Wiktionary talk:Votes/pl-2010-06/Number vs. numeral

So um, clarify. What happens if I vote support, or oppose? -- Prince Kassad 22:11, 12 June 2010 (UTC)Reply

Clearer hopefully. I'm not sure how to identify the set of effected entries without relying on the terms to be voted on. Should 1/1st also be included? Conrad.Irwin 22:45, 12 June 2010 (UTC)Reply

I really think this is putting a lot into a single vote. Trying to decided the number/numeral issue at the same time as the Cardinal X/(plain) X header issue could make interpreting the results very messy and leave only a minority satisfied. --EncycloPetey 22:58, 12 June 2010 (UTC)Reply

Mayhaps this vote should solely be category names - then, if there is still faction over whether to include the Cardinal/Ordinal in the header, we can decide at a later date. Conrad.Irwin 23:18, 12 June 2010 (UTC)Reply

I think, rather than trying to disentangle category and header use, we might phrase the vote to create a decision between Number and Numeral in both, but make it clear that this doesn't require the use of either one, as such. That is, a vote that approves "Number" over "Numeral" does not preclude the use of "Cardinal number" as a header / category name, nor does it preclude the possibility that a later vote would opt for simply "Cardinal". So, the vote selects one word over the other (in headers and categories), in the event that we use that word at all. We might still need a third vote to determine how we categorize numerical/symbolic forms like "8" and "V".

However, I'm not entirely certain that this would be the best option for handling the vote, as it would automatically eliminate at least one possibly workable solution. --EncycloPetey 17:20, 14 June 2010 (UTC)Reply

Are we supposed to vote just for one option, or for as many as we like? Ƿidsiþ 12:16, 14 June 2010 (UTC)Reply

My understanding is that each voter can vote "support" for as many options as he wishes; see also the following section on the voting mechanism. --Dan Polansky 12:22, 14 June 2010 (UTC)Reply

I have placed my understanding to the text of the vote to "Voting mechanism:", to be mercilessly revised. But let us be explicit. --Dan Polansky 12:29, 14 June 2010 (UTC)Reply

This vote seems at this point to be structured as a preference vote with six options. As a result, the option that wins most supporting votes needs to be confirmed in a subsequent approval vote. Having a subsequent approval vote is IMHO a good thing, and a sound way of setting up preference votes. The first vote, a preference one, does the job of finding out what the community thinks; the second vote confirms that a supermajority of editors support the results of the preference vote. --Dan Polansky 12:20, 14 June 2010 (UTC)Reply

What a horrible way of doing things, I suggest either we use a voting procedure that gives us a winning option in one go (Instant-runoff voting perhaps), or we restrict this vote back to just the boolean "cardinal/ordinal number" "cardinal/ordinal numeral" in category names. In the latter case we can have a second vote for part-of-speech headers. Conrad.Irwin 14:02, 14 June 2010 (UTC)Reply

I thought you in particular were going to claim this two-vote procedure was horrible. But I do not expect you to explain why exactly it is horrible. It is a procedure that makes it possible to have multiple-option votes, and abide by the consensus principle at the same time. In a multiple-option vote, the winner can easily have 30% support or even less. --Dan Polansky 14:53, 14 June 2010 (UTC)Reply

It is horrible because we're clearly doing something wrong - many organisations have votes to determine the winner from a set of multiple options, and, as far as I know, none of them need two votes to find one result. If it's actually two results, which would require two votes, then we shouldn't try and confuse everyone by voting on the same two issues twice, instead vote on one issue and then on the other - surely that's a less convoluted method of progression. Conrad.Irwin 15:36, 14 June 2010 (UTC)Reply

You are saing that the procedure is horrible (very bad rather than merely bad or suboptimal) because (a) "we're clearly doing something wrong" (this is IMHO no reason at all but just a rephrasing of the vague assertion that something is horrible), and (b) because the method is convoluted (a rather vague assertion). You are proposing Instant-runoff voting, but only under the head of "perhaps". You are actually rather noncommittal about (c) which method of voting this vote should use, despite that it was you who has changed this vote into a multiple-option one, and (d) what specific, not vague, advantages and disadvantages the various proposed methods have. The procedure that I have proposed has the advantage of being a multiple-option one and supermajority-based at the same time; its disadvantage is that it takes rather long because two votes have to be setup. The disadvantage of Instant-runoff voting is that it is not based on supermajority, and is subject to fallacies of various slicing of options. The two-vote procedure has the further advantage that, after the first vote is completed, the result of the first vote can be interpreted in various ways, and anyone can propose an approval vote based on his own interpretation of the first vote, taking into account side effects of option slicing. Let me point out that it is the need to achieve 70%-majority that sets Wiktionary voting apart from voting in other organizations.

