Wikipedia talk:Manual of Style/Archive 202

Archive 195

←

Archive 200

Archive 201

Archive 202

Archive 203

Archive 204

Archive 205

RfC: Linking to wikidata

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.

---

I have noticed another editor adding links to wikidata for people who fail notability requirements to have articles on wikipedia. In some cases, the he links to wikidata after the article on the person was deleted in WP:AfD. Before I start reverting his edits, I like to know if this is an acceptable practice. I don't know much about wikidata and what gets included there, but it would seem to me if a person is mentioned in an article (usually as part of a list) and is not notable enough for his/her own article, then there shouldn't be a link to anything. Here's an example of what I'm talking about: Mayors of Teaneck, New Jersey That article has a lot of links to wikidata. I appreciate input on this.--Rusf10 (talk) 05:15, 14 January 2018 (UTC)

---

If someone wants to see how the article looked with the Wikidata links: https://en.wikipedia.org/w/index.php?title=Mayors_of_Teaneck,_New_Jersey&oldid=817759493 ―Justin (koavf)❤T☮C☺M☯ 08:07, 16 January 2018 (UTC)

---

• Such links strikes me as a bad idea for multiple reasons:
  1. WikiData doesn't comply with en.WP's WP:Biographies of living people policy or our core content policies.
  2. Wikidata is not prose, so doing this is going to sharply transgress the principle of least surprise, and the reader may really have no idea what's happening at all.
  3. This isn't like linking to Wiktionary for WP:DICDEF material; that's linking to a sister project for material that should not be on Wikipedia at all because of what kind of material it is. Bios do not constitute such material.
  4. If a subject is arguably notable and should have an article here, it should be a redlink. If it's not, it should not be linked at all.
  5. It's WP:GAMING the system to keep promoting non-notable subjects.
      — SMcCandlish ☏ ¢ >^ʌⱷ҅_ᴥⱷ^ʌ<  14:02, 15 January 2018 (UTC)
• I agree with SMcCandlish here. I followed the wikidata links and my first thought was that this is a very elaborate and space-consuming way of presenting absolutely no information whatsoever. I think these links are a terrible idea. Reyk _YO! 14:06, 15 January 2018 (UTC)

How can a database of information contain "absolutely no information" unless it is an empty record, which this entry is not? --RAN (talk) 20:46, 16 January 2018 (UTC)

• I would consider indeed to revert this completely. I already shiver when a person who fails our local inclusion standards, but is notable for e.g. Spanish is cross-wiki linked to the Spanish article (which is rather difficult to see even). I see this as nothing short of an inline external link to the website or social media account of non-notable subjects. We could consider an EditFilter for this, at least to detect such cases. --Dirk Beetstra ^{T C} 14:08, 15 January 2018 (UTC)

Wouldn't it make sense to keep the information without the link, though? Wikidata is also a sister project, and having items decorated with information about their meaning is a basic principle of the semantic web; it could be invaluable for automatic treatments of our content, or creating an improved browsing tool in the future. Our WP:BUILD policy supports the principle of creating a connected network of information that brings related items together.

I can imagine a specific template for this purpose, similar to how we already have WP:INTERWIKI tags. It should be a systematic effort though, and not the initiative of a single editor, so this should be discussed at the village pump, not at the manual of style. Diego (talk) 14:30, 15 January 2018 (UTC)

Generally WP:EL discourages external links within article bodies. The problem is EL operates on the basis that the EL's are about providing more information on the subject of the article, not unrelated information about someone who appears in the article. When it comes to lists, the only links within the list should either be a link to the relevant Wikipedia article, a properly cited reference, or an external link at the end to an appropriate resources on the subject of the article. In the above example, the wikidata EL's don't provide any further information on the subject 'Mayors of Teanick' than is already covered by an appropriate reference. Only in death does duty end (talk) 14:39, 15 January 2018 (UTC)

Links to Wikidata are under Wikipedia:Wikimedia sister projects, not WP:EL. The question here is whether we can make such links useful to readers (the requirement by that guideline); this doesn't need to be with a direct link to the URL, maybe we can explore more useful ways. Perhaps we should simply follow WP:SOFTSISP and create a soft link, to avoid WP:Astonish wighout losing the utility? Diego (talk) 14:53, 15 January 2018 (UTC)

It's unclear to me what use there is in linking to completely empty wikidata entries, as is the case here. Reyk _YO! 15:00, 15 January 2018 (UTC)

Can you give an example of a linked empty Wikidata entry? --RAN (talk) 20:30, 15 January 2018 (UTC)

The Karl Wagner entry was an example of a link to a contentless entry. It's a big sprawling page with pretty boxes everywhere but no information. Reyk _YO! 21:24, 15 January 2018 (UTC)

• Which Karl Wagner? Karl Edward Wagner (1945–1994), American writer of fantasy stories

Karl Wagner (bobsleigh) (born 1907), Austrian bobsledder who competed in the early 1950s Karl Wagner (luger), German luger who competed in the 1920s Karl Willy Wagner (1883–1953), German pioneer in the theory of electronic filters?

I'm not sure if you're being sarcastic, but those big boxes are the information stored in Wikidata (properties and values), together with the unique ID. Diego (talk) 11:31, 16 January 2018 (UTC)

Wikimedia sister project links are also covered by WP:EL. All external links are covered by WP:EL except for the named exceptions. See WP:ELMAYBE. The 'external' in EL is 'external to wikipedia' not 'external to wikimedia'. -edit- to expand, in an article about a person on Wikipedia, it may be acceptable to add a link to wikidata per ELMAYBE if wikidata contains more/useful information that is not covered within the article (leaving aside the issue of wikidata being unreliable for the moment). In an article about something else, in which the person is named, its almost certain the wikidata entry will not contain useful information to that article, as the wikidata material would be linked to the person. Only in death does duty end (talk) 15:24, 15 January 2018 (UTC)

But WP:EL covers links within an article; it does not cover soft redirects. Diego (talk) 15:41, 15 January 2018 (UTC)

Those are not soft redirects. Only in death does duty end (talk) 20:44, 15 January 2018 (UTC)

I know that. My proposal was changing them to soft redirects, because that's how the relevant guideline WP:SOFTSISP says that links to sister projects should be handled. Diego (talk) 10:24, 16 January 2018 (UTC)

• Much of this discussion is redundant, it seems to me. Links to Wikidata now appear automatically under "Tools" in the left margin of the desktop version as soon as there is a Wikidata item that links to the article. So there's never a need to add a link to Wikidata (or Wikispecies) in the External links section. Peter coxhead (talk) 15:28, 15 January 2018 (UTC)

See the above example RE Mayors of Teaneck, the wikidata links added to the Wikipedia article are not to wikidata items that link to the article page. (The wikidata link under tools goes here. Only in death does duty end (talk) 15:33, 15 January 2018 (UTC)

Rusf10 already considers the case closed and has removed the links. An example would be Frank White Burr. -RAN (talk) 21:07, 15 January 2018 (UTC)

User:Richard Arthur Norton (1958- ), the response here was a near unanimous consensus against linking to wikidata, but if you want to drag this on, go right ahead. It is also worth mentioning that the article for Frank White Burr was deleted as per WP:Articles for deletion/Frank W. Burr. The result was not redirect either, but that's another issue.--Rusf10 (talk) 21:19, 15 January 2018 (UTC)

Taking out the wikidata links might be jumping the gun a little, although I don't think consensus will change, but I do agree with taking out that big blob of hidden text. I'm not sure what it was even for. Reyk _YO! 21:34, 15 January 2018 (UTC)

"the response here was a near unanimous consensus" Well, given the one-sided comments posted above, that's perhaps understandable, but the RfC has only been open for two days, and has not been centrally advertised (which is required for something so sweeping), so it's too soon to suggest consensus has been reached. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:20, 16 January 2018 (UTC)

Re. "centrally advertised": sorted --Francis Schonken (talk) 12:35, 16 January 2018 (UTC)

While the links in the Mayors of Teaneck list are indeed to different items than that of the equivalent to the page, the links mentioned by Peter do indeed render moot any suggestion that Wikidata is not a suitable target for us to link to, or that there is not consensus to regard it as such.

"WikiData doesn't comply with en.WP's WP:Biographies of living people policy or our core content policies." Neither does Commons, nor Wikisource, nor any of our sister projects, including other Wikipedias. This point is completely irrelevant. Wikidata, does though comply with WMF policy on BLPs.

Link to Wikidata in existing tables if they are not getting an article

• Link to the name in Wikidata if they appear in a table. Mayors of Teaneck, New Jersey is a good example of people that will not get full articles. If someone thinks they can expand an entry into a full biography, nothing stops them if they meet WP:GNG. Linking to Wikidata allows us to disambiguate people of the same name and link to other authority control identifiers like VIAF and LCCN. No one is arguing that being in Wikidata makes you Wikipedia notable. This is for people who do not have a full biography and most likely never will. It is no different than linking to the Italian Wikipedia for people there, that will not get an article in the English Wikipedia. Do not link to people in Wikidata for "Notable people from X" in location articles. Being in Wikidata does not make you notable, but notable people can be in Wikidata where the biographical data is too thin for a full article. --RAN (talk) 20:26, 15 January 2018 (UTC)

It is different, because Italian Wikipedia contains useful information to a reader which is more compliant with our local policies. Wikidata does not. Only in death does duty end (talk) 20:47, 15 January 2018 (UTC)

What useless information is in Wikidata? --RAN (talk) 21:00, 15 January 2018 (UTC)

Let me turn that question around.... what USEFUL information is at Wikidata? Blueboar (talk) 06:15, 16 January 2018 (UTC)

Interwiki links. And, um, ... I think that's it really. As an example of what's wrong with Wikidata, its page on one of the Teaneck mayors [1], an immigrant from India to Tanzania and then the US, lists his citizenship as US, based on zero sources. It's a plausible guess that he's a naturalized citizen (maybe Teaneck requires its mayors to be citizens, I don't know), but as far as I can tell it's only a guess, and if true is only true for a certain period of time and doesn't tell the whole story. That sort of thing would not be acceptable in a BLP here, and I think it violates WP:ELNO #2 and #12. —David Eppstein (talk) 06:22, 16 January 2018 (UTC)

You do realize that there is a template {{Unreferenced}} in the English Wikipedia that is posted in over 5,000 articles. I stopped at 10 pages of 500. There are over 10,000 paragraphs in biographies of living people with no reference and there are over 10,000 dates of birth in biographies of living people that are unreferenced. I stopped the search at 10K, you can run it longer. "Unreferenced at this point in time" is not the same as "inherently unreferencable". --RAN (talk) 22:18, 19 January 2018 (UTC)

• Only link to the interwiki links section of Wikidata items in cases where there are articles for something in three or more Wikipedias but not in English. This is the way {{Interlanguage link}} does it (though the choice of Wikidata link vs. direct link(s) to other Wikipedias is at editors' discretion), solely for the purpose of allowing the reader to use the links, and it avoids presenting the reader with too many or too few language links and allows the reader to skip past the offensive, objectionable and completely untrue piles of useless data large number of tangentially relevant statements and links to other databases that some items have. Jc86035 (talk) 03:14, 16 January 2018 (UTC)

It's useful to specify when two Wikipedia articles speak about the same person. Pernell Roberts, Helen Schucman and Saybrook University all speak about the same Eleanor Criswell. Wikidata links allow this information to be conveyed to the reader and otherwise it wouldn't be expressed in Wikipedia. I see no need for a subject to be notable for us to specify that when a subject is talked about in multiple articles to specify that it's the same person. ChristianKl ❪✉❫ 10:16, 16 January 2018 (UTC)

Like this: Eleanor Criswell? --Francis Schonken (talk) 10:24, 16 January 2018 (UTC)

It would be possible to create a redirect for every person that's mentioned in a Wikipedia article and let that point to that Wikipedia article but I'm not sure that's a clean solution either. If a person is only mentioned on two pages and on one of those pages there's a link to a redirect to the same page, that feels more like an hack. ChristianKl ❪✉❫ 17:25, 16 January 2018 (UTC)

• Just thinking out loud: what if our 'page' of a non-existent article (the red-linked page) is interwiki-linking to other Wikipedia's articles (including wikisource/wikidata/etc) of said article if articles exist there (it may help the reader to more information, or editors to a quick start if they wish to create the article)? --Dirk Beetstra ^{T C} 07:39, 16 January 2018 (UTC)
  • Did you mean what we do with {{ill}} templates, which makes linking to more than one page on other projects possible, starting from a redlink? --Francis Schonken (talk) 07:47, 16 January 2018 (UTC)
    • Roughly, though without telling the page which interwikis to show (though I then don't know how a page would detect the existence etc.). --Dirk Beetstra ^{T C} 07:51, 16 January 2018 (UTC)

      Technically, one can call multiply Wikidata entries from the same page, and actually Wikivoyage heavily uses this.--Ymblanter (talk) 07:54, 16 January 2018 (UTC)

      E. g. "De Danske, Norske og Tÿdske Undersaatters Glæde, TWV 12:10" would link to multiple articles on other Wikipedias & Commons (via the Wikidata item). Would that approach what you had in mind? Note that at the TWV list I had used "De Danske, Norske og Tÿdske Undersaatters Glæde, TWV 12:10" instead, with three direct links to the sister projects instead: this has the advantage of one-click reach of these other pages, but the disadvantage of not linking to any future other Wikimedia project which may have an article on this before en.Wikipedia (although it would also not take more than two clicks via any of the linked pages to any future additional page on the same topic elsewhere). --Francis Schonken (talk) 08:15, 16 January 2018 (UTC)

With regards to Wikidata links within tables of items: a while ago I created Template:Wikidata icon which allows for this in a visually clear way (without having to have the Q-number showing to the reader). This is useful for list articles where not every item in the list is necessarily appropriate for a standalone Wikipedia article [yet?] but THEY ARE ALL applicable for Wikidata items (practical example), or when the infobox is about two separate things described in the same article (practical example). Wittylama 12:51, 16 January 2018 (UTC)

Like his? --Francis Schonken (talk) 13:20, 16 January 2018 (UTC)

The horror :-) Adding meaningless (to readers) "flags" to items in lists or in the middle of text just to indicate that some other unreliable site has "information" (well, some structured data) on this item, which in most cases will leave them disappointed anyway, is a bad idea. There is no reason to spam one specific site simply because it is a "sister site". That specific example sends readers to an unsourced page with an external link to Findagrave, where the Findagrave page doesn't even include the information that he was mayor of that town. Why would we ever want to promote such links (either disguised as bluelinks, or by giving them a pretty icon)? Fram (talk) 13:29, 16 January 2018 (UTC)

Francis Schonken, yeah - like that would be an example use case. And Fram, setting aside your hyperbole, the value of having wikidata links in this kind of circumstance (list of items where not every item will/does have a WP article) is to give readers access to more, relevant, information about the subject than could be appropriately added within the context of the given table. It is easy to find a relatively empty item to point to as proof that no such links should be added. Equally, it would be easy to find a high quality item to 'prove' the reverse. The point of these links in tables is that we can give our readers access to further free-licensed information in a structure that we can control (as opposed to external links). We can ALSO, when we desire, add a column for external links, references, other authority-control numbers if applicable, and in such cases it would be possible to move a WD item reference to that column in addition/replacement to the external. That would need to be seen on a case by case or article by article basis. In the mean time, having an unobtrusive WD icon (instead of its WD Q-number) is a convenient way to indicate to readers that more information exists and STILL leaves open the possibility of the words in WP being red-linked, blue-linked, or not linked at all. Wittylama 15:58, 16 January 2018 (UTC)

I'd avoid doing that with a flag-like icon though, it seems too counter-intuitive. E.g. at the Mayors of Teaneck, New Jersey page: what club does this gentleman belong to?... this red-green-blue striped flag is not the icon of a political party I know. --Francis Schonken (talk) 17:22, 16 January 2018 (UTC)

Anyhow, I listed the template for deletion, see the template's entry on the Templates for discussion page. Don't think this kind of WP:EGG flag link is something we should be doing. --Francis Schonken (talk) 21:04, 16 January 2018 (UTC)

I find it premature to list a template for deletion that pertains precisely to the discussion being held here in this subsection: about the use of WD links within tables - clearly one outcome will prejudice the other. If you have an alternative image/icon to propose, go ahead, but the WD icon is standard visual iconography across our sister websites - unless we want to treat en.wp as a special exception (like we do with the Wiktionary logo). Wittylama 10:31, 17 January 2018 (UTC)

