Community Wishlist Survey 2016/Categories/Citations

6 proposals, 94 contributors, 110 support votes

Label erroneous Harv-style references

  • Problem: This may apply only to English Wikipedia, but linked Harvard-style short references (e.g. "Martel 1998") without corresponding full citations are not flagged as errors. A Javascript, en:User:Ucucha/HarvErrors.js, helps readers find these errors, but without it, the errors exist silently, with no error message or tracking category.
  • Who would benefit: Readers trying to verify sources, Good Article assessors trying to make sure that sources are present, gnomes who fix pages.
  • Proposed solution: Implement a way to flag these errors in Wikimedia. Maybe turn the script into a gadget?
  • Phabricator tickets:

Community discussion

Hi Jonesey95 - just a heads-up, I've moved your proposal to this Citations page. -- DannyH (WMF) (talk) 01:48, 23 November 2016 (UTC)[reply]

  • "Erroneous Harv-style references" can cover a multidie of errors, whereas short cites (of any kind) without a full citation is one kind of error. What other kinds of errors are contemplated here? ~ J. Johnson (talk) 23:11, 7 December 2016 (UTC)[reply]

Voting – Label erroneous Harv-style references

  1.   Support as a gadget, at least. Stevie is the man! TalkWork 03:21, 28 November 2016 (UTC)[reply]
  2.   Support This script has been really helpful and I would support integrating it into the normal CS1 error reporting or, secondarily, adding it to the gadget list czar 00:49, 2 December 2016 (UTC)[reply]
  3.   Support Libcub (talk) 02:18, 2 December 2016 (UTC)[reply]
  4.   Support--Kiril Simeonovski (talk) 13:58, 2 December 2016 (UTC)[reply]
  5.   Support Emir of Wikipedia (talk) 15:13, 2 December 2016 (UTC)[reply]
  6.   Support -- SBaker43 (talk) 23:43, 2 December 2016 (UTC)[reply]
  7.   Support Let's keep the inline citation style consistent. -- Kndimov (talk) 00:12, 3 December 2016 (UTC)[reply]
  8.   Support --Ranjithsiji (talk) 03:28, 4 December 2016 (UTC)[reply]
  9.   Support per User:Czar --Hedwig in Washington (talk) 02:52, 6 December 2016 (UTC)[reply]
  10.   Support- Useful. DPdH (talk) 19:36, 6 December 2016 (UTC)[reply]
  11.   Support Rdrozd (talk) 22:25, 6 December 2016 (UTC)[reply]
  12.   Support --Elmidae (talk) 15:11, 7 December 2016 (UTC)[reply]
  13.   Oppose, until it is clarified whether the proposed labelling is for just this one condition of incompleteness, or would cover various other errors. ~ J. Johnson (talk)
  14.   Support I have come across this issue a few times and have sometimes been stumped because the editor who placed the citation has retired and there is nothing that might relate to the citation elsewhere in the footnotes (like Martel, Pierre-Albin (1999)... in the present case). Flagging these as an error may help editors who wander onto the page, know nothing of en:User:Ucucha/HarvErrors.js, and recognise the citation to create/amend the bibliographical entry as appropriate Iadmc (talk) 04:57, 8 December 2016 (UTC)[reply]
  15.   Support Miniapolis 20:53, 8 December 2016 (UTC)[reply]

