Talk:Approval voting/Archive 2

Historical use

The article claims several historical uses of approval voting (or similar methods):

• 13th century Republic of Venice: I am unable to find anything in this article which suggests approval voting (or any variation thereof) was used. Does anyone have any external references?
• Parliamentary elections in 19th century England: Again, I was unable to find any other references to approval voting.
• United Nations Secretary-General: Although the Secretary-General has to be approved by all 5 permanent members of the security council, they also have to be voted in by the majority of the security council, and the majority of the general assembly, so I don't think this really counts.
• Although it warrants a mention as a similar concept, I don't think Bucklin Voting can be considered as simply a variation of approval voting, as voters are not required to approve of individual candidates (they may not approve of any of them, and may simply be ranking the least worst alternatives).

If the first two cannot be substantiated, I suggest they be removed. Any other thoughts on the last two? -3mta3 (talk) 22:10, 28 November 2007 (UTC)

I'm reasonably confident that a published source for the Venice claim can be found and inserted. It's fairly common as a claim about Approval Voting; what would be appropriate would be to place a source needed tag and to remove the claim if no source is provided within a reasonable time. This same editor removed the same claim from the Range voting article, I reverted it there. The claim about the UN Secretary-General is also commonly asserted. What is described above isn't relevant, the question would be the voting method for the General Assembly. Essentially, the 5 permanent members each have a veto power (if that's true, and it may well be). Bucklin voting is "Instant Runoff Approval." That is, if it proceeds to the last round, it has become a full, unrestricted Approval election. First round only, it's simple Majority. Second round, it's two-votes-allowed Approval (but only one per candidate!). The similarity is important: if Bucklin violates one-person, one-vote, which has been asserted by FairVote and which is an interpretation of Brown v. Smallwood (an incorrect one, in my opinion), then so too would Approval.

--Abd (talk) 06:22, 29 November 2007 (UTC)

I easily found multiple references by googling "Venice voting." I have cited in the article a paper that might not itself be fully qualified as "reliable source," but which does cite several published papers on the topic in peer-reviewed journals. Ideally, this article should cite the fully qualified sources ... but, for now .... Fascinating to me is that the method used was, more fully, multilevel Asset Voting, a method so new and advanced that I can't probably create an article on it! --Abd (talk) 13:47, 29 November 2007 (UTC)

I changed the range voting article simply because it seemed rather stilted prose (the product of too many edits, I assume). The Venice reference is indeed a fascinating one, and the paper seems reasonably objective (i.e. it doesn't seem to be the product of an advocacy group). The UN secretary-general is often claimed, but it is a much more complex (and political) process. No luck on the English Parliament one? -3mta3 (talk) 18:22, 29 November 2007 (UTC)

I've rearranged the article a bit, moving some of the past uses from the heading to their own section. -3mta3 (talk) 18:50, 29 November 2007 (UTC)

As to Parliament, not yet. Brams makes the claim, and he's pretty sober. (But a paper that is simply a talk he gave, even if sourced, is less desirable than published sources, and those are mostly held behind protection. I.e., one has to pay to see the papers. He also claims that Approval was used to elect Popes, and I've seen a source for that. There is a description of the Venetian usage in Voting system, with references. I agree with moving historical usages, for the most part, out of the introduction to a History and Usage section. Brams has also noted various large societies which have adopted Approval (American Mathematical Society, for example), plus he has described the IEEE case, basically saying what I've said in various forums: the IEEE adopted Approval when the board wanted to avoid vote splitting that might have elected a dissident candidate, then they dropped it when the board didn't need it any more. The argument that has been given that Approval was dropped by the IEEE because most voters weren't using it is clearly a specious one. In some elections, votings don't need to cast additional votes, and in some elections, only a minority of voters would do so. Such as Greens in Florida in 2000. The *option* to cast extra votes costs nothing. Actually counting those votes adds a small expense or trouble. With internet voting, no trouble at all.... (And if there are only two candidates, voting for more than one would be quite rare and quite useless, except that one might be more likely to avoid majority failure, if a majority is required for election.)

One more item. Found an election announcement, Sept 2007, for the American Mathematical Society: [1]. Indeed, they use Approval Voting --Abd (talk) 19:08, 29 November 2007 (UTC)

The AMS link seems to suggest that approval voting is only used for those two committees (not for the Board of Trustees or other positions). Anyone able to clarify? -3mta3 (talk) 20:31, 29 November 2007 (UTC)

Approval voting is used by several scientific societies including the "American Statistical Association". According to its bylaws:[2]

2. Balloting. For all of the Association's elections, the system known as approval voting shall be used. Regardless of the number of candidates or the number of places to be filled, the voter may vote for any number of candidates but may not cast more than one vote for a candidate. Winning candidates are those with the highest numbers of votes. Any tie shall be broken by random selection; no runoff elections shall be held.

JRSpriggs (talk) 04:02, 30 November 2007 (UTC)

To the claim about 19th century England, Abd recently added a source and attribution (diff) with the edit summary “Historical use - until I can find something better, attributed to expert.” I agree with Abd, as he also commented above, that the current source for this claim is not a high quality source for this claim, in part because of the lack of specifics or other references in the source. Moreover, the current source does not make any mention of parliamentary elections. I'll allow some time for Abd or others to follow up and find a better source soon. However, unless a supporting source is found, the mention of parliamentary elections should be tagged or deleted. Without any sourced information regarding further specifics about how, when, and where approval voting was allegedly used in 19th century England and without a reliable source, the article should probably either explicitly note these deficiencies or the claim should be entirely deleted. DCary (talk) 21:17, 21 January 2008 (UTC)

I'd disagree. It's a statement by a known expert, and it is attributed. In other words, what the article currently has is verifiably true. Brams did state that. Obviously, we'd prefer to have better source, in which case it would not need attribution. "According to Brams" covers the situation for the moment. Now, if there is any notable claim that it was *not* used, the balance would shift. We'd have to report both, or neither. I'll stand by the text at this point. But I'll also see if I can ping Brams. (Absent that, the article can note that his claim has not been confirmed, pending confirmation, that's one possible solution. In a "real encyclopedia," the editors would contact him and ask for a source. So.... --Abd (talk) 03:51, 31 January 2008 (UTC)

The first thing at issue about the Brams claim is what exactly he is talking about. What happened in 19th century England that Brams claims happened? Did the events actually happen, the way Brams thinks they did? Those are history questions, not political science questions. As far as I know, Brams is not a recognized expert in the history of 19th century England, so his self-published, non-peer-reviewed claims about unspecified historical events do not qualify as a reliable source. If Bram's claim is true, I suspect there will be a reliable source to back up the historical part of the claim. If we can't get a reliable source for the historical part of the claim, there is a strong case for deleting any mention of it. DCary (talk) 00:42, 8 February 2008 (UTC)

The claim regarding 13th century Venice, is still without a reliable source. The currently noted source (Mowbray and Gollman) claims approval voting was used. However, the two references Mowbray and Gollman give for descriptions of the Venetian voting do not support that claim. The two references are:

• J. J. Norwich. A History of Venice. Allen Lane, The Penguin Press, 1982, pp 166-167.
• E. M. Tappan. The World’s Story: A History of the World in Story, Song and Art, volume 5. Boston Houghton Mifflin, 1914, pp 51-54.