Never link to Wikidata

• I thought we had this discussion a few months ago at the Village Pump... any way... until wikidata can comply with our Policies and guidelines, we should not link to it. At all. Blueboar (talk) 22:28, 15 January 2018 (UTC)
• Support. Had enough of WikiData shenanigans. Curly "JFC" Turkey 🍁 ¡gobble! 01:23, 16 January 2018 (UTC)
• Support No wikidata links as per  SMcCandlish above.— Preceding unsigned comment added by Rusf10 (talk • contribs)
• Support Unless and until Wikidata can prove its security is as good or better than here at Wikipedia, we should not use it. A good model here is Commons; Commons works for use here because they've gotten good at stopping abuse. Unless and until Wikidata can establish themselves as good as Commons, we should not use it here. --Jayron32 01:37, 16 January 2018 (UTC)
• Support - I am not sure that we need to use Wikidata in any manner, but certainly not as a replacement for articles. Moreover, if an article might be plausible, it is better to have a redlink so that we can tell that an article is missing. — Carl (CBM · talk) 01:49, 16 January 2018 (UTC)
• Support avoiding wikidata links in article text. They do not have anywhere near the quality control or useful content of other inter-language links provided e.g. through {{ill}}, and seem to have been included in some of the examples discussed above primarily as a way to sidestep Wikipedia's BLP requirements. The link from an article to its own wikidata item (provided through wikimedia software and not under our control) should stay, of course, because we need it to make our links to other language versions of the same article. —David Eppstein (talk) 01:51, 16 January 2018 (UTC)
• Support per Jayron32 and David Eppstein. · · · Peter (Southwood) ^(talk): 06:12, 16 January 2018 (UTC)
• Support - There hasn't been a good case made for using Wikidata at all. And if its use in the Teaneck mayors article is typical, it seems to be just a means to dodge our content requirements while sending readers to a series of perplexing and useless dead ends. Reyk _YO! 06:28, 16 January 2018 (UTC)
• Support - No good case has been made for use of Wikidata, perhaps other than to replace the old interwikis. - Sitush (talk) 07:04, 16 January 2018 (UTC)
• Support per all the good reasons above. Fram (talk) 07:17, 16 January 2018 (UTC)
• Support not using WikiData (for transcluding data at all, only interwikis), and certainly not in this way. --Dirk Beetstra ^{T C} 07:26, 16 January 2018 (UTC)
• Support per Jayron - and for largely the same reason I have been harping on for months about. The comparisons to commons don't hold up under the slightest scrutiny because Commons has content policies that are equal to, or stricter than ENWP - which means content hosted there is 99% of the time able to be used as-is here. We don't have to use it, and commons doesn't force itself upon us. But should be choose to, we can be assured they (almost always) meet our free content requirements. Wikidata has no policies that come close to matching our content policies. It doesn't have a userbase remotely as large as is required to administer it effectively *to the point where we would consider it adequate*. And the userbase it does have is more concerned with collecting vast amounts of factoids rather than making sure they are accurate. Only in death does duty end (talk) 08:37, 16 January 2018 (UTC)
• Oppose as luddism (though I believe that in the case in question such links were not needed).--Ymblanter (talk) 10:08, 16 January 2018 (UTC)
• Oppose It's useful to specify when two Wikipedia articles talk about the same person even when that person isn't notable. That's for example the case for Pernell Roberts, Saybrook University and Helen Schucman which all talk about the same Eleanor Criswell. Disambiguation is also useful for list articles in general where Wikipedia might not want to show the date of birth/death for every person in the list but where that's still useful information to be able to disambiguate the listed people. ChristianKl ❪✉❫ 10:20, 16 January 2018 (UTC)
  • @ChristianKl: What? That is why Pernell Roberts, Saybrook University and Helen Schucman should all three link to Eleanor Criswell (and if that article gets created, the editor should make sure they all incoming links are actually to this specific Eleanor Criswell), or that all three articles should have a proper reference on the statement that talks about Eleanor Criswell from which readers can conclude it is the same Eleanor Criswell. We don't need WikiData for that (especially since we know that there is so much unreliable data on WikiData, that I would not be surprised that the WikiData Eleanor Criswell could very well be a mix of several persons with that name). --Dirk Beetstra ^{T C} 10:28, 16 January 2018 (UTC) (screwed up ping. --Dirk Beetstra ^{T C} 10:29, 16 January 2018 (UTC))
    • A normal redlink doesn't provide any information about how multiple links with the same name are actually about the same person. The ability to disambiguate between multiple people with the same name (and handle multiple names of the same person) is useful. ChristianKl ❪✉❫ 17:14, 16 January 2018 (UTC)
  • Created Eleanor Criswell and Eleanor Criswell Hanna as redirects to Saybrook University, where this person is now mentioned in the lead as founder of the university. Note that the Wikidata item of that university (Wikidata:Q7429134) links to the Wikidata item about this person. --Francis Schonken (talk) 10:56, 16 January 2018 (UTC)
    • In this case I think that's an okay solution but I'm not sure that it's a good general solution for every case in which two Wikipedia articles talk about the same person. ChristianKl ❪✉❫ 17:23, 16 January 2018 (UTC)
• Comment I've made a proposal below to use the information in Wikidata without sending our readers there. I'd want to ask editors supporting the removal of links what they think about this alternate proposal. Diego (talk) 10:44, 16 January 2018 (UTC)
• Oppose Though I think the link should be in some way distinguishable from a standard wikilink. We're sending readers to a totally different looking page and that's pretty confusing; most readers will have never heard of Wikidata. A small icon or something that could flag this as a link that leaves Wikipedia would be a solution. As is often the case with Wikidata discussions on Wikipedia, 'dont use it at all ever' just isn't a convincing argument for me here. Sam Walton (talk) 12:40, 16 January 2018 (UTC)
• Support per Jayron32 and David Eppstein. I'm fine with the link in the sidebar, but we shouldn't be linking in the article text or external links to wikidata until it's closer to meeting our policies. Ealdgyth - Talk 14:01, 16 January 2018 (UTC)
• Support Partial support. I've tried making nuanced arguments over the last year, I've tried engaging people in discussion. Nothing works. I'd be happy to see honest discussion over possible integration with Wikidata, but we never seem to be able to get there. In general, I distrust drastic measures like this one, but in this case, we've reached an impasse between two communities with different, and often conflicting, visions. The best way forward is to push the reset button and start over. - Dank (push to talk) 15:08, 16 January 2018 (UTC) Tweaked to "partial support", more later. - Dank (push to talk) 01:41, 19 January 2018 (UTC)

  Could you please explain what do you mean to "start over"? Start what over? If all links to Wikidata get banned (which is what you just supported), there is nothing to start anymore.--Ymblanter (talk) 15:18, 16 January 2018 (UTC)

  Across hundreds of discussions, it's become clear that opinions and positions have solidified into two or more camps that are intent on warring with and defeating all opposing camps. This isn't the way to construct a workable, shared community; it's the path to degrading and demeaning the people and policies of an existing community (such as it is, with all its flaws). I'm not psychic and I can't predict if there is some less intrusive plan for how to integrate Wikidata into Wikipedia that would respect Wikipedia's policies and ultimately work for everyone involved, but I hope there is. I feel your pain, and I'm not happy that it's come to this. I only know that being realistic and honest about how we've failed so far is the only way forward. - Dank (push to talk) 15:50, 16 January 2018 (UTC)

  I do not have any pain, I am not heavily using Wikidata in my English Wikipedia activities, and as a Wikidata user we have a number of projects actually quite happy with Wikidata such as Wikivoyage or external users. It is just my point, which I tried to get across below but only got personal attacks so far, is that the only way data could propagate from Wikidata to the English Wikipedia is that the Wikipedia users must find some way of filtering acceptable data. Right now, for whatever reason, they expect that Wikidata would filter it for them. If the result of the RfC is "to ban Wikidata", these discussions are not going to happen, and there obviously will be no import in any form. This might be ok (at least one user stated he is quite comfortable using Google) but we all must be quite conscious what this outcome actually implies.--Ymblanter (talk) 16:02, 16 January 2018 (UTC)

  The same old invalid arguments get made over and over. The first few times, I'm thinking, maybe there's information I don't know, or maybe they misspoke by accident. I'm past that point now, and I think probably others are too, which is why this has turned into a battleground where people aren't responding to each others' points, just as you didn't respond to mine. But I'll respond to yours: an individual editor can easily get information from Wikidata into Wikipedia, by copying that information into Wikipedia and taking responsibility for their edit; there are also ways to automate edits at high speed, but again, the individual editor is taking responsibility for the edits, following established policies and cultural norms, or else they face the consequences. That's not what happens when information is imported from Wikidata without being held to Wikipedia's standards. And what's "obvious" to you, that the only possible outcome if this RfC passes is that everyone stops working on these problems, isn't obvious, or even likely. And no, my position isn't black-and-white, even though the RfC statement that we're voting on is black-and-white: that's how it works around here. "Support" comes closest to my position, so I'm supporting. I'm not hostile to Wikidata or to people who identify with Wikidata. I want to see successful integration. I want to see Wikidata people getting a better understanding of what it means to be a Wikipedian, and learning more about our policies. I want to see higher quality discussions on these topics. Over the past year, it's been a solid wave of IDHT, across a wide variety of topics. If we can't start from a place of honesty and shared values, then there's no point in even discussing integration. - Dank (push to talk) 17:31, 16 January 2018 (UTC)

  I have no issues with you supporting, I just found it strange that you think smth would "start over" if this RfC is closed as "ban". I now see what you mean, though I clearly disagree with you. Btw I do not think you responded to my point, not that I expected this. And, to be honest, most of the "Wikidata people" you are referring to are established Wikipedia editors. If you are specifically addressing me, I am here for 11 years and have made 105K edits, most of them in the articles.--Ymblanter (talk) 17:40, 16 January 2018 (UTC)

• Oppose blanket ban. Can clearly sometimes be useful with {{Ill}} to identify which of various people with a particular name a redlink refers to, especially if the entry ties that identity to unique identifiers in multiple reputable biographical databases, eg VIAF, ULAN, RKD, Library of Congress etc etc. That's a useful form of clarification. Jheald (talk) 14:35, 16 January 2018 (UTC)
• Oppose Though we should be distinguishing between different types of links (like we do with other Wikimedia projects), wholesale removal of these links seems counterproductive: we should be pointing people to any reasonable information about them... Wikidata is one of those places. Sadads (talk) 14:35, 16 January 2018 (UTC)
• Comment - This subsection could use some clarification at the top. Surely it doesn't actually mean Wikipedia should never link to Wikidata, as we have handy links in the sidebar that don't create many of the problems people have listed. It's worth mentioning that this, as I understand it, is specifically about linking to Wikidata via links in the body and/or external links section of an article. If that's the case, then I've yet to see a compelling reason to do so, but at the same time I'm reluctant to support a "never". — Rhododendrites ^talk \\ 14:37, 16 January 2018 (UTC)

  Note that there are a few more cases, such as for example {{Commons category}}, which will have to stop retrieving data from Wikidata if this proposal succeeds.--Ymblanter (talk) 14:44, 16 January 2018 (UTC)

  @Ymblanter and Rhododendrites: The problem is this isn't a proposal. This is a sentiment that has ballooned into a poorly constructed "vote." -- Fuzheado | Talk 16:12, 16 January 2018 (UTC)

  This is clear, but I do not see any procedural way to stop this, and most supporters are clear that they support the blanket prohibition with a possible exception of interlanguage links.--Ymblanter (talk) 16:16, 16 January 2018 (UTC)

• Oppose. Absolutely absurd to hold Wikidata to a higher standard than any other website linked to by Wikipedia, other Wikimedia projects like Commons and Wiktionary, or Wikipedia itself. Gamaliel (talk) 15:17, 16 January 2018 (UTC)

This seems (or atleast started about) to be about body links to wikidata; we do not allow external links in the body. Galobtter (pingó mió) 18:35, 16 January 2018 (UTC)

we do not allow external links in the body. - Except when we link to Wikitionary, or Wikisource, you mean? Diego (talk) 19:08, 16 January 2018 (UTC)

• Oppose and Protest this badly formed RfC. This so-called Request for Comment was a casual question that started with "I don't know much about wikidata." And now we're voting support/oppose on a global policy in English Wikipedia that would effectively ban linking forever? No way. -- Fuzheado | Talk 15:55, 16 January 2018 (UTC)

@Fuzheado, Rhododendrites, Pigsonthewing, Guerillero, Galobtter, and Mike Peel:- I wrote the question at the beginning to get some feedback but I did not create the RFC. The RFC was created by User:Richard Arthur Norton (1958- ) in a way that was designed to skew the results. The issue should be whether or not we can link to wikidata in the body of the article, not never link to wikidata.--Rusf10 (talk) 18:43, 16 January 2018 (UTC)

• You can't read minds so please AGF and do not write that my motive "designed to skew the results". You took four people's opinions in 24 hours and began removing links. A formal RFC runs for 30 days. --RAN (talk) 19:38, 16 January 2018 (UTC)

I know; The RfC tag should be in the "Never link to Wikidata" section as the neutral statement and should be limited to linking in the body. Galobtter (pingó mió) 18:47, 16 January 2018 (UTC)

#Link to Wikidata in existing tables if they are not getting an article should remain part of the RfC imho: this part of the RfC would become more important again if the #Never link to Wikidata part would fail to find consensus one way or another. --Francis Schonken (talk) 18:59, 16 January 2018 (UTC)

• Oppose this non-neutrally worded and ill-conceived RfC, per Andrew Lih and others. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:59, 16 January 2018 (UTC)
• Support for the *style* issue, based on WP:EGG (which is guidance found on another manual of style page): clicking a bluelink in the body of a mainspace page (article), and getting somewhere outside Wikipedia is a WP:EGG problem and we should avoid it. *Transclusion* or other invisible/unclickable links are not a style issue: they are content issues, and thus fall outside the remit of the Manual of Style discussion page. Links that clearly identify themselves as external (e.g. by being in an EL section) or as going to Wikidata (e.g. when the {{ill}} template is correctly applied) are not problematic for style guidance such as WP:EGG afaik. --Francis Schonken (talk) 16:48, 16 January 2018 (UTC)
  • Modified to broader support: I do think "(Wikidata)" appearing mid-text or in a table or list to be bad style: Wikidata is not a household name (yet) like most two- or three-letter abbreviations for languages are, so the "Wikidata" term appearing with a link in the body of an article would still be more confusing to the average reader than most other usages of the {{ill}} template. --Francis Schonken (talk) 13:27, 22 January 2018 (UTC)
• Oppose Could you have a more non-neutrally worded RFC? I can't roll my eyes enough at the idea that we should never link to a sister project --Guerillero | Parlez Moi 16:51, 16 January 2018 (UTC)
• Oppose per Sadads, Fuzheado, Gamaliel, Guerillero, and others. --Rosiestep (talk) 17:00, 16 January 2018 (UTC)
• Oppose per many people above. There's a big disconnect between the initial question and the headers here, so the original requester may want to pull this RfC and try again with the narrower question. In the general case, those objecting to them could always create the missing Wikipedia articles/redirects and link to those instead (where notable)... BTW, are we supposed to !vote in the section above as well? Thanks. Mike Peel (talk) 18:25, 16 January 2018 (UTC)

• @Mike Peel: The cases with which this discussion opened - which may not even matter since we've gotten so far off that topic, but FWIW - were where the articles had been deleted at AFD, and links to Wikidata were added instead. In such cases, creating the article is not a good solution, unless the reason for deletion could be addressed. Nikkimaria (talk) 18:36, 16 January 2018 (UTC)

  In the case of AfD, policy is very clear. In other cases, Mike's suggestion is workable. --Guerillero | Parlez Moi 19:01, 16 January 2018 (UTC)

• Yes, AfD policy via ATD is clear, but because administrators are not requiring policy-based !votes, we get the dysfunction in which redirects that wouldn't be deleted at WP:RfD, are deleted without due process at AfD.  Unscintillating (talk) 21:56, 16 January 2018 (UTC)

• This RfC really needs clarification. Galobtter (pingó mió) 18:35, 16 January 2018 (UTC)
• Oppose per Yaroslav, Andrew, and Guerillero specifically. mahir256 (talk) 19:06, 16 January 2018 (UTC)
• Oppose and speedy close as patent nonsense and per snowball, per Ymblanter, Fuzheado, and Guerillero — OwenBlacker (talk; please {{ping}} me in replies) 20:22, 16 January 2018 (UTC)
• Oppose I see no reasons mentioned except for the ignorance and stubbornness regarding the Wikidata (like that it does not contain useful information) or unrelated arguments (like that Wikidata can be vandalized or that users don't like the way information from there is presented in the article)--Ilya (talk) 21:12, 16 January 2018 (UTC)
• Oppose  I'd never looked at Wikidata until about an hour ago, but fail to see why I should be able to link to articles on de.wikipedia.org, which does not support our policies; and add links to wiktionary, but not wikidata.  Unscintillating (talk) 22:03, 16 January 2018 (UTC)
• Oppose. Without addressing the specific merits of this case, procedurally, a blanket ban on linking to a sister project needs a better formatted and advertised RFC. ---- Patar knight - ^chat/_{contributions} 22:50, 16 January 2018 (UTC)
• Support. Interwiki links served up on the left sidebar are fine. Replacing red links in articles with links to Wikidata is not good. Wikidata is a free, collaborative, multilingual, secondary database, collecting structured data to provide support for Wikipedia, Wikimedia Commons, the other wikis of the Wikimedia movement, and to anyone in the world. Right, it's a secondary database. Secondary databases obtain data from primary databases; they don't deliver secondary data to primary databases. That's the equivalent of "citogenesis". Here's how Wikidata can use its structured data to provide support for Wikipedia: I want Wikidata to tell me when "the birthdate of John Doe in English Wikipedia is not consistent with his birthdate in the French, Spanish and German Wikis. When the place of birth of Jane Doe in English Wikipedia is different than Jan Doe's birthplace in the Portuguese, Japanese and Russian Wikis. Flag these for human, or bot attention. Wikidata can be useful as a quality-control vehicle. Its data elements could have a measure of confidence. When there is universal agreement on the value of a particular data item from a large number of independent wikis, then there is high confidence in the item's accuracy. When there is disagreement, the level of confidence is lower. This is needed support to ensure higher quality of a reference that anyone can edit. – wbm1058 (talk) 04:39, 17 January 2018 (UTC)
• Oppose. I'm okay with the idea of "no links to Wikidata in prose from the article space", but the way the question is worded is vague and could potentially be interpreted to mean no use of Wikidata at all, which I most certainly do oppose. Lankiveil ^{(speak to me)} 11:00, 17 January 2018 (UTC).
• Oppose procedurally, and also on the grounds that Wikidata links are better than nothing. As above I would support only linking to the interwiki links sections on Wikidata items, since Wikidata is a developing project, concerns about unsourced content etc. Do we also have to ban linking to articles (in other Wikipedias, and/or this Wikipedia) with {{unreferenced}}? (Note that Wikidata currently has an open RfC on privacy and living people.) Jc86035 (talk) 12:22, 17 January 2018 (UTC)
• Oppose. Perhaps if this were phrased differently and didn't give the impression of banning Wikidata everywhere ("[...] we should not link to it. At all."), I might be able to support this. I disagree with the idea of prose Wikidata links, but this may set a precedent for other interwiki links (and Wiktionary/etc can be useful some times). Anarchyte (work | talk) 12:35, 17 January 2018 (UTC)
• Support principle, oppose RFC statement. I quite agree that these WD links should not be in article text - but I thought we already had a long-established consensus to avoid, say, inline links to Spanish Wikipedia articles in the middle of sentences? Surely all we need to do is to formalise that existing position with a mention of WD - "Links to entries in other language Wikipedias or Wikidata items should not be used in running article text", put it in the MoS list of links to avoid, and we're done.

