all and it flies in the face of standard democratic theory, might have something to do with it being preserved in Bouricius' version. Then "there is almost always significant drop-off in either the first round or the second round: is added. This is an entirely new meaning for "drop-off." "Almost always" is weasel words, and simply not true for the reversed timing.

If Bouricius wants to quote FairVote, or reasonably paraphrase FairVote argument, it must be done with attribution.

And "with the majority threshold ignoring the number of votes cast in the first round" is true. Also the number of votes in the last ten elections are ignored, and the price of eggs in China as well. A lot is ignored by a "majority threshold." What counts is the number of valid ballots in an election, and that is what a runoff is. An election. Depending on various factors, more voters may turn out or less voters, than in previous process, but there is no particular reason to believe that results are worse because of lower turnout: lower turnout means to me that voters who did't care didn't show up. Yes, it can mean other things, people have difficulty voting sometimes, but ... there are absentee ballots for that, and I don't see that differential turnout, overall and on average, indicates anything other than variations in voter apathy or interest. And neither does Robert's Rules; quite simply, it's not an issue for them. Lower turnout? Fine! Maybe that will result in a majority for one. (For the benefit of other readers who may not be aware, Robert's Rules prefers repeated balloting without elimination of any candidates (and even with, possibly, new nominations), until there is a majority winner. That can sometimes take a lot of ballots, but Robert's Rules is primarily designed for assemblies that meet in person, and so taking many ballots is much more possible. Robert's Rules would not recommend, at all, preferential voting for an assembly meeting in person. They reluctantly allow that it is a lesser evil that election by plurality for mail ballots, and, note, this is another clue that they consider preferential voting to still require a true majority, that old interpretation otherwise was totally bogus. (And they were explicit, fortunately, and I absolutely don't know how I overlooked that before.)

Is this really relevant, that primary elections, held before a regular general election, will have lower turnout? Do we think it a problem that party primaries are held separately? Social utility theory would indicate that lower turnout can improve election results by giving, effectively, more weight to the votes of voters who care. In any case, repeated balloting is inconvenient. And Robert's Rules considers the basic rule of democracy, that decisions are made by a majority of those voting on that specific decision, is more important than that inconvenience. Jurisdictions that have runoff process have obviously made that decision, and FairVote sells them IRV on the claim that it will find majorities. But it generally fails to do so, and generally, where majority failed in the first round, is electing by plurality (77% of the time). The language about drop-off in turnout is smokescreen to cover that up, and the mention about comparing the number of votes from one poll to the next is redundant." Does Bouricius or FairVote have any data to show that turnout is different between, say, an October primary and a November runoff? In Cary, they were the same. And we commonly hold primaries as special elections other than with the general election. The lower turnout in a primary is never called "drop-off," and, in this case, the claim about fewer votes for the winner in the runoff than in the primary is then utterly irrelevant.

By the way, I'm noticing a pattern of attempting to substitute the more complex language of "majority threshold" for "majority" or "election by majority." I don't wonder at it. But I'm not likely to accept it. No question, the language here can be improved, and I still intent to provide that Cary source, though it is not critical, the San Francisco ballot arguments would suffice.

In any case, we cannot invent arguments to place here. If I've done that, that was not my intention, my intention has been, wherever I've inserted language here that wasn't sourced, to represent what I'm familiar with from experience in this field, and I've never edit warred to keep something like that in, though I get fairly stubborn once it has been sourced. Generally, removing sourced material is not a good idea.--Abd (talk) 03:49, 25 May 2008 (UTC)

Mr. Bouricius edited the section on runoffs, essentially restoring language I'd removed and removing language that had been there (I think from me). While this has all been discussed above, and there was no response, clearly further discussion is needed. I reverted it. This is what I restored:
In comparison, a two round system (with a majority requirement) will always allow interested voters to participate in the runoff round, and the result satisfies requirements of election by majority. However, depending on election timing, it is argued, there may be a significant drop-off in voter turnout in the second round and it is then possible that the nominal "majority" winner of the second round may receive fewer votes than the loser received in the first round of voting.[5]
This drop-off does not necessarily occur in runoff elections where the primary is the special election and a runoff is held, if necessary, with the general election, as, for example, in Cary, NC, in 2003,[6][7] and this kind of shift of election timing was proposed as an alternative in the debate over Ranked Choice Voting in San Francisco.[8]
Mr. Bouricius had made it:
In comparison, a two round system (with a majority requirement) will require interested voters to participate in a second runoff election, and the result satisfies requirements of election by majority, since it does not consider the number of votes cast in the first round. However, there typically is a significant drop-off in voter turnout in either the first or second round and it is possible that the nominal "majority" winner of the second round may receive fewer votes than the loser received in the first round of voting,[9] which is not possible with IRV.
And his edit comment was: (It is not only drop-off in the second round that is of concern, It is dropoff in EITHER round compared to a unified election. citations of selected dropoffs are not appropriate.)

First change. "Allow" is changed to "Require." In the U.S., anyway, voters are never required to vote. They are allowed to vote. Turnout reflects, typically, the degree of interest. The language of "allowed" was my summary of the concern expressed by Robert's Rules over preferential voting: "Although this type of preferential ballot is preferable to an election by plurality, it affords less freedom of choice than repeated balloting, because it denies voters the opportunity of basing their second or lesser choices on the results of earlier ballots...." As Richie has pointed out so many times, RRONR calls for Repeated balloting, without eliminations, but the same principle surely applies to runoffs with a reduced candidate set on the ballot. In any case, taking what RRONR considers increased freedom, hence "allow" and turning it into "require" is a piece of political spin. Freedom is slavery. War is peace.

A majority requirement, indeed, "requires" some further process if no majority is found. That is a tautology. But what are the implications of this? On the one hand, there is the additional expense and inconvenience of voting. On the other, there this from the Vermont Senate recently:[22]

(1) The principle of majority rule is fundamental to the concept of democracy. When possible, election laws should be structured to uphold and facilitate this basic principle.

(2) In a multicandidate race, when no candidate receives a majority, the candidate with the most votes (the plurality) may actually be the candidate most opposed by the majority of voters.

(7) It would be desirable, and there is a popular preference, to have a direct popular election by majority vote in all elections for the offices of U.S. senator and U.S. representative.

(8) A voting system known as “preferential voting” in Robert’s Rules of Order Newly Revised, and popularly known as “instant runoff voting,” which has been used for governmental elections for over 80 years in Australia, as well as in the Republic of Ireland, can fulfill these goals of majority rule, with direct popular election.

They were sold a bill of goods. IRV does *not* tend to find majority winners, compared to top-two runoff, which always finds them, in practice. (Where write-ins are allowed, or sufficient spoiled ballots are cast, real runoffs can still fail to find a majority as RRONR defines it, but it's rare.) Consider an IRV election. The phenomenon mentioned as undesirable in item (2) can happen with IRV, rather easily, but it cannot happen with top-two runoff if the voters care. (If due to vote splitting in the first round, the the worst two candidates end up in the runoff, and voters don't care much about either over the other, it's possible the worst could win, but, again, that would be rare. Even when top-two fails to include the best candidate, it would almost always include the next best, and by definition, the next best paired with the worst would prevail.)

(2) Actually satisfying majority rule by requiring majority vote guarantees that (2) won't happen, in real life, because voters get to vote explicitly in a runoff. Since IRV can elect by plurality -- unless protected against that by a majority requirement -- it is bizarre that IRV is proposed as if it solves the problem. It ameliorates it, probably, only a little, whereas top-two runoff actually fixes it. And, guess what? There is a cost to fixing this problem. But there are other methods which apparently are better at finding majorities. Bucklin doesn't discard votes or conceal them. When a candidate wins in IRV by a majority in the last round, it's possible that the candidate has a majority, but those votes are concealed underneath the remaining candidate. In a majority-required IRV, one would continue the elimination another step, thus revealing (and counting, at some point) more of the votes. The IRV winner might then actually have a true majority. If majority vote was desired, this is exactly what would be done.


From FairVote Vermont: "With Instant Runoff Voting we will always know the will of the majority of Vermont Voters and we will eliminate the problems associated with winning office on a plurality." This is the propaganda. Noble goal. But extremely misleading. IRV, like regular runoff voting, tends to encourage additional candidates to run, which tends to cause majority failure. And, in actual practice, whenever "instant runoffs" are taking place, even with only a few candidates, most of them (7/9 elections) are failing to find a "majority of voters" for the IRV winner. It is, of course, finding a "majority of last round votes," because that is mathematically guaranteed by the method. IRV tends, strongly, to produce the same results as Plurality, exceptions are relatively rare. In fact, the only one in the U.S. that I'm aware of was Ann Arbor in the 1970s, which was an unusual situation: partisan election with a strong third party taking most of its votes from the Democrats, so the Republicans were winning with a plurality. IRV *is* election by plurality, but it generally finds a larger plurality and *sometimes* a majority where Plurality voting might only have found a plurality. I see no reason why voters in San Francisco would have voted differently if their three-rank ballot had been Bucklin instead of IRV (real voters don't pay attention to the Later-no-harm criterion as much as some think, particularly in nonpartisan elections), and Bucklin analysis, applied to the SF RCV vote data, finds majorities better than IRV. It, too, has a famous "comeback" election, where the Minnesota Supreme Court reversed what was clearly the outcome chosen by the voters in Brown v. Smallwood. But I digress.

I'm not stuck on the particular language here, and, especially, I'm interested in seeing sourced argument and fact from the FairVote editors. What's been done here is to make up arguments, as is happening elsewhere. I don't generally take out made-up arguments, if I think they are true, merely because of sourcing deficiencies, but there is a question of balance here. The sourced San Francisco material, which was about actual arguments presented in the implementation of RCV there, has now been taken out twice, by Mr. Richie and by Mr. Bouricius, with the edit summary only referring to the Cary material. The argument about the latter has,perhaps, some validity, though the Cary results are only cited to show that it happens, that both elections have similar turnout, not to make any claim about frequency, but, in fact, I don't know any study of overall turnout differential where the effect of one election being a general election isn't involved. FairVote's study is of both elections being primaries, if I'm correct, so they were both special elections. Now, turnout in their study is lower in the runoff. What does that indicate? Does it indicate some injustice or does it reflect that some voters will be equally happy -- or equally unhappy -- with either candidate? Historically, in the U.S. IRV saw some use in primaries, and was repealed and replaced with runoff voting. It would be interesting to find, from, say, newspaper archives, what arguments were presented at the time. (This factoid is not one we are likely to see where FairVote is trying to sell IRV to jurisdictions using runoff voting!). From my own examination of results with runoffs (one-third, roughly, of runoffs result in "comeback" elections, whereas in 9 recent IRV elections that went to "instant runoff," there were no comebacks. FairVote finds, in runoffs, 29% comebacks.)

