Following this discussion in mid-2014, there was a determination of consensus that the cite doi template should be deprecated. Now, a bot is making large-scale substitutions of the template in article space, bringing this to the attention of more editors. Should the cite doi template be deprecated? —Alex (Ashill | talk | contribs) 00:06, 7 September 2015 (UTC)[reply]

Deprecate[edit]

• Support Per explanation in discussion section. -- Ricky81682 (talk) 00:26, 7 September 2015 (UTC)[reply]
• Support. The problem here is the exponential growth of templates. When you create a separate template for every single reference you create, as this one does, you create a number of templates that is impossible to maintain. What happens when the reference is removed? We have a left over template with no transclusions that is merely taking up space. All of these templates that you're creating are just wrappers for Cite journal anyway, so this is heavily redundant. The end result is templates-within-templates, often being discarded and left laying about, and every single one of them could have been accomplished by a single easy-to-use template that leaves no mess. The accessibility issues for new users are quite serious as well; we can not simultaneously hold up the "anyone can edit" philosophy as our ideal and encourage the use of a cite journal inside a single use template inside a cite DOI. Those are completely contradictory, as the latter makes it impossible for anyone unfamiliar with the fine details of the project to figure out how to edit the thing. Lastly, to Curly Turkey who unhelpfully said "fix it", there's nothing to fix here. The fundamental nature of how this template was used is flawed. The fix is to use the template that's actually in use under two layers of wrapper here: Cite journal. It's a very easy to use citation template that is extremely friendly to any editor, especially when using refToolbar. ~ Rob^Talk 02:41, 7 September 2015 (UTC)[reply]
  • How was my comment "unhelpful"? There have already been proposals to reimplement the template using WikiData rather than subtemplates—the suggestion is not idle. And if redundancy (?) is a concern, keep in mind the whole reason these templates came into being in the first place: if a book is used as a reference on 20 pages, and a user corrects an error in the citation on one of those pages, the other 19 will remain in error. Editors are thus unnecessarily burdened with tedious, hard-to-track maintenance issues when they could be generating or improving content. Curly Turkey 🍁 ¡gobble! 03:33, 7 September 2015 (UTC)[reply]
• Support We need to make things easier not harder. This is a poor template that hides some of our most important content (references) off on subpages. It makes it harder for new and old editors alike. People can keep using the template but a bot should come along and replace it with "cite journal" We also need to make translation easier. This template is not supported in most other languages. For medicine there was consensus back in Mar 2014 [2] for "cite pmid" for similar reasons Doc James (talk · contribs · email) 02:54, 7 September 2015 (UTC)[reply]
• Support. I don't see a good reason to change the decision, and a lot of reasons to keep it deprecated. These reasons include potential for long-term vandalism because all the little template pages aren't watchlisted; problems keeping formats consistent between pages in citation style CS2 vs citations using these templates which are in CS1; restrictive author name format (if I remember correctly, cite doi disallows spelling out author first names, and this is enforced by bot reversion of such changes); difficulty making changes to references that use this especially when noedit is set; etc. I do think we should eventually move to a more centralized and reusable citation system but it should be done through wikidata, not this way. —David Eppstein (talk) 03:32, 7 September 2015 (UTC)[reply]
• Support. I don't see any reason to revert the earlier decision to deprecate this template. --Randykitty (talk) 09:04, 7 September 2015 (UTC)[reply]
• Support per comment in discussion section. Graham87 10:31, 7 September 2015 (UTC)[reply]
• support it would be best to deprecate it--Ozzie10aaaa (talk) 13:26, 7 September 2015 (UTC)[reply]
• Support I agree with the removal. All references should be given with a concrete way and not by creating dozens of subpages. -- Magioladitis (talk) 16:48, 7 September 2015 (UTC)[reply]
• Support I hate the giant clots of wikisnot left in article markup by cite templates and migrating citation data to wikidata can't come fast enough, but this was never a great solution to the problem and is well past the point where action is needed on the deprecation. Opabinia regalis (talk) 18:42, 7 September 2015 (UTC)[reply]
• Support Transcluded templates (1) are often at odds with consistent citation formatting and contrary to WP:CITEVAR, (2) separate statements from the sources used to support those statements making it more difficult to see what the ref is when editing, and (3) unneeded. Other tools such as RefToolbar and Wikipedia template filling tool make it easy for editors to add full citations directly to the Wiki Text. List-defined references are preferable to transcluded templates to reduce clutter in wikitext. Finally if this template is retained, the |noedit option needs to be removed since it completely hides where the citation is stored. This makes it difficult for conscientious editors to undo vandalism. Boghog (talk) 19:42, 7 September 2015 (UTC)[reply]
  • How are they contrary to CITEVAR? Nobody's forced (or even encouraged) to use them. Curly Turkey 🍁 ¡gobble! 23:02, 7 September 2015 (UTC)[reply]
    • The reason why I don't like this template is that it enforces a certain citation style that often conflicts with WP:CITEVAR (see this discussion). One could add a ref name tag to provide more details about the reference and one could add passthrough parameters to allow for variations in formatting, but then the template and associated ref tag starts to become so long that it starts to defeat the purpose of using a transcluded template. Better to substitute ((cite pmid)) with ((cite journal)). Boghog (talk) 13:11, 8 September 2015 (UTC)[reply]
      • (a) I repeat: nobody is forced or even encouraged to use these templates, so CITEVAR doesn't even come into the picture (b) the point of these templates is not to make the citation short, but to keep it maintainable (especially across large numbers of articles), so adding parameters is no issue. I add a |ref=harv to every ((cite isbn)) I use so I can use ((sfn))s. This does not "defeat the purpose", as the purpose is not to save space. Curly Turkey 🍁 ¡gobble! 21:15, 8 September 2015 (UTC)[reply]
        • If no one is encouraged to use these templates, then why not deprecate? The strongest argument in favor of this template is to reduce clutter in the Wiki text. The strongest argument against this template is that the citation data that is stored in template space is not regularly monitored and therefore is subject to vandalism. Hence these templates degrade, not enhance maintenance. Boghog (talk) 19:12, 9 September 2015 (UTC)[reply]

• support With the bot this scheme all but relies on broken, moving the references inline is probably for the best, or at least stop new pages being made. Additionally, the reference toolbar and VE can now insert references using ((cite journal)) or similar that make this easier - I can't see what use having these as subpages does, other than increasing load times (as transcluding from extra pages can have a higher impact on load times); I know WP:DWAP, but we still need to be mindful. Also, if content is migrated to Wikidata (which I feel is unlikely, that's not really what it is for), then it is easy enough to move this over later. Mdann52 (talk) 16:36, 8 September 2015 (UTC)[reply]
• Support refToolbar already has a DOI search function when using ((cite journal)). There is no reason to have a separate template that duplicates that functionality, especially when the refToolbar fills in all the information for you! Primefac (talk) 16:02, 9 September 2015 (UTC)[reply]
• Support: for me, having the reference information embodied within an article is useful for maintaining reference format consistency. Praemonitus (talk) 22:42, 9 September 2015 (UTC)[reply]
  • Nothing prevents you from doing so—((cite doi)) is entirely optional. Curly Turkey 🍁 ¡gobble! 23:06, 9 September 2015 (UTC)[reply]
    • The problem is other people using it. They can be elimenated, but it would have been less work if cite doi was not used in the first place. Graeme Bartlett (talk) 10:53, 10 September 2015 (UTC)[reply]
• Support deprecation unless cite doi can be instantly filled by some tool - in place (and most certainly not as an unwatched sub-page) using some smart subst-ing and use of external resolver tools if needed. Shyamal (talk) 05:19, 10 September 2015 (UTC)[reply]
• Support - everyone else summed this up; I see no need for a detailed explaination since it's basically a combination of other reasons. KieranTribe 08:56, 10 September 2015 (UTC)[reply]
• Strong Support It is extra complicated to maintain these when an article has many uses. This is especially so when there are errors in the template involved. A bot does not have the eyes of a person, and plenty of low quality material has been imported automatically. It might only get checked on a FA review. And I suspect there are thousands of errors scattered throughout these templates. By-the-way substing does not work well, and to get them included inline the copy and paste is the way to leave out the crud. Graeme Bartlett (talk) 10:46, 10 September 2015 (UTC)[reply]
• Support deprecation. ((cite journal|doi=...)) does the same job but better, and avoids creating huge numbers of subtemplates. I was not involved in the earlier discussion, but would have argued for deprecation and nothing has changed since then. Let's finally kill ((cite doi)) off; substing and then deleting the subtemplate would seem to do the job. Note I came here via an invitation on WT:AST Modest Genius ^talk 11:06, 10 September 2015 (UTC)[reply]
• Strong support I was in favour of keeping them last time we had an RfC on the issue, and I've switched my to deprecate and subst. The main reason is simply that while ((cite doi)) is a valid style, it has a very rigid structure and is a nightmare to maintain, even across one single article. For instance, the author names must be in 'Last, F.' format. This means that if someone later adds a ((cite book)) in the article, then need to make sure that they also declare author names in 'Last, F.' format. Or if someone else uses an automatic ref-filling tool, which instead uses 'Last, F', 'Last F', 'F Last', 'F. Last', 'First Last', 'Last, First' or any other variant.