But titling the RFC position "Never link to Wikidata" seems to be just asking for trouble - in a year or two some people will look back at that header and choose to interpret it as an RFC on ever crosslinking to Wikidata, whatever the context, even if most commenters here are considering the inline links issue. That won't help anyone and will simply serve to entrench further conflict. I don't think it was deliberately intended that way, but as written it's playing into some very established community rifts (as Dank notes above), and we should try and avoid encouraging that. Andrew Gray (talk) 13:12, 17 January 2018 (UTC)

• Support Wikidata is a slow cancer eating away at WP. Lugnuts ^{Fire Walk with Me} 18:26, 17 January 2018 (UTC)
• Support. Wikidata can be used to host original research and unsourced claims, and allow such faulty information which cannot come in by the front door, instead enter by the back door. If there are valuable data missing from an article, it should come in only if properly sourced and added directly. And as to the assertion that something is better than nothing: No. It is not. Better to have no assertion than one based on faulty information or personal opinion. We do not need to compound errors by repeating them. Kablammo (talk) 19:03, 17 January 2018 (UTC)
• Support per my comments here. ‑ Iridescent 19:17, 17 January 2018 (UTC)
• Support per numerous editors above pointing out the dangers of such links more calmly than I can. Yngvadottir (talk) 22:30, 17 January 2018 (UTC)
• Support. Wikidata is a deeply flawed turd in the project, with too much unsupported information to be safe to use. - SchroCat (talk) 23:18, 17 January 2018 (UTC)
• Support. Wikidata should be clearly identified as an unreliable source and never ever used for anything on WP. It is absolute rubbish.Smeat75 (talk) 00:56, 18 January 2018 (UTC)
• Support - I'm very wary of Wikidata and the information it hosts, and until further notice I think its best to keep it away from this project. -Indy beetle (talk) 03:59, 18 January 2018 (UTC)
• Oppose blanket ban. I'm not specifically defending the links at issue; I haven't looked into them. But I don't think it's a very good argument to say that if a subject doesn't pass WP:N, we can't have a link to it. The point of N, or at least part of the point, is to avoid cluttering the index space on en.wiki — too many articles make for excessive disambiguation and reduce the utility of completion in the search box. That's not really an issue with links. A WD link could help users find the right instance of a search term, which might be more difficult with a general web search. Side note: Isn't this RFC in the wrong place? It's not clear that it's a style issue. A lot of people are making arguments that seem to be more about general policy than style. --Trovatore (talk) 05:00, 18 January 2018 (UTC)
• Support not linking to Wikidata in prose in mainspace. This is purely from a "path of least astonishment" perspective. Does any reader expect to wind up on Wikidata when they click a link in mainspace? Of course not. Nor do they really want to end up on Wikidata. Wikidata is great for many things, but human readability is not one of them. ~ Rob₁₃^Talk 09:16, 18 January 2018 (UTC)
• Support per Jayron. ​—DoRD (talk)​ 12:23, 18 January 2018 (UTC)
• Oppose. It is absurd to have a policy on a Wikipedia edition that it shall blacklist another Wikimedia sister-project. Shall we refuse to link to Commons next? It's not like they don't have their own content problems after all, so the same 'their policies aren't compliant with our policies, and some of their content is bad quality' argument could be used against Commons too. Wittylama 13:16, 18 January 2018 (UTC)
  • ​Er, we almost never do link to Commons. The question being asked isn't "do we include links in the EL section as we currently do with Commons?", it's "do we include wikilink terms within the text of an article to their Wikidata entries?". ‑ Iridescent 15:51, 18 January 2018 (UTC)
    • Iridescent: There are other parts of this discussion which have somewhat more nuanced points (e.g. about usage in tables, or when there's a redlink, or within the prose), but THIS section of the discussion is voting on a blanket ban on ANY Wikidata links - the subject heading is literally "Never link to Wikidata". Wittylama 15:59, 18 January 2018 (UTC)
• Oppose. The solution proposed is not a proportionate response to the issue raised. We should deal with compatibility issues on a case-by-case basis, not reject entire Wikimedia projects where a significant amount of work is being done by long term Wikipedians with the goal of supporting Wikipedia. Battleofalma (talk) 16:00, 18 January 2018 (UTC)
• Oppose—Linking to Wikidata is smart and logical way of dealing with the fact that Wikipedia is not a directory and has higher standards of notability than Wikidata. Especially in cases where Wikipedia provides a table or list of people, entities, or places, connecting them to individual Wikidata entries is the appropriate way to archive factual information about those entities, while preventing them from cluttering up things here. (To take an example I worked on, the muncipal council members of Sucre should get Wikidata entries, but not Wikipedia pages.) Yes, Wikipedia and Wikidata have some different policies. In some cases Wikidata policies should be moved towards Wikipedia's (Verifiability, BLP), but not on this front. Less notable data should not be part of Wikipedia, but it should be linked so as to enrich our content here.--Carwil (talk) 16:16, 18 January 2018 (UTC)
• Oppose per WP:CREEP. Per WP:BOLD, we should never say never. Andrew D. (talk) 16:21, 18 January 2018 (UTC)
• Oppose: I am sure there are times when liking to WD would provide the reader with useful additional information to compliment what WP has to offer. Data-related stuff is not really in my area of knowledge, though. –Sb2001 19:53, 18 January 2018 (UTC)
• Oppose. WD is the only chance for WP to deal with certain very „dynamic“ data. Ziko (talk) 05:15, 19 January 2018 (UTC)
• Support Comment as the RfC is not sufficiently well worded. The links currently do not specify that they are to Wikidata per here which will confuse our readers / editors. We should not IMO be replacing what looks like links to articles here with WD links. Once this is fixed we can than discuss the uses that may be reasonable. These should be spelled out before implemented however. Doc James (talk · contribs · email) 11:02, 19 January 2018 (UTC)

  Looking at this issue further and links to all sister sites appear the same as internal Wikipedia links. Not sure if their is a phabrictor ticket to have this adjusted?

  And to clarify further we already have Wikidata links in the left sidebar of all our articles plus we have wikidata links in some infoboxes which we should keep. When WD links appear as external links that is also fine. Doc James (talk · contribs · email) 19:31, 19 January 2018 (UTC)

• Oppose blanket ban. Links to Wikidata should be used cautiously as with any other inline interwiki link. Deryck C. 18:11, 19 January 2018 (UTC)
• Oppose. I share a number of concerns about how Wikidata gets used in Wikipedia. (Indeed, I just generally have concerns about the level of sourcing on Wikidata). But banning links from en.wp to Wikidata isn't in the best interests of either project in the long term. The Land (talk) 19:41, 19 January 2018 (UTC)
• support. The sidebar link serves to link the article to its Wikidata page, so there is no need for any more in the body of the article. Linking to any other Wikidata page is an even worse idea; if the Wikidata page has a corresponding article then link to that if it makes sense. If there is no such article then a red link should be used, if appropriate. In neither case is linking to Wikidata better. Such would just confuse and frustrate readers, and disrupt the process of building the encyclopaedia that red links are part of.--JohnBlackburne^words_deeds 21:43, 19 January 2018 (UTC)

  @JohnBlackburne: I think the idea is in some cases to show a red-link and an icon or reduced-sized link to Wikidata, eg in a list of people, to clarify who the redlink is for, as identified in external databases. But not a replacement for red-links, and only to be used occasionally and with discretion. Jheald (talk) 22:02, 19 January 2018 (UTC)

  That still has two problems. First the link to Wikidata will still confuse and frustrate readers, given Wikidata’s current UI of a badly formatted multilingual database. Second it encourages editors to create redlinks for the wrong reasons. Currently a redlink means someone thinks it could be an article. But if they are associated with Wikidata links then editors will add red links not because there might one day be an article, but because there is a Wikidata item and they want to link to it.--JohnBlackburne^words_deeds 00:57, 20 January 2018 (UTC)

• Oppose this overly broad proposal. Enterprisey (talk!) 09:09, 20 January 2018 (UTC)
• Comment the formatting of this RFC is leading an ambiguous outcome, the RFC question is about not linking to Wikidata in the body of the text, but the heading of this section is Never link to Wikidata. Comments from people supporting this RFC suggest that some people think they are voting for an overall ban of Wikidata linking on English Wikipedia and some people think they are voting on not having links within articles. One way to remove this ambiguity is to change the subsection heading from Never link to Wikidata to a concise statement that summarises the RFC statement. John Cummings (talk) 10:12, 20 January 2018 (UTC)

We cannot use Wikidata as a source in English Wikipedia, just as we cannot use Wikipedia or other user-generated sites. It is one think to have a link in the left sidebar; but another thing entirely to use Wikidata in Wikipedia infoboxes or any other part of the article. Our standards for reliable sources do not allow that. Kablammo (talk) 16:11, 20 January 2018 (UTC)

Using Wikidata to disambiguate people who don't have their own Wikipedia articles isn't using Wikidata as a source. ChristianKl ❪✉❫ 23:03, 20 January 2018 (UTC)

You are confusing using Wikipedia/Wikidata as a source for a fact, and using it to identify someone uniquely. We link people to their names in other Wikipedia articles all the time, that is what a blue link is for. That is not using Wikipedia as a source for a fact. --RAN (talk) 15:00, 24 January 2018 (UTC)

• Oppose blanked ban. Links to wikidata should be ok least for the cases where target is clearly marked. Zache (talk) 10:16, 20 January 2018 (UTC)
• Support Wikidata's mission is admirable, but it's too nascent to reliably support en.Wikipedia with supplemental data. While sourcing has been improved (according to WMF), there's still a massive gulf between its current state and what I'd consider a reliably sourced data repository. I'm also concerned with Wikidata's reported use by PR professionals and SEO firms for their own ends. Widespread support on en.Wikipedia would only aid their efforts. While I'd like to see something like this adopted in the future, it's too soon to greenlight this practice. - GS ⋙ ☎ 06:07, 21 January 2018 (UTC)
• Support. Wikidata is uncurated and unreliable. It can be host incorrect perhaps defamatory BLP claims. Strong support for blanket ban to biographical links and links in BLPs per WP:BLPSOURCES. LK (talk) 00:55, 22 January 2018 (UTC)
• Support Wikidata is a wiki that anyone can edit. It is thus an unreliable source and cannot be used to support any content in article space. I support a bar from using Wikidata in article space. --Malcolmxl5 (talk) 14:17, 22 January 2018 (UTC)
• Support. Per JohnBlackburne and GrapefruitSculpin. Kaldari (talk) 21:43, 22 January 2018 (UTC)
• Poorly worded - I cannot support such generic statements without a clearer proposal. But I understand and can agree with parts of the sentiments expressed. You'll have to come up with something more nuanced to get my support though. effeietsanders 22:14, 22 January 2018 (UTC)
• Support for all the good reason above. Wikidata is fundamentally flawed and until its deep-seated problems are sorted out it should only be dealt with at arms length. - SchroCat (talk) 23:16, 22 January 2018 (UTC)
• Support. The status quo is intolerable. It's a continual disruptive battleground, with both sides throwing all their work down the toilet as enthusiasts roll wikidata forwards and critics roll it back. We either need to go full-on wikidata wherever it's usable, or not at all. Those are the stable outcomes. And if we go full-on wikidata, god help us. Per abundant arguments above, nuke it from space (i.e. phab rollback the wikidata commands that were added to wikitext). It's the only way to be sure. We can keep the left-sidebar wikidata interlanguage links. Alsee (talk) 17:57, 27 January 2018 (UTC)
• Support per SMcLandish's astute observations at the top of this, and others'. The principle of least surprise is the biggest one, and it's a HUGE fucking surprise to land on a page that is effectively not readable without a bunch of investigation into where you've landed, what it means, etc. Aureliano Babilonia (talk) 04:48, 28 January 2018 (UTC)
• Support per SMcCandlish. -- Tavix ^(talk) 22:05, 1 February 2018 (UTC)
• Oppose. Linking to an item in Wikidata provides a kind of informational anchor and a helpful starting point for readers who want to understand particular names or phrases in a Wikipedia article. Here is a recent example of how Wikidata linking improves a Wikipedia article: Budapest Open Access Initiative. -- Oa01 (talk) 09:43, 13 February 2018 (UTC)
• Support blanket ban on principle, but IMO this is not a "style" issue and should be enshrined at WP:ELNEVER. Hijiri 88 (聖やや) 09:52, 13 February 2018 (UTC)
• Support. It's not useful for a name in a list to offlink to WikiData in lieu of an internal Wikipedia article. The Wikidata link doesn't add any useful information above and beyond what was already gleaned from the list that the person's name was included in — and, in fact, a WikiData link for any given person will only exist if we formerly had a Wikipedia article but it got deleted. A Wikidata entry doesn't get created without the prior existence of a Wikipedia article on at least one language wikipedia, yet the deletion of the Wikidata entry does not automatically follow from the deletion of the article. The primary purpose of Wikidata is to assist with the authority control data and the crosslinking of Wikipedia articles in different languages. An entry in there isn't informative at all, and it's not a resource that people independently consult in its own right for information about people or things — it's just a technical tool to simplify some Wikipedia editing tasks, nothing more, so offlinking a former article topic to their ghost wikidata entry as a replacement for the deleted article isn't contributing anything useful to anybody's knowledge of anything. I'm not suggesting that it's useless, as it clearly does have some very useful purposes — but serving as an alternate body text link for a person or thing that doesn't have a Wikipedia article to link to is not one of those. Bearcat (talk) 21:32, 13 February 2018 (UTC)

• @Bearcat: Your point about the creation of Wikidata items is incorrect - it's quite possible for an item to exist without a Wikipedia article having existed previously. Items can be created which meet one of three criteria outlined at WD:N - one of those is a valid sitelink, but the other two don't rely on that. That's why, for example, Wikidata items exist for individual scientific articles which would never meet our notability policies. Nikkimaria (talk) 02:24, 14 February 2018 (UTC)

• The context for my comment was that when articles about smalltown mayors are deleted, certain people are inserting links to their ghost wikidata entries to the town's overall list of mayors as a replacement for the deleted redlink. I can't speak to wikidata items existing for scientific articles, because I don't work in that domain — in the contexts where I encounter Wikidata links being inserted directly into body text, the prior existence of a now-deleted Wikipedia article is how the Wikidata number came into existence. Bearcat (talk) 02:30, 14 February 2018 (UTC)

• Fair enough, just wanted to make clear that it would be possible to go create a Wikidata item on a small-town mayor without ever having an article here, deleted or no. Nikkimaria (talk) 02:41, 14 February 2018 (UTC)

• Oh, I realize that it's possible. It's just that nobody ever would, because in the absence of a Wikipedia article about the smalltown mayor there'd be no purpose to doing so. Bearcat (talk) 02:51, 14 February 2018 (UTC)

• Well, User:Bearcat, that is in a way what WikiData is supposed to be. If it is a fact that John Doe is the king of FarFarAway, then that is data that WikiData wants to capture. Their database could contain ALL US mayors for every imaginable town from start of time till now, with tenure, town, etc. WikiData does not exist only for use on other Wikipedia's, its purpose is to be a database. (and part of the problem is that people on WikiData would make those entries, because it is data they want to capture and to be used - and for some data is it (was) more important to capture it than to have it locked down as reliable - they have no clue that the John Doe that was the mayor in FarFarAway is the same John Doe that invented the light bulb - they may have two the same John Does in their database, or two independent John Does who are the same person - and one might not be able to discern them in some cases). --Dirk Beetstra ^{T C} 05:55, 14 February 2018 (UTC)

• How frightfully pointless. Still doesn't mean that a list of mayors in Wikipedia should be linking to their Wikidata entries as substitutes for Wikipedia articles, though. Bearcat (talk) 06:02, 14 February 2018 (UTC)

• Indeed. Knowing that it may take extensive work to link (or de-link) the two John Does from the two different lists of people, and that that is probably not extensively done by WikiData people (if it is even possible in some cases) that may result in significant BLP issues. Now, on the other hand, I don't know whether it is significant for en.wikipedia to know whether the Mayor John Doe mentioned in the town article is the same as the inventor John Doe that is mentioned in the light-bulb article. --Dirk Beetstra ^{T C} 07:28, 14 February 2018 (UTC)

• Re. "It's just that nobody ever would" – incorrect, and I can't stress this enough: e.g. this Q item was created yesterday at Wikidata, by an editor who is currently blocked at en.Wikipedia, without *any* article or item on *any* other Wikimedia project. By this indirect technique Wikidata has been used to smuggle topics that wouldn't pass WP:GNG into en.Wikipedia (just dig a bit deeper than I just did and you'll see that some of the recent successful AfDs had a similar origin, just absorbing editor time over and over with no net benefit to Wikipedia). --Francis Schonken (talk) 06:54, 14 February 2018 (UTC)

• Support. Per SMcCandlish. Mike Christie (talk - contribs - library) 22:40, 18 February 2018 (UTC)

Discussion (linking to Wikidata RfC)

What I would like to comment about here is the statement (actually, several statements) in the above section that Wikidata is universally not useful except possibly for interlanguage links. May I please remind the community that we had (and still have) a severe problem that our information quickly gets outdated. If (a random example) a mayor of Diyarbakır gets reelected (the example is so random, that I did not even look up whether they have a mayor and whether this is an elected post), the information about a new mayor is unlikely to propagate to the English Wikipedia any time soon. It can be propagated first to the Turkish Wikipedia (which is understaffed due to the Wikipedia ban in Turkey), and then someone may be so nice and come here to update the article - or may not, depending on whether they speak English and feel comfortable to edit the English Wikipedia, whether they had enough coffee in the morning, and whether they have promised their girlfriend to go to a movie tonight. Indeed, I am currently going through articles on Ukrainian localities and occasionally remove information about their mayors which someone added there 10 years ago and never cared to update. In the pre-2012 era it was actually recognized as a problem and there were some discussions on how to handle this problem - not so many, because who cares about mayors of cities outside English-speaking world. One example I remember was the project organized by WereSpielChequers with the goal to follow whether a person is designated as living on some projects and as dead on other projects. At the time I thought this was a brilliant idea. It worked by using a bot which was checking info on different language versions on Wikipedia and made project-specific lists. It was fully superseded by Wikidata (which is, at the end of the day, still more reliable than just a collection of data from different language versions on Wikipedia), and, as far as I understand, stopped working. One (not the only one, but I would dare say the main one) idea of the early Wikidata was to host all this information centrally, so that updates do not take ages to propagate from one project to each other and get stuck there, possibly without sources. Now, whoever votes above "never use Wikidata", basically undersigns that they failed to accomplish this task. The task does not necessarily have to be accomplished by direct import of data via templates (and at this stage I would discourage the direct import because of the vandalism issues), but it can be dome differently, for example by using bots and creating lists off the main space. But by voting "never use" you guys send the project - not Wikidata, but the English Wikipedia - back to the middle of the 2000s. (For the record, I do not think there are many advantages in using Wikidata in the case from which this RfC started - the Wikidata entries about mayors may have useful info such as for example DOB reliably sourced to databases, but indeed nobody cares, since they are not notable for Wikipedia, and Google gets the same results faster; if they are notable, I agree it should be a redlink).--Ymblanter (talk) 07:50, 16 January 2018 (UTC)

Courtesy @WereSpielChequers:, the ping failed the first time.--Ymblanter (talk) 07:52, 16 January 2018 (UTC)

I do agree to the sentiment expressed here - but thinking in real BLP-level situations: what if a living person's data here is extracted from WikiData, and someone is changing the WD data it results in BLP violations here; I can block someone from changing data here but I can't block someone on WikiData from changing data here. Then there are cases of incongruent data (the way that WD is displaying data is not how we want to display it, or the data can be displayed in multiple ways while all being correct - http://example.org; https://example.org; http://www.example.org .. all correct, all different - try to get that 'aligned' with a local choice, while other wikis chose another option), or data that we cannot display but comes in through the WD-backdoor (blacklisted external links are one example of that).