Shall we focus on this section? I suggest first trying to make it complete. If we worry about "too much detail" first, we will never find consensus. Then, when it is complete and accurate in what it says, and it is all sourced, we can prune it down by making the language tighter. If there is an anecdote that is sourced, add a contrary anecdote, if one thinks it imbalanced. I've been following discussions on WikiEN-l, and some general comments there is that NPOV is almost never achieved by removing sourced material. Where there is imbalance, add. Where there is error, correct it. "Too much detail" is an argument that was rejected as almost certainly reflecting some POV-pushing. But there are always exceptions to every rule, including this one.--Abd (talk) 20:10, 26 May 2008 (UTC)

[edit] Various edits to Abd's new section

Like others, I question Abd's decision to move the controversies article into the article. It's very US-centric, for example, and rather arbitrary. It also makes the article a lot longer with a lot more on this than in some other areas But there are ways to improve it, and I added more resources and corrected at least one very clear error, which was in the section on third parties in the Con section and created a new con argument to address it. RRichie (talk) 14:33, 25 May 2008 (UTC)

Richie apparently has not been paying attention. It was not my decision. I opposed it. It is not "my new section." It's an old section that was in this article at one time and that became the Controversies article, with reference here; the plan was always to bring a summary back. The section was already here, but with a simple reference to the Controversies article. The decision to merge and redirect the Controversies article, however, was made in an AfD at Wikipedia:Articles for deletion/Instant-runoff voting controversies (2nd nomination). I also pointed out here that it was going to cause this article to become too large, and, in my opinion, there was no consensus to merge, but it was, at the end, a majority that way, and that's what the admin decided, incorrectly, in my opinion. I've also, above, suggested that I would support a review of the deletion, but not as the Lone Ranger, so to speak. No support has appeared, only carping about so much being brought in at once. But the alternative was to leave all the material effectively deleted. Perhaps some would like that! -- but it's not legitimate. If something is incorrect or violates guidelines, sure, take it out or fix it. Generally, the same standards apply here as they applied to that article, though, and if that makes this article too large, guidelines are again clear. Create a new article and summarize it here. Apparently the larger community doesn't think that time has been reached. (I think that most voting Merge and Redirect thought that most of the content would disappear because it looked to them like original research. In fact, it was simply writing without sources, and when that is fixed, there may be more content, not less. And then we will have a very good case for a subarticle.)
In any case, I thank Mr. Richie for making his edits piecemeal, it makes it much easier to review them. I'll discuss them here. But first, I'm going to create a new Talk section, because Mr. Richie first made an inappropriate edit to the Robert's Rules mention in the introduction.--Abd (talk) 15:13, 25 May 2008 (UTC)

[edit] Richie's edit to RRONR in the lead

I reverted Richie's edits to the lead with this: diff

His insertions are shown in italics here: "The manual goes on to note that if voters don't rank all candidates, this may prevent any from receiving a majority of ballots cast in the first round and, if established in the by-laws, "require the voting to be repeated."

This is, simply, incorrect. RRONR is not talking about rounds. It is talking about repeating the voting. And the default in Robert's Rules is that election by a plurality is not allowed. This does not need to be in the bylaws. Using preferential voting does not remove this fact, by itself, and the language of the procedure as given in the rules uses "majority" in the ordinary way. If the bylaws specify that the last-round "majority" determines the winner, even if it is not a majority of valid ballots cast, then the Bylaws would indeed have allowed election by plurality. But Robert's Rules is not recommending that. The language inserted by Richie is not found in Robert's Rules. The "voting to be repeated" language is not so qualified, that is interpretation by Richie, and incorrect to boot.

To make this even more clear, failure to add ranked votes cannot affect, at all, any candidate from receiving a first round majority. It only affects later rounds, and an ultimate possibility of majority failure. Richie, in his apparently strong desire to keep hold of this Robert's Rules recommendation fantasy that he was so long able to promote, and even to see enshrined in legislation passed in Vermont recently, that he projects completely absent meaning onto the text, in an attempt to interpret it consistently with the fantasy. (That's quite a propaganda coup! legislators should be careful when they declare "findings." They might be setting up pi to be exactly 22/7.)

I'd suggest that Mr. Richie review, again, the conflict of interest guideline, and that he refrain from editing the lead entirely; the guidelines suggest that COI editors make no controversial edits to articles at all, but participate in finding consensus in Talk.--Abd (talk) 15:13, 25 May 2008 (UTC)

Abd, let's deal with the substance of our difference. I realize this isn't the first time that has come up, but maybe, just maybve, we can get to consensus here.
First, it's clear that many groups that follow Robert's Rules use IRV without repeated balloting for their elections. We also know that many of these groups don't do repeated balloting even if the majority winner in the final round of counting does not have a majority of the first round vote.
There can be different reasons for this. For example, at some party conventions, people don't stick around throughout the day. Repeated balloting might not maintain the same turnout. Or organizers of such an in-person meeting might make a decision that they only have so much time to devote to elections. More typically, it is done when voting by mail and it is too costly to do repeated round of votoing.
So, for whatever reason, they have made a decision not to use repeated balloting. Tthey then use IRV as detailed with precision in Robert's Rules. Note that this description clearly does not say "invalidate the election if the final round winner does not have a majority of the initial vote." That's because so few groups that choose to use IRV seem to use this practice, I assume. (In fact, I don't know of where it's done. Abd, do you?
Now Robert's Rules does say to be clear in your bylaws about your method of voting. So I think rather than YOUR interpretation -- always fail the election if the final round majority winner does not have a majority of the first round vote -- I think it's quite sensible to read Robert's Rules to say that you specify this in the bylaws and then stick to that.
Note: you're right that it urges groups to do good voter education in part so that valid ballots don't fall off as the count progresses. But I think it's reading far too much into it to say that all IRV elections should fail if you don't a majority-of-first-count winner. Even repeated ballot will often have fall-off in turnout if the final candidates are not appealing to some voters. There's NO way to ever guarantee that winners will have a majority of the first-round count no matter what system you use.
I think my description is fair, Abd. It includes your warning (one that no group we know of seems to follow when using IRV, by the way), so I would think that would please you. But true, it doesn't include your interprtation that would suggest that all the groups that don't have this "failed election" provision are violating robert's rules.

RRichie (talk) 19:04, 25 May 2008 (UTC)

You have a clear point-of-view, Abd, but it is NOT a consensus point of view. Your interpretation of RRONR is different than how it's used in all the NGO's we know that decide to use IRV for elections. It's hardly surprising why they make that decisions, when you read the RONR passage, which has been quoted above. The description is of classic sequential IRV. It does not end with anytthing like "if there is not majority, you have to hace a new election." Rather it ends with "Again the number of ballots in each existing pile is recorded, and, if necessary, the process is repeated—by redistributing each time the ballots in the thinnest remaining pile, according to the marked second choice or most-preferred choice among those not yet eliminated—until one pile contains more than half of the ballots, the result being thereby determined. The tellers’ report consists of a table listing all candidates or propositions, with the number of ballots that were in each pile after each successive distribution."
That paragraph would be the obvious place to say your election has failed if not an initial first choice majoirty rather than a couple paragraphs down when suggesting good voter education. Remember, the context is that this is with groups that have chosen not to use repeated balloting. Given that decision, they aren't going to suddenly use repeated balloting. They are to use IRV instead of plurality voting. And that's what they do.
If your interpretation is so definitively the right one, can you please find a single NGO using IRV today that will fail an IRV election if it does not generate a majority of first choice winner? I can find many that don't do this. So if yours is the norm, surely you find it spelled out somewhere.
RRichie (talk) 20:15, 25 May 2008 (UTC)
I have warned RRichie at his talk page about edit warring as a COI editor. Even though I could justify additional reversions with regard to this particular set of COI editors (notice that Tbouricius has been warned by others for being a COI SPA, that he could be blocked), I reverted the last incorrect and POV-pushing change by Richie and then self-reverted. I could go ahead now, but I'm waiting for a token of good faith: if someone reverts my own reversion, I'd not need to go to AN/I or to the admin who previously blocked Richie and who is presumably already aware of the implications. This isn't a threat, it's an attempt to avoid Richie being blocked; his behavior here would very easily result in a block from any administrator aware of his COI status and his edits today.
Richie is confusing what Roberts Rules says and describes and actual practice of NGOs. If he wants to say that "actual practice of NGOs using IRV is to complete the election with only a plurality," fine. It's probably true. But it is also true that far more NGOs simply use Plurality, and the comment probably wouldn't belong in this article, though it's possible, if it could be sourced and properly balanced. What he calls "my interpretation" of RRONR is blatantly clear from the text, so clear that I don't need to do any synthesis at all, I'm totally happy with simple quotation of the text, and if someone thinks I've quoted out of context, they are completely welcome to add anything I've omitted. I have *not* claimed that it is "the norm" for organizations to use preferential voting and then repeat the election if no majority is found. What I'm saying is that this is what RRONR describes. Maybe it's done. But since I am not describing "the norm" but and instead simply insisting that the reference to RRONR represent what is relevant in the source, I don't need to find any such organization. The only reason I have done *any* interpretation in this section, or have left in of interpretation by others (Richie pretends it is only me, simply because I'm the most persistent editor on this), is to make it concise, and any concise language that we can agree upon is fine with me, but if we can't agree, then there is direct quotation. It's not *that* much text.
In other words, Richie demands, If your interpretation is so definitively the right one ... What interpretation? What Richie keeps inserting is interpretation. What I've left in or put in is reporting. And I don't mind exact quotation with no interpretation at all, not even the slightest. I don't find it necessary to explain away RRONR. Just let the readers know what it says about "preferential voting." And do add stuff like "if provided in the bylaws," which was, quite simply, not true. RRONR, I've seen this explicitly stated again and again, requires election by majority, and in another section (quoted in the RRONR section of this article?) it specifically states that plurality elections are not allowed unless allowed by the bylaws. No bylaw on the topic, majority election required. Implement preferential voting "as described by RRONR," no additional bylaw, election by majority is required, which is made very, very clear by the comment about educating voters about what can happen if they truncate.
Further, it explains the comment about preferential voting being preferable to "election by plurality." If a majority is not required, it is election by plurality. But that is quite clearly not what they are thinking of. They are saying -- look, this is interpretation, not what I'm insisting upon in the article --, "Instead of simply abandoning the principle of majority, use some form of preferential voting; these are more efficient at finding a majority. But, of course, no voting method can guarantee a majority in a single ballot, so repeated balloting may still be necessary. Want to avoid repeated balloting at all costs? We don't recommend that, but if you must, then you may add a bylaw allowing plurality election, or implement preferential voting with rules that elect by a 'last round majority.' But, please, don't say we told you to do it."

