Wiktionary talk:Votes/2019-03/Defining a supermajority for passing votes

Also...

• If this passes, could it be added to Wiktionary:Voting policy.
• Could Wiktionary:Voting policy also include a statement under Closing the vote along the lines of, "A vote is closed only by an administrator." as this is not clear from the policy.

Thanks, Stelio (talk) 09:54, 12 March 2019 (UTC).

• I didn't feel the need to make WT:VP a long, bureaucratic slog. It's supposed to be a page about policies for making votes and casting votes, not the minutiae of closing votes.
• This is a good point. It's not a rule, and we've let non-admins close votes before, but it's the usual community practice, so I think I'll slip it into this vote. —Μετάknowledge^{discuss/deeds} 21:46, 12 March 2019 (UTC)

Why should a non-administrator not be allowed to close a vote? I would rather the policy be that the vote can only be closed by a person who was eligible to participate. There are numerous individuals who contribute regularly and well to this project who are not (often by their choice) administrators. Also, it is very easy to change the closure if it was done inappropriately (whether by and admin or anyone else) so there isn't much harm in allowing anyone who wants to close a vote to do so, especially now that there will be a verifiable definition of what the result ought to be. - TheDaveRoss 18:03, 14 March 2019 (UTC)

There isn't much harm in letting non-admins close votes, but there also isn't much advantage to it, and it cuts down on closures that have to be fixed or get questioned. —Μετάknowledge^{discuss/deeds} 21:38, 14 March 2019 (UTC)

If past proposals have been passed with a simple majority (or less than a 2/3 majority), my only concern would be that it would take a 2/3 majority to overturn a smaller majority, which counterintuitively would end up being less consensus-based. But I'm not sure how things have been handled in the past. And either way I'm still inclined to vote for this. פֿינצטערניש (talk) 11:26, 13 March 2019 (UTC)

This is theoretically possible, but such votes were mostly passed a very long time ago. In the unlikely event that this would come up, this vote allows for the community to override it and allow for a consistent cutoff between a past and present consensus. —Μετάknowledge^{discuss/deeds} 18:46, 13 March 2019 (UTC)

@Metaknowledge: Maybe also prohibiting conditional support votes. —Tom 144 (𒄩𒇻𒅗𒀸) 12:12, 13 March 2019 (UTC)

This vote is meant to codify existing practice and introduce one new thing, the specific cutoffs. That is a contentious issue that would best be treated in a separate vote. —Μετάknowledge^{discuss/deeds} 18:46, 13 March 2019 (UTC)

I agree that keeping the two separate is a good idea. But I do like the idea of clarifying that putting your mark in the support section means you support the vote as written and not as amended in commentary, similarly with oppose. - TheDaveRoss 18:03, 14 March 2019 (UTC)

+1: Prohibiting conditional support votes is a separate subject fit for a separate vote. --Dan Polansky (talk) 18:13, 15 March 2019 (UTC)

@Metaknowledge, TheDaveRoss, Dan Polansky: Done. Feel free to edit it. I hope adding examples isn't too informal. – Tom 144 (𒄩𒇻𒅗𒀸) 15:41, 16 March 2019 (UTC)

What does this apply to?

Does this suggested policy apply to polls on discussion pages, including request pages like "Wiktionary:Requests for deletion", which I have been told are technically not votes? — SGconlaw (talk) 11:05, 15 March 2019 (UTC)

Your question seems to be contain its own answer. This vote applies to votes, and polls are not votes. —Μετάknowledge^{discuss/deeds} 14:37, 15 March 2019 (UTC)

Well, I was told that this percentage applied when closing deletion discussions, so if that is indeed what editors feel it would be desirable to add this to the vote, I feel. — SGconlaw (talk) 14:59, 15 March 2019 (UTC)

I don't understand your sentence. In any case, I added a clarification to remove any lingering confusion. —Μετάknowledge^{discuss/deeds} 19:27, 15 March 2019 (UTC)

I mean that I think the proposal should be extended to any situation, including polls and discussions, where it is necessary to assess whether there is a consensus among editors. — SGconlaw (talk) 22:12, 15 March 2019 (UTC)

I see. I would also consider that to merit a separate vote (but I would not create such a vote, as I disagree). —Μετάknowledge^{discuss/deeds} 22:18, 15 March 2019 (UTC)