The second reason is that while <ref>((cite doi|10.1234/foobar))</ref> certainly looks cleaner in the edit window, it not user friendly. It means that on an article using 50 ((cite doi)) templates, you need to keep 51+ different pages in your watchlist if you want to fully know what's happening to that one article. It also means that two different articles using the same ((cite doi)) cannot be customized independently of one another, for instance if in one article it would be warranted to wikilink the authors, and if in the other this is undesired. Headbomb {talk / contribs / physics / books} 17:08, 10 September 2015 (UTC)[reply]

• Strong support DELETION deprecation and delete ASAP. The ((cite doi)) system turns Wikipedia into a DOI database (a clear violation of WP:NOT; specifically not a repository; these are loosely associated content related only by the fact they have a DOI number) since the bot that maintains the DOI citation template generates a new subtemplate for each time a new DOI is used, as ((cite doi/doinumberhere)) containing citation information, instead of replacing the ((cite doi)) with a ((cite journal)) transclusion on the page it is used on. Since many items have DOIs as well as other UIDs such as PubMed IDs, we have multiple different sets of the same information in the various databases that the various bots generate for each of these template systems, depending on which bot created what subtemplate into which database entry on whichever template system the bot maintains. This makes multiple subtemplates in multiple systems redundant and replicative of each other, failing TEMPLATE DUPLICATION deletion criterion. Many of these fall out of use, but are not deleted, making Wikipedia a permanent database of DOIs, something it should not be. -- 70.51.202.113 (talk) 04:50, 11 September 2015 (UTC)[reply]
• Mostly oppose deprecation, but definitely remove |noedit=. I support for some ideas below. If someone does the work to make the template safesubst'able, and the subst is required, that would probably be all we need.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  11:39, 12 September 2015 (UTC) Updated: 14:58, 12 September 2015 (UTC)[reply]
• Strong support, per what I said in the last discussion. APerson (talk!) 13:46, 21 September 2015 (UTC)[reply]
• Support - Not practical to maintain and creates style inconsistencies. Kaldari (talk) 05:53, 22 September 2015 (UTC)[reply]

Don't deprecate[edit]

• Don't deprecate (as proposer). I object to the deprecation of this template both on process and substance grounds. First, I don't think that the closure of the previous discussion accurately reflected the consensus in the discussion but instead imposed the closer's views; it is very clear to me that there was either no consensus or a consensus to not deprecate at the time. But I'll focus on the substance here.

1. The strongest rationale the closer used for deprecating in the previous discussion was the principle that article content should be in articles, not templates. But the editorial decision made by editors is which reference to use, and a DOI by its very nature uniquely, completely, and permanently identifies the reference. Therefore, from a content point of view, ((cite doi 10.1234/56789)) is the entire portion of the citation that is content and fully satisfies WP:V. Everything else in the ((cite journal)) template is just metadata. (It's very important metadata, of course, completely necessary for a human reader to read the article, but still just metadata, not article content that has any compelling policy need to be in the article source.)
2. Vandalism of the large number of lightly-watched doi templates was raised as a concern in the previous discussion. As far as I can tell, this is a purely hypothetical concern with no basis in experience, so it doesn't worry me much.
3. Concerns about style consistency were also raised. If the cite doi template does not produce consistent style on an article, it shouldn't be used in that article, just like the citation and cite XXX templates technically shouldn't be mixed. That doesn't mean that cite doi shouldn't or can't be used in articles which do use the cite xxx family (since cite doi used the cite journal template). Moreover, worrying about the differences between the formatting created by different citation templates is only for the small minority of very good (typically featured article candidate) articles; much more common is to worry about the lack of any references at all or bare urls. To the extent that cite doi makes it easier for editors to add citations, that's a far bigger benefit to the project than this issue.
4. Even if the consensus is that the cite doi templates should never be used in article space, it would be useful to keep them around with a requirement that they always be substituted. It would be easier to type ((subst:cite doi 10.1234/56789)) than to type out the entire reference; I see no reason (except performance, which we're not supposed to worry about) not to keep the cite doi templates around and allow bots or automated tools to create new ones just as a useful tool for this reason.
5. Fundamentally and honestly, my preference for cite doi comes down to WP:ILIKEIT: I find it very useful to be able to just copy and paste the doi from a source and let an existing template or a bot handle putting the metadata in place (and bots tend to be better than humans at handling metadata). Also, I find wiki source easier to edit when it's not cluttered by very long references. For example, Dexbot's transclusion of cite doi templates at Milky Way added 25 KB (over 20%) to the length of the article source. Again, those who don't like the cite doi templates don't have to use them.

Ordinarily, WP:ILIKEIT is not a valid rationale. But the whole point of a template is to be useful; at some level, every template exists because someone likes it, and no template is every mandatory. Many of the arguments against use of this template, to me, boil down to WP:IDONTLIKEIT (and I believe I've addressed the major arguments that aren't IDONTLIKEIT above). Therefore, the decision whether to use this template should be left to individual editors or individual articles and projects; I see no need to make a site-wide decision to prevent editors who like the template from using it. —Alex (Ashill | talk | contribs) 00:37, 7 September 2015 (UTC)[reply]

Do you want to have a singular discussion section or not? It's impossible to have a coherent conversation if you want to comment up here. -- Ricky81682 (talk) 01:02, 7 September 2015 (UTC)[reply]

• Don't deprecate—if there are issues with how the template is implemented, fix it, don't disrupt the work of productive editors. Curly Turkey 🍁 ¡gobble! 01:33, 7 September 2015 (UTC)[reply]
• Keep' don't let the Luddites and morons win this one. Everyone is better off learning how to program and use templates. If new users don't get it, so be it. 166.170.51.68 (talk) 01:51, 7 September 2015 (UTC) "Vote" removed. This user is blocked under various IPs including 166.176.58.220, and is therefore not able to participate in this discussion. Comment preserved for historical purposes and due to responses. Feel free to ask on my talk for more information. ~ Rob^Talk 20:27, 9 September 2015 (UTC)[reply]

I wonder if this comment is being sarcastic? Doc James (talk · contribs · email) 02:55, 7 September 2015 (UTC)[reply]

no, editors should learn how to operate here. We shouldn't coddle people. — Preceding unsigned comment added by 166.176.57.125 (talk) 03:02, 7 September 2015 (UTC)[reply]

