Voting on: Disallowing extending votes beyond the initial "Vote ends" date provided on the vote page.
Duration note: The vote is set for three months, and is not expected to be extended, to prevent discussions about circularity or recursiveness of the vote.
Support. My first choice is the traditional practice of closing votes which show no consensus as "no consensus", and then if need be having more discussion and a better vote (with refined wording, or after working out some of the side issues that blocked the first vote, as happened with the "well-known work rule" votes). What Dan suggests on the talk page (a set of rules for when to extend votes in a way that minimizes selection bias) would be my second choice, although I think it might be good to have some cutoff to keep votes from dragging on indefinitely as several have already shown a tendency to; e.g. "no vote shall last more than 3 months (even if this means it is closed as no consensus)". - -sche(discuss)17:40, 8 August 2015 (UTC)[reply]
A vote that keeps on dragging on is better than a vote that is closed as "no consensus", IMHO. I do not see any harm im votes lasting more than 3 months. For instance, Wiktionary:Votes/2014-08/Migrating from Template:term to Template:m would have benefitted from being extended even longer than it was; I stopped extending it because people started to object. For some reasons, some people came to vote only after several months. From what I have seen, the fact that the vote was opened for a long time did not in any way diminish its consensus based character. More votes passing via repeated extensions, even those suffering from some small degree of something like selection bias, is much better than volume non-consensual changes indirectly supported by many here. When the same people who did not oppose and take action against non-consensual volume changes made without a vote now find a little confirmation bias issue with a real, transparent vote, I don't really know what to think. The important thing to me is to have votes that work better, and the repeatedly extended votes do work better, from what I can see. --Dan Polansky (talk) 14:34, 9 August 2015 (UTC)[reply]
Support in general (weak support). Perhaps there could be cases when extensions should be allowed, e.g. when nobody voted on a topic, which is of interest to at least a few editors and some editors expressed interest in extending a vote. --Anatoli T.(обсудить/вклад)01:14, 13 August 2015 (UTC)[reply]
God, one year is too much. (at least IMO) If a vote takes one year to reach a result, maybe the voted proposal was not so popular in the first place? Feel free to prove me wrong on this.
@Dixtosa and @Yair rand: Also, I like this idea (diff) "All we should do is to show active votes more actively. For example, putting them in the watchlist page below Wanted Entries will dramatically increase the awareness of new votes. This has been suggested by YairRand ••Dixtosa (talk) 10:36, 19 July 2015 (UTC)" . --Daniel Carrero (talk) 08:43, 20 September 2015 (UTC)[reply]
The continuously-extended vote gets extended for a full year only if there is a continuous interest in the vote. The algorithm for the continuously-extended vote is this: if at least one new vote was cast to the vote in the last extension period, extend further by one more month, unless one year has already passed. The algorithm was used in Wiktionary:Votes/2015-03/Templatizing topical categories in the mainspace, which is no longer being extended, but while it was, it gathered more votes and became more representative. Let me point out that nominations in RFD and RFV do linger for a year or longer, and it is not even necessary that new votes are cast there. --Dan Polansky (talk) 09:13, 20 September 2015 (UTC)[reply]
Ok, your algorithm makes sense. Still, one year is really too much in my opinion, even if there is a continuous interest in the vote. Maybe shorten it to 6 months or 3 months. Since most of the supporters here are commenting their votes along the lines of "relutanctly", "weak support", perhaps we shouldn't actually disallow extending of votes after all. I am going to change my vote. --Daniel Carrero (talk) 06:37, 27 September 2015 (UTC)[reply]
@Daniel Carrero: One year under continuous interest is fine; policy chances are more important than individual RFDs and RFVs and yet we let RFVs linger for a year. But I am fine with 6 months if that's the community preference. 3 months is too low, IMHO, based on our experience with continuously extended votes. Right now, it looks like the community wants to have the end date frozen. If that comes to pass, I will have to create votes lasting for 3 months to give them a good chance. And I feel quite embarrassed to do that since that makes even obviously failing votes linger for 3 months. I appreciate your having actually considered what I am saying and its relation to selection bias. Furthermore, if you have objections to some particular vote-extension practice, please say so. Like, you might object to extensions that do not state any extension algorithm. I am guilty of extending without rationale since that seemed too formal to me, but you never know: people who are informal enough to want to do everything without votes are suddenly very formal when it comes to extending votes. I guess I know why. --Dan Polansky (talk) 07:41, 27 September 2015 (UTC)[reply]