(ec, the comment above was added by Richie while I was writing this response.) First things first. We cannot come to consensus if you are edit warring, Rob. I just reverted your last edit to the lead, which was incorrect, and you are clearly pushing your political agenda, and you are formally COI, not even marginal. In order to avoid crossing the 3RR "bright red line," I'm asking you now, or any other editor, to undo my subsequent reversion of myself. We can continue to work on compromise language here, but, absent what I've just asked for, I will be taking this to WP:AN/I for administrative response, or possibly to the administrator who previously blocked you. The guideline is clear, and you are breaking it, and you could once again be blocked. Do not edit the lead. I would say "except to make noncontroversial edits," but, if I assume good faith, you may not be capable of recognizing what is controversial. For others reading this, User:RobRichie is Rob Richie, the Executive Director of FairVote. His participation is welcome here, but not to the extent of making it necessary for me to cross the 3RR line to maintain article neutrality. Additionally, it should be known that User:Tbouricius is associated with FairVote as well, and has been a co-author with Richie. Mr. Bouricius began editing here in connection with edit wars being conducted on this article by Richie, as an IP editor, and he was blocked at the same time as Mr. Richie. In the hopes that he, with legislative experience and not being as single-focused as Mr. Richie, I intervened with the original blocking admin to support lifting Bouricius' block; he declined, being rather peeved by the prior behavior, it would seem, but then Newyorkbrad lifted the block, and my intervention was cited. I mention Bouricius here because in the current mixup caused by the merge result of the AfD on the controversies article, he has somewhat behaved as the COI SPA that he is, and some of my limit of 3RR today came from reverting him.

Now, as to the substance. I am not describing, nor does the reference to RRONR in the lead, actual practice, except in the sense that RRONR is entirely about general practice, but preferential voting in NGOs is not general practice, it's a special and unusual practice. Nor have I implied that RRONR says that "elections should fail" if no majority is found. That's true, i.e., that is what just about any parliamentarian would say, but it isn't what the source says. The language that Mr. Richie has altered is RRONR's language, summarized in a hopefully neutral way. If something is there, in the source, that has been omitted in my summarization, by all means, put it in, but please don't make up what you think RRONR is saying. The language which Mr. Richie has twice attempted to distort through synthetic insertion does not say that an organization is "violating Robert's Rules" if it elects by plurality, but, indeed, it is, unless the organization has a bylaw permitting it. The practice that Richie is referring to does not contradict this. Essentially, if an organization implements Robert's Rules, and then adds that a final round majority, disregarding exhausted ballots, will pick the winner, that's a mathematical certainty to complete, and election by plurality has thereby been implemented.

But if they simply state, "preferential voting by the method described in RRONR," they would not have implemented election by plurality, they would have done something *better* than plurality. That's what RRONR also says, and that is exactly what it means. And part of being better is that if a majority of those participating in one poll has not agreed on a candidate, that candidate has not been elected.

No amount of smoke and handwaving from the FairVote editors can change this. The spin stops here!

Abd. I don't see my interpretation as spin, obviously. I see this as reality. And nearly all (indeed all that we know of) of the NGO's relying on Robert's Rules when choosing to use IRV-style preferential voting read it the way I read it. If your reading of it is so objectively true, why don't others seem to read it your way? Might it just be possible that my reading isn't spin, but a very fair way to read it?
Anyway, moving beyond charges of "spin", etc (and yes, I know you believe that's what I do, when I see this as defensible differences of opinion about the same facts), I didn't just revert your edit of my edit back to your edit. Rather, I simplified it to quoting straight from the manual without your "spin" or my "spin." People can read the text themselves -- it's sourced. But to put your highly disputable interpretation in the lead of the article does not seem the correct resolution. How about letting the RONR text speak for itself?
RRichie (talk) 02:50, 26 May 2008 (UTC)
Richie, your argument isn't related to the issue. The issue is what Robert's Rules describes. Does it describe electing by a plurality? Do you have any evidence that these NGOs have interpreted RRONR as you claim? By that, I'd mean parliamentarians working for them. Sure, it is terribly easy for someone not familiar with parlipro to make that misinterpretation. What you are doing is, apparently, claiming that because organizations decide to put election by plurality in their bylaws, for convenience, this is what RRONR recommends? For IRV, as it has almost always been described, *is* election by plurality. If the majority requirement is left in place, it is not election by plurality, it is a better procedure that may avoid runoffs or repeated balloting, by finding majorities otherwise hidden underneath the more standard vote-for-one information.
What I'm insisting on is that the paragraph state what Robert's Rules says on the topic, clearly and succinctly. It could be done with even fewer words, but every time I attempted to do so, the FairVote editors would take it out as my interpretation. That's why we have an almost-quote, it's been the only way to unquestionably get it past you and your friends. And what you wanted to add was not from the source. You implied in that inserted text that election by majority would require some special bylaw. No. If an organization has adopted Robert's Rules, election by plurality is *prohibited.* It states that to use preferential voting, a bylaw is necessary, but it does not state the content of that bylaw. So if the bylaw simply read that preferential voting is to be used as described in Robert's Rules of Order, Newly Revised, 10th Edition, pp such and such, what would we have? It is crystal clear. It would have already been clear to a parliamentarian, but the voter education reference makes it openly clear, no expert opinion is needed. A majority is required. Don't want to risk runoffs: you'll have to make a bylaw for that, and it is easy for those supporting IRV to slip one in that never actually says "we won't wait for a majority," because most people just don't realize the implications. All the bylaw has to do is describe the single transferable vote process and exclude the exhausted ballots from the basis for the majority. And, I'm sure, that's exactly what has been done. But this was not the RRONR recommendation. Period. It was a method that allows election by plurality. It's obvious. What if every voter declined to add any ranked votes? Simple: exclude all but the top two, and presto! a majority (or a tie). RRONR is very clear what would happen with plumping like that: election failure, repeat the election. (unless one gets a majority, of course)--Abd (talk) 05:06, 26 May 2008 (UTC)
Instead of taking my suggestion that he undo my revert of myself, Mr. Richie continued to argue, here and on his Talk page, that he was right and I was wrong. He's missed not just one but several points, starting with COI editor! Yo! Do not edit the intro to this article. He did make what I could only assume is a good faith attempt to find compromise language by allowing him to completely miss the issues, and he still left out material he's been struggling mightily to exclude. On the possibility that he did not understand the request, I was suggesting that he look at History for the article, and next to my last edit, click on Undo. That edit was my undoing of my own reversion of his edit. I did that because the fuss here had caused me to reach the 3RR limit, and, when you do that, the advice is to revert yourself. Reverting yourself is also a move in WP:BRD, a method of dispute resolution. In any case, I shouldn't have to push 3RR with an edit-warring COI editors. Nobody is supposed to edit war with anyone, but COI editors are violating guidelines if they make *any* controversial article edits. And reverting me, as he did, was obviously controversial. And continuing to edit the lead when I'd asked him not to, because he was consistently slanting it, was also obviously controversial. There are other editors who will help, COI editors are supposed to make "suggestions," not battle for control of content. Anyway, I've now undone my reversion of myself, as I wrote I would do if time lapsed without helpful response. I won't go to WP:AN/I tonight, I'll see what it all looks like tomorrow. Wikifuss and wikidrama is a royal pain; if Richie would respect the guidelines and drop his consistent assumption that I'm trying to wreck his nice article, and notice that other editors are able to work with me, we might start to make much more efficient progress.--Abd (talk) 05:32, 26 May 2008 (UTC)


Abd.... wouldn't it be sensible to just present the information of Robert's Rules in the introduction without your spin? I was trying for a compromise -- not just reverting to my suggested edit, but not to the way you want it. What was factually wrong with my suggested edit? Nothing, I would argue. You want it just the way you want it without trying to compromise.
The simple fact is that you are suggesting that Robert's Rules is saying that elections with IRV "fail" if the final round majority winner does not have a majority of the initial round of votes. I think that is disputable and think the text supports my view. And I think the fact that all the NGO's we know of using IRV don't have your "failure" trigger suggests that I'm on solid ground with my interpretation. But for you, it seems to be "my way or the highway", which I think unfortunate and certainly not in the spirit of trying to reach consensus. Instead you just keep falling back to saying "COI editor" and interpreting that to mean you you have no need to try to reach consensus. That seems unfortunate.
RRichie (talk) 13:05, 26 May 2008 (UTC)
To your first question, certainly, Rob. That's what I've insisted on, in fact. The word "fail" is not in there. What is in there is exact language from Robert's Rules. Haven't you noticed that? Robert's Rules is not saying *anything* about "elections with 'IRV'". It is describing a method which can be used that uses the same STV technique as IRV but which *obviously* may "require the election to be repeated" if truncation causes majority failure. What's unclear about this? What a few NGOs do isn't actually relevant, unless you can show reliable source that experts interpreted Robert's Rules differently from the plain language of the text. Some organizations want to avoid runoffs. So they allow election by plurality, whether straight Plurality voting or Preferential voting with no majority requirement, or some other technique. What does this have to do with what RRONR describes? --Abd (talk) 18:58, 26 May 2008 (UTC)

I edited the porition under discussion. RONR gives an example of preferential voting without a majority requirement on page 413 ("bottoms-up" variant), so clearly acknowledges that preferential voting can have varrying forms and thresholds, DEPENDING ON WHAT THE BYLAWS STATE. Since RONR also states that the use of preferential voting MUST be authorized in bylaws, it is also reasonable to expect an organization to be clear in its bylwas whether an election can "fail" due to lack of a quorum, lack of a "majority," etc. and most critically, what to do if there is a failure. Tbouricius (talk) 16:33, 26 May 2008 (UTC)