Okay thanks. IMO we should keep things simpler when possible. Doc James (talk · contribs · email) 03:04, 7 September 2015 (UTC)[reply]

"Keep this because people should use templates" is of a strange argument, when the question is "Should we use Template X or Template Y?", rather than "Should we use this template or no templates at all?" WhatamIdoing (talk) 03:24, 7 September 2015 (UTC)[reply]

Sounds more like people should learn how to use the layering of templates if need be. I mean, I'm consider getting a bare url a bronze medal, getting the ref tags right is a silver, and getting the full citation template (from the Reftoolbar if nothing else) a gold medal winning work. Saying they need to learn that cite doi is actually a hidden page with a cite journal citation if they want to fix something is just asking for the moon from everyone. -- Ricky81682 (talk) 09:11, 8 September 2015 (UTC)[reply]

• Keep as an option for people who want to use it. (I'm not one of those people.) WhatamIdoing (talk) 03:24, 7 September 2015 (UTC)[reply]

But what about the other editors on each page? They won't know how it operates and shouldn't have to learn another layer of complexity just to help. People have for years created their own citation templates and argued that everyone should let them be but they become harder and harder to rip out at time progresses. -- Ricky81682 (talk) 05:24, 7 September 2015 (UTC)[reply]

From the gist of the discussion that template was completely redundant to ((cite book)), and nobody appears to be defending it. You don't seem to understand the whole point of the template—it is not to move data into templates (where the data resides is irrelevant). The whole point is to reduce redundancy and ease maintenance (so people can spend more time on improving content). I hope you don't have it in your head that editors are up in arms over not being allowed "hide away" data in templates—or do you? Curly Turkey 🍁 ¡gobble! 05:44, 7 September 2015 (UTC)[reply]

That would make sense if these citations were in mass use (see Category:Citation templates for plenty of examples). Here, it's only being used once if at all so it's not reducing any redundancy here. To me, it seems like editors are so focused on "this is easy for me to get so I just do it once and no one will ever need to edit it" and presume that the citation bot or whoever fills in the citation will do it correctly so much over whether or not other editors "get" how this works. This is just layers of wrapping over a cite journal citation that could easily be copied and pasted. -- Ricky81682 (talk) 05:51, 7 September 2015 (UTC)[reply]

Perhaps that's the case with ((cite doi)), but it's not with ((cite isbn)) (or wasn't—perhaps it is now after so many have been destroyed). The only reason I ever went through the trouble of setting up a ((cite isbn)) was when I knew I was going to use it in multiple articles (as with all my ukiyo-e, Winsor McCay and Chester Brown-related articles). That was the whole point of the template. With ((cite doi)) the subtemplates were bot-generated, so it shouldn't be surprising that so many single-use ones were created. Solution? Stop the bot. Not too many people are going to go to the effort of hand-creating a single-use ((cite doi))—it's extra work. Curly Turkey 🍁 ¡gobble! 06:17, 7 September 2015 (UTC)[reply]

The bot (AFAIK) is not running any more - hence why this has come up! Mdann52 (talk) 16:08, 7 September 2015 (UTC)[reply]

• Keep. As a previously-new user, I can testify that it was much easier for me to use Cite doi, which only requires a single parameter, than to use all the parameters of the other citation templates. Having a template such as Cite doi is important if we want to encourage new users to use citations in their articles. --Erel Segal (talk) 05:33, 7 September 2015 (UTC)[reply]

Except it doesn't do that. It used to when citation bot filled in the details when people put in a cite doi. It's possible to create a citation bot in the future that takes cite dois that just use the citation parameter and then fill in the details in the article itself at which point all of these arguing is moot. -- Ricky81682 (talk) 05:51, 7 September 2015 (UTC)[reply]

It's not moot—we still have maintenance and redundancy issues. Curly Turkey 🍁 ¡gobble! 06:22, 7 September 2015 (UTC)[reply]

User:Erel Segal you can still use it. A bot will just come around and fill in the rest of the parameters for you and switch it over to cite journal. Doc James (talk · contribs · email) 09:23, 7 September 2015 (UTC)[reply]

User talk:Doc James "you can still use it" - but the Template page clearly says that "Template:Cite doi/doc is being discussed for deletion in accordance with Wikipedia's deletion policy". This means that I (and other editors) will not be able to use it at all.

If the intention of the deprecators is that Cite doi will remain in place, and the only change will be in the behaviour of the bot - then this should clearly stated in the template body. --Erel Segal (talk) 10:11, 7 September 2015 (UTC)[reply]

Yes it says deprecate up above rather than delete. Yes we should update the template to say this. Doc James (talk · contribs · email) 21:34, 7 September 2015 (UTC)[reply]

"Deprecate" is also not a good term because it implies that it may be deleted soon.

I think Cite doi should be free to use by anyone who finds it useful, without any "deprecation" notice. The discussion should be only on the behaviour of the bot. --Erel Segal (talk) 06:34, 8 September 2015 (UTC)[reply]

• Keep I've found the cite doi template useful. Also I disagree with the statement that other editors

find the cite doi template confusing. For example, the article Meridian arc used to contain

    a result which was first obtained by
    Ivory.<ref>((cite doi
    |10.1017/s0080456800030817
    |comment = Ivory 1798
    |noedit
    ))</ref>

which was succinct and, provided you know what a DOI is (a given in most scientific fields), informative. Now it's

    a result which was first obtained by
    Ivory.<ref>((Cite journal| ref = harv| author1-link = James Ivory (mathematician)| doi = 10.1017/s0080456800030817| title = A New Series for the Rectification of the Ellipsis| journal = Transactions of the Royal Society of Edinburgh| volume = 4| issue = 2| pages = 177–190| year = 1798| last1 = Ivory | first1 = J. |url = https://books.google.com/books?id=FaUaqZZYYPAC&pg=PA177))</ref>

which is distinctly more intimidating.

...But your second example uses ((Cite journal)), which most of the supporters are advocating as the alternative to ((Cite DOI)). ~ Rob^Talk 14:50, 7 September 2015 (UTC)[reply]

Right, this is the result of the bot replacing ((cite doi)) with ((cite journal)), and thereby junking up the article, in my opinion. (Sorry that I didn't make this clear.) cffk (talk) 15:37, 7 September 2015 (UTC)[reply]

I don't find it "distinctly more intimidating at all", because any correction I need to make can be made right there, and all the info I want to see is right there in the code. What's intimidating is "how TF do I correct this typo in the title when I can't even tell how it's getting into the article?" Heh.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  14:56, 12 September 2015 (UTC)[reply]

And I think the second one is less intimidating if someone had just read the text and see all those names and words first. Cite journal is ugly in full but not impossible to figure out. Cite doi is hidden in full. If I read the text, and the citation is wrong for some reason and so I click "edit" and I get this cite doi and I can't figure out how to help, how are we better off? The first option only works if it was done perfectly the first time and even citation bot wasn't that perfect. -- Ricky81682 (talk) 09:15, 8 September 2015 (UTC)[reply]

It's even worse than that, for the example above, because the noedit option was used. So there's no "edit" for you to click on and no way to figure out where to go to fix the incorrect citation. —David Eppstein (talk) 06:21, 14 September 2015 (UTC)[reply]

• Keep, for the readability reason just given above (most of the time, citations can be as short as

  ((cite doi|10.1017/s0080456800030817))

