Jump to content

Wikipedia:Village pump (proposals)/Archive 172

Add links
From Wikipedia, the free encyclopedia
Village pump (proposals) archive  

This page contains discussions that have been archived from Village pump (proposals). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.

< Older discussions · Archives: A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y, Z, AA, AB, AC, AD, AE, AF, AG, AH, AI, AJ, AK, AL, AM, AN, AO, AP, AQ, AR · 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214

Adding the Edit Request Wizard to various information and policy pages

Some users with a conflict of interest or being paid to edit struggle to properly format edit requests. The WP:Edit Request Wizard handles the formatting automatically, and is supposed to encourage proper wording of requests. Should this tool be mentioned in policy and information pages regarding WP:Conflict of interest and paid editing? Specifically, I propose linking to it in these pages:

Comments also welcome on improvements to the wizard itself.

Sam-2727 (talk) 20:02, 4 September 2020 (UTC)

  • I had no idea this even existed! I'd actually suggest adding it to Talk Pages as well, in some fashion Nosebagbear (talk) 14:53, 5 September 2020 (UTC)
  • As said above, had no idea this existed. Something like this should be publicised more. I support this page being linked to from the aforementioned pages, it would make clearing up formatting much less of a pain. — Preceding unsigned comment added by Berrely (talk • contribs) 18:18, 5 September 2020 (UTC)
  • Re:No idea this existed:That would be because it's new, please see Wikipedia_talk:Teahouse#Edit Request Wizard for more context. I would like to hear from editors capable of reviewing and testing bugs. If independent reviewers say it's ready, I would support too. Usedtobecool ☎️ 20:29, 5 September 2020 (UTC)
  • Forgot to put my support for my own idea. I agree, some testing is needed. Hopefully in the course of this RfC someone will be able to confirm that all is working as expected. Test cases I've performed are here: User:Sam-2727/ERW Test Cases. Sam-2727 (talk) 00:12, 6 September 2020 (UTC)

Migrate college color data to Wikidata

There is a discussion ongoing about migrating some sports team color data from a local Lua module to Wikidata: Module talk:College color#Proposal: Migrate to Wikidata. –IagoQnsi (talk) 07:35, 11 August 2020 (UTC)

  • Yet another case of “Hey, we could use Wikidata to do this thing we already do”. No thanks. Blueboar (talk) 21:50, 11 August 2020 (UTC)
  • Oppose would serve only to make subtle vandalism easier. No benefit. DES (talk)DESiegel Contribs 02:14, 13 August 2020 (UTC)
    • @DESiegel: As I discussed on the module talk page, there are plenty of ways to monitor the data and combat vandalism. Here's a query that shows the most recent changes to color properties for college sports teams. A bot could be easily configured to update a page here on Wikipedia with those query results, and show those colors visually so that incorrect colors could be easily spotted. If vandalism did become rampant, Wikidata has the same protection tools we have on Wikipedia, and iirc, policy there says highly-visible items are eligible for indefinite protection if needed.
    As for the "no benefit" part, it's true that this has limited benefit to English Wikipedia. But it has great potential to benefit other Wikipedias and the rest of the Wikimedia community. Having a freely-available regularly-updated reference-backed database of college sports team colors is of great value to the general public; I can foresee these being used for cool data visualizations. –IagoQnsi (talk) 00:09, 14 August 2020 (UTC)
    If those running Wikidata want to import this data, they are free to do so and other projects are free to use it. At this point, I do not trust sourcing and validation done on Wikidata, and I strongly object to any information in any en.Wikipedia article being automatically imported from Wikidata. No exceptions. I opined in an RFC some years ago that we should make no use at all of Wikidata in articles, and i stand by that view today. I object to any and every expansion of use of Wikidata here. DES (talk)DESiegel Contribs 00:15, 14 August 2020 (UTC)
  • Support The benefits are many and I think IagoQnsi could have more to say. Centralizing the data so it is accessible to all 300+ wikis just makes sense. Only need to change data in a single place instead of 300. Monitor for vandalism in one place. Fix mistakes in one place. etc.. There are tools and methods for vandalism monitoring, that would be part of the deal for acceptance a robust method. As noted by IagoQnsi, there are future applications not currently available that could be built with a central color database, not just college sports teams. -- GreenC 02:54, 13 August 2020 (UTC)
  • Support Module:College color/data is a horrible hack to store data, it should never have been created, and it's a usability nightmare for anyone that wants to change something in it. Wikidata can handle this in a more user-friendly and cross-wiki way, with better monitoring tools via constraint violations and data queries. Thanks. Mike Peel (talk) 19:45, 13 August 2020 (UTC)
    • Question: How would the 794 citation templates in Module:College color/data be stored in Wikidata? --RexxS (talk) 23:11, 13 August 2020 (UTC)
      • @RexxS: Wikidata supports attaching references to statements; see wikidata:Help:Sources. I ran a script against all the reference templates to get the full list of parameters they use, which are as follows: access-date, archive-date, archive-url, date, language, page, publisher, title, url, url-status, website, work, year. All of those have corresponding properties in Wikidata, so they can be imported pretty easily. The properties 'publisher', 'website', and 'work' will have to be mapped to Wikidata items (because the corresponding Wikidata properties takes an item rather than a string). However, I just checked, and there are less than 100 uses of all those properties combined, so they'll be easy to reconcile with OpenRefine. –IagoQnsi (talk) 23:45, 13 August 2020 (UTC)
        @IagoQnsi: Please have a look at MOS:INDENTMIX. You've just made a mess of this thread for anybody using a screen reader. (reply: Fixed! –IagoQnsi (talk) 01:35, 14 August 2020 (UTC))Not quite. --RexxS (talk) 01:50, 14 August 2020 (UTC)
        Yes, I'm sure most of the parameters can be stored on Wikidata, even if you have to create hundreds of new Wikidata entries to accommodate the mismatch of datatypes (and more every time new citations are added). How do you intend to store the type of citation template used, so that the citation can be accurately reconstructed? Given those sort of problems, you need to ask yourself if Wikidata really is the best place to store this kind of data. --RexxS (talk) 00:51, 14 August 2020 (UTC)
        @RexxS: I don't expect I'll have to create many new items, if any at all. As I said, those 794 citations combined have <100 uses of parameters that have to be mapped to Wikidata items. And most/all of those uses won't require new items – most of the references were published by a university, and we already have WD items for every university.
        As for the type of template used, all of these citations use either {{cite web}} or {{cite manual}}. The distinction seems a bit arbitrary (i.e. many of the 'cite web' ones seem like they should be considered manuals), but I could store type of reference (P3865)=user guide (Q1057179) for the 'cite manual' ones if you'd like. But ultimately, the goal here is not to be able to generate the exact same underlying wikitext, but to generate a citation with the same semantic meaning. Really, the outcome is that the citations will be much cleaner – on Wikipedia, you often end up with information rather randomly distributed between 'work' and 'publisher' and 'title', but Wikidata forces you to be more consistent. –IagoQnsi (talk) 01:35, 14 August 2020 (UTC)
        @IagoQnsi: I think you may find that the editors using college colours are likely to be quite interested in recreating the exact underlying wikitext. Have you considered the workflow for an editor adding an entry to the Scribunto module (not simple, but there are plenty of examples to copy and the citation uses a format they will be familiar with), with that for adding a college colour to Wikidata, including all the work in deconstructing a citation into multiple non-obvious reference qualifiers. Worth thinking about. --RexxS (talk) 01:50, 14 August 2020 (UTC)
        @RexxS: It's not that we couldn't recreate them exactly as written; it's that it makes far more sense to clean them up so they appear better than how they're currently written. There's no reason to perfectly preserve messy stuff like |title=Color {{!}} Brand and Messaging {{!}} University of Colorado at Boulder. For example, take this citation: {{cite web |url=https://brand.cornell.edu/design-center/colors/ |title=Colors |publisher=Cornell University Brand Center |accessdate=July 17, 2019}} It's not worth creating a new item for "Cornell University Brand Center", so when I transfer it to Wikidata, I will likely transform it to look like this:
Wikidata example reference
Property
Normal rank no value edit
▼ 1 reference
+ add value
      • This isn't exactly the same, but it's basically a neutral change, if not an improvement. And if I plug that data back into {{citation}}, you get this perfectly cromulent citation: "Colors", Cornell University Brand Center, Cornell University, retrieved July 17, 2019 (for comparison, here's the original citation: "Colors". Cornell University Brand Center. Retrieved July 17, 2019.). –IagoQnsi (talk) 02:29, 14 August 2020 (UTC)
        @IagoQnsi: Please try to observe MOS:INDENTMIX. It's not very kind on screen readers to screw up the indenting as you do.
        If you read WP:CITEVAR, I think you'll find that the changes you have made are far from neutral. Aside from the obvious re-assignment of title and publisher, you've changed the citation style from {{cite web}} to {{citation}}, which should never happen. You also set the accessdate in mdy format. How did you recover that piece of information from your Wikidata record? These are the sort of problems I've been grappling with for the last seven years with Module:WikidataIB and I'd be fascinated to see how you think you've solved them. --RexxS (talk) 03:51, 14 August 2020 (UTC)
  • outdented — GhostInTheMachine talk to me 10:16, 29 August 2020 (UTC)
  • @RexxS: I understand the need to not change citations that are already perfectly fine, but that is not the case here. If you look through the citations in this module, there's little consistency between them. They alternate between quotes and italics; some are randomly in all caps; the name of the publisher is sometimes in the title field, sometimes in the work/website field, and sometimes entirely absent. In general, these all look like the authors pasted in their URL, clicked "autofill" in their citation tool of choice, and called it a day. I would be more than happy to preserve any existing citation style, but there just isn't one here.
  • Separating content from style is really the whole purpose of Wikidata You can display info from Wikidata with any citation style, any date format, even any language you want. If you want to show one of these citations in an article that's using {{citation}}, you can do that; if you want to use {{cite web}} or anything else, you can do that too. The goal of migrating to Wikidata is not to force any particular presentation style; it's simply to separate the data from the presentation. –IagoQnsi (talk) 06:18, 14 August 2020 (UTC)
    @IagoQnsi: I'm afraid you're mistaken about the citations stored in Module:College color/data. Apart from one access-date in ISO format, they are remarkably consistent. There is not one single example of italics hard-coded into the citations. Your complaint about "quotes and italics" is a consequence of MOS:ITAL: "Use italics for the titles of works (such as books, films, television series, named exhibitions, computer games, music albums, and paintings). The titles of articles, chapters, songs, episodes, research papers and other short works instead take double quotation marks. The citation templates are smart enough to use the appropriate formatting to match our Manual of Style. The citation style used is WP:CS1 and you'll run into huge problems if you think you can flip to another style at will. You're going to have to preserve the citations style used however it is stored, so that problem will need to be cracked before Wikidata becomes a viable storage mechanism for citations.
    Separating content from presentation is an artefact of Wikidata, not its purpose (that would be "storing data"). Unfortunately, Wikidata is currently structured to accommodate import of plain data, with little thought given to how it is to exported and re-used. If it is to be used as a database back-end for a Wikipedia article, it needs to be capable of respecting any presentation options that are currently in use in that article. The citation templates can do a lot of that work for you (they can even detect the date format in articles that have set one), but deciding on the style of citation that should be used in a particular article is currently the stumbling block for attempts to store citations in any format other than wiki-text. --RexxS (talk) 12:45, 14 August 2020 (UTC)
    @RexxS: If citation style was stored on Wikidata, it would cause violations of WP:CITEVAR, not prevent them. Here's how that would play out: ClaimX on Wikidata is used in two Wikipedia articles – ArticleA (which uses {{cite web}}) and ArticleB (which uses {{citation}}). Now, suppose User:Example updates ClaimX to use {{cite web}}. In this hypothetical, we are storing style information in Wikidata, which means ClaimX will now appear in ArticleB with {{cite web}}. Oh no, this violates WP:CITEVAR!
    Thus, this is how we do it instead: ClaimX on Wikidata has a citation with no defined style. ArticleA and ArticleB both import ClaimX, but they each render it in a different style. User:Example makes an edit to ClaimX. Because Wikidata does not store citation style information, the citation will continue to appear in {{cite web}} on ArticleA and in {{citation}} on ArticleB. The citations remain consistent, WP:CITEVAR is obeyed, and everyone is happy. –IagoQnsi (talk) 05:31, 18 August 2020 (UTC)
    @IagoQnsi: You think "ArticleA and ArticleB both import ClaimX, but they each render it in a different style." so the citation has to be hand-crafted for each article, thus defeating the object of the Module:College color, which is intended to render the information automatically. Just look at Module:College color/doc #Test table for an obvious example. The table is generated by one line: {{#invoke:{{BASEPAGENAME}}/contrast |testtable |style=background:#FDFDFD }}, but your solution would require 1132 rows to be written out individually so that you can supply the references in the format you want.
    The whole point of Module:College color is that it stores the citation style for each team along with its colour scheme. The advantage of that is that Abilene Christian Wildcats doesn't have to have the same reference style as Youngstown State Penguins. The downside is that if another article about Abilene Christian Wildcats is written, it needs the same citation style as the base article, but that's the route that the writers and maintainers have taken, and I don't see that you should be forcing them into doing a lot more work in each article just to avoid storing the citation style, which your solution can't deliver. --RexxS (talk) 17:17, 18 August 2020 (UTC)
    @RexxS: The citations wouldn't have to be rendered individually; you'd have a template/module which renders all of them, and you just tell that template how to style all of them. Something like: {{#invoke:{{BASEPAGENAME}}/contrast |testtable |style=background:#FDFDFD |citation-style=CS1 |date-format=mdy }} –IagoQnsi (talk) 20:11, 18 August 2020 (UTC)
    @IagoQnsi: So you write another template/module with 1132 rows just to display all of them? We already have a module that does that in one line. Otherwise, how does {{#invoke:{{BASEPAGENAME}}/contrast |testtable |style=background:#FDFDFD '''|citation-style=CS1''' |date-format=mdy }} know what the value of |citation-style= is? You've told me that you're not going to store it on Wikidata with the rest of the data. --RexxS (talk) 02:24, 19 August 2020 (UTC)
    @RexxS: My understanding of your argument is basically that the style of any given citation should never change. However, that's not quite what WP:CITEVAR says. It reads Editors should not attempt to change an article's established citation style (emphasis added). Citation style is set on a per-article basis, not a per-citation basis. Based on that policy, there is no reason that any given citation should have a set style; instead, it should be rendered in the style of the article it appears in. That's what the current policy would indicate, and frankly, what makes the most sense.
    Let me give an example. Suppose that I (a human editor) saw a citation-supported fact in ArticleA, and I decided that that fact (and its citation) would also be useful to include in ArticleB. In the process of copying this information over, I would surely edit the citation to convert it to the citation style of ArticleB, wouldn't I? If the citation from ArticleA is in {{citation}} style, but ArticleB uses {{cite web}}, then I would need to convert that citation to use {{cite web}}, right? If it is acceptable for me to change the style of a citation when moving it to a new article, then why shouldn't it be okay for Wikidata to do this as well? –IagoQnsi (talk) 17:59, 26 August 2020 (UTC)
    @IagoQnsi: You seem to misunderstand how this module is used. Each of the colour entries applies to effectively one article, and the data module stores not only the colour information for that article's team, but the citation style used in that article. That's how it works now. You want to change how it works to fit in with the ability of your scheme to store information, and to throw away the information about the style of citation to use. The editors who created and use this module make use of it in a simple way now; and you are going to ask them to either store locally a separate list of articles with their citation styles, or to insert the citation style manually in each article, for no reason other than your proposed solution can't function in the way that they are working now. That's the tail wagging the dog.
    Let's look at your example. You suppose that the colours for Team A, used in ArticleA might somehow be useful in ArticleB. I see no evidence of that ever being the case. Under the present system, that is still trivial to achieve when ArticleB has the same citation style as ArticleA; you just copy the relevant template over and it generates the citation for you. It would require manual updating of the citation if the styles were different. However, in your scheme, the citation would have to be locally set up in each ArticleA's case to match its existing style (you haven't given any examples of how that might be accomplished), and then set up again locally if you wanted to use it in ArticleB whether the style matched or not. I don't see that as an advantage. There is no demonstrated benefit of decoupling the citation style used in each team's article from the information stored in the module. At present there is effectively a one-to-one correspondence between article and entry in the data module, which allows a simple storage method. You are advocating using a more complex storage system (some info on Wikidata; some info locally) to allow for the possibility that somewhere the editors might want a two-to-one (or more) correspondence between article and colours info, but you haven't demonstrated any need or demand for that. --RexxS (talk) 19:06, 26 August 2020 (UTC)
    @RexxS: This module is not at all used on a single-article basis. Each school's colors are used on probably hundreds of articles each, as we have many many articles covering college sports. To give an example, the colors for Duke Blue Devils are transcluded in Duke University, Duke Blue Devils, Duke Blue Devils men's basketball, 1990–91 Duke Blue Devils men's basketball team, Atlantic Coast Conference, 2018–19 Atlantic Coast Conference men's basketball season, and every variation of those articles (i.e. other years/seasons, other sports). If we were to transfer the data to Wikidata, the colors would also be used by many pages on other editions of Wikipedia, which would all have their own citation styles presumably.
    Setting the style wouldn't be difficult. Presumably, we would default to CS1 style, which is used on an overwhelming majority of college sports articles that I've encountered. But this could be easily overridden either on a per-use basis (e.g. {{College color boxes|citation=cs2}}) or on a per-article basis, using something like {{Use CS2}} (which could work in the same way that templates like {{Use mdy dates}} work currently). Presently, the module's citations are not transcluded onto articles at all, so this is all hypothetical for now. –IagoQnsi (talk) 19:52, 26 August 2020 (UTC)
    @IagoQnsi: At present, any article that wishes to use the citation stored in the module can do so trivially. The citations stored in the module are an almost equal mixture of {{cite web}} and {{cite manual}} with a couple of {{cite book}}.
    You claim "Setting the style wouldn't be difficult ... this could be easily overridden ... using something like {{Use CS2}}" Here's a citation: {{cite manual |url=http://www.thepacwest.com/documents/2015/6/22//pacWest_colorways_spring2015.pdf |title=Pacific West Conference Visual Identity Standards |accessdate=March 23, 2017}}. I'd like you to show me how it isn't difficult to set CS2 for it. Please let me know roughly how long I'll need to wait for your solution. --RexxS (talk) 21:30, 27 August 2020 (UTC)
    @RexxS:If you have that citation stored in Wikidata, it's very easy to spit it out in either style. The parameters are exactly the same between {{cite manual}} and {{citation}}; in fact, they both rely on the same underlying module. Our template would pass all the reference parameters to Module:Citation/CS1, then set the CitationClass to "citation" if the page uses CS2, or to one of the CS1 values (e.g. book, web, news, etc) if the page uses CS1. There would have to be a little bit of logic to decide which CS1 value to use — basically, if the reference includes a type of reference (P3865) value, we'd just use the CitationClass that corresponds to that media type; otherwise, we'd make a best guess based on the included properties, much like how {{citation}} does. –IagoQnsi (talk) 22:45, 27 August 2020 (UTC)
    @IagoQnsi: Okay, show me. I'm really interested in that "little bit of logic to decide which CS1 value to use". Then I'll ask you how you'll deal with chapter and article titles that are defined differently in CS1 and CS2. --RexxS (talk) 23:20, 27 August 2020 (UTC)
    @RexxS: If type of reference (P3865) is set, then we can just have a 1-to-1 array mapping of reference type items to CitationClass values (e.g. map book (Q571) to CitationClass='book', etc). If P3865 isn't set, then we look at what properties are set. A very rudimentary solution would be to use 'book' if volume/edition/page are set, and otherwise use 'web'. That's not a robust solution, sure, but it's already more than enough to handle all of the existing references on this module. I'm not proposing we migrate every citation on Wikipedia to Wikidata; I'm just proposing we migrate the ones in this module. The implementation will get better with time if it's being used and edge cases begin to crop up. –IagoQnsi (talk) 23:56, 27 August 2020 (UTC)
    @IagoQnsi: You don't seem to have shown me anything. And I'm afraid that type of reference (P3865) doesn't mean what you think it does. --RexxS (talk) 00:14, 28 August 2020 (UTC)
  • Oppose, as I will continue to oppose any of the many attempts to cause this project to rely more on Wikidata via the back door. DEsiegel has it right in this instance. However bad things might be here, they are not better there. - Sitush (talk) 19:48, 13 August 2020 (UTC) But, hang on, the discussion seems to be elsewhere? - Sitush (talk) 19:49, 13 August 2020 (UTC)
  • Support I'm usually suspicious of migrating things to WikiData, but this seems low risk in terms of vandalism. Most people will recognize if the value is not a color, and even if subtle, it should be obvious from all the images what the correct colors are. Others note that there are content improvements that can be made if migrated, so seems like a net improvement. — Wug·a·po·des​ 19:57, 13 August 2020 (UTC)
  • Support. Exactly the sort of data Wikidata works best for. – Finnusertop (talk ⋅ contribs) 02:34, 14 August 2020 (UTC)
  • Oppose. Above, IagoOnsi claims "there are plenty of ways to monitor the data and combat vandalism." and adds a query that one can use. Problems; this is an extra step one must take (not integrated with the watchlist), on a different site, and gives poor results. Not only does it not show what was changed or in which diff (only a date), but it includes changes having nothing to do with changing the colour. E.g. there is an entry for 24 May 2020 for the Portland Pilots; but the only change on that date is a bot adding a Chinese label[1]. The query doesn't filter out or indicate bot changes. Apparently the query shows pages where the official colour has been changed at some time in the past, and shows the date of the most recent change to the Wikidata item, no matter what that change was. Which means that contrary to the claim, it is impossible to monitor this with that query, as the most recent entries will not show recent changes to colour, but any recent change whatsoever. I see no reason to change a woking system to one that is harder to monitor. It would be better, if there is a need for the other Wiki-languages, to write a bot that copies changes to the module here, over to Wikidata, keeping Wikidata in sync with us. But giving control of this to Wikidata? No, thanks. Fram (talk) 11:51, 14 August 2020 (UTC)
    • @Fram: Apologies for that issue with the query; I didn't catch that. However, the approach of having a bot update a list with a query would still work as intended. If we had a bot update a locally-stored list with the results of this query every day/hour/whatever, then it would be easy to monitor changes by adding the list page to your watchlist. The list page would only get updated when a color is changed. –IagoQnsi (talk) 04:46, 18 August 2020 (UTC)
  • Oppose per DES. and Fram. Suggest someone from Wikidata make a copy of this material and post it there for use on other wikipedias if they want it. But zero benefit for en-WP in migrating our own hosting of this data, and non-zero risk in terms of reduced editability by local contributors. -- Euryalus (talk) 12:10, 14 August 2020 (UTC)
  • Oppose in practice. It's a noble idea which could work well if Wikidata had as many eyes on it as enwiki does. However, it's so tempting to change one's rival's colour scheme to poop and vomit, and a casual editor who notices may not find their way through the maze of code and data to the source of the problem. mw:Multilingual Templates and Modules may be a good alternative. Certes (talk) 12:40, 14 August 2020 (UTC)
    Why would multilingual templates on MediaWiki.org be less prone to vandalism than statements on Wikidata? * Pppery * it has begun... 13:52, 14 August 2020 (UTC)
    A good question. Perhaps they are harder to find and edit, and would be on the coder's watchlist. If mw: is prone to vandalism then we should leave things as they are. Certes (talk) 13:58, 14 August 2020 (UTC)
    (ec) I guess that, unlike the free for all Wikidata is, the intention is to make those multilingual templates only editable by a group comparable to our template-editors (but obviously not restricted to enwiki editors)? If not, then this is not a solution of course. Fram (talk) 14:00, 14 August 2020 (UTC)
    I'm probably not the best person to answer this because I strongly oppose Multilingual Templates and Modules for reasons unrelated to their proneness to vandalism (which I explained best at mw:Global templates/Discuss/oppose), but I'm not aware of any established plan to restrict who can edit them. Furthermore, non-trivial multilingual templates often depend on tabular data on Commons, which likewise might need to be protected to prevent vandalism. * Pppery * it has begun... 14:39, 14 August 2020 (UTC)
    It appears that that project has shifted to using a tool instead of a bot, which obviates the concern about vandalism to the templates themselves. There is still the, probably more important, concern about vandalism to Commons datasets the template uses, and of course my independent concern about perpetuation of code bloat. * Pppery * it has begun... 16:24, 14 August 2020 (UTC)
  • Support, per GreenC, Mike, and others. Rehman 13:26, 14 August 2020 (UTC)
  • Oppose This seems very much like a solution in search of a problem, since, as DESiegel says, Wikidata could import data from the English Wikipedia module and therefore help other wikis even if this proposal fails. * Pppery * it has begun... 13:52, 14 August 2020 (UTC)
    • @Pppery: It's easy enough to have a bot import the data once, but it's hard to keep the data in sync as time goes on. Any new edits on enwiki need to be re-imported, either manually or by a periodically running bot (but there's no easy way for a typical Wikidata user to be sure that such a bot is still up and running). And if new edits are made on Wikidata, it creates a conflict that can't be resolved by a bot. The overall effect is that anyone using the color data via Wikidata is getting an inferior/untrustworthy version of it. –IagoQnsi (talk) 05:14, 18 August 2020 (UTC)
  • Oppose per Fram, Certes, Euryalus, and Pppery. The whole point of creating this in the first place was to have all the color data collected at one centralized location. If WD wants to make a copy of this to help other wikis, that's fine, but spreading this data out over hundreds of pages is literally what we had before this module was created. Nobody wants to go back to that chaos. Ejgreen77 (talk) 07:37, 16 August 2020 (UTC)
    • @Ejgreen77: This wouldn't create chaos like existed in the pre-module days. The issue we used to have is that the colors were stored in dozens of articles, and had to be manually synced. On Wikidata, just like in the module, they will only be stored in one place: the team's item. All the dozens of articles about, say, Duke Blue Devils teams will pull the school colors from Duke Blue Devils (Q2907160). And it's easy to query Wikidata to show all the colors on one page, such as with this query. –IagoQnsi (talk) 05:01, 18 August 2020 (UTC)
      • Even so, there are still hundreds of entries in the module currently, so all of those entries will each be stored at their own separate location. The whole point of creating this in the first place was to store all the data at one centralized location to make it easy to monitor. Ejgreen77 (talk) 10:42, 26 August 2020 (UTC)
  • Oppose While there may be some valid uses of Wikidata to store information for use on Wikipedia, it is becoming increasingly apparent that it can no longer be trusted to maintain data in a usable structure. This collection of related data is best kept in a single location under enwiki control. --RexxS (talk) 18:42, 16 August 2020 (UTC)
    • @RexxS: Besides our disagreement on the issue of citation style, I haven't seen any discussion about the structure of the data. On the module talk page, I showed how I expected the data to be structured with an example: wikidata:Q2892868#P6364. I welcome criticism, but I haven't seen any yet. Cheers, IagoQnsi (talk) 20:38, 27 August 2020 (UTC)
    • @Mike Christie and Johnbod: You both cited RexxS's comment above as your reason for opposing the proposal. I am curious to know what parts of the Wikidata structure you find to not be usable. Is it the citation style concerns RexxS raised earlier in the discussion, or is it something else? No one else has weighed in on myself and RexxS's citation style thread, so I have no sense of the broader community opinion. Thanks, IagoQnsi (talk) 20:38, 27 August 2020 (UTC)
      • Mike said "per many of the above comments; RexxS sums it up well" & I said "per RexxS and others". I think RexxS's summary is clear enough, & there are many points others have made. Johnbod (talk) 21:16, 27 August 2020 (UTC)
        For me, the key point RexxS made is it is becoming increasingly apparent that it can no longer be trusted to maintain data in a usable structure. Updating short descriptions is a huge use of enwiki resources that could be better directed elsewhere, but it has to be done because vandalism to the corresponding Wikidata descriptions is, practically speaking, invisible to watchers here. That soured me on using Wikidata for anything that could be equally well done here. Unless there are obvious major benefits to the move, I don't want to see enwiki move information to where it cannot be monitored and managed (and protected) locally. Mike Christie (talk - contribs - library) 21:36, 27 August 2020 (UTC)
  • Oppose now Proponents of this idea are free to copy color information to Wikidata and to develop modules for use on other Wikipedias. After six months of successful operation, this proposal could be considered for enwiki. However, until then, having all the information on one page is known to work and is known to resist subtle errors and outright vandalism. At enwiki, anyone can propose a color change but only template editors can edit. At Wikidata, any IP can change colors in any of the 1140 locations. Johnuniq (talk) 09:52, 18 August 2020 (UTC)
    • @Johnuniq: Wikidata's protection policy says that all items used on over 500 pages should be semi-protected. Each school's colors are used on several hundred pages, and with the likely advent of other wikis making use of these colors, it is likely that a request for semi-protection would be approved if this proposal happened. –IagoQnsi (talk) 20:59, 27 August 2020 (UTC)
      • Anyone is welcome to copy all the data to Wikidata and set up whatever is needed to make it work in articles. Then test it at various Wikipedias (not enwiki) and see how it goes for six months. After that, come back and make a proposal to use it here. Johnuniq (talk) 23:48, 27 August 2020 (UTC)
  • Support per all of the support !votes above. I only recently started interacting with the whole college colors apparatus, and it's a horrible hack that requires a template-protected edit request just to add a missing school. This is an ideal value to store at Wikidata, where it can be available to all languages and others who seek to use it. Regarding abuse, I just don't see vandals really wanting to change Foobar University from red to blue or succeeding at doing so off the radar. {{u|Sdkb}} talk 07:46, 24 August 2020 (UTC)
  • Support this is a value in a template that a user currently can't edit in the way templates are usually edited in Wikipedia. Having to go to https://en.wikipedia.org/wiki/Module:College_color/data to edit the data is completely counterintuitive. ChristianKl ❪✉❫ 15:04, 26 August 2020 (UTC)
If having to go to another page on Wikipedia is counterintuitive ... wouldn’t having to go to a completely different WEBSITE (Wikidata) to edit be a lot MORE counterintuitive? Blueboar (talk) 16:41, 26 August 2020 (UTC)
@Blueboar: Using a Wikimedia sister site shouldn't be too hard; Commons doesn't seem to trip people up too much. If you're able to navigate the current chain of {{CBB roster}} --> {{CollegePrimaryStyle}} --> Module:College color --> Module:College color/data, you'll probably be able to figure out Wikidata too. –IagoQnsi (talk) 17:32, 26 August 2020 (UTC)
  • Oppose, per many of the above comments; RexxS sums it up well. Mike Christie (talk - contribs - library) 17:36, 26 August 2020 (UTC)
  • Oppose, per RexxS and others. Johnbod (talk) 18:06, 26 August 2020 (UTC)
  • Comment - I have been thinking about why I am so negative about Wikidata. And have realized that it comes down to some very simple reasons. First, I am text oriented and not data oriented. I simply don’t see the world in terms of bits of data to be stored and retrieved. Second (And more important), I am an older person, and easily confused by all this whiz bang technotify stuff. I hate templates (with all their confusing parameters and fields) and avoid them if I can (for example, I don’t use citation templates, and still do all my citations by hand using <ref>text</ref> formatting.) However, I find templates a breeze compared to editing Wikidata. I tried, and very quickly gave up. To put it bluntly... I feel that I am UNABLE to edit wikidata. When we move things to Wikidata, they become out of my reach as an editor. If I needed to edit information pulled from Wikidata, I COULD not do so. So much for “anyone can edit”. Blueboar (talk) 21:53, 27 August 2020 (UTC)
  • My opinion is that the citation discussion above is largely a red herring. That's a question of implementation and I'm pretty sure Trappist if no one else could provide a functioning citation in CS1 or CS2 for the reference data related to each color (and I can pick out at least one other user who could do it as he's basically done it already with another template).