No. Page 413 does not exclude the exhausted ballots. We've been over this before, and the problem is that RRONR isn't completely explicit here. The key phrase is this: "... until one pile contains more than half the ballots, the result thereby being determined." (p. 412). Half the "ballots," not half the "ballots containing votes for candidates not eliminated." The next paragraph, p. 413, states language that could then be thought to contradict this gloss: an exhausted "should not be placed in any pile, but should be set aside." It does not explicitly state that these ballots are of no further effect, but I can see how someone who has been thinking of IRV for years might think it is implied. However, the teller's report consists of the number of ballots in each pile for each round. The basis for majority would be the number of valid ballots counted in the first round, plus illegal votes counted at that time.
Robert's Rules sets up the procedure that, in recording the votes cast, the tellers count not only valid votes, but also illegal votes, such as illegible ballots or overvotes. "Illegal votes cast by legal voters ... are taken into account in determining the number of votes cast for purposes of computing the majority." RRONR, p. 402. Using preferential voting doesn't change this one bit. The practice of political jurisdictions, the practice of NGOs that have decided that efficiency is more important than majority rule, is all irrelevant. (Note that Plurality *usually* comes up with the same winner as IRV or other preferential systems. So the trade-off isn't necessarily a bad one. Usually.) The issue here is what RRONR describes, and what would happen if this is adopted without some other modification. Very, very clearly, a true majority is required. If not, RRONR is setting aside what it has elsewhere called a fundamental principle of democratic process, and without even mentioning that they are doing so. And, of course, the comment that the election "will have to be repeated" then makes no sense at all.
Mr. Bouricius, please don't edit the lead. You've shown again and again that you are affected by your affiliation, that you, as an SPA and probably COI editor, are inclined to maintain FairVote positions. Definitely, in the past, you have shown yourself more willing than the socks and Richie to compromise, but we have a lot of work to do on the Controversies section, and this RRONR thing in the lead is distracting us. What is there should be what is in RRONR, and any summarization should be a matter of agreement and common consent. But I'm not going to allow summarization to make it unbalanced as a presentation of the RRONR position on "preferential voting." I have *not* inserted my interpretations, I've just acted such that the text is represented accurately, without interpretation. What I do of interpretation here is by way of background.
Mr. Bouricius' last edit to the lead, which I have reverted, added what I here put in italics: "The manual goes on to note that if voters don't rank enough candidates, this may prevent any from receiving a majority, if bylaws do not allow a winner with a majority of merely unexhausted ballots, and "require the voting to be repeated."
You made that up. It's true, i.e., the conditional clause is, in itself correct, and the bylaws could provide for that (they can provide for anything, bylaws abrogate the Rules of Order, generally), but Robert's Rules does not note that. You did. It is entirely unnecessary, a slant, synthesis, not faithful to source, and, apparently, seeking to create some slant in the reader's mind. With no bylaw other than the implementation of what is in Robert's Rules, exactly as stated, a majority is required. That bylaws could state otherwise is moot. They could state whatever they want. If the bylaws allow such a last-round winner, however, the method is not what RRONR is describing as "better than a rule that plurality shall elect," that is plurality, merely a more sophisticated form of plurality. It is election without obtaining a majority of ballots cast in favor of a candidate. Shuffling the votes around doesn't change that.
Remember, what Roberts Rules is trying to do is to provide a better means of avoiding runoffs than by simply allowing a plurality to elect. But it is not, and never states that it is, recommending or allowing such an election (by plurality of votes, no matter what method is used to analyze the votes).
As an aside, RRONR would clearly disapprove of Range voting, as normally presented. If an Approval cutoff were included, and if the winner were required to have a majority of voters approving, RRONR might be brought to accept it, because then there could be majority vote expressed. And where that majority was not clear, there would have to be a runoff. So multiple candidates receiving a majority might possible require a runoff, though there is reasonable contrary opinion, and that is a whole other question, I'm just mentioning that the majority problem is not unique to IRV, for sure, and that there is *no* method which can guarantee a majority without coercing voters or depriving them of the right to essentially vote Yes or No to any question. In a multi-candidate election, if a candidate receives a majority of valid ballots as votes, then those voters have certainly implied a Yes vote for election, so majority rule is satisfied. (RRONR's inclusion of illegal ballots in the basis means that RRONR allows NOTA votes. No election method can satisfy majority rule, guaranteed, in a single ballot. Indeed there is at least one Range Voting advocate who is strongly opposed to the concept of majority rule, believing that overall social utility as shown through Range votes should prevail. I don't agree. The majority must always consent, if we are to have democratic process. In a healthy society, the majority will generally consent to maximized social utility, and I prefer not to delegate to any method the power of decision of what is truly overall maximized social utility. I'm not a technocrat. Again, a whole other issue, I'll repeat, an aside here.)
There have been RfCs before on this, and they quite simply don't match Richie's opinion that I'm out on a limb here. The text is clear, and others have agreed. The only exceptions have been Nrcprm2026 sock puppets, Mr. Richie, Mr. Bouricius, and sometimes, Tom Ruen, as I recall, he tends to waffle about. All of these are affiliated in some way with FairVote --Abd (talk) 18:58, 26 May 2008 (UTC)
Abd is wrong that the text of RONR is clear in supporting his interpretation. I believe Abd has missed my reference to RRONR giving an example of preferential voting WITHOUT a majority requirement on page 413. This non-majority example is in the paragraph BEFORE the one with the language he quotes. RRONR gives an example of using the single transferable vote to determine winners WITHOUT a majority threshold using "bottoms-up" IRV. So, yes, RRONR generally prefers majority thresholds, and arguably, MAY prefer not to discount exhausted ballots. But that is not at all certain, as ballots with blanks or abstentions are NOT to be used in the denominator to calculate a majority (see RRONR definition of majority on page 387) and exhausted ballots are widely (though not by Abd) viewed as abstentions on the question of which finalist is preferred. Also, the fact that the key sentence about "requiring" a "new election" has the word "may" in the lead, makes it ambiguous what is meant. The quotation Abd is choosinbg to insert in the introduction creates the false impression that RRONR insists that a majority (defined by including exhausted ballots in the denominator) is the only fully propper threshold, but this ignores three facts -- 1. the fact that RRONR also gives a non-majority example, 2. that RRONR says to discount abstentions in calculating a majority, and 3. that the word "may" makes the requirement less than certain and subject to bylaws. Therefore it is appropriate to make clear that the threshold should be specified in the bylaws, as RRONR would advise.
I will let Abd and others ruminate on this before I edit the section to remove Abd's layer of interpretation with his obsession on the majority issue. The key fact is that this piece is still too long for the introduction (and essentially repeats what is in the later section on non-governmental organizations). I would hope a short neutral sentence can be substituted as I tried to offer previously. Here is another attempt at offering such a compromise.
'"The single transferable vote methodology used by IRV is presented in Robert's Rule of Order as an example of preferential voting."'
Tbouricius (talk) 01:35, 27 May 2008 (UTC)

Nope. The appearance of recommendation is preserved without the modifying negative comment. Now, as to duplication of what is later, that could be fixed. The following is the paragraph as it is now. What part of this is my "layer of interpretation?

Robert's Rules of Order calls preferential voting "especially useful and fair" when more than one ballot is impractical, such as elections by mail. "In such cases it makes possible a more representative result than that under a rule that a plurality shall elect." "Preferential voting has many variations;" the single transferable vote technique used by IRV is the example given. The manual goes on to note that if voters don't rank enough candidates, this may prevent any from receiving a majority and "require the voting to be repeated." "Although this type of preferential ballot is preferable to an election by plurality, it affords less freedom of choice than repeated balloting, because it denies voters the opportunity of basing their second or lesser choices on the results of earlier ballots, and because the candidate in last place is automatically eliminated and may thus be prevented from becoming a compromise choice."[2]

This is almost entirely exact quotation. If any of the small amount of summarizing text in there is incorrect or distorted, please, point it out and I'm sure we can fix it. I don't see any. What Bouricius wants to do is to remove exact quotations, while he is calling it removing my "interpretation," and he then substitutes total interpretation. It's accurate, that's true, but it has turned a balanced examination in a manual of procedure into an apparent recommendation that has already, elsewhere, been twisted into a justification for eliminating the majority requirement, which is the exact opposite of the RRONR principles and intentions.

Now, as to interpretation of the text. We really aren't allowed, strictly, to "interpret" except by consensus. That is, if we can agree on an interpretation of the text, we can use it, at least until someone else objects. Given that the paragraph as it is now does not interpret, it presents, mostly through exact quotation, the interpretation argument may be moot. But I'm going to answer, anyway, about interpretation, hoping that this doesn't distract from the simple clarity of the text.

Abd is wrong that the text of RONR is clear in supporting his interpretation. I believe Abd has missed my reference to RRONR giving an example of preferential voting WITHOUT a majority requirement on page 413.

No, I didn't miss it. I, in fact, referred to this section. I noted that the language here was not crisply explicit. It is an interpretation that this is an election allowing a plurality winner. But it does not actually say that, nor does it specify the basis for majority, it just says "majority of the ballots." Given that this is what is always the basis for majority in Robert's Rules, and it takes a special interpretation to make it refer to "majority of the unexhausted ballots," I claimed before that Bouricius' interpretation was contrary to the general principles of parliamentary procedure; but then I noticed the mention of the election having to be repeated if ballots were exhausted.

This non-majority example is in the paragraph BEFORE the one with the language he quotes. RRONR gives an example of using the single transferable vote to determine winners WITHOUT a majority threshold using "bottoms-up" IRV.

I don't see that, and I'm looking again and again at that page. There are line numbers on the page. Where, pray tell, does the text revert to "without a majority threshold"? What line numbers and what language?

So, yes, RRONR generally prefers majority thresholds, and arguably, MAY prefer not to discount exhausted ballots.

It is crystal clear and explicit that if a ballot contains what may be a vote, even if illegible, it is part of the basis for a majority. How much more is it so if it contains a vote for a known and accepted candidate. Did Bouricous read what I wrote? Indeed, perhaps I write too much, but, I'm sure, it takes less time to read it than to write it! To convert clear and precise prescription into a vague "generally prefers" and "may prefer" is pure political spin. Bouricius was a politician, a Vermont legislator. Experienced.

But that is not at all certain, as ballots with blanks or abstentions are NOT to be used in the denominator to calculate a majority (see RRONR definition of majority on page 387) and exhausted ballots are widely (though not by Abd) viewed as abstentions on the question of which finalist is preferred.

No, RRONR is quite clear. Ballots that are blank are abstentions and are not counted. "Scrap paper" is what RRONR calls them. But an illegal ballot is to be included if the voter was legal but the candidate wasn't or was illegible. I quoted the definition above!

And what is "widely viewed" isn't relevant to what RRONR describes. Quite simply, they are two different things. It is correct that such a ballot is equivalent to an abstention on the question of which candidate is preferred, but to take, for example, an RCV ballot from San Francisco, with three ranks, and a voter ranks three out of over twenty candidates, and none of them are the top two, and call that an "abstention" is preposterous. The voter voted against both of the top two. Robert's Rules -- and the Vermont Senate -- affirm the necessity in a democracy that decisions be made with the explicit consent of a majority voting, and RRONR makes it very very clear that a ballot with any kind of mark on it at all that might be an attempt to vote is considered a part of the basis for the majority. Not to mention exhausted ballots which have votes for actual candidates on them. but merely the "wrong" ones.

Also, the fact that the key sentence about "requiring" a "new election" has the word "may" in the lead, makes it ambiguous what is meant.

It is not ambiguous at all. My, my, Bouricius is getting inventive! If I dared to even look like I was doing something like this, some months ago, he'd have been all over me with derision. I wonder! Is it deliberate that Bouricius did not quote the relevant sentence, could it be because if he did, the argument falls to pieces?

Here is the "key sentence," [if a voter declines to indicate additional choices], "it may prevent any candidate from receiving a majority and require the voting to be repeated." The "may" does not refer to the repeating of the voting. The word "require" is used for that. It is "may prevent," not "may require." And "may," there, is a reflection of the fact that usually such does not cause majority failure, unless there are many candidates and many voters who don't add additional preferences.