70 percent for CFI and EL is too high

I deem 70 percent for CFI and EL to be excessively high; 2/3 should suffice. --Dan Polansky (talk) 11:56, 15 March 2019 (UTC)

I'm pretty sure I got that idea from you. I'm of two minds about it; perhaps @Stelio, TheDaveRoss, Tom 144 could weigh in on this specific part? —Μετάknowledge^{discuss/deeds} 14:39, 15 March 2019 (UTC)

I'd go for 2/3 in all cases; simpler and consistent. And I'd go a bit further and say that a failure is a 2/3 majority the other way in all votes, with "no consensus" in between. [0,1/3]=failure, (1/3,2/3)=no consensus, [2/3,1]=pass. Otherwise you could technically stack a vote by phrasing it in an opposite way (it's easier to get a fail than a pass: compare "Change to marketing Wiktionary as a Wikipedia project?" which needs 2/3 majority for a name change, with "Keep Wiktionary as a Wikimedia project?" which requires a 1/2 majority for a name change); although maybe that's not an issue since we word votes in such a way that an active outcome requires a pass, and a failure and a lack of consensus have the same practical status quo outcome. -Stelio (talk) 15:50, 15 March 2019 (UTC)

My position for multiple recent years has been that 2/3 should usually be enough. Wiktionary:Votes/pl-2018-04/Image policy is an EL changing vote closed at 2/3 and I supported that closure in the vote as per diff. --Dan Polansky (talk) 16:28, 15 March 2019 (UTC)

I would be content with >= 2/3s being the measure for all official votes, consistency is nice. - TheDaveRoss 17:31, 15 March 2019 (UTC)

You mean simplicity, not consistency, I guess. Since, is any rule that differentiates inconsistent? Same speed limit in the city as in the countryside? I like the U.S. different requirements for votes that change constitution. --Dan Polansky (talk) 18:11, 15 March 2019 (UTC)

I mean consistency, which is also simplicity in this case. A consistent standard for what a pass/fail/no consensus. Since the stakes for all votes here are pretty low, there is no real harm in just using a single standard. The U.S. Constitution, on the other hand, has some fairly high stakes. - TheDaveRoss 18:22, 15 March 2019 (UTC)

• Nobody seems to love the distinction, so I have removed it from the vote. —Μετάknowledge^{discuss/deeds} 19:27, 15 March 2019 (UTC)

Votes can only be closed by administrators

Nope. --Dan Polansky (talk) 11:57, 15 March 2019 (UTC)

As an explanation: a vote is a quasi-measurement to see whether there is consensus. Administrators are not the sole people able to determine a vote outcome based on the votes cast and based on the previous vote closure practice, which can be examined in the wiki pages and their revision histories. Closing a vote is not a technical operation and does not require any technical authorization. --Dan Polansky (talk) 12:03, 15 March 2019 (UTC)

I raised it only due to a comment on WT:STOCK ("Generally admins close votes") which made me think it's a literally-unwritten policy, and so worth documenting if it is standard practice (I didn't know about it which is why I closed the vote myself). -Stelio (talk) 15:52, 15 March 2019 (UTC)

If this vote is actually about removing ambiguity about when a vote passes then "only administrators can close votes" is utterly pointless. A computer program could close a vote. DTLHS (talk) 16:03, 15 March 2019 (UTC)

True. -Stelio (talk) 16:40, 15 March 2019 (UTC)

I said this above and will repeat it here, totally agree that this limitation is too strict, I would prefer that only those eligible to participate can close a vote. - TheDaveRoss 18:19, 15 March 2019 (UTC)

• Per this feedback, I have removed this statement from the vote. —Μετάknowledge^{discuss/deeds} 19:27, 15 March 2019 (UTC)

I would prefer that admin-only closure remained a part of the vote. DonnanZ (talk) 20:05, 15 March 2019 (UTC)

It is best removed from this vote and enacted via another vote if desired. This vote is about the threshold, and who closes the vote is orthogonal to that. --Dan Polansky (talk) 20:18, 15 March 2019 (UTC)

Alternative proposal

This is an important subject, but very tricky. To me, a vote like this is quite scary. Here is my alternative proposal:

The threshold at which a vote is considered to pass is not set in stone. For the more ordinary votes, when the ratio of supports to the sum of supports and opposes reaches 2/3 or more, this usually indicates a pass. Some votes may require a higher threshold, some a lower threshold. Abstains do not count toward the threshold, but are useful to show participation. For example, in a usual vote, 10 supports, 5 opposes and 13 abstains would be a pass since 10 in 10 + 5 reaches the threshold of 2/3.