    I have at least one or two interests of discussion, the first of which is the question of implementation of the colors. From memory (mobile > me), the college colors module doesn't always provide true colors in the interest of accessibility and being able to ensure that fore and background colors meet AA compliance (or sometimes we add colors in the particular school theme which moderate this question but which are not official). This is not data that Wikidata should want to take and I certainly would not want to foist it on them. How would you propose working around that (assuming my question is not off base)?

    My second interest pertains to use not on the main school page (navboxes among others), which can sometimes add to the expensive operation count. Do you have an implementation in mind which would sort that?

    Given the use in navboxes, I am unsure if this is data we should use from Wikidata. It is almost certainly data they would be interested in (pending q1) and so moving/copying the data there would no doubt be appreciated. --Izno (talk) 21:57, 27 August 2020 (UTC)

    • @Izno: Good questions! I was not aware that any of the colors in this module were not official. How are these unofficial colors chosen? If they are being systematically derived from the official colors, then I imagine we could store only the official colors in Wikidata, and have our module on Wikipedia derive the accessible colors as it retrieves them. If they're being chosen some other way, then we might want to figure out some schema for storing them, perhaps using nature of statement (P5102) to distinguish official and unofficial colors. I think it'd be useful to have these unofficial colors stored in Wikidata as well; AA compliance is something that all Wikimedia projects would be interested in, not just us. It's just a matter of finding a good way to delineate them from the official colors.
    • As for the expensive operations issue, I didn't have a particular implementation in mind. One method that would be easy enough would be to have a bot update a local page every hour with the results of a query for all the college colors, and then have the template/module pull the color from there instead of directly from Wikidata. That wouldn't be such a great solution in the long-term if we were to expand this to be used for more than just college sports, but perhaps there are other ways to efficiently cache this that I'm not aware of, or perhaps it's something that could be raised with the Wikidata Lua developers.
    • We could also just allow a single expensive query to be used here. Most articles only use colors from a single school, so it wouldn't be a very big deal. There are a few articles such as Atlantic Coast Conference which use colors from many schools, and we'd want to have a better solution for them. We could perhaps have a bot use a SPARQL query to update articles that use many schools' colors on an hourly basis. But in the short term, I don't think there are any pages that use enough different schools' colors to hit the expensive function limit, so we don't necessarily have to have a finalized solution for this just yet. –IagoQnsi (talk) 23:42, 27 August 2020 (UTC)
  • Oppose for now per Johnuniq. I am willing to reconsider after seeing it demonstrated and proved to work better than the current system in practice, including ease of use to people unfamiliar with either system, lower workload for editors, compliance with MoS, and resistance to vandalism. · · · Peter Southwood (talk): 13:55, 28 August 2020 (UTC)
  • Oppose for exactly @Blueboar:'s reasons. Wikidata is highly technical. I understand text on English Wikipedia. I don't understand Wikidata at all. The more you move data from English Wikipedia to Wikidata, the more you are reducing my ability to edit English Wikipedia. Mr Serjeant Buzfuz (talk) 14:29, 6 September 2020 (UTC)
  • Oppose you can state that WD is catching vandalism, but vandalism stays for days, weeks if you are lucky enough that the local wiki does not catch it. Buckingham Palace was officially opened in 1826, 2018, 2013, 1999, 2018 (again, now for 22 days), 2017, 2012 (for 4 months), 2012, 2012, 2011 (for a month), 1703 (for 8 months), (deleted) (for 2 months), 1200, (no value), 1020, 1720, 5555, 1720, 1715. Take your pick. Now, that is a number that people actually may know (and still we can't decide), but for colors? #FEFEFE or #FDFEFF? Just keep it local, easier to protect and easier to monitor. --Dirk Beetstra T C 15:17, 6 September 2020 (UTC)

Updating some of the protection padlocks to be more consistent

Please see WT:PROTECT#RFC: Changing protection icons instead. Goose(Talk!) 17:38, 7 September 2020 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

One thing that bugs me about the protection padlocks is that they consist of a mix of letters and shapes, which comes off as inconsistent to me. Therefore, I am proposing to update some of the protection padlocks to match their Commons counterparts, specifically the symbols for full protection and template protection:

Template protection icon from Commons
Full protection icon from Commons

I would also like to see the office protection and extended confirmed protection padlocks be updated to match the rest of the protection icons. The office protection padlock can be updated to have the logo of the WMF, while I am not too sure about how to update the extended confirmed protection padlock. Please provide your input on this idea. Goose(Talk!) 04:30, 7 September 2020 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Cleaning up after #WPWP edits

As a result of the Wikipedia Pages Wanting Photos campaign running on Meta since July and ending this 31 August, there have been a lot of poor-quality edits by editors who are unfamiliar with the image use policy/MOS and/or just don't care about quality. Some had incorrect image placement, some introduced bad captions, some inserted totally irrelevant images, and some completely broke infoboxes and other templates. Since a lot of the affected articles are of obscure unwatched topics, a coordinated clean-up effort might be needed. Maybe a centralised noticeboard to identify problematic editors and check their contributions? There was earlier discussion at AN but it didn't seem to reach any conclusion. --Paul_012 (talk) 09:27, 28 August 2020 (UTC)

I'd note that we do have 1073 to log these, so it's not hard to track, but there's 32k hits. ProcrastinatingReader (talk) 14:13, 28 August 2020 (UTC)
Thank you, ProcrastinatingReader. I missed that the edit filter had already been implemented. Would you know if there's a way to get a de-duplicated list of users who have triggered the filter, so that spot checks can be done? --Paul_012 (talk) 17:02, 28 August 2020 (UTC)
That edit filter ought to be changed and set up to disallow the edit and then from this point forward this event should by all means be banned on this project, It's disruptive and to be very blunt it's a pain in the ass for us who have to go through everyones contribs checking each and every edit. –Davey2010Talk 20:51, 30 August 2020 (UTC)
To be clear, the consensus on ANI was that it was not disruptive, and you don't need to check everyone's edits. Our normal community workflows probably caught many of the mistakes already, and there is no evidence that the quality of the edits is significantly different than other newcomer edits, Sadads (talk) 20:02, 9 September 2020 (UTC)
It's great that we have new editors as part of this, even if they aren't doing things quite right yet. I hope that you're welcoming them and teaching them how to best edit Wikipedia. Thanks. Mike Peel (talk) 19:35, 3 September 2020 (UTC)
+1 to what Mike said: most of the edits I have seen are constructive, and build on the work of other Wikimedia communities -- so please be thoughtful in engaging with folks. Sadads (talk) 20:02, 9 September 2020 (UTC)

Link to following Accessibility features on the top of home page

There should be an accessibility shortcuts to the following items on the top of the first and foremost page.

  1. https://en.wikipedia.org/wiki/Help:Multilingual_support , Font and rendering support, IPA, Braille etc.
  2. https://www.mediawiki.org/wiki/Universal_Language_Selector
  3. https://en.wikipedia.org/wiki/Wikipedia:Spoken_articles
  4. Wiki markup language support. (What is a markup language?, what is the Wiki markup language, its List of codes, cheatsheet etc.)

This is a keen request. This should be added on top of all pages, as an accessibility hyperlink. Even this should be added in all other Wikis and as well as Wikipedia home pages. Thank you in advance. RIT RAJARSHI (talk) 22:23, 13 September 2020 (UTC)

2 is already linked in the sidebar. 3 is irrelevant for the vast majority of articles without spoken versions, and there is already an appropriate template for articles that have spoken versions. 1 is also already linked by similar templates. 4 is irrelevant for everyone but users that edit with markup, which is an extremely small portion of all visitors to Wikipedia. – Teratix ₵ 01:59, 14 September 2020 (UTC)
Agreed. {{u|Sdkb}} talk 03:58, 14 September 2020 (UTC)

I agree some of the feature exists but the "how to use them" is not available directly on first page. The pages I linked are mostly some "help pages" and they should be linked in the main pages. RIT RAJARSHI (talk) 10:03, 14 September 2020 (UTC)

But why should the help pages be displayed on every article, where they will be irrelevant clutter the vast majority of the time? – Teratix ₵ 11:03, 14 September 2020 (UTC)
Not to distract, but "Spoken articles" should be linked to less, not more. Screen readers are fine nowadays. People who navigate Wikipedia this way want the current article spoken by their screen reader 99% of the time, and a random snapshot of unknown quality of the article as it was 6 years ago 1% of the time. And in the rare 1% of the time they might prefer a human spoken version (highly technical article? Lots of foreign words?), it usually doesn't exist, because very few WP articles have manually spoken versions. SnowFire (talk) 00:26, 15 September 2020 (UTC)

Migrate some uses of Template:Commons to Template:Commons category

Hi all. We have {{Commons}} that provides general links to Commons and {{Commons category}} that links specifically to Commons categories. When used without parameters, {{Commons}} links to Commons search pages; when used with a parameter it uses that parameter to link specifically to that Commons gallery or category. Meanwhile, {{Commons category}} when used without a parameter uses Wikidata to link to Commons categories (where they are defined through sitelinks on Wikidata) or the named parameter where that is provided (which is cross-checked with the Commons category sitelinks on Wikidata to add it to tracking categories).