Add button for inserting blank citation templates in wikitext editor

note: WTE == wikitext editor

  • Problem: flawed functionality for insertion of citation templates in current WTE.
  1. Copy/paste of citation templates is annoying and feels easily automated. Automating insertion of citation templates should make cite creation more accessible (i.e., more folks will add citations) and less error-prone (i.e., creators will make fewer errors).
  2. The currently-provided functionality (popup dialog) is inappropriate for WTE users, esp those who mostly use external editors when composing, and who use WTE for its specialized value-adds (notably preview, diff, and the all-important save). GUIs are for Visual Editor.
  • Who would benefit:
  1. Everyone who adds citations by WTE.
  2. Everyone who reads articles, since (c.p.) more/better citations tends to improve article quality.
  • Proposed solution: add to current and future WTEs a menu (à la Insert>Citation template>{book, journal,news, web, etc}) that allows one to insert the empty wikitext for the chosen template. Note this proposal differs from functionality provided by the current WTE menu=Cite, which pops-up a GUI for filling in only a part of the chosen template. (And which therefore seems to me more appropriate for the Visual Editor--YMMV.) I want automated insertion of the empty template wikitext into the editing textarea, so that I can fill-in the template (and do the rest of my editing) in an external editor: see section=Usecase following.
  • For candidate common templates, see (e.g.) column=Common usage in the body table under Citation template examples.
  • Regarding WTE versions, see discussion by User:Neil P. Quinn-WMF below.
    Usecase: FWIW, I'm not only the kind of {dinosaur, poweruser} who uses the WTE, I also do the actual editing off-line, and just copy/paste into the WTE for preview, diff, and save ... except that I will use the WTE GUI for one sort of task: inserting things for which I don't remember the syntax. This I do currently almost entirely for redirects and the very-occasional table, but would appreciate GUI automation for the following currently-manual usecase:
  1. In wikitext for article I'm editing (in separate buffer), create or reuse empty <ref> tags. I don't use the GUI button=Reference for this, because I can remember that :-) and because the important thing is usually to find if someone else is already citing that datasource, and if so to reuse their ref name.
  2. Search Wikipedia for the type of citation matching the datasource to which I want to make reference.
  3. Browse to the chosen page (e.g., this).
  4. Browse that page, looking for something like Full parameter set in vertical format.
  5. Copy that template text, paste between my <ref> tags.
  6. Fill-in template, continuing with usual {edit, preview, save} loop (copying to WTE as needed).
Instead, I'd prefer to
  1. In wikitext for article I'm editing (in separate buffer), create or reuse empty <ref> tags.
  2. Copy my editor buffer's contents to WTE, move cursor between my <ref> tags.
  3. Choose Insert > Citation template > whatever from the proposed WTE GUI.
  4. Copy WTE editing area (C-a C-insert) back to my editor's buffer.
  5. Fill-in template, continuing with usual {edit, preview, save} loop (copying to WTE as needed).
  • Phabricator tickets: not at this time, lemme know if that needs done.

Community discussion

  • enWikipedia has a gadget whose name I don't remember that can do this for DOI and ISBN. Citoid can do so as well. However, only a few identifiers can be processed this way. Jo-Jo Eumerus (talk, contributions) 15:27, 11 November 2016 (UTC)
    • Well, for all Medicine-related articles PubMed would be useful. And this could also be extended to other data sources that have unique identifiers and a clear data structure. These are quite a lot. There exist some reference management softwares that are capable of retrieving citations from a wide variety of databases. --Furfur (talk) 17:16, 11 November 2016 (UTC)
      • @Furfur:, @Tlroche:, @Jo-Jo Eumerus:, PMCs and PMIDs are both already supported by citoid, as are URL and DOI. ISBN is not currently supported, although probably will be within the next few weeks or so. In wikitext editor in en wiki, go Cite -> Template -> Cite journal -> and there's a little magnifying glass you can use to fill the citation based on a supplied identifier (PMID/PMC, DOI, URL). It's not available in other languages in wikitext editors, but universal support is planned for the new wikitext editor, here is the task for it: phabricator:T130400 Mvolz (WMF) (talk) 20:45, 28 November 2016 (UTC)[reply]
        • @Mvolz (WMF): thank you, that looks interesting. I did not find this functionality in the German language wikipedia. I tried it in en.WP with an example PMID but it apparently does to retrieve the full citation (e. g. the authors’ names, title, year, journal, volume, number, etc.), although the PMID is a clear and unambiguous identifier. This tool only inserts the PMID number with a link to PubMed. --Furfur (talk) 21:07, 28 November 2016 (UTC)[reply]
  • Aye, having a much broader system would be very desirable. Typing out the citation info is among the most boring parts of article writing. Jo-Jo Eumerus (talk, contributions) 09:24, 12 November 2016 (UTC)
  • Is the problem here then just that to add Citoid support to the wikitext editor, T130400, or something else ? --Snaevar (talk) 23:48, 12 November 2016 (UTC)[reply]
  • On the English Wikipedia, from the editor menu, we can already fill in forms for a few of the cite templates. If this could at least be expanded to include other cite templates, that would be great. Stevie is the man! TalkWork 16:26, 13 November 2016 (UTC)[reply]
    • @Stevietheman: Thanks for pointing out the existing functionality, which I don't use for reasons now edited into this wishlist item. Yes, I would definitely like more templates exposed via menu; no, I don't want more GUI forms--I just want to insert wikitext into the editing area, for editing externally (e.g., in God's Own Editor :-) Tlroche (talk) 20:15, 17 November 2016 (UTC)[reply]
  • Isn't this already planned for the 2017 wikitext editor? {{Nihiltres |talk |edits}} 18:43, 13 November 2016 (UTC)[reply]