If the WD's own policies are not violated, then there is nothing that we can do to not have our policies (or even global decisions) violated except for breaking the link with that WikiData item (which, at some point, becomes impossible to maintain). --Dirk Beetstra ^{T C} 08:08, 16 January 2018 (UTC)

If you want my personal opinion, we should have bot transfer and a system similar to pending changes (or may be even incorporate it into pending changes). Such edits typically need reliable sourcing anyway, which can seldom be recovered from Wikidata.--Ymblanter (talk) 08:12, 16 January 2018 (UTC)

May be I should expand on this a bit. The main problem of Wikidata is that it is understaffed, and I do not expect this to change any time soon. I do not expect Wikipedia editors to start editing Wikidata on a regular basis, even if the interface becomes very simple (which is kind of what we are close to). On the other hand, this is a real problem which needs to be solved. IMO the only way in principle to solve it is to shift it to the projects whereas kiiping at the same time Wikidata as a central repository of information. Bots are probably the easiest way to do it, though I am pretty open to other workable solutions.--Ymblanter (talk) 08:20, 16 January 2018 (UTC)

Btw I can block someone for vandalism on Wikidata (and do). If you see instances, especially coming from registered users, which require administrative intervention, pls let me know.--Ymblanter (talk) 08:13, 16 January 2018 (UTC)

That is great .. but just a small fraction of the problem (though in case of BLPs and banning cases a rather big problem - though I doubt you can block an editor on WikiData who is blocked here just because he is blocked here for making such changes). You can not solve the dissimilarity problem (except through big hacks in every single wiki's translusion template for each parameter that would suffer that problem, or defining the same parameter on WikiData 800+ times (once for each Wiki)), nor the problem that WikiData can have data that we transclude here that here gives problems (e.g. solve this: blacklist an url here, then add it to WikiData so that it transcludes here and then try to edit a page that transcludes the data here ..).

Regarding sending Wikipedia back to the middle of the 2000s .. that would not have happened if editors would not have been so eager to delete the data here after having it on WikiData - then breaking the link would not have resulted in loss of data here (except that we don't display the data anymore that was since added to WikiData .. which actually we never had here anyway). --Dirk Beetstra ^{T C} 08:51, 16 January 2018 (UTC)

I do not know - in the example with Shrivallabh Vyas the original vandalism was on the Wikipedia side [2], and if the guy were still alive, we would return to the same vandalized article now. In any case, all this information can be imported back to Wikipedia overnight, I just do not see any point in doing this. We can outline complex templates as well and go back to 2003 - but would anyone benefit from this?--Ymblanter (talk) 09:10, 16 January 2018 (UTC)

Short answer: yes. Except for a couple of things (interwikis), most data on WikiData is similarly incorrect as here, and as I and Fram outline, there is data on WikiData that we decided already not to use (but which is difficult to filter out - is WD's date-of-death referenced to a NYT obituary: transclude - is WD's date-of-death referenced to Find-a-Grave: do not transclude). Is http://www.example.org the same as what WD mentions to be https://www.example.org?). Go on and on. And all local editors have to find a solution, while it is so easy: can I find a reliable source for the date-of-death: no - don't edit, yes: write it and reference it. Someone adds a date-of-death with a Find-a-grave reference: revert (or better, check and use a proper reference if available), now I have to jump through hoops to see whether the date-of-death is properly referenced on WikiData (and as it is referenced, we will display it). Can you see how easy it is to vandalise en.wikipedia? And I haven't even started about those excessively high-speed bots that are running on WikiData that could wreak havoc on thousands and thousands of pages on 800+ wikis in a short time if they would have an undetected bug. --Dirk Beetstra ^{T C} 09:25, 16 January 2018 (UTC)

I do not see how it contradicts to the soultion I proposed. One possible implementation (not the only one and possibly even not the best one): Take a bot; let it find an article which has no date of death, whereas the Wikidata item has; if Wikidata item is unsourced or sourced to Wikipedia -> make a bot leave a message at the talk page; no infobox -> make a bot leave a message at the talk page; if the Wikidata item is sourced and there is an infobox in the Wikipedia article -> have a bot edit the infobox and mark the edit as needed a human attention (via pending changes or differently, needs to be figured out) -> a human comes and sees whether this is NYT of Find a grave -> a human edits the article, everybody is happy.--Ymblanter (talk) 09:32, 16 January 2018 (UTC)

'Take a bot; let it find an article which has no date of death, whereas the Wikidata item has; if Wikidata item is unsourced or sourced to Wikipedia -> make a bot leave a message at the talk page; no infobox -> make a bot leave a message at the talk page; if the Wikidata item is sourced and there is an infobox in the Wikipedia article -> have a bot edit the infobox and mark the edit as needed a human attention (via pending changes or differently, needs to be figured out) -> a human comes and sees whether this is NYT of Find a grave -> a human edits the article, everybody is happy.' .. or just have a tracking category with 'hey, there is a WD item but no local data .. ' and have a human editor coming and add it here. 'have a bot edit the infobox and mark the edit as needed a human attention (via pending changes or differently, needs to be figured out)' .. so yes, whether through transclusion or through a bot, you still want to run the risk of transcluding BLP violations ...

Your example of Shrivallabh Vyas is nice .. say I vandalise a page here, my vandalism gets transported to WikiData and then transcluded on 800+ other wikis, here my vandalism is wiped, but because we are transcluding WikiData data, it gets back-transcluded until we also wiped the WikiData entries (and in the meantime, on 800+ other wikis N00bs might want to edit the data themselves because something wrong is transcluded). I am getting a headache. --Dirk Beetstra ^{T C} 09:40, 16 January 2018 (UTC)

Another solution might be: make sure that the WikiData data is of such a level that we can trust it. But that is not the quality control that WikiData has, worse, it is demonstrating to bot-import notoriously unreliable sources or plain spam. --Dirk Beetstra ^{T C} 09:47, 16 January 2018 (UTC)

Realistically, it is not going to happen without engagement of a much larger set of users than current Wikidata users, and they are obviously not going to be angaged if Wikidata is banned here. This is not a solution.--Ymblanter (talk) 09:57, 16 January 2018 (UTC)

(EC)Well there is the point. ENWP requires information that is compatible with its policies. Until Wikidata can provide that, its of minimal usefulness. I doubt anyone seriously expects that to happen anytime soon as I don't see a horde of wikidata users waiting to leap into action to tidy up its unsourced/badly sourced/non compliant info, so yes, there is no solution other than to not use Wikidata for almost all purposes. Only in death does duty end (talk) 10:17, 16 January 2018 (UTC)

Well, the notion that Wikidata is a service provider and Wikipedia is a customer is just entirely wrong. Wikidata is the data repository, and Wikipedia is not the only user of this data, and I believe not even the biggest user. Everybody uses data in different ways. It is up to the user to decide which data is acceptable and which is not. The data selection must happen on the Wikipedia side. It will never happen on the Wikidata side; it is perfectly acceptable in Wikidata (and will always be acceptable) to have different (contradicting) data cited to different sources. And actually Wikipedia users do select the information. Now it just happens via the Google search. If anybody thinks that Google is more reliable than Wikidata - fine, but this is just short-sightedness. You guys just believe that someone needs to bring you data and ask you to accept it - and when it does not happen, you are pointed out to a repository and get asked to select data yourself, you get upset and say that the repository is evil. Do you know Wikipedia is unreliable? Yes, sure. Is this morally wrong to cite data referenced to Wikipedia? I guess not. Do you expect this data to magically become reliable because it is not on Wikipedia anymore but on Wikidata? No, I guess not. Just do not use it. Use smth which is reliably sourced, and ideally check the source before using it.--Ymblanter (talk) 10:26, 16 January 2018 (UTC)

Yet another point to made. Exactly. And I think it is time that this customer decides not to use the data. --Dirk Beetstra ^{T C} 10:32, 16 January 2018 (UTC)

This is ok. Use Google if you think it is better. It sure never gets vandalized and always reports reliable information.--Ymblanter (talk) 10:34, 16 January 2018 (UTC)

No, but as I said just below, if I have the choice between going to WikiData and then having to use google to check whether WikiData is correct, or I have to use Google directly, then I chose the latter. --Dirk Beetstra ^{T C} 10:41, 16 January 2018 (UTC)

You are a physicist, right? You should know then that finding a solution and checking the validity of an existing solution ate two completely different things, of quite some different degree of complexity and necessary computational effort.--Ymblanter (talk) 10:45, 16 January 2018 (UTC)

Chemist, to be precise. I know, but I am also aware how a wrong DOE can completely go wrong. And I am sorry to say, but I am afraid that WikiData took the wrong approach - if they would have started to serve the Wikimedia community by serving them properly they would have engaged Wikipedians to supply more reliable data. And with that reliable data they would automatically have started to serve the community outside of Wikimedia as well. It is now such an entangled mess that it is hard to use for the Wikimedia community, and anyone outside of that. --Dirk Beetstra ^{T C} 10:56, 16 January 2018 (UTC)

(ec) I actually do agree that Wikidata from the beginning should have concentrated more on importing data from databases and less on importing data from Wikipedia, and only sort out these issues later. However, I do not think it is inherently wrong to store there data imported from Wikipedia. They should just not be reimported back here, and it is not actually difficult to prohibit such import for example at the template level, but of course nobody would do this if the template is likely to be attacked by a member of an anti-Wikidata brigade, with all this work destroyed overnight. Definitely, if one views Wikidata just as an extension of the English Wikipedia, which must deliver data in exactly the same format and which would conform with exactly the same policies (changing as well when policies change) as the English Wikipedia, then it is not reasonable to have it as a separate project, it should be just part of the English Wikipedia. Concerning more engagement, I disagree, there was no additional engagement on Commons (which is understaffed as well and would seriously benefit from an influx of more constructively-minded and less trigger-happy users) even despite the fact that the Commons requirements are stricter than the Wikipedia requirements.--Ymblanter (talk) 11:11, 16 January 2018 (UTC)

On the contrary, Wikipedia is the customer, and Wikidata is the service provider. The service they wish to provide is a data repository of content factoids to be used on all projects. Much like commons is a media repository. As the service provider wikidata needs to comply with our, the customer's, requirements. Until it does so, we can choose not to use its services. That wikidata editors do not understand that wikidata exists to serve the other projects is not our problem. Only in death does duty end (talk) 11:01, 16 January 2018 (UTC)

I am afraid you completely misunderstand what Wikidata actually is.--Ymblanter (talk) 11:12, 16 January 2018 (UTC)

From the front page of WikiData: "WikiData acts as central storage for the structured data of its Wikimedia sister projects including Wikipedia, Wikivoyage, Wikisource, and others." That is all completely true. As such, the data should all flow in the direction of WikiData, and not necessarily the other way. And I would argue, lets just have it flow that way, and not back - I am sorry, but it simply doesn't work, the data cannot be reliably used, even for simple, non-sensitive data. We've tried, we failed. --Dirk Beetstra ^{T C} 11:27, 16 January 2018 (UTC)

I see indeed that you argue this way. I just disagree with this vision and find it extremely short-sighted.--Ymblanter (talk) 12:41, 16 January 2018 (UTC)

Well you need to tell your fellow wikidata editors to stop attempting to use it that way. As it stands (as Beetstra above quotes) wikidata is a central data repository - the intent of which is that data is used on Wikimedia projects. I cant put it any simpler to you, you clearly don't even understand the scope of your own pet project. Only in death does duty end (talk) 11:34, 16 January 2018 (UTC)

Well, sure, you are not active on Wikidata, you are way less active than me have on the English Wikipedia, but you understand better than me and also better than everybody else how it should work, and you are sure what you say would be taken seriously. Fine, I can survive this.--Ymblanter (talk) 12:39, 16 January 2018 (UTC)

Much like your response to Fram this is another example of your 'NA NA NA CANT HEAR YOU' bullshit when you hear something you don't like and cant refute. Perhaps you over-estimate how seriously you are taken. Saying 'that's not wikidata's purpose' when it is both the stated purpose (as a central data repository), and the purpose its currently (being attempted by wikidata editors) to be used for on ENWP (drawing information from Wikidata for use in articles) is either extreme ignorance or flat-out falsehood. Only in death does duty end (talk) 13:25, 16 January 2018 (UTC)

Well, both communities felt confident enough to award me administrator privileges, something which I have not seen you to achieve with either of them. But, as I said, you are certainly entitled to have your opinion on the subject, even if it is completely uninformed and aggressive. This is ok with me. I am not even going to report you for a personal attacks. But I hope you will excuse me if I stop spending my time replying you.--Ymblanter (talk) 13:33, 16 January 2018 (UTC)

No you have said Wikidata's stated purpose is not in fact its purpose. Unless you back it up with some evidence that's a false statement. Please provide the requested information as to what wikidata's true purpose is. Clearly with all your experience you should be able to satisfy a simple question as to what wikidata is for. Only in death does duty end (talk) 13:37, 16 January 2018 (UTC)

I do not feel I should be communicating with someone who (i) calls me a liar thus lying themselfves; (ii) on top of this have difficulties understanding elementary text.--Ymblanter (talk) 13:44, 16 January 2018 (UTC)

Right, I will take that as 'No I cant state wikidatas purpose'. Glad we are clear. Only in death does duty end (talk) 13:55, 16 January 2018 (UTC)

Ymblanter, Only In death is not the only editor wondering whether the purpose of Wikidata is compatible with Wikipedia’s purpose. so, while I respect your decision not to respond to his/her comments, I would appreciate an answer to the question... what IS the purpose of Wikidata? Blueboar (talk) 16:04, 16 January 2018 (UTC)

The link which I provided right below says Wikidata is a free, collaborative, multilingual, secondary database, collecting structured data to provide support for Wikipedia, Wikimedia Commons, the other wikis of the Wikimedia movement, and to anyone in the world.--Ymblanter (talk) 16:07, 16 January 2018 (UTC)

Taking this out because it was hatted: d:Wikidata:Introduction. With this, I better quit the discussion, since I was already claimed to say smth I did not say, and this is probably not going to stop.--Ymblanter (talk) 14:18, 16 January 2018 (UTC)

Losing focus here: please comment on content ("Wikidata"; "links"; "Wikidata and links") not on fellow editors. --Francis Schonken (talk) 14:10, 16 January 2018 (UTC)
The following discussion has been closed. Please do not modify it.
True that. Understanding elementary texts is a basic requirement. Correctly writing such texts apparently not though... Fram (talk) 13:50, 16 January 2018 (UTC) Yeah, pulling rank to win an argument seems like a concession of defeat to me. Reyk YO! 13:53, 16 January 2018 (UTC) Only in death, you seem to have misunderstood the discussion. Ymblanter is an admin, you are not, so you are wrong and lose. Arguments and evidence only count between people from the same class, but people from a lower caste should never doubt the correctness of the higher castes. Please stop wasting Ymblanter's time by offering arguments and the use of a brain. Be glad that they are magnaminous enough to grant you the right to an opinion; if you don't stop, I'll start an ANI discussion Tbanning you from having an opinion and directly addressing admins. Fram (talk) 13:48, 16 January 2018 (UTC) I'm not sure even ANI has (yet, despite some attempts in the past) tbanned someone for having an opinion... Expressing it yes ;) Only in death does duty end (talk) 13:55, 16 January 2018 (UTC) I see, you guys have chosen a nice tactics: First, to misinterpret what I said, second, accuse me in lying and throwing a bunch of personal attacks on me, and then stating that I should prove that I am not lying. May be I can then explode, and sure you would be happy to bring me to ANI, or may be even to Arbcom. However, this is not going to happen. Who is genuinely interested in the subject, can go to d:Wikidata:Introduction, and then discuss it there if you are unhappy. Here, I do not see much point. I tried to be polite, it did not work. This is unfortunate, but I am not going to start call people names.--Ymblanter (talk) 14:05, 16 January 2018 (UTC)

I agree, they are obviously not engaged if WikiData is banned here, they are now also not engaged because of the massive problems that WikiData has because of the way that they started to collect data (import, import, import .. import ..). If all Wikipedia editors with some remote interest in the data would all walk over to WikiData to get it all correct (which is now a gargantuan task) we would not be editing Wikipedia anymore, and because of the mess WikiData is people would not get engaged there to get it correct. WikiData would have engaged many more editors if they would have put quality first, now people don't get engaged because of the lack of quality, and because of the task at hand to get it up to scratch. And in the meantime you expect, with the current policies of importing WikiData has, to just take it because it is the best you can do? 'Format C:'/'are you sure?'/'yes', and start over .... --Dirk Beetstra ^{T C} 10:13, 16 January 2018 (UTC)

See my response above. Right now you are using Google, and nobody suggests to ban Google. You just have wrong expectations from Wikidata. What you want will never happen.--Ymblanter (talk) 10:28, 16 January 2018 (UTC)

I know that it is not going to happen .. Jimmy Wales sending me 1 million dollar is not going to happen either. I will just have to live with it. And if I have the choice between checking with Google to see whether WikiData is correct, and if it is not using Google to get the correct data, or skip WikiData and just get the correct data from Google directly, then I choose the latter option. WikiData has its place in this world, but no place on en.wikipedia. --Dirk Beetstra ^{T C} 10:39, 16 January 2018 (UTC)

"in the example with Shrivallabh Vyas the original vandalism was on the Wikipedia side [3], and if the guy were still alive, we would return to the same vandalized article now." The vandalism was here first, true, but it was corrected here as well before the removal of Persondata happened.[4] Neither that correction from 2016 nor the addition of the correct date of death this month were done at Wikidata, which still showed the original error or vandalism from enwiki (which was never visible to readers here in the first place). Fram (talk) 09:48, 16 January 2018 (UTC)