I would like to bot-migrate uses of {{Commons}} to {{Commonscat}} in circumstances like these:

  1. {{Commons}} is used without a parameter (= links to the search page) but a Commons category is sitelinked to the page/category
  2. {{Commons}} links to a Commons category
  3. {{Commons}} links to a non-existent or redirected gallery but a Commons category is sitelinked
  4. Optional: if {{Commons}} links to a non-existent or redirected gallery, with no Commons category, then a) the link could be removed or b) the parameter removed..

This could be done as a one-time run or run regularly to catch new cases. There are currently ~65k uses of {{Commons}} (compared with ~780k uses of {{Commons category}}). If this is OK, then I can submit a bot proposal - I already have semi-automated pywikibot code to do this that I can easily convert to a bot, I have been running it interactively in the last few days but given the number of cases and high accuracy rate this is probably better as a bot. (e.g., for case 1, [2], or for case 2, [3] - cases 3 and 4 aren't yet coded). This is part of my wider work to clean up enwp <--> wikidata <--> commons sitelinks, I can provide more background if needed.

(BTW, I'm not sure if this should be an RfC or just a straw poll, if anyone thinks it should be a RfC then feel free to add the tags.) Thanks. Mike Peel (talk) 18:17, 30 August 2020 (UTC)

question mark Suggestion I have an alternative, somewhat overlapping proposal that I think will achieve the same ends and require less bot editing and make things better for readers. I propose turning {{Commons}} back into a generic "link to the best item at Commons" template that it was before the semantics of Commons search changed. We could have {{Commons}} use Module:Commons link, as {{Commons-inline}} does. When {{Commons}} has no parameters, Module:Commons link would use Wikidata to find one of a Commons gallery or Commons category (if they exist). The module is "fail soft": if Wikidata is inconsistent, it falls back to Commons search. Using this module would cover Mike Peel's cases (1) and (3), above. I enthusiastically support using a bot to handle case (2) above (because the intent of the editor is clear: {{Commons|Category:Foo}} should clearly be {{Commons category|Foo}}). I bet that would be a one-time-run bot.

The use of this module is prototyped at Template:Commons/sandbox, with test cases at Template:Commons/testcases

Some further notes to answer questions that other editors may have:

  • About 2 years ago, the meaning of Commons search changed. When there was a search term "Foo", if there was a gallery called "Foo", readers would be taken directly to that gallery. If not, and if there was a category called "Foo", readers would be taken directly to that category. Otherwise it would fall back to a generic search for "Foo". This was much friendlier than the current situation. Module:Commons link (function getGalleryOrCategory) is designed to restore that behavior (as far as it can), using Wikidata. If it cannot, it will default back to search.
  • Module:Commons link is ready for prime-time. It is used on 25,000+ articles by {{Commons and category}}, {{Commons and category-inline}}, {{Commons-inline}}, and {{Sister links}}. It has test cases and I have received no complaints over the last several months.
  • Module:Commons link is fail-soft and non-intrusive:
    • If a parameter is provided, it uses that instead of searching through Wikidata
    • It looks in multiple Wikidata fields (e.g., sitelinks) for possible galleries and categories: if a gallery or category is available, it uses it. If Wikidata is self-contradictory, it gives up and falls back to Commons search.
    • Many thanks to RexxS and Mike Peel to advice on how to build this module: I relied on Module:WikidataIB for inspiration!
  • In case Mike Peel was wondering: if P373 is deprecated, Module:Commons link would still function without errors. It uses P373 as a possible source of gallery links, but if it is empty or removed, it would simply find fewer galleries.

I believe that Module:Commons link fulfills most of the goals of the proposal by Mike, while being automatically up-to-date when Commons sitelinks (e.g.) change.

What do other editors think of this? — hike395 (talk) 19:44, 30 August 2020 (UTC)

  • @Hike395: I was hoping to keep this focused on the specific issue of whether to use the {{Commons}} vs. {{Commons category}} templates as that's a practical step that we can take now. If we're going to turn this into a broader discussion, then I would like to see us use a single template that links to the gallery and/or category as they are available through the sitelinks on Wikidata, possibly through Module:Commons link if @RexxS: can confirm that this is compatible with Module:WikidataIB). Thanks. Mike Peel (talk) 20:02, 30 August 2020 (UTC)
    • @Mike Peel: Looking at Module:Commons link, I can confirm that it interacts with Wikidata in just the same way that WikidataIB does. Essentially, it adds multiple utility functions for working with categories and galleries on Commons, and two main calls to get the associated Category and Gallery respectively into Wikitext. --RexxS (talk) 23:11, 30 August 2020 (UTC)
@Mike Peel: Apologies for possibly randomizing the discussion. Some more notes:
  • If we get consensus around this, we can promote Template:Commons/sandbox to Template:Commons right away, and achieve part of what you desire from a bot run immediately, without waiting. It's a practical step we can take today. In this case, {{Commons}} would become what you are suggesting (a "template that links to the gallery and/or category as they are available through the sitelinks on Wikidata"), with extra checks from other properties for fallback and consistency.
  • Back in March, RexxS generously did a code review for Module:Commons link, and didn't see any problems going into production. Mike: did you have specific concerns about how Module:Commons link was different than Module:WikidataIB? There are four main entry points: getGallery, getCategory, getGalleryOrCategory (return one), getGalleryAndCategory (return both). Instead of checking Wikidata in series, bailing after the first item is found, Module:Commons link looks in all of the same places as Module:WikidataIB, and does not return a Commons link if any inconsistency is found.
  • In that same discussion, Mike brought up an interesting idea for combining Commons templates. Can we defer that discussion in favor of a simple call to Module:Commons link in {{Commons}}? As you said, the module call is a simple practical idea we can implement today. The Commons templates can always be merged later, if we get consensus.
More comments/suggestions/issues are welcome! — hike395 (talk) 23:09, 30 August 2020 (UTC)
@Hike395: Looking at this in more detail, I think the changes at Template:Commons/sandbox are a useful improvement to reduce the number of bad links, and I think you should make that change immediately (I can make it live if you don't have the permissions to do so). However {{Commons cat}} includes the tracking categories, in particular for cases where there is no defined parameter and no sitelink on Wikidata (= link to the search in this template), and where the local text doesn't match Wikidata. These are really useful for maintenance, but they become more complex to handle when you have both categories and galleries (in particular if you don't know which the editor wanted to link to). Pi bot is also set up to maintain the links to categories, which is part of why I want to migrate cases over to use {{Commons category}} where possible, without removing links to galleries. In the long term, I would like to see us using both templates without parameters (just using Wikidata), and linking to either/both the gallery and category (and maybe related categories), but that's a separate discussion.
Bottom line, I think you work is good and should be made live, but it's parallel to the specific proposal I've put forward here. I think both could go forward in parallel. Thanks. Mike Peel (talk) 19:01, 1 September 2020 (UTC)
@Mike Peel: Thanks! I just made it go live. I agree that Module:Commons link is complementary to your proposal. Your proposal contains a valuable set of cleanups that I think we should implement. I think we're just discussing the details of the cleanups at this point.
I'm more than happy to implement the tracking category logic from {{Commons category}} in Module:Commons link. That should be easy in Lua. If you agree, I can add that logic to {{Commons}}. I can also expose it as an entry point, so we can make {{Commons category}} a bit more compact (if you wish). I'll start work on that in Module:Commons link/sandbox. After implementing that, we won't need to change the template to get the tracking categories.
I am thinking that {{Commons}} and {{Commons category}} express different editor intent. The latter means "I really want to link to a category", while the former means "I want the best Commons thing, of some sort". I'm just trying to be super careful about having a bot make massive changes that may alter editor intent. Let me break down the possible cases you've outlined, and make a suggestion that should preserve editor intent.
  1. {{Commons}} has a first parameter that is a category --- clearly move to {{Commons category}}
  2. {{Commons}} has a first parameter that is a redirect to a category --- move to {{Commons category}} is very likely to be correct
  3. {{Commons}} has a first parameter that doesn't exist at Commons. There are two possibilities:
    • The editor made a mistake (likely) --- the right answer is to delete the first parameter and let Module:Commons link do its Wikidata magic.
    • The editor may have intended to invoke a Commons search (unlikely, but possible)
    We could simply delete the first parameter and ignore the unlikely case? Or I can add a |fallback= parameter to {{Commons}} and Module:Commons link, and have the Commons search execute after Wikidata fails. What do other editors think?
  4. {{Commons}} does not have a first parameter --- Now that we are using Module:Commons link, which is specifically designed for this case, I would leave this case alone. (Especially after I add the extra tracking categories).
To summarize, I support using a bot in my cases (1) and (2), above, oppose using a bot in my case (4), and let's discuss my case (3). I'll get to work on the tracking categories.
What do other editors think? Mike? — hike395 (talk) 14:37, 2 September 2020 (UTC)
@Hike395: I think we agree on your first one (my #2), and the second one that is part of my #3. Your third and fourth points assume that the link to search was correct (I'm automatically excluding any that link to an existing gallery here), I'm not sure that is a reasonable assumption. Particularly in categories, I think it makes sense to link to the Commons category rather than the search results, where (and only where) such a category exists. If there's no such sitelinked category, then my proposal wouldn't alter those cases. Perhaps the general question here is: do we want to link to search rather than an existing Commons category on the topic? Thanks. Mike Peel (talk) 17:55, 2 September 2020 (UTC)
@Mike Peel: I agree that search links are a bad user experience. However, now that Module:Commons link is used in {{Commons}}, my case #4 doesn't use search links any more. My case #4 now directly goes to the Commons category (if there is no corresponding Commons gallery, otherwise it will use that). I've written (but not yet tested) the tracking category logic in Lua. Soon there'll be no compelling reason to move case #4 over to {{Commons category}} (I think). — hike395 (talk) 19:16, 2 September 2020 (UTC) I
@Hike395: OK, that's good. The question then becomes, for the cases where the template links to Commons categories only, whether it's better to use {{Commons category}} to make it clear that it's a link to a Commons category. Then we keep the gallery and category lik maintenance separate. But the other way of looking at it is, it's the first step to merging to a single Commons template that sorts it out automatically. Thanks. Mike Peel (talk) 17:34, 3 September 2020 (UTC)

@Mike Peel: Instead of changing the template (which possibly alters editor intent, and would be incorrect if a new gallery shows up), should we change the output of the templates to warn editors (and readers) that Wikidata is taking them to a category? That is, if {{Commons}} (or related templates) find a category in Wikidata instead of a gallery, should we populate the sister link box with the label of Category:Foo instead of Foo ? Or would readers find that too confusing? 14:56, 6 September 2020 (UTC)

I think the general convention is to not show 'Category:' - the same as the categories at the bottom of an article don't have the prefix. My preference would still be to change the template. Thanks. Mike Peel (talk) 14:16, 12 September 2020 (UTC)
@Mike Peel: I still wonder if we can come to a compromise acceptable to both of us. How about this: can we change the output text for these kind of templates?
If the template links to a gallery, it could read something like
Wikimedia Commons has a gallery for Foo
If it links to a category, it could read something like
Wikimedia Commons has a category for Foo
If it falls back to some sort of search, it could read as it does today
Wikimedia Commons has media related to Foo
and if {{Commons and category}} returns both, it could read
Wikimedia Commons has both a gallery and category for Foo
I like this, because it makes the Wikidata logic more transparent to both editors and readers. What do other editors think? — hike395 (talk) 00:58, 13 September 2020 (UTC)
@Hike395: That's the kind of solution I'd like to see us reach in the long term, but it's kinda beyond what I was proposing at this step. I'm puzzled why you don't like the idea of moving the category links over to the other template for now, though, where there isn't a defined parameter? We can always merge the templates in the future, if there is consensus to do so (and I think that needs wider discussion than we're seeing here so far). Thanks. Mike Peel (talk) 20:16, 13 September 2020 (UTC)

@Mike Peel: Here's one case: Editor A uses {{Commons}} on Foo, with only Commons:Category:Foo existing at Commons. Bot comes through and converts it to {{commons category}}. Nothing has changed. But then Editor B creates Commons:Foo. Without the bot, the page would auto-update to link to Commons:Foo. With the bot, the page would not auto-update. Why do we want to turn off auto-updating?

Here's another example. I come across an article that violate WP:IG: I transwiki the gallery over to Commons, and try to make it nice (include FPs, organize into sections, etc.). If the article has {{Commons}}, then I leave that alone, because some prior editor wanted some sort of Commons link. If the article has {{Commonscat}}, then I convert it to {{Commons and category}} because I know that some prior editor explicitly wanted to link to a Category, and they would be unhappy if I took their category link away. If a robot converts {{Commons}} to {{Commonscat}}, I don't know what editors intended. I could look back into the history, it's true, but why make me do the work for no functional difference?

Hope this helps explain. — hike395 (talk) 05:13, 16 September 2020 (UTC)

Technical notes

As this is now just targeting namespace:1 ("Talk:") the technical implementation would be by setting (noindex,follow) in $wgNamespaceRobotPolicies. This can be done with a phabricator request such as in phab:T104797, such a request should include a permanent link to a successful, closed community discussion. — xaosflux Talk 23:27, 19 August 2020 (UTC)

A secondary decision that can be made is if the use of __INDEX__ should be allowed to purposefully allow indexing on a page-by-page basis (this is an update for $wmgExemptFromUserRobotsControlExtra) - by default manually tagging for indexing would be allowed, but it can also be forbidden if necessary. Pages manually tagged for indexing will appear in Category:Indexed pages for tracking and review. — xaosflux Talk 23:27, 19 August 2020 (UTC)

On a related note, currently most talk pages of BLP's are noindexed by way of having a template that inlcudes the NOINDEX directive, as can be seen in Category:Noindexed pages. — xaosflux Talk 23:30, 19 August 2020 (UTC)

From below, some talk page copy-paste "archives" may not be tagged like this. A bot may be suitable for this task if wanted. — xaosflux Talk 13:09, 20 August 2020 (UTC)
The most talk pages of BLP's are noindexed by way of having a template that inlcudes the NOINDEX directive is because those talk pages transclude Template:BLP, which contains a __NOINDEX__. Conventionally, the {{BLP}} is not used directly, but is done by means of either or both of {{WikiProject Biography|living=yes}} or {{WikiProject banner shell|blp=yes|1=...}}. --Redrose64 🌹 (talk) 19:37, 20 August 2020 (UTC)
  • Something I've just observed for a new page I created is that, even though it's not listed on Google yet since it hasn't been checked off at NPP, its talk page is listed. That's definitely not what we want. {{u|Sdkb}} talk 23:49, 31 August 2020 (UTC)

Measures to curb sandbox abuse?

I know this may have been proposed before, but would it be feasible to fix the MediaWiki software so that sandbox edits won't count towards autoconfirmed status, or alternatively, make the autoconfirmed privilege a little stricter while keeping to the project's goals of a free encyclopedia? I've seen vandalism from a certain troll whose name I won't mention through sock accounts abusing the sandbox just to get around the autoconfirmed protection even on pages with indef semi-protection in place. I am honestly incensed at this considering how persistent and relentless said troll was in disrupting the project and sowing discord amongst those who he rubs elbows with, but I digress. Blake Gripling (talk) 13:45, 10 September 2020 (UTC)

It would be pretty useless to limit, and counterproductive. I'd rather have someone goof around in their sandbox than insert random, incorrect commas on live articles. Must consider the purpose of autoconfirmed, which is a very minimum requirement and not meant to be hard to get. If you try change that, these "trolls" won't stop, you'll just inadvertently send the issues into mainspace. ProcrastinatingReader (talk) 15:38, 10 September 2020 (UTC)
It is possible to use other types of thresholds for granting access to automatic groups, but as PR notes above - the bar for "autoconfirmed" is meant to be very low - primarily as a screen against non-wikipedia bots and casual spammers/vandals. — xaosflux Talk 16:06, 10 September 2020 (UTC)
  • I know this doesn't directly address your original question, but if there's a specific page (or small set of pages) that are being abused, WP:ECP, could be applied. You can request this at WP:RFPP, or if you really feel it can't be discussed on-wiki, feel free to email me. -- RoySmith (talk) 16:20, 10 September 2020 (UTC)
  • This question comes up fairly regularly, most often "to limit the permission to the mainspace edits" in some form (I'm am pretty sure it's on WP:PEREN at this point). It would require significant rework in how automated permissions are granted. Users gaming permissions are routinely taken care of after the fact. --Izno (talk) 16:37, 10 September 2020 (UTC)
Strongly Oppose The sandbox is a sandbox. No damage is done to the wiki if someone vandalizes the Sandbox. Like yeah, you're a jerk if someone has made an experimental page there, but it's not permanent. This doesn't make sence. Also, if a sandbox is like practice, then it should count towards edits. No explanation needed. Arsonxists (talk) 00:19, 11 September 2020 (UTC)
Comment: To be clear, this isn't to prevent sandbox edits from being logged and counted towards one's contribution history, bur rather to keep abusive users from gaming the system by making test edits to qualify for autoconfirmed status. Blake Gripling (talk) 02:01, 11 September 2020 (UTC)
I am pretty sure that you have to wait 4 days before autoconfirmed status, even if you made 10 edits. I do get what you're saying, but ussually, abusers don't go to the Sandbox to get autoconfirmed status, they ussually just vandalize ones that anyone can edit.Arsonxists (talk) 14:50, 11 September 2020 (UTC)
And those who do, will find other ways to bypass the auto confirmed status another way. It's an open encyclopedia, this is the side effect of that mission statement. —TheDJ (talk • contribs) 14:36, 14 September 2020 (UTC)
  • I am inclined to agree with User: ProcrastinatingReader. It is better to have some one goof around in a sandbox than vandalise main articles, and if a Wikipedian does, for example, insert a blatant hoax, s/he is likely to get a message saying the sandbox is meant for experimentation Vorbee (talk) 15:30, 14 September 2020 (UTC)
  • @Blakegripling ph: Won't help, generating 10 edits without using the sandbox is a piece of cake. I think the 10 edits are more to stop sleepers from gaining autoconfirmed after 4 days. And well-meaning newbies, of course. I doubt it was ever intended to stop (serious) vandalism. — Alexis Jazz (talk or ping me) 16:57, 14 September 2020 (UTC)
    • It's actually kinda helpful, because if vandals want to vandalize a part of Wikipedia that's auto-protected, it may force the vandal to either not vandalize that part, delay the attack 4 days, or be a contrustive part of Wikipedia for a moment. So yeah, removing sandbox edit counting won't help that much. Arsonxists (talk) 18:46, 14 September 2020 (UTC)
  • Is what the proposer is saying that one's contributions to the sandbox should not count towards one edit count? Would there be a way to indicate, if a Wikipedian clicks on "Edit count", how many edits were merely contributions to the sandbox? Vorbee (talk) 07:50, 16 September 2020 (UTC)
  • I have just been on the Edit count function at Wikipedia, and whereas it tells one how many edits are to the talk pages, how many are to the main pages, how many are to User talk, how many are to category talk and so forth, it does not have a separate section saying how many edits are to the sandbox. I am not sure what category edits to the sandbox would fall under, and while I appreciate this might be technically difficult, I am wondering whether it would be feasible to introduce a section "Sandbox" in Edit count. Vorbee (talk) 06:41, 17 September 2020 (UTC)

CentralNotice banner for Indic Wikisource Proofreadthon 2020

Dear colleagues, please comment on CentralNotice banner proposal for Indic Wikisource Proofreadthon 2020 for the Indic Wikisource contest. (1 Oct2020 - 15 Oct, all IPs from India, WPs,only). Thank you. Jayanta (CIS-A2K) (talk) 12:32, 18 September 2020 (UTC)

Implement "smart" custom toggles which act based on their state

Do you know how long an enhancement request takes to be reviewed at Phabricator? I created a task, phab:T262622, last week and it still hasn't been commented. I don't know the workflow there, so if such change would likely take more than 2-3 weeks, would it be adequate to implement the request (given it's accepted) here (at the MediaWiki namespace)? The script linked in the task is my sandbox. Alexiscoutinho (talk) 14:06, 18 September 2020 (UTC)

@Xaosflux and TheDJ:

Get rid of stub tags

I know this is a bold proposal, and is likely to be controversial, but stub tags aren't useful. They don't get people to edit stub articles, which is their stated purpose. They do have a number of negatives: They often remain in articles long after they has been brought up to Start-class and higher. They conflict with the classes stated on talk page banners, which are often more up-to-date. They add a useless image and clutter to the articles. It's time to begin to think about eliminating them. For those people who might be feeling reticent, perhaps an experiment should be run where stub tags are removed from a random subset of articles, then they are compared in, say, one year's time, to a subset of similar articles and measured for levels of (destubifying) improvements. Any thoughts? Abductive (reasoning) 00:35, 21 June 2020 (UTC)

I think you have actually identified two distinct issues here: 1) we don't have any evidence that stub tags are achieving their stated purpose, and 2) stub tags are conflicting with WikiProject assessments. For #1, I think an experiment could yield from interesting results. For #2, I think this would make a nice bot task to remove stub tags when a WikiProject assessment is upgraded from stub. It might be worth upgrading {{WPBannerMeta}} to generate a reassess flag for the bot to add when a stub tag is removed, but a WikiProject still has it assessed as a stub. For the experiment on #1, I would suggest:
  1. Get a list of something like 10,000 random articles, and remove any that are not assessed as a stub by WikiProjects, or have neither a stub tag nor a WikiProject banner on its talk page.
  2. Do an assessment of the remaining articles and remove any that aren't actually stubs.
  3. Split WikiProject claimed articles by stub-tagged or not, and remove/add stub tags to get each WikiProject's article count roughly even (shoot for at least 1/3 of each). Remove stub tags from the unclaimed articles to even out the total count, again going no less than 1/3 for either group.
  4. For the two groups, drop the top and bottom (quartile? ⅒th?) in terms of edit size, ignoring any bot edits, and compare the two groups by interest (number of edits by unique editors), engagement (number of non-revert, non-patrolling edits by registered editors), and overall article change (total prose added to article during experiment run time). VanIsaacWScont 14:17, 21 June 2020 (UTC)
    Thanks for your input. Abductive (reasoning) 23:07, 21 June 2020 (UTC)
  • Oppose The OP doesn't present any evidence to support the proposal. Me, I'd rather eliminate project templates and assessments as I've never seen these do any good and they tend to stick around for longer. The stub templates are comparatively unobtrusive and have an encouraging and pleasant tone. Andrew🐉(talk) 22:47, 21 June 2020 (UTC)
    So, I present no evidence, then you say talk page assessments don't do any good, and also present no evidence. Abductive (reasoning) 23:07, 21 June 2020 (UTC)
    This is not my proposal and so it's not my responsibility to be providing evidence. But here's a couple of examples – both being articles that I created recently. Firstly, consider Ambarnaya. I created this with the {{river-stub}} tag. User:Catan0nymous added two more stub tags: {{Russia-stub}}, {{Siberia-stub}} and user:Abune then consolidated the stubs into {{Russia-river-stub}} and {{Siberia-stub}}. These tags seem to have three functions:
    1. Placing the article into the categories: Category:Siberia geography stubs and Category:Russia river stubs
    2. Displaying some appropriate icons -- maps of Russia and Siberia
    3. Advising the reader that the article is just a start and inviting them to help expand it
    My second example is List of longest-running radio programmes. In that case, I started out with the {{under construction}} tag because the page needed a good structure as a foundation and I wasn't sure what would be best. Once that was done, I removed the tag. By that time, the list structure was established and I used the {{dynamic list}} tag to indicated that the list was quite open-ended. That tag also invites the reader to add sourced entries and so I didn't see the need for a stub tag too.
    Neither of these cases indicate that stub tags are a problem that needs fixing. Sensible editors use them as they see fit and they don't seem to cause any trouble. One feature which helps is that, by convention, they are placed of the foot of an article, where they don't get in the way.
    Andrew🐉(talk) 08:00, 22 June 2020 (UTC)
    Andrew Davidson, I agree with the assessment system. It's absurd. Articles can be Stub, Start, C, B, GA, A, or FA. That's seven levels, which is absurd. The gradations are way to small, and the assessment criteria way to subjective. Class A articles are "Very useful to readers", but GA are, "Useful to nearly all readers". That's absurd. -- RoySmith (talk) 02:15, 22 June 2020 (UTC)
    The minimum acceptable standard for an article on the front page at ITN, OTD or DYK is effectively B class due to the requirement that they be fully referenced (the same goes for the de-stubathons), although we can't state that explicitly because B class ratings are in the hands of the projects. GA and FA are community assessments. B class articles are generally capable of becoming GA, but there is little incentive to submit them for review (GA is already backlogged enough), unless you are seeking to take it to DYK, FA or FT. Similarly, A class articles are generally capable of becoming FA after a review, but most cannot because each editor is limited in the number that can be submitted. So we have three categories (stub, start and C) for poor quality articles. The differences between them are fairly objective, but the value of the distinction is questionable. Hawkeye7 (discuss) 22:57, 20 July 2020 (UTC)
  • Comment I doubt the suggested experiment would produce any clear results frankly. People are often reluctant to remove stub tags, or simply don't notice them. The wikiproject tags are just as prone to under-rate as the ones on the article in my experience. What might be more useful is a list of articles tagged as stubs where the article is over a certain size (not sure what). Reviewing these would I expect show most can be removed. Of course people still have to do this. Johnbod (talk) 23:59, 21 June 2020 (UTC)

Considering I've run contests with over 1000 articles destubbed directly from stub categories like WP:The African Destubathon and Wikipedia:The Great Britain/Ireland Destubathon and am running Wikipedia:The 50,000 Challenge which directly feeds off stub tags this is one of the dumbest, most ignorant proposals I've seen for some time and that's saying something!! There is an issue with updating the project tags once an article is no longer a stub and stub tags remaining in articles not stubs but that's hardly a reason to get rid of them entirely. Rosiestep and myself proposed a bot to sort that out but the Wikipedia community being the divided way they are wouldn't get us the consensus we need to sort it out.† Encyclopædius 08:19, 22 June 2020 (UTC)

I don't see why, User:Encyclopædius. I won some sort of prize in one of your excellent contests, but when looking for articles to improve, I remember just removing the stub tag on about five for every one that actually was a stub. I don't agree with complete abolition, but they are in a serious mess & should be sorted out. Johnbod (talk) 13:58, 22 June 2020 (UTC)
Yup, and can you believe when I proposed a bot to control the inconsistency and remove stub tags from articles with over 1.5 kb readable prose and update the talk page tags some people opposed? Andrew Davidson for a start. † Encyclopædius 14:04, 22 June 2020 (UTC)
Can a bot at least go around and remove stub tags from all really huge articles that have talk page templates that claim Start-class or above? Then maybe work its way down to smaller articles until it on the verge of making errors? Abductive (reasoning) 04:18, 24 June 2020 (UTC)
Yes. In fact, the MilHistBot is already capable of carrying out this task. Hawkeye7 (discuss) 22:57, 20 July 2020 (UTC)
AWB does something similar, too, by removing stub tags from articles that are above a (generous) size. WhatamIdoing (talk) 00:35, 7 August 2020 (UTC)
  • "stubs" help navigate articles that need attention. They typically are located at the very bottom of article and do not interfere with regular reader who has no intention to improve article. Basically nothing is wrong with "stubs" as far as I'm concerned. User:Abune (talk) 13:04, 22 June 2020 (UTC)
    Very true, User:Abune. Readers rarely notice them, and even if they do, it is hard to see that they have any negative impact. Some editors find them useful. What's the problem? (A: there isn't one). Edwardx (talk) 14:34, 22 June 2020 (UTC)
  • Qualified support. I generally agree with the issues raised by the proposer. Stub tags are not particularly aesthetically pleasing, and do tend to linger after the article has been expanded to the point where they are no longer appropriate. Efforts to just find and remove them at that point become busywork. I am wondering if there is some template magic that can be applied so that stub tags on articles that are likely not stubs can turn invisible, and just show up as categories. BD2412 T 15:11, 22 June 2020 (UTC)
    @BD2412: no problem: Template magic. 10000 might be a tad much. Also you'd have to make all stub templates pass the forcestub parameter in order to be able to force the stub template on articles longer than 10000 bytes. - Alexis Jazz 01:16, 21 July 2020 (UTC)
  • Simpler solution: If you come across an article that you think is no longer a stub... remove the tag. Problem solved. Blueboar (talk) 15:19, 22 June 2020 (UTC)
    Not really - pretty much no one who would know how to do this ever looks at these articles, all xxx,xxx of them (what is the number, does anyone know?). Johnbod (talk) 21:44, 22 June 2020 (UTC)
    There are 3,399,601 stubs as of now. You can see the stats here. TryKid [dubious – discuss] 23:28, 22 June 2020 (UTC)
    Yikes! Over half our articles. These are project ratings though. I see 4,310 "top importance" stubs! Thanks, Johnbod (talk) 01:28, 23 June 2020 (UTC)
    I'd like to see one of those top importance stubs. To make a start I followed TryKid's link to find Category:Stub-Class_articles. From this, I selected Category:Stub-Class Accounting articles because there is a systemic bias against business articles. Accounting standard looked promising and I found that this is assessed as High importance but Stub class. This is all coming from the project template as the article page doesn't have any tags. It could use some because I immediately noticed a blatant howler, "Accounting standards were largely written in the early 21st century." What I also notice is that while its talk page only had 7 readers in the last 30 days, the article itself had 2,481. I could now tag-bomb the article but what it really needs is improvement... Andrew🐉(talk) 08:55, 23 June 2020 (UTC)
    Yeah, 3M project rating stubs. The number of "tagged" stubs seems to be 2,265,086, from Category:All stub articles. This number looks more correct since many of the "top importance stubs" aren't stubs anymore and are already detagged but the talk page wasn't updated. As previously pointed out, even many of the "tagged" stubs aren't stubs. Maybe a bot that automatically upgrades project rating of stubs to "start" if a tag isn't found on the main page is needed. Blofeld's idea of automatic detagging if the article is above a certain size was also good. TryKid [dubious – discuss] 12:59, 23 June 2020 (UTC)
  • Support In my early (2005) days, stub-sorting was one of my favorite hobbies, and it saddens me to finally deprecate the stub templates, but nowadays they are duplicative of the assessment templates on the talk page and thus unnecessary. -- King of ♥ ♦ ♣ ♠ 01:58, 23 June 2020 (UTC)
  • Oppose, although a possible alternative is to link the stub tags to the WikiProject banner, if the article is assessed as a Stub class a bot adds the tag, if (hopefully when) the article is upgraded to Start or higher a bot removes the tag. Cavalryman (talk) 02:19, 23 June 2020 (UTC). Amended comment to oppose retaining alternate. Cavalryman (talk) 06:55, 27 July 2020 (UTC).
    • This is a very good idea. It fixes the discrepancy between main page and talk, while keeping some level of friendly encouragement at the main page. - Nabla (talk) 14:52, 27 June 2020 (UTC)
  • Oppose While I agree that many of the problems you listed are real and affect Wikipedia, stub tags are necessary. They may not be very effective at getting readers/editors to add content to or destubify articles, but they make a vital contrast between what is and is not a reasonable length for an encyclopedic article. Most readers don’t care to browse Wikipedia’s myriad informational pages on article length, the different classes, prose, etc. Stub tags are simple, easily understood, and to-the-point. If we’re going to get rid of stub tags, why not just get rid of the stub classification altogether? It doesn't make sense. Improvements should be made, but we need them. Their importance to the project overrides any negative aspects. MrSwagger21 (talk) 02:26, 23 June 2020 (UTC)
  • Oppose per MrSwagger21. - ZLEA T\C 02:52, 23 June 2020 (UTC)
  • Comment on balancing editor and reader needs. Regarding editor needs (which we always tend to overprioritize, given the systemic bias introduced by who we are), my takeaway here is that it seems there's no evidence that the tags draw editors, and while it's perfectly plausible they do, this might be a good thing for someone to do a research study on. Regarding reader needs (which I don't see really getting proper attention here), the way we indicate article quality is a little quirky — we indicate GA/FA with a topicon, but stubs with a tag at the bottom, and nothing in between. I think there's a reasonable case to be made that stubs, given their low quality, ought to have the tags as a sort of warning. The counterargument would be WP:NODISCLAIMERS and the fact that there's a distinction between low quality and just short, and most stubs in my anecdotal experience are not lower quality than start/C class pages, just shorter. I'm not sure where I land on the necessity of stub tags, but I hope we'll consider how they impact the experience of non-editing readers, not just editors. I have brought up before that there is room for improvement in how we present content assessments to readers more generally (particularly, the difference between GAs and FAs is not made clear), and I'd like to see more work in that area. {{u|Sdkb}} talk 07:53, 23 June 2020 (UTC)
  • Oppose no. They are helpful and low-key. Not in anybody's way. Regards, GenQuest "Talk to Me" 10:47, 23 June 2020 (UTC)
  • Comment: sorry for the somewhat self-promo, but this is something that could hopefully be done with relatively little consternation were my proposal for an extension that adds categories from tags on the talk page successful. Such a move would allow the pages to retain the stub categories, without the duplication of effort in tagging the article with stub tags and also doing the assessments on the talk page. Naypta ☺ | ✉ talk page | 13:07, 23 June 2020 (UTC)
  • weak oppose for now - People I talk to (non editors) are often not aware a talk page even exists for an article. If a stub tag encourages the occasional person to add content then I see that as a Good Thing. If there were some way of showing that there was no evidence that this occurs, then I'd support their deprecation. I should add that the proposal was worth bringing up and I did consider supporting, and I do think the topic is worth exploring. Cas Liber (talk · contribs) 04:54, 24 June 2020 (UTC)
  • Conditional support. Stub templates are messy and outdated, and does not do what they are supposed to do, although they have other important uses. I suggest removing the templates, but replacing them with categories or a function within WikiProject banners. Rehman 05:21, 24 June 2020 (UTC)
  • Support - simplification is good. What stubs do can/is/should be done in a talk page banner, e.g. wikiproject assessments. Bottom line is we should have one place where we record what state an article is in, and that one place should probably be in a talk page banner. Levivich [dubious – discuss] 19:26, 26 June 2020 (UTC)
  • Oppose per Cas Liber. Stub tags are at worst benign. They're the literal bottom of the article and if a reader wants to ignore it, they can. I really can't understand how they can be seen as harmful. On the other hand, if they get 1 reader to help expand an article, the encyclopedia benefits greatly at pretty much no cost. They're a clear net positive. Putting them on the talk page may be convenient for us and our internal categorizations, but that's counterproductive for article improvement. Most people don't know about talk pages (seriously, ask your friends the last time they looked at an article talk page), so hiding the stub tags where only insiders will find them is in my mind a net negative. — Wug·a·po·des​ 21:37, 27 June 2020 (UTC)
  • Oppose. Stub templates are a type of maintenance template. Because of their ubiquity, we've decided to put them at the bottom instead of the top of the page. They share all the pros and cons of other maintenance templates. One one hand, we think that – given the dynamic nature of Wikipedia – we should alert readers and editors if something is not right with an article. On the other hand, we don't know if more time is spent tagging or actually fixing articles. The main problem I see with stub templates is what others have pointed above. We have two systems to mark an artcle as a stub: stub templates and talk page banners. These are not always in synch. When this happens, the fault is with the editor who (de)stubs an article using only one method. The guideline at WP:DESTUB says to do it using both. I'm sure we're moving toward automatic article assesment (WP:ORES) at some pace. In the meantime, we should automatize getting rid of the discrepancies between stub templates and talk page assesments using a bot. – Finnusertop (talk ⋅ contribs) 23:20, 27 June 2020 (UTC)
    "we don't know if more time is spent tagging or actually fixing articles" Probably the first one, in my considered opinion. RandomCanadian (talk / contribs) 03:43, 5 July 2020 (UTC)
  • Support A good solution, as already proposed by others, would be for categorisation via wikiproject banners on the talk page. RandomCanadian (talk / contribs) 03:43, 5 July 2020 (UTC) Striking sockpuppet.P-K3 (talk) 22:47, 8 July 2020 (UTC)
  • Oppose I'm not joking in saying that this is normally how I find GAs to write. Stub tags are invaluable. TonyBallioni (talk) 02:17, 6 July 2020 (UTC)
  • Oppose I hate creating articles (for some reason), I don't think I've created an article yet. I do help massively improve existing articles, and have plans to edit a couple of stub articles and make them more full. Stubs have their benefits, help find topics that may look interesting, and give one a chance to expand them. Also, what Tony said above. ProcrastinatingReader (talk) 12:23, 6 July 2020 (UTC)
  • Comments: Leaning toward oppose, as the WikiProject placement on Talk pages make them less obvious to readers and editors, to the point that I tend to forget about them. I would also to interested to know:
    • Do all stub categories have associated WikiProjects?
    • Is if there an easy way to get a list of articles pages (as opposed to the talk pages; c.f. Category:Stub-Class video game articles to Category:Video game stubs) from the WikiProject categories?
    • Could WikiProject stubs be organized into subcategories like the stub templates?
  • —Ost (talk) 22:27, 9 July 2020 (UTC)
  • Weak Conditional support. As per @Rehman above. However, few editors may still use those stub templates. If they were at the top, maybe it would be diffwerent, but I say drop them and go with what Rehman said above. --Guitarist28 (talk) 14:32, 14 July 2020 (UTC)
  • Oppose removal. I think the format of the tag should be changed, though. I am coming from the perspective of a wikiHowian, where stub tags are placed on top of stub articles. I think it would be more useful if the stub tag was in the form of other maintenance templates. Something like:

    That way, readers see what they can do to help Wikipedia. By having them at the top, readers can work on expanding the article, and then remove the stub tag once the article provides more coverage. Aasim 08:06, 18 July 2020 (UTC)
    Awesome Aasim, I like the idea of a more clear tag, although it should be ambiguous that it isn't a major issue with the article itself, and should actually encourage people to build the page, like:
    Ed6767 talk! 00:21, 21 July 2020 (UTC)
  • Comment One badly needed step would be to go thru all of these stubs & verify that they are, in fact, stubs. In my own browsing thru articles I've found that a quarter should be re-graded to "Start" or better. And about 1 in 10 of the "Start" class should be regraded as "C" or better. (Since my experience is at odds with Johnbod above, maybe my evaluation is more strict than his?) In any case, I figure re-evaluating as many stubs as possible would help reduce the number of stubs in Wikipedia to less than half the total number. --llywrch (talk) 21:52, 20 July 2020 (UTC)
    There are two types of articles here: (1) articles that have already been re-graded as start or higher, but still have Category:Stub message templates on them; and (2) articles that are classed as stubs, but really are start or higher. Hawkeye7 (discuss) 22:57, 20 July 2020 (UTC)
    Possibly, or we're looking at different articles. I feel most assessors pretty much only take length into account, whereas for some subjects a few lines might be C or even higher. The official definitions for the classes are actually pretty generous, & I think most assessors are rather too strict (as well as judging mainly on length). Johnbod (talk) 00:08, 21 July 2020 (UTC)
    Responding to both comments together. Hawkeye7 the first case could easily be addressed with a bot -- & probably needs to be done in any case -- while the second describes the phenomenon I was talking about. Johnbod I suspect we might be looking at different articles, or groups of articles. When calling an article a "stub", I look more to how completely the article covers the topic than the length -- there are some notable historical personages about whom all we can write is limited to 2 or 3 sentences, which means we are stuck with what I call a "permastub" -- although if an article is, say, 5,000 bytes in size & marked as a "stub", I'm going to look hard at why it has that assessment instead of a "start" or "C" class. And sometimes it requires expert knowledge to know what information is available about a given topic, in order to know if the article completely covers it. -- llywrch (talk) 16:55, 21 July 2020 (UTC)
    Ah, perhaps different standards then - I stand with User:Grutness/Croughton-London rule of stubs, and indeed the official definition, though of course "the breadth of coverage expected from an encyclopedia" (which stubs lack) is wonderfully vague. But most articles in most print encyclopedias would be called "stubs" by most WP assessors imo. Johnbod (talk) 17:23, 21 July 2020 (UTC)
  • While I support the idea, there are a few things we should consider:
    1. The public. Stub came to have meaning outside the editing community: The list gradually changing colour on Haskell's screen represented hundreds of women scientists who have either never had a Wikipedia entry, or whose lives and work are dismissed in a stub a few lines long.
    2. I'd favor something more descriptive, like {{Television needs production section}}. Stub tagging is kinda lazy. (guilty as charged)
    3. I very much support an experiment. I would suggest we pick a number of articles. For half of them a hash is created and only the hash is published. For the other half you remove or hide the stub tag, but don't publish a list. Sure one could look at the contributions of the bot that does it, but we can't control that. We'll have to wait until a decent number of articles on the list has been improved beyond stub status, at which point we'd publish the list of still-stubbed articles and compare.
    4. As BD2412 suggested, Template magic. - Alexis Jazz 01:16, 21 July 2020 (UTC)
  • Oppose the editors at College Football Project often use the stub tag to focus editing efforts and assist with improving articles.--Paul McDonald (talk) 13:20, 24 July 2020 (UTC)
  • Oppose removing the stub templates, they do have a function. We possibly also need to:
    1. add an option for notes to the stub template
    2. explore a way to merge the stub flagging and the assessment systems
    3. create a bot task to find stub articles that have grown / had n edits / had a change of assessment. — GhostInTheMachine talk to me 13:05, 27 July 2020 (UTC)
  • Comment - I have edited hundreds of music stub articles, mostly by adding RS to them. I find stubs useful for sorting, at least in this specific regard. Caro7200 (talk) 15:55, 29 July 2020 (UTC)
  • Support completely rethinking what stub tags are and what they are supposed to achieve, and whether they do so. As I see it, there are three things we're trying to achieve: (i) tell people that the article isn't complete (ii) ask them to contribute and (iii) we sort the stub tags so people can concentrate on stubs in an area they're interested in. Point (i) is obvious without the tag, yet it often stay s on the article long after ceases being useful. Given how long some articles remain stubs, we are obviously not doing a great job with (ii). Stub sorting has been partially superseded by wikiproject tagging and a generally much improved category system. I must admit I like the fact that stub sorters are there (tagging an article as "stub" is a quick way to get an extra pair of eyes, as it summons a stub sorter within a few hours at most). But I haven't seen any recent evidence that tagging and categorizing stubs actually leads to much improvement at all, or that the duplication of stub/non-stub tagging on article plus talk page does anything useful. Having an easy way to display article quality (not just stub/not stub status) in any category or for any link in a list would be far more helpful. Not just stubs need improving. —Kusma (t·c) 22:47, 1 August 2020 (UTC)
  • Comment I find it interesting that no one has mentioned Wikipedia: WikiProject Stub sorting, whose members might have something to contribute to this discussion (I also don't recognize any of the commenters here from the project). One of our members apparently just found this discussion yesterday (August 1) and posted a note on the WPSS talk page. I'll put my 2p in later when I have a chance to read through all the entries here. Cheers, Her Pegship (?) 20:32, 2 August 2020 (UTC)
  • Oppose: I personally bookmark several stub categories relevant to my interests and periodically expand them, contradicting the claim that stub tags "don't get people to edit stub articles". I think the only actual problem here is articles that are no longer stubs, still being tagged as such. Sometimes I'm not sure when I'm improving articles very incrementally at what point they have stopped being stubs. I would support a bot to remove outdated stub tags based on talk page assessments, or on a length barrier beyond which an article is clearly not a stub. ~ oulfis 🌸(talk) 01:11, 3 August 2020 (UTC)
  • Support Stubs are messy and the visual clutter is immense for the reader.Epelerenon (talk) 05:25, 3 August 2020 (UTC)
  • Support The stub tags are a relic of an earlier era. Now that all articles get quality ratings from multiple projects on the talk page, having them also noted as stubs on the main article page is redundant. Time spent adding tags and finding the right categories, and then the time removing them, is time wasted. CaptainEek Edits Ho Cap'n!⚓ 06:57, 3 August 2020 (UTC)
  • Oppose. While WikiProjects cover many topics, not all stub articles fall under the purview of a WikiProject, and not all the projects use assessment tags. I have no opinion on project assessment tags, but stub tags/types/cats are not only applicable and useful across the entire encyclopedia, but also a way to gauge which topics need maintenance of other kinds. (I do believe that the parameters of what constitutes a stub article should be revisited.) Last but certainly not least, the editors of Wikipedia:WikiProject Stub sorting have been faithfully maintaining, sorting, and standardizing stub types for over 15 years (that I know of), and eliminating the fruits of that work across the encyclopedia would be counterproductive. Her Pegship (?) 17:36, 3 August 2020 (UTC)
  • Oppose: I'm frequently disgruntled by tags and tag-bombing. But I find the stub tags the least-intrusive and one of the most valuable of all the tags. Bot jobs of updating either the article or the talk page when one is changed would be valuable, but the tags do help editors find stubs in a category they're interested in. The stub categories and WikiProjects don't quite align, as I'm not the first to point out, and I think this is a feature not a bug. — Bilorv (talk) 23:13, 3 August 2020 (UTC)
    Inconsistency between article page tags and talk page project tags is certainly not a "feature" - it is purely the result of carelessness/sloppiness/ignorance on the part of editors. The stub definition is general - other quality ratings may vary with the relevance of an article to a particular project, but an article should either be or not be a stub for everyone. What would be the benefit of such a "feature"? Johnbod (talk) 01:28, 4 August 2020 (UTC)
    I believe what Bilorv means by "stub categories and WikiProjects don't quite align" is not inconsistency between ratings in the stub tags and the WikiProject assessments (ie the same article being called a stub one place and start-class another, which is a "bug"), but differences between stub categories and extant WikiProjects (ie an article being tagged as '18thC-novel-stub' even though there's no wikiproject for 18thC novels). I agree with Bilorv that this second kind of misalignment is a feature and not a bug, as it is useful to have stub categories be specific, whereas wikiprojects need to be rather broad. For the Great Britain and Ireland Destubathon, for example, it was extremely useful to have stubs categorized by quite specific geographic regions, even though it would be nonsense for a whole Shropshire wikiproject to exist. ~ oulfis 🌸(talk) 19:51, 4 August 2020 (UTC)
    Thank you Oulfis, this is exactly what I meant, and apologies for the confusion. — Bilorv (talk) 20:47, 4 August 2020 (UTC)
    Ok, thanks for the clarification. Johnbod (talk) 21:07, 4 August 2020 (UTC)
  • Support. The stub tags are a subset of the project rating system (and predate projects). Several people have stated that they use the stub tags to find work to do. That's great, but they could do exactly the same thing by bookmarking the applicable project stub categories, such as Category:Stub-Class animal articles, or even Category:Stub-Class articles. As a software guy, I believe in DRY, and now that we have project ratings, stub tags violate that. -- RoySmith (talk) 01:55, 4 August 2020 (UTC)
    • The reference to DRY is amusingly misdirected. In my experience, articles don't usually get multiple stub tags but one often find a great proliferation of project tags on the talk page. and so I often use {{WikiProjectBannerShell}} to cut down on the clutter. For example, I recently updated Stonehenge which has 11 different project templates ranging from WikiProject Alternative Views to WikiProject World Heritage Sites – a project that is explicitly inactive. The various project ratings are inconsistent and there are other independent assessments too such as a Vital rating and GA review. If you want to cut down on repetitive clutter then the place to start is the talk page. Why, for example, is there not a single template for projects, which lists all the relevant projects in one combined list? Why does each project have to have its own separate template? It's not invented here, I suppose. Andrew🐉(talk) 09:59, 4 August 2020 (UTC)
  • Comment: "all articles get quality ratings from multiple projects on the talk page"? No. They don't. I sort a lot of stubs, and frequently create the talk page and add project banners. Many readers don't know much about projects, they don't belong to one but just edit their own chosen subset of articles, or perform their favourite subset of Wikignoming. If the stub categories were abolished in the belief that projects do it all, we would need a maintenance category such as "Pages (other than dab pages or redirects) whose talk page has no project banner" - and even then the pages which have {{WP Bio}} but nothing more specific would end up lost. PamD 19:02, 4 August 2020 (UTC)
  • Support - stub tags are just visual clutter. You can usually get the same information from WikiProject categories. Maintaining the same information in multiple places is a waste of time and effort (and screen space). Kaldari (talk) 20:35, 4 August 2020 (UTC)
  • Support. No longer necessary. We presumably have quite a lot of left-over functionality that's no longer needed, and this would be a good start in clearing them up. DGG ( talk ) 00:26, 5 August 2020 (UTC)
  • Strong Support - Useless garbage. Schierbecker (talk) 14:22, 5 August 2020 (UTC)
  • Oppose - The stub tags are out of the way, do not interfere with reading the article and help in categorizing and drawing attention to articles in need of more content. Removing them would be counterproductive. A better use of our time would be a bot that removes stub tags from articles that are over a certain prose size in length. Ciridae (talk) 16:30, 5 August 2020 (UTC)
  • Support. The sole potential benefit is ease of searching up very short articles to expand, and that can already be done with Wikiproject ratings. Meanwhile, I doubt readers in 2020 need stub tags anymore. SnowFire (talk) 21:39, 5 August 2020 (UTC)
  • Oppose. The stub tags are linked into other portions of the encyclopedia, and work with the Wikiproject ratings, so removing them would cause a whole set of other problems. Frankly, I like the stub tags, they definitely have encouraged some readers to positively contribute to the project. In an era when Wikipedia is losing new contributors, shouldn't we be trying to attract more? It's also not like they are taking up space on servers, and don't clutter articles as nearly as much as the issue tags at the top. Modifying those seems to be a better way to achieve what the original poster wants. User:Heyoostorm_talk! 16:26, 6 August 2020 (UTC)
  • Support — My opinion is the stubs are graffiti and serve no useful purpose. —¿philoserf? (talk) 08:07, 8 August 2020 (UTC)
  • I think I fall on the oppose side of the s-o spectrum. I don't think it makes sense to rely on WP ratings (as noted above, many pages end up with stub tags but no WP tags). For utility, I do tend to agree that the "find short articles that could use expansion" side is fairly convincing from a utility-aesthetics spectrum. I do find it a bit obnoxious that sometimes these don't match the WP ratings, but that's a job for a bot (as noted above). If I have any pain point with stub tags, it's that they basically duplicate our category structure (whether on the article side or the WP side), and could use a much smaller tree from which to select a stub tag. (I might even suggest that we should remove all sub-stub tags and put the weight on proper categories and proper WP tags; editors interested in stubs for the categories of interest can use one of our intersection tools such as WP:PetScan.) From a visibility perspective, I actually kind of like the idea of going full {{ambox}} on them.

    On a tangential note, I'm kind of saddened that so many of the calls to action are so often attempted to be hidden on the talk page, or converted to hidden categories. We want new editors and maybe the best way to get them is certainly not going to be found hidden away. (Never mind all the other ways the world has changed in two decades, like the now-60% of readers on mobile.) --Izno (talk) 22:58, 9 August 2020 (UTC)

  • Support. Mandatory reading of Banner blindness as to why those things are useless for the average reader. Popo le Chien throw a bone 18:44, 10 August 2020 (UTC)
  • Oppose deprecating stub templates. The visible part of the template is for the reader, not the editors. It sends the message that Wikipedia is not finished. It also soothes the reaction "WTH, is this all you have on the topic?" by saying "yes, we know this is incomplete". Banner blindness may apply to frequent readers, but we should also consider more casual readers. Compare {{User page}} and {{User sandbox}} which are also aimed at informing the uninitiated. Even though the call to action probably doesn't result in immediate action, it reinforces the "anybody can edit" idea. Pelagic ( messages ) – (17:23 Tue 11, AEST) 07:23, 11 August 2020 (UTC)
  • Comments. Talk-page templates are suppressed on mobile, whilst stub tags aren't (though you may have to expand References or External Links, whatever the last section is). Also, it's not immediately obvious that the talk page should be where one looks for other article meta-data like quality rating. Pelagic ( messages ) – (17:23 Tue 11, AEST) 07:23, 11 August 2020 (UTC)
  • Questions. If Encyclopædius and Rosiestep couldn't get consensus for a bot that directly updates stub and quality tags, how about a bot that adds them to a category for review, like Category:Not-stub stubs? Or some other interactive query? How resource-expensive is it to run an article through ORES, is the score calculated automatically on save/publish and stored with the revision, or would a bot that compares ORES scores to stub tags be costly to run? Pelagic ( messages ) – (17:23 Tue 11, AEST) 07:23, 11 August 2020 (UTC)
There is Wikipedia:Database reports/Long stubs, which has been updated recently. Don't know if there's a bot that can check the entries, or if it can only be done by human. Her Pegship (?) 17:36, 12 August 2020 (UTC)
  • Support per DGG. I agree that they have outlived their usefulness. They also are a pit of make-work for gnome editors (stub sorting) who would otherwise be doing more useful things. Calliopejen1 (talk) 00:16, 16 August 2020 (UTC)
For some of us gnomes, stub sorting is the only aspect of editing WP we feel confident about. Not a reason to keep stubs...just saying. ;P Her Pegship (?) 00:52, 16 August 2020 (UTC)
  • Support There are several articles on Wikipedia, which are long enough not be a stub article, but still classified as stub. Go to https://en.wikipedia.org/wiki/Wikipedia:Database_reports/Long_stubs While some small articles aren't classified as stub, which reduces the value of Stub tags. In less widely used Wikipedias like the Hindi Wikipedia and Bengali Wikipedia, the problem is even more pronounced, where most of the smaller articles (some with less than 100 words) are not classified as Stub, whereas some previously Stub articles, which have since been improved, still contains the stub tags. Soukarya (talk) 13:07, 18 August 2020 (UTC)
    What the Hindi and Bengali Wikipedias do with stub tags is their own business. Maybe they have rules that don't just concern article length - such as the amount of referencing, or the quality of the writing. Or maybe the person who creates these sub-100-word articles simply doesn't know about stub tags. Whatever the reason, it's nothing to do with us; we cannot tell them how to handle stub tags on their Wikipedias. --Redrose64 🌹 (talk) 15:14, 18 August 2020 (UTC)
    Yes, not many editors know how to add or remove Stub tags, what criteria must be fulfilled in order to classify some article as Stub or remove the Stub tag, which is the reason why I support the removal of Stub tags altogether. After all, the Stub tag in English Wikipedia can be very misleading sometimes (please go through this weblink https://en.wikipedia.org/wiki/Wikipedia:Database_reports/Long_stubs, which I had already mentioned above, to view the long, well-referenced articles which are still tagged as Stubs, and most Stub tags in popular Indian vernacular language (like Hindi & Bengali) Wikipedias are all useless, and in articles where it is required, editors do not put it there. An example of sub-100-word Stub article in Hindi Wikipedia is this, https://hi.m.wikipedia.org/wiki/%E0%A4%A7%E0%A5%8D%E0%A4%B0%E0%A5%81%E0%A4%B5_%E0%A4%B0%E0%A4%BE%E0%A4%A0%E0%A5%80 which is about a popular Indian youtuber. Soukarya (talk) 15:34, 18 August 2020 (UTC)
    Hindi Wikipedia is nothing to do with us. If you want to change their practices, do so at their discussion forums, not here. --Redrose64 🌹 (talk) 17:00, 18 August 2020 (UTC)
  • No? Someone can correct me if I'm mistaken, but I believe more than half the articles on the project are stubs? So... We're going to get rid of hundreds of templates and make edits to millions of articles because?...There is a general feeling of indifference to their usefulness? Sorry. That's not a project you undertake when the main argument is "meh". If you don't like em, don't use em. No one is going to drag you to ANI because you're just not particularly into stub sorting. GMGtalk 13:34, 18 August 2020 (UTC)
  • Oppose, solution looking for a problem. Stifle (talk) 16:05, 21 August 2020 (UTC)
  • Oppose. I'm unconvinced by the arguments for. I might be convinced that the stub information should not be visible on the article page, since readers aren't going to care -- they can already tell it's a short article, without needing us to help them out. I'd also be OK with some of the other suggestions made, such as making the stub tag invisible if a WikiProject has assessed the article above stub on the talk page, or hiding the tag above a certain article size. But getting rid of them is throwing out the baby with the bathwater; there are legitimate uses for stubs which outweigh the minor benefits discussed above. Mike Christie (talk - contribs - library) 16:20, 21 August 2020 (UTC)
  • Oppose Having more than one way to group and access stub articles to work on is totally sensible, so this is a daft proposal. Removing them would only weaken the Project. That said 'stub' categorisation on an article page, whilst unnoticeable to the vast majority of readers, really does need to match with Talk page WikiProject assessments. I confess to often amending the latter, but forgetting to remove the former. If there isn't agreement for a bot to do the work of removing outdated stub tags, then at least a tool to identify the mismatch between talk pages might be valuable for some editors. Personally, what I find frustrating is to have to individually change three or four 'stub' assessments on Talk pages - one for each WikiProject, when that assessment is going to be the same across all of them. Quality assessments could really doing with being integrated so that only one needs to be changed, leaving just the 'importance' assessment to be specific to each individual WikiProject. Just thought I'd throw that out there. Nick Moyes (talk) 19:18, 21 August 2020 (UTC)  
Although in most cases they should be the same, & it is indeed a pain to have to change several, there are many cases where it is wholly appropriate to have different ratings for different projects, especially where a project only applies to a part of the subject. Obviously that is much more often the case for importance ratings, so you might well need to edit every line anyway. Johnbod (talk) 20:33, 21 August 2020 (UTC)
  • Oppose - Stubs still have their uses and as noted above they don't interfere with readers as they're directly at the bottom of articles, They're harmless and IMHO I see no valid reason to remove or deprecate them. — Preceding unsigned comment added by Davey2010 (talk • contribs)
  • Oppose. It is a long-standing policy that Wikipedia is a work in progress, and stub tags are one of the most visible reflections of that. However, there has been a trend towards intolerance of stubs, and stubs that would have been perfectly acceptable in 2006 are now routinely draftified without discussion. Stub tags serve as a reminder that stubs are a natural part of Wikipedia, and I'm concerned that removing them would further drive that intolerance. --Paul_012 (talk) 09:57, 28 August 2020 (UTC)
  • Oppose - As a user of User:Anomie/linkclassifier.js the stub tags, and the corresponding categories, are what cause these links to appear a different colour as I am browsing Wikipedia, in turn I use this to find articles that I might be interested in improving. I would not be opposed to a redesign though.CSJJ104 (talk) 14:54, 30 August 2020 (UTC)
  • Oppose – it's simply an invitation to expand the article, which is what wikipedia needs really. Removing stub notes is hiding the embarrassment that much of wikipedia has – inadequate research & topic coverage. – ishwar  (speak) 05:05, 31 August 2020 (UTC)
  • Oppose – I agree with earlier comments that stubs are useful and help encourage readers to consider expanding the article. I've also seen too many perfectly acceptable stubs get AFD'd instead of someone taking the initiative to expand them. I really like Awesome Aasim and Ed6767's suggestion above for a "friendly" maintenance tag that encourages readers to expand the article. Carter (talk) 00:29, 1 September 2020 (UTC)
  • Oppose – for five reasons: (1) I think they are useful for new editors. All that "Be Bold" stuff? I was terrified of making edits when I first signed up. I thought anything I did would get me called in front of the Wikipedia Star Chamber. But then I started seeing stubs notices on articles I was reading, actually inviting me to try to improve them. That's what got me started making (cautious) edits, to stub pages that I knew a bit about. Stub notes are invitations to new editors: "We really mean it — try to improve this article — give it a go!" (2) They're unobtrusive. Right at the bottom of the page, they don't interfere in reading the article at all. If you read that far, you get the acknowledgement that this article could be better, and the invitation to try to fix it. (3) By contrast, adding big hatnotes at the top of a page, as suggested upthread, is saying "This article is shite, you shouldn't rely on it, but feel free to try to make it better, if you can." It isn't really much of an inducement to read the article. (4) There's Talk pages? Who knew? Why wasn't I told? (5) As I've learnt more about stubs and the intricate connection with the categories, I find them very useful to see the inter-connections between different articles within the same topic, and particularly ones that need work. --Mr Serjeant Buzfuz (talk) 03:36, 1 September 2020 (UTC)
  • Oppose – this is a huge issue and should be reviewed with its long-term effects in mind. While stub tags may not be directly effective in expanding stub articles, I think that stub tags are highly useful for Wikipedia, because they create a sense to every reader that "People like you can join Wikipedia". This is a vital part of how some readers become editors, and I'd say a good percentage of editors wouldn't have joined if it weren't for stub tags or other similar tags indicating how they could help; they wouldn't have known you can edit Wikipedia. If it weren't for stub tags Wikipedia would feel more like a book than an open community, and if it didn't have this inviting feel, if the number of editors in Wikipedia reduced, then the quality of articles in general would fall. I decided to oppose instead of to wait for the survey because I think that a survey couldn't show the effects I have in mind, as they would only be felt if a huge percentage of articles were used. While it's true that stubs may remain in articles after their class has increased, this is generally not an issue for the reader, especially since in larger articles stub tags are hard to realise. They only catch your attention if you're actually looking at a stub, when they are some of the few chinks of text in that article. While destubifying may be a hustle for editors, it is a necessity which arises from a necessary part of Wikipedia's structure.