In fact, there is not even a mechanism in place to choose in this talk page the voting method of this vote. The two-vote procedure has the advantage that the second vote in effect confirms in a formally clean way the details of the voting method used by the first vote. At the same time, the two-vote method lets each editor indicate whether he finds the winner of the first vote acceptable. If an editor finds the winner of the first vote unacceptable rather than good enough, he may still vote oppose in the second vote. --Dan Polansky 16:06, 14 June 2010 (UTC)Reply

What happens when the second vote fails? What's the difference between voting for the "another" option, instead of merely against the second vote? It is convoluted because you every user will cast between 0 and 7 votes spread over a total of 9 closely related options (i.e. 0-6/6 of the current options + 0-1/3 on the second vote -assuming that is the normal S/O/A). The vote cast on the second vote page is going to be strongly correlated towards the votes on the first vote page; so it's quite clear that you're asking people to submit duplicate information (even if the options on this page were all distinct, which they're not). For contrast, choosing between "*dinal number" or "*dinal numeral" in categories for this vote would require 0-3/3 (depending on voting method), and then choosing the PoS once we've established the category name (assuming we want to maintain all the options given here, which I'm not sure we do) will require 0-4/3-4 depending on the first outcome. So, the user may cast up-to the same number of votes; but they will have two fewer options to chose from.

I know next to nothing of voting procedures - instant run-of is just one of the few names I know :p - but I've read enough to realise that, like security, trying to invent your own, unanalysed, methods is hubris. It's been proven that there's no perfect way of voting for one choice among more than 2 , so I suggest we go back to having a boolean choice, at least for this vote. Alternatively, of course, we could run this first vote on WT:BP informally, if (as you suggest) it's outcome will be effectively meaningless. Out of interest, where did you pluck the 70% from? We've, very deliberately, never had a policy on pass percentage, and votes have passed with less support in the past. Conrad.Irwin 16:38, 14 June 2010 (UTC)Reply

Re "What happens when the second vote fails?": If the second vote fails, there is no consensus and the result is indeterminacy. That is the way it should be, rather than picking a 25%-winner of a multiple-option vote without being certain that the 25%-winner is good enough for a supermajority of editors.

Re "What's the difference between voting for the "another" option, instead of merely against the second vote?": The difference is that a user can vote for the specific option 2 in the first vote, and oppose the winning option 3 in the second vote. This is not the same as voting for the option "Find another way" in the first vote, and opposing the winning option 3 in the second vote. In the first vote, I vote for the option that I see best fit. In the second vote, I can approve of the winner of the first vote if the winner is good enough. So it makes perfect sense for a voter to vote for, say, the option 2 in the first vote, while supporting the option 3 in the second vote on the ground that the option 3 is the winner of the first vote, and it is good enough.

There is nothing unanalyzed or insecure about the two-vote procedure. Its second part is the familiar, analyzed and well understood approval voting.

The threshold of 70% has been repeatedly used in past votes. You say that "votes have passed with less support in the past", yet the only such vote that I know of is the poorly closed vote on voting eligibility. I am not saying that the threshold of 70% is in a written policy, nor am I saying that there is a formal documentation of consensus in its favor. I am merely saying that this threshold has actually been used in actual votes. My understanding is that there is some discretionary flexibility regarding this threshold given to admins who are closing a vote.

This vote does not need to be run in BP informally only because of your personal dislike of voting. --Dan Polansky 17:05, 14 June 2010 (UTC)Reply

On another note, if a multiple-option vote turns out to give a supermajority (70%, 66%, whatever) support to one of the options, the second vote is unnecessary. --Dan Polansky 17:09, 14 June 2010 (UTC)Reply

Since this vote never opened, I've taken a hacksaw to it and revamped it. The problem Dan mentions remains the same: do we have support for each option or both support and oppose? If Cardinal numeral/Ordinal numeral passes, does that mean that by extension, all the others become invalid? Tough one, I await your thoughts. Mglovesfun (talk) 13:00, 26 July 2010 (UTC)Reply

The vote is now structured as two independent votes, one for categories and one for headers. It's (IMO unlikely but) certainly possible that these votes will go in opposite directions, one deciding on "number" and the other on "numeral". Nothing inherently wrong with that, but some voters might consider that the worst scenario of all, though they don't care which option is chosen: that is, some might not prefer "number" or "numeral" as long as the same word is used for both categories and headers. The votes should IMO be tied together somehow.​—msh210℠ (talk) 15:56, 26 July 2010 (UTC)Reply

Msh, I was going to comment on exactly that - I don't have an opinion regarding vote 2 (I'm undecided about vote 1 currently) other than if vote 1 decides on "Cardinal number" or "Number" then the category should be "Cardinal numbers", similarly if vote 1 decides on "Cardinal numeral" or "Numeral" then the category should be "Cardinal numerals". I think the best way to solve this would be to set the options for vote 2 to be:

• Option 1: Categories should use whichever of "number" and "numeral" wins vote 1
• Option 2: Categories should use whichever of "number" and "numeral" looses vote 1
• Option 3: Categories should use "number" regardless of which wins vote 1
• Option 4: Categories should use "numeral" regardless of which wins vote 1
• Option 5: Oppose all - categories should be named something different to what is proposed here
• Option 6: Abstain

Each would have an example showing what the actual category names would be of course. Thryduulf (talk) 08:42, 27 July 2010 (UTC)Reply

