Talk:Voting methods in deliberative assemblies/Archive 1

Approval Voting

It is clear that Robert's Rules do not contemplate Approval Voting. Approval Voting is, however, merely a matter of counting all the votes cast, instead of assuming that votes for more than one in a single-winner election are errors. The instructions to the clerk are based on an assumption that they are errors, but nowhere in the Rules have I seen a justification of the practice of discarding overvotes that is not simply this circular argument: they are errors because they will be discarded, and they are discarded because they are errors. *why* are they errors?

In making multiple choices by show of hands, or standing, there is no rule to prevent voting for more than one. I have never seen any objection to such voting, and I'm sure it has happened. Now, I'm pretty sure that it has been assumed that voting by ballot would produce the same results as voting by show of hands, but if overvoted ballots are tossed, the results may indeed be different.

I would interpret the section of RR where overvotes are to be discarded as dicta, because there is a precedent condition that may not be true: "because it is not possible for the tellers to determine for whom the member desired to vote". If an ordinary motion passed to specifically allow overvotes (more votes than winners), then such votes would no longer satisfy the condition that it is not possible to determine the intention. The votes would be taken as writ, which would have been a better rule in the first place.

If overvoting is to be considered contrary to bylaws, requiring a 2/3 majority to amend, then the matter should actually be debated! It's a bit odd that Robert's Rules doesn't consider Approval Voting, because it has been used in governmental elections, in the early days in the U.S., and, as I mentioned, overvotes are not prohibited with show-of-hands. What should the clerk do when recording the totals counted? What if someone objects that a voter voted for another candidate previously?

The claim that the clerk cannot tell the intention was not true. We presume that voters who cast interpretable votes intended to vote as they voted. They can make errors, to be sure. But it is closer to democracy to count the votes they cast, unless overvoting has been made explicitly prohibited, and not merely through an assumption of error.

It is actually a difficult question; no overvoting rules are, in public elections, a matter of statute. It's been noticed that one can implement a relatively advanced election method by simply striking a couple of lines from the election code, one that resolves the objection made to preferential voting in the more complete discussion that is not included on this page. What an amazing thing: it is actually a very deep and significant reform to simply Count All the Votes! Abd 03:16, 27 September 2007 (UTC)

Approval voting is better than plurality but nothing beats multi-member single transferable vote when it comes to getting that perfect blend of proportionality and being able to elect the specific candidates that you want. I think the reason we don't have it in America is that that the politicians basically believe we are too dumb to be able to understand the concept of putting a "1" next to your first choice, a "2" next to your second choice, etc. They say it would confuse us so much we wouldn't bother to vote anymore, and voter turnout would suffer. What's more, they apparently think immigrants are the dumbest of us all, because they say that that voter turnout among those groups would be harmed even more than among the rest of the population. Captain Zyrain 04:25, 27 September 2007 (UTC)

Approval voting iSN'T better than repeated balloting. Approval voting can't determine a singular majority choice. If approval voting is used in repeated balloting, it would allow a strategy of false support to give a false impression of popularity to other voters. Counting "overvotes" in a plurality/bloc ballot can be done by fractional counts to the tied choices, if the voter is not able to recast their vote. Anything else isn't acceptable. Tom Ruen 04:31, 27 September 2007 (UTC)

No claim has been made that Approval Voting is better than "repeated balloting." Robert's Rules of Order recommends repeated balloting. The proposal that overvotes be divided is unjust, and there is legal precedent for considering it so. (Robert's Rules instructs the clerk to discard ballots with overvotes, but only on the grounds that they are "errors," not that it would be unjust to count them.) Approval is a form of "alternative vote," i.e., if one votes for A and for B, one is voting for A in every pairwise election not involving B and for B in every pairwise election not involving A, and is abstaining from the A/B pair. While it is true that it is possible for more than one candidate to gain a majority in Approval, it is standard democratic practice with initiatives and referenda, when Ballot Questions are presented for simultaneous vote, and more than one conflicting Question gains a majority, for the one with the most Yes votes to prevail. That's Approval Voting, as to the handling of multiple majorities, in a nutshell. I haven't reviewed the cases, but there were apparently forms of Bucklin Voting that did divide up multiple approvals, possibly as Ruen is suggesting, and that was ruled unconstitutional, and I'd agree. On the contrary, Brown v. Smallwood ruled Bucklin, in a form that was like Instant Runoff Approval, to be a violation of the Minnesota Constitution, on the basis that it allowed "alternative votes," and here I would disagree. In any case, we should be so luck as to have the multiple majority problem, it would be extremely rare in public elections (it would require a significant number of voters to vote for both frontrunners). And none of this has much to do with this article, but rather with attempts of FairVote activists to use the mention of Preferential Voting in Robert's Rules for promotion of Instant Runoff Voting. --Abd 05:57, 6 November 2007 (UTC)