KnolGua (talk) 09:53, 2 September 2020 (UTC)

  • I have always thought that they were pointless. Support. Foxnpichu (talk) 14:43, 3 September 2020 (UTC)
  • Support. Stub tags create the impression something's wrong with the article, although it may explain the topic sufficiently for most readers. I agree with commenters on the other side that a friendly positive invitation to readers to expand or improve an article would be fine. -- econterms (talk) 01:02, 10 September 2020 (UTC)
  • I think the problem is less to do with stub tags at end of articles than with WikiProject ratings of articles on the articles' talk pages. If one goes to the article Escape to the Country, and looks on its talk page, one will see that is of interest to WikiProject BBC, but is only rated as stub class by this WikiProject. If one reads the article, one will see it is too long to be counted as a stub. I think the problem may be less to do with stub tags than how active WikiProjects are. Vorbee (talk) 15:02, 11 September 2020 (UTC)
    • Yes, the problem has to do with WikiProjects and how active they are. But, we should still get rid of stub tags, because finding and identifing the non-stubs that have stub tags is an annoying job. Especially throughout one of the largest websites in history. It's endless. And, with WikiProjects going inactive, comes alot of stub tags on expanded pages. Arsonxists (talk) 15:18, 11 September 2020 (UTC)
  • Support To be honest, I never got the Stub rating. Half of Wikipedia articles are Stubs, so that means alot of Stub tags. Also, Removing Stub tags won't do anything to WikiProject ratings, which still have Start, C, B, and A ratings. Arsonxists (talk) 15:47, 11 September 2020 (UTC)
  • Oppose. Editors find them useful, in such a way that cannot be replaced by WikiProjects (for example, 18th century novel stubs tag is useful as there is no WikiProject for those novel articles). In addition, they are probably the biggest invitation to edit Wikipedia on articles. I probably wouldn't be here without stub tags. A redesign may be a good idea. --Danre98(talk^contribs) 03:09, 14 September 2020 (UTC)
  • Support I have manually removed stub templates from thousands of articles; they stay around for far too long usually. Also too many articles with multiple stub templates. (as a side-note is there any one-click solution for removing all stub templates from an article?) --WS (talk) 13:10, 14 September 2020 (UTC)
  • Oppose per Mr Serjeant Buzfuz et al, but let's keep talking about ways to improve the template and possibly integrate better with article ratings. Retswerb (talk) 03:50, 15 September 2020 (UTC)
  • To avoid confusion, I point out that the proposalisto get rid of stub tags, not to remove articles which are so short as to be stubs. I dont think anyone is proposing that. DGG ( talk ) 22:32, 18 September 2020 (UTC)

Designating current seasons in infoboxes

Moved from Template talk:Infobox award § Using image for current award rather than text
 – {{u|Sdkb}} talk 23:12, 16 September 2020 (UTC)

Looking at {{Infobox award}} as linked from e.g. Pulitzer Prize, it is not immediately clear that the globe and timer indicate the most recent year in which an award was given. It presents possible accessibility problems, and also causes confusion for editors because of the image's frequent use in the {{Current}} maintenance template, leading me to initially believe that there might be something in need of update. I propose that we switch to instead use "Current:", as I've demonstrated in the sandbox (see testcase here). Is that alright? Update: per below, I am expanding this proposal to include analogous changes at other infoboxes with similar circumstances. {{u|Sdkb}} talk 18:32, 16 September 2020 (UTC)Minor refactoring done to retain context during move.

Notified: Wikipedia talk:WikiProject Awards. {{u|Sdkb}} talk 18:35, 16 September 2020 (UTC)
Just as a note, {{Infobox football league}} and a dozen other sports seasons templates use this or similar versions of the "clock" image. I would suggest that the discussion get a little wider, maybe at one of the Village Pumps, to get consensus about the use of these images in general. Primefac (talk) 19:04, 16 September 2020 (UTC)
Primefac, thanks for pointing that out! I'll move this to VPR, since I think similar considerations would apply to all these templates. If anyone wants to list out the templates or propose specific alternative wording for some of the others (e.g. should it be Current season for {{Infobox football league}}?), that would be helpful. {{u|Sdkb}} talk 23:06, 16 September 2020 (UTC)
Support - Is that what that image means? I had no clue, and I often wondered what the heck it was. Clicking on the image simply takes you to the file itself, still providing no context. Even the infobox is edited, it's not very clear, and the template documentation doesn't really describe that an image is going to be used—it takes some serious connecting of the dots. At least in the {{Current}} template, the verbiage provides an explanation. If the discussion is widened, I'd have to see other cases, but with {{Infobox football league}} at least, I still don't find the image particularly helpful when I look at Premier League, but at least the football league template documentation uses the image, which is a step up from the Awards infobox. -2pou (talk) 19:31, 16 September 2020 (UTC)
Perhaps the word Current could also be in italics? I don't know if the rest agree. I've implemented it and in my opinion it looks better. El Millo (talk) 00:06, 17 September 2020 (UTC)
Support - I do not believe I have ever seen this before, but its meaning to me as a sighted user is fully opaque and therefore worth exactly zero. For unsighted users the usefulness is surely less than that (more confusion/frustration trying to use our site). If people don't like the word Current:, it seems we could drop it altogether: no image, no word; just the link to whatever current thingy would be there. — JohnFromPinckney (talk) 10:21, 17 September 2020 (UTC)
Support The fact that it took me 10 minutes to figure out what these people were trying to propose really just gives it away that it needs change. Arsonxists (talk) 16:25, 17 September 2020 (UTC)
Support any clarification over the old. I read the proposal thrice and looked at old and new and I still couldn't understand what the thing is supposed to be. I imagine the current version also violates accessibility in all sorts of ways. —  HELLKNOWZ   ▎TALK 22:41, 18 September 2020 (UTC)

Always Show Column Headers in Tables

Tracked in Phabricator
Task T42763

Viewing long tables (both x and y axis) is really annoying, because the column headers are alllll the way at the top, but you're viewing a list of "yes" and "no" values in the fields, so you have no idea what they mean. You would need to memorize all of the table headers before viewing an entry several tens of entries down (such as "US Mobile" in this page's table: https://en.wikipedia.org/wiki/List_of_United_States_mobile_virtual_network_operators )

It would be very nice if the column labels would remain on screen at the top of the page, while scrolling through the table.

In modern web pages it is possible and somewhat common to show the "search" bar at the top of the page even when scrolling through contents, so this technique could easily be used.

Since this behavior could pose an issue on small screens with column headers that are very tall, this option should be implemented as a toggle button at the top left corner (first cell) of the table. — Preceding unsigned comment added by CambridgeBayWeather (talk • contribs) 23:56, 15 September 2020 (UTC)

I don't know if that placement is the best, but potentially fixable col headers would be a welcome option (esp. if off by default, e.g., for smallish tables). — JohnFromPinckney (talk) 01:53, 16 September 2020 (UTC)
Maybe the point where it's enabled by default is 15 sections down, and 5 columns. A limit like that would organize the tables even better by eliminating the tables that don't need it and having it for the ones that don't. Arsonxists (talk) 02:02, 16 September 2020 (UTC)

Issues raised by Citation bot

The following discussion is an archived record of a request for comment. Please do not modify it. No further edits should be made to this discussion. A summary of the conclusions reached follows.
While there were three distinct questions, there was enough interplay between them that I am closing this based on the consensus of the entire discussion rather than merely question by question.


Whether to link
While some wished to do so only in certain circumstances (e.g. if there was a free version), there is consensus for including a link, when an online source is available, from the title text. There is consensus that removing a link requires human judgement.


What to link
There is consensus that what to link should be accorded respect when made by humans (see more in the following section). When deciding what to link in the title, the consensus is that sources which verify the information being cited and sources which are free are top priorities; a lesser factor to consider is whether it is a link to full text or partial (e.g. abstract only). At no time should we be linking to copyright violations.


Duplicate parameters and when to remove
There is consensus against removing a parameter just because it is a duplicate. However, there is an agreement that sometimes removing duplicates can be appropriate. This decision should be made on an article by article basis, unless it is a copyright violation for which there is consensus to remove. Questions that can help guide a discussion include:

  1. Does the functional impact, as currently experienced by readers, of the parameter exactly duplicate another parameter?
  2. Does including a duplicate link help to establish intent of the person adding the citation by helping them to say where they got it or otherwise enhance verifiability in some way?
  3. Is the parameter free or behind a paywall?
  4. Is the link dead? If yes does it make sense to rename it to a different parameter in keeping with question 2?
  5. What will best respect WP:CITEVAR?

These questions should be asked in concert with the consensus around whether and what to link.

Best, Barkeep49 (talk) 04:42, 19 September 2020 (UTC)

Many of you will note that the erstwhile User:Citation bot was blocked about two months ago. After copious discussion, the block has revealed some underlying issues about how we deal with citations, that seem to lack community consensus. The initial block was over Citation bot unlinking sources in reference titles if it was already linked to something like a DOI, or other unique identifier. I thank User:Levivich for the wording of the questions, whose text I snatched verbatim from WP:BOTN :) CaptainEek Edits Ho Cap'n!⚓ 20:39, 5 August 2020 (UTC)

I object to the nature of these questions. Each one muddles a principle:
  1. Should the titles in citations be linked, and if so, to what?
  2. Should the titles in citations be linked if that link is a duplicate of an identifier?
  3. Under what circumstances should a title link be removed from a citation, by human or by bot?
with its implementation:
  1. Should |url= be the means of linking to a source?
  2. Should |url= be allowed to duplicate another parameter (or vice-versa)?
  3. Should bots be allowed to make the decision on whether a title link is redundant?
--RexxS (talk) 14:21, 5 August 2020 (UTC)
@RexxS: You are welcome to ask those questions below, though it will make closing all of this quite interesting. I asked these because Levivich seemed to have done a good job summarizing the issues, and nobody objected to them in the last two weeks that they languished at BOTN. Again, I figured that someone needed to get the ball rolling, and that person has consistently been me. CaptainEek Edits Ho Cap'n!⚓ 18:46, 5 August 2020 (UTC)
@CaptainEek: An "interesting" close is usually a consequence of not asking straightforward questions. Levivich certainly did a good job of giving an overview of the multiple issues that needed to be decided, but that does not necessarily make a good choice for RfA questions, which should be as clear and focused as possible. I made this point in the prior debate. I'm grateful for you getting the ball rolling, but I still fear that it now be difficult to pick apart editors' opinions on the broad principles from debate on how those might be implemented. It's all too easy to divert attention from the "issues raised by Citation bot" by focusing on the coding of CS1/2 citations, which is a complete red-herring. --RexxS (talk) 21:25, 5 August 2020 (UTC)
  • Comment – there are various ways to link titles in cite templates, that is, apart from |url=,
    • |chapter-url= – for chapter titles
    • |title-link= and, alternatively, a wiki-link of the title
    • |series-link=, for titles of series
    • etc
I assume that these other methods of linking, and not exclusively |url=, are covered by this RfC too. --Francis Schonken (talk) 05:23, 24 August 2020 (UTC)
@Francis Schonken: your comments encouraging specific other people who have expressed opinions elsewhere to bring them here look like WP:CANVASSing to me. —David Eppstein (talk) 06:46, 24 August 2020 (UTC)
No intention to canvass specific people, just trying to get the discussion in one place, as I have done here too. Anyhow pinged the other participant in that talk discussion too, to avoid a perception of unequal treatment. Note that that other participant in that talk page section, i.e. Trappist the monk, had already been pinged here (by Headbomb), without anyone considering that, apparently, selective pinging inviting to participate here. --Francis Schonken (talk) 07:04, 24 August 2020 (UTC)
I was 100% unaware of this discussion until now, but that it quite possibly my own incompetence, I have added a link on the citation bot talk page. AManWithNoPlan (talk) 20:52, 24 August 2020 (UTC)
@AManWithNoPlan: you were notified of this discussion (via ping) on 5 August; and I posted a link to this ongoing RfC on the bot's talk page on 22 August. --Francis Schonken (talk) 04:14, 25 August 2020 (UTC)
I would like to emphasize that despite the fact that I was invited to this discussion, it didn't influence me. I already had unfinished replies laying around in my drafts folder for more than a week and was just lacking the time to finalize them.
As a general remark I would like to see much more constructive and solution-oriented thinking, assumptions of good faith, and less wikilawyering. It's not good for our communication culture and not for the project.
--Matthiaspaul (talk) 13:44, 1 September 2020 (UTC)

Title linking

1. Should the titles in citations be linked, and if so, to what (in other words, what should be in |url=)?

  • The title should be linked to free full versions of records when possible, with paywalled links as a distant 2nd option. That doesn't mean that |url= should be populated, however, if it's redundant to an identifier. In the case of a free full version of record, the identifier should be marked as free, so that the title is automatically linked. If the identifier doesn't link to the full full version of record, it can be provided too, but not in |url=. There's partial implementation for this in CS1, but this should be fully implemented. Headbomb {t · c · p · b} 01:46, 5 August 2020 (UTC)
  • Yes, to a free copy of the source or the closest thing to it. "Click on the title, get to the source" is something every reader knows and expects, and we should provide it. If there is a free (non-copyvio) version of the source, that should be linked under the title. If there isn't, then a copy of the source that provides a free preview, if available (e.g. Google Books or Amazon Books). Or, to an "official" copy of the source, if no free copy is available (e.g., whatever the DOI points to). Or, to a library catalog entry, to help the reader obtain the source (e.g. the Worldcat entry). Whatever we can do to get the reader to the source, that's what should be linked under the title in a citation. Lev!vich 01:47, 5 August 2020 (UTC)
  • Titles should be linked whenever there is free full text, as that is what our readers expect to find in linked titles. Our readers don't necessarily know what a PMC, PMID, DOI, SCID, or any other identifier is, but they know if they see a blue title, they can expect to read free full text. Titles should NOT be linked when they ONLY duplicate an abstract available in a DOI. PMC should continue to automatically link to the title, and editors should be able to manually add free full-text links via the URL parameter that are not copyvios but that are free full text. Titles should NOT link to book listings like at amazon or google books unless free full text is available; if the page is available on line, that can be linked in the page parameter. SandyGeorgia (Talk) 01:49, 5 August 2020 (UTC)
    • How sure are we that readers know blue title = free full text? I've asked several academics who are regular Wikipedia users whether they know why some refs have the titles linked and others do not. None knew. Adrian J. Hunter(talk•contribs) 06:57, 5 August 2020 (UTC)
      • They may not expect it to go to exclusively to the free, full text, but I believe that people do realize that it's meant to take them to the article/source (paywalled or otherwise). WhatamIdoing (talk) 14:59, 5 August 2020 (UTC)
      • How can you be an academic without knowing how to look up what is DOI, HDL, PMID, PMC, ArXiv, ProQuest, ISBN, OCLC etc. That is just mind boggling that someone who is supposed to be good at researching and looking stuff up can just throw up their hands at what link to click based on their institutional subscriptions and access.  — Chris Capoccia 💬 18:37, 6 August 2020 (UTC)
    • there is already a clear way of noting whether something is free text and it's the several icons added with parameters like url-access, doi-access etc. title link does not mean it's free text. it only means someone populated the url parameter with something. it could have even been populated with a broken link.  — Chris Capoccia 💬 14:31, 5 August 2020 (UTC)
  • Yes We link the article title of other sources (newspapers, websites) so no reason why we shouldn't also for journal papers. Readers should not have to follow a blue-underlined random collection of digits and letters (a code). As for only linking to free papers, it may be the only option at the moment to help the reader avoid disappointment that the link provided is not worth clicking. If we had a better way (icon?) to indicate paywalled links (which also applies to newspapers) then this may be less of an issue. -- Colin°Talk 09:27, 5 August 2020 (UTC)
  • Other: This question is confusingly stated. URL should only be populated with something unique that is not duplicate of parameters. The place to complain about parameter access = free not creating a title link is in the CS1. for example, doi-access=free currently creates a title link from the doi. there could be something similar for s2cid-access=free. Users are not stupid. they can click in more places than just the title. It will be perfectly fine for many articles to not have any title links displayed. This is no different than for books. lots of books just have ISBN and users have no problem clicking on that and navigating to how they want to find the book.  — Chris Capoccia 💬 12:07, 5 August 2020 (UTC)
  • Yes per SandyGeorgia. It's fine to do an identifier, but if there is a free version it should be linked in the title whether or not there is an identifier that links to the free version also. --Ealdgyth (talk) 13:40, 5 August 2020 (UTC)
  • Yes The title of a citation is the place where readers are accustomed find a link to the source. Other than a link to a copyright violation, there is no good reason why a link from the title should be removed. The title link should go to the most useful version of the source available. This is a different principle from its implementation (via one or another parameter, which is irrelevant to the principle). --RexxS (talk) 14:14, 5 August 2020 (UTC)
  • Yes, I prefer having links on the title, even if they're "duplicates" and "redundant". I prefer having the most useful URL, but "most useful" depends upon the reader's circumstances, so in the end, any link that makes sense to the editor is okay with me. And before we get too bothered about any of this, let's take a moment to remember that almost nobody reads the refs, and the better the article is, the fewer refs they click on. So let's do something sensible, but let's not spend too much effort trying to achieve perfection, and let's definitely not yell at other editors for putting the "wrong" duplicate and redundant link into the URL. WhatamIdoing (talk) 15:04, 5 August 2020 (UTC)
  • Wrong question. Titles should be linked when the source is a url that is used as a reference. Titles should not be linked when the reference is offline. Titles of references available both offline and online can sometimes be linked as a courtesy to readers. Titles of references that have better online identifiers than urls, such as dois or hdls, should probably not be linked in my opinion, because the link is redundant, but I know others disagree. Titles should not be linked when the link goes to a preliminary version of a publication that is missing the information the reference is used to source. The real answer is "it depends" and that we cannot make a rule for this sort of thing because any rule we make would be broken. In the end there is no substitute for human editorial judgement and intelligence. Trying to replace judgement and intelligence by rules is a mistake. —David Eppstein (talk) 06:38, 6 August 2020 (UTC)
  • Yes, the title should link to the "best" (most useful, free if available, etc.) source if it is online. I don't see an issue with duplicate links: if someone knows/cares about online identifiers and wishes to visit those directly, those links are available for them to specifically click on. Human judgment is going to be necessary in many cases to determine the "best" link, and in those cases a human can put that into the url= field; once that is populated, a bot shouldn't remove it even if it duplicates one of the online identifier links. If there is no manually chosen URL, I'm fine with the title being automatically linked to one of the online identifier URLs if one is available. I agree with WAID above that we really shouldn't overthink/overcomplicate this. CThomas3 (talk) 20:30, 6 August 2020 (UTC)
  • No Wikipedia is an encyclopedia not a search engine. It's the text of articles which is our main output, not the references, which readers pay little attention to. We're already seeing aggressive attempts by bots to insert link-spam; trying to pull readers to their site rather than the competition (e.g. Google Books vs Internet Archive, neither of which wrote or published the books that they are now claim-jumping.) We should not encourage this. Andrew🐉(talk) 11:55, 7 August 2020 (UTC)
  • yes there should be a link if the article can be referenced online, but not always needed. Whether free to access or not does not matter, but we should prefer free where it can be provided. Graeme Bartlett (talk) 12:00, 7 August 2020 (UTC)
  • Yes title link to the actual reference used for the citation. No harm in duplication. · · · Peter Southwood (talk): 14:28, 7 August 2020 (UTC)
  • No — The wikipedia is over-linked as is. A linked title is a link too many. —¿philoserf? (talk) 08:36, 8 August 2020 (UTC)
  • Yes, and to the specific page in question where possible. Adam Cuerden (talk)Has about 7.4% of all FPs 22:39, 9 August 2020 (UTC)
  • Titles should be linked whenever there is any online version of the source available that is not a copyvio. If ther is a free-to-read version that is complete, the link should go to that. if there is no fully free version, but there is a version with part of the source free, such as an abstract, the title should link to that. Failing either of those, the title should link to a fully paywalled version. All of that should be true regardless of whether or not the title link goes to the same place as some identifier link goes. DES (talk)DESiegel Contribs 00:57, 12 August 2020 (UTC)
  • I don't know. For umpteen years I've been linking article names to the source I found it at. An example of my practice can be seen here. Some have been to JSTOR or persee.fr, some to academia.edu, some journals have a website with free copies, such as Bulletin of the American Society of Papyrologists, & some to slightly sketchier sources. (No, none to SciHub, nor do I plan to. At least not yet.) I've always felt this was not entirely right, & sometimes thought setting out a source like JSTOR or doi separately would look more professional; it would match current practice with the ISBN template; of course if I ever submitted an article I largely wrote to the GA or FA process I would have to rewrite it beforehand to use one of these template, so there's that. Lastly, I have to admit that none of the votes above have convinced me to either continue my practice or change it. -- llywrch (talk) 06:36, 12 August 2020 (UTC)
  • Yes, if there is an a URL link inserted in the template, it should be dispalyed in the title, despite all other identifier. Like that, when you click on the title, you access the link inserted by the editors. --Anas1712 (talk) 00:00, 13 August 2020 (UTC)
  • Wrong question ... modules should be used to generate urls when there is a full text version, such as |pmc-access or |doi-access. No, there should not usually be any link to a paywalled source in the URL because it does not help our readership, very few people are willing to pay $$$ in order to check a WP reference. (Unless the content can be verified to a preview or there is no other identifier to uniquely identify the source). (t · c) buidhe 03:29, 13 August 2020 (UTC)
  • Example: I was just reading holocene extinction. I wanted to read one of the sources. The cite ref looks like this:

    Ceballos, Gerardo; Ehrlich, Paul R. (8 June 2018). "The misunderstood sixth mass extinction". Science. 360 (6393): 1080–1081. Bibcode:2018Sci...360.1080C. doi:10.1126/science.aau0191. OCLC 7673137938. PMID 29880679.