Both Tappan and Norwich agree that the Doge was elected with a super-majority requirement (25 out of 41). Tappan is silent about the specific voting system used and offers no suggestion that approval voting was used. Norwich on the other hand makes it clear that approval voting was not used to elect the Doge:

• Each of the 41 electors was allowed to secretly nominate one candidate.
• A list of candidates was created, with duplicate nominations ignored.
• Candidates were selected in random order to be:
  • discussed and questioned
  • voted on
• If a candidate so considered received 25 votes, he was declared Doge, otherwise, another candidate was randomly chosen for the next consideration.

Even regarding the selection of the 41 electors, which was a multi-phase selection process, neither Tappan nor Norwich give any indication that approval voting was used. They both agree there were phases in which a small (9-12 member) deliberative body would select a larger group for the next phase. They also agree that a super-majority was needed, but give no other indication of how voting was done.

Given that there are many ways, without using approval voting, that a deliberative body can go about choosing another multi-member body and that approval voting is not particularly suited for such a task and context, these references are notably unsupportive of the claims in the article. I have growing doubts about the veracity of the claim. It may be appropriate to remove the claim from the article, at least until some truly supportive, reliable sources can be found. DCary (talk) 00:46, 31 January 2008 (UTC)

Mowbray and Gollman use the term "approval voting" because this is a very recent source (2007) and the term has become common. As with prior considerations of the application of "approval voting," it is necessary to understand that it is Approval voting with special rules added. In this case it is a supermajority requirement to win. The exact sequence of voting is also not specified, only a necessary outcome: to become an elector, the nominee had to receive a minimum number of approvals. This was a complex process used to elect the Doge, but the core resemblance to Approval is that each nominee had to receive that support, the votes are thus independent. As far as anything I've read, there was not necessarily a single ballot; and the supermajority requirement would almost certainly require multiple balloting to elect the required number. That is, I'd speculate, there would be a ballot and every nominee who gets the number of approvals needed would be elected. I'd also surmise that if more than the needed number of electors got that number, the ones that got the most would be elected. The key thing that the sources agree on is that the candidate must receive so many votes. Above was a description of the actual election of the Doge himself. This is *not* claimed to be Approval. Rather, what is in the article is that Venice used "a complicated system of approval voting and random lots to select the committee that would then select the Doge ..." In any case, the point of citing Mowbray and Gollman is that they actually report their conclusion that this was Approval.

However, there is more. Above, I speculate. Turns out that we have explicit report. From Mowbray and Gollman:

As far as we are aware, just two other papers investigate the mathematical properties of the protocol. Lines [11] discusses approval voting, which is the method of voting used in each round which increased the college size, and in the final round. Candidates for the next college (or for the Dogeship, in the case of the final round) were proposed, and a ballot was held in which the current college members signalled either their approval or disapproval of each candidate, with no limit on how many candidates they approved or disapproved. Candidates receiving the required minimum number of approval votes joined the next college. If not enough candidates received the minimum, the college repeated the process, holding another ballot.

Now, this talks about "a ballot being held." If this was a sequential presentation of candidates, with a winner being declared as soon as one reached the required margin, this would be, in theory, different from Approval (because a key feature of Approval is that multiple winners -- nominees reaching the required margin -- are possible). However, in practice, with a high supermajority required, and the number to be elected, the difference would be academic, for it is quite likely that multiple ballots would be needed (and thus every nominee would receive a vote opportunity). It may be that there is some ambiguity in the sources. Above is reference to Lines, and the note refers to [11] M. Lines. Approval voting and strategy analysis: A venetian example. Theory and Decision, 20:155–172, 1986. There's the source! Now, somebody with access to a library.... Here, though, obviously, not even primitive synthesis is necessary to call this "Approval voting." An academic published source calls it that. By the way, Lines is also given as a source for the Venetian claim by Brams in his "Mixed Success" paper. I've found other papers which reference Lines. If we are going to find and use Lines, we might as well also take from that source, if it has it, the years over which it was used. It's cited in another paper as "between 1268 and 1797." If that's true, that is astonishingly stable.

From Springerlink [3]: Approval voting and strategy analysis: A Venetian example; Journal: Theory and Decision; Publisher: Springer Netherlands; ISSN: 0040-5833 (Print) 1573-7187; Volume 20, Number 2 / March, 1986; DOI 10.1007/BF00135090; Pages 155-172; Marji Lines, Facoltà di Economia e Commercio Ca Foscari, Universita Degli Studi di Venezia, 30123 Venice, Italy

Abstract The author presents a historic reconstruction of the single-member constituency election system known as approval voting which was used to elect Venetian dogi for over 500 years. An interesting procedure theoretically, concurrent approval voting is the only sincere single-winner election system. Central issues concerning strategy choice under uncertainty are investigated using a contingency-dependent framework of individual behavior given prior probability distributions over decision relevant propositions. Extensions are then proposed for the use of approval procedures in modern elections and other collective decision-making situations. Finally the advantages of trichotomous preferences in decision and strategy analysis are argued.

Bingo. The point is that this author calls it Approval voting, that's a published conclusion. So, while it will be very interesting to see, if possible, what the exact details were (and then we can note any variations from standard approval voting, we have authority for the use of the term. Brams didn't just make it up. (And even if he had, we could still have attributed it to him and kept it.)

But there is more. Mowbray and Gollman also have some other sources.

In fact, there is a case (described in [15] p.300) in which tactical voting allegedly determined the outcome of one election for Doge. In 1423 Francesco Foscari, an underdog candidate, received 17 approval votes out of 41 in the ninth ballot by the final college and 26 approval votes in the tenth ballot, thus winning the election. It was claimed that his supporters had engineered this win by voting in earlier ballots for a candidate that no-one wanted, thus enticing others to vote for Foscari, and then suddenly switching their votes. Presumably in 1423 concurrent voting had not yet been introduced.

This addresses the sequential voting problem, and why it might have been changed to concurrent voting. Was it changed? If so, pretty much any obstacle to calling this, flat out, Approval voting, would have disappeared. [15} is J. J. Norwich. A History of Venice. Allen Lane, The Penguin Press, 1982. The description above re sequential voting may be referring to earlier elections. I'd still call the sequential variation "Approval," and, in fact, I have many times referred to an "approval poll" that actually took place in an organization I belonged to. And the voting was sequential, because it was by show of hands. Indeed, I've frequently claimed that plurality was actually Approval, and that the freedom to vote for more than one was only lost when secret ballot was introduced. In Venice, we can suspect, they may have gone to written ballots (secret or not), but kept the allowance to vote for more than one. (I've seen a lot of single-winner elections by show of hands. I've never heard an instruction, "vote for one," and I've never seen a challenge to a vote, "But you already voted for another candidate."

Robert's Rules disregards overvoted ballots because they are errors. But why are they errors? Because they will be discarded. Oops!