Hmmmm... long time for a reply.... I claim approval voting is incompatible with a repeated balloting process. So under such a process, I suggest at best counting overvotes as divided votes as a more fair approach than not being counted at all, and mainly assuming people WANT to support multiple candidates in a given round, basically generalizing to a repeated balloting by a fractional Cumulative voting system. Tom Ruen 06:06, 6 November 2007 (UTC)

First of all, the most relevant aspect of this, here, is "repeated balloting," which is recommended by Robert's Rules. Repeated balloting, in Robert's Rules, means entirely new elections, these are *not* runoffs, no candidates are excluded on the basis of the first round (RRONR p. 426), unless they voluntarily withdraw, and the vote received in the first election is irrelevant to the votes in subsequent ones, so it would seem that whether or not the first round allows multiple votes is not affected by "repeated balloting." If the votes manage to find a majority winner, then there would be no further rounds (under Robert's Rules unless there are specific rules in the Bylaws otherwise). And if they do not, they are moot. I think Mr. Ruen is confusing repeated balloting with the rounds in an Instant Runoff Voting election.

Secondly, with regard to Approval Voting, RR instructs the clerk to discard overvotes, but the only reason given is that "it is not possible for the tellers to determine for whom the member decided to vote." Thus RR is actually silent on the question of whether or not voting for more than one, in a single-winner election, could be allowed, for if it were allowed, there would be no question raised by multiple votes, the tellers would count the votes as cast. Dividing the votes into fractions is, as Mr. Ruen noted, a form of Cumulative Voting, which is quite a different method, and it is clearly inequitable for single-winner elections. Consider that a voter favors A but also votes for B. If it turns out that the frontrunners are B and C, the vote for A is moot. In an Approval election, it could be eliminated without changing the result (which is why Approval Voting is a form of alternative vote: the vote for A and B is a vote for A in case that is relevant, or a vote for B in case that is relevant, and if both votes are relevant, it is an abstention between them. In no case do the double votes represent an extra *effective* vote that unjustly harms the election of a third candidate; it is always true that one of these votes could be eliminated without changing the outcome. Now, if the votes are divided, then the vote count for B is reduced by one-half vote, which could change the outcome. By voting also for A, the voter has reduced the vote for B; if B and C are the frontrunners, but A is the favorite, Favorite Betrayal is then strategically forced. It defeats the purpose of Approval Voting, which is to avoid such a need for Favorite Betrayal; were this necessary for equity, fine, but Mr. Ruen has simply claimed that it is "incompatible" -- and elsewhere I've seen the claim that full counting would be "unfair" -- but no logic or evidence has been shown where an actual inequity or unfairness would occur. And especially not with repeated balloting. If the extra votes do any harm, it is only to the voter casting them, just as any ill-considered vote can harm the voter. But in contested elections where there are two clear frontrunners, such harm would be extraordinarily rare: it requires that the voter vote for both major candidates. It happens, but rarely would this be a deliberate vote under current conditions unless the voter truly intends to abstain from the pair. --Abd 05:01, 7 November 2007 (UTC)

The only reasoning that makes repeated-balloting makes sense is for voters to influence each other's choices by the vote tally to encourage compromise. An approval count offers no secure information for compromise because voters who support more candidates will have a larger influence on the apparent distribution of support. Strategic voting under this circumstance rewards insincere overvoting to falsely encourage voters to abandon stronger candidates under the illusion that a compromise is actually stronger. No sane person who understands this would support repeated ballots and approval voting. This is merely my opinion, but it convinces me that approval isn't compatible with any runoff process. you can delude yourself otherwise as you like. Tom Ruen 06:08, 7 November 2007 (UTC)

Tom Ruen is an activist with FairVote, promoting Instant Runoff Voting, and it has been a tactic of FairVote to invent objections to other voting methods, such as Approval Voting, Range Voting, Bucklin Voting, and Condorcet methods (in addition to legitimate objections). Here he promotes an entirely new one, that voters will allegedly, in a context where they don't know each other's preferences, "insincerely overvote." Yet the most common criticism of Approval Voting, repeated often by FairVote, is that most voters do not add extra approvals (this was the excuse of the IEEE board in stopping the use of Approval). There is no evidence from the literature that Approval voters would vote as Ruen describes, it is something he has apparently made up -- I'd be happy to see sources to the contrary! -- and yet he describes an opinion "otherwise" as being a matter of "deluding yourself."