Really, Mr. Bouricius, this may all be examined later by ArbComm, if this nonsense seriously keeps up. Do you think that, in the much more deliberative environment of an ArbComm procedure, that statement would be seen as anything other than an attempt to maintain a biased position?

The quotation Abd is choosinbg to insert in the introduction creates the false impression that RRONR insists that a majority (defined by including exhausted ballots in the denominator) is the only fully propper threshold, but this ignores three facts

RRONR doesn't insist. It prescribes. Organizations can do what they like, but if they decide to implement Robert's Rules of Order as their rules of procedure, majority means what they have defined it to mean, and it isn't ambiguous. They do not describe any other threshold.

1. the fact that RRONR also gives a non-majority example,

If this is a "fact," please quote it. I'm suspecting that Bouricius is relying, for this "fact," on some possibly ambiguous language on p. 413, but he has yet to be explicit, all the while insisting that he's proven his point.

2. that RRONR says to discount abstentions in calculating a majority,

Sure. And it defines what an abstention is, and what it is not. And I already quoted that, and Bouricius has totally ignored it.

An absention is the absence of a vote of any kind, other than "I abstain," if we want to call that a vote. Since we are talking about written ballots, an abstention is a blank ballot. Above, this is what I wrote: Robert's Rules sets up the procedure that, in recording the votes cast, the tellers count not only valid votes, but also illegal votes, such as illegible ballots or overvotes. "Illegal votes cast by legal voters ... are taken into account in determining the number of votes cast for purposes of computing the majority." RRONR, p. 402. Using preferential voting doesn't change this one bit.

Illegal votes don't count for any candidate, so they don't express a preference between any. If these are still included in the basis for a majority, how much more would we think it appropriate to include ballots which *do* express some specific preference that is counted, even though the candidate is later eliminated.

and 3. that the word "may" makes the requirement less than certain and subject to bylaws. Therefore it is appropriate to make clear that the threshold should be specified in the bylaws, as RRONR would advise.

This is even more confused. The "may" is referring, quite clearly, to the conditional and uncertain nature of the effect of truncation... Not at all to the repeating of the election; the word "require" is used there. Truncation may cause majority failure and require the election to be repeated. "May" is not attached to "require," but to "cause." Majority failure requires repetition of the election under Robert's Rules, and this is exactly what is standard. They don't mean a runoff. They mean a new election.

Absolutely, organizations can change this if they don't like it. But *nowhere* does Robert's Rules recommend that, and they argue against it, in fact. It is very, very clear that if there is no specification otherwise in the Bylaws, a majority of those voting is required to elect, by Robert's Rules. So what Robert's Rules describes is preferential voting with a majority requirement for election, recommended, I'd agree, over the more common decision: dropping the majority requirement and making a rule that "a plurality shall elect." If an IRV election is allowed to elect without finding a majority, it is election by plurality.

Here is a simple example. There is an election, three candidates, IRV ballot. But every voter plumps for his favorite, and none gains a majority. What does IRV do? Pretty clear, if it considers only the last round majority, the candidate with the most votes wins. Plurality. If it is Robert's Rules, new election. If it is top-two runoff, the two candidates with the most votes face each other in the runoff. Now, which of these is the most democratic, next most democratic, and least democratic?

Robert's Rules prescribes the most democratic solution. Top-two runoff is a compromise, and IRV is really a plurality method.

(If write-ins are allowed in the runoff -- and it seems they often are -- it really is a new election, it's just been set up in a way that makes it less likely to again experience majority failure. And probably it doesn't require a majority, given that most organizations, by this time, really want the damn thing to be over with. Still, majority failure is rare, I suspect, in top-two runoffs.) --Abd (talk) 04:07, 27 May 2008

[edit] strange edit

When I clicked on recent changes I saw this edit. Can someone explain this. If not, I will revert. QuackGuru 22:47, 26 May 2008 (UTC)

Thanks. I wouldn't call the edit vandalism, though it was consistent with one side in some disputes that have been going on here. The editor has 17 edits since June 2007, nearly all in the Voting systems area, and did previously edit this article in March this year, adding a link to a local IRV organization (probably should be removed, I'll look at that.) I already addressed the content of this edit and wrote a gentle request to the user on his Talk page to be careful about removing sourced content. Particularly as an SPA with an easily inferred bias. He's welcome to contribute here, as far as I'm concerned. Nice to know that someone else besides us chickens noticed the removal of sourced content.--Abd (talk) 23:41, 26 May 2008 (UTC)


The "strange edit" person here. I've been watching the IRV page for some time and have agreed with nearly every edit, but Abd's recent editing of the IRV vs runoffs section struck me as very odd, so I thought best to revert most of it to the previous version. I thank Abd for discussing this with me in a professional manneer.
My biggest problem was that Abd's edit removed all mention of potential "drop-off" in the first round of two round systems. On my User talk page, he rejects the use of the word "drop-off" in that context, which is a fine semantic point that I won't argue with. We can choose a different word, but the point remains that participation spread across multiple rounds is potentially less than that which might be achieved in a single round. Some voters may fail to show up for the runoff (aka "drop-off"); others may skip the first round (aka "some term that means first-round drop-off"). Either scenario is a potential benefit of IRV over two rounds systems due to the single ballot.
The data on Cary doesn't dispute this point. In fact, it actually supports it. If participation goes up significantly in the runoff, then there were many interested voters who skipped the first round, bringing into question the "legitimacy" of the first round. That was exactly the text that was removed, and that's the claim being made by advocates of IRV. I'm fine with restoring the content on Cary, but it shouldn't be presented as if it's negating the claim being made. Then again, any data that shows a significant difference in turnout between the two rounds would support this point, so I would think it odd for the article to single-out Cary.
Abd also removed the note that the winner of the second round of voting may receive less votes than a loser of the first round, because it was not sourced. First of all, the article said "may", so I don't see why a citation is necessary for something that is clearly possible. Second, it has in fact happened; for example, in the 2001 race of City Attorney in San Francisco, where Herrera (the winner) won fewer votes in the second round than Lazarus did in the first. Progressnerd (talk) 01:09, 27 May 2008 (UTC)
That runoffs held as special elections often have lower turnout than "primaries" held at a general election is common. That special elections usually have low turnout is commonly understood. However, what Cary had was a primary held as a special election and a runoff held with the general election. Though, in fact, it was an off-year general election, so it doesn't necessarily have high turnout in either case. The problem with this whole section is that, attempting to make desired points, it seems, arguments started to be synthesized. I've taken a lot of flack for allegedly putting original research in articles, and here we are seeing it coming from another direction. I've never seen a reference to "drop-off" in a primary, and it seems to me that the term arose because some kind of point is being made that there is a turnout problem no matter what you do, if you have two elections. I can, in fact, make the reverse point: suppose there is a party primary with six candidates. Two of them make it to a runoff. Both of these candidates are about equally acceptable overall to the party voters, many of them really don't care which of them it is, they will support either. Turnout will be low. Is this a problem? Now, take it in reverse. Primary with six candidates, none of them very well known, and there isn't a lot of campaign money being spent. Turnout in the primary is low. Two candidates left in the runoff, now they know they have a chance, their supporters get fired up, turnout improves. Is this a problem? (This is so-called "drop-off" in the primary.) What I've been asking for, now, is *sources*. What arguments are notable here?
Meanwhile *sourced* and *relevant* information is being removed. Bouricius' last edit summary:
(clarified that drop-off can be in either round, and removed Cary and SF, as they do not offer any counter-example.)
First of all, this argument that there is a "drop-off" problem with holding an early primary, runoff if needed with the general election seems to be entirely new, I've not encountered it. People do have an idea that runoffs have low turnout, and MilesAgain ridiculed the idea that it could *ever* be different. Cary is an example of it being different. And the San Francisco reference, which has been removed once again, is about an *argument.* You know, what this whole section is supposed to be about. What *arguments* are being notably presented.
Ah! I get it! You guys think we are arguing about whether IRV is a good method or not! This is a total misconception of what we do in this article.' We, properly, are reporting about the public debate. We cannot, here, introduce new arguments. Because some of us have written about what we know is being presented in the debate without sources, it may have seemed to some that we were, ourselves, arguing. That's not what we are supposed to be doing in the article. This is not a section about the characteristics of IRV, as such. It's about the debate, first. What arguments are presented. And then what relevant, sourced facts can we present to illuminate this. Speculation isn't a part of this. We don't try to resolve the debate here.
This particular section is problematic because it is about IRV finding majorities, which is definitely a claim made for IRV. Eventually we come to the problem text: This is what Bouricius just put in:
In comparison, a two round system (with a majority requirement) allows interested voters to participate in the runoff round, and the result satisfies requirements of election by majority because the number of votes cast in the first round is not considered when calculating the majority threshold for the second round. However, there may be a significant drop-off in voter turnout in either the first or second round of a two round system, so the winner may not win a majority of those who participated in one or both rounds.[10] It is possible that the nominal "majority" winner of the second round may receive fewer votes than the loser received in the first round of voting,[11] which is not possible with IRV.
The runoff round, with two candidates, satisfies majority requirements as they have been understood for centuries. Whenever a vote on anything is redone, the number of voters in the previous vote is not relevant. More, less, the same; the same voters, an entirely different set of voters, it is all moot. There is a properly called election, and whoever votes votes, and "majority" means a majority of valid ballots cast. But because some point is desired here, is being pushed, an entirely new line of consideration is introduced. It's worse here, but this is really the same with the FairVote argument that I have seen, which is the last sentence of the paragraph. The first part is a complicated way of saying that the second round is a new election. All the rest of it follows from that. The number of votes in the first election has no bearing whatever on the second election. Voters in the second election can vote the opposite of how they voted in the first. Looks to me like Mr. Bouricius is trying to make it complicated when it is simple.
So I'm completely disallowing that first monstrosity of a distorted explanation of why the second round satisfies the requirements of election by majority, because it's all unnecessary. It satisfies the requirements of election by majority, normally, because there are only two candidates. Period. The phase of the moon doesn't matter, nor does the number of votes in the first election. It is a runoff *election*, and voters in it are completely independent from the first election (and in many places, they can even write in new names if they want). The majority in an election is determined by the number of voters voting in that election, and you can't compare this with the previous election, partly because you don't know which votes are what. Two elections, 1000 voters each. What's the total number of voters voting? At least 1000 and not more than 2000. Not terribly useful information.
Now, as to the second part, the comment that the number of voters for the winner in the runoff can be less than the number of votes for the loser in the primary. I'll agree that it's true, but is it notable? My own answer is yes, but why it is notable isn't being recognized in the text above. It is notable because this is an argument being presented by FairVote or FairVote supporters. If you can find reliable source for that, put it in. I didn't find it right away, and because it is spin and therefore *must* be attributed, I took it out.
You also removed the attribution to FairVote of the study. You are replacing sourced material with unsourced POV argument, without consensus, edit warring over it.
This whole thing stinks, Mr. Bouricius. Stop it. Because Richie stopped reverting, I'd hoped we could move on. Apparently not. Give it some consideration, I'll look again tomorrow. Do understand that when admin attention comes here, it could get fairly ugly. I'm not looking forward to that, and it can be pretty unpredictable, but, since I'm prepared to go to ArbComm over all this if necessary, I will do what is necessary to prevent COI/POV domination of the article. --Abd (talk) 02:39, 27 May 2008 (UTC)

[edit] Proportionality and IRV

Abd. You are citing the Jenkins commission to support your argument that IRV isn't friendly to third parties. IRV of course isn't going to help third parties win when they can't win majorities, but that's not the point the Jenkins commission makes. Note that the Jenkins commission recommended IRV for the great majority of seats, by the way -- a very important point that your selective use of it would not suggest possible -- but with "add on" seats to ensure more proportionality.IRV along as a winner-take-all system will not necessarily provide proportionality.