--Abd (talk) 03:51, 31 January 2008 (UTC)

The Marjit Lines article is a reliable source. Based on that source, I withdraw my suggestion that a claim about the Dogi of Venice should perhaps be removed from the article. I'll add the Marjit Lines source as a reference and leave the Mowbray and Grollman source since it is more accessible. An open issue is whether and how the claim might be changed. DCary (talk) 00:42, 8 February 2008 (UTC)

I've been looking for a source for the claim that Approval Voting is used to elect the U.N. Secretary-General. What I've found is a reference that it was used in 1996, that this election of Kofi Annan followed the Wisnumurti Guidelines. The claim, as it has been stated, isn't correct. The Secretary-General, if these guidelines are followed, is nominated by majority vote of the Security Council, with the five permanent members having veto power, and then confirmed (or, theoretically, rejected, though it has never happened) by majority vote in the General Assembly. So where does Approval Voting come in? Well, the guideliness provide for Approval *polling,* where members of the Security Council, with papers colored differently for permanent members and regular members, vote to "encourage" or "discourage" proposed nominees. The poll is not binding, so it is not correct to call it an election, it's a straw poll. The guidelines provide for additional polls to be taken if necessary. Interestingly, the only personal experience I have had with Approval Voting was with just such a straw poll, used to *suggest* a choice between five or six alternatives. While the status quo, in that poll, received something like a 65% approval -- and thus could have been retained by a majority, and was probably the majority's first preference -- there was another option which received 98% approval. This was immediately submitted as a motion to adopt, which passed unanimously. So a similar process resulted in the election of Kofi Annan by acclamation. But an election, it was not. The election was by ordinary motion to nominate, and a majority would probably have legally sufficed, but unanimity was probably, also, considered desirable, as it was in my own experience. In any case, I'm going to remove the claim, as such. I might edit it to reflect the true facts. --Abd 22:31, 30 November 2007 (UTC)

The item about Wikipedia Arbitration Committee should be changed, given better sources on some points, or perhaps even removed. If the item is kept, some key issues that need to be addressed are:

• Describe it as an election, as Wikipedia does, or provide reliable sources for not doing so. The fact that the election does not make the final decision does not, by itself, make it any less of an election.
• Be consistent about whether this item is about the 2006 election or ArbCom elections in general. If the latter case, provide appropriate sources.
• Describe the election more accurately as:
  • a multi-winner election
  • a variation of 3-slot range voting (Yes=+1, Abstain=0, No=-1)
  • eligible winners are required to have a positive average vote total, which also means more Yes-votes than No-votes. The part about the unselected candidate should be dropped.
• Make the links http links rather than wiki links to avoid some of the problems with self reference.

Regardless of whether one thinks of this as a poll or an election, since this is not a use of approval voting as otherwise generally discussed in this article and it is either an ongoing practice or it is historically a very recent, not particularly noteworthy event, a strong case could be made for not including it in this article. The example might be of interest to the range voting article instead. DCary (talk) 20:46, 13 February 2008 (UTC)

Unless we have reliable sources supporting the claim regarding Bucklin voting that it becomes an approval voting election in later rounds, the claim should be weakened, for example by saying that in some ways it becomes more like an approval voting election. DCary (talk) 20:46, 13 February 2008 (UTC)

Burr Dilemma Dilemma

There is mention in this article of the Burr dilemma, which is a neologism coined by Jack H. Nagel in an article published in The Journal of Politics, where he claims that "it has not previously been recognized that the first four presidential elections (1788-1800) were conducted using a variant of approval voting." (from the abstract at [4]. The article itself is at [5]. He's right that it wasn't previously recognized; however, he's off the wall in calling that election "Approval Voting" merely because electors had two votes. A basic characteristic of Approval is that voters are unrestricted in voting for or against all candidates. The election he examines is not Approval at all, it is described on Wikipedia as Plurality-at-large voting, in the article Block voting. It was a "variant" of Block voting because the two offices being elected were not identical, they were the President and Vice-President, but the voting process held a single election for them both, with the recipient of the largest number of electoral votes becoming President and the runner-up becoming Vice-President.

Frankly, I find it appalling that a peer-reviewed publication published the article at all; however, this is the journal of the Southern Political Science Association, and my experience has been that political scientists are not *usually* experts on innovative election methods.

In the Burr/Jefferson election, there was no dilemma in the election itself. The electors cast their votes and Burr and Jefferson tied, i.e., the electors voted for them as expected, given that the party system was developing, and this threw the election into the House of Representatives, which has different rules. That is, both Burr and Jefferson won, but the House had to decide who became President. The framers of the Constitution do not appear to have anticipated the arising of electors dedicated to slates. The "Burr Dilemma" is Nagel's imagination: could Burr or Jefferson have attempted to arrange matters so that one or the other would have won? The dilemma, supposedly, is that if one of them attempts to influence electors to vote for him only, instead of two, the other would have similarly attempted to influence electors to vote for him only, thus spiraling into a retaliatory cycle leading to single-vote Plurality. Which, frankly, makes no sense, since this could quite easily throw the election to their opposition.

None of this has anything to do with real Approval elections, as would take place if Approval were to be adopted for political elections in the U.S., except in a quite peripheral way. Yet Nagel translates this into an indictment of Approval Voting, which is then used by FairVote as an argument against Approval. Even Warren Smith of the Center for Range Voting ([6]) seems to have been taken in, for he confirms the "dilemma" as a strategic problem with Approval Voting, and, of course, suggests Range Voting as a solution. (He's not correct. If the dilemma were real, normal Range Voting would not solve it, because these are strategic voters, and Range voted strategically generally becomes Approval.) Not one of Smith's finer moments, in my opinion.

The point here is that the Burr Dilemma is discussed in this article as if it had something to do with Approval Voting. The only connection is that Nagel calls the U.S. Presidential election process in those days a "variant of Approval voting," which is preposterous: just look at the definition of Approval voting on this page -- or anywhere except Nagel's article. I think the mention should be taken out.

There was language in the Burr dilemma article making pretty much the same point as I just made, until it was taken out with this edit: diff. I've been finding, over the last few months, as I become familiar with the universe of voting systems articles on Wikipedia, that edits like this pop up frequently: a user registers and immediately makes a drive-by edit like this. And nobody is watching. That language was not sourced, that's true, but it was clear and easily verifiable, and thus was defensible (or at least editable to be defensible). I can't be sure that this editor was a sock puppet, but I've seen behavior quite like this from definite socks. There is one sock master who has been very active with the voting methods articles, and I think I've identified socks associated with him that haven't yet been tagged and blocked. In some cases, too much time has elapsed, checkuser wouldn't be able to prove sock puppetry.

The Burr Dilemma has, in fact, nothing to do with Approval Voting. On the other hand, because Nagel does use the name "Approval Voting" in his published article, it could be claimed that the controversy should be included; but I question its notability at this point. Generally, though, I prefer to leave in articles that may be of use to researchers, even when notability is questionable. One of the things that all those socks have been doing is to put up voting methods articles for deletion, and when none of the voting systems editors notice it and defend the article -- only a few may have edited the more obscure articles, and perhaps they don't log in during the AfD process -- the articles get deleted. Such as Proportional approval voting, though it's definitely notable, and there are others. The deletion proposer will claim -- perhaps accurately -- that there has been no mention in peer-reviewed journals, confusing "reliable source," necessary to state fact without attribution, with notability, which is the general standard for article survival. And all it takes is one administrator to buy the argument. See the AfD for PAV: Wikipedia:Articles_for_deletion/Proportional_approval_voting. --Abd 03:07, 4 December 2007 (UTC)

Significant changes to ballot

I removed the reference to IRV requiring major changes to a ballot. In many jurisdictions (eg Canada, UK, etc...) where paper ballots are used, there is no change to the format of the ballot required - all that is needed is that voters put numbers instead of a single 'X'. This does not constitute a significant change. 24.150.226.39 (talk) 22:40, 9 December 2007 (UTC)