  ), and for only having to fix citations once (e.g. changing ((Cite journal)) to ((Cite conference))). Because there are articles cited more than once, even if many ((Cite doi)) may currently only be used once (and sometimes, they could be used more than once - just some citations predate the template maybe? then the fix is to use the template everywhere, instead of inlining). --Chire (talk) 09:19, 8 September 2015 (UTC)[reply]
• Kinda-sorta Keep, conditionally: I appreciate the template's ease of use, but it would need overhauling one way or another to put all the citation data into the actual article (and to remove the |noedit= that hides things). I do not at all buy the idea that a DOI by itself "is" the citation content; the human-readable portion is, because WP is written for human reading. Many people have no idea what a DOI even is. Without safesubst, subst'ing would be a terrible, sloppy way to address the problem (since any conditional code would also end up in the article after the subst, e.g. ((#ifeq:((str left|(({1|))}|3))|10 and much, much more. With safesubst this would be viable. Alternatively: I would suggest that bot replacement by a properly-constructed ((cite journal)) is what needs to happen.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  11:38, 12 September 2015 (UTC)[reply]
  • @SMcCandlish: Actually, the code can be rewritten prior to substitution to make use of "safesubst", which will parse the parser functions prior to substitution instead of substituting the parser functions themselves. Over at WP:TFD/H, we substitute complicated templates all the time with great success. ~ Rob^Talk 14:50, 12 September 2015 (UTC)[reply]

If someone wants to do that work. As long as the citation details are in the article where people can work with them (and where they'll be retained if someone ports our article code to another wiki where they've duplicated only our basic citation templates), and it doesn't result in a mess in the article, that would probably resolve the issues with this template. I've clarified my comments above.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  14:56, 12 September 2015 (UTC)[reply]

It's fairly easy to do. The end result would be the insertion of whatever "cite journal" template is currently being used under the wrapper anyway. ~ Rob^Talk 14:59, 12 September 2015 (UTC)[reply]

• Keep it for all the reasons already mentioned. --Saimondo (talk) 19:13, 18 September 2015 (UTC)[reply]

Alternative 1[edit]

Alternative 1: Create a sub-page, not for every DOI, but for every journal (the journal ID is usually part of the DOI, so it should be easy for a bot to know which DOI belongs to which journal). PRO: much less pages to watch; easier to detect vandalism. CON: large pages, might be hard to edit.

• Support. I think this is a reasonable compromise. --Erel Segal (talk) 09:28, 9 September 2015 (UTC)[reply]
• Oppose as not technically feasible. There is no consensus for a bot to complete such a task, and it's unlikely any will be obtained. We would need not only a bot to go through all the mainspace and add a journal parameter, but we'd also need a bot to create fairly complicated template syntax either on a one-time basis (to convert all existing uses) or an ongoing basis. This also worsens the accessibility issues that are a major argument to deprecate. ~ Rob^Talk 11:17, 9 September 2015 (UTC)[reply]
• Could you clarify what you mean by this? What would this sub-page contain, how would it be called, how would users add and edit individual citations? And in light of the comment above stating that vandalism is a purely hypothetical concern (I for one have never seen a vandalized Cite Doi template, and in the running of Citation Bot I have looked at a great many), what will this accomplish? Martin (Smith609 – Talk) 07:02, 10 September 2015 (UTC)[reply]

Alternative 2[edit]

Alternative 2: Do not create sub-pages - insert the citation into the paper body - but leave the template Cite DOI as it is - without deprecation or deletion. PRO: editors will have an option to insert citations with a single parameter, which is arguably easier than inserting citations with many parameters. CON: after the bot replaces the citation, the article looks messy.

• Conditionally support so long as this is implemented in two phases. First, deprecate. Second, seek wider community consensus for such a bot and implement it. It is not reasonable to hold up deprecation regarding current uses while a bot is approved. ~ Rob^Talk 11:19, 9 September 2015 (UTC)[reply]
• Conditionally support, exactly as Rob said.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  11:33, 12 September 2015 (UTC)[reply]
• Conditionally support As long as we are not creating new subpages and putting the data in the article with a bot replacing it with "cite journal". This is what we are doing right now though so not sure how it is different than the proposal above? Doc James (talk · contribs · email) 11:53, 22 September 2015 (UTC)[reply]

Alternative 3[edit]

Alternative 3: Create sub-pages only for DOIs that are used in at least 5 different articles. PRO: these papers are probably sufficiently important to worth their own page. Number of sub-pages will be greatly reduced. CON: someone still needs to monitor these sub-pages.

• Oppose. Doesn't address really any of the deprecate arguments other than proliferation of subtemplates. Accessibility issues remain, as do vandalism issues. ~ Rob^Talk 11:20, 9 September 2015 (UTC)[reply]
• Technical concern. How will one count how many pages a DOI appears in, and who should do this counting? How will this person or script know that it is appropriate to use the Cite Doi template in these pages (they might not match the formatting, or the editors who 'own' the pages may prefer not to use cite doi). And why 5? Martin (Smith609 – Talk) 07:05, 10 September 2015 (UTC)[reply]
• Oppose Do not see this as useful. We want the ref info in the article. Doc James (talk · contribs · email) 11:20, 22 September 2015 (UTC)[reply]

Alternative 4[edit]

Alternative 4: Leave things as they are and concentrate effort on getting this template to work with WikiData as a backend. Meanwhile, add to the documentation that the template should be used only when the number of articles the source will be used in would justify the overhead.

• Support. Based a comment by Daniel Mietchen from below, it seems that Wikidata handling of citations is just around the corner: "On Wikidata, WikiProject Source MetaData deals with bibliographic metadata, there is a collection of bibliographically relevant Wikidata properties, a list of items with DOIs, and with Arbitrary access finally scheduled to come to the last wikis next week, we could start pulling in bibliographic metadata from Wikidata, ideally through some Lua modules (Module:Cite doi is already in the making). As for displaying the full metadata or just an identifier (be it a DOI, ISBN or Wikidata ID) in the Wikipedia source text, perhaps we can build an opt-in Javascript tool that expands the identifier-only variant into the full-metadata variant, feeding changes back to Wikidata.".
• Oppose. Right now, we don't know what citations from Wikidata will even look like. I'm not opposed to a complete conversion to Wikidata should it be accessible to new users (for all references), but that can be looked into when we have a better idea of what that will entail. For now, we don't lose any opportunity to use Wikidate by converting to Cite journal and solving immediate problems. ~ Rob^Talk 11:21, 9 September 2015 (UTC)[reply]
• Support. We should be working to make things more consistent and easier for users. Deprecating these templates leads to inconsistency, more difficult and time consuming maintenance, and more error-prone effort switching to WikiData as a backend. Curly Turkey 🍁 ¡gobble! 23:10, 9 September 2015 (UTC)[reply]
• Neutral: Seems too soon, per BU Rob13's comments. Let's see a test implementation.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  11:34, 12 September 2015 (UTC)[reply]
• Oppose if we decide to move refs to Wikidata having one less type of ref to deal with would be a good thing not a bad thing. If we do decide to do this we can than concentrate on pulling "cite journal" from Wikidata. Doc James (talk · contribs · email) 11:21, 22 September 2015 (UTC)[reply]

Discussion[edit]

• As I said before, this template's construction is unique in how we do citations. No other citation format does it by creating subpages and then calling them. We don't do that for books, for newspapers, for websites, for most journals, for anything but this set of citations. Most of these citations are just wrappers for cite journal anyways. This way of citation (burying the details of the citations into hidden subpages) is not remotely helpful to editors who would have to learn how this thing worked to even begin to add or review citations on a page. It's off-putting as hell and currently there are almost 60k largely unwatched subpages created (more than 10% out of all templatespace) out of approximately 67 million doi in existence. Further, each one of these citations are called maybe once or twice at most. These could all be subject to very sneaky unseen vandalism. The only gain is that some people will very extensive computing and programming knowledge can save a few bytes from articles is not overall conductive to the encyclopedia. Deprecation is adding thousands of bytes text into the pages that frankly should be there in the first place. If that information was cited to books, to webpages, to newspapers, to other journals, even to doi articles that didn't use this system, that text would be in the article. The overall potential gains from this vastly more complicated scheme that only some editors here understand is nice but I'm more concerned that we already have to teach new users how to edit here, that they need to cite things, civility rules but we shouldn't need to teach them that they also much understand complex computer template formatting to figure out how a particular citation is done just because this stuff seems easy to a very few users here. We have already installed cite journal into one of teh template in the browser, there's no need for people to call cite doi which then call a template subpage which then calls cite journal. That level of complexity is entirely unnecessary. -- Ricky81682 (talk) 00:26, 7 September 2015 (UTC)[reply]