Support per -sche predominantly. Although I'm not here that often, I'd prefer that a vote close before I have a chance to get to it than that a vote drag on for ever and ever and get mired in terrible politics as often happens. --Neskayasprecan?01:22, 22 September 2015 (UTC)[reply]
@Neskaya: "Before I have a chance to get to it": if you don't want to be getting into votes, you don't have to, right? Just ignore WT:VOTE and you're done. As per WT:VOTE, there are always some votes that you can get into, so better not look at WT:VOTE. Also, if the end dates become frozen, it may lead to more votes created to run for 3 months, so it is unclear whether the mean vote length increases or decreases as a result. --Dan Polansky (talk) 07:45, 27 September 2015 (UTC)[reply]
I don't think she's saying she wants to be prevented from voting, I think she's saying she'd rather be prevented from voting (=undesirable) than have votes be extended (=more undesirable). Creating votes to run for 3 months would deviate from the usual practice (status quo) of votes lasting one month or less, so it's uncertain that such a practice would become widely allowed even if some user(s) wanted to implement it. Other community members might decide to overrule a vote-drafter's attempt to prescribe a 3-month vote just as they might decide to overrule a vote-drafter's attempt to prescribe a nonstandard closing formula like "51% = pass". - -sche(discuss)18:43, 27 September 2015 (UTC)[reply]
This is really getting too far. First people complain of confirmation bias, claiming that the vote should be created to run longer instead of being extended. And now someone insinuates that it is somehow forbidden to create votes running for three months. It is not forbidden; it is actually already happending and it does not suffer from confirmation bias. And the practice is not that policy votes last "one month or less"; one month is the very minimum, and we have extensive previous practice of votes running much longer than a month. --Dan Polansky (talk) 06:33, 3 October 2015 (UTC)[reply]
@Ungoliant: I agree that what happed at Wiktionary:Votes/pl-2014-07/Allowing well-attested romanizations of Sanskrit was bad, and I remind the reader that I, while being the person who did the extensions, was first to protest as per User_talk:Stephen_G._Brown#Allowing_well-attested_romanizations_of_Sanskrit ("I think it inappropriate for a vote to be closed on the same day on which a vote was cast that brought the balance just above the 2/3 threshold. That is especially so if the vote was already running for 5 months.") I was also the person who reopened that vote (diff) once it was no longer locked. The evidence is now clear that I was not a lone dissenter, and that multiple people see the problem of selection bias that the closer of the vote did not see. On the page of the present vote, I posted an extension mechanism or algorithm that reduces selection bias to minimum. --Dan Polansky (talk) 08:09, 31 October 2015 (UTC)[reply]
@DCDuring: May I ask whether this vote is driven by the concern with confirmation bias? And may I ask whether there is any appreciable confirmation bias in the algorithm for the continuously-extended vote, which is this: if at least one new vote was cast to the vote in the last extension period, extend further by one more month, unless one year has already passed. --Dan Polansky (talk) 09:04, 3 October 2015 (UTC)[reply]
If we require that the vote duration specified at the time of the start of the vote is binding and cannot be changed during the run of the vote, then we either have to create relatively short votes that get too little chance for gaining cast votes or we have to create long votes such as for 3 months, which gives more chance for people to come to the vote but also creates an avoidable delay for those votes that get a clear result fairly early. And then, even 3 months can prove too short a period. I think the following is a good principle for repeatedly extended votes:
If a vote is getting the result of no consensus, and if at least one vote was cast to it during the last month or extension period, the vote can be extended in good conscience. A result is "no consensus" if it is 50% in support or higher but lower than a pass.
The above principle involves a certain amount of something that resembles (but is IMHO not identical to) W:selection bias: without its application, more votes would be closed as "no consensus" which in practice is similar to "fail". On the other hand, the votes that pass as a result of extension are those that gained more cast votes as a result of the extension and are thus more representative of the positions of all editors. Furthermore, the longer the vote lingers on the WT:VOTE page, the less plausible it is for a regular editor to claim that they did not see the vote or that they did not find the time to consider its proposal, do the research and thinking, and oppose if applicable.