• And yet you left the reference to Condorcet methods requiring major changes in place, twice [7]. From the voter's perspective all ranked-choice ballots are the same. However, IRV requires massive adjustments in counting. It is not summable and thus the information contained in a ballot is factorial in the number of candidates. Summability criterion used to show this fact but it was deleted. Instead this information is available at Electowiki. As far as whether requiring rankings is a significant change, this is your opinion, and while all of Canada may be using paper ballots, at least 2/3 of the United States is voting on electronic machines, for good or for ill. Changing these voting machines and educating voters would be a major issue with ranked choice voting, while it would be a minor adjustment with approval voting. More importantly, why did you find it necessary to bias that paragraph in favor of IRV versus Condorcet methods? - McCart42 (talk) 23:59, 9 December 2007 (UTC)

Yes, I thought it was strange to remove IRV while leaving Condorcet, same issue. (A complete IRV ballot is the same as a complete Condorcet ballot, it is merely a different analysis that is used.) "All that is needed is that voters put numbers instead of a single 'X'. That's a radical ballot change, in many places. In some, sure, not. The argument is a general one, and like all general arguments, may not apply in some specific place. There is no serious contention that I've ever seen that the change to IRV, in many places, requires changes to voting procedures and counting procedures, and, where relevant, to voting equipment. Why does San Francisco with it's "Ranked Choice Voting," which IRV proponents call "IRV" when they want to point to success stories -- though it's a mixed bag there, but that's not relevant here -- have only three ranks? It's quite possible that the majority failure in some of those elections (winner had less than 50% of valid ballots cast) results from that restriction, which is due, it's been alleged, to voting equipment restrictions. I changed this language back. I'm not personally adding citation tags for arguments that are well-known, but I think we might pick up a citation from the IRV article.

I'm not happy with the state of this article with respect to NPOV balance. There is criticism of Approval that exists that is not shown here, but favorable arguments are. The Instant-runoff voting article has a controversies section and even a fork has been created for in-depth treatment of arguments, parallel treatment might be appropriate here.--Abd (talk) 03:02, 10 December 2007 (UTC)

• The problem is that outright comparisons between IRV and Condorcet, or between IRV and approval, have been deleted when they've been attempted in the "controversies" articles. IRV advocates have called it POV-pushing to compare IRV to approval in their article. You can't discuss one method in a vacuum without comparisons. One perfectly reasonable way to compare methods is by simulations, such as those done by Ka-Ping Yee [8] or Brian Olson [9]. - McCart42 (talk) 03:09, 10 December 2007 (UTC)

Added Wikipedia usage of Approval Voting

The elephant in the living room. Approval voting is used to elect members of the Arbitration Committee (I sourced this on the article page). Technically, it is not an election, it is a poll. Any registered Wikipedia user who registered prior to a cutoff date may vote Yes or No on each nominee. Each vote is independent of all the other votes, there is no "vote for N." It's been proposed, and rejected, as has Condorcet voting. "Winners" of the election are not announced as such; rather, the election results are apparently reviewed by Jimbo as advice, and he makes the decision; in 2006, seven members were appointed; they were the seven highest vote recipients in the poll, in terms of vote percentages (percentage Yes, each candidate vote considered in isolation). Fourteen candidates had more than 50% Yes. The candidate with the most Yes votes had, however, a 25% No percentage, leaving him with a lower net Yes vote if Nos are subtracted, or a lower percentage Yes if that is the considered basis.

Technically, this is Range Voting (Average), this is not ordinary Approval, where the candidate with the most Yes votes wins. Net Yes voting (Yes minus No) is more like Range 3 (-1,0,+3), except that there is no 0 vote. Abstentions are not counted, so this is "blanks-excluded" Range.

In fact, the raw votes are available, the identity of the voters is known, and tools could show and compile, for example, the "age" of each voting account, number of edits, etc. The results page shows total Yes, total No, and percentage Yes, i.e., Yes/(Yes+No). In a normal Approval election, we'd want to compile slightly different data. Each "ballot" would be counted, so we'd want to know the number of unique voters. However, I've many times noted that Approval is equivalent to a series of Yes/No Ballot questions, each one of which must pass with a majority. (Apparently, a Yes over No majority is required.) However, with that, it is the one with the most Yes votes which prevails if more than one pass, in the state rules I have seen.

Given that the goal is apparently an Arbitration committee with the broadest confidence of the community, and that there can be severe participation bias, considering the Yes percentage (Yes/Yes+No) instead of the absolute Yes vote, would seem to be appropriate, to me. The danger in this of a "dark horse" winning because most voters, not recognizing the name and having no opinion, leave it blank, is avoided without having any rule about it, is avoided by the fact that this is just a recommendation. Further, in addition, given the easy possibilities of packing the vote, holding the election as a poll feeding an appointment by a "trusted servant" also makes complete sense to me in an environment like Wikipedia. Jimbo can also use any criteria he sees appropriate to analyze the votes. If he wanted to, he could look at user histories for voters (voting is public), to see if there was some balancing necessary, or he can just look at the past behavior of arbiters nominated by the process to see if they are acceptable in his view. Wikipedia is not a democracy, except in certain ways. The founder, so far, retains effective veto power. Eventually, I expect, that may pass. --Abd (talk) 02:05, 10 December 2007 (UTC)

I'm fine with this, but per WP:SELFREF you need an outside reliable source which talks about wikipedia elections, I think. MilesAgain (talk) 23:02, 11 December 2007 (UTC)

Other issues and comparisons: Tolerances, not preferences advantage or disadvantage?

An IP editor added the sentence about "other political scientists" to this claim about Approval. Now, the whole section, in my opinion, needs rewrite to satisfy NPOV, and we should move in that direction,

• It allows voters to express tolerances but not preferences. Some political scientists consider this a major advantage, especially where acceptable choices are more important than popular choices. Other political scientists consider this a major disadvantage, especially where expressing the will of the people is important.

First of all, I think the comment about tolerances, not preferences, is an analysis of what Brams has claimed. I'm not sure of his exact language, and we should bring that here, or else we might be distorting. I do think that Brams repeats what I see as an error of considering Approval votes to be anything other than votes. The assumption that they represent some sort of feeling about the candidate, ("Approval") leads to many problems. In the end, approval votes are just votes. You vote for a candidate you are willing to support, and not for one, or against one, you are not willing to support. It's an *action*, not a sentiment. However, I'm not sure I can find a source for this argument that isn't me. (I've argued this in many places, and, without peer review, I can't use it for the article.) However, this makes me sensitive to what might be distortions of what Brams, or others, have actually written.

Now, as to other political scientists, I don't know who is being referenced. Because all political scientists, including Brams, agree that "expressing the will of the people is important," it is not clear how this translates into some sort of "disadvantage." The question is how we determine or measure that will. Compared to plurality, Approval adds flexibility of expression. Compared to ranked methods with equal ranking disallowed, it adds flexibility in one way while not allowing it in another, and this would be the argument in this piece of the section, that the addition is more important than the restriction. There is an argument that I, unfortunately, have not seen anywhere but my own analysis that IRV and Condorcet methods do violate one-person one-vote, in a way that Approval does not. It relates to this, so I may need to look more closely at Brams to see if he has expressed it, it underlies the theory of Approval Voting and why it might be better than a more expressive (of preference, but not of "approval") ranked method. For starters, it's my view that no candidate should *ever* be elected if a majority of voters have not explicitly approved that election. This is actually fundamental parliamentary procedure, and it is set aside only due to a belief that elections must complete or else the sky will fall. (IMHO!).