    I don't know which one to click on. Turns out none are free, but the second one (doi) will take you to a first-page preview. That second one should be the link for the title, so no one else has to check all four to figure that out. I will add the |url= to this citation. I hope no bot comes along and removes it because it duplicates an identifier. :-) Lev!vich 04:10, 16 August 2020 (UTC)

  • Yes. The title of a citation is the most intuitive place for readers to find a link to the source, and they shouldn't be expected to understand all the citation word salad to find out where the link otherwise might be. If there's no free source available, it should link to the most useful - and which that is should be down to the editor discretion. Boing! said Zebedee (talk) 16:33, 17 August 2020 (UTC)
  • Yes, to the most useful link, and failing that, at least to the WP:SAYWHEREYOUGOTIT link, per current guidance, if that was the source of the content that was added to the article. --Francis Schonken (talk) 04:28, 24 August 2020 (UTC)
  • Yes, but question simplifies too much. Titles should, whenever available, be linked to the most relevant, specific and suitable online resource. While it is nice if the link points to a free resource, being free is not (and never was) a requirement for a link to be provided, however, being relevant in the context of the citation is. By default, the link should point to where the cited source was found by the editor who added a citation in order to satisfy our core principle of verifiability (WP:V). If that happens to be a paywalled link or limited in other ways, this may make verification more difficult for readers, but this is irrelevant in regard to being a proper link in a citation or not. If multiple links are available, it should be in the editor's judgement which link to include, ideally it would be a deep link to the actual document (f.e. a PDF, ZIP, etc.), not only to a meta-page, because the document itself is the most relevant thing in regard to verifiability and the readers' interest who want to study that resource. Ideally, this document should be archived and an |archive-url= link provided as well. Since archived links cannot be given for identifier links, since |archive-url= and |url= should remain in sync, and also because identifier links often point to meta-pages only rather than the actual document supporting a statement, the link associated with the title should be provided by |url= and have priority over secondary links derived from identifier parameters. (Likewise for |chapter-url=.) So, ideally, |url= would point to the actual document, and the identifier links further down in the citation would point to various meta-pages providing additional context (and from which instances of the actual document are typically also available). If, however, the actual document would be already provided by a deep link through one the provided identifiers, |url= would be free to possibly provide an alternative (better qualitity, cheaper, fallback etc.) source for the document (as a convenience link) or to yet another relevant page not covered by the identifier links already. So, what's best in the interest of the readers and the project in regard to verifiability, readability, completeness, stability, availability, etc. actually depends on the circumstances, and it is necessary to apply good common sense on a citation-by-citation basis. --Matthiaspaul (talk) 20:00, 24 August 2020 (UTC)
  • A few people are answering the wrong question - but not all. This discussion is not about removing all title links, just redundant ones. AManWithNoPlan (talk) 11:30, 25 August 2020 (UTC)
    • If you read what people are telling you in this section, you'll realise that there are no redundant title links. Title links, where available, are desirable and your bot should not be removing them. --RexxS (talk) 22:22, 25 August 2020 (UTC)
      • I did read them and I am correct about some. I rufused to point out who since that would be be rude to squash peoples contributions durig a discussion and i trust any counting votes to not be so stupid as to just add up yes and no. AManWithNoPlan (talk) 00:51, 27 August 2020 (UTC)
  • Often Impossible to say where you got it or more correctly, add people and bots add more identifiers, there is no way that a reader can know this. Many citations start with a DOI, and then someone later adds the URL (research gate, handles, etc) or visa-versa. Only by looking over the editing log can you find where the very first reference is in the WP:SAYWHEREYOUGOTIT sense. BUT, that only tells you the first person to read the reference. You have no idea how later editors found it. Secondly, the original editor might have use a publisher URL, and then linked to the pubmed instead because the PMID was in the URL. AManWithNoPlan (talk) 16:14, 28 August 2020 (UTC)
  • Yes, using the title-link parameter, URL for clickable arrow. Titles in citations should be linked to a Wikipedia article about this source using the title-link parameter (e.g. |title-link=The Great Gatsby) and only in this case should the title appear in blue (not a common case). Clicking the arrow after the title should open the website defined in the Url parameter (e.g. |url=https://archive.org/...) showing the title page of the source text. Other unique identifiers may be given such as DOI, ISBN, JSTOR, etc. but always in addition to the URL, as the reader may not understand them. The bot may update or correct the URL but should never remove it. Johannes Schade (talk) 16:12, 31 August 2020 (UTC)

Duplicate linking

2. Should the titles in citations be linked if that link is a duplicate of an identifier (DOI, PMC, PMID, etc.) (in other words, should we have |url= even when it is duplicative of, e.g., |doi=)?

  • No, in the sense that having a |url=https://doi.org/... that duplicates a |doi= is pointless redundancy. This is why we have |doi-access=free, to mark the DOI as free and have them automatically linked, and remove the necessary duplication. Headbomb {t · c · p · b} 01:41, 5 August 2020 (UTC)
  • Yes if and only if the DOI, PMC, or whatever contains freely accessible full text, no if it is only to an abstract or full text is paywalled. SandyGeorgia (Talk) 01:50, 5 August 2020 (UTC)
  • Yes, titles should be linked even if the link is duplicative of an identifier such as DOI, and regardless of whether it's free or paywalled. The reader knows and expects to click on the title to get to the source. The reader will probably not know what the links marked "DOI", "PMC", etc. are. Even if we put lock symbols next to them to identify which ones are free, and links to "DOI" and "PMC", etc., to explain what they are, this is just asking the reader to do more work to get to the source: (1) they have to know what DOI/PMC are or else look it up, (2) if there is more than one, they have to choose which one to click on, (3) if some are free and some are paywalled, they have to learn what the lock symbols mean, and (4) if there is more than one free one, they have to choose which free one to click on. "Click the title, get to the source" is what the reader knows and expects; whether the same thing is linked elsewhere in the citation is irrelevant to the reader. It is we, the editors, who should "pick" which version of the source is the "best" version for the reader, and we should put that as a link under the title. Then the reader just knows to click on that one to get to the best one. And even if it's not free, they'll know that's the place where they can obtain a copy. On top of the reasons to link the title, there are really no reasons AFAIK not to link the title to the best available source--who cares if the title link is duplicative of an identifier? Lev!vich 01:54, 5 August 2020 (UTC)
  • Yes. Ordinary readers assume that clicking on a title link will take them to that document. They may not know what a "DOI" is and what clicking on it achieves. – Finnusertop (talk ⋅ contribs) 08:54, 5 August 2020 (UTC)
  • Yes and no per Sandy. -- Colin°Talk 09:22, 5 August 2020 (UTC)
  • Other only by way of parameter - access = free. for example, doi-access=free. URLs that duplicate parameters should not be manually populated in URL and bots should be free to remove duplicate URLs and move URLs to parameters.  — Chris Capoccia 💬 12:09, 5 August 2020 (UTC)
  • Yes per Sandy and Colin. --Ealdgyth (talk) 13:41, 5 August 2020 (UTC)
  • Yes the links from identifiers are not relevant to most readers and the citation title should always link to the most useful online source that is not a copyright violation. The parenthetical question makes assumptions about the implementation that may or may not be true and muddies the issues, which should be a clear principle. --RexxS (talk) 14:26, 5 August 2020 (UTC)
  • Yes, I prefer including a "duplicate" URL. If an editor has placed a "duplicate" URL in the citation, then it should generally not be removed as being redundant. WhatamIdoing (talk) 15:08, 5 August 2020 (UTC)
  • I prefer no but I recognize that other editors might disagree. This is not the sort of thing we need rules over. As long as it is consistent within an article, WP:CITEVAR should be controlling. In particular we should not be hacking citation templates to force one side of this debate to get its way, and we should not be using robots to add or remove these links. Incidentally, it is important to recognize that some doi-linked citations DO NOT HAVE A TITLE to which a url can be linked, and are valid as citations with a doi but become invalid if a url is added; a bug in some citation templates related to this that broke the references in multiple articles was fixed only this week. —David Eppstein (talk) 06:40, 6 August 2020 (UTC)
  • Yes, I think it is important not to delete a URL that has been deliberately placed in the url= field, regardless of whether it duplicates one of the other available URLs. Even if the behavior of the template is currently such that the desired link will still appear without the explicit URL, that may not always be the case: for example, a different link may be automatically substituted if a new identifier is added, or someone may change the default behavior of the template. The only upside I see of deleting the link is saving a few bytes of storage space. CThomas3 (talk) 20:43, 6 August 2020 (UTC)
  • Yes no need to remove if a formatted ID duplicates the link. But also no need to add an identical link later if something like DOI covers it. The URL if given is what the reference supplier used to access the page. Graeme Bartlett (talk) 12:02, 7 August 2020 (UTC)
  • No Raw links are inferior to persistent identifiers because they tend to break over time. And, again, Wikipedia is an encyclopedia not a search engine. Andrew🐉(talk) 12:06, 7 August 2020 (UTC)
  • Yes link the title to the source actually used. No problem if it duplicates a DOI. Not needed if DOI is there from the start, but do not remove if it exists. · · · Peter Southwood (talk): 15:14, 7 August 2020 (UTC)
  • No — The wikipedia is over-linked as is. A redundant link is a link too many. —¿philoserf? (talk) 08:36, 8 August 2020 (UTC)
  • Yes, this is an enhancement of WP:V. Reducing steps between readers and sources is a good thing to do, and a link in the title is more accessible to those who are unfamiliar with persistent identifiers. (If a field other than url= automatically links the title to a specific webpage, I find that comes to much the same outcome for the reader and have no view on whether url= is preferable if it would link to the exact same page.) CMD (talk) 17:25, 9 August 2020 (UTC)
  • Yes It's simpler and can often avoid multiple clicks to actually get to a text. Adam Cuerden (talk)Has about 7.4% of all FPs 22:43, 9 August 2020 (UTC)
  • Yes The title link should always be present, even if it exactly duplicates an identifier link. If this link will be generated automatically by the template/module based on the other parameters, the |url= parameter can be left blank (or better yer, set to an HTML comment explaining why it is blank), but the title link should always be present, even if it is redundant to an identifier link. DES (talk)DESiegel Contribs 01:03, 12 August 2020 (UTC)
  • No. Per Headbomb. --Bsherr (talk) 02:17, 12 August 2020 (UTC)
  • It depends, if the URL is only "doi.org" or similar like that, no. But if it's the link is the one from the paper's own page on the academic journal's website, then I feel that it's not necessary to remove it, even if it's duplicating the "Doi" parameter . In fact, in several cases, the insertion of the web address of the paper in question makes it possible to add the date of access. However, when we apply CitationBot on a Wikipedia page, most of the time, in the cases mentioned above, the URL and the accessdate are deleted. Therefore, we cannot know when the reference was added. -- Anas1712 (talk) 00:00, 13 August 2020 (UTC)
    • Access dates are not relevant for papers published on specific dates. If you accessed doi:10.1007/lrr-2016-1 2 years ago, if you access it today, or access it in 2 years, you will always be presented with the same version of the paper. Access dates are relevant for websites that can change, like this story for NY Times. If you access it in three days, you may not see the same content you are seeing now. Headbomb {t · c · p · b} 20:30, 25 August 2020 (UTC)
      • [Slightly off-topic] Oddly enough, the documentation for {{Cite web}} claims "Access dates are not required for links to ... news articles with publication dates. I agree with your interpretation, Headbomb: even an online newspaper with a publication date can change after a while, and that documentation is badly flawed. --RexxS (talk) 22:31, 25 August 2020 (UTC)
        • Should be revised to cover online access of print sources. A NY Times article from 1993 can have a web version, but it'll still be stable, because then the NY Time was a print publication and there is no difference between the two. In 2020, when you cite the NY Times, you are likely citing the web version, which is not necessarily stable like the print edition is. Headbomb {t · c · p · b} 22:39, 25 August 2020 (UTC)
          Headbomb, Even in the print version of the NY Times, there would be several editions printed during the day, and coverage would get updated for breaking stories. I used to get the early edition of the NY Times at about 10 PM the night before the nominal publication date. I guess that ended in 1997, though: https://www.nytimes.com/1997/08/02/nyregion/about-early-editions.html
          There were also regional editions; the version sold in the city was slightly different from that sold on Long Island, suburban New Jersey, etc. The bottom line is that just citing a date and page number for the NY Times isn't actually enough to definitively locate a printed article. You also need to know which edition. I assume this was equally true for other papers. I suspect the vast majority of people who cite articles don't do this. -- RoySmith (talk) 22:56, 25 August 2020 (UTC)
Then that's what |edition=Evening/Morning is for, not |access-date=. Headbomb {t · c · p · b} 22:57, 25 August 2020 (UTC)
  • Wrong question this functionality can and should be obtained using |doi-access=free and similar, rather than using a duplicated URL parameter. They should not be linked where it just takes the reader to a paywall, because that doesn't help them. It's wrong to assume that links to URL have been placed deliberately; most of them are likely automatic where people use the url to generate a citation. (t · c) buidhe 03:33, 13 August 2020 (UTC)
  • Yes. The title, as the most intuitive place to find it, is the most important place for the link. That other parameters might also link to the source is not a good justification for not having the link in the title. Boing! said Zebedee (talk) 16:38, 17 August 2020 (UTC)
  • This question doesn't mean anything, or the first wording means the opposite of the second (after "in other words"). "Linking" and "filling a certain parameter" are not the same thing. Discard all the answers above. Nemo 11:04, 21 August 2020 (UTC)
  • Yes it is appropriate to link the title even if that link is already present in the list of identifiers. As a reader I prefer clicking on a title than having to know how bibliographic identifiers work to select the most appropriate one. − Pintoch (talk) 11:34, 21 August 2020 (UTC)
  • Yes, among a choice of links in the template, the link of the title can, and probably should, duplicate the most useful one, or at least the WP:SAYWHEREYOUGOTIT one. --Francis Schonken (talk) 04:28, 24 August 2020 (UTC)
  • Yes We link titles as a standard practice, I think it would be confusing to the reader as to why some titles are linked and others are not. I think Rexx's point below about teaching computer skills is also relevant here. I like the WP:SAYWHEREYOUGOTIT point, I think the URL in the title should be the one where the source was originally procured from, even if that is going to duplicate things. CaptainEek Edits Ho Cap'n!⚓ 09:11, 24 August 2020 (UTC)
  • Yes whenever the DOI etc. leads to a paywall but the text is freely and legally available at a URL, even if that URL is a preprint etc. Certes (talk) 10:31, 24 August 2020 (UTC)
  • No unless it is a completely free copy, which should be generated by a template parameter. WP:SAYWHEREYOUGOTIT is not applied correctly here as it explicitly states "...it does not matter whether you read the material using an online service..." AManWithNoPlan (talk) 20:41, 24 August 2020 (UTC)
    • WP:SAYWHEREYOUGOTIT requires the editor adding content from an online resource like a PubMed abstract to give a link to the resource they used. The title link is the obvious place for that. Yet you're advocating removing those title links contrary to policy. --RexxS (talk) 21:06, 24 August 2020 (UTC)
  • It depends. If the link provided by |url= is pointing to exactly the same page as one of the links derived from the provided identifiers, we don't need to duplicate that link in |url=. (I'm impartial in regard to if links only resolving to exactly the same target page should be regarded as being "equal" as well, because the behaviour of resolvers and redirectors may change over time, and it may be good to have an alternative link.) As a convenience to users who don't know what identifiers are and never click on them, the title could still be linked to this page through (manual or priority-based automatic) auto-linking, however, it is important that editors always have full control if auto-linking should be provided by the template at all and if so, which identifier should be selected. If |url= or |title-link= are provided, they must always have priority over identifier-derived links. However, if the URL and the identifier link point to the same site, but not to exactly the same page (f.e. if the identifier link points to a meta-page, and the URL to the actual PDF), these links are by no means "equal", and the link provided through |url= should never be considered "a duplicate", "redundant" or "equivalent" to the identifier link. Since some users lump these definitions together semantically the exact wording is relevant here. --Matthiaspaul (talk) 20:43, 24 August 2020 (UTC)
Following up on my comment above, I would like to add that even in cases where the |url= is an exact duplicate of an identifier-derived link, I consider the removal of |url= inappropriate if |archive-url= is given as well. Some identifier-derived links are intended to be semi-permanent, but ultimately they are subject to link-rot as well, so it is good to keep archived versions around. If someone has gone the extra mile to provide an archived snapshot for such a link, we should be thankful for this contribution instead of destroying it. I would change my mind on this only if our citation templates would introduce means to provide link archives for identifier links as well. --Matthiaspaul (talk) 13:25, 1 September 2020 (UTC)

3. Under what circumstances should a title link be removed from a citation, by human or by bot (in other words, when should |url= be blanked)?