Oppose I think that extending votes is a good idea if (1) there were votes cast close to the closing date, as in hours or just a few days before, and (2) the status is not clear because the tally is right on the dividing line. Even in the case of a technically late vote, though one made before a decision, it is better to keep things open than to either disregard the late vote or include only it, either of which could be seen as preferential treatment. However, votes should not be extended indefinitely, especially if there is a clear outcome (meaning 60% is failure, end of story), but even if there is no recent activity they should just be closed as no consensus. Also, it's not necessary to extend them beyond 1 or 2 weeks at a time (at least 1 week from the date of extension, no more than 2 weeks beyond the initial closing date). DAVilla06:27, 10 August 2015 (UTC)[reply]
I would argue that extending by a mere week rather than a month creates more of a selection bias. And it does very little in giving votes significantly more chance to gather participants. --Dan Polansky (talk) 07:05, 10 August 2015 (UTC)[reply]
There is selection bias inherent in the term of the vote, so of course keeping it open for a month more is going to give a better chance for participation. If you think it is necessary for the vote to be open that long, though, then it should have had a longer timeframe in the first place. The point of extending the vote is not to bring correction to an interval of time that was too short, it is to round out an arbitrary cutoff date when doing so would be useful to determination of the outcome.
I should also say that I don't believe it's the author's prerogative to extend a vote in order to achieve consensus, regardless of any argument about decreasing selection bias over a longer period of time. If this is the level of control you desire over votes you've initiated, it might be wiser to step back, recuse yourself in a way, and simply let others close your votes for you. DAVilla02:46, 11 August 2015 (UTC)[reply]
Re: "There is selection bias inherent in the term of the vote": I cannot confirm this. Please read W:Selection bias and clarify why any vote should automatically (inherently) have a selection bias. Selection bias can be exemplified on a clinical trial. It occurs when you run a clinical trial and end it at a date different from the one originally set up for the trial; you keep observing results at various prospective dates of termination, and pick the date of termination that looks best for you; and because there is certain random variation, your ability to pick the termination date translates into your ability to pick a more favorable result, and skews the result. If a clinical trial ends on the date originally set up for it, there is no selection bias; similarly, if a vote ends on the date originally set up for it, there cannot be a selection bias: the closer does not choose the termination date, and this lacking choice cannot translate to choice of more favorable result. Generally, the higher the number of candidate termination dates, the higher the selection bias. On the talk page of this vote, I have presented a mechanism that lets the choice to extend or not extend be made only once by a human: after the first month. Thereafter, the extension is under control of the algorithm, and is result-agnostic, depending merely on whether new votes are coming to the vote. --Dan Polansky (talk) 08:38, 11 August 2015 (UTC)[reply]
The timeframe establishes a bias in that "some members of the population [would] be less likely to be included than others", namely those who are less active. Since that delineation between contributors isn't usually a major concern, I'm sorry if I misunderstood the kind of bias you meant. On your point, I agree that selecting a termination date based on the data is a stronger bias, which is why I only consider extending the deadline when the result is not clear. Extending a failed vote in the hope of gaining more support, even if it's done once, is fishing for the result you want. That's a more important point than the duration of the extension, which I'd only mentioned as an aside. DAVilla05:31, 12 August 2015 (UTC)[reply]
The above user Zo3rWer is an incarnation of Wonderfool, a user who has had many incarnations over the years, and seems to enjoy being blocked once in a while only to get a new incarnation. He is not blocked and is currently in his productive stage, expanding the content of en wikt without being blocked. Through his incarnations, Wonderfool has contributed a significant volume of lexicographical content to en wikt. I don't know what it means to be "banned" on en wikt and whether he is, but given he edits, the supposed "ban" does not seem to be in effect. The gravest fault with the vote that I see is that it supplies zero rationale, but the same fault is in multiple support votes. --Dan Polansky (talk) 08:57, 31 October 2015 (UTC)[reply]
Oppose: the current one-month recommended opening time for votes appears to be slightly on the short side for some proposals. I am in support of disallowing indefinite extension of votes (e.g. six months or one year could be a hard limit), but lesser individual extensions do not seem like a problem. --Tropylium (talk) 17:24, 28 October 2015 (UTC)[reply]