On the other hand, ranked methods with equal ranking allowed are in some ways equivalent to Approval, allowing the voter flexibility in how to vote, in "expression." My own understanding has led me to the conclusion that there are two principles to be respected: broadest acceptance and majority rule. That is, Approval arguably fails the majority criterion, but, at the same time, a majority has, in that case, explicitly accepted that "failure," which preserves majority rule. (Thus the common usage of dealing with multiple conflicting ballot questions.) It can't be done with Approval, unless some "favorite" indicator is added to the ballot, but if such were added, the ballot could be examined for both the Approval winner and the preference winner, if they are different. That is, if there is a candidate who, considering the Favorite marks, beats the Approval winner, a runoff would be triggered. The electorate explicitly decides, by majority vote, between the options of widest approval and majority preference. The combined method satisfies the majority criterion, and, from simulations, the Approval winner is almost always the Condorcet winner, so ... runoffs would not be common. (Think about what it takes for those two winners to be different: it takes people who have voted for both frontrunners.... not common.)

I'm certainly friendly to improving the language of this section, and I consider it unbalanced as it is. But something more clear than what was attempted is needed; and it may well be best to actually quote one or more of those "other political scientists." Presumably the anonymous editor, if he or she reads this, has some actual comment in mind and can thus easily supply it. --Abd (talk) 03:49, 10 December 2007 (UTC)

POV tag

This article has apparently had little attention by those who are not advocates of Approval voting in recent weeks, and is not now NPOV. It is certainly more unbalanced than the IRV article, which wears the POV tag. For example, the section on strategy presents the issue of tactical voting in the most positive light poossble. What about Nicolaus Tideman's analysis that shows Approval is among the voting methods that are absolutely the MOST subject to tactical mahipulation? I'm not sure when I or other voting methods experts with a skeptical eye will have time to do the kind of review this article needs, but in the mean time Wiki readers need to be warned that it is biased. Tbouricius (talk) 15:23, 11 December 2007 (UTC)

I agree that the article has a lot of problems. The POV tag is appropriate. Let's work on it! By the way, as to history, substantial editing on this article was done by one of the sock puppets blocked in the Instant-runoff voting affair earlier this fall. --Abd (talk) 22:03, 11 December 2007 (UTC)

I agree, even if I am a sock! SockPuppetForTomruen (talk) 22:41, 11 December 2007 (UTC)

Ruen was blocked too, at one point. but only by a humorless administrator. Well, really, I suppose the admin can be forgiven for wanting to confirm that SockPuppetForTomruen was actually a sock puppet for Tomruen; unfortunately, I think the admin blocked the suspected sock's IP address, i.e., Ruen's IP. Anyway, socks are welcome, *but* not for contentious edits, probably -- however, an identified sock isn't a problem *at all*. --Abd (talk) 00:21, 12 December 2007 (UTC)

As to Tideman's analysis, well.... let's say that I haven't seen any confirmation of that. Now, googling I find an interesting article.... rangevoting.org/TidemanRespA.html. The language seemed pretty familiar, then I look at the top .... yes, indeed, I did write this, and Smith took it and put it up on the site. Is that "peer review"? (Well, of a kind, but no cigar.) This is about Tideman's views on Range, but they are related, for "strategy" in Approval consists of voting it like sincere Plurality, whereas, supposedly, suckers will add approvals for other than their favorite. When criterion failures and strategy are being considered by IRV advocates, and the Favorite betrayal criterion, for example, is pointed out, they will argue, in return, that such failures would be rare in practice, which is true *in a two-party system*. Yet when it comes to criticizing Approval, they cry "vulnerable to strategy." What is this strategy? Vote for your favorite! Uh, isn't that a sincere vote? The fact is that there is no clear definition of sincere vote in Approval, other than not reversing preference, and Approval never rewards preference reversal. Same with Range. Mr. Bouricius must be aware of this, it is the core of the Majority Criterion problem and whether or not Approval passes that Criterion, and we have discussed this much elsewhere.

When Approval students discuss Approval voting strategy, it is to consider methods of where in a preference order to set an "approval cutoff." It does not mean insincere voting, which is what it means with all ranked methods.

The conditions where Approval does not elect a majority favorite are only those where two candidates gain a majority. Is it possible to imagine this happening in anything close to a two-party system? It requires that many voters vote for the two frontrunners, like Bush and Gore.

Then there is rangevoting.org/RichieRV.html. This is Warren Smith's own response to a critique of Range and Approval by Rob Richie. I'd say its worth reading, for sure.

Now, what does Tideman actually write about Approval strategy? I'm not at all sure what source Bouricius has in mind, but, from Collective Decisions in Voting, 2006:

The general strategy that works under Approval voting is to identify the two options that are most likely to win, and give only one of them a vote. This involves "burying" the less attractive option if one had been inclined to vote for it, and "directly hoisting" the more attractive option if one had been inclined not to vote for it.

Tideman then proceeds to consider the susceptibility of Approval Voting to this "strategy."

First of all, the "general strategy that works" isn't precisely what Tideman says. It's just a bit more than that. I.e., one votes for the two frontrunners as he describes, but, then, also votes for any other candidate whom one prefers to the preferred frontrunner. So, presumably, the Nader supporter votes for Gore, which is what Tideman describes, but also for Gore.

For major party supporters, the strategy is to bullet vote. In a two party system, or with two major candidates, therefore, we might expect most voters to bullet vote. As the experience with *some* Bucklin elections shows, this is common; likewise, it's been asserted that a flaw in Approval is that most voters will vote for only one candidate, as with the IEEE tactical application of Approval. (It appears that Approval was implemented, not to enhance democracy, but to head off a possible vote-splitting problem that could have lead to the loss of a board-supported candidate; then, when the danger was past, approval was rescinded because it could then allow dissident candidates some traction. The argument that it was rescinded because most voters bullet voted is preposterous. That's expected when there are two major candidates, it is a sincere vote and it is rational and strategically effective.)

This is the strategy that Tideman finds Approval "vulnerable to." Sincere voting, with a realistic approval cutoff. Consider the election Tideman is considering, but let's make it clear: there are only two candidates. How does Tideman's comment apply? Is it "burying" to not vote for the less attractive one? Or "directly hoisting" the more attractive one if one was not pleased that there is no better candidate? No, it is simply doing what elections do: choose between options.

Now, this commentary is, in at least some senses, Original Research. We can't just take this and put it into the article; but, I would submit, the arguments here, the important ones, are directly verifiable. I don't think we need to find a peer-reviewed source that points out that voters will only rarely vote for both frontrunners in an Approval election, in order to understand that it's true. The French report that Smith refers to, though, is available. Likewise, I think Tideman says the same thing.

In any case, Tideman's description of Approval being particularly vulnerable to strategy is quite strange. Note that Bouricius used the term "tactical voting." The Wikipedia article Tactical voting defines it as: "Tactical voting (or strategic voting or sophisticated voting) occurs when a voter supports a candidate other than his or her sincere preference in order to prevent an undesirable outcome."

Now, what does "support" mean? Let me give some examples: suppose I prefer some relatively unknown person to fill a political office. I can write in his name. But, instead, I vote for a candidate on the ballot. That's tactical voting. Suppose I'm voting in an IRV election, and I really only like one candidate. But I fear that this candidate will not win, so I add a second preference vote for a candidate whom I don't like, but I prefer to a third candidate whom I fear will win. That's tactical voting. In Approval, I vote for my Favorite, but, because my Favorite is not likely to win, I add a vote for my preferred frontrunner. Again, that's tactical voting.

