Wikipedia:Proposal to revise CSD R3
This is a failed proposal. Consensus for its implementation was not established within a reasonable period of time. If you want to revive discussion, please use the talk page or initiate a thread at the village pump. |
Should speedy deletion criterion R3 be revised to better address redirects in a language other than English? — Scott • talk 16:46, 9 July 2024 (UTC)
This is obviously a failed proposal, so marking it as such. I had no expectation of it succeeding while writing it, and did so mainly to highlight how a micro-segment of our existing policies is incoherent and unfit for purpose. The responses it's attracted over the last two weeks also provide an illustration of how there's an unspoken agreement among many contributors that redirects exist in a parallel universe where the project's core policy on sourcing does not apply. Until there is a broader understanding of the act of creating a redirect, progress is not possible. — Scott • talk 19:30, 24 July 2024 (UTC)
Synopsis
[edit]Since 2011, the list of possible reasons for deletion given by our guideline on redirects has said:
If the redirect is a novel or very obscure synonym for an article name that is not mentioned in the target, it is unlikely to be useful. In particular, redirects in a language other than English to a page whose subject is unrelated to that language (or a culture that speaks that language) should generally not be created.
CSD R3, governing unlikely and malformed redirects, is nearly 20 years old, and has grown by accretion over that period. Despite this extended opportunity to provide guidance, it remains poorly-defined, and its unresolved ambiguities and gaps have lead to a large amount of wasted editing time. In particular, it inadequately addresses this issue of redirects in languages other than English, which typically exist for multiple years in total obscurity before being uncontroversially deleted at RfD.
This document presents the history of CSD R3, accompanied by case histories of how such redirects have been addressed by community practice in deletion discussions. It proposes revisions to the criterion for improved clarity, including the extraction of conditions relating to non-English redirects into a new, refined criterion requiring the presence of the redirected term in the target article.
Note: A similar proposal was briefly discussed in March 2024 but did not attract much participation. This document goes into much greater detail and has been tagged to gain a broader range of community feedback.
Criterion R3
[edit]Current text
[edit]R3. Recently created, implausible typos
This applies to recently created[1] redirects from implausible typos or misnomers. However, redirects from common misspellings or misnomers are generally useful, as are some redirects in other languages. This criterion does not apply to redirects created as a result of a page move,[2] unless nothing was at the title until recently. It also does not apply to articles and stubs that have been converted into redirects, including redirects created by merges,[3] or to redirects ending with "(disambiguation)" that point to a disambiguation page.
- 1. The definition of recent is intentionally flexible since some pages may receive more notice than others. Pages older than about 3–6 weeks are unlikely to be considered recently created; pages older than about 3–4 months almost never are. Higher-profile pages are considered recently created for shorter periods than those with a lower profile.
- 2. Page moves are excluded because of a history of improper deletions of these redirects. A move creates a redirect to ensure that any external links that point to Wikipedia remain valid; should such links exist, deleting these redirects will break them. Such redirects must be discussed at Wikipedia:Redirects for discussion before deletion. However, redirects that were obviously made in error can be deleted as G6, technical deletions.
- 3. See Wikipedia:Merge and delete for an explanation as to why redirects created by merges cannot be deleted in most cases.
History
[edit]Revision history of CSD R3
| ||||||||||||||||||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Proposal of revised R3 and new R5
[edit]R3. Recently created, implausible typos
This applies to redirects from implausible typos or misnomers which are a maximum of 2 months old, as opposed to redirects from common misspellings or misnomers, which are are generally useful. It does not apply to redirects which:
- were created by moving pages that are more than 2 months old;
- used to be articles;
- were created by merges;[1]
- have names ending with "(disambiguation)" and point to a disambiguation page; or
- are in other languages (see criterion R5).
- 1. See Wikipedia:Merge and delete for an explanation as to why redirects created by merges cannot be deleted in most cases.
R5. Redirects in languages unrelated to target article
This applies to redirects from languages other than English to articles which do not mention the redirected term. Redirects from alternative languages are useful when the target article demonstrates a connection to the language of the redirect, or culture associated with that language. Before deleting, consider whether the needs of readers would be better met by including the redirected term in the article, or converting the redirect to a disambiguation page if there are sufficient search results for it to merit one.
This criterion does not apply to redirects which:
- used to be articles;
- were created by merges; or
- are to templates.