This is a weaker proposal than the one made by the vote creator. It leave open the possibility that votes that are like changing a constitution would require a higher threshold but also that votes that are about matters of taste could require a lower threshold. --Dan Polansky (talk) 12:17, 15 March 2019 (UTC)

I don't like this proposal, because it's worded so weakly. This vote is designed to remove debate about how to close a vote. If there is a reason to have a lower or higher threshold, that threshold should be established before the vote begins, else the vote's closure is at the whim of the closer. This vote allows such modification by community consensus only. —Μετάknowledge^{discuss/deeds} 14:42, 15 March 2019 (UTC)

The weak wording is a strength, in my view. It is dangerous to remove all debate from closing votes. On the other hand, it is preferable to provide hinting, and this is what the proposal does. Until now, we only had a changing preexisting practice evidenced in closed votes. To provide an express hint at a customary threshold is a good thing, but I fear going beyond that. Curiously enough, I find the current somewhat chaotic situation reminiscent of the British common law quite tolerable. --Dan Polansky (talk) 16:18, 15 March 2019 (UTC)

I am not a fan of having to debate the results of a vote. If the voters are clear on what the rules are then there will be less reason for debate (e.g. is a "weak support" worth less than a "strong oppose"?, no). Ambiguity here is just buying trouble. - TheDaveRoss 17:36, 15 March 2019 (UTC)

Flexibility is also buying escape from trouble. Further reading: W:WP:IAR and W:Wikipedia:Five_pillars#WP:5P5. --Dan Polansky (talk) 18:07, 15 March 2019 (UTC)

Yeah, those policies are lip service and nobody really thinks that these projects would improve if anyone who didn't agree with a policy unilaterally decided not to follow said policy. In fact there are people who do that all of the time here, and we block them constantly. - TheDaveRoss 18:17, 15 March 2019 (UTC)

Not really. Let us recall the hot words: these were brought about by someone's initiative and then supported by an apparent consensus yet never made it to CFI. And the new synonym formatting was used far earlier than it was codified into policy. There are more examples from the English Wiktionary. --Dan Polansky (talk) 19:02, 15 March 2019 (UTC)

What does 2/3 support : opposition mean

It seems very unclear. If I did not know what was intended, I would not understand. Like, support : opposition is what I would read as the number of supports to number of opposes, and that is 2:1 when it is to correspond to the intended 2/3 threshold. --Dan Polansky (talk) 20:20, 15 March 2019 (UTC)

I'm open to suggestions for rephrasing it. Would "2/3 support/(support+opposition)" be clearer? —Μετάknowledge^{discuss/deeds} 22:10, 15 March 2019 (UTC)

I think that would be clearer. I agree that the current phrasing is confusing. —Granger (talk · contribs) 01:06, 16 March 2019 (UTC)

That would be clearer, and I don't see any problems with it right now. The wording I use in my proposal is this: "when the ratio of supports to the sum of supports and opposes reaches 2/3 or more, this usually indicates a pass"; you would drop "usually" since you want to have it inflexible. --Dan Polansky (talk) 06:25, 16 March 2019 (UTC)

Vote-specific overrides

The vote page says "Any future vote can override these rules by specific community consensus concerning that vote", as an administrative note. I am not sure I understand what it means, and I am not sure in what sense this is an administrative note. Does it mean that in any particular vote the voters can decide that they are going to apply a different threshold for that one vote? If so, it needs to be made part of the policy, not an "administrative note", whatever that means. --Dan Polansky (talk) 06:34, 16 March 2019 (UTC)

Flexibility is required

In #Alternative proposal, I made a proposal that includes flexibility. Let me argue in favor of at least some flexibility in a separate section.