Yet "tactical voting" is used as if it were some indictment of the practice. What it really means is that the voter considers not only his or her personal preferences, but also the realities of the situation. In small direct democracies, choices are made as a process of negotiation, where the individual and collective preferences interact, often with a seeking of supermajority approval. In large democracies, however, elections are used, but it is not particularly desirable to collect raw preferences from voters that have no relation to the social context. If one were to somehow build a machine that could extract from me my personal preferences to be amalgamated with others, there would be a problem: my personal preferences are idiosyncratic, and my top one hundred list for President might not include the common candidates on other's lists.

We truncate our list and combine it with what we know of others, most notably as expressed by the limited choices on the ballot. Because this involves compromises, it is "tactical voting." If I voted with true sincerity, I would write in a name at each rank, probably, and I'd probably exhaust the ranks before getting to anyone actually on the ballot.

With all ranked methods, tactical voting involves preference reversal. With Plurality, for example, tactical voting involves voting for a candidate when one prefers another, and not voting for one's favorite, and that is preference reversal. It is, thus, misleading, to use the term "tactical voting" with respect to Approval, when the vote *is not insincere*, it is simply a choice made by the voter of where to place an Approval cutoff. What I approve does, indeed, depend on what I think I can get. --Abd (talk) 04:39, 12 December 2007 (UTC)

You were recently in an edit war with a banned user, removing these references which apparently discuss the same subject:

• Niemi, R.G. (1984) "The Problem of Strategic Behavior under Approval Voting" American Political Science Review 78(4) pp. 952-958.
• Saari, D.G. and Van Newenhizen, J. (2004) "Is approval voting an ‘unmitigated evil?’ A response to Brams, Fishburn, and Merrill" Public Choice 59(2) pp. 133-147.

So, clicking on those links, which are clearly peer-reviewed sources there are some pretty harsh criticisms, such as, "AV is one of the most susceptible systems to manipulation by small groups of voters (for example, small, maverick groups could determine the AV outcome)." How do you respond to those critiques? MilesAgain (talk) 04:35, 13 December 2007 (UTC)

My response: "Huh?"

It's late, but I suppose I can say more. There are "peer-reviewed" criticisms of Approval that clearly did not involve "peer review" by election methods experts. We'd have to look at the specifics, but the claim repeated above is way outside of what is generally accepted about Approval. If I had time, I'd look at the sources, but, instead, I'd invite some more information to be brought here. *How* is Approval, allegedly, so "vulnerable to manipulation." As an example of the patent nonsense that is sometimes bandied about regarding Approval, "vulnerability to tactical voting" is common. What is "tactical voting" in Approval? Voting for your favorite is one form. Another is adding a vote for a frontrunner so that your vote won't be moot. So what's the claim here?

I can't tell, because the articles aren't accessible without payment.... and I can't travel to a library that would have them. Maybe someday.

I gotta love the "peer-reviewed" title of that second article: "Is approval voting an 'unmitigated evil'?" When did you stop beating your wife? Put it another way, if the answer is Yes or No: Yes, it is unmitigated, pure evil, invention of the grand panjandrum himself, calculated to destroy humanity in a single stroke by allowing a voter to actually cast a vote for more than one alternative, horrors! Or, No, it's mitigated by being mostly not used, if any organization is so unfortunate as to actually use it, it will vanish in a sulfurous cloud of smoke. --Abd (talk) 05:13, 13 December 2007 (UTC)

To MilesAgain: As Abd pointed out, neither of your references is available free. Since I am very poor, this means that I cannot read them to verify their truth. Please summarize the arguments here (in your own words, to avoid copyright infringement) so that I can decide whether they are rational or not. Thank you. JRSpriggs (talk) 07:37, 13 December 2007 (UTC)

How about if we start by cleaning up the strategy section. There is a lot of unreferenced assertions, many of which may simply be untrue. I'm tagging a number of statements in the section that need verification and/or citation. I consider them candidates for deletion unless someone thinks they can be salvaged. I'm also ready to delete some of the speculation or nonsubstantial assertions, such as the last two sentences of paragraph 2, maybe the whole paragraph. Similary, the third paragraph can be reduced to the first two-thirds of its first sentence less the "essentially". The Condorcet loser example is overly narrow and obscure, all it takes is for voters to bullet vote. Let me know which parts of this you think can be salvaged but you just need some extra time before I start cleaning up. DCary (talk) 05:35, 7 January 2008 (UTC)

Note the citations above in this section. MilesAgain (talk) 23:13, 7 January 2008 (UTC)

I restructured and improved parts of the strategy section. The last two paragraphs about Condorcet just before the example still need some work. They should probably be moved to another section, Effect on elections for example. The information from Laslier needs to be more accurately presented if it is kept. The information from Brams needs some qualification to make it NPOV. DCary (talk) 21:27, 15 January 2008 (UTC)

I have tagged a number of items in the article that are candidates for improvement or removal. Some explanatory notes:

• What is approval voting? Brams defines it. Claims that anything else is approval voting need similar quality citation. Otherwise it should be distinguished as not being approval voting. Claims of equivalency need to be supported.
• Claims about the benefits of approval voting need to be supported by citations.
• Suggestions for extending/modifying approval voting and/or using such modifications in non-election situations need citation, in part to distinguish them from OR. This is particularly true of the discussions of so-called multi-winner approval voting and approval polling, both of which I would consider as candidates for deletion.

DCary (talk) 21:27, 16 January 2008 (UTC)

As to claims of equivalency, when a fact may be verified simply by looking at the definitions of things, or the uncontested consequences of what is known, it is not necessary to cite a source for it. Now, if someone wants to challenge an equivalency or analogy, to claim that the elements are *not* equivalent or analogous, then we'd have to see if there is any basis for this, or if, alternatively, it's a mere attempt to exclude explanatory text from the article on the technical grounds of a lack of specific source. In the absence of actual contention, I'm not going to give examples yet; I'd prefer to let someone who thinks one of the similarity or equivalency claims in the article is not verifiable by the kind of examination I've described propose it and explain why. Brams indeed defines Approval voting, but then generalizes the term. For example, he claims that various professional societies have adopted Approval Voting. By going to their bylaws, you can see what they do. It is not Yes/No voting, it's simply allowing each voter to vote for as many candidates as they like, even though only one is going to win. Another way of stating this is that such a method assumes No for all candidates not voted Yes. Or that marking a candidate is voting Yes and not marking the candidate is voting No. --Abd (talk) 03:46, 20 January 2008 (UTC)

If it helps move us to actually improving article, please consider each of the items I tagged as contested as to some combination of accuracy and/or neutrality. Rashly deleting the tags without any improvements to the article is not constructive, nor is offering insinuations about my intentions. Those tags should be restored. Deleting the tags does not make the content uncontested, it only reduces the opportunity for others to see that there are problems and for others to have a chance to offer their improvements. Deleting the tags without any improvements suggests a willingness to vouch for the 100% accuracy and neutrality of the content. I'd have been happy to further explain the reasons for any of the tags, if you had indicated your interest. I'll mention a few as starters and will follow up with more as appropriate.

• The source provided later in the article does not support the assertion in the second paragraph of the introduction that approval voting is used to select the Secretary-General of the United Nations. In fact, that reference supports the opposite: repeated approval polling, not voting in an election, is recommended as a part of a consensus building process in preparation for a non-approval election to nominate a candidate for the post. The tagged, unreferenced statement in the introduction is significantly misleading if that reference is the only basis for its support. The statement later in the article makes a different claim that is more appropriate.