Breakdown of proposed changes
[edit]Criterion | Change | Reason |
---|---|---|
R3 | Formalize definition of "recently created" as "2 months" and remove first footnote | Eliminate disagreements over definition; value is midway between vague ages given in current footnote, which is no longer needed |
R3 | Remove "and stubs" | Redundant, stubs are articles |
R3 | Rephrase page move condition and explicitly define "recently" as 2 months; remove second footnote | Same reason as first change; second footnote, added in 2012, is outdated and muddles matters by adding a dependency on criterion G6 |
R3 | Remove mention of other languages | Move to new criterion |
R5 | Define requirement for articles to demonstrate connection to redirected term | Reduces available options to simple choice between removing unmentioned non-English redirects, including the term in the target article, or creating a disambiguation page |
R5 | Suggest comparing value of deleting the redirect vs adding the term to the target article or creating a disambiguation page | Reminder that the options are useful in different ways and subject to editorial judgement. Deletion is an appropriate action for editors without a strong working knowledge of the language in question, and harmless (see #Deleting such redirects is harmless, below) |
R5 | Don't exempt page moves from criterion | There is precedent of uncontroversial deletion, after a significant period, of redirects created by moving pages away from unrelated titles in other languages (example deleted in 2022 after 10 years) |
R5 | Exempt templates from criterion | Template redirects will probably require special discussion (recent examples of template RfDs: deleted, not deleted) |
Rationale
[edit]Uncontroversial deletion of old redirects in other languages is accepted practice
[edit]The following is a non-exhaustive list of examples of redirects in other languages which were deleted uncontroversially after a long time between 2022 and the present date. It clearly illustrates the extended periods which such redirects typically survive for, with an average age of 11.68 years. Several RfDs in that period also took place for non-English redirects younger than a year.
Title with link to RfD discussion | Deleted in | Created in | Age in years |
---|---|---|---|
Foreign Semendyayev redirects | 2024 | 2021 | 3 |
جمادى الآخر أو جمادى الثان | 2024 | 2006 | 18 |
Tara (rijeka) | 2024 | 2016 | 8 |
امضارمز | 2023 | 2009 | 14 |
Mononoke hima | 2023 | 2006 | 17 |
إنفنيتي ورد | 2023 | 2011 | 12 |
PartiaTë Gjelbërit e Shqipërisë | 2023 | 2009 | 14 |
Attha | 2023 | 2006 | 17 |
Kurze Oktave | 2023 | 2007 | 16 |
Schachmatt | 2023 | 2009 | 14 |
Aryaee | 2023 | 2007 | 16 |
تراجم شعراء السودان | 2022 | 2014 | 8 |
Hıristiyanlık | 2022 | 2011 | 11 |
Ὀφίουσα | 2022 | 2016 | 6 |
Slovénie | 2022 | 2008 | 14 |
Virginie-Occidentale | 2022 | 2009 | 13 |
Mazoki Lietuva | 2022 | 2016 | 6 |
Sistema solar | 2022 | 2012 | 10 |
3 Croatian phrases about cities | 2022 | 2015 | 7 |
Bogengrad | 2022 | 2015 | 7 |
"Chess" in 27 languages | 2022 | 2009 | 13 |
Mathematical terms in 6 languages | 2022 | 2009 | 13 |
Inspection of older RfD archives will reveal an extensive amount of historic precedent for similar deletions. One of the items featured in the examples of common RfD outcomes is Страдание, which was created in 2009 and deleted in 2014.
The lack of a formal definition of relevance results in original research
[edit]Rationales for both deletion and retention frequently involve comments along the line of "As far as I can tell...". As Wikipedia:Redirects in languages other than English wisely notes, "The only language we can rely on our editors speaking is English. Often it requires a strong working knowledge of a language to evaluate and understand redirects".
Where keep outcomes have occurred, they have often resulted from flimsy speculative or implausible (e.g. 1, 2, 3) arguments, which this proposed criterion eliminates in favor of an unambiguous requirement for inclusion in the target.