But categories depend on headers, not vice versa. What if someone has a strong opinion on categories but merely wants headers to match whatever choice of categories wins? Perhaps there should be a "whichever header wins" choice for categories (also a "whichever header loses" I suppose, per your suggestion, though I can't think why anyone would choose that), and also a "whichever category wins" choice for headers (and a "loses" one), with the explicit rule that if such a choice wins for both categories and headers then there's no decision and that, moreover, if support for a particular category choice is stronger than support for the winning header choice (which is not the matching choice for headers as the aforementioned category choice), but "whichever header wins" wins for categories, then there's no decision with respect to categories. (And the same for headers, m.m.) (I don't think that that can happen under our current voting procedures, as we require a supermajority, but it'd be a nice rule to have in place just in case. Actually, the vote page should make explicit how votes are going to be counted.)​—msh210℠ (talk) 16:23, 27 July 2010 (UTC)Reply

Here it is not only evil, but an absurdity. It seems that a new low has been reached when a dictionary based community needs to vote on the meaning of terminology. It makes absolutely no sense to say that the articles are about numerals unless the emphasis is on something like the relative merits of how they might appear in different fonts. The articles are about the meaning behind the forms and not about the forms themselves, in other words about "numbers." Trying to determine truth with a vote is an abuse of process. Eclecticology 10:56, 31 July 2010 (UTC)Reply

You seem you be very sure of your terminology, but it's at odds with linguistic publications and modern grammars. A "number" is a concept or meaning, but the words that represent them are called "numerals" in the literature. In linguistic and grammaical literature "number" is a property of the grammar (e.g. singular, dual, plural), and hence the distinction. However, because there is a strident division in the community about using this terminology, we must resort to voting. --EncycloPetey 19:52, 31 July 2010 (UTC)Reply

Yes I am sure of my terminology. I agree that numbers are about concepts and meaning, and numerals are about the words themselves; that is exactly why "number" would be the correct term to use. The articles are about concepts and meaning. "Part of Speech" relates to meaning. The indicated grammatical use of "number" is additional to that currently in dispute, so going down that road doesn't help anything. The solution to a strident division is to find an acceptable alternative to both proffered versions. Why not just use "adjective", and get around the problem that way? Eclecticology 07:03, 1 August 2010 (UTC)Reply

While wikipedia articles are centered on meanings, our entries are centered on terms (ie meaning is just one of the several things we connect to a term). PoS is a property of the term, not necessarily the meaning (eg language name terms don't have uniform PoS's across enwikt). "Adjective" has been proposed before, but as these terms don't always function as adjectives in English (let alone other languages), I don't think it has a good shot. --Bequw → τ 14:43, 1 August 2010 (UTC)Reply

Meaning is a combination of all the features that give life to a word or term, including part of speech. The numbers mostly function as adjectives, but if in a given context it acts differently a separate part-of-speech heading would be warranted. Eclecticology 07:21, 4 August 2010 (UTC)Reply

Meanings of words (what you call "meaning behind the form") are covered in the definition lines. How the words themselves are treated as parts of speech is a different and orthogonal category. Different languages and different grammar traditions use different criteria to classify words into parts of speech. What this vote hopes to establish is not to determine the "meaning of terminology" (which we cannot do anyway, as we are descriptive and define terms as they are used), but to build a consensus (by majoritarian voting, if necessary) on what term should be used to uniformly classify numbers/numerals as parts of speech Wiktionary-wide. It is better to enforce one uniform solution rather than having chaos in naming headers and categories. There is certainly a number of advantages to the term numeral, one of which is its preciseness and clarity (compare the 10 meanings of ] to 2 meanings of ]). --Ivan Štambuk 20:47, 31 July 2010 (UTC)Reply

It reminds me of that Winston Churchill quote saying that democracy is "the best of a bad lot", yes voting is evil, but what do we have that's better? Mglovesfun (talk) 20:51, 31 July 2010 (UTC)Reply

Majoritarian voting does not produce consensus. It imposes the tyranny of the majority or of those who are most persistent in pushing their point of view. Eclecticology 07:03, 1 August 2010 (UTC)Reply

This is not a matter of a point of view. Either of the terms number or numeral is perfectly fine and proper. Choosing between them is not a matter of point of view, but of terminological preference. Neither of those terms carries a particular subjective perspective. PoV is an issue with Wikipedia articles which are written in prose and where presentation can make a great deal with respect to certain subjective aspects of the items of discussion. Here that aspect is absent. The thing is, we are pretty avantgarde when it comes to naming parts of speech. For example, we already use headers such as ===Participle=== (completely unnecessary as IMHO those are normal adjectives and nouns built from verbal roots) or ===Determiner=== (which I've heard of on Wiktionary for the first time, and still I'm not sure what it means). Now, according to some recent English grammars prefer the term numeral over the more "bloated" term number. There have already been several discussions over which term should be preferable for standardization, but unfortunately have't been resolved successfully. So far editors followed their own preference which resulted in couple of thousands of entries for some languages having one scheme of naming headers and categorizing entries, and couple of thousands of entries for other languages having other scheme of naming headers and categorizing entries. Unified solution is needed, and (super-)majoritarian voting seems to be the only way to achieve it. --Ivan Štambuk 18:29, 2 August 2010 (UTC)Reply

If either term is fine and proper, then this is a matter of point-of-view, just like the choice of British or American spellings in Wikipedia. I see little benefit to engaging in avantgardism when the average reader is not engaged in the latest debates about linguistic theory. Perhaps you're right about "participle", and "determiner" was not any part of the grammar that I learned in school many years ago.