• The non-equivalence of voting for none, 1, or more versus voting for those candidates one approves of is recognized and is a distinction Abd himself has promoted. For example, even when there are just two candidates, I might vote wisely, even sincerely in some sense, for exactly one candidate, even though in another sense I don't approve of either one.

• Given the first non-equivalence, it is unclear which of the two is being referred to by the second equivalence. Further more, the equivalency of either to range voting is at best ambiguous since range voting itself offers multiple ways to count votes and declare winners, even if voters vote are allowed to vote only values of 0 and 1. Also there is the whole issue about what is considered a sincere vote and how that may change depending on what a ballot is interpreted to mean. This mention of sincere voting shouldn't be taken as an invitation to resolve here the issues about sincere voting, rather it should only serve as a reminder as to what an unrestricted claim of equivalency should be able to handle.

DCary (talk) 08:15, 20 January 2008 (UTC)

There was no intention to comment at all on the motives of Mr. Cary. I mentioned a possible bad faith situation as a purely hypothetical situation, not one that had arisen, and I apologize for the inadvertent implication. As to the substance, each issue is separate. To separate this from the old POV section discussion, I'll open up a new section at the end of this page with subsections for each issue. Before moving on, though, I just want to say, about removing the tags, that I don't ordinarily do that without discussion unless I change the text to something that I think should not require a citation or I provide a citation. I'm not sure why I did it that way this time, beyond what I wrote in the summaries or above, I may have been confused about how long they had been there. Any user who thinks that what is wrong with the text is that it needs a citation is welcome to place a cn tag again, and any user who thinks the text is incorrect is, of course, welcome to change it. I would not ordinarily place a cn tag on text I think is incorrect, unless it's merely a suspicion and there is no source. --Abd (talk) 01:15, 21 January 2008 (UTC)

[ This thread is continued in the Unsourced text in article section. DCary (talk) 07:37, 21 January 2008 (UTC) ]

Removal and replacement

As far as I can tell from a quick look, all of the concerns in this section have been addressed. So I removed the POV tag. MilesAgain (talk) 02:02, 28 January 2008 (UTC)

Look again, both more carefully and more broadly. The removal (diff) of the POV tag was premature and precipitate. There are a number of outstanding NPOV issues with the article, some of them documented and/or recently commented on before, in, and after this section, while others undoubtedly have not yet been specifically mentioned. The removal of the tag was undermined by the unsourced or POV-laden edits that followed (for example, diff, diff), including some (from diff to diff) that seem oblivious to some recent discussions aimed at improving accuracy and neutrality. The nature of NPOV disputes should discourage quick, unilateral action in removing the tag. Better instead to first continue improving the article, demonstrate some NPOV stability in the article, then seek and achieve some consensus on this page about removal of the tag. DCary (talk) 19:07, 29 January 2008 (UTC)

I'm sorry, I was being a little WP:BOLD. Do you feel that any of those edits you provide diffs for introduce POV problems? I feel quite strongly that they are all completely neutral. Would you please make a brief list of the specific NPOV problems you believe remain? MilesAgain (talk) 17:46, 30 January 2008 (UTC)

I appreciate boldness and occasionally practice it myself. There is a knack for knowing where boldness is appropriate. You make many good contributions, some of them are bold. As to your questions: 1) yes, I do think the edits I provided diffs for certainly involved and compounded, if not introduced, POV problems; that's why I pointed them out; 2) yes, I did, see my previous comment. Perhaps that was too brief. So, do you want a brief list or a complete list? Kept up-to-date? Reiterating or summarizing the points of views and assessements from me and others? So other editors don't have to read what other editors have recently said on the issues? I'm not volunteering for that. In my previous comment, I offered some suggestions on how to find some of the outstanding NPOV issues. I'll note that I also suspect every one of the "historical" examples in Uses section has some NPOV problems. One could also read through the article and find the existing fact tags. Or just note all the claims made without referencing a reliable source, making particular note of those claims that might be controversial. Other starting points would be to write some comments in and relating to the existing discussion explaining how the recent changes to the first sentence (now first two sentences) in the article were "all completely neutral." Or start a new section that describes how making an edit change (boldly, but without prior discussion or consensus), hinging on value judgements of what one editor thought was more important and less important, is an example of making completely neutral point of view changes. DCary (talk) 06:32, 31 January 2008 (UTC)

I would like to see a specific list of the outstanding NPOV issues, with clear reference to the text in dispute for each, including those issues that I may have contributed to, please. It would be very helpful to have something more than references to other unspecific sections of this talk page. My understanding that POV issues usually pertain to the wording of the article body text, and much less often to the citations or lack thereof. As for the first paragraph, here is what I did:

before	after
Approval voting is a binary, single-winner voting system used for elections, which allows each voter to vote for, or approve, as many of the candidates as desired. The winner is the candidate receiving the most votes. Each voter may vote for any combination of candidates and may give each candidate at most one vote.	Approval voting is a single-winner binary voting system used for elections. Each voter may vote for (approve of) as many of the candidates as they wish. The winner is the candidate receiving the most votes. Each voter may vote for any combination of candidates and may give each candidate at most one vote.

Where is the neutrality issue? MilesAgain (talk) 11:41, 31 January 2008 (UTC)

Sometimes improving the wording of an article has NPOV implications, sometimes it doesn't. Avoiding unintended consequences requires some understanding of the NPOV issues.

NPOV disputes are about which points of view are presented in the article and how they are presented. The resolution to those issues depends heavily on the points of view that are expressed in reliable sources. A point of view does not qualify for inclusion in an article just because an editor subscribes to POV or knows people who do. Once sources are identified, editors can evaluate them for reliability, examine their content, and make decisions about whether and how to include in the article any material based on the sources. Asking for reliable sources can help provide a good basis for establishing consensus about NPOV and can help avoid irrelevant discussion.

I'll address a specific point of concern in the Talk: subsection where it had been recently discussed. DCary (talk) 04:34, 6 February 2008 (UTC)

The silly Binary terminology was added by an anonymous Approval-supporter overriding my efforts to organize the election methods with a one vote category on the Voting system page, replacing another silly yes/no category. [10] The anonymous swarm wouldn't accept that one vote systems existed, so they came up with a binary category to make approval fit in with plurality because they can both use identical ballots. (I guess it was changed from yes/no because that implied 2 types of marks, while he wanted to make it clear it was mark/no-mark binary.) Of course all nonsense since (1) plurality doesn't need a binary ballot and (2) approval can use a 2-mark ballot as well as a 1-mark ballot.) Happy editing! Tom Ruen (talk) 18:14, 1 February 2008 (UTC)

"Binary" was added to this article very recently by Dcary[11]. However, I assume he simply found the category and used it. It *usually* is binary (as is Plurality), that is, for each candidate the voter has two options: a vote, or no vote. (But, of course, if abstentions have a different effect than No, then we have three options. Otherwise the explicit No is dicta or there for non-electoral purposes or security.) The only difference between Plurality and Approval is the restriction in Plurality requiring, effectively, a No vote (blank is the same as No) on all candidates but one. I really consider Plurality and Approval to be the same basic system, and in parliamentary practice, with direct voting by persons, sequentially, they are the same. (But where votes are recorded member by member, an objection *might* be raised. Anyone ever see this? It is with written ballots that the difference pops up.) Perhaps I should take a look at the category debate. I do think "binary" a useful classification. "one vote" is ambiguous -- or different from binary. --Abd (talk) 18:56, 1 February 2008 (UTC)