The weak formulation of R3 leads to inconsistent outcomes
[edit]Here are some recent examples of how the existing R3 criterion's failure to adequately address foreign-language redirects has resulted in unpredictability and excessive use of editors' valuable time:
Year | RfD | Result |
---|---|---|
2023 | Kamala Harris's name in three varieties of Chinese | No consensus. The second RfD for them. The first, in 2021, had resulted in their being kept despite there being no use of the names in the article. The 2023 RfD had several keep votes which were conditional on the names being added to the article, which did not happen |
2023 | Five Chinese transliterations of the names of Korean people | No consensus. Had been relisted multiple times over two months. The latest in a series of RfDs, following a delete result in 2011 and keep result in 2018 |
2023 | Four tremendously long transliterated names of a Russian publisher | No consensus after a month. Received a single delete vote and a single highly speculative keep vote which couldn't even identify the languages involved |
2022 | A German name for Greece | Retargeted to a section of Name of Greece for being mentioned there. By contrast, a simultaneous nomination of several French names for Greece resulted in no consensus due to poor attendance and speculative keep arguments |
2022 | Four names for Poland in other languages | Two retargeted to Names of Poland, one to an article on a Scandinavian dance where it was mentioned as a plural form, and one, in Latvian, allowed to remain pointing to Poland despite not featuring in that article. The latter decision was made on the basis of a comment mentioning the Polish–Lithuanian Commonwealth having existed, even though that article also does not contain the Latvian name for Poland |
Deleting such redirects is harmless
[edit]A reader being routed to an article which doesn't contain the word that they were looking for is a recipe for astonishment. Deleting a redirect because its title is not present at the target eliminates that risk and provides an avenue directly to potentially useful search results. Visitors from other sites landing at the former location of the redirect receive the unambiguous message "Wikipedia does not have an article with this exact name. Please search for The term in question in Wikipedia to check for alternative titles or spellings."
As a speedy deletion, it would also present no prejudice against the later creation of either a new redirect targeted to an appropriate article containing the term, the same article again if the term had been added to it in the meantime, or a disambiguation page.
Compliance with guidelines for new CSDs
[edit]The four guidelines for new CSDs are:
Guideline | Compliance of this proposal |
---|---|
Objective: Most reasonable people should be able to agree whether a page meets the criterion. Often this requires making the criterion very specific. | This proposal defines simple and unambiguous tests for whether certain redirects qualify for speedy deletion. |
Uncontestable: It must be the case that almost all pages that could be deleted using the rule, should be deleted, according to consensus. CSD criteria should cover only situations where there is a strong precedent for deletion.... | Extensive precedent for uncontroversial deletion has been demonstrated and the new criterion is worded to precisely target such instances. |
Frequent: Speedy deletion is intended primarily as a means of reducing load on other deletion methods such as Wikipedia:Articles for deletion and Wikipedia:Proposed deletion.... | The examples given above illustrate the frequent occurrence of redirects in languages other than English, which require a disproportionate degree of effort to resolve at RfD under the current R3 criterion. |
Nonredundant: If the deletion can be accomplished using a reasonable interpretation of an existing rule, just use that. If this application of that rule is contested, consider discussing and/or clarifying it. New rules should be proposed only to cover situations that cannot be speedily deleted otherwise. | The current R3 criterion has been demonstrated to be insufficiently precise, and this proposal makes the necessary clarification by extracting elements into a new criterion. |
Discussion
[edit]Support
[edit]- Support, as the author of this proposal. — Scott • talk 16:46, 9 July 2024 (UTC)
- Support although this was discussed at Wikipedia talk:Criteria for speedy deletion/Archive 87#R3: redirects from foreign this proposal contains more detail so its worth another consideration. Crouch, Swale (talk) 17:27, 9 July 2024 (UTC)
- Support, although this would've benefited from more discussion before going straight to RfC. Queen of Hearts talk 15:34, 11 July 2024 (UTC)
Oppose
[edit]- Oppose as written and send back to the drawing board (i.e. we should discuss the wording before !voting). On R3, I would replace "used to be articles" and "were created by merges" with "contain article revisions in their history". Note that "used to be articles" is ambiguous since it could be interpreted to include the case where the page was moved (which would conflict with the first exception since it doesn't have a 2-month grace period). On R5, I'm afraid "do not mention the redirected term" is overly broad. Maybe this is a valid reason to argue for deletion at RfD, but CSDs need to be uncontestable, and the examples in "The weak formulation of R3 leads to inconsistent outcomes" show that it is not. That is, R5 would cover situations where an RfD failed to find consensus, and adding a new CSD to address the "existing R3 criterion's failure to adequately address foreign-language redirects" would be the tail wagging the dog here. -- King of ♥ ♦ ♣ ♠ 18:35, 9 July 2024 (UTC)