If you read the section of the Jenkins commission report that you've cited, note that it says this: "The Commission's conclusions from these and other pieces of evidence about the operation of AV are threefold. First, it does not address one of our most important terms of reference. So far from doing much to relieve disproportionality, it is capable of substantially adding to it. Second, its effects (on its own without any corrective mechanism) are disturbingly unpredictable. Third, it would in the circumstances of the last election, which even if untypical is necessarily the one most vivid in the recollection of the public, and very likely in the circumstances of the next one too, be unacceptably unfair to the Conservatives."

Not a word about third parties. Rather, it is exactly a commonly made critique against the French runoff system -- it often results in highly disproportional outcomes for the major parties.

As a general point, by the way, the Australian Greens aren't against IRV compared to winner-take-all systems. They indeed use it for their internal elections, as do many parties. Rather, they would like PR -- who wouldn't if your national share of the vote is 8%. RRichie (talk) 11:35, 27 May 2008 (UTC)

As an important addendum, I think it quite important to note that the Jenkins commission recommended IRV (called "AV", the "alternative vote" in this report) for the great majority of seats. For more on this, see the cite mentioned above, along with this one, sections 125-131:

http://www.archive.official-documents.co.uk/document/cm40/4090/chap-7.htm

RRichie (talk) 11:48, 27 May 2008 (UTC)

Remember, the subsection is about the effect on third parties. The Jenkins report, in what you quoted above, indicated that AV was "capable of substantially adding to" disproportionality. In context, that means harm to third parties, and possibly to proportionality between the major parties as well. I'm happy to see balanced quotation from the Jenkins report. What I took out was reinterpretation of the Jenkins report to make it appear to be recommending "proportional representation," which, in the context of our article, would be a system of PR that they aren't comparing AV with. Find a balancing quote from the report, put it in, I expect I'll be fine with it. But, given the history, I'll ask you to put the language here. You aren't very good at avoiding controversial edits to the article. Avoid "interpreting" the report, unless you are sure that we'll agree on the interpretation.
Most election reformers in the U.S. would agree that PR would be a major improvement. STV is quite a reasonable way to accomplish it, with the right vote deweighting. There is a better and simpler way (simpler as to what takes place at public expense, but more complex in that it isn't a mindless translation from ballots: Lewis Carroll's invention, rediscovered by Warren Smith in 2000, called Asset Voting. To Carroll, it was simply a way to do PR. --Abd (talk) 12:44, 27 May 2008 (UTC)
Abd -- Jenkins is focused most decidedly about disproportionality for the major parties. Please reread it - you're the one who is interpreting it here and doing it inaccurately. Note that Jenkins recommends IRV as a general method for the great majority of seats, and then a top-off, semi-proportional representation system to provide more proportional representation that no winner-take-all system is likely to provide.
RRichie (talk) 23:03, 27 May 2008 (UTC)

[edit] Criticism section

In section 8 "Is IRV better than other winner-take-all systems?" there are pro arguments and then con arguments. I think it would make more sense to do what many other articles do and just create a "Criticisms" section that includes criticism and rebuttles. Or, better yet, create an arguemt section for IRV and an argument section against IRV and leave out the intro. I think the intro just makes the whole thing more confusing. QuirkyAndSuch (talk) 03:59, 28 May 2008 (UTC)

You suggestion makes sense to me. I think the whole section is far too long. Abd inserted the content after it was decided it didn't work as a stand-alone article. He opposed tbouricius' suggestion of making it much shorter. Presumably doing so would take a lot of time and editing. If someone has that time, more power to them.

RRichie (talk) 09:23, 28 May 2008 (UTC)

It wasn't decided that "it didn't work as a stand-alone article," not exactly. It was decided that the content belonged here. It's complicated, though. Many editors voting apparently thought that the content was original research, etc., and I suspect that this was on both pro and con sides. So they were thinking that the content, merged, would be much shorter after improper material was removed. My own opinion is that, with age, it will expand, not shrink. (Some voters seemed to have the position that there wasn't any controversy, that this was all made up. I think you know that's not true. There is controversy over Ware's method going back into the nineteenth century....]

So, fret not. If we tighten up the section, making it reliably sourced for everything that isn't blatantly obviously true, and it's still too long, we will then have a very good basis for moving it back to a subarticle, and then bringing back a summary here. I'm not saying it will be easy, I don't expect us to naturally agree on the summary! But I think we can get there. --Abd (talk) 23:49, 29 May 2008 (UTC)

Definitely, some of us thought the controversy stuff needed drastic shortening. Not because of a lack of controvery, but because there is a lack of reputable sources summarising the controvery. The argument that the controversy argument was a spinout was flawed because there was a lack of brief summary in the main article, and the spinout was in such bad form. The listing of arguments is not encyclopedic. It is a poor attempt to satisfy WP:NPOV. It resembles a point form draft. It resembles a summary of a debate, not of a subject. I suggest noting for each issue that “there is a debate” without regurgitating the debate. Be satisfied with linking to major proponents. I still think WP:SYN needs a strong application here. --SmokeyJoe (talk) 07:51, 30 May 2008 (UTC)
Originally, this article contained a bullet list of arguments. Given that the arguments themselves can be highly misleading, this was problematic. So I, and, I think, some others, started to expand that section to (1) source the arguments. *who* or *what organization* argued this way? Now, we do run into a notability and RS problem here, it's true. However, there was no general attempt to insert arguments that weren't already well-known, i.e., there are, almost certainly, multiple independent sources that, all by themselves, show that the argument is being made. There isn't any academic study of the arguments presented (though there is academic study used as part of arguments), but, if we look in newspapers, books, and the web pages of notable organizations, we can see the arguments being made. For example, FairVote presents arguments. They aren't RS for fact, generally, because they are an advocacy organization, but they *are* RS on the narrow point: is this argument being notably made? Likewise a letter to the editor of a newspaper may, particularly if the argument is well known and not isolated, serve as verification that an argument is being made, i.e., being made as part of the public controversy.

Anyway, making the section NPOV caused it to expand until objections arose that this was imbalanced. An experienced editor suggested the fork, and then created it. He wasn't involved in any disputes in this article; all involved editors accepted the move. Now, the plan was to bring back summarized content, not to indefinitely leave all of it there. There is no bullet list, as such, here; rather the section headings are categorized by argument. We *could* mix the arguments, and, given that there are a few cases where the *same* argument is made for both pro and con, that might work. It's also possible, perhaps, to examine issues rather than arguments, as such, but this is far more difficult and would probably require original research and lots of synthesis.

The debate has not been regurgitated. Here is the problem: these arguments, such as "IRV finds majority winners," are highly misleading (the opposite argument "IRV does not find majority winners," by the way, can be made; in fact, that's what I've found, studying real election results, but ... that is original research, to state it like that is something I've never attempted to do in the article. It's a conclusion. We have, in the Implementations article, presented the results of actual IRV elections, and the data is there, most of it, but drawing the conclusion is what can't be done. Anyway, if we say "IRV finds majority winners" as an argument, but don't examine the details, with what can be sourced, we are essentially propagating a propaganda point. To say "there is controversy over this" isn't enough if there are no details on the nature of the controversy. The reader may ask, "Well, does it?"

My approach has been to, first of all, present sourced text that makes the claim understandable. What does "majority winner" mean. There are, in fact, standard definitions, but part of the problem is that when IRV advocates say that IRV finds majority winners, they are using a different definition, a very nonstandard one, based on an unstated analogy. By the way, I'm not, here, trying to argue about IRV< but to explore the principles involved in editing the article. And I definitely am not trying to push the readers into some conclusion. I just want them to have as much of the same information as an expert would have as possible, presented neutrally. Propaganda campaigns (on all sides) tend to seek quick, sound-bite arguments, particularly ones that sound true and only start to fall apart if examined closely.

As to WP:SYN, it isn't necessary, generally, to synthesize. Synthesis arises as a problem here when an editor synthesizes in an attempt to be concise, and the synthesis is, or is seen as, biased. Classic example in this article, the former "Robert's Rules recommends Instant-runoff voting." You can find a ton of source on that, though, I don't think, none of it is academic, and it isn't coming from parliamentarians. It's coming from FairVote, but also from others who simply didn't read the handbook carefully. What Robert's Rules actually does can be presented with a few sentences of quotations, and I'm fine with that. However, FairVote editors have rather steadily and consistently attempted to whittle it down to remove detail necessary to make it a balanced report, on the argument that it was "confusing" or "too much detail." And what they keep, of course, even though the "recommends" language was dropped, implies recommendation. "Roberts Rules describes IRV as 'especially useful' etc." But is what RR describes "IRV," and is this all there is to it? And it isn't, see the current brief mention in the lead and the longer one in the article section on NGOs.

Now, right now, there is no "argument" in the article over Robert's Rules and IRV. It's a fact, as I mention above, that the claim is everywhere, presented as an argument for IRV. It's in Vermont legislation as a "finding." It's on the web sites of election officials explaining IRV. Etc. Yet what that legislative body or those election officials are implementing is quite different from what RRONR describes. It is really classic spin. Problem is, the difference is subtle. Crucial to any parliamentarian, but dismissed by FairVote editors as "pendantic," "nit-picking," and "POV-pushing."

I'd love to be able to present conclusions, but I can't. However, I can insert the very few words necessary of sourced quotations, very clear, and not controversial among parliamentarians, and apparently satisfying otherwise uninvolved editors in the two RfCs we have had over this (the RRONR issue) that it's correct and not biased, and I can, and have, done it without any synthesis at all, which means that if it is misleading, it would be a matter of being incomplete, i.e., selective quotation. But efforts to change this have only involved inserting synthesis or removing sourced quotations.