Agreed, that BLP data is not transcluded (unless by choice, we have special templates for living people that can completely rely on WikiData data), but twitter feeds are. It is just a matter of finding one that is not on WikiData, transclude some vague account here, wait for WikiData to take my value (that is why we have WikiData-only maintenance categories like Category:Twitter username_not_in_Wikidata, right, to tell WikiData that they can take our data? I still fail to see how that maintenance template is improving en.wikipedia, we have the data ..), and then wipe the data here. It is good we are more careful with BLP data, but we transclude a lot from WikiData. --Dirk Beetstra ^{T C} 10:22, 16 January 2018 (UTC)

There is no evidence that in general, Wikidata is more up-to-date or correct than e.g. enwiki. Shrivallabh Vyas died on 7 January 2018, as correctly noted and sourced in enwiki. According to Wikidata[5] he was dead since 2011, which was sourced to "imported from English Wikipedia" in July 2014. Which is weird because at the time the enwiki article[6] said (correctly) that he was alive. So Wikidata is not simply outdated, but filled with false information, referenced to an unreliable source, which didn't even have that information at the time. Saying "never use Wikidata" is not "send the project - not Wikidata, but the English Wikipedia - beck to the middle of the 2000s." but protect the project from using a generally inferior, unreliable source. Fram (talk) 08:20, 16 January 2018 (UTC)

To make it clear, I am not going to respond to any arguments provided by Fram, since in my experience this is a loss of time. Other people may try if they want.--Ymblanter (talk) 08:22, 16 January 2018 (UTC)

"My claims are rubbish, so I'll try a personal attack instead". If you can't stand criticism of your hyperbolic claims and don't know how to react when it is shown that even the imports from enwiki by Reinheitsgebot (10,947,043 edits at Wikidata!) can't be trusted, then just don't participate in these kind of discussions, it would be even less lost time for all of us. Fram (talk) 08:36, 16 January 2018 (UTC)

• Respectfully, yes, I do think that Wikidata fails at the task it's intended to perform. From what I have seen it is cluttered with empty, out of date, or simply erroneous entries. There's enough of that on this site already, and it is no solution to delegate responsibility for it onto another. That doesn't solve the problem, it just makes the problem present in two locations instead of one. Reyk _YO! 08:38, 16 January 2018 (UTC)

My point was actually that Wikipedia (specifically, the English Wikipedia) fails to perform the task - to solve the problem I outlined.--Ymblanter (talk) 08:51, 16 January 2018 (UTC)

Is a lot of data on enwiki outdated? Yes. Does Wikidata solve this problem? No. Correctly identifying a problem doesn't make your solution automatically correct. Wikidata may be an improvement for small, rarely edited wikis (and certainly for those wikis which are largely bot-populated in the first place, like Cebuano or some invented languages). No one here is arguing to abolish Wikidata or to restrict how other wikis may use it. But enwiki, with all its problems, generally is more up-to-date and reliable than Wikidata is (for starters, we don't accept wikis or findagrave as sources, while Wikidata explicitly supports bot importing such sites as references), so replacing our data with data from Wikidata is in general a bad idea (and yes, you can find counterexamples). Using bots on enwiki to import selected, vetted, well-sourced series of data from Wikidata may be useful in some cases, and bots providing in projectspace lists of articles where e.g. the date of birth or the date of death mismatches between enwiki and Wikidata is something few people would object to. But none of this means that not using Wikidata live in articles is throwing enwiki back in time, it means that people look at the actual situation, not the ideal one, and realise that Wikidata is not (yet?) an improvement. Fram (talk) 09:05, 16 January 2018 (UTC)

• @YMBlanter thanks for the plug for the death anomaly project. Yes I designed it and it did identify a lot of anomalies between wikipedia language versions. A large group of us across about 11 languages then fixed those we could across many many language versions of Wikipedia. It didn't stop because Wikidata superseded it, it stopped because our bot writer retired. If someone is interested in writing a new bot for it please get in touch, it might be easier to code now that we have Wikidata. What I'm not sure of is whether Wikidata has equivalent volunteers using it to identify and list anomalies such as "people who are alive according to one language version of Wikipedia but not according to another". Call me overly cautious, but I'd prefer a data integration project that had such anomaly finding at its core. You would still find cultural anomalies because, for example, different language versions of Wikipedia have made different decisions as to how old someone needs to be before you assume they are dead; and more embarrassingly, if we have two unsourced bios for the same mid twentieth century sportsperson one describing the person as dead and the other as living, it is tricky to resolve if you can't find a source in a language you know. But I for one would be more comfortable with Wikidata at this stage if it led on anomaly finding rather than populating infoboxes. ϢereSpielChequers 12:40, 16 January 2018 (UTC)

I think there are a lot of pages of a similar intent on Wikidata (I am not immediately sure though whether there is one specifically on death anomaly - @Pasleim: probably knows everything about it, but even if there is not one, it can be I guess easily written. The problem is then more on the projects' sides - who is going to use this bot-generated lists to actually use it and how this information would propagate to the projects.--Ymblanter (talk) 12:46, 16 January 2018 (UTC)

That requires making the data available on the those language versions of Wikipedia that could be persuaded to take part. ϢereSpielChequers 13:50, 16 January 2018 (UTC)

Using Wikidata to identify entities unambiguously

In the discussion above I've proposed using the knowledge that some entity (person, building, art work...) has a Wikidata entry and keeping that information withing our article when relevant, not as a link to that entry, but as metadata for the entity as described in our article. I want to ask those editors supporting a ban of links to Wikidata what do they think about this proposal. Such metadata could take the form of a template that creates no visible link, but that is visible when editing the article - and maybe in a Special page or at Page information.

The idea would be to unambiguously identify topics of discussion which are not notable enough to have a Wikipedia article, but that are nevertheless verifiably identified as an existing entity that has a unique identifier at Wikidata. Adopting this practice throughout the project would provide invaluable to automated and semi-automated tools for processing articles, and could help research projects to better analyze the knowledge contained in our encyclopedia.

Note that the only information in Wikidata that we would be using by this approach is the existence of the entity registered, not any of the properties stored there. We are already doing this for topics that have a Wikipedia article; and we have parameters in the infoboxes templates that allow us to specify fine-grained usage of information in those topics, or avoid using them when we don't trust them. Diego (talk) 10:43, 16 January 2018 (UTC)

@Diego Moya: But how sure are you that all the data collected for WikiData-personality Donald Trump is talking about ONE physical incarnation of Donald? The way WikiData is sometimes importing data I would not be surprised that they could very well have imported data from Donald Trump (yes, I know, a Google search can also here on Wikipedia result in information being included about another Donald). And with a less notable personality (especially the ones which do not pass our bars) the risk of confusing the data is only bigger. --Dirk Beetstra ^{T C} 10:50, 16 January 2018 (UTC)

(edit conflict) The editor adding the template should make sure that d:Q22686 correspond to the entity[humang being] that has property[45th President of the United States of America], of course. That responsibility would belong to a Wikipedia editor, not a Wikidata one. It's no different than adding an internal wikilink to an article; the editor should make sure that the target topic is the same one being described at the origin article. This would allow us to increase this hyper-linking to entities that are not notable. Diego (talk) 10:56, 16 January 2018 (UTC)

@Diego Moya: but that is in core the same problem as discussed above. When transcluding an official website from WikiData you have to make sure that it is actually the one of the subject here. But we can do the same thing without WikiData. We now link to a redlinked person, and if we know that there are two distinctly different we make sure we disambiguate. With WikiData you just add a layer of complexity to that. --Dirk Beetstra ^{T C} 11:01, 16 January 2018 (UTC)

We seldom disambiguate redlinks, and that wouldn't work accross the interwikies for items that have different names in different languages. The idea would be to use an unambiguous identifier (the bit that is provided by Wikidata and not redlinks), and make sure that all our articles in all language 'pedias point consistently to the same ID when they refer to the same object. A good place to do this is at WP:CSC items that have been included in a list because they fail the notability criteria or short, complete lists of every item that is verifiably a member of the group - precisely the case in the example provided in the discussion. Also, aren't redlinks only for entities that we deem likely to be notable at some point? Diego (talk) 11:06, 16 January 2018 (UTC)

And we don't disambiguate those because we don't care whether non-notable subject A is the same or not as other non-notable subject A or that they are the same non-notable 'subject A'. But it is a function of WikiData, having two 'subject A's who are disambiguated by identifiers. But until WikiData IS that reliable database that does that, we should not use it on Wikipedia, as the unreliability of WikiData data may very well mix 2 'subject A's, or split one 'subject A' into two persons, or display very wrong information on the correct 'subject A'. --Dirk Beetstra ^{T C} 11:33, 16 January 2018 (UTC)

Have you just said that we don't care about the items described within our own articles if they happen to be not notable? Note that the reliability would be provided by us using the same identifier consistently in several of our articles; not by looking at what's stored at Wikidata for that identifier. We'd be using Wikidata only as a placeholder to make sure that we're not needlessly replicating identifiers. This could benefit both projects, I see it as Win-Win. Diego (talk) 11:41, 16 January 2018 (UTC)

No, we use references to make the statements about the person reliable. Whether the John Doe in article A is the same or different from the John Doe in article B is from a greater perspective not important until they link to the same article. WikiData is the central repository of data, en.wikipedia is, per pillar, not. --Dirk Beetstra ^{T C} 11:47, 16 January 2018 (UTC)

And I do agree, it would be great to be able to disambiguate those two John Does .. but again, if they are both non-notable on en.wikipedia grounds, then how I am to expect that WikiData is capable to properly deconvolute the data between those two John Does .. they will each link to two different WikiData IDs and have no data to identify them properly with sufficient reliable data. At best you know that one is from here, and the other from there (if it is not accidently the same person who moved inbetween). There will be John Does out there that cannot be sufficiently distinguished. It may be what WikiData aspires to do, but I would not keep up the hope that Wikipedia will at some point turn into a more reliable source based on such assumed data. --Dirk Beetstra ^{T C} 11:56, 16 January 2018 (UTC)

(edit conflict)The way I read WP:Build the web, registering that John Doe is the same in articles A and B is important "to enable readers to access relevant information on other pages easily" and "deepen their understanding of a topic by conveniently accessing other articles", which can't be done if we don't have the information to begin with. We do that with wikilinks to articles for items that are notable, and wikilinks to sections for items that are discribed within a larger article. I don't see why the same principle couldn't be extended as well to items that are described within our articles but don't have whole sections of their own; this information would allow us to reliably "find all articles where John Doe (Q1234567) is mentioned", or know that the Eleanor Criswell at Saybrook University is the same one listed as the spouse of Pernell Roberts. (^{, by the way}).

P.S. Again, this information would be added only for items that we can reliably identify as being the same according to our reliable sources; not according to what is stored in Wikidata. Diego (talk) 12:08, 16 January 2018 (UTC)

Creates a lot of extra clutter in the wikicode for very little actual benefit. We had things like persondata for years because of similar claims of this being invaluable metadata, but apart from then being mass-imported to Wikidata it was in the end a massive effort to have it all, and very little was actually done with it. Fram (talk) 10:54, 16 January 2018 (UTC)

And there won't be much use in the future either, if every time we propose a way to use the data, it is summarily rejected ;-)

Apparently these organizations are using Wikidata identifiers to unambiguously identify entities. Diego (talk) 11:11, 16 January 2018 (UTC)

Good for them (or not, but anyway that's up to them). But that's not really what we are discussing, what you propose is adding metadata to specific bits of enwiki article text so some hypothetical organisation or person may use this data for whatever. While this is not impossible, it seems like a lot of clutter and effort for little result. There are countless problems and issues that need fixing on enwiki, and finding reliable references for our articles seems like a lot more necessary than tracking down which Wikidata item, if any, belongs with which bit of article text, and then adding this invisible link for the benefit of, well, not the readers, not the editors, but some nebulous hypotheticals specifically interested in subjects notable enough for Wikidata but not for enwiki. Fram (talk) 12:52, 16 January 2018 (UTC)

not the editors says you. I've quoted the part of our current guideline that explains precisely how this may benefit our readers, given the right tools for navigation (just not a plain link to Wikidata). People supporting a ban on this should explain why this potential benefit is to be forbiden universally. Diego (talk) 12:59, 16 January 2018 (UTC)

I agree with Fram on this. The proven downsides in terms of reliability and maintainability far outweigh the claimed benefits. Reyk _YO! 13:00, 16 January 2018 (UTC)

Care to ellaborate? What are these proven downsides in terms of reliability and maintainability? Diego (talk) 13:02, 16 January 2018 (UTC)

You've quoted that guideline, but I don't see how it applies. To quote that guideline: "Ask yourself, "How likely is it that the reader will also want to read that other article?"" I simply don't see a potential benefit, and I see many problems. Main problem is that we send readers to a page which is not intended for readers. But for editors as well, we send people to a page at an unreliable wiki as if it is a good resource to use. We wouldn't allow such links to findagrave or ancestry to be placed in article text next to names, so why would we allow or encourage such links to Wikidata? I'm not a fan of such links to Wikipedia articles in other languages either for that matter, but it's not because those are allowed for the moment that we should accept the same for links to a site with a different purpose, policies, structure, ... Adding such metadata around terms makes editing harder, as it is more visual clutter one has to ignore, avoid, maintain, ... And it encourages people to recreate articles, links, on Wikidata which we don't want here, and to add sources like findagrave which we don't want here; Wikidata is too often used as a way to avoid blocks, deletions, policies on enwiki. Fram (talk) 13:16, 16 January 2018 (UTC)

You've quoted (have you? I don't see it) that guideline - I did it here.

Main problem is that we send readers to a page which is not intended for readers. - Then you haven't understood my proposal, which was to remove direct links to WikiData; but keep the fact that two parts of text in our articles refer to the same Wikidata object.

This would be used for navigating between our articles, without ever leaving Wikipedia, just like wikilinks; but withour requiring directional links ("connect this article to that article), and for topics that are not notable. Navigation within Wikipedia is the main purpose of MOS:LINK.

Adding such metadata around terms makes editing harder, as it is more visual clutter one has to ignore, avoid, maintain, ... - Fair point; yet the amount of clutter is similar to that of reusing references within an article, so hardly unmanageable.

Wikidata is too often used as a way to avoid blocks, deletions, policies on enwiki - Can you please explain that? How is data on Wikidata misused to disrupt Wikipedia? Diego (talk) 13:48, 16 January 2018 (UTC)

I self-corrected my "have you?" question before your post here, but thanks for taking the time to point it out anyway. "This would be used for navigating between our articles, without ever leaving Wikipedia" how? Not by readers apparently, but how would editors use this to navigate from one article to another without leaving wikipedia? Even when you follow the link to Wikidata, you don't have the option of going back to other Wikipedia articles which link to the same item. The issue is not that the clutter would be unmanageable, everything is manageable, but that the cost/benefit ratio would be bad. We already have a lot of necessary clutter like refs, please don't add unnecessary clutter as well (the same goes for things like hidden categories, which on many articles have lost their usefulness completely now that they are swamped with cats generated by authorty control and other Wikidata stuff). As for the "Wikidata for avoidance of enwiki issues" RAN (the editor at the start of all this) is a good example. When he got into trouble here for a whole range of issues, he created Wikidata items instead, added Findagrave there, and then added links to his Wikidata items to articles here. Fram (talk) 13:57, 16 January 2018 (UTC)

Readers would be able to find connected articles by searching for the Wikidata item ID (maybe with Search, preferrably by building a tool that searched articles containing the tag with that ID). What count as unnecessary is ultimately subject to personal preference; I've stated my case of how the people writing WP:BUILD found the idea valuable, just like myself.

he created Wikidata items instead, added Findagrave there, and then added links to his Wikidata items to articles here. - Only the last part would be problematic for us. If we restrict ourselves to in-Wikipedia links and didn't expose readers to the link to Wikidata, that strategy wouldn't do much good. Diego (talk) 14:08, 16 January 2018 (UTC)

But the readers wouldn't see the Wikidata ID, if I understand you correctly. So how would they do this search? In any case, you find it worthwhile, I find it a lot of hassle for something that would be rarely if ever useful (with a redlink, you can already easily click it and then use "what links here", so your proposal would only potentially help for items where we have multiple subjects for one redlink, which all have a Wikidata item, and where all of them have that Wikidata item correctly added to their text entry). Fram (talk) 14:33, 16 January 2018 (UTC)

But redlinks should not be added to items that are not notable...