It is very strange to be categorized as an activist with Fairvote, while my opinions are my own, even as influenced by other individuals. Overvoting for strategic advantage on a nonbinding vote is easy and safe. Repeated balloting already contains incentives to imply false support (and I've seen it used in practice, with anonymous paper ballots anyway), and overvoting would just expand this incentive. The only thing that makes Approval "harmless" (as a varation of plurality) to me is if the results of a single vote are binding. Tom Ruen 20:14, 7 November 2007 (UTC)

Generally, a single vote is binding under Robert's Rules if a majority is obtained. Further, if a majority is not obtained, unless there are rules allowing election by plurality or there is other special process under the Bylaws, the vote is normally as if it had not happened. There is an election de novo, candidates are not dropped; Robert's Rules recommends strongly against such dropping. There are fairly strong forces against insincere approvals; as FairVote often points out, voting for another can nullify your vote for your Favorite (as against the additional approval). Ruen mentions "incentives to imply false support," and that he has "seen it used in practice," but this is about repeated balloting, a basic democratic process, and one clearly recommended by Robert's Rules. I have some suspicion that Ruen is not talking about the repeated balloting that RR has in mind, but repeated balloting in rounds with candidate dropping. There is known strategy in that configuration, which, by the way, is strategy to which IRV is vulnerable, and is precisely the reason that RR dislikes candidate dropping. --Abd 23:18, 7 November 2007 (UTC)

I've seen repeated balloting done, requiring 60% support for a winner, and no forced elimination in the first 5 rounds or so, with voluntary withdrawal, and later progressively higher threshold, 15, 20, 25% support to avoid elimination in later rounds. If RR were to use repeated balloting, with approval overvoting, terminated when a single candidate exceeds 50%, then I accept insincere overvoting is dangerous, and sincere overvoting is also dangerous. The only GAIN I see is people can continue to support a favorite who can't win, while supporting a compromise, and feel good or something. I still have no faith in the value of a vote where some people have multiple votes, still seems just opening unnecessary power for mischief. I wouldn't support it if someone suggested, I'd argue against it, and I'm be very surprised if a room full had a majority of people to agree to it. My sense is if a majority only see value in one vote, the question is moot, and the minority approval lovers can pout if they like. Tom Ruen 23:54, 7 November 2007 (UTC)

The relevance here (albeit thin in itself) is simply whether or not, as Ruen claims, allowing overvotes is incompatible with repeated balloting as described in Robert's Rules, and I haven't seen any evidence presented other than this bizarre speculation, which runs entirely in contrast with centuries of experience with Approval Voting (as was used with repeated balloting for the election of Popes, for example); the common problem is *refusal* to add additional approvals, i.e., usually the claim of "strategic voting" with Approval is the opposite of adding extra votes, that voters will, even though they supposedly "approve" of a candidate, vote only for their favorite. Voting for a dummy candidate to raise him or her to involvement in compromise status would be very dangerous, unless you actually do approve of that candidate, in which case it is not insincere.

My hope is that the editors of Robert's Rules will include more alternative election method information in the next edition. I do find it a bit odd that I've been unable to find, in Robert's Rules, any justification for prohibiting overvotes (beyond what I've mentioned, which *assumes* that they are errors), it is as if the idea simply never occurred to the editors. Yet Approval *has* been used for elections by deliberative bodies, and any concern about Majority Criterion failure can be addressed by holding another poll whenever it is possible that a majority has not explicitly consented to the result, which would be majority failure *or* multiple majorities. But I think most organizations would not bother with the latter and would instead simply elect the candidate with the most votes, unless an additional vote were easy to take. --Abd 16:31, 7 November 2007 (UTC)

Language moved from Instant Runoff Voting, to be integrated into this article at some point

IRV is discussed in Robert's Rules of Order Newly Revised (RONR) as an example of "preferential voting." RONR recommends the use of repeated balloting to obtain a majority vote for the winning candidate in officer elections, describing preferential voting as a "more complicated" system that "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 or proposition in last place is automatically eliminated and may thus be prevented from becoming a compromise choice". It does state that preferential voting is "especially useful and fair" if it is impractical to take multiple ballots (e.g. if the election is held by mail), stating that "in such cases it makes possible a more representative result than under a rule that a plurality shall elect". IRV (or any other alternative voting system) can only be used if the bylaws specifically authorize it. RONR details the vote tabulation procedures for IRV as an example of how to conduct preferential voting for single-member offices. RONR notes that this description is provided only by way of illustration and that there are many other possible methods of preferential voting. It also cautions that voters unfamiliar with preferential voting should be instructed as to how the system works so that they will be able to vote properly. Captain Zyrain 15:40, 10 October 2007 (UTC)