As I've stated, we can synthesize if we have consensus that the synthesis fairly represents reliable source. Finding such consensus is what is needed to make articles other than a set of quotations. It's harder when there are editors with a POV they are attached to. But it can still be done. So ... SmokeyJoe, the problems of this article aren't simple or easy to solve; if they were, they would have been solved before now. There is controversy over IRV, that is, there is a campaign *promoting* IRV, and there is resistance to that campaign coming from two different camps: one being those who believe that the status quo is generally better than IRV, and the other being those who see IRV as possibly better than the status quo, but definitely not as good as certain other proposed reforms. These other proposed reforms are notable; as an example, Approval voting, which has quite a lot of academic source, it was proposed, most notably, by an academic. Or, as a variation on this, IRV is most commonly being proposed and implemented in the U.S. as a replacement for top-two runoff. Yet much of the promotion of IRV compares it with plurality. Top-two runoff and IRV are similar in quite a few different ways, but they can produce different results, and arguments over this can be found in the media. In San Francisco, one of the arguments presented for IRV related to the lower turnout in runoffs. The primary was held with the general election in November, and the runoffs, if needed, in December. (In some San Francisco districts there were many candidates, over 20 in some cases, so there was lots of majority failure.) Opponents to the IRV measure suggested moving the primary up, so that the runoff would be held in November. I'd predict that this would result in different results than IRV, based on the performance of top-two runoff compared with the performance of IRV, but, again, I can't put that into the article, at least not here and not stated like that. But we *can* present the arguments being made, and I believe that, with good faith, we can find summary language that is acceptable. Sources are needed, but this is where consensus comes in: how can they be summarized? This summarization would also take place if we had a subarticle.

To me, though, the problem with trying to summarize *now* is that we don't have conensus on the basic material, much less on a summarization of it. To me, "There is debate over this" is a bit weasely. It's possible, though, with sources. What sources? In any case, I do not see any reason why we cannot present the debate in some detail, together with reliably sourced relevant fact. The "debate" means quotation or consensus summarization of notable arguments, preferably attributed. I've done some of this, "According to the City Attorney...." or the like.

If our goal is adherence to Wikipedia guidelines, literally and strictly, we will have, in fact, departed from the goal, which is NPOV and verifiable content, there is a reason for Ignore all rules being Rule Number One: the overall goal of the project trumps the specifics of guidelines. In the hot environment of existing political campaigns, simply reporting the most notable arguments or "facts" can create a POV bias in favor of whatever position has the most funding, the most "momentum."

In many many areas I've encountered a conflict between NPOV and Notability guidelines, extending as well into reliable source problems. The basic policy is verifiability, and it's possible to state many "non-reliable source" facts in such a way as to make them totally, uncontroversially true and verifiable. Controversies articles have, in a number of places, been a way to avoid balance problems (i.e., Notability problems).

Suppose there is a position accepted by general consensus in the scientific community, with few exceptions. Say, evolution. In an article on evolution, to present much at all about "intelligent design" creates a due weight problem. However, an article on the criticism of evolutionary theory becomes a new topic, one within which intelligent design is notable enough to be presented more completely. That article must itself be NPOV, it can't just be a collection of arguments why evolution is a Marxist conspiracy, or whatever. Controversy is its own topic, and it would ordinarily include both sides of the controversy, i.e., there are creationists and there are scientists and other who argue against the creationist arguments. The article on the controversy can state that the critical views are "pseudoscience," and such articles often do state that, if there is RS for it.

This was what we did with the Controversies article: create an article specifically on the controversy. So even if the most notable arguments we see in the media are pro-IRV, we can examine *all* the arguments that are sufficiently notable, on their own, without impacting due weight. The problems with the Controversies article were three-fold: it was, in some respects, poorly edited and had been worked over by editors with a POV. (pro-IRV, for the most part, by the way). Secondly, both pro-IRV and critical editors, and perhaps a few neutral, had also written from what they knew, without sources, and a fair amount of this material stood because everyone involved knew that it was true, even if not sourced yet. During the AfD, I started adding some of the sources, there is quite a bit that remains to be done. I took out one section that is really about a quite notable (and academically sourced, I think, that is, such source can be found) because it was simply too confused. It will come back, I'm sure.

The third problem was that what I'd suggested at the beginning, that summarization, per WP:SUMMARY would come back into the article, never happened. My own attention was elsewhere for some months.

One thing, though, should be clear. The Controversies article wasn't a POV fork, contrary to much opinion expressed in the AfD. There was a question asked there that nobody ever answered: if it was a POV fork, what POV did it favor? WP:FORK, as part of the definition of a POV fork, refers to intention, and that improper intention wasn't there. It's possible, I'd grant, that the IRV supporters were happy to get the critical material, which is less than half of the controversies material, out of the article, but that wasn't going to be sustainable. It was going to have to come back in summary, eventually.

By the way, fresh editors who would help with this are quite welcome. Pro-IRV, critical, or neutral (which often means "ignorant," but that is helpful too, ignorant people may ask good questions, such as why the emperor has no clothes). If a section is wrong, take it out. If it is true, but needs source, or if you doubt it but are willing to assume good faith, tag it. If it needs fixing, fix it. If "neutral" is willing to learn -- which might end up creating a POV! --- then it will all help. I generally interpret Neutral Point of View to be a matter of consensus, it is *not* the POV of any editor. I've come to think that IRV is a terrible method, actually, but that doesn't prevent me from acknowledging whatever is factual about it. It would, for example, deal with the Spoiler effect, which I would normally consider a good thing.

This is what I suggest, and what I hope we can agree on. Let's continue cleaning up the section in this article, either sourcing or removing unsourced material, making it balanced, etc. This *may* result in increased size, not decreased, we won't know until we have done it. To the extent that we can agree on balanced summarization of sources, with reference to the sources, it may become less text. Then, if at some point it becomes apparent that a legitimate fork is needed, we can either just do it, or we can go to DRV. The AfD consensus wasn't clear, the closing admin repeated the WP:FORK argument as if it were an established fact, even though a crucial element of that alleged fact was never provided: evidence of intention to create a POV imbalance, or, possibly short of that, evidence of actual POV imbalance not remediable by ordinary edits. So we'd have a good basis for a reversal of the Merge decision; but my opinion is that we can simply bypass that review, a merge decision in an AfD does not trump editorial consensus. It simply sets up a presumption that is rebuttable. --Abd (talk) 01:16, 31 May 2008 (UTC)

[edit] Con section on monotonicity (voting for a candidate can make them lose)

An unregistered editor shortened the explanation of the issue, making it still harder to understand (the section already has a banner about it being inadequately explained). I am restoring the explanation and adding a bit more. Tbouricius (talk) 14:04, 28 May 2008 (UTC)

I didn't see anything wrong with your explanation, Terrill. I may need to look at this questioning of the real-world issue, that's more complicated. I'll get to it, I expect. --Abd (talk) 23:44, 29 May 2008 (UTC)

[edit] Does RRONR require strict majority for preferential voting?

Editor Abd inserted this statement again in another section of the article (in the pro-con section about majority in one ballot). As I have pointed out to Abd before, RRONR is not categorical about this. On page 413 RRONR gives an example of preferential voting that absolutely does NOT have a majority requirement. In this implementation of preferential voting to fill multiple seats, candidates are removed from the bottom and the single transferable vote is used to redistribute votes "until all but the necessary number of candidates have been eliminated (that is, in this example, all but three)." It is mathematically impossible for three candidates to receive a majority. This version of preferential voting is often called "bottoms up, and is neither proportional nor majoritarian. If RRONR intended to stress the majority threshold, it could have presented another preferential voting multi-seat method in whcih the seats are filled by a majority one at a time, with the ballots re-tallied, excluding the previously elected candidates for the subsequent seats. Instead, RRONR used an examp-le of preferential voting without a majority requirement.

I ask Abd to back off his relentless editing to insert his unique interpretation of RRONR throughout the article. Tbouricius (talk) 14:37, 31 May 2008 (UTC)

Yes. Terrill has indeed attempted to "point out an example that absolutely does NOT have a majority requirement." And it is true. This time Terrill gave enough argument that I got the point. But the example is multiwinner. The question of "majority" does not arise in multiwinner elections, because factions are being represented (with the method described). That is correct, it's an example of preferential voting without a majority requirement, indeed, it has no requirement at all, no quota. It's just the top three candidates. As Terrill knows, there are better ways to do proportional representation, but that section has *no* relevance to single-winner elections. IRV is a name applied exclusively to single-winner sequential-elimination preferential voting, and with single-winner, the majority question arises, and is clear and simple (by standard Robert's Rules).
(This little disagreement is related to the mischief done by the abandonment of traditional names in favor of the neologism, Instant Runoff Voting, though this particular one would remain, because there is indeed a natural confusion between various methods of preferential voting. Preferential voting, properly and historically, refers to the ballot design, and says nothing about specific counting rules. Many methods use a preferential ballot, but in some places, a specific method is called "preferential voting," for example, Bucklin voting was known as that. In fact, nearly every method could use the same ballot. (Plurality voting: only the top rank vote is counted, ballot spoiled if more than one of them is marked. Approval voting: Same as plurality, but more than one top rank vote is allowed. Borda count: Assign each rank a value from one to N, with N candidates, add the values up. Range voting: same as Borda but N may be larger than the number of candidates, and more than one candidate may have the same rank. They *all* can use the same preferential ballot, but the precise rules and the method of vote analysis are different.)
We mention Robert's Rules because they describe a single-winner usage of sequential elimination. With that method, the one they describe, they clearly require a majority. They then *also* describe a multiwinner method. Majority doesn't apply to that, this is broadened representation, a very simple form. It behaves quite differently from IRV. Fewer votes are wasted. If RRONR was going to express the "majority" concern with this multiwinner method, they would be opening a huge can of worms. It's an example where "majority rule" is inequitable. The reason to use this multiwinner method is to broaden representation, so the applicable concept would not be "majority" but "quota," and this is one very complicated issue, and Terrill, I presume, knows that. That section is not about the subject of this article!
In the new mention Terrill describes above, it was appropriate to make a general statement about usage. That statement has an exception, which is if a majority is required. Without the exception being noted, the statement becomes untrue. Now, I'm not trying, really, to "rub it in." I'm actually disturbed by the additional mention here, and if I'd had more time, I might well have put in something different. Please don't assume that I'm trying to make some POV point. I'm not. I just happen to understand a whole series of points of view here, and so I'm sensitive to the arguments and the spin that can arise. As are, I'm sure, Terrill and some of his friends. COI editors are allowed to make non-controversial edits, and, in actual practice, here, I've been interpreting that liberally. As long as I can continue to assume good faith, I'm not at all about to raise a fuss simply because I disagree with an edit, as long as I can see it as a true attempt to find consensus. Where it gets dicey is when there multiple reverts that clearly represent entrenched disagreement. So, Terrill, fix it: try to understand the purpose of the mention, in terms of accuracy, and find better language or example. We've taken a lot of flack over this for unsourced statements, and this edit arose in the process of adding sources.--Abd (talk) 16:16, 31 May 2008 (UTC)

In any case, I now see the edit Terrill made. I left the explanation in the text out, it isn't necessary. The footnote I put back in, it points to the exception, which is majority failure, and it is source for the weasel word "most." The statement "don't count for or against" isn't true if majority failure is relevant. If a majority is required, adding a second preference can give that second candidate a majority, and thus the election completes. If no second, the candidate's favorite might go on to win the election. If the election is repeated, the runner-up seems to win about one-third of the time ("comeback," a FairVote study found 29%).