Not havink a link to Wikidata doesn't prevent us from having a small internal link to the search page, like those for refs or notes. Diego (talk) 15:26, 16 January 2018 (UTC)

Ah, I see what you mean. Again, this would add visual clutter to what the reader gets, which to me is not worth the end result. I can't imagine many readers having the question "in what other enwiki articles is this non-notable subject used as well", and considering that it would only be really useful if it was somewhat complete, I see a gargantuan effort for little positive result. A nice idea, but (as you probably guessed by now) not something I support. Fram (talk) 15:36, 16 January 2018 (UTC)

@Diego Moya: Overall, I think there's merit to the idea. The accusation of "Luddism" by YMBlanter is nonsense; this is about ability of en.Wikipedia articles to comply with en.Wikipedia policies, and not have them skirted by "offshoring" to other projects. If WD information is pulled into WP, then it can be made to comply here. I'm not sure how that gets done, e.g. the template suggestion below. However, this RfC/firehose has obviously turned into a mess almost overnight and is unlikely to result in a consensus on anything. I would suggest letting the battleground fall to silence and you can float this idea again, with some specifics later, and not here. This is a policy debate, not a style one.  — SMcCandlish ☏ ¢ >^ʌⱷ҅_ᴥⱷ^ʌ<  00:29, 17 January 2018 (UTC)

@Diego Moya: Perhaps you are aware of this, but at the moment, searching for Qnumbers used in enwiki through our search interface doesn't seem to work. Q6167205 is used in Mayor of Verona, New Jersey, but a search gives no results. Fram (talk) 07:49, 17 January 2018 (UTC)

Searching for insource:Q6167205 works, though. —David Eppstein (talk) 07:55, 17 January 2018 (UTC)

Interlanguage link template

We already have both established precedent, and a template, for links to Wikidata, albeit they should take a different form to that used in the example quoted above, by Koavf. That template is {{Interlanguage link}}, and for Wikidata it works like this - {{Interlanguage link|William Harold Bodine, Sr.|WD=Q45371935}} renders as William Harold Bodine, Sr.. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:11, 16 January 2018 (UTC)

"Established precedent" is meaningless if this RfC decides otherwise. That we have a template is nice, but again hardly relevant if this RfC decides that such links are not welcome. Fram (talk) 12:33, 16 January 2018 (UTC)

Maybe, but as Andy said, this should be announced centrally for it to override the precedent. Diego (talk) 12:41, 16 January 2018 (UTC)

It is announced centrally now (not by me), and I don't think this requirement was valid anyway. An RfC on this page is not really an obscure location. The "precedent" is one or two users adding it to a template and starting to use it, not something that was centrally announced and decided either... Fram (talk) 12:46, 16 January 2018 (UTC)

The precedent to override is the WP:Sister projects guideline. Diego (talk) 12:51, 16 January 2018 (UTC)

I see that this was labeled a guideline by Whatamidoing in 2012, I can't find any discussion to establish whether this really was and is a generally accepted guideline or not (but I may well have missed said discussion). The blanket statements in that page about inline links to sisterprojects seem not to match generally accepted practices. Considering how even for search results, there were sister projects we explicitly rejected (i.e. sister projects which are not allowed by enwiki to be shown i na search here), I doubt the community at large would be happy with the promotion of inline links to such projects. Anyway, central notice was asked for, central notice has been provided, so whether the precedent is a valid guideline or not is not really an issue, this RfC can decide what to do with this specific issue. Fram (talk) 13:01, 16 January 2018 (UTC)

I'd also like to see the discussion to promote this to a guideline. If it's been just kind of snuck in without anyone noticing, that would be reason enough to revisit which of these sister projects we should really be linking to. Reyk _YO! 13:10, 16 January 2018 (UTC)

@WhatamIdoing:--Ymblanter (talk) 13:12, 16 January 2018 (UTC)

I believe that you'll find that SISTER was first labeled as a part of the Manual of Style by a logged-out editor in October 2005, which predates my involvement rather significantly. WhatamIdoing (talk) 19:53, 17 January 2018 (UTC)

That would be a misuse of the template since there are no other language articles as far as I can see (if there were, we would interlanguage link directly unless there were multiple ones), and by both EL and sister projects we wouldn't link to a page with no information. Only in death does duty end (talk) 13:30, 16 January 2018 (UTC)

It would not be a misuse, since that's part of what the template is designed to do. If you see an item on Wikidata "no information", by all means nominate it for deletion there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:06, 16 January 2018 (UTC)

Why would we send anyone willingly to a page where the only external sources (given in general, not as specific references) are to Familysearch (I have no intention to register to a wiki from the Church of Latter Day Saints, thank you) and Rootsweb ("is currently unavailable")? Just because it is a sister site doesn't mean that it is an acceptable site. Fram (talk) 14:56, 16 January 2018 (UTC)

On the other hand, why would you object, if the links establishing identity were to eg VIAF, ULAN, RKD, Benezit, Library of Congress? The motion above aims to ban all links via {{Ill}}, without any scope for editorial judgment, no matter how well identified the entry is. That's WP:CREEP, and over-reach. Jheald (talk) 15:08, 16 January 2018 (UTC)

How's that WP:CREEP? Creep is "Instruction creep is often a result of editors producing too much instruction, resulting in very long, complicated pages. " Here a very simple rule is proposed. You are free to disagree with the proposal, but I don't think "creep" is a valid argument here. I oppose such links in general because I see way too many problems at Wikidata. This doesn't mean that no entries at Wikidata are good, but that not enough of those without an enwiki article are. I see the "it's a sister project, so we should link to it (or be allowed to link to it) as a non-argument, we are not here to promote other WMF projects, and certainly not ones which basically bring us everything we don't want in enwiki (sourcing, policies, lack of vandal fighting, ...). It would be better, if we want to provide readers useful pointers to outside sources for subjects without an enwiki article, to allow such links directly to a small number of reliable sources (like the LoC) instead. I am not really fond of that either, but it would at least be better than linking to Wikidata. Using an unreliable intermediary to direct readers at best to reliable sources is not something I support. Fram (talk) 15:25, 16 January 2018 (UTC)

Yeah, this is WP:CREEP. You're taking something that is currently a matter of editorial discretion, you want to take that discretion away, and in the process you want to one more to WP's ever more voluminous pile of rules. The fact you don't even recognise this for the WP:CREEP that it is makes it even worse.

Now, your claim: not enough of those without an enwiki article are [any good].

Let's see. Currently 549,004 entries without an en-wiki article, but with a VIAF identifier. (tinyurl.com/yb8xkcg4). None of them worth being able to distinguish? 200,459 with a Library of Congress ID (tinyurl.com/y7yq8lkk); 29,351 with a ULAN identifier (tinyurl.com/y8cglcao); 15,075 entries with an Art UK identifier but no article here. (tinyurl.com/y8f8xxjl)

Now let's look at some of those Art UK examples (ie painters with works in UK public collections). tinyurl.com/y9nwpttk. The top one (most likely to be referred to?) has links to entries no fewer than 32 different biographical databases. Even down at number 3200, we find the identity is backed by links to four different external databases.

Yet none of this you think is useful, not ever, to uniquely identify somebody in a list or a citation section that is the subject of a redlink? Jheald (talk) 16:37, 16 January 2018 (UTC)

Please provide some useful links, instead of plain text tinyurls which turn out to be Wikidata queries. It's not really helpful this way. Anyway, no idea how you can quote my "not enough" statement and then translate it to not ever, none of this... Anyway, number 1 is Arnold van Westerhout. He is linked from two enwiki pages, as can easily be found. If we had used links to Wikidata instead (the type RAN used and which started this discussion), I would not have been able to find this. Fram (talk) 16:50, 16 January 2018 (UTC)

The proposal above is for a total ban on Wikidata links, so let's not mess about here. A redlink is all very well, but is hardly the point. It doesn't identify AvW uniquely -- there might be multiple articles that have redlinks intending different AvWs -- nor does it attach the identity intended to external databases. In contrast, the {{ill}} template gives the option to output a redlink with a Reasonator link Arnold van Westerhout that does all of that, and gives editors a heads-up that there is a Wikidata item to link to if/when they do choose to create an article. The proposal above would take away editor discretion to include such a link in all circumstances. Jheald (talk) 17:20, 16 January 2018 (UTC)

Ugh, please, not Reasonator. Those fake machine generated articles are just terrible. but that page (Reasonator and Wikidata) is again a good example of why we shouldn't use them. They both describe him as "Belgian", which is of course false. But it is sourced! Yes, to this which correctly describes him as Flemish. Has the source changed since it was added to Wikidata, and is Wikidata yet again not so up-to-date as is claimed? Or was it a wrong addition right from the start? No idea. But it's not something we should willingly expose our readers to. Fram (talk) 08:05, 17 January 2018 (UTC)

When did "external sources" on sister-project pages we link to become a requirement for linking? How many Commons pages, for instance, include "external sources"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:06, 16 January 2018 (UTC)

External sources are not a requirement, no. Reliable sources should be a requirement. Linking to poorly sourced pages with the intention to provide additional data to our readers goes against what enwiki should do, i.e. provide and summarize information from high quality sources. And I don't consider structured data and images to be similar enough to require them to be treated the same. Feel free to start a discussion about Commons of course, but don't expect me to base my position wrt Wikidata on what we do with Commons, or how Commons deals with problems. Fram (talk) 16:41, 16 January 2018 (UTC)

It more than "goes against what we should do". It is an explicit violation of WP:ELNO #2, which disallows linking to "Any site that misleads the reader by use of factually inaccurate material or unverifiable research". —David Eppstein (talk) 19:03, 16 January 2018 (UTC)

The same reasoning should then apply to links to other language Wikipedias ({{ill}}), which I would support. Btw it would also apply to links to the English Wikipedia which are formatted as external links (which does happen, though not very often).--Ymblanter (talk) 19:08, 16 January 2018 (UTC)

Links to Wikidata, such as those discussed above, are not "formatted as external links". HTH. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:11, 16 January 2018 (UTC)

OK (though that's not what was said above). When did "reliable sources" on sister-project pages we link to become a requirement for linking? How many Commons pages, for instance, include "reliable sources"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:11, 16 January 2018 (UTC)

You mean, aside from the licensing templates and requirement to note the sources the images etc. were taken from? Curly "JFC" Turkey 🍁 ¡gobble! 23:21, 17 January 2018 (UTC)

• The question of WP:ELNO#2 bothers me. We don't use {{ill}} to link to other-language Wikipedias with a reputation for uncorrected errors; why should we link to Wikidata while it has a reputation for uncorrected errors? — Arthur Rubin (talk) 23:20, 17 January 2018 (UTC)
  • We don't link to other wikis as references, but we do link to them for navigation. Different purposes may be covered by different guidelines. Diego (talk) 09:25, 18 January 2018 (UTC)

Wikidata link options

Here are some options:

1. Frank White Burr (rejected, should be more distinguishable from an internal link, per principle of least astonishment)
   • [[d:Q5490147|Frank White Burr]]
2. Frank White Burr (no redlink with Wikidata link icon)
   • {{Wikidata icon|Q5490147}}
3. Frank White Burr (redlink with Wikidata link text)
   • {{Interlanguage link|Frank White Burr|WD=Q5490147}}
4. Frank White Burr (redlink with Wikidata link text and reasonator link text)
   • {{ill|Frank White Burr|WD=Q5490147|reasonator=1}}
5. Frank White Burr
   • {{Interlanguage link|Frank White Burr|D|Q5490147}}
6. Frank White Burr
   • {{Interlanguage link|Frank White Burr|d|Q5490147}}
7. Frank White Burr
   • {{Interlanguage link|Frank White Burr|wikidata|Q5490147}}
8. Frank White Burr (Wikidata)
   • Frank White Burr ([[d:Q5490147|Wikidata]])
9. Frank White Burr (Wikidata)
   • Frank White Burr <small>([[d:Q5490147|Wikidata]])</small>
10. Frank White Burr (wikidata)
    • Frank White Burr <small>([[d:Q5490147|wikidata]])</small>
11. Milton Gideon Votee (the soft redirect, this would create a blue link but stop at a page that offers to send you to Wikidata, the message can be customized for Wikidata like the one for Wiktionary)
    • {{Soft Wikidata redirect|d:Q5490147}} (→ See Wikipedia:Templates for discussion/Log/2018 January 19#Template:Soft Wikidata redirect --Francis Schonken (talk) 05:38, 19 January 2018 (UTC))
12. Frank White Burr (invisible text to preserve the unique identifier, but no useable link)
    • Frank White Burr <!--Wikidata:Q5490147-->
13. W. Homer Axford (the standard soft redirect with no special Wikidata message)
    • {{Soft redirect|d:Q47462594}}
14. William Harold Bodine, Sr. (this option redirects the name to a list, but reminds the reader that more data is available at Wikidata on the redirect page)
15. National Register of Historic Places listings in Somerset County, New Jersey (this model uses a link to Wikimedia Commons along with symbol for commons and the phrase "more images". It would be adopted to read "more information" and use the symbol for Wikidata. It could be displayed in its own column in the table.

Comments on link style

• I oppose 1, 2 (flaglike icon doesn't work: should be something else for the link, see above), 5 and 6. --Francis Schonken (talk) 19:41, 16 January 2018 (UTC)
  • 8: not OK imho, parenthesis with link should have small text, so that it is graphically different from surrounding prose, which may also have parentheses. --Francis Schonken (talk) 21:24, 16 January 2018 (UTC)
  • Oppose #11: those that don't deserve a separate page in Wikipedia can be bluelinks, but then as a redirect to the Wikipedia page that has most information on them (which might very well be the page which contains the bluelink you started from). --Francis Schonken (talk) 21:49, 16 January 2018 (UTC)
  • #12: imho, no, not what we should be using invisible text for. --Francis Schonken (talk) 22:16, 16 January 2018 (UTC)
  • Oppose all, after having given this some more thought. --Francis Schonken (talk) 12:52, 22 January 2018 (UTC)
• 5 and 6 looks like he is a member of the Democratic party, remember cursor rollovers do not work in mobile Wikipedia. There is really no point in creating a red link especially since his entry was already deleted, we do not need a chart that is a sea of red links. The whole point of this is that these people will probably never get a full Wikipedia entry. We just want to uniquely identify them from others of similar name like VAIF and LCCN do. --RAN (talk) 20:26, 16 January 2018 (UTC)

8. Frank White Burr (Wikidata)

-- Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:52, 16 January 2018 (UTC)

Note that Francis has edited his first comment, since I replied to it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:35, 16 January 2018 (UTC)

? Didn't change anything to my first comment after you replied to it. --Francis Schonken (talk) 21:49, 16 January 2018 (UTC)

Yes you did; and now someone else has deleted half of my comment (now restored). WFT is going on? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:41, 16 January 2018 (UTC)

Again, I didn't change anything to my first comment after you replied to it. I expanded the list with the additional examples presented elsewhere by others (none of these examples were changes to a comment of whoever: I just copy-pasted the examples in the list, adding a few variants which were not "comments" nor "changes to comments" whatsoever). My first comment was untouched from the asterisk that preceded it to the signature that ended it. Then I also added another comment about one of the new examples, with an additional new signature. So, no, I did not "edit my first comment after you replied to it". --Francis Schonken (talk) 22:50, 16 January 2018 (UTC)

"I expanded the list" Yes. You changed your first comment, after I had replied to it. Stop denying that you did so. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:29, 16 January 2018 (UTC)

Sorry, again: I did not change my first comment after you had replied to it. This is the entire comment:

• I oppose 1, 2 (flaglike icon doesn't work: should be something else for the link, see above), 5 and 6. --Francis Schonken (talk) 19:41, 16 January 2018 (UTC)

It remained unchanged after you replied to it. --Francis Schonken (talk) 04:55, 17 January 2018 (UTC)

• Oppose all- wikidata should not be linked in the body of an article. This is the case especially for non-notable subjects such as Frank W. Burr which already had an article deleted at AfD.--Rusf10 (talk) 21:43, 16 January 2018 (UTC)
• Oppose all per my contributions to the discussion above. Wikidata links fail WP:ELNO #2 (far too much of the "data" on Wikidata is unsourced speculation; Wikidata editors have explicitly rejected policies analogous to WP:BLP [7] so Wikidata should not be linked in a way that makes its data appear definitive). Additionally, Wikipedia's policies and guidelines on whether an article should be redlinked should be based purely on potential notability of the subject, not on whether some other project has a link. —David Eppstein (talk) 22:35, 16 January 2018 (UTC)
  • Wikidata is compliant with the Wikimedia Foundations policy on living people. WP:ELNO #2 forbids inherently "unverifiable" data, which is not "unreferenced at the moment", they are not synonyms. My quick calculation says that we would lose about 20% of the English Wikipedia if we removed every paragraph that is currently unreferenced. Correct me if you calculate a different number. --RAN (talk) 22:38, 16 January 2018 (UTC)
  • "Wikidata editors have explicitly rejected policies analogous to WP:BLP" This canard is as untrue now as it was at all the many previous points when I have pointed out its falsehood. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:44, 16 January 2018 (UTC)
    • "Wikidata is compliant with the Wikimedia Foundations policy" is a highly disingenuous way of saying that they don't follow the much stricter English Wikipedia policy. (As a separate project, they're allowed to set their own policies, but if their policies are incompatible with ours then we should not link to them.) And the link I posted, showing them rejecting a policy analogous to WP:BLP, speaks for itself. —David Eppstein (talk) 22:49, 16 January 2018 (UTC)
      • Your link does not "show them rejecting a policy analogous to WP:BLP" (and you've gone from "policies" to "a policy"); it shows the rejection of a weak, badly worded and ambiguous first draft of a policy, far from being the equivalent of enWp's, for the good reasons stated there - effectively sending it back to the drafters with the message that a better policy is needed. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:24, 16 January 2018 (UTC)
  • There are almost 300 Wikipedias, of which the English is one, all must comply with the WMF policies. (now corrected above) We link to Wikimedia for images and Wikisource and Wikiquote, all with their own policies. And oddly we already link to Wikidata for people who already have articles, despite it having its own policies. We also link to a half dozen outside Authority Control identifiers, each with their own policies. We also link to newspapers in references which may disclose a living person's age, and the name of their spouse and children. This is simply what to do about notable people without articles that have data at Wikidata.--RAN (talk) 22:53, 16 January 2018 (UTC)
    • This discussion started with links to NON-notable people (deleted at AfD) who have data at Wikidata so it is, again, disingenuous to shift the goalposts to talk only about the notable ones. And when we link to Wikisource and Wikiquote, it doesn't create the false impression that unsourced speculative birth and citizenship data at the link is reliable. And Wikimedia commons has shown itself to be good at making sure its images comply with policy (their licensing requirements are stricter than ours and well enforced); Wikidata has both weaker requirements and very weak enforcement. —David Eppstein (talk) 23:20, 16 January 2018 (UTC)
      • AFAICT, none of the other sister sites you mention - Wikisource, Wikiquote, Wikimedia Commons - have "policy analogous to WP:BLP", nor indeed do they require any sourcing for birth dates of living people, nor the death dates of the recently deceased. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:27, 16 January 2018 (UTC)
      • NO one has changed the goalposts, the initial statement was: "links to wikidata for people who fail notability requirements to have [standalone] articles on wikipedia. In some cases ... links to wikidata after the article on the person was deleted" (emphasis mine). The argument is people notable to be included on a list but not notable for a standalone article, whether deleted, or never created. The notabilty of the people included on the list has been an accepted fact in multiple AFDs. --RAN (talk) 00:00, 17 January 2018 (UTC)
• Oppose all Ugh, per David Eppstein, just looking at that link - more concerned about limiting expansion than having accurate information, and also rejecting stuff like The repeated addition of unverifiable information (such as "John Doe is the world's lamest man") can result in a block of the offending user. Not compliant with our policies on BLP. Galobtter (pingó mió) 07:22, 17 January 2018 (UTC)
• Oppose all per David Eppstein. Because Wikidata does not accept many of our more strict content requirements, particularly for BLPs, using Wikidata is equivalent to sneaking BLP violations in through the back door. The use of Wikidata should be the exception, not the rule, and should be done sparingly and with responsibility if at all. I oppose anything that would make linking to it routine. Reyk _YO! 07:25, 17 January 2018 (UTC)
  • addition- since it's been pointed out that this is the MoS talk page I should also say something about the style concerns. I oppose the invisible wikitext option, since that should be deprecated generally, and I don't approve of any of the other options because they are unnecessary clutter- especially given the low value of what's being linked to. Reyk _YO! 09:18, 17 January 2018 (UTC)
• Support 3, 8 and 9 (Wikidata should be capitalized; redlinks are fine in article text but can be hidden if a template is used; <small> can be used in body but not in infoboxes), assuming that links are kept. Jc86035 (talk) 12:16, 17 January 2018 (UTC)
• 3 or 9 could be acceptable in some cases, possibly replacing the direct Wikidata link with a soft redirect. Although, the question of whether the link would necessarily violate WP:EL bothers me. — Arthur Rubin (talk) 23:15, 17 January 2018 (UTC)
• Support 3rd, oppose others, because it's the only mobile-friendly way, for users that "oppose all", I would ask you all that if there are missed concepts here, are they all ineligible in English?! --Liuxinyu970226 (talk) 11:49, 18 January 2018 (UTC)