Perhaps some of us don't attach much importance to standardization, and are quite happy to live with ambiguity for as long as the situation warrants. That different languages should have different schemes for identifying parts of speech is perfectly acceptable. The structure of some languages does not fit into the neat cubicles designed by Indo-European philologists who saw ancient Latin and Greek as the arch-models of linguistic perfection. In a wiki there is no need for a unified solution; respect for the work and understanding of others is far more important. Eclecticology 07:13, 4 August 2010 (UTC)Reply

I completely disagree - standardisation and consistency of the structural elements of our entries (headers, context labels, etc) is one of the most important things we need to do if the project is to useful to the vast majority of readers. Having multiple headers for the same things is a sure-fire way of both providing a source of confusion and looking distinctly unprofessional. This isn't about forcing all languages into the same holes as those that fit for Latin and Greek, it's about proving a means for readers of a dictionary written in English to understand the content. Thryduulf (talk) 18:57, 4 August 2010 (UTC)Reply

Voting is not evil. Unification or standardization of arbitrary differences and disunities in Wiktionary through voting is good, especially when enough time for discussion is given before the vote starts. --Dan Polansky 11:33, 1 September 2010 (UTC)Reply

Have we discussed it enough? It's already past the start date. —Internoob (Disc•Cont) 21:31, 31 July 2010 (UTC)Reply

I don't think there's a clear feeling of what the vote should entail to best produce a meaningful result. I agree with those who think the current vote setup is overly complicated. My feeling is that we should first settle what POS header we want, and forget about the category implications until after that's settled. Otherwise, the arguments will focus overly on categorization, which is really more a tool for facilitating our content rather than content. --EncycloPetey 23:41, 31 July 2010 (UTC)Reply

I don't think this should be resolved by a vote at all. Eclecticology 00:44, 1 August 2010 (UTC)Reply

How do you propose that this should be resolved? --Ivan Štambuk 18:31, 2 August 2010 (UTC)Reply

Mostly by leaving it unresolved. If past editors have chosen one term or the other when the article was started, leave that term in place. Perhaps in the distant future there will be real consensus; I'm prepared to wait that long. Eclecticology 07:24, 4 August 2010 (UTC)Reply

I disagree. One of the qualities a dictionary has to have is consistency. Consistency makes it easier for users to understand and use. We've received complaints in the past from anonymous users about the categories being several times less useful because of their duplication, and while I don't care about the outcome of the vote, I think it should be decided forthwith. —Internoob (Disc•Cont) 17:34, 4 August 2010 (UTC)Reply

IMO the issue I and Thryduulf raised at ] must be resolved before the vote begins. The vote as currently structured requires people to choose among choices not including their own.​—msh210℠ (talk) 18:07, 4 August 2010 (UTC)Reply

"One", "hundred", and "half" are all cardinal num(ber|eral)s in some sense, but in English they don't have the same grammar: one serves as a determiner ("I see one person", "I see one"); hundred requires a determiner, and it plus its determiner serves as a determiner ("I see one hundred people", "I see one hundred"); and half behaves either like hundred ("I see one-half people", "I see one-half") or like all ("I see half of a person", "I see half"), or like a blend of the two ("I see one-half of a person"). Does this vote presuppose that they all get the same POS header? Or are we just voting specifically on the entries for ] and ], leaving room for future discussion about what other entries should match those entries? Or what? —Ruakh_TALK 20:26, 2 August 2010 (UTC)Reply

That's one question to be answered. "One" is a cardinal but it also has additional functions that other cardinals do not have, since it can also function as an article (determiner). This is commonly the case for the word "one" in Romance languages as well. "Half" is a fractional numeral, and fractionals do not behave like cardinals, nor like ordinals. Neither do ordinals and cardinals behave alike. This is true in many languages, and not just in English. In fact, the primary reason I haven't started doing Latin ordinals is that I'm uncertain whether they belong in the POS numeral or simply adjective. For most Romance languages, the ordinals behave more like a special class of adjective than anything else. This is one reason I set up the template {{ordinal}}, so the inflection line could be marked even in situations where the part of speech for an ordinal was not "Numeral".

So, given this variability, do we need multiple part of speech headers for the numerals? Not necessarily, and I favor a single simple header. Determiners, adjectives, pronouns, and other parts of speech all have subgroupings with differing grammatical behavior. Adjectives may be comparable or not, predicable or not. Some pronouns only function as a subject, some only as an object, and some only in other specific situations. Determiners can specify location, number, or neither. Some determiners have plurals, some do not, and some are only plural while some are never plural.