@Ashill: I'd ask you to put your comments here so there's a coherent discussion but it's your RFC and I'll respond to your points anyways. First, as to the consensus, it's been there a year and other than DePiep Johnmperry's childish insults at those who disagree, there hasn't been any real disagreement. Insulting those who disagree with your personal ease of use concerns is not conducive to collaboration in my view. Second, ignoring the possibility doesn't mean it's resolved. As I note below, these subpages are being edited but only by a very small handful of people who even know they exist. It is a potential for concern and the point is, this system adds an additional layer of complexity that for little gain. Your third point confirms my earlier concern: this is just a wrapper for cite journal so if there's a need to change cite journal, we're just adding additional work to be done on this wrapper. As to whether ((subst:cite doi 10.1234/56789)), that's fine and cute but that isn't the point. You can still have that, and still have a citation bot fill out the full details (I believe a bare urls bot tries that) but that bot puts the text into the article where it belongs. It doesn't take that text, put it in a subpage (which Special:RecentChangesLinked/Category:Cite_doi_templates people still need to fix) and then goes on its way. The fact that a small number of people can even find these pages is a real problem. Finally, the fact that adding the full citations adds text to the article is precisely what it does, it adds text to articles, text that in my opinion should be there in the first place. The text would be shorter if we pulled every section into separate subtemplates but we've chosen to give up on making it easier for a small subset of editors in exchange for the larger majority of editors to be able to work here. -- Ricky81682 (talk) 01:20, 7 September 2015 (UTC)[reply]

Further, for examples of stupid vandalism, see the histories of Template:Cite doi/10.1136.2Fbmj.d4488 (which is cited nowhere) and Template:Cite doi/10.1021.2Fjf801670p (which is cited at a sandbox). That's just on a quick check of recent changes to the template but again, the vast majority of these citations are to a single page or maybe two pages and we would be better off if the text for these sixty thousand or so pages were incorporated into the articles that are better watched. -- Ricky81682 (talk) 01:46, 7 September 2015 (UTC)[reply]

Supposedly this info is all going to be moved to WikiData. Will that not result in the exact same problem you here describe? Should we propose barring the data being moved there? Curly Turkey 🍁 ¡gobble! 01:48, 7 September 2015 (UTC)[reply]

That's been a proposal for how long? I think there were other proposals starting at meta in March 2014. Is you suggestion that we wait until it's finally agreed upon? And as difficult it could be, it would be possible to reverse this into WikiData which we'll need to do for all the non-doi stuff. -- Ricky81682 (talk) 02:02, 7 September 2015 (UTC)[reply]

I'm saying the "who will watch these?" problem will be exactly the same. What's your opinion on that? Curly Turkey 🍁 ¡gobble! 02:05, 7 September 2015 (UTC)[reply]

If the data is moved to WikiData? I presume WikiData but what does that have to the current set-up? -- Ricky81682 (talk) 02:11, 7 September 2015 (UTC)[reply]

Let's try this again: you've argued that moving the data into templates will create a situation under which the data will not be watched by many and thus open to vandalism. How would that be different with the data at WikiData? Curly Turkey 🍁 ¡gobble! 02:16, 7 September 2015 (UTC)[reply]

I don't know, we aren't moving it to WikiData now so why does this matter? Maybe I would oppose a move then, I don't know. We have a current situation with tens of thousands of unwatched pages and the response has basically been 'it doesn't really matter' which is a viewpoint. Is your point that we should ignore it because we could have another set-up with the same problem? -- Ricky81682 (talk) 02:22, 7 September 2015 (UTC)[reply]

@Ricky81682: Re: "unwatched pages": this applies to ((cite doi)) as the subtemplates are bot-generated. It does not apply to ((cite isbn)), though, as all those subtemplates must be hand-generated and thus are on people's watchlists (like mine). The "consensus" at this page is being used to subst out all the ((cite isbn)) templates even though the "unwatched templates" argument doesn't apply. This is disruptive, and those of us who are affected didn't even get a say in the "consensus": I don't watch ((cite doi)) because I don't use it, but the "consensus" is being applied to ((cite isbn)), which I do (and have watchlisted). Curly Turkey 🍁 ¡gobble! 03:41, 7 September 2015 (UTC)[reply]

I'm not sure why you are arguing based on cite isbn. You there also started a discussion which resulted in a consensus to deprecate so unless your plan is to argue that we should ignore that discussion and ignore the prior RFC, I'm not seeing your point. -- Ricky81682 (talk) 05:21, 7 September 2015 (UTC)[reply]

If you're not sure why I brought up ((cite isbn)) then please read what I wrote again. The problems you bring up simply don't apply to that template. Curly Turkey 🍁 ¡gobble! 05:45, 7 September 2015 (UTC)[reply]

@Curly Turkey: I tend to think this should not be moved to WikiData. They moved interwikis and various other things there, and I and many other editors have no idea where it is, how it works, or how to change it. I shudder to think about all our citation data being buried in some external WMF database where so many of us won't be able to correct it. Unless and until I see viable tools that make citation easier with WikiData not harder, I find the proposal troubling. Citation templates are complex for new users (and for complicated citation cases), but at least they're right there in the article.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  15:51, 12 September 2015 (UTC)[reply]

@Ricky81682: I was explaining the rationale for my !vote; I didn't mean to cut off discussion and sorry if I did. Could you please clarify how I insulted editors who disagree? I just reread my comments and don't see it; I certainly intended no insult.

1. No one has contested the consensus before, but it's only relatively recently that this old discussion has started affecting large numbers of pages and, therefore, editors. And I'm contesting it now (as are others).
2. Yes, it's well known that this is a wrapper for cite journal. The question is whether that's a problem. I've said clearly why I don't think it is. Substituting is exactly what Dexbot is currently doing to thousands and thousands of pages, without discussion involving the editors of nearly all of those pages. If there are any changes that are made to the syntax for using cite journal, such changes would have to be made (presumably by a bot) in every article that calls cite journal; I don't see what's more difficult or more effort to have the bot do the same in cite doi templates.
3. One can't use subst:cite doi xxx if the templates are deleted, as you have been trying to do. ([3] [4]).
4. Thanks for pointing out a couple examples of vandalism (which were corrected within minutes, which to me is strong evidence that even those isolated examples of vandalism aren't harmful).
5. Again, it's fine if you prefer to have the citation text interspersed with the source text. So don't use cite doi! Others don't like the metadata interspersed (and there are multiple citation approaches that don't mix the metadata in with the text, such as harvtxt; it's not just cite doi), and there's no policy requirement to have the metadata interspersed.
6. As I recall, DOI bot was blocked for inputting the metadata into the article text. There was not objection at the time (as I recall) to the bot creating the cite doi subpages, partly because it's a cleaner approach for a bot: much easier to ensure that the code doesn't have unexpected side effects if it's only editing an isolated subpage. —Alex (Ashill | talk | contribs) 02:38, 7 September 2015 (UTC)[reply]

@Ashill and Ricky81682: See my comment in the discussion section about vandalism to these templates. Graham87 10:38, 7 September 2015 (UTC)[reply]