  Anyway, is there be possible to restore {{Internal link helper}}, but only for the Wikidata propose, so we can see a green link, and by moving our mouse we can see "The article XXX is not yet created, you may visit Wikidata for viewing its summary-of-details and other languages editions" and link changed to normal redlinks. --Liuxinyu970226 (talk) 11:04, 22 January 2018 (UTC)

• Support 3 and 9, with same-size text (8) used in infoboxes where text is already small. We should only redlink things that we think will pass as notable and be the subject of a Wikipedia article, and leave non-notable items un-redlinked.--Carwil (talk) 16:41, 19 January 2018 (UTC)
• Support 2, 3 and 9. The flag is already used by Commons for links to Wikidata so it shouldn't be too unfamiliar to our readers. Deryck C. 18:14, 19 January 2018 (UTC)
• Support 2, 3 + 9: if people really doesn't know what the flag is, they can hover over the link, or—if on Mac—can preview the page. I do not see how it could cause a problem. –Sb2001 18:30, 19 January 2018 (UTC)
• Support 9, if the subject is notable 3, and less so 2: Otherwise if we already have an article on the subject than linking to WD will occur on the article page itself. Doc James (talk · contribs · email) 09:17, 20 January 2018 (UTC)
• Support 2, as that matches the convention already in use in Wikidata-enabled infoboxes, on Wikimedia Commons, and elsewhere. Ideally #2 should also include a redlink. After that, 3 and 7 are OK, 8-10 aren't great since they're missing the redlink but are also sort of OK. 5 and 6 are rather meaningless, and 11-13 seem pointlessly complex or just pointless (in the case of #12). #4 would be nice, but it's getting too long (particularly on mobile devices), and that's also a concern with the ones that spell out 'wikidata' rather than showing the logo. Thanks. Mike Peel (talk) 13:51, 20 January 2018 (UTC)−
• Support 2, 3 + 9 All three work well and for non-Wikpedia notable people 9 does a good job at disambiguation. ChristianKl ❪✉❫ 23:09, 20 January 2018 (UTC)
• Support soft redirect I don't care if we make a special soft redirect just for Wikidata. Visually it gives the best look, a clean blue link, and gives the reader an option of not going to Wikidata. It links all the appearances of the person in Wikipedia should the decision be made later to start an article. If an article is made it will insure that it is properly linked to Wikidata, and a new entry not created as a duplicate. --RAN (talk) 01:07, 21 January 2018 (UTC)
• Oppose all, a redlink will do. We are not writing a database, that is WikiData. --Dirk Beetstra ^{T C} 05:29, 21 January 2018 (UTC)
  • @Richard Arthur Norton (1958- ): "this option redirects the name to a list, but reminds the reader that more data is available at Wikidata on the redirect page" - see WP:EL. We do not link lightly to external sources for 'more information'. As the situation is now, I don't think there is any legit reason to link to WikiData as a source of information. The option is worse than linking to the corresponding page on another language Wiki. --Dirk Beetstra ^{T C} 09:40, 31 January 2018 (UTC)
• Oppose all except 12. There should be a blanket ban to biographical links and links in BLPs per WP:BLPSOURCES. LK (talk) 01:01, 22 January 2018 (UTC)
• Oppose all... I have no issues with linking in the sidebar to Wikidata for the subject of the article - but we shouldn't be linking in the body of the article. Ealdgyth - Talk 13:22, 22 January 2018 (UTC)
• Oppose all We should not be linking to an unreliable source in article space. --Malcolmxl5 (talk) 00:11, 23 January 2018 (UTC)
• Oppose all. The link in the sidebar is all that’s needed. These just highlight the serious problems with putting links inline. 1 looks like an article link. 2 looks like a decoration. Most look like links to the Wikidata article. Two look like links to another language (d[e].wp e.g.). Worst of all is the soft redirect that looks like a normal link but instead takes readers a page with no content, just the link to the page they wanted and a data link they will probably regret clicking on. All of them have serious issues and so show how Wikidata links within articles are just a bad idea.--JohnBlackburne^words_deeds 06:18, 31 January 2018 (UTC)
• Oppose all Sidebar link is sufficient. Anything else is covered by ELMAYBE. Only in death does duty end (talk) 15:34, 31 January 2018 (UTC)
• Oppose all per the above arguments. -- Tavix ^(talk) 22:08, 1 February 2018 (UTC)
• They're all pretty awful. I guess 12 is the least bad, in its uniquely awful way. Aside from the usual Wikidata issues, the link usually falls under WP:ELNO #1 or a really lousy WP:ELMAYBE. Wikidata was designed as a database for values to retrieve, not for human reading. Alsee (talk) 09:03, 4 February 2018 (UTC)
• Oppose all, that way madness lies. Kaldari (talk) 06:30, 11 February 2018 (UTC)
• Support 8. Possible variation: Frank White Burr ^(Wikidata). (Frank White Burr ^{([[d:Q5490147|Wikidata]])}) -- Oa01 (talk) 09:50, 13 February 2018 (UTC)

Inline flag-like icon for Wikidata, at TfD

FYI

 – Pointer to relevant discussion elsewhere.

Please see Wikipedia:Templates for discussion/Log/2018 January 16#Template:Wikidata icon.

The fact that the nominator was blatantly accused of bad faith in even daring to open this template for discussion says a lot about the bloc vote going on over there right now.  — SMcCandlish ☏ ¢ >^ʌⱷ҅_ᴥⱷ^ʌ<  00:36, 17 January 2018 (UTC)

No one should be nominating for deletion something being actively discussed at an RFC. --RAN (talk) 01:07, 17 January 2018 (UTC)

This is not a block vote. The nominator himself advertised the TfD here, and whoever was interested went there to vote. There was no conspiracy, at lest not than I know of. I agree though that nominating at TfD smth which is being discussed at RfC is not smart and serves no purpose.--Ymblanter (talk) 06:52, 17 January 2018 (UTC)

if people agree that the template shouldn't exist, then discussing any solution with this template (or anything similar) in it can be closed. Seems like a rather good purpose for a TfD. Fram (talk) 07:33, 17 January 2018 (UTC)

@Wittylama: Since the TfD was "suspended" (not sure if that's a valid procedure), commenting here: Assuming Wikidata links in articles are retained, would {{Interlanguage link}} adequately replace {{Wikidata icon}} (example: Douglas Adams vs )? Jc86035 (talk) 12:13, 17 January 2018 (UTC)

• Yeah, I do not agree with the "suspending" of the TfD discussion either. Reyk _YO! 12:23, 17 January 2018 (UTC)

Thank you Jc86035 for your productive contribution to this discussion! I would suggest that it's not quite the same for a couple of reasons. Firstly is the visual aspect (the use of the word "wikidata" rather than the logo) but that is of smaller significance in the grand scheme of things. I used the logo when I built the template because it is consistent with our sisterproject templates and consnstent with the way we link to wikidata items on Commons - a common visual vocabulary across projects. But, as we've seen in the 'template for discussion' debate, this stylistic choice is not appreciated by all... The more substantial point I feel that makes it different is that the Wikidata Icon template is independent of whether there is a blue, red, or no-link to the item in question. The ILL template has, as its main purpose: "So long as the page on this wiki does not exist (i.e. is "a red link"), this template temporarily displays" and then there is [[Template:Interlanguage_link#Forcing_links|the display=1 feature to "force" the wikidata link even if it is a bluelink. This template was designed specifically and deliberately to NOT interfere or interact with the normal link mechanics and policies on WP. So... it can be placed next to a bold word in a table or infobox, or a bluelink, redlink, or no-link and no template-alteration needs to take place if an article about that subject is created in the future (in the event of a redlink). See, for example, how it is used in the infobox in: The Offerings of Peace and The Offerings of War. This is an article about a pair of sculptures that are both notable for WikiDATA's perspective, but would not be worth writing about independently in WikiPEDIA - in this case the ILL template using the wikidata-field wouldn't be appropriate (I believe) because the title of the artwork in its own infobox shouldn't be linking to itself. See what I mean? Wittylama 13:03, 17 January 2018 (UTC)

But why would we link to the Wikidata pages for these two sculptures in the infobox / image? That's not the purpose of an infobox or image. In those, the link to Wikidata should go, the accession number should go (or at least should be correct!), and the overly precise coordinates should go (have one coordinate at the top right of the article instead). Then you end with readable, attractive, informative images without the barrage of needless information it has now, coupled with two meaningless "flags" (I understand you created these with the best intentions, and that they are being used in Commons and so on, but to most of our audience they will be confusing and meaningless). What this page could use, if you want to include links to reliable, relevant databases about the two sculptures, is something like the authority control template at the bottom (but a bit more restrictive than what is usually shown). Then again, the Wikidata pages about these sculptures are at the moment only sourced to the Art Gallery of New South Wales, a link which is already in the article . Basically, the links to Wikidata add nothing here but visual clutter and redundant information in a reader-unfriendly format. Fram (talk) 13:18, 17 January 2018 (UTC)

I've fixed the Accession number error, and removed the geo-coords from the image/infobox in that article - good points. Wittylama 21:13, 17 January 2018 (UTC)

I've removed all three Wikidata links (icons) from tha article, as none of them contained any information not already present in the article. It makes no sense at all to link to sister projects if that links doesn't give the reader any additional information. Fram (talk) 08:16, 19 January 2018 (UTC)

​Almost all of the non-Wikipedia sister project links I've seen have text explaining what the link does (e.g. "Media related to article subject at Wikimedia Commons"), aside from inline English Wiktionary links, which are often unmarked per Wikipedia:Manual of Style/Linking#Interwiki links. I think usage on Commons is different because there's no Manual of Style (if you're referring to c:Template:Artwork the addition of the logos without accompanying text seems to have been without formal consensus), and since Wikipedia has a guideline which advises against unexplained icons ("[Repeated use of icons in non-body text] should only be done if the icon has been used previously with an explanation of its purpose.") I think it would be better to at least have accompanying text to help readers understand what the icon is. Jc86035 (talk) 13:33, 17 January 2018 (UTC)

Hi again Jc86035, you make a fair argument that the other sister-project links are often surrounded by explanatory text. The purpose of this template was to UNclutter the space visually and leave the associated word UNencumbered to be (or not to be) a red or blue link. Alternatives in a table format would be to have: the full wikidata number (which is just an arbitrary number and not informative of anything); a more elaborated template (as you described like the commonscat template) but that would be very bulky; the word 'wikidata' (like in the ILL template you mentioned) but that also defeats the purpose of being small and unbotrusive to fit inside table cells. Perhaps, when the template is used repeatedly and throughout the article, such as in List of public art in the City of Sydney, then it should be prefaced with a short explanatory comment near the beginning which describes its purpose? This would be just like many articles take the time to explain the scope and features of complex tables before launching into the details. I realise this suggestion won't placate the people who dislike any wikidata links as a matter of principle.. but haters gonna hate, what can you do! Wittylama 21:13, 17 January 2018 (UTC)

What you could do is institute actual strict sourcing requirements on Wikidata instead of insisting that there is no problem other than "haters". —David Eppstein (talk) 21:26, 17 January 2018 (UTC)

On "complying with BLP policies"

There seems to be a general misunderstanding here on what Wikidata actually is. My attempt yesterday to get the point across resulted in a series of ad hominem arguments, so that these had to be hatted. I will try the second, and the last time. Whoever wants to listen, may get a chance; whoever has already an opinion, I do not expect them to listen. I will try to be brief.

Wikidata hosts data. It hosts a lot of data with very different provenance (which sometimes is not even recorded). There is data with good provenance and data with bad provenance. The data is used by external users, which include Wikimedia projects. All users use the data in very different ways. The usage of Wikidata by the German Wikipedia is very different from the usage of Wikidata by the Italian Wikivoyage and is different from the usage by the Smithsonian.

I see claims that "Wikidata must comply with the BLP policy of the English Wikipedia". By itself it does not make sense. If this means that Wikidata should only host data which comply with the BLP policy of the English Wikipedia (or with other policies) and delete all other data, this is not going to happen. Never, in any scenario. For a multitude of reasons. One of the reasons is that the English Wikipedia is not the only user of Wikidata.

The correct statement must be "The English Wikipedia must not import data which does not comply with its policies, including BLP". This is a perfectly valid statement. Indeed, Wikipedia must comply with its own policies.

The question is how this is compatible with Wikidata hosting, among others, data with bad provenance. It is clear that only data with good provenance (and by good I mean in this case compliant with the policies of the English Wikipedia) must be imported. Now, there are different scenarios how this is possible.

One scenario is that Wikidata itself must supply the data in the form which complies with the policies of the English Wikipedia, and then Wikipedia might want (or might not want) to import the data. Vandalism fighting on Wikidata aside, this is an unlikely scenario, because, as I mentioned, Wikidata also hosts and will host data which are just not suitable for import here. The conclusion that many users make is that Wikidata is badly designed, and all import must be banned. This is a perfectly valid conclusion, but it must be taken consciously. Indeed, one user here said they would be more comfortable by using Google than by using Wikidata. It is ok. I do believe that the community in general does not support this point of view (and in particular why this RfC is now heading to no consensus leaning oppose), but this is a valid point for an RfC.

Another, a way more realistic scenario, is that end users, including the English Wikipedia, must themselves select which information is suitable to import. This can be done by different means. It can be done by writing complicated lua templates, which would only import information which is sourced to sources from a whitelist, or not sourced to sources from a blacklist. It can be done by using bots and checking the transferred information manually (and in the meanwhile, it could be hidden similar to pending changes). It could be done by bots posting at the talk pages so that human users might later check it and transfer to the articles. There could be more options, or a combination of those. However, this must happen at this side, not at the Wikidata side. (Wikidata might help by better structuring data, for example, by promoting data of certain provenance to a preferred rank, but this is about it. All of these things need some development. Development is complicated and takes time. Nobody would start knowing that after two months work a member of the anti-Wikidata brigade would show up, nominate the developed template on TfD, make an appropriate ad campaign and get the template deleted. If we are going to behave like this, we can just forget about any integration with Wikidata. May be it is not needed and everybody is happy using Google, I do not know. But expectation that all of this will be done at the Wikidata side, if anybody has them, are unreasonable.--Ymblanter (talk) 08:09, 17 January 2018 (UTC)

"My attempt yesterday to get the point across resulted in a series of ad hominem arguments, so that these had to be hatted." You are the one who refused to engage further with another experienced editor because you are an admin and they aren't.

"I see claims that "Wikidata must comply with the BLP policy of the English Wikipedia"." I don't think anyone here has said what Wikidata must do. They can do (and actually do) whatever they like. But whether they comply or not has an impact on whether we will use their data or even link to them.

• "Indeed, one user here said they would be more comfortable by using Google than by using Wikidata. " You have been going on and on about this. As far as I can tell, User:Dirk Beetstra (who is that "one user") has never argued for the import of data from Google, or for the direct linking to Google (search results and the like) in articles. What they claim is that, to find good sources (which is often given as an argument for the links to Wikidata, that it can be used as a pointer to good sources), they'ld rather use Google than Wikidata. Using this "Google vs. Wikidata" again and again in this discussion is an obvious strawman. (see also: "May be it is not needed and everybody is happy using Google, I do not know.")
• "this RfC is now heading to no consensus leaning oppose" No "leaning oppose" is really noticeable. More recent votes have no more value than earlier votes. And in any case, if there is this much opposition to using Wikidata, it should give some pause to the proponents.
• "Nobody would start knowing that after two months work a member of the anti-Wikidata brigade would show up, nominate the developed template on TfD, make an appropriate ad campaign and get the template deleted." Any examples you have in mind? For someone complaining about ad hominem arguments (and I presume you were not complaining about your own ad hominem arguments there), you sure know how to present things in a neutral way which will sway people from all sides. The only "ad campaigns" I have seen is posting a "neutral" statement at Wikidata every time some Wikidata-related discussion happens on enwiki, after which suddenly the "opposes" and "keeps" increase. Or of course "neutrally" mentioning an RfC at some Wikidata convention, after which suddenly people who rarely or never edit here and never have shown any interest in the topic come to the defense of the Wikidata side. If you know of similar "ad campaigns" from people supporting some deletion of Wikidata uses on enwiki, feel free to post examples here. But don't present this as a battle between a poor policy-compliant community who love Wikidata, and some raging "anti-Wikidata brigade" mounting "ad campaigns" to delete your hard work out of sheer malice. Fram (talk) 08:31, 17 January 2018 (UTC)

nothing "on content" here. Second warning to stay on topic. --Francis Schonken (talk) 08:43, 17 January 2018 (UTC)
The following discussion has been closed. Please do not modify it.
I am not going to reply to Fram, as I mentioned earlier, but let me note they developed an unfortunate habit lying about my behavior, as if nobody can check what I actually said. One can easily check also whose statement goes first in the hatted section - surprise-surprise.--Ymblanter (talk) 08:34, 17 January 2018 (UTC) Another personal attack? Either retract the claim of lying, or tell us what exactly is a lie. As an admin who is so proud of their admin status, you should certainly know that accusing people of lying requires some good evidence with it (at least on enwiki, perhaps it's different on Wikidata?) Fram (talk) 08:41, 17 January 2018 (UTC)

I'd close this subsection without further ado: it's only and exclusively about content guidance, without relation to style guidance (must I rememberremind everyone that we are here on the Manual of Style talk page? – discussion of content guidance without style aspect is hardly appropriate here). With the diatribes veering off in even less related realms while, really, people seem to have little to discuss about the style aspect, this subsection should in fact be closed ASAP. --Francis Schonken (talk) 08:50, 17 January 2018 (UTC)

People vote above about the content guidance, not about the style guidance. But anyway it already did not go in a good direction, which I, to be honest, could have expected in advance.--Ymblanter (talk) 08:54, 17 January 2018 (UTC)

Re. "People vote above about the content guidance" – I didn't, so please let everyone speak for themselves.