I think therefore that what we are trying to do with this vote is to say: "If a separate part of speech for numerical words is recognized in a language, what then should be the label applied in the header?" The two key issues I see are (a) Number vs. Numeral, (b) should a qualifier of "cardinal", "ordinal", "fractional", "adverbial", "distributive", etc. be included in the header or not. --EncycloPetey 23:06, 2 August 2010 (UTC)Reply

Thanks. I think, based on my own previous thoughts as well as what you say, that cardinal numers and ordinal numers are really completely separate topics, in that, for example, we could well be justified in treating ordinal numers under ===Adjective=== while treating cardinal numers under some sort of dedicated header. In fact, I think that my preference is exactly that; but the current vote setup makes that impossible. —Ruakh_TALK 00:14, 3 August 2010 (UTC)Reply

Possibly, but I think the choice of whether to use "Number/Numeral" or something else will likely vary by language. English ordinals are a borderline case in my estimation, and legitimately could be listed as either "Number/Numeral" or as "Adjective" with an (ordinal) tag. The English ordinals do have some properties not found in any other "adjectives"; more so than in Latin or Spanish. --EncycloPetey 03:37, 3 August 2010 (UTC)Reply

One of the wonders of the English is the flexibility of the grammar. A word that starts as a noun can be used as a verb, and the resulting sentence can be understood even if no dictionary lists it as a verb. Eclecticology 07:35, 4 August 2010 (UTC)Reply

Yes, but I don't think that's relevant to this. The question isn't what parts of speech a word can have, but what parts of speech it does have. —Ruakh_TALK 17:23, 4 August 2010 (UTC)Reply

I made some more changes to this vote. Most importantly, I added the section === Votes dependent on other factors === for those who wish to vote for whichever option wins the other round, or whatever else. This hopefully addresses some of the problems. —Internoob (Disc•Cont) 18:43, 19 August 2010 (UTC)Reply

If vote 1 decides on "numeral" and vote 2 decides on "whichever wins vote 1", then categories will use "numeral". Similarly if vote 1 decides on "whichever wins vote 2" and vote 2 decides on "number" then the headers will use "number".

If vote 1 and vote 2 both decide on whichever wins the other vote, what happens then? If, "numeral" is more popular than "number" (ignoring those voting for whichever wins the other) in both sections then the only sensible course of action (imho) is to use "numeral" for both header and category. If others agree to all this it should be specified explicitly.

If however (again ignoring conditional votes), vote 1 prefers "number" and vote 2 "numeral", what happens? Thryduulf (talk) 13:19, 20 August 2010 (UTC)Reply

• One problem with this vote is that it doesn't seem to be based on any understanding of what options actually have any supporters. For example, is there actually anyone who wants POS headers using "number" and categories using "numeral", or vice versa? If not, then structuring the vote as two semi-interlinked subvotes is needlessly complicated (see Thryduulf's comment), and likely prevents the addition of an option that someone really does support. (For example, I'd be happy with any of these options for cardinal numerals, but I think we should use ===Adjective=== for ordinal ones; but the vote's current structure presupposes that cardinal and ordinal numerals are to be treated in parallel ways.)

  To address this, I think we need to start with what you might call a "brainstorming" phase, where everyone states what their preferences are. We can even structure it a bit like a vote, where everyone's preference is indicated in a one-line list item, if that will make you feel better about it.

  —Ruakh_TALK 14:06, 20 August 2010 (UTC)Reply

  • I second that.​—msh210℠ (talk) 16:52, 20 August 2010 (UTC)Reply
    • I also think so, so I made WT:Beer parlour#Numbers and numerals. —Internoob (Disc•Cont) 02:54, 21 August 2010 (UTC)Reply

The structure that the vote had when I was last editing it was IMHO okay: in this revision.

The options offered in that revision:

• Use "Cardinal number" and "Ordinal number" in headings and in names of categories
• Use "Cardinal numeral" and "Ordinal numeral" in headings and in names of categories
• Use "Numeral" for heading and "Cardinal numeral" and "Ordinal numeral" for categories
• Use "Number" for heading and "Cardinal number" and "Ordinal number" for categories
• Use "Numeral" for heading and "Cardinal number" and "Ordinal number" for categories
• Find another way

I see no need to complicate the thing by separating headings and categories as two independent topics: they are not really wholly independent.--Dan Polansky 09:37, 1 September 2010 (UTC)Reply

I have reverted the vote to a simple one multi-option vote revision, having an agreement from Mglovesfun. I have added an option for adjective-supporters, one that I do not really support right now, so please adjective-supporters edit the option to match your real preference. --Dan Polansky 10:06, 2 September 2010 (UTC)Reply

I think we should simplify this vote even further. We don't need to tackle the entire issue at once; a previous vote has already resolved one aspect of the question (it's Category:French cardinal Xes, not Category:fr:Cardinal Xes) without the world exploding due to all other aspects being left unresolved.

Given that (1) the title of this vote is "Number vs. numeral" and (2) everyone seems to agree that we should either consistently use "number" or consistently use "numeral", the "number" vs. "numeral" question seems like a good place to start. I get the impression that everyone would support at least one of these options:

1. Categories will be named Category:English cardinal numbers, Category:French ordinal numbers, etc.
2. Categories will be named Category:English ordinal numerals, Category:Spanish cardinal numerals, etc.