- Oppose. This imposes arbitrary hard limits on things that need discussion - the inconsistent outcomes demonstrate this perfectly: some redirects get deleted, some get kept, some get retargetted. RfD is not overloaded, and redirects to foreign languages make up only a small proportion of redirects discussed there. In short I see no benefit and a lot of potential harm from this proposal. Thryduulf (talk) 18:38, 9 July 2024 (UTC)
- To expand on this slightly, this fails WP:NEWCSD points 2 and 3. It fails point 2 (uncontestable) because it would cover redirects that have been kept or retargetted (per the proposer's own evidence). It fails point 3 (frequent) - there is no evidence that these take up a disproportionate amount of time at RfD with most days not seeing any such redirects nominated (the number handled by existing criteria is irrelevant because an expanded criterion would not change that). Thryduulf (talk) 15:51, 11 July 2024 (UTC)
- Oppose. I agree with King of Hearts that the changes to R3 need more workshopping for improvement and clarity. As for R5, I'd rather scrap the criterion entirely than make it a new and less bounded CSD. Eluchil404 (talk) 03:48, 10 July 2024 (UTC)
- Oppose any CSD including language like "target page does not mention redirect", as that can be easily changed in either direction by editing the target page. Whether the target page should mention the redirect (and whether and how precisely it should) is a totally different question better suited for discussion than for speedy deletion. —Kusma (talk) 13:33, 10 July 2024 (UTC)
- Oppose Prior discussion on A10, which also uses the recent-creation rule, agreed that there should not be a hard cutoff. I don't have a problem with R3's foreign-language clause, but it should be clarified that it should also be plausible in the foreign language — an example would be redirecting Le Hexagon (instead of the correct L'Hexagone) to Metropolitan France — perhaps by rewriting it in terms of a language being obviously impertinent to the topic, such as Мексиканцы (in Russian) to Mexicans. I also agree that there should have been prior negotiations over the proposal. –LaundryPizza03 (dc̄) 12:35, 11 July 2024 (UTC)
- Just a quick response to say that proposals don't need "prior negotiations". The community is literally discussing the proposal right here in these comments, which may lead to modifications and resubmission for consideration, or total abandonment, as consensus sees fit. — Scott • talk 12:48, 11 July 2024 (UTC)
- Proposals don't need prior discussion, but they very often benefit from it. Thryduulf (talk) 13:08, 11 July 2024 (UTC)
- Just a quick response to say that proposals don't need "prior negotiations". The community is literally discussing the proposal right here in these comments, which may lead to modifications and resubmission for consideration, or total abandonment, as consensus sees fit. — Scott • talk 12:48, 11 July 2024 (UTC)
- Oppose. R5 as written will lead to the deletion of valid redirects in several cases. Those include 1) pages that include a romanization of a redirect but not the redirect itself (e.g. すばひび) 2a) pages where a term is mentioned and an unmentioned abbreviation is redirected to them (e.g. СФРЈ) 2b) pages where a modified form of an abbreviation is mentioned (e.g. Р.С.Ф.С.Р) 3) pages where the redirect is mentioned in an included file (e.g. Le Fruit de la Grisaia). In all of these cases, the Ctrl + F test fails but the redirect should be retained anyway. I'm also opposed to adding a hard limit to R3's recently created per LaundryPizza. Nickps (talk) 12:27, 18 July 2024 (UTC)
- Oppose R5. If we ask whether a redirect is related to a specific language, it's often easy to answer with a conclusive "yes", but it's often hard to answer with a conclusive "no". See LaundryPizza's comment about Le Hexagon, for example. While these redirects are frequently deleted, we need to continue sending them to RFD because they frequently need further examination and frequently aren't bright-line cases. Nyttend (talk) 22:30, 18 July 2024 (UTC)
- Oppose R5 but Support R3 reform The point of a speedy deletion criteria should be that it can be applied without controversy and nearly mechanically. R5 leaves too much vagueness and discretion, and the problem is not so frequent as to require a new CSD. But I always felt that R3 was a poorly defined criteria and am happy to see some more thought get put into it. Side note: if you're putting more than one question up for vote, it's best practice to have separate sections for each question, cus otherwise you get folks like me who like part of your idea but not all of it :) CaptainEek Edits Ho Cap'n!⚓ 16:20, 23 July 2024 (UTC)
Comments
[edit]- It seems good to keep the discussion phase for non-English redirects, as there might be a connection to the subject of the article in some way the would-be deleter doesn't realize. This is covered under WP:RFD#DELETE number 8, so that's probably fine as-is without creating R5. Tightening up the definition of "recently created" in R3 might be helpful. I'd err on the low side, and also define it in terms of days so there's less ambiguity in edge cases. Maybe 30 days? At worst a bad redirect will just get deleted after no one comments on the nomination. I would keep the footnotes except for the one giving the vague definition of "recently created". -- Beland (talk) 18:26, 11 July 2024 (UTC)
- I think most people will favor the 30-day cutoff, if we must adopt this. –LaundryPizza03 (dc̄) 05:11, 16 July 2024 (UTC)