• It seems to me that the biggest failure of a consensus-based decision-making system is when endless calls for consensus come in the way of implementing it. There was a discussion. Consensus was reached. No-one challenged said consensus at the time by asking for another opinion. We're now being asked for consensus on whether or not to implement the previous consensus. Frankly, if this closes as don't deprecate, I'm tempted to open a discussion on whether to implement the consensus not to implement the first consensus to make it obvious how silly this is. ~ Rob^Talk 02:49, 7 September 2015 (UTC)[reply]
  • WP:Consensus can change. One of the (many) reasons it can change is that interested editors may not know about the first discussion. The English Wikipedia is far too big of a project for everyone to know what's happening, or even what's happening that significantly affects their own interests and work. WhatamIdoing (talk) 03:29, 7 September 2015 (UTC)[reply]
    • That doesn't make it any more efficient to continuously debate the same point. In a close discussion, consensus almost always will change if debated often enough, merely by virtue of who shows up. Deference should be given to the first outcome to prevent inconsistencies across time. ~ Rob^Talk 15:43, 7 September 2015 (UTC)[reply]
• I should state that I don't care about the mechanism. If all of the cite doi sub-templates are a problem and there's a better way to store the metadata, that's great (though I thought we aren't supposed to worry about performance issues like that). If it makes sense to create a new cite doi template with no subtemplates but a single parameter, so ((cite doi 10.1234/56789)) becomes ((cite doi | 10.1234/56789)) and the cite doi template finds the metadata in some database that's not stored as a template, that's fine by me (and I assume it would be trivial to have a bot make those changes site-wide), as long as the existing templates aren't taken out in the meantime. The "problem" of unused templates has also been raised; how is that a problem? It's useful if the same source gets cited again. Again, a database may be a more efficient way to store that information than templates, but that's a technical issue that I don't think is within our purview as editors and would take time, per the discussion above. —Alex (Ashill | talk | contribs) 03:00, 7 September 2015 (UTC)[reply]
  • That's how it already works with ((cite isbn))—no bot needed. Curly Turkey 🍁 ¡gobble! 03:23, 7 September 2015 (UTC)[reply]
  • @Ashill: What is your actual concern? Is it the markup being a mess or having that information at all in the articles? You use the term "metadata" to I'm guessing referring to the author and other actual citation information which is a bit odd usage. If you wanted to, every citation could be done with a ref name tag for the doi and then put the full references in under the reflist so that the markup isn't so ugly. If your goal is that all citation information (other than the doi) should be hidden off somewhere else, well, that's not a lot of support for that. If you were to, say, we are now going to put all New York Times articles into a cite NTY|Author|Date template which will then call a subpage somewhere else with the full title, url and other information, that's up to you and you could argue that no one else has to use it but I'm guarantee that'd be taken up for deletion because the ultimate point is that other editors are stuck with your choice. -- Ricky81682 (talk) 05:21, 7 September 2015 (UTC)[reply]
    • I didn't say anything about "metadata" and I didn't say anything about hiding any information. Curly Turkey 🍁 ¡gobble! 05:35, 7 September 2015 (UTC)[reply]
      • My comment was at your level because I was responding to Ashill's comment. -- Ricky81682 (talk) 05:53, 7 September 2015 (UTC)[reply]
• I see a lot of "cite template data is content" arguments, but how does that not apply to navbox templates? Lists arranged by editorial whim are far more content-like than the well-defined data fields of a book citation. Why is there no campaign to obliterate navboxes? Curly Turkey 🍁 ¡gobble! 03:25, 7 September 2015 (UTC)[reply]
  • Probably because you're not watching the right pages. WhatamIdoing (talk) 03:30, 7 September 2015 (UTC)[reply]
    • Does the smiley mean you're kidding, or are people actually trying to deprecate navboxes? Curly Turkey 🍁 ¡gobble! 03:34, 7 September 2015 (UTC)[reply]
• Vandalism or ill-informed edits (especially blanking) to the cite DOI/PMID templates is a real problem that has caused real harm. See the edit histories of Template:Cite pmid/9252594 (which was broken for seven months), Template:Cite pmid/12805553 (which was broken for almost a year), and Template:Cite pmid/24308656 (which was broken for over sixteen months). These instances of vandalism caused problems that I fixed in the articles about Vacanti mouse, Veristrat, TRPN, and Proprioception (where I first found this problem). See my message at the talk page of Dexbot's owner, Ladsgroup. Graham87 10:31, 7 September 2015 (UTC)[reply]

• We should keep in mind that the decision here also affects Template:Cite isbn, which was deprecated after the deprecation of Cite doi. However, many arguments raised here are not relevant to Cite isbn. Particularly:
  • Cite isbn is usually used in several articles, since usually a textbook contains material relevant to several articles.
  • Cite isbn is edited by an editor - not by a bot. Hence, it is in the editor's watch list and the vandalism argument is irrelevant. --Erel Segal (talk) 17:02, 7 September 2015 (UTC)[reply]
    • The argument isn't irrelevant. The point is that the articles themselves are going to be watched by many more eyes than random individual subpages like these. Someone could just as easily create these doi templates by hand (and I think some people are) just like the ISBN ones are (although I'm finding a lot of them are orphaned now) and watch them. People could watch all of the them, that's not the issue. The principle isn't that "hey the bot is creating these things so it's terrible", it's whether or not references should be kept off-article in this way. -- Ricky81682 (talk) 22:10, 7 September 2015 (UTC)[reply]
      • No, the primary argument has been that having subtemplates for "article content" (citation data) goes against the guidelines. I'm finding a lot of them are orphaned now: of course you're finding them orphaned—they're being susbted out! Every ((cite sibn)) I've created is now an orphan! Curly Turkey 🍁 ¡gobble! 22:54, 7 September 2015 (UTC)[reply]
        • And now they're being nominated for deletion. To destroy the evidence? Curly Turkey 🍁 ¡gobble! 00:08, 8 September 2015 (UTC)[reply]
          • I'm not the one orphaning them. But they are orphaned templates. There's a variety of them at this and this page I'm finding from various lists. I'm not sure what your argument is: there's proposals (RFCs, discussions, etc.) to deprecate these and then people (or bots) come by to enforce that but no one has yet to deal with the orphaned templates. Is your argument that no one should be implementing these proposals simply because you find them useful? That all discussions are bunk and we should discuss and discuss until you get what you want? -- Ricky81682 (talk) 00:56, 8 September 2015 (UTC)[reply]
            • I'm saying it's extremely bad form to go around deleting these templates while they are under discussion—if the discussion were closed with "keep" would you volunteer to undelete these templates and undo the substs to all the articles or leave the work to those of us who have already been put out? Curly Turkey 🍁 ¡gobble! 02:30, 8 September 2015 (UTC)[reply]
              • They're orphaned templates. I wasn't the one who orphaned them. They haven't been "under discussion" since 2014 when there was a discussion and someone in the meantime chose to deprecate their use (and that wasn't me). For over a year, the policy was to make orphaned. The doi is easily searchable. Go find the original article and put the template back if you want and then argue keep for it, I don't care. I've listed orphaned ISBN templates, orphaned wdl template, orphaned templates of all types. And for none of those, I was the one who orphaned the citation so I'm not sure what "bad faith" is there in that someone else substitutes a template according to a consensus for over a year ago and then I list the orphaned template for deletion. Should we keep all orphaned citation around in the small possibility that after enough repeated discussions, they'll be support for these template subpages and then they won't be orphaned? Why haven't you ask the bot user to reverse all their edits? I have no idea if this was orphaned due to the bot or because an article created nine months ago or three days ago changed a citation. That's precisely my point, nobody know what's going on with these things. -- Ricky81682 (talk) 04:22, 8 September 2015 (UTC)[reply]
• As a note on process, given that past consensus was to deprecate the template, I assume that unless this is closed as consensus to not deprecate, then the template would still be deprecated. In other words, the burden of showing consensus should be with those seeking to change the previous outcome. A "no consensus" result should be interpreted as "no consensus to stop the deprecation". This probably won't matter, because it appears a consensus to deprecate is forming, but it's worth noting in case the tide changes. ~ Rob^Talk 18:54, 7 September 2015 (UTC)[reply]