(Though we should still have an "oppose both" option, in case I'm wrong.)

We'll then have completely resolved the question of category names — which is arguably the most important question, anyway, since it's what makes it possible to navigate among these entries — and, given that most people seem to oppose mixing "numbers" and "numerals", we'll have made some progress toward resolving the question of header names (in that we'll have ruled out some possibilities).

—Ruakh_TALK 15:23, 2 September 2010 (UTC)Reply

That sounds like a good idea: I second. We can discuss headers thereafter.

Note though that category names is arguably not the most important question (though I'll grant that it also arguably is the most important question), because we can have subcats. For example, if ordinals were in English adjectives but in French ordinal numbers, they'd still be relatively findable if the latter were in French adjectives.​—msh210℠ (talk) 15:42, 2 September 2010 (UTC)Reply

I do not quite agree with (2), as the following hyponymy structure seems quite common in English textbooks:

• Numeral
  • Cardinal number
  • Ordinal number

And if I am right in assuming that this option is rather traditional, then this option has the benefit of being accepted by the Anglo-American readers as usual and unsurprising. It was me who has added the option 'Use "Numeral" for heading and "Cardinal number" and "Ordinal number" for categories', currently one of my favorites. Having this option on the list harms not: if it turns out to be unsupported by people, so be it. --Dan Polansky 15:47, 2 September 2010 (UTC)Reply

An addendum: OTOH, voting only on the deepest-level categories as you have proposed seems doable and simple. So while there can be a mixing between "numeral" and "cardinal number", there can be no mixing between "cardinal number" and "ordinal number": it makes no sense to have Category:English cardinal numbers and Category:English ordinal numerals. --Dan Polansky 15:52, 2 September 2010 (UTC)Reply

That's interesting; it hadn't occurred to me that "cardinal numbers" and "ordinal numbers" might both be "numerals". Ah, well. At least we'll have sorted out one level of categories. It's a start. :-P   —Ruakh_TALK 19:52, 5 September 2010 (UTC)Reply

I've edited the vote to do something like what I describe above. Please take a look and confirm that it seems O.K. —Ruakh_TALK 20:11, 5 September 2010 (UTC)Reply

I have responded at #Category-only vote. --Dan Polansky 09:37, 6 September 2010 (UTC)Reply

A response to a post in the previous section #Return to simplicity:

Looks basically okay to me. I guess I liked my version better :p, as it seems better to me to couple the several decisions into one and judge each combination of decisions as a whole, but this revision of the vote is okay, and I will not revert.

I am not sure I fully understand the voting mechanism. ("Approval voting, with the caveat that an option must have a supermajority (something like 70%, though as usual this is left to the closer's discretion). Each voter can support both options, or support exactly one option (implicitly opposing the other), or oppose both options. Abstentions, as usual, do not count in either direction.") If this really were an approval voting, then each of the two options should have a support section and an oppose section. In fact, I would design four subvote headings: (a) Approve option 1: support, oppose; (b) Approve option 2: support, oppose; (c) Prefer option 1 to option 2: support, but no oppose; (d) Prefer option 2 to option 1: support, but no oppose. This would cleanly separate two acts: approval, and preference. I may prefer option 1 to option 2, and yet I may disapprove of both of them. The evaluation of the vote: an option is a candidate winner if it has been approved by a supermajority. If there is only one candidate winner, it is the ultimate winner. If there are two candidate winners, the ultimate winner is the one who has got more preference votes. If no option has been approved, there is no winner and no follow-up renaming of categories can be made. Hopefully, people would realize the names of categories are neither right nor wrong but merely mostly conventional, both options would get approved, and the winner would be determined using the preference votes. --Dan Polansky 09:39, 6 September 2010 (UTC)Reply

In Approval voting, you can approve of any subset of the candidates. Since we only have two candidates, that means four options: approve of both; approve of "number" but not "numeral"; approve of "numeral" but not "number"; approve of neither. Those translate to: {{support}} both; {{support}} "number" but not "numeral"; {{support}} "numeral" but not "number"; {{oppose}}. I don't see the problem.

If you don't want approval voting, and would prefer a preference-based voting system, I'm fine with that in principle; but your specific system seems likely to confuse voters: in my experience voters don't cope well with multiple correlated options.

If we're querying both approval and preference, then the options are these:

• number = numeral > FAIL (support both, no preference)
• number > numeral > FAIL (support both, prefer "number")
• numeral > number > FAIL (support both, prefer "numeral")
• number > FAIL (support only "number")
• numeral > FAIL (support only "numeral")
• FAIL (support neither)

and the question is how to represent that in the clearest way for voters so that they know how to translate their opinions into the appropriate votes.

—Ruakh_TALK 12:03, 6 September 2010 (UTC)Reply

The problem is that approval alone does not capture preferences. If all voters approve of both options, but one half of voters prefers option 1 while the other half prefers option 2, then they should each vote support for only one option, thereby implicitly opposing the other option, or that at least is what you write: that by not supporting, the voter implicitly opposes. But a voter who prefers option 1 to option 2 does not necessarily oppose option 2. An implicit opposition is IMHO a bad thing; either an opposition is explicit, or it should be discarded as opposition.