Hi Tlroche - just a heads-up, I've moved your proposal to this Citations page. -- DannyH (WMF) (talk) 01:46, 23 November 2016 (UTC)[reply]

@Neil P. Quinn-WMF and Nihiltres:, I'm a bit confused by the above discussion. Can anyone definitively state this proposal (inserting the wikitext blob of a citation template with empty fields) will be in the 2017 wikitext editor? Is it a plan or a definite will-do? If a definite will-do, I see no need for the WMF to take this on per this survey. Stevie is the man! TalkWork 15:35, 29 November 2016 (UTC)[reply]

  • @Stevietheman: sorry for the confusion—some of it is my fault for misunderstanding the original proposal! I'm not 100% certain of the team's plans for the 2017 wikitext editor (I work with them but I'm not on that team itself), but I'm pretty confident of the following:
    • the 2017 editor will include a dialog box allowing the user to fill in text boxes with citation information (with the help of Citoid if desired) and then insert the finished wikitext into the page. This will be directly parallel to the feature in the visual editor (though for fundamental reasons it will not be possible to edit the generated citation as you can do in the visual editor).
    • the 2017 editor will not include the precise feature described in this proposal, which would allow the user to insert the unfinished wikitext for a citation into the page, with blank fields to be filled in later.
So, this very specific feature won't be available in the 2017 wikitext editor, and already isn't in the existing wikitext editor. In that sense, it's a reasonable wishlist request.—Neil P. Quinn-WMF (talk) 20:02, 29 November 2016 (UTC)[reply]
Ideally, the string emitted when pressing that button should be user-configurable (in Preferences), so it could be used for other repeating stuff as well. As such this proposal is somewhat similar to another proposal "Better support for commonly used sources" listed further above. --Matthiaspaul (talk) 23:57, 29 November 2016 (UTC)[reply]
  •   Done? This already exists by clicking {{источник}} at the bottom of, say, [1].
Источники: * {{источник|url=|Название=|Автор=|Издатель=|Дата=}} {{источник?}} {{нет источников}} {{нет источников в разделе}} {{VOA|url=|title=|author=}} {{ЧК|}} {{kremlin|url=|title=}} {{IIP Digital|url=|title=|author=}} {{GlobalVoices|url=|title=|author=|original=}}
Clicking any of these templates inserts them into the article body at the cursor position. --Gryllida 23:34, 1 December 2016 (UTC)[reply]