@Tropylium If your concern is that one month is just "slightly" short, what if we just set votes (from the outset, so that no extending is needed) to run for a slightly longer time, say two months? A concern with allowing extensions is that while you may not intend for long extensions to result from the vote you've cast here, proponents of extending votes have said they're fine with extending things for even a whole year (Ctrl+F "year" on this page). - -sche(discuss)17:37, 28 October 2015 (UTC)[reply]
Something of the sort sounds reasonable in principle. That doesn't appear to be what we're voting on here though. Extensions running for several months doesn't strike me as as big a problem at all, while votes being cut short before reasonably active editors have had a chance to weigh in would be. (Perhaps we need a more nuanced follow-up vote.) --Tropylium (talk) 17:57, 28 October 2015 (UTC)[reply]
I said that I am okay with the hard limit being one year but I also said "But I am fine with 6 months if that's the community preference." I don't see anyone else making any comment about one year. I will take this opportunity to ask -sche, what is your rationale for supporting the vote's proposal? I read your vote comment and it does not read like a rationale; it reads as a statement of preference. --Dan Polansky (talk) 18:01, 28 October 2015 (UTC)[reply]
Oppose action as currently stated. I would support imposing a limit, such as 3 or 6 months, on the maximum duration of a vote and perhaps imposing specific criteria under which a vote may be extended. --WikiTiki8914:53, 29 October 2015 (UTC)[reply]
Oppose. I think votes should generally be closed after the standard duration has elapsed, but I can't support an absolute ban on the extension of votes. I think there are cases in which extending votes may be warranted. But we ought to establish guidelines on the circumstances in which extending a vote is appropriate and how the process should be carried out. -Cloudcuckoolander (talk) 01:55, 31 October 2015 (UTC)[reply]
Abstain Although, as noted above, I think there are cases where this can be useful, I wouldn't want to stand in the way of a simple solution to a simple problem, and wouldn't have any objection to following the rule myself. DAVilla00:02, 21 August 2015 (UTC)[reply]
And what is the simple problem? Selection bias? Selection bias has much better, also simple, solution: "new votes added in the last extension period => extend further by one month". --Dan Polansky (talk) 19:50, 21 August 2015 (UTC)[reply]
I'm not saying it's happened yet to this extent, but potentially there could be votes that seem to go on forever, to the point where people who are for go against just to make it die already. DAVilla06:01, 22 August 2015 (UTC)[reply]
This could be prevented by setting one year to be the maximum duration. And it could not go forever, since there is a relatively small number of editors capable of adding new votes. --Dan Polansky (talk) 08:43, 22 August 2015 (UTC)[reply]
Abstain In general terms, I support. However, on a tactical basis I'm prepared to allow brief exceptions. (Example: vote is scheduled to end December 24. Many people are away/off-wiki around then. Wouldn't have a problem allowing a vote to be extended to after the holidays.) StevenJ81 (talk) 13:07, 14 October 2015 (UTC)[reply]
Ineligibility in an abstain vote is unimportant. In any case, every civil comment that is to the point is welcome, even from non-eligible editors. Thank you, StevenJ81, for your comment. --Dan Polansky (talk) 15:32, 25 October 2015 (UTC)[reply]
Abstain We shouldn't be making important decisions by means of such a blunt instrument as a vote in the first place. —Aɴɢʀ (talk) 14:33, 25 October 2015 (UTC)[reply]
@User:Angr: This page makes it possible for you to use an instrument that you consider less "blunt", such as provision of arguments and evidence. It's a pity more people do not use these less blunt instruments on this vote page; I did. Would you like to share with us your arguments and evidence concerning the proposal here? --Dan Polansky (talk) 15:19, 25 October 2015 (UTC)[reply]
I have no opinion on the proposal at hand. Issues should be discussed until a consensus is reached, in which case a vote is superfluous, or until it's clear that no consensus will not be reached, in which case a vote is pointless. —Aɴɢʀ (talk) 15:49, 25 October 2015 (UTC)[reply]
Support. My first choice is the traditional practice of closing votes which show no consensus as "no consensus", and then if need be having more discussion and a better vote (with refined wording, or after working out some of the side issues that blocked the first vote, as happened with the "well-known work rule" votes). What Dan suggests on the talk page (a set of rules for when to extend votes in a way that minimizes selection bias) would be my second choice, although I think it might be good to have some cutoff to keep votes from dragging on indefinitely as several have already shown a tendency to; e.g. "no vote shall last more than 3 months (even if this means it is closed as no consensus)". - -sche (discuss) 17:40, 8 August 2015 (UTC)
Oppose --Dan Polansky (talk) 07:46, 1 August 2015 (UTC)
Abstain Although, as noted above, I think there are cases where this can be useful, I wouldn't want to stand in the way of a simple solution to a simple problem, and wouldn't have any objection to following the rule myself. DAVilla 00:02, 21 August 2015 (UTC)