  • When redundant to identifiers, per Template:Cite journal#Identifiers, or when it's a paywalled link that doesn't add anything to the links already present e.g. a EBSCO paywall link does not add anything that a DOI already provides. Headbomb {t · c · p · b} 01:48, 5 August 2020 (UTC)
  • If the URL linked to the title does NOT go to free full text, and is repeated in another identifier, it can be removed. Other than that, no one (human or bot) should be removing links to free full text, particularly per WP:CITEVAR and the need to provide accessibility to readers. SandyGeorgia (Talk) 01:52, 5 August 2020 (UTC)
  • Not when redundant to identifiers for reasons in my !vote to Question 2 above. Never by a bot because the URL was likely placed there by a human being, who decided that this link was the best link for the reader. A bot shouldn't overrule that human decision. (Note: replacing dead URLs with live or archived URLs is not removing or blanking |url=, and thus could be done by a bot.) Only with good reason by a human - if it's copyvio, if it's a dead link, or if there's some other good reason. I trust humans to make these decisions on a case-by-case basis in a way that a bot cannot. I'm particularly concerned that people will take the time to carefully select URLs to link to, and it'll be erased en masse by a bot. (Which is kind of what led to this bot being blocked in the first place, I believe.) Lev!vich 01:58, 5 August 2020 (UTC)
    • There is zero reason to have both |pmc=91234 and |url=https://www.ncbi.nlm.nih.gov/pmc/articles/PMC91234/. Likewise for having both |pmid=91234 and |url=https://pubmed.ncbi.nlm.nih.gov/91234. There is likewise no issue to clean this pointless redundancy by bot. Headbomb {t · c · p · b} 02:00, 5 August 2020 (UTC)
      The reason to have both is what I said in my !vote in #2 above: to always make it so "click the title, get to the source", so the reader doesn't have to think about which of multiple links is the best one to click on. And I know that you've said that we should accomplish that by removing the |url parameter and instead having a blank |url parameter "auto-fill" effectively with the |doi or whatever, but that's not how CS1 templates work, and the maintainers have previously said that it's not how they're going to work any time soon. So until then... Lev!vich 02:04, 5 August 2020 (UTC)
      That neglects three things. 1) PMC already autolinks the title. 2) A PMID link will take the place of a free version of record, like PMC, if it's declared in the URL, or one that could be found, but isn't added yet. 3) If there's a consensus to always link the title, no matter what, then that can happen automatically and there's no need to have the pmid url in the url parameter when it can just be generated from the pmid parameter. Headbomb {t · c · p · b} 02:11, 5 August 2020 (UTC)
      When |pmc= already exists, and results in the same reader-facing output as a |url= with a link to the same PubMed Central page, then removing the redundant URL would fall afoul of WP:COSMETICBOT. WhatamIdoing (talk) 15:12, 5 August 2020 (UTC)
      Cosmetic doesn't prevent edits, if other stuff is happening. -- GreenC 15:18, 5 August 2020 (UTC)
      Yes, but that little "if" adds a layer of complexity, both in programming the bot and in all of us humans reading the unnecessary changes via diff. WhatamIdoing (talk) 23:14, 5 August 2020 (UTC)
  • What Sandy said. I don't quite understand what Headbomb is saying about autolinking. From a user point of view, I want article titles linked to the URL of the paper if it is available freely, with the official publisher URL in preference to the PMC URL. I don't want users to have to click on links that are only available in codes (PMC/PMID/DOI) because they won't all understand that, and isn't consistent with how other sources (newspapers, websites) work for linking in references. Any URLs in the codes are a bonus and I don't care if they duplicate the link in the title. -- Colin°Talk 09:16, 5 August 2020 (UTC)
    I think Headbomb means Help_talk:Citation_Style_1#Auto-linking_titles_with_free_DOIs and Wikipedia:Village_pump_(proposals)#Auto-linking_titles_in_citations_of_works_with_free-to-read_DOIs. The CS1|2 template will autolink the title. -- GreenC 14:25, 5 August 2020 (UTC)
  • Guiding principle must be that URLs can break and parameters are always preferred. URLs that should more properly be represented by parameters should be moved from the URL field and translated to the correct parameter. If the correct parameter is already present, the redundant URL should be deleted. URL should also be deleted for facilitating copyright violation, although I'm not sure this will be able to be managed by bots.  — Chris Capoccia 💬 12:12, 5 August 2020 (UTC)
  • Should not be removed if the url is to free text ... whether or not there is an identifier link and whether or not that identifier link is redundant. --Ealdgyth (talk) 13:43, 5 August 2020 (UTC)
    What about 'redundant autolinked title links'? -- GreenC 15:18, 5 August 2020 (UTC)
  • The citation title should always link to the most useful online source. The only reason to remove a title link is when it points to a copyright violation. That is a decision that can only be made (and debated) by a human. A bot should never make an edit that has the result for the readers of removing a link from a citation title. --RexxS (talk) 14:31, 5 August 2020 (UTC)
    • There is zero point in having both |pmid=91234 and |url=https://pubmed.ncbi.nlm.nih.gov/91234 just for the sake of preserving a link. Or redundant links to paywalled resources. Headbomb {t · c · p · b} 14:36, 5 August 2020 (UTC)
      • But is there any point in flooding everyone's watchlists to remove the URL? WhatamIdoing (talk) 15:14, 5 August 2020 (UTC)
        So, the hard-coded URL dies or changes. It's trivial to fix the autolinking in the central CS1|2 cfg. But, IABot is going to detect the dead static URL in the individual cite, and add an archive URL. Now we have a cite that is not properly autolinking because there is a dead URL in |url= field taking priority, with an unneeded |archiveurl= that probably should be removed along with the static URL. OR, we can proactively remove the redundant URL to prevent a future mess. -- Green<span style="color:there should be a link on the title (and note that some of the people who voted No were answering a different question, whether |url= should be filled). #093;">C 15:28, 5 August 2020 (UTC)
        Or we could assume that scenarios based on PubMed Central (part of the US National Institutes of Health) going offline, or them not being able to find their own documents when given their own ID numbers, are just not very compelling. WhatamIdoing (talk) 23:20, 5 August 2020 (UTC)
      • There is every point in having a link from the title, and it shouldn't have a bot removing it. Nobody cares about the detail of the implementation except a bunch of techno-geeks. The reader comes first and they don't care what parameters we have in the wikitext, they just want to click the title to get to a source, and a bot should should never be making an edit that removes that from them. --RexxS (talk) 16:35, 5 August 2020 (UTC)
        If it's so trivial to fix the autolinking, why hasn't it been done after three months? Even then, the hard-coded url can still be needed when a free directly-accessible source is not linked from any identifiers. It's also necessary to allow an archived url to be directly linked from the citation title, which the bots seem incapable of sorting out – a far better solution than preemptively removing links in case they become dead links. --RexxS (talk) 16:35, 5 August 2020 (UTC)
        Because CS1 updates are basically contingent on User:Trappist the monk deciding it's time to do an update. It's not, in theory, a limitation, but you need to know LUA and to be an admin to update the CS1/2 modules. But this isn't a particularly hard thing to do for those with the skills. Headbomb {t · c · p · b} 18:06, 8 August 2020 (UTC)
        Ah, but RexxS does know Lua, and is an admin - indeed, when I have a lua problem, I prefer to go to RexxS and not Trappist. --Redrose64 🌹 (talk) 19:44, 8 August 2020 (UTC)
        Nevertheless, Trappist has put a huge amount of effort into creating and maintaining the CS1/2 modules. That means Trappist has a far better overview of the intricacies and features of that module than I have. I would not feel comfortable editing any of those modules, principally because I would automatically defer to the maintainer of a module to know better than I. The same would apply, for example, with Johnuniq's convert modules. --RexxS (talk) 20:55, 8 August 2020 (UTC)
        Can you verify if this is autolinking?
        In this example it would be redundant to add |url=https://doi.org/10.12927/hcpol.2009.21005 because it is the same URL as is already generated - most everyone agrees about this. In terms of archives, doi.org is unlikely to completely die rather change its URL syntax (the base URL) which can be easily "fixed" in the template cfg, the same way we could with external link templates like {{gutenberg author}} should they ever change base URLs. If it did completely die we'd probably change to a different ID provider as the default autolinking. The bot's don't currently archive ID URLs at this point but given the uncertainty they should wait to determine the outcome, and there is no technical reason they could not do so. BTW my professional training is History. I suspect you are more techno geek than myself. But I don't hold it against you :) My interest is the humanities and the technology that makes it more accessible to a broader public. -- GreenC 20:35, 5 August 2020 (UTC)
        @GreenC: I can verify that the citation you quote is linking the title, but I seriously doubt that there's anything automatic about it. You'll appreciate that
        also has the title linked. Does that make the doi redundant? They all link the title to the source, so does messing with those actually benefit the reader?
        What about these citations? Why would we change the first for the second?
        How did this edit benefit the reader?
        FWIW my first degree was in electronic engineering, but that was in 1972 and there were no geeks in those days. :( --RexxS (talk) 22:15, 5 August 2020 (UTC)
        The |pmc=2732656 instructs the template to automatically generate a title URL. Which is my understanding of autolinking. The ID |pmc=2732656 and the |url=https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2732656 generate the same title link thus redundant, do we keep both anyway? You are correct we should not change the first for the second, since that would eliminate the title link with no new link being generated. Well you are about a generation, I'm 92' .. they were nerds when it meant something. -- GreenC 23:56, 5 August 2020 (UTC)
        @RexxS: The edit you questioned, removing a direct link to Semantic Scholar and replacing it by an S2CID id, benefits the reader by not misleading them into thinking that they can follow the link to find the actual reference, when in fact they will only find a landing page there telling them that they should have followed the DOI instead. —David Eppstein (talk) 06:47, 6 August 2020 (UTC)
        @David Eppstein: removing a link to a version of the source (which will then take them on to a full version) and replacing it with nothing does not benefit the reader. Most readers don't understand s2cid or doi, so editors should not be forcing them to fit their preferred way of working. When will they realise that the change needed in that edit was to link the title to a better site in order to improve it, not to remove it altogether and make it worse? --RexxS (talk) 16:24, 6 August 2020 (UTC)
        When you say that replacing a link on the title to a link on the S2CID is removing the link altogether, I think you are ridiculously overexaggerating the issue. And when you say that readers are incapable of understanding links described as identifiers and are only capable of finding and following links when those links are placed on the title of a reference, I think you are grossly underestimating the competence of readers. —David Eppstein (talk) 19:01, 6 August 2020 (UTC)
        And I think your personalisation of genuine issues detracts from the seriousness of the problems. Removing a link from a citation title should not be happening because readers expect the link to be there. You are attempting to brush the issue aside as if doesn't matter. Well it does. Likewise you are not displaying any understanding of the difficulties that ordinary readers experience in trying to make sense of a plethora of different acronyms and symbols. Not everybody is as technically gifted as you and you overestimate the willingness of a reader (who just wants to see a source) to learn jargon designed for use by academics who use it on a daily basis. --RexxS (talk) 21:51, 6 August 2020 (UTC)
        It's not a question of learning jargon. It's a question of being capable of noticing that something in the reference is linked and seeing where you go when you follow that link. If you can't do that, you also can't use Wikipedia, use the web, or follow links in titles. —David Eppstein (talk) 01:12, 7 August 2020 (UTC)
        If you'd spent as many years I have teaching adults, especially those as old as I am, basic computing skills, you'd soon revise your "being capable of noticing that something in the reference is linked and seeing where you go when you follow that link. If you can't do that, you also can't use Wikipedia, use the web, or follow links in titles. Giving folks predictable pathways, such as "clicking on a title takes you the source" is precisely how they become able to use Wikipedia, and use the web, and follow links in titles. --RexxS (talk) 16:49, 8 August 2020 (UTC)
  • Remove per Chris Capoccia and Headbomb, if the CS1|2 template is already autolinking the title field (per RfC consensus and ongoing discussions at Help_talk:Citation_Style_1) there is no logical reason or purpose to have a redundant URL in the |url= field. If the URL is not redundant ie. they are different URLs, it should not be removed since the |url= is a manual override of the autolink. -- GreenC 14:37, 5 August 2020 (UTC)
  • Only ever by a human, and only for the purposes of making the citation style within an article consistent with its prior state, per WP:CITEVAR. —David Eppstein (talk) 06:41, 6 August 2020 (UTC)
  • The bot shouldn't, and a human should only do so to consciously choose to a better/more appropriate link, not because it's redundant. CThomas3 (talk) 20:52, 6 August 2020 (UTC)
  • Only remove the link if it goes to a deprecated site, (such as copyright violating or dead) or is wrong. Don't bother removing it just because of PMC or DOI or the like. That is just cluttering up watchlists for no benefit. Of course the other parameters can be added even if there is a link in the title. Graeme Bartlett (talk) 12:07, 7 August 2020 (UTC)
  • Ban bots We can't have bots run by paid SEO editors fighting to insert their preferred links. Per WP:CITEVAR, the issue should be left to discretion of the human editor who actually used the source in question to support some text in the article. Andrew🐉(talk) 12:11, 7 August 2020 (UTC)
  • If it aint broke don't fix it. · · · Peter Southwood (talk): 15:18, 7 August 2020 (UTC)
    To clarify: I don't think it is broken and don't think it needs fixing. If the title link url works, leave it alone. · · · Peter Southwood (talk): 10:57, 9 August 2020 (UTC)
  • It is broke. Let’s fix it. Any redundantly linked title should have the |url= removed. The wikipedia is over-linked as is. A redundant link is a link too many. —¿philoserf? (talk) 08:36, 8 August 2020 (UTC)
  • The link from the title should never be removed, evven if it is redundant. If, and only if, the template/module will generate this link automatically based on other parameters, |url= may be set to blank, or better to an HTML comment explaining why, but not otherwise. All this is assuming that the article uses CS1 or CS@, of course. DES (talk)DESiegel Contribs 01:08, 12 August 2020 (UTC)
  • Never by a bot bots should not be going around second guessing decisions made by actual humans. I would support a bot identifying cases that need human attention, but never making these sorts of changes themselves. - Nick Thorne talk 23:54, 13 August 2020 (UTC)
  • Per RexxS above, the citation title should always link to the most useful online source (barring copyvio). I don't know if there might be cases where there's a good argument for not having the link (unlike Holmes, I don't claim to be able to eliminate all possibilities). But if there is, then it needs to be decided by a human. The link should never be removed by a bot. Boing! said Zebedee (talk) 16:44, 17 August 2020 (UTC)
  • Don't change, unlike there's an open-source access to the full paper, but that's the purpose of OABOT, so maybe allows in precise cases, but not to generalise to all. The primary policy is to not remove utile information that is already present. The objective of the bots in citation maintenance is to help the wikipedians to add informations and to automatise the search of informations, to make sure the citation is more precise and more accessible. Voilà, voilà. --Anas1712 (talk) 14:42, 18 August 2020 (UTC)
  • Don't remove title links on the basis that they duplicate identifiers. It is fine to remove a |url= if the template will generate the same link on the title via auto-linking, but not otherwise. − Pintoch (talk) 11:37, 21 August 2020 (UTC)
    • It's also more than fine to remove links that lead to databases without any possibility of free versions (like pubmed), to paywalled papers (like EBSCO and proxies), to preprints (like arxiv). Headbomb {t · c · p · b} 15:27, 22 August 2020 (UTC)
      • It's not fine to remove the link to where the editor found their source per WP:SAYWHEREYOUGOTIT. If they found their information at a PubMed abstract article, we should be giving the editor the ability to indicate that via the title link, and we should be giving the reader the opportunity to follow the intuitive link to the source used in writing the text. We should not be having those swept away by the bucketload by a capricious bot. --RexxS (talk) 21:03, 24 August 2020 (UTC)
  • The content of |url= should be decided on a case-by-case basis, by human editors, not modified by bot – a bot is welcome to address malformed links, permanently dead links etc, but should not modify the intent of human editors. --Francis Schonken (talk) 04:28, 24 August 2020 (UTC)
  • Never by bot and very rarely by human I cannot think of a good reason for a bot to remove the URL parameter. I cannot think of a reason beyond copyright violation for a human to remove the URL, though I assume there must be some reason I haven't dreamt up :) In general, Rexx's point about having consistently linked titles wins me over, and thus I think this is a good reason to keep titles linked. Humans can always use their best judgement to figure out if a link needs to go, but I see no need to have CitationBot doing that. CaptainEek Edits Ho Cap'n!⚓ 09:17, 24 August 2020 (UTC)
  • When redundant to identifiers Linking to non-free copies with the title leads to reader confusion when they expect a title link to actually link to something useful. Cannot believe that I was not aware this discussion was going on until now. AManWithNoPlan (talk) 20:44, 24 August 2020 (UTC)
    • Identifiers are inferior to title links because readers understand and expect title links. Your bot is happy to remove title links that go to pmc urls. Those sort of links – the ones used by the editor to add text – are useful title links, and no reader is going to be confused if they are allowed to follow those sort of links. --RexxS (talk) 21:03, 24 August 2020 (UTC)
  • Only when identical to other links. There is no point to keep an |url= link if it points to exactly the same page as an identifier-derived link. This is the only case when (automatic) removal of |url= is justified (except for if it would point to illegal contents which always justifies its removal), and this is also the only case for which we actually had community consensus. If |url= resolves to exactly the same page, |url= may be removed by a human after applying editorial judgement, but it should not be removed by a bot because a bot cannot apply the necessary editorial judgement, if the resolver will, with high certainty, resolve to the same page into the long-term future, or if it may change over time. A link pointing only to the same site (like links to the direct document vs. to a meta-page) is not "equal" and must therefore never be considered "redundant", "equivalent", "duplicative", etc., and consequently it must not be removed. (Since some users lump these definitions together semantically, the exact wording is important.) Also, |url= pointing to a non-free or restricted-access resource is never a valid reason to remove a link per our core policy on verifiability WP:V and the principle of WP:SAYWHEREYOUGOTIT. For that reason, also a dead |url= must not be removed (but |url-status=dead be set), as this would invalidate the citation. Of course, if it is possible to replace a (dead) link by a similarly authorative or better live URL, that's allowed for humans (and in some very limited cases it may also be allowable for a few highly regarded bots -- however, not for Citation Bot, which has a long track record of messing up citations and over the years has created way too much damage to the project to still have any credibility, IMO -- the quality of edits is much more important than the quantity). --Matthiaspaul (talk) 21:39, 24 August 2020 (UTC)
I would like to add that even in cases where the |url= is an exact duplicate of an identifier-derived link, they should not be removed if |archive-url= is given as well (unless citation templates introduce parameters to specify archive links for identifiers as well). Some identifier-derived links are intended to be semi-permanent, but ultimately they are subject to link-rot as well, so it is good to keep archived versions around. --Matthiaspaul (talk) 13:31, 1 September 2020 (UTC)
  • The bot never; a human might be justified to do so in rare special circumstances. Johannes Schade (talk) 16:26, 31 August 2020 (UTC)
  • Matthiaspaul says above As a convenience to users who don't know what identifiers are and never click on them, the title could still be linked to this page through (manual or priority-based automatic) auto-linking. I think in such cases removing |url= is OK; the parameter is freed up for some potential other URL, and in the meantime readers see no change since the title still has the auto-linked ID-derived URL. Of course the edit-summary should provide [a link to] a reassuring explanation of this. jnestorius(talk) 14:13, 3 September 2020 (UTC)
Yes, just in case this didn't became clear in my post above, if |url= points to exactly the same page as provided through an identifier link and there is no |archive-url=, then |url= serves no purpose and can be freed for better uses. (With |archive-url= being present I would not agree with the removal of |url= unless the citation templates would be improved to support archive links for identifier links as well.) What's also a problem is that Citation Bot has been caught to remove |url= also in cases where the identifier-derived links were not identical, but only pointed to the same site (meta page vs. deep link to document) or even to a different site altogether which just happened to contain similar information. This is something that I consider to be disruptive and unacceptable. That's why I pointed out that people have vastly different ideas what terms like "duplicative", "equivalent", or "redundant" mean, and that therefore the unambiguous terms "identical" or "equal" should be used. --Matthiaspaul (talk) 15:34, 3 September 2020 (UTC)
  • We need a better CS1/CS2 template, if title links are important I think that when no title-link is present, the template should always try to make one. If something-free is not set, then auto-link doi with the assumption that it is not free in the icons. I think wikipedia over links, but if we are going to insist on links, then let us make them just appear by magic. AManWithNoPlan (talk) 14:27, 3 September 2020 (UTC)
  • AManWithNoPlan has the right idea. I don't care if there's a url= field, but I do care that the title links to be best available url for the source. (We should make it as easy as possible for readers and editors to verify the content of our articles.) If titles can be linked automatically based on some of the other fields: great. If not, leave the url field alone. pburka (talk) 23:03, 5 September 2020 (UTC)
  • I also agree with AManWithNoPlan. It is the job of the CS1/CS2 template to create the best title link (with the appropriate |url-access=), not a bot. But this is off topic for Link removal. The |url= link should be removed if it is equivalent to the link created by a parameter used in the cite: |oclc= (|url=https://www.worldcat.org/...), |doi= (|url=https://doi.org/...), |pmc= (|url=https://www.ncbi.nlm.nih.gov/pmc/...), |pmid= (|url=https://pubmed.ncbi.nlm.nih.gov/...), |jstor=, etc. It should also be removed if the bot adds the equivalent parameter to the cite. — User-duck (talk) 17:41, 6 September 2020 (UTC)
    • Off topic. Based on some of the WP:SAYWHEREYOUGOTIT arguments, |title-link= should not be a cite parameter. Wikipedia is not considered a reliable source and the book article seldom contains the referenced content. Personally, I think it should be more like "say where you can find it". — User-duck (talk) 17:41, 6 September 2020 (UTC)

Summary

I've unarchived this discussion as the issue is now back at WP:AN #Citation bot again. The discussion needs a summary of the consensus shown in the discussion.

I believe the consensuses demonstrated here are:

Title linking
Yes to a free link. Yes to other links.
Duplicate linking
Yes
Link removal
No

I suggest that an uninvolved admin evaluates the discussion and closes it with their own conclusion, but if that doesn't happen in the next few days, I propose to close this discussion in the manner I've outlined above. --RexxS (talk) 01:06, 19 September 2020 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Village_pump_(proposals)/Archive_172&oldid=1092751231"