Voting in general is vulnerable. The voter eligibility requirements can be worked around: an army of attackers can start editing by making trivial yet non-harmful edits, reach the required edit count, and then start impacting votes. This threat in my view requires a certain open-endedness and indeterminacy of the vote mechanism. --Dan Polansky (talk) 07:33, 16 March 2019 (UTC)

This has never happened since our current eligibility requirements were put into place. There are many flaws in our democratic system, but you're choosing to focus on one that is extremely unlikely ever to occur. —Μετάknowledge^{discuss/deeds} 15:51, 16 March 2019 (UTC)

The fact that a security hole has not been exploited does not mean it is not there, and should be disregarded. Should such an event happen, my guess is that established editors of the English Wiktionary would treat the voting system as if it were flexible, regardless of what is codified. --Dan Polansky (talk) 20:05, 16 March 2019 (UTC)

I do think it's only a matter of time before a vote here is canvassed the way votes on Wikipedia often are (in ways sometimes clearly identifiable—e.g. when we/WP sees the posts on reddit or twitter that sent voters over—and sometimes not). While I'll mind WP:BEANS, I can think of things outside groups might want to push; a few votes e.g. on unifying Serbo-Croatian seem like they already attracted some outsiders/meatpuppets. We could make it more difficult by increasing the threshold of eligibility, for example to the "500 edits, existed for 30 days" that Wikipedia uses for some canvas-prone topics, but even that can be circumvented with determination. Perhaps our best defense is our orneriness: if users were trying to push through something in bad faith, we might block them (or at least say "this vote was clearly canvassed, it's not valid"), without all the due process Wikipedia can get tangled up in; I don't think this vote prevents that (since "this vote is invalid because it seems to have been canvassed" seems somewhat orthogonal to "the ratio of supports to opposes meets the threshold for classification as X"). Then again, bad/good faith can be subjective, and one of the users in this very thread has protested blocks admins have given trollish users, so maybe our theoretical 'safeguards' would fail. "Civil" trolls do have a track record of taking over a lot of sites (unless they're proactively banned) and theoretical safeguards against bad actors have a track record of failing in the real world recently (though I won't name examples as that'd get political). - -sche (discuss) 19:10, 21 March 2019 (UTC)

Wikipedia (and real-world) elections are bigger and therefore have fundamentally different problems than we do. Voting structures do not scale up neatly, but exhibit different behaviours at different scales. At our current scale, basic eligibility requirements can prevent the Serbo-Croatian mess from recurring. The idea of someone successfully canvassing on Reddit or Twitter is absurd unless you have reason to believe that our editor base will suddenly grow at a rate much higher than what the rate is now. —Μετάknowledge^{discuss/deeds} 19:48, 21 March 2019 (UTC)

Re: "At our current scale, basic eligibility requirements can prevent the Serbo-Croatian mess from recurring": Not really; it's just that Serbo-Croatian editors were not smart or persistent enough to work around the eligibility requirements in a very straightforward way. All they need to do it make 50 edits in the mainspace without getting into trouble, and there is so much entirely uncontroversial work to be done in the mainspace that this is rather easy to do. Very-low-frequency very-high-impact events are not absurd to comtemplate at all; ask various failed insurance companies to tell you more, or ask the nuclear power industry. Induction is a trap. --Dan Polansky (talk) 10:47, 22 March 2019 (UTC)

You ignored my entire deductive point, concerning scale. Serbo-Croatian bad actors aren't stupid so much as relatively few in number, making the fraction of them that are committed even fewer in number. —Μετάknowledge^{discuss/deeds} 14:37, 22 March 2019 (UTC)

My point is that if those Serbo-Croatian editors who voted oppose in Wiktionary:Votes/pl-2009-06/Unified Serbo-Croatian want to become eligible for voting, they practically can; and therefore, the "basic eligibility requirements" are not guaranteed to prevent anything. That has nothing to do with scale. Admittedly, they were not bothered so far. All they need to do is become bothered. And separately from that, there is my point with low-frequency high-impact risks. One should not dismiss risks while claiming to know things one cannot possibly know well enough. --Dan Polansky (talk) 15:07, 22 March 2019 (UTC)

Anyway, even though I agree "bad-acting" will eventually happen again, I'm not persuaded of the argument that this proposal makes it easier. Even now, if enough bad actors register and edit enough to meet eligibility and make the final tally on a vote be, say, 22 supporters to 6 opponents, I don't expect anyone would close it as having too small a ratio of support to support+opposition, which is what this vote sets. If we would throw it out as invalid or decline to implement it despite it meeting the letter of the "law" for passage, I think we could still do that. - -sche (discuss) 20:07, 23 March 2019 (UTC)