Manually deprecation rather than a bot then? If people who support the template fight a bot deprecating and also fight manual attempts at deprecation, that's basically just people refusing to accept consensus. -- Ricky81682 (talk) 22:10, 7 September 2015 (UTC)[reply]

• BU Rob13: that argument may not hold water—issues have been reaised as to how the RfC was conducted, and in each subsequent RfC new voices have appeared opposed to the deprecation. Appealing to the "status quo" may very well be gaming the system by appealing to a "consensus" whose validity is in question. Curly Turkey 🍁 ¡gobble! 22:57, 7 September 2015 (UTC)[reply]
  • If there is question about the validity of the consensus, take it up through WP:CLOSECHALLENGE, something that no editor has done as far as I can tell. We have processes to determine the validity of a closure. If you choose not to exercise them, then it is presumed valid. You cannot unilaterally state the consensus is invalid without asking for review. ~ Rob^Talk 02:48, 8 September 2015 (UTC)[reply]
• By the reasoning that the templates are usually just used once, we also should remove Template:LSR and Template:LPR, which create 1-2 subpages for any software just to store the version number... (see Category:Latest stable software release templates and Category:Latest preview software release templates, about 1500 templates total) I really liked cite doi, because it usually lead to higher quality citations (except for the fact that CiteBot never used "cite conference", but always used "cite journal", unfortunately); in particular if they are shared across multiple articles. I do not see much of a benefit of getting rid of all these articles, but I do see the benefit of better quality. I don't think inlining the citations is the way to go - errors from automatic citations then need to be fixed in multiple places; instead the data should probably be migrated into WikiData and included from there, if templates are really too expensive. --Chire (talk) 09:13, 8 September 2015 (UTC)[reply]

• Articles that are heavily cited do have templates. That's what Category:Specific-source templates is for. As to those templates, there were silly versions of those too in the past. It's all about use. The problem is these aren't used very often (once at most) and if we do deprecate again, I think we should have a bot go by and just count the number of translucations so that the commonly used articles are kept. But that can always be done after the fact since there's almost 60k pages there now. I had to start a second RFC because citation bot was still running around doing its job. -- Ricky81682 (talk) 09:33, 8 September 2015 (UTC)[reply]
• I agree with Chire — I found ((Cite doi)) very useful but think we should now concentrate on migrating citation metadata into some more structured form that could be fed into templates here. On Wikidata, WikiProject Source MetaData deals with bibliographic metadata, there is a collection of bibliographically relevant Wikidata properties, a list of items with DOIs, and with Arbitrary access finally scheduled to come to the last wikis next week, we could start pulling in bibliographic metadata from Wikidata, ideally through some Lua modules (Module:Cite doi is already in the making). As for displaying the full metadata or just an identifier (be it a DOI, ISBN or Wikidata ID) in the Wikipedia source text, perhaps we can build an opt-in Javascript tool that expands the identifier-only variant into the full-metadata variant, feeding changes back to Wikidata. -- Daniel Mietchen (talk) 21:46, 8 September 2015 (UTC)[reply]

• I posted notices of this discussion to WikiProjects for astronomy, physics, biology, chemistry, history of science, science, and mathematics. I notified academic journals but it had already been notified. Blue Rasberry (talk) 14:41, 9 September 2015 (UTC)[reply]
• @SMcCandlish: I'm a bit confused by the comments you've made. You've commented on both Deprecate and Don't Deprecate, which are sections that have been treated as support/oppose, meaning they're usually mutually exclusive. Would you mind making a single comment here clarifying your position to make sure it isn't misinterpreted? Sorry for the extra work, I just can't discern what your position is. ~ Rob^Talk 15:18, 12 September 2015 (UTC)[reply]

Sorry; I've clarified my !votes. Trying encapsulate all the following: a) the idea I support requires work; b) if it's not done, then I revert back to supporting deprecation; and c) one thing should definitely be deprecated (|noedit=); d) one of the other ideas also has its merits but is not mutually exclusive; and e) there's a bot-ish possible approach, too. I'm now favoring: safesubst + required substitution − noedit = viable.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  15:46, 12 September 2015 (UTC)[reply]

• It's about time this discussion is closed by an uninvolved editor. It's been open for a week and consensus is extremely clear and in favor of deprecation. From a simple vote count, we have 19 for and 7 against deprecation and removal (counting SMcCandlish as against). This is a discussion, of course, but a 5:2 ratio in favor of something is nearly impossible to overcome unless the arguments of the majority are against policy or extremely trivial, which is not the case here. Keeping this RfC open longer just further delays the deprecation of Cite DOI, something that has been pending for a year. It's time to get this over with. ~ Rob^Talk 01:49, 16 September 2015 (UTC)[reply]
  • You could try wp:Requests for closure, but we may still have to wait the full month.LeadSongDog come howl! 15:50, 17 September 2015 (UTC)[reply]
    • It's a bit premature for a formal request for closure, I think. We'll probably have to wait out the month, which is annoying. That leaves just enough time to start another RfC about whether there's really really really consensus to deprecate! ~ Rob^Talk 18:12, 17 September 2015 (UTC)[reply]
      • I would say its a bit early too close. I just had a message from the feedback request service asking me to participate. We should not close while people are still being actively invited.--Salix alba (talk): 07:11, 24 September 2015 (UTC)[reply]
• I've not seen Wikipedia:TWL/Citoid mentioned much in this discussion. Visual Editor uses this to generate citations from DOI's on the fly. You can access this server from outside VE as well, I've a script User:Salix alba/Citoid, which does just this. I'd like to see work to improve the editing tools to use the Citoid server. --Salix alba (talk): 07:11, 24 September 2015 (UTC)[reply]

Alternatives[edit]

I believe that conflicts in Wikipedia should be resolved by "consensus", which means that we should not just decide between two opposing alternatives, but rather make our best efforts to find a new alternative which will be accepted by everyone. I would like to suggest some alternatives which relate some of the issues raised above. --Erel Segal (talk) 08:52, 8 September 2015 (UTC)[reply]

Alternative 1: Create a sub-page, not for every DOI, but for every journal (the journal ID is usually part of the DOI, so it should be easy for a bot to know which DOI belongs to which journal). PRO: much less pages to watch; easier to detect vandalism. CON: large pages, might be hard to edit.

Alternative 2: Do not create sub-pages - insert the citation into the paper body - but leave the template Cite DOI as it is - without deprecation or deletion. PRO: editors will have an option to insert citations with a single parameter, which is arguably easier than inserting citations with many parameters. CON: after the bot replaces the citation, the article looks messy.

Alternative 3: Create sub-pages only for DOIs that are used in at least 5 different articles. PRO: these papers are probably sufficiently important to worth their own page. Number of sub-pages will be greatly reduced. CON: someone still needs to monitor these sub-pages.

Other alternatives are welcome.