Also, reminding everyone that this talk page is under DS (see banner above on this page). The DS resulted (at least in part) from diatribes conducted at talk pages such as this one. So I'd carefully, carefully ask everyone to stay on topic. --Francis Schonken (talk) 08:59, 17 January 2018 (UTC)

This is ok, but would you cross out the support votes which had the motivation "Wikidata is unreliable"? Actually, very few votes deal with the style (mine did, for the record).--Ymblanter (talk) 09:03, 17 January 2018 (UTC)

Wikipedia:Manual of Style/Linking#Overlinking and underlinking is part of the style guidance. Some people might find this:

• Frank White Burr

terribly underlinked; others might find this:

• De Danske, Norske og Tÿdske Undersaatters Glæde, TWV 12:10

terribly overlinked: which links to retain can be argued with whatever one thinks a viable rationale in a !voting area (I'll not be the one closing the RfC but any closer would usually weigh relative validity of reasons adhered to by various participants). My problem with the current section is that it is the second one initiated about "general" characteristics of Wikidata (we already have #Discussion (linking to Wikidata RfC), in the OP: "... that Wikidata is universally not useful ..." – emphasis added, which clearly sets the topic of that section as pro and contra that *general* appreciation), often with little connection to how that affects *linking* (this time even less than the previous time). In sum, my rationale for closing this section would be WP:NOT#BLOG or some such (and if not blog-like, at least a duplicate of the topic in #Discussion (linking to Wikidata RfC)). General advantages or disadvantages of Wikidata are hardly the topic here: the topic is whether, and if so under which conditions, we should link to Wikidata (not scores of characteristics of that website which have little or nothing to do with potential linking from Wikipedia). --Francis Schonken (talk) 07:21, 18 January 2018 (UTC)

Speaking of "elementary text", shouldn't the section header be "complying" instead of "complaining"? "Complaining with BLP policies" is rather weird... Fram (talk) 09:16, 17 January 2018 (UTC)

Weird perhaps, but we see it all the time. EEng 23:29, 17 January 2018 (UTC)

• Is there a way from a Wikidata entry to see if there is a link in a Wikipedia to that entry? --RAN (talk) 03:03, 18 January 2018 (UTC)

There's a whole section named Wikipedia with interwiki links on each item... Sabas88 (talk) 14:41, 18 January 2018 (UTC)

I think you are telling me that Wikidata has a field for links for articles in Wikipedia, that I know. When we create a link like d:Q000000 is that findable from within Wikidata? --RAN (talk) 17:44, 18 January 2018 (UTC)

When (and how) SHOULD we link to Wikidata?

From the comments above, it seems that many (not all) editors agree that we should not include links to Wikidata in the TEXT of an article, but many (not all) editors think it appropriate for an ARTICLE to include a single link to the corresponding page (on the article’s topic) in Wikidata... perhaps in a sidebar. Essentially, a note saying “FYI, Wikidata has a page on this topic”, but no more. I could live with that... but perhaps we should discuss this in more detail.

So far, we have focused on when (and how) NOT to link to Wikidata, so perhaps it would be helpful to discuss when (and how) TO link to it. Blueboar (talk) 15:32, 18 January 2018 (UTC)

We already have such link, it's the "Edit links" button under the language interlinks to other Wikipedias. I've seen the Spanish Wikipedia using more explicit links, with an Authority control template at the bottom of the page showing the Q number at Wikidata for the article's topic, but both point to the same place.

I'd prefer the discussion to be more fine-grained, and we studied when and whether we could create links to Wikidata for entities that are not whole articles, but are described within our articles and for which a Wikidata page exist. Authority control is a basic tool in library science, and Wikidata could be a perfect complement for incorporating it to our encyclopedia, if we find a way to keep in check its shortcomings regarding reliability. Diego (talk) 16:54, 18 January 2018 (UTC)

P.S. The links could take the form of a dfn tag, which doesn't create a link to the URL, but instead can be used to show the item's definition as a tooltip. I think this usage would be consistent with our WP:JARGON policy; we certainly often use Wiktionary as an external link in that way. Diego (talk) 17:12, 18 January 2018 (UTC)

Where this conversation seems to get stuck is along these lines: "There are recurrent problems with Wikidata information in Wikipedia articles, in part because the readers are being deceived about where the information comes from." "But Wikidata only exists as a database to feed information to other projects, so we can't send readers to Wikidata, or segregate Wikidata information in Wikipedia articles, and anyone who says different is trying to erase Wikidata from Wikipedia and kill it off." But ... who says the Wikidata community isn't responsible for presenting information in a way that would make sense to readers? Who says we can't send readers to Wikidata, rather than deceiving readers about where information on Wikipedia pages is coming from? I get that that's not their plan. "I'm only responsible for this one thing that I know how to do, you guys are responsible for everything else, and if we fail, if it causes problems, then that's your fault". I get why people want to say that, I'm even sympathetic, and I admit that there are specific exceptions where Wikidata links work without causing too much trouble ... but as a general philosopy of just sticking Wikidata links in wherever, whenever, it's not working and I can't support it. - Dank (push to talk) 17:00, 18 January 2018 (UTC)

@Dank: Where do get this Wikidata only exists as a database to feed information to other projects and the Wikidata community isn't responsible for presenting information in a way that would make sense to readers from? I think there are very few people working on/with Wikidata who would subscribe to either of those views. Although User:Fram might not be a fan, most of us I think would regard Reasonator as a doing really a pretty good job of formatting Wikidata information in a reader-friendly way, extensively use it ourselves to 'read' Wikidata, and routinely include columns of links to it from workpages such as eg Talk:List of Royal Academicians/RAs. Given many of the entries on that page have Wikidata items with in excess of 30-40 matches to external databases, I for one find that Reasonator presents them in a very effective and usable way. But, hey, we're just editors, trying (across en-wiki, Commons, Wikidata, other wikis) to improve coverage and add information, and making pages to help ourselves and other editors to do that, so what do we know? Jheald (talk) 18:35, 18 January 2018 (UTC)

But to add to the above, there are some very valuable external sites with links that are only somewhat stable and need to be regularly checked, because routinely some get changed (without forwarding). It is far easier to manage, check and routinely update those links centrally, and then present them via a template here. So I make no apologies that links to at least one site like that are now run entirely through Wikidata. Jheald (talk) 18:48, 18 January 2018 (UTC)

I'm sorry it sounded that way. I can't respond right now ... I just got a month's worth of TFA blurbs to do, I'll be back when I get done with those. - Dank (push to talk) 22:18, 18 January 2018 (UTC)

Sorry, it's going to be a while before I can get back to this. - Dank (push to talk) 19:35, 21 January 2018 (UTC)

Living people and Wikidata

The main concern for not linking to Wikidata is concerns about living people.

• Wikipedia already links to Wikidata for living people that have Wikipedia entries, whatever concerns there are, already apply.
• There is concern about errors in Wikidata, they are handled the same way they are in Wikipedia, by correcting.
• There is concern about referencing in Wikidata, "unreferenced at the moment", is not the same as "inherently unreferencable". There are over 10,000 Wikipedia articles on living people that have unreferenced paragraphs. There are over 10,000 Wikipedia articles on living people in which the birth date given is unreferenced. I stopped the search at 10K, someone else can do a more accurate count by letting the script run longer.
• Wikipedia already links to VIAF and LCCN as well as other Authority Control databases for living people that lists their birth year.
• Wikipedia already links to news articles as references that may give a person's birthday, and name their minor children, and name the city they live in.
• Wikipedia can just ban links to living people at Wikidata, not dead people, and not reference works.
• Wikipedia already allows interwiki links to any other language Wikipedias using {{Ill}}, when an article appears there.
• There was a concern that Wikidata's policy on living people does not 100% match the English Wikipedia policy, no other language Wikipedia does. Nor does Wikimedia Commons, Wikiquote, or Wikisource, all of which we link to. However, all do follow the Wikimedia Foundation's policy on living people.
• People also want to link to reference works in the reference section, that do not have ISBN entries.
• There are over 50,000 biographies of living people in which there is not a single reference to any data point displayed in the infobox.

--RAN (talk) 14:57, 19 January 2018 (UTC)

At least for me, the main concern about linking to Wikidata is

• Wikidata is not a reliable source
• Wikidata is poor in vandal fighting
• Wikidata is not reader-friendly (and Reasonator reads like what it is, a bot-created surrogate)
• All this is the most worrying for BLPs, but problematic enough in general
• For many links we currently have to Wikidata, there is no excuse for them: often they are links where the linked page already exists on enwiki anyway, and the sources in Wikidata are already listed at enwiki as well, so the page adds nothing at all

I'm no fan of using {{Ill}} to other wikis either, for much the same reasons. But at least these pages are structured the same as enwiki pages, and as long as you link to the bigger language versions (like de), you may be pretty sure that they do adequate vandalfighting, notability checks, BLP checks...

No one (I think) has brought up the issues of birthdays, minor children, ..., that seems like a strawman. Furthermore, there is no problem to link to reliable sites with this information. LCCN is considered a reliable site, Wikidata isn't.

It has never been a problem linking to reference works which don't have an ISBN, so no idea why you bring this up. Fram (talk) 15:11, 19 January 2018 (UTC)

No, my main concerns are not BLP. However it's inefficient to bother with complex or debatable issues when BLP is such a simple and blatantly fatal point to terminate any debate. I also need to call out the fundamentally false comment about "Wikidata's policy on living people does not 100% match the English Wikipedia policy". No. Wikidata does not have any such policy at all. Wikidata is in violation of the mission-wide resolution for all projects to create such a policy. And fundamentally, it's not even the lack of policy that's the BLP issue. The real BLP issue is the pathological community. All of Wikidata's attempts to create such a policy, up to and including today, are FUBAR because of the pathological community. A majority are opposed to living-persons-violations being listed as a blockable offense. Anything less than 90% support on that is appalling. Alsee (talk) 09:53, 4 February 2018 (UTC)

More on styles

• To me, the major issue is that blue-linking to Wikidata in article text creates the impression that there is a en.Wikipedia article on the linked subject/topic... when, in fact, the reader is sent to a page at Wikidata (or to put it another way... the link is presented as a link to a page inside en.Wikipedia, when it is actually a link to a page outside of en.Wikipedia). Blueboar (talk) 01:03, 20 January 2018 (UTC)
  • Maybe the MOS issue would be clearer if we set Wikidata aside and formulated it more generally as "Never assign direct interwiki links to the text of an article. All such links should be made either as a soft redirect or as a special marker that is not part of the main article text (such as a footnote or parenthetical interlanguage link)."? Because much of the discussion here has focused on "it's useful" vs "it violates BLP", neither of which is really a style issue, so it would help if we could decouple link style from the more general issue of whether we should link to Wikidata at all. —David Eppstein (talk) 01:18, 20 January 2018 (UTC)
    • Hmm... yes... I like that (because the issue isn’t limited to just Wikidata). It also keeps the focus of the concern to the appropriate page... Deal with the STYLE problem here at the MOS, and deal with any OTHER problems that people have with Wikidata (like the BLP concern) at the appropriate other policy and guideline pages (where they relate). Blueboar (talk) 02:30, 20 January 2018 (UTC)
    • I'd have no objection to that for prose, but we need to change "(such as a footnote or parenthetical interlanguage link)" to "(such as a footnote, table, or parenthetical interlanguage link)". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:33, 20 January 2018 (UTC)
      • How are interwiki links in tables any different than interwiki links in prose? Blueboar (talk) 12:25, 20 January 2018 (UTC)
        • They can have a column header. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:15, 20 January 2018 (UTC)
          • That would only be a valid explanation if you intend for *all* entries in a table column to link to the same interwiki site, even when there is an internal article link that could be used instead. When would that ever be a useful thing to do? —David Eppstein (talk) 20:32, 20 January 2018 (UTC)
            • When we have tables containing subjects significant enough to be in list articles but not necessarily the subject of their own article, and where there is more content relating to those subjects on sister projects such as images of listed buildings (monuments), texts by authors, data on species, or indeed on people. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:50, 20 January 2018 (UTC)
• I thought it was actually a slightly different blue colour from internal links, although not by much (but I haven't checked the colour codes recently). Presumably the link *could* be more of a different colour, but I'm not sure what the history / other uses of this are. Thanks. Mike Peel (talk) 13:55, 20 January 2018 (UTC)
  • Hmmm... I might be more accepting of interwiki links if the link were a significantly different colour (say bright yellow). My primary STYLE concern is avoiding surprise.... we don't want our readers to be confused when they click on a link, and end up being taken to a page in a sister project (when they were expecting to be taken to an article here on WP.en). We need something that makes it very clear to the reader (before they click on a link) that they are about to be taken out of WP.en... so they will know: "ah... if I click this, I will be taken to a sister project". Blueboar (talk) 14:46, 20 January 2018 (UTC)

Of course, only we cognoscenti will know what the yellow means... EEng 15:04, 20 January 2018 (UTC)

A valid point... but you could say the same about the difference between redlinks and bluelinks. Yet, everyone figures that out quickly. If we used a distinct colour for interwiki links, the non-cognoscenti would at least see that something was different about the link (and be less surprised when taken to a different project). Sure, at first, they might not know what the different colour indicated, but they would figure it out quickly. They would see a link in yellow (or orange or whatever), and wonder: "hmm, why is this link a different colour?"... (CLICK)... "ah, I see. Links in that colour take you to an interwiki page. Got it." Blueboar (talk) 16:42, 20 January 2018 (UTC)

I hate to break it to you, but I doubt more than 1 in 100 readers have any concept of an interwiki link, or understand that we have different "wikis", or anything like that. Red vs. blue is easy to grasp -- you get nothing, or something. EEng 19:17, 20 January 2018 (UTC)

I doubt that (or at least doubt that they'd have any difficulty making the leap of understanding Blueboar outlines after an initial experience or two). Even children understand how domain names work, e.g. that www.steam.com, steamcommunity.com, and support.steampowered.com are all different "places" within the Steam gaming system. At Meta, there's already a feature (optional in the preferences) to use different link colors for different projects. (It's not perfect; it uses more than color, including underlining and other style, which can be confusing, since the underlines look like those produced by <abbr> and by <element title="...">; but with some additional work it would be better, and it's already more useful than it is a hindrance).  — SMcCandlish ☏ ¢ 😼  19:04, 13 February 2018 (UTC)

Children understand domain names if they happen to look at them, or have some reason to. Most people don't. EEng 03:02, 14 February 2018 (UTC)

Not allowed to remove useless links to Wikidata?

Apparently even utterly useless links to Wikidata may not be removed, if the lead from User:Wittylama and User:Pigsonthewing is to be followed. The Offerings of Peace and The Offerings of War is an article given above by Wittylama as an example of where the Wikidata icon is used. I commented above with some remarks about this, after which Wittylama acted on some of the not-Wikidata related remarks but ignored the Wikidata ones. So I removed the icons, only to get reverted by the above two editors, despite asking repeatedly at User talk:Wittylama#The Offerings of Peace and The Offerings of War what these Wikidata links added for the reader.

As you can see from the diffs, there are three links to Wikidata items[8]. The first two are for the two artworks described collectively in this article (they have an individual Wikidata item, and the two together also have a Wikidata item). The third is to a later artwork made as a hommage to the works, which again is described in the article but has a separate Wikidata item.

The three Wikidata items contain no information not already present in the article. The only source at Wikidata is given in the same line in the enwiki articles.

Insisting that some link to an unreliable site, which adds literally nothing of information for any reader, must be kept, is a tell-tale sign of spam: not the enwiki article is important, not the reader is important, but the link, the website linked to, is the main concern. Spam gets routinely removed, and insistent spammers routinely blocked. Can someone indicate what information for readers these specific links actually add which isn't already present in the article, and if none, why these shouldn't be removed? Fram (talk) 15:02, 22 January 2018 (UTC)

The Wikidata item contains the same information, but it's in tabular format. So, it's no different than having an infobox, except that it's placed in a different website. But given the disputes provoked by infoboxes, it's no wonder that the same debate about the usefullness of structured information is repeated about Wikidata. Diego (talk) 15:22, 22 January 2018 (UTC)

"it's no different than having an infobox, except that it's placed in a different website." Which basically makes all the difference. If I make "wikiinfoboxes.com" where I repeat information from articles, I wouldn't be allowed to link to it either. Plus: infoboxes are supposed to be at-a-glance information, which is hardly the forte of Wikidata. An infobox should be reader-friendly, and Wikidata isn't. Fram (talk) 15:44, 22 January 2018 (UTC)

Comprehensive responses to each of Fram's complaints have been made over on my talkpage, at the subsection which Fram started: User talk:Wittylama#The Offerings of Peace and The Offerings of War. It does not automatically make Wikidata links "spam" just because Fram don't agree with these responses and does not believe Wikidata to be a useful supplement to Wikipedia articles in general. Anyone wishing to weigh in here might like to read that conversation on my talkpage first. Wittylama 15:33, 22 January 2018 (UTC)

"Comprehensive responses"? No indication has been given of what information a reader can find at these Wikidata pages which isn't already present directly in the article. "Anyone wishing to weigh in here might like to read that conversation on my talkpage first." Well, yes, that's why I linked to it. I didn't claim that Wikidata links are automatically spam: I claimed that these links are spam, because these links, in these articles, don't add anything but are only useful if you want a link (no, multiple links) to Wikidata no matter what. If Wikidata proponents want to include Wikidata links in the body of articles, preferably with a fancy flag-like icon, even if they serve no purpose for our readers and don't offer any aditional information, then it is no surprise that there is such a heavy backlash against these. Fram (talk) 15:44, 22 January 2018 (UTC)

• Comment- these really do seem like just linking to Wikidata just for the sake of linking to Wikidata, adding extra clutter and an unexpected perplexing diversion to an unreadable page for the hapless reader. Reyk _YO! 15:46, 22 January 2018 (UTC)
• Comment- let's wait for the RfC above to see whether there is consensus to link to WikiData in article content, then we can start a second RfC (because that is what it likely is going to need) to gauge what to do with the already existing links and possibly remove the fait accompli, and possibly an RfC for data that is pulled from WikiData (if it is unreliable to link to, it is also unreliable to pull data from them - and note that on BLPs data is pulled from WikiData to transclude in our content). --Dirk Beetstra ^{T C} 06:37, 23 January 2018 (UTC)

Another template for discussion

For anyone interested in this discussion, there is also a discussion on the Soft Wikidata recirect template that User:Richard Arthur Norton (1958- ) recently created as another way to link to wikidata within the body of an article--Rusf10 (talk) 01:26, 27 January 2018 (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.