By the way, in the Australian material from Antony Green, he points out that Optional Preferential Voting which is what is called "IRV" in the United States, is resulting in high rates of "plumping," i.e., votes that don't rank beyond first rank, and that this tends to award victory to the primary winner (since they don't require an absolute majority). This is the same result I've seen in the U.S., that rankings are being preserved through vote transfers, his analysis is just about the same as mine.

So ... this has opened one o them cans o worms. There are actually two methods used in Australia: The most common is Preferential voting (STV singlewinner), with full ranking required, so a majority will *always* be found, even if they had to pry the vote out of the voter with a crowbar. (It is illegal not to vote in Australia, and if you don't fully rank on a preferential ballot -- in most places -- your ballot is spoiled as to that election (though it's a little more complicated, there is a kind of "backup vote" that is allowed). The other method used in Australia is Optional Preferential Voting. And this is exactly what is being used everywhere in the U.S. (in public elections). And, apparently, it behaves the same in the U.S. Yet lots of argument about IRV has been based on the behavior of full-ranking required preferential voting. The fact that U.S. "IRV" is actually Optional Preferential Voting should certainly be in the article! --Abd (talk) 16:37, 31 May 2008 (UTC)

[edit] SECTION: IRV gives voters a wider range of choices

I dropped the last sentence, unable to make it coherent to the paragraph. IRV gives choices. Eliminating a compromise doesn't mean choices didn't exist, only that voters had other preferences!

IRV gives voters a wider range of choices

  • Like the two round system, IRV tends to give voters a wider range of choice among candidates than plurality. More independent and third party candidates are likely to run because the spoiler problems are less severe. [12] The sequential method of IRV accommodates choices differently than runoff voting by not immediately reducing the field to two in the second round, as typically done in runoff elections like the French presidential election, 2002. As noted by Robert's Rules of Order, however, the elimination technique used by both methods can prevent a candidate "from becoming a compromise choice." This possibility is ameliorated if the runoff is not limited to two candidates, such as by allowing write-in votes.[13][14]

[edit] Like plurality, IRV requires the winner to have core support.

I removed this section. All its saying is "IRV is like plurality". There's no objection to the truth of this fact that makes it worthy for this section. It's written in a backwards way, "here's a fact, and here's people who don't care." That's not a great argument. Tom Ruen (talk) 02:48, 1 June 2008 (UTC)

FairVote has asserted, "We believe a winner should be at least one voter’s first choice, meaning they would receive more than 0% in current rules."[15] Essentially, if a candidate is dropped in the first round, even if this candidate is everyone's second choice, that candidate cannot win, and, of course, a candidate with no votes would be immediately dropped.

Other reformers, such as Range voting and Condorcet method advocates, don't consider "core support" a meaningful criterion.[16]

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

I removed this section below. First of all, the title above could be considered a pro or con depending on your interest. Secondly, the entire supporting information below is a statement against MAJORITY/single-winner methods in general, in favor of proportional systems. It's a waste of space here in a single-winner method to offer a weakness for something IRV never claimed to support. Tom Ruen (talk) 02:56, 1 June 2008 (UTC)

The ACE Encyclopedia reports that the alternative vote "was introduced by the Nationalist government in Australia in 1918" when "it became clear that several aligned conservative candidates all standing in the same electorate could split their vote between them under first-past-the-post, thus handing victory to the less popular but more disciplined Labor Party forces." "Commentators on Australian politics historically tended to regard the alternative vote as a variation of FPTP, in most cases giving results nearly identical to that system in terms of election outcomes and the structure of party systems. Douglas Rae, for example, in his seminal work on the consequences of electoral laws, stated baldly that 'the Australian system behaves in all its particulars as if it were a single-member district plurality formula'. A number of other commentators have argued that preferential voting makes little difference to Australian electoral results and have not been central in determining how governments are constituted." However, in recent years minor parties have become more significant, and thus spoiler prevention has become more important. The report concludes that "There is widespread agreement that AV has facilitated coalition arrangements such as that between the Liberal and National parties, and that it works to the advantage of centre candidates and parties, encouraging moderate policy positions and a search for the 'middle ground'."[17]

The British Independent Commission on the Voting System, known as the Jenkins Commission (UK) recommended that the UK adopt IRV (called the Alternative Vote) but with a "top-off" of additional MP seats to reflect party proportionality. The report noted that IRV would "increase voter choice," but, being a majoritarian system, without additional mechanisms, "it offers little prospect of a move towards greater proportionality."[18]

[edit] Instant runoff voting fails the Monotonicity criterion

I removed these web references:

  • Visualizations of IRV behavior are available from multiple independent sources.[19][20]

These appear to be personal websites. They are attempts to "model" voters and no clarity to what they mean in practice. Since the effects of nonmonotonicity IRV are identical to the effects of any runoff process, if they deserve reference it ought to be against all runoffs with forced elimination. I just can't see that this adds any clarity for evaluating the practical reality of IRV and real voter behavior. Tom Ruen (talk) 03:17, 1 June 2008 (UTC)

I'm going to note a counter-argument, but not push it at this time. The best known of these sites is that of Ka-Ping Yee, who is a notable expert. Olson is mentioned as a confirming source. Yee diagrams, as they are called, are well-known in the voting methods world, but it is also correct that the sites are "personal web sites." The diagrams illustrate visually the phenomena that various election methods show, and some of them, at least, are well-defined. Remember, this is the Controversies section, so a rewrite would note Yee diagrams as an argument, or support for the monotonicity argument. I think there are also Yee diagrams on the Center for Range Voting web site.
The claim that the effects of IRV are identical to the effects of any runoff process is a separate matter, and a huge amount of nonsense has been written on this topic. IRV and top-two runoff produce different results, that is apparent, and voting systems theory leads me to conclude that top-two is superior to IRV. Little-known fact, and FairVote isn't going to tell you, Tom, but IRV was used in the U.S. for political primaries for a time. What was it replaced by? Top-two runoff. Read Robert's Rules on preferential voting compared to runoff elections, then realize that the "elimination" in top-two, generally, isn't. That is no candidate is actually "eliminated." What happens is that the top two get ballot position in the runoff. Voters can still vote for anyone. Now, this may be a moot difference, but the fact is that "comeback" elections are rare with IRV, and take place about one-third the time with TTR. They are *different* methods, and the results show it. TTR is closer to what Robert's Rule recommends. (And, in fact, if Preferential Voting -- a better form than STV -- is used with runoff if there is majority failure, you *have* what Robert's Rules recommends. Fewer runoffs (probably about half or more, depends) and still adherence to democratic norms. --Abd (talk) 12:37, 11 June 2008 (UTC)

Well, myself, I advocate a "top-two IRV" vs "bottom-up IRV" as a more conservative reform from Plurality, but BOTH are nonmonotonic in the same rare 3-way cases (and even more rare 4+ way cases). The primary reason I object to the diagrams is that they require a degree of understanding that is fully unclear to an average reader. Explaining nonmonotonicity in a simple way is hard enough. Throwing up graphs that have nothing to do with real voters, real elections, don't illuminate. They just confuse. I'd be VERY GLAD if there was a way tp explain these diagrams on Wikipedia but I don't see it. Tom Ruen (talk) 21:47, 11 June 2008 (UTC)

[edit] About that American Political Science Association.

I've come to the conclusion that the alleged "adoption" by APSA of "IRV" is truly non-notable, but taking it out will rather gut that section. This factoid has been heavily promoted by FairVote, [23] produced 584 hits today. For example: Massvote.org, a rather confused document, as a "bullet point": "American Political Science Association uses it to elect their president."

Citizens for Instant Runoff Voting in New York State,[24] lists APSA as "endorsing" IRV.

However, Brams notes in Mathematics and Democracy (2008), p. 21, the following, confirming much of what I've written here in the past:

Not all societies that have been approached about adopting AV [Approval Voting], including two that I belong to -- The American Political Science Association (APSA) and the International Studies Association (ISA) -- have been amenable to election reform, much less the adoption of AV. Significantly, these societies are dominated, or heavily populated by, academic political scientists; none holds competitive elections unless a petition candidate challenges the official slate (this has never happened in the ISA; in the APSA, the last challenge to a presidential candidate occurred almost forty years ago.

APSA's alleged "adoption of IRV to elect its president" has been promoted heavily by FairVote, as we can see from the google hits, because of the obvious implication: if political scientists use IRV, must be great stuff. But I don't think we would want to imitate, in public elections, or even in private elections such as student body governmental ones, what APSA *actually* uses. Go to the APSA web site and see what they do.

There is a nominating committee appointed by the President. It, presumably by majority vote, names a single nominee for the office. If there is no petition candidate -- which has been the case for the last forty years, according to Brams -- that nominee is formally elected at the annual conference, becoming the "President-elect," and the President the next year. This is how APSA actually elects their President. APSA is about a hundred years old, and at the time it was formed, "preferential voting" was fairly common in the U.S. So they put it into the constitution, to cover the contingency that there is not just one, but two petition candidates. Given that there has been one petition candidate in the last forty years, I can speculate that the frequency is, say, one per fifty years. Two? Well, a rough guess would be one every 2500 years. (It is very hard for petition candidates to win, usually.) And more could be written on this, but I'll spare you. The APSA factoid isn't notable for the purposes of this article, except in the Controversies section, where it just might belong, since the situation is being used as an argument for IRV, a thoroughly deceptive one. And that can be shown by simple reference to reliable source.... Meanwhile, I may start working on this, what to do, with some edits.

One more comment on what APSA actually does. They conduct elections by electronic ballot, so "repeated balloting" as is actually recommended by Robert's Rules of Order would be quite practical. In the election for President, it isn't the President being elected, technically, but the President-elect, there is a whole year until the winner actually becomes the President. Used to be a link to the actual APSA constitution, it has apparently been removed. --Abd (talk) 13:16, 11 June 2008 (UTC)

I think it is objectively important to point out that the national association of political scientists will use IRV if it has a three-candidate race for president. It's quite reflective of where political scientists are on IRV. Also, can you find actual association elections that do mail voting or internet voting with repeated balloting if no on wins a majority of initial choices? Far more use IRV and accept a majority of the final round as determinative.
Also, Abd, check out FairVote's link on universities and colleges. We now know of at least 41 colleges and universities where student governments use instant runoff voting, as documented on our site -- that's more than "several." I see someone else changed this, but note that you can verify yourself.
RRichie (talk) 22:03, 11 June 2008 (UTC)