It is NONSENSE to call plurality a binary method unless there's only one choice. Binary implies 2 responses per choice, which is 2^n possible votes, while plurality has n possible votes. It's an accident of ballot design convenience that plurality and approval share ballot types. Original plurality voting, everyone has ONE MARK, whether writing a name on a paper ballot, or physically voting with their bodies by moving around a room to represent their vote. There's nothing binary about it at all. Only approval can offer this claimed term. Tom Ruen (talk) 21:37, 1 February 2008 (UTC)

Okay, this essentially asks me to look at this in a very picky way. What is the source for "binary"? How is the article improved by having it in the introductory definition? Take out "binary" entirely? Why would this make any difference? As has been noted, Approval may be seen as a special case of Range Voting, with binary votes; however, the reverse is also obviously true: Range Voting is Approval with fractional votes allowed. Range Voting and Approval Voting are rather artificially divided, based on history. The people who started promoting Range thought of Range as involving higher resolution, such as 0.0 - 10.0 used in Olympic scoring (and Warren Smith would prefer real numbers in the range of 0-1, and that is what his simulations -- and those of some others -- use in estimating performance of elections, though I think the actual votes are reduced to the allowed resolution). The core definition of Approval is that the voter expresses an action of approval or disapproval (the latter often by default) for each candidate. It's true that in practice, the term Approval is limited to binary votes, and Range is used for higher resolution. There is a blended case, though, which is Range 2, one step up from Approval, where voters cast Yes/No votes, but abstentions have an effect. There are thus three votes; typically the No votes are subtracted from the Yes votes. Thus the three possible votes are -1, 0, +1. This is what is used for some Wikipedia polling, as an example. ArbComm elections are not actually determined by the poll results, the complete results, as is, which do not declare winners, is then advice to Jimbo Wales. He could decide, if he wanted, to use average vote, or net positive vote, or maximum Yes vote. From this choices, though, it appears he is using net positive vote, which makes it Range 2 in practice. (Range 1 is Approval). Because the actual votes are Yes/No and are independent for each candidate, however, most would call this an Approval election. This is the introduction, and some pickiness is appropriate for it. I don't see "binary" as necessary, I'd leave the subtle ambiguity, or explicitly explain it. Since the explanation might be considered original research (though it's pretty obvious and thus might be defensible), I'd not put that in the introduction, for sure.

Then the other picky point. With Range Voting, there are two basic ways of determining the winner: Sum of votes and average vote. Averaging can be used with Approval just the same (same reason for doing it, same problem with doing it). Averaging probably improves the outcome as long as there is some rule preventing write-in votes from automatically winning! -- but this has never been shown conclusively, it's just an opinion that some Range Voting supporters have. (Averaging in Approval is a little like IRV discarding moot votes (not for the top two) and not considering them as part of the basis for a majority. Again, Approval Voting can (and probably will) be used together with a majority requirement, with further process (runoff, exhaustive balloting with or without eliminations) ensuing if there is majority failure. This would probably be the most common exception to what the introduction reads. If Approval is implemented in the U.S. for public elections, my guess is that the first implementations will be as runoff replacements (as is happening with IRV), but with Approval, I'd assume that the majority requirement would remain. It has not with IRV, I believe, primarily because IRV has been oversold as a runoff *replacement*, and that it can fail to find a majority hasn't been a factor in the debates; as I've written elsewhere, the voter information pamphlet, in the putatively neutral summary, explicitly stated that "a winner will still be required to gain a majority." That is directly contrary to what the proposition actually did, unless we accept that a final round majority is a true majority, and, as I've pointed out, we might as well say that unanimity is required under the new system; just take it one more round, and it will always show unanimity for the IRV winner.

In any case, Approval with a majority requirement, Approval with No votes subtracted from Yes votes, Approval with vote averaging, all would be called (and I think are called, but I haven't searched) "Approval." Key, again: independent voting for each candidate, in a manner that expresses "Approval" -- in the action sense, not necessarily the emotional or moral sense -- in the range of 0-1, or reducible to that range. (Yes - No Approval has votes of -1, 0, +1, which linearly maps to 0, 0.5, 1, so the social ordering from one is exactly the same as the social ordering from the other, these are all "one vote, one person" systems. (And there are other possibilities twittering out at the fringes of what people discuss).

The fractional vote issue is quieted by simply leaving out "binary." Most people will, in fact, assume binary votes, we don't have any habit of thinking of fractional votes (even though that is what Olympic scoring is).

The issue of determining the winner could be covered by weaseling it: The winner is generally the candidate receiving the most votes. Contrary to some expressions I've seen, weasel words are quite appropriate for introductions, so that a summary can remain accurate, fairly distilling the rest of the article; in this case, the exceptions should be mentioned in the article, and the notable one would be that a majority might be required, hence the one with the most votes might not win in the primary, and could lose in subsequent process. I think we have to remember that voting systems are not only used in government and, in fact, the vast majority of elections are held outside of government, in countless organizations. Robert's Rules would clearly suggest a majority requirement. (It appears to prohibit Approval Voting, but, in fact, it merely instructs clerks to discard overvoted ballots because they are obvious "errors." There is no rule prohibiting overvoting in elections by show of hands or rising, only with written ballots does the problem appear. So, for written ballots, any organization could specifically allow Approval voting, and, as we know, some significant ones have.

--Abd (talk) 16:54, 1 February 2008 (UTC)

The first wikilinked paragraph from the phrase containing "binary" says, "Binary voting systems are those in which a voter either votes or does not vote for a given candidate." I've never heard the term "binary" applied to voting systems, so I added a fact tag, and I'm going to remove it from the intro sentence here because it doesn't say anything that the rest of the paragraph doesn't. MilesAgain (talk) 17:57, 1 February 2008 (UTC)

Good points and good edit. Such are the pitfalls of my relying on and replicating material from other pages that is not demonstrated to be supported by reliable sources. DCary (talk) 04:34, 6 February 2008 (UTC)

I agree that it was premature to remove it. An action like that should be *explicitly* approved by the active editors, even only one who, with a sustainable assumption of good faith, objects, should be enough to keep the tag, pending further process. There should be, however, a specific and active list of POV issues, either text in the article that should not be there, or missing text that should be there, but perhaps we haven't yet come to consensus on it. So if anyone thinks the tag should remain, please start a new section below with a specific list of issues. At least one! It does not have to be complete, more can be added later, but we need something to work on. (I could identify possible POV issues, indeed, but it's a matter of available time.) Further, not all editors check their watchlist every day. Time should be allowed. *At least* a week! More would be better. I appreciate the apparent intention of MilesAgain, but ... not quite yet.

--Abd (talk) 19:56, 29 January 2008 (UTC)

Creating new sections for issues that have been discussed recently elsewhere on this page may only fracture the discussion. In many cases, I'd prefer to build on what we have than start anew. I agree about expecting quick responses and being limited by available time. I'm not volunteering to start lists or keep them up-to-date for people who can't find the numerous issues themselves or don't have time to even list their own issues. Creating lists would also tend to encourage more fractured discussions by discouraging consideration of what has already been said. There is a lot of discussion to keep track of. In the big picture, concise, well-considered, and edited writing is the ultimate time saver. It is considerate writing as well. I'll continue my work on NPOV issues, necessarily not all at once, keeping in mind that much of the effort does not involve writing on Wikipedia. DCary (talk) 06:32, 31 January 2008 (UTC)