Alternative 4: Leave things as they are and concentrate effort on getting this template to work with WikiData as a backend. Meanwhile, add to the documentation that the template should be used only when the number of articles the source will be used in would justify the overhead. Curly Turkey 🍁 ¡gobble! 09:11, 8 September 2015 (UTC)[reply]

What's the difference between Alternative 4 and "Don't deprecate" other than you have a theoretical (if it happens) endpoint? The first discussions on this at Meta were in I think March 2014. Alternative 2 is basically what's being argued for here under the deprecate option, few people have supported fully deleting the template, just how it functions. The better way is just to create a new template that does that and then once this thing is cleared up, we can redirect it back and fix it over time. This isn't my RFC but you'd be better off putting these into separate subheadings at the top of the page and asking for people to strike and move their votes. However, just randomly spouting out ideas without actually supporting one looks chaotic and like you're just grasping at straws here. -- Ricky81682 (talk) 09:27, 8 September 2015 (UTC)[reply]

Uh ... I've already !voted "don't deprecate" ...? Every proposed solution is theoretical until it's implemented. Shooting down ideas because they're "theoretical" is the opposite of helpful. And year after proposal and it hasn't been implemented yet? God! the sky'll fall before it gets implemented! Better obliterate those 60k citations now! Curly Turkey 🍁 ¡gobble! 11:50, 8 September 2015 (UTC)[reply]

@Curly Turkey:, there is no serious proposal to "obliterate" citations, which would obviously break wp:V even worse than cite doi and its ilk. The intent of deprecation is to change those citations from scattered, hidden, unmaintainable, and incomplete implementations in template space to local, visible, maintainable, and complete implementation in article space until such time as a better approach is available. I can't understand why anyone still thinks this is even a matter for debate. The templates cite doi and cite pmid are utterly dependent on user:citation bot which has been unmaintained for a year and has been broken even longer. LeadSongDog come howl! 15:01, 8 September 2015 (UTC)[reply]

because citation bot could come back so the best thing to do is be preventative and keep the status quo until our bot returns. 166.176.59.181 (talk) 11:34, 9 September 2015 (UTC)[reply]

@166.176, would you please either register an account or, failing that, indicate at which of these many IPs you can be reached? Hopping IP makes it very difficult for people to use talk and other facilities in conversation with you. You also miss out when someone replies to you at an address you no longer use. LeadSongDog come howl! 16:34, 9 September 2015 (UTC)[reply]

@LeadSongDog. WP:HUMAN. 166.176.59.153 (talk) 17:52, 9 September 2015 (UTC)[reply]

Yes, that's hardly news, but we can't help you if we can't talk to you. wp:WHYREGISTER LeadSongDog come howl! 18:40, 9 September 2015 (UTC)[reply]

WP:NOTHUMAN Staszek Lem (talk) 16:58, 28 September 2015 (UTC)[reply]

FWIW, a new team are now working to restore Citation Bot, which should soon be operational again. Martin (Smith609 – Talk) 18:11, 8 September 2015 (UTC)[reply]

Regarding the Wikidata option, the reason I downplayed that that in my previous closure was that there apparently hadn't been any discussion (at least none I could see) that demonstrated this was even something the community wanted on a larger scale. This question of whether it's even desirable to shove all this off into WikiData probably needs its own RfC type discussion at some point, before it's presented as an option to solve the problem at hand. Gigs (talk) 22:08, 9 September 2015 (UTC)[reply]

@Jo-Jo Eumerus: It's my understanding that refToolbar already has that functionality with the positive effect of not needing anything temporary. Someone mentioned here that you can plug a DOI into refToolbar and it will populate the other fields for you automatically. What you're proposing is essentially a new template and a new bot. It would need approval from the community, and it goes beyond the scope of this discussion I think. This is more about what to do with the current state of Cite DOI (i.e. should we deprecate and remove all these templates-within-templates). ~ Rob^Talk 16:52, 10 September 2015 (UTC)[reply]

Ah, indeed it does. Such a thing could help with deprecating the current way the template is used, but you are right, this task would require a separate approval/discussion yet.Jo-Jo Eumerus (talk, contributions) 17:11, 10 September 2015 (UTC)[reply]

Since a bot created these, a bot can create them on WikiData. There is no need to keep this here while awaiting a decision, since WP:NOT Wikipedia is not a database for DOIs (or PMIDs, or anythign else) They should be deleted forwith. Alternatives 1/3/4 all fail WP:NOT and are not proper alternatives as they already contravene What Wikipedia is NOT. Wikipedia is not WikiData, and we should not be a holding bin for Wikidata. We do not hold definitions, they are frequently deleted at XfDs regardless of what happens on Wiktionary. -- 70.51.202.113 (talk) 04:08, 13 September 2015 (UTC)[reply]

Excellent point, but I would go even further. Why should WikiData replicate data already found in external databases such as AdsAbs, arXiv, CrossRef, and PubMed? The Citation bot queries these databases to automatically create new citations as needed. These external databases maintain this data better than WikiData could ever hope to. WikiData should concentrate on unique data not contained in these external databases such as translation of titles (e.g., |trans-title=, |trans-chapter=) into the native language for each of the official Wikipedias and corrections of any errors in the external databases. (It should also be noted that PubMed corrects errors if they are pointed out to them.) While the RefToolbar can create citations using DOIs or ISBNs by scrapping data from web pages, the citations frequently contain errors that must be corrected by humans. Human edited citations in their entirety should be imported into WikiData. Boghog (talk) 08:55, 13 September 2015 (UTC)[reply]

The WP:NOT point doesn't apply, since it applies to content. We're not creating the list of citations as a repository for readers; we've created them as a tool for editors of data that needs to be in the encyclopedia anyway (whether it's inline in the articles or farmed off to the cite doi templates). Every reference on Wikipedia is content that exists elsewhere; that's the point of a reference. —Alex (Ashill | talk | contribs) 00:31, 16 September 2015 (UTC)[reply]

NOT certainly applies, because otherwise you could create your own social network outside of articlespace, which is something you should not be doing. Similarly, this NOT applies, because Wikipedia is not a hosting site for any use if it isn't in articlespace. We are creating a repository, because we never end up deleting references that are unused, and we replicate citations across different UIDs like PubMed and DOI can end up with duplicate citaitions. So it's all a WP:NOT failure. The point of a references is to be used on an article, and people seem to object to deleting unsued ones, so, it is clearly a compiliation for creating a DOI database, instead of actually using it for references. DOIs are not the only type of UIDs so it shouldn't even be housed in a DOI system (there being others such as PubMed, bibcodes, LOC, ISBN, etc) as it promotes duplication of content (already evident with duplicate citations in differen systems). Encyclopedic content is stored in the articles themselves, so the references should not be in this DOI system, since they should be directly placed in the articles, instead of hidden away and unreadable in source mode when people actually try to edit articles, thereby having some random number without text explaining which reference was here, without needing to flip back and forth between edit view and rendered view. These templates are detrimental to examining articles on an article by article case. -- 70.51.202.113 (talk) 03:43, 20 September 2015 (UTC)[reply]

NOT applies only to those not being used in the encyclopedia, I'd think. ~ Rob^Talk 04:44, 20 September 2015 (UTC)[reply]

Systematically building this makes it creation of a database for which it applies to all, since they could just be substituted directly into the articles instead as almost all of them are single and very low use templates. Per WP:TFD precedents, such low usage is a reason for deletion. So, combining TFD-precedent and NOT, this is still a database system that should not be held on Wikipedia. And indeed, these are like WP:T3 CSD deletable templates, being hardcoded instances of a generic form. Those that are used should be substituted into the articles that use them and then deleted. Those that are not used should be simply deleted. -- 70.51.202.113 (talk) 04:48, 21 September 2015 (UTC)[reply]

otherwise you could create your own social network outside of articlespace: so first these tmeplates are all article content and that's why they should be deleted, and now they're a social network? If NOT applies, it ain't because of your very confused reading of it. Curly Turkey 🍁 ¡gobble! 08:47, 20 September 2015 (UTC)[reply]

The statement was that NOT does not apply outside of articlespace. My example is something that shows that NOT applies outside of articlespace. I never said that DOIs are socialnetworks. NOT applies outside of articlespace. -- 70.51.202.113 (talk) 04:48, 21 September 2015 (UTC)[reply]