That is, if 5 voters vote in support for option 1, while 6 voters distinct from the 5 other voters vote in support for option 2, and no voters vote in "oppose both", what happens? Does the option 2 win, as it IMHO should? Or is there no winner, as option 2 has not gained support of a supermajority? --Dan Polansky 12:59, 6 September 2010 (UTC)Reply

Re: your first paragraph: That's more or less correct. To clarify, the emphasis on approval over preferences is not a bug, but a feature: the intent of approval voting is to find the solution that the maximum number of editors is satisfied with, rather than a solution that a minimal majority of editors is maximally happy with. In the abstract, that aligns pretty well with our notion of consensus; but as I said above, if you do not like approval voting, and would prefer a preference-based voting system, I'm fine with that.

Re: your second paragraph: No winner, since 6 people support option 2 and 5 people oppose it.

—Ruakh_TALK 15:01, 6 September 2010 (UTC)Reply

An addendum: the vote Wiktionary:Votes/pl-2009-12/Treatment of toneless pinyin syllables was IMHO a fairly well set up multi-option vote, in spite of some negative comments on the setup. --Dan Polansky 13:05, 6 September 2010 (UTC)Reply

On another note, I admit that I have misunderstood what "approval voting" is, having failed to read W:Approval voting. Thank you for correcting my misunderstanding. I wrote "If this really were an approval voting, then each of the two options should have a support section and an oppose section", which is wrong per W:Approval voting". --Dan Polansky 14:19, 6 September 2010 (UTC)Reply

I am amazed that you consider the toneless-pinyin-syllable vote to have been well set up. First of all, it offered only a limited concept of preference, which you seem to be a big fan of; for each option, the possible votes were only "support", "oppose", and "abstain". There was no way to support multiple options while indicating a greater preference for one over another. Secondly, in theory it might have seemed straightforward, but I think practice proved otherwise:

1. One editor commented that he literally did not know how to vote; another partially voted, and asked what else he had to do to vote.
2. All but two voters either (1) voted in each sub-vote or (2) voted "support" in one sub-vote and ignored all others. To me this implies that, even among those who did vote, there was not a very sophisticated understanding of the voting mechanism: they either treated each sub-vote as completely independent, or they treated all the sub-votes as mutually exclusive. (In point of fact, five of the sub-votes represented mutually exclusive options, and the sixth sub-vote represented an option that was mutually compatible with four of those five; but the relationship between the six sub-votes themselves, as opposed to that between the options they represented, was subtle at best, and depended on how the person closing the vote decided to count votes. I'm still not sure I understand it, and I was that person!)
3. As I recall, it took me upwards of an hour to collate the results of that vote, figure out what the possible outcomes were, type up a full justification of my conclusions — note that normally the justification of a vote outcome takes the form "Passes 8–1–2 (89%)" or something, whereas here it took the form of several paragraphs of ad hoc reasoning — and even then there were some open questions. We got lucky in two respects:
   1. The specific votes cast made it possible to reach a conclusion. In the general case, that might not have been possible.
   2. It wasn't an actual policy vote that everyone would have to obey from then on, but rather, a vote on what BD2412 should do with various existing entries. He was willing to proceed with his edits based on a weakly justified vote result, but I would not have been willing to edit a policy page based on my interpretation of those vote results.

Fortunately, we can expect to be lucky in this vote as well: with only two options, it's unlikely that votes will be so convoluted as to make closure difficult; and Wiktionary:Entry layout explained/POS headers#Debated POS level 3 headers is obviously not as set-in-stone as some of our policies, such that edits to it are not a huge deal as long as the "debated" stays put. But I'm amazed, simply amazed, at the idea of taking that vote as an model for any new vote!

—Ruakh_TALK 15:42, 6 September 2010 (UTC)Reply

(unindent) The Pinyin vote has IMHO been fairly well setup, not perfectly well setup. For me it seems clear that the way you have closed the Pinyin vote was the right one: an option was a candidate winner if it got approved by having a supermajority support among the votes cast in that particular option, and it was the ultimate winner if it gained most support votes among candidate winners. But I admit that it was you and not me who had to carry the burden of closing that vote.

With the current setup, the vote we are currently discussing is fairly likely to result in an impasse, requiring a subsequent vote that confirms the winner. The setup that I have proposed at the top of this thread increases the chance that there will be no impasse. If we want to get the thing decided in one round, the vote should better expressly reflect the distinction between approval (a thing is good enough) and preference. Whether we choose the method I have proposed or another method, we should use some sort of voting that combines approval of an option and preference between options, or else the situation is not much unlike in the vote Wiktionary:Votes/pl-2009-08/Voting eligibility 2. --Dan Polansky 16:25, 6 September 2010 (UTC)Reply

Re: "an option was a candidate winner if it got approved by having a supermajority support among the votes cast in that particular option, and it was the ultimate winner if it gained most support votes among candidate winners": I don't see why. That seems like a way to determine the winner, but nothing in the vote setup indicated that it was the way, and I can think of other ways that make as much a priori sense. In particular, I don't see any obvious reason that an option with more "support" votes, but also more "oppose" votes, should be the winner. For example, if we take our "supermajority" cutoff to be ≥⅔ (just for the sake of argument), your approach would have 11–5 (69%) beat both 30–16 (65%) and 10–0 (100%), which is hardly intuitive.