Voting – Add button for inserting blank citation templates in wikitext editor

  1.   Support--Strainu (talk) 10:12, 28 November 2016 (UTC)[reply]
  2.   Support -- Ariconte (talk) 19:20, 28 November 2016 (UTC)[reply]
  3.   Support per my somewhat more general suggestion above --Matthiaspaul (talk) 23:57, 29 November 2016 (UTC)[reply]
  4.   Support this general idea per clarifications above. The use case makes sense. Stevie is the man! TalkWork 17:28, 30 November 2016 (UTC)[reply]
  5.   Support -- Tournasol7 (talk) 18:18, 1 December 2016 (UTC)[reply]
  6.   SupportGryllida 23:34, 1 December 2016 (UTC)[reply]
  7.   Support -- CFynn (talk) 05:26, 2 December 2016 (UTC)[reply]
  8.   Support--Kiril Simeonovski (talk) 13:58, 2 December 2016 (UTC)[reply]
  9.   Support optional comment MartinThoma (talk) 14:50, 2 December 2016 (UTC)[reply]
  10.   Support I love this idea. Would make my life easier Hires an editor (talk) 01:54, 3 December 2016 (UTC)[reply]
  11.   Support Please do this. Sesamehoneytart 04:01, 3 December 2016 (UTC)
  12.   Support --Ranjithsiji (talk) 03:29, 4 December 2016 (UTC)[reply]
  13.   Support --By erdo can • TLK 17:00, 4 December 2016 (UTC)[reply]
  14.   Support This will help to improve the rest of the articles. Yes, it's good to improve any article, especially the worst cases in current seasons. - Nino Marakot (School time here, in computer lab) 07:45, 6 December 2016 (UTC)[reply]
  15.   Support DPdH (talk) 19:39, 6 December 2016 (UTC)[reply]
  16.   Support Miniapolis 20:55, 8 December 2016 (UTC)[reply]
  17.   Support This is relatively easy to implement, and the payout is comparatively large. As an editor who primarily uses the wikitext editor, I've also wished for such a (default) tool many times in the past. --Waldir (talk) 10:45, 10 December 2016 (UTC)[reply]
  18.   Support - Bcharles (talk) 15:35, 10 December 2016 (UTC)[reply]
  19.   Oppose There's already a Gadget for this.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  18:02, 11 December 2016 (UTC)[reply]
  20.   Support--Mikheil Talk 20:58, 12 December 2016 (UTC)[reply]

Automatic import and formatting of literature references from selected database sources

  • Problem: Literature references have to be formatted manually, even if they are accessible through unique identifiers (e.g. PubMed ID, DOI). This is time-consuming, leads to errors and a diversity of styles that is sometimes not helpful.
  • Who would benefit: Editors (and readers)
  • Proposed solution: Add support in such a way that the contributor simply has to supply the literature identifier for the article to be cited (e.g. the PubMed ID or DOI) and the software (not only VE but also the source editor) autocompletes and formats properly the literature reference.
Example: typing the PMID number "27765289" should insert a reference like this:
<ref>Masukume G, Kipersztok L, Das D, Shafee TM, Laurent MR, Heilman JM: Medical journals and Wikipedia: a global health matter. Lancet Glob Health. 2016 Nov;4(11):e791. DOI:10.1016/S2214-109X(16)30254-6doi:10.1016/S2214-109X(16)30254-6. PMID: 27765289. </ref>

Community discussion

enWikipedia has a gadget whose name I don't remember that can do this for DOI and ISBN. Citoid can do so as well. However, only a few identifiers can be processed this way. Jo-Jo Eumerus (talk, contributions) 15:27, 11 November 2016 (UTC)[reply]
Well, for all Medicine-related articles PubMed would be useful. And this could also be extended to other data sources that have unique identifiers and a clear data structure. These are quite a lot. There exist some reference management softwares that are capable of retrieving citations from a wide variety of databases. --Furfur (talk) 17:16, 11 November 2016 (UTC)[reply]
Aye, having a much broader system would be very desirable. Typing out the citation info is among the most boring parts of article writing. Jo-Jo Eumerus (talk, contributions) 09:24, 12 November 2016 (UTC)[reply]
A half measure might be to have a template (e.g. {{find-cite |PMID= |DOI= | ISBN= }}) that would flag a bot to fill in the details. Bcharles (talk) 22:24, 15 November 2016 (UTC)[reply]
enWikipedia does actually support it through their visual editor; the "Cite" tool has a tab for automatic generation of citations from URL, DOI or PMID identifiers. The output uses a combination of <ref /> tags and the "Citation Style 1" family of templates (e.g. Cite journal, Cite book, etc.).
For example, submitting the PMID of '27765289' given above autopopulates the following citation in the "Cite journal" template:
<ref>Masukume, Gwinyai; Kipersztok, Lisa; Das, Diptanshu; Shafee, Thomas M. A.; Laurent, Michaël R.; Heilman, James M. (2016-11-01). "Medical journals and Wikipedia: a global health matter". The Lancet. Global Health 4 (11): e791. ISSN 2214-109X. PMID 27765289. doi:10.1016/S2214-109X(16)30254-6. </ref>
So it may just be a matter of incorporating this existing function into the wikitext Edit toolbar. — Sasuke Sarutobi (talk) 18:25, 1 December 2016 (UTC)[reply]

Hi Furfur - just a heads-up, I've moved your proposal to this Citations page. -- DannyH (WMF) (talk) 01:46, 23 November 2016 (UTC)[reply]
Hello DannyH (WMF), ok, thank you! I still think this would be useful ... --Furfur (talk) 02:03, 23 November 2016 (UTC)[reply]

Voting – Automatic import and formatting of literature references

  1.   Support--Wesalius (talk) 06:30, 28 November 2016 (UTC)[reply]
  2.   Support Léna (talk) 11:09, 28 November 2016 (UTC)[reply]
  3.   Support--Alexmar983 (talk) 17:12, 28 November 2016 (UTC)[reply]
  4.   Support JAn Dudík (talk) 21:35, 28 November 2016 (UTC)[reply]
  5.   Support --Izno (talk) 00:00, 29 November 2016 (UTC)[reply]
  6.   Support Stevie is the man! TalkWork 15:09, 29 November 2016 (UTC)[reply]
  7.   Support, seems like something which can be implemented to good use via Citoid.--Telaneo (User talk page) 21:30, 29 November 2016 (UTC)[reply]
  8.   Comment This is possible to realize technically, but the Wikipedia guideline w:WP:CITESTYLE states "Wikipedia does not have a single house style, though citations within any given article should follow a consistent style.". There would be necessary to change the guideline first. It is improbable that it can be resolved within a year. I think, it is against the Wikipedia policy to create the tool, that will contravene wikipedia guideline. Therefore I formally request to withdraw this proposal from Voting. --Snek01 (talk) 21:37, 29 November 2016 (UTC)[reply]
    Since each editor is responsible for their own edits, this means they also must abide by consistent citation styles in an article, no matter which tool is available to them. So, I disagree with this proposal being withdrawn. Stevie is the man! TalkWork 17:31, 30 November 2016 (UTC)[reply]
    The use of this tool should not be a must. It should be a useful tool for the contributors. But I agree that some consensus has to be found for the citation format of various data sources. However, these kind of – if you like, "enforced" – consensuses already exist in the form of various citation templates like {{Cite news}}, {{Cite journal}}, {{Cite web}}, etc., and these templates are widely used. The discussion about the format must in my view not be concluded before the introduction of such a tool. Since it would be a centrally defined format it could be changed at any time, if desired, after suitable discussion. I also would not find it helpful if every single data source would have a different citation format and we would have to discuss this. --Furfur (talk) 21:48, 8 December 2016 (UTC)[reply]
  9.   Support Alexei Kopylov (talk) 08:44, 30 November 2016 (UTC)[reply]
  10.   Support Connor Behan (talk) 18:37, 1 December 2016 (