Also, I guess I just don't see why there will be an impasse. My impression from people's comments is that most people are O.K. with either option; so both options should have a clear supermajority, and the only question is which option will have more supporters (i.e., fewer opponents). Or are you saying that, since approval voting gives no way to indicate a preference if you support both options, people will decide to (pretend to) oppose one option in order to indicate that they prefer the other?

—Ruakh_TALK 17:04, 6 September 2010 (UTC)Reply

P.S. You can stop arguing against approval voting and for preference-based voting. I've already said repeatedly that I'm O.K. with preference-based voting, so you're arguing with the wind. :-P   —Ruakh_TALK 17:07, 6 September 2010 (UTC)Reply

(unindent) My answer to the last question is yes. In detail: Supposing now that most people are okay with both options, the only formal way they have to indicate their preference is to support one option while abstaining in the other option by not giving it a supporting vote (rather than by explicitly voting "abstain" to that option, as there is no such possibility). And, if an abstinence in an option is read as an opposition in that option, this is what is fairly likely to produce an impasse. --Dan Polansky 17:46, 6 September 2010 (UTC)Reply

I guess I just have more faith in our voters than you do. I think that voters are often confused by overly complicated votes, for the simple reason that they don't invest much time in figuring out how the votes are intended to work (and sometimes the people who set up the vote don't seem to have invested as much time as they should have, either), but when they understand the voting system — and I think most of the editors who participate in votes do understand approval voting — I think they generally vote pretty honestly. I haven't seen much evidence of strategic voting. (Outright vote canvassing has happened, of course, but that's another matter altogether.) —Ruakh_TALK 18:21, 6 September 2010 (UTC)Reply

(unindent) I am okay with both options, and tend to prefer one of them, and I really do not know how to vote in the current setup. By your advice, I should vote in support of both options I guess. But then my preference is lost, which I find angering. I find myself in a prisoner's dilemma: the most cooperative thing for me to do would be to vote in support of both options, but I need to defect in order to let my preference be heard. So this is IMHO a poorly setup vote, the way it is right now. --Dan Polansky 18:41, 6 September 2010 (UTC)Reply

I have changed the stated consensus threshold from "70%" to "70% or 66%", increasing the indeterminacy. The indeterminacy is painful, but only fair, it seems to me.

Above on this talk page, I have stated that "The threshold of 70% has been repeatedly used in past votes", but this seems wrong, as I cannot find votes that would demonstrate the "repeatedly" claim. The threshold of "70%" has been mentioned in the vote Wiktionary:Votes/pl-2009-03/Removing vote requirements for policy changes. Various people have voiced support for a lower threshold in votes that do not change the rules for voting.

The vote referred to above refers to W:Wikipedia:Guide to requests for adminship#If you disagree with consensus, which as of now says:

It is the job of bureaucrats to determine consensus when closing a request for adminship. As RfA is not a straightforward majority vote, there is no precise "pass" or "fail" percentage, and the bureaucrat may discount comments which were made in bad faith or are of questionable validity. However, as an approximate guide, you are likely to pass if you achieve at least 75% support. Nominations which receive less than 70% support are unlikely to be successful, except in exceptional circumstances.

--Dan Polansky 14:34, 8 September 2010 (UTC)Reply

I have confirmed that the vote has started today, as planned. While I have some issues with the current setup of the vote, the setup is good enough for the vote to start. Ruakh has IMHO done a great job in simplifying the vote. I would think that people should vote according to their preference, voting support only in that option that they prefer. If it turns out that no option is supported by a supermajority, we can create a subsequent vote that will confirm the plain-majority winner of this vote. But this is only an if: this vote can still generate a clear supermajority winner. --Dan Polansky 17:40, 17 September 2010 (UTC)Reply

Is the "Oppose both" option for people who don't like either "number" or "numeral", or for people who disagree with the vote? — lexicógrafo | háblame — 12:15, 20 September 2010 (UTC)Reply

Either one. This vote will have one of three outcomes: number; numeral; or no decision. "Oppose both" is for people who prefer no decision to either of the others, either because they dislike both "number" and "numeral", or because they dislike this vote and want the decision to be made by a better (future) vote. (Of course, either possibility is flexible; someone might dislike both "number" and "numeral" but still prefer that we make this decision now, or they might dislike this vote but support one of the outcomes enough that they'll set aside their distaste.) —Ruakh_TALK 13:07, 20 September 2010 (UTC)Reply

Actually, looking at it a different way — I suppose it's only for people who disagree with the vote, but one reason to disagree with the vote could be that you don't like either "number" or "numeral". —Ruakh_TALK 14:09, 20 September 2010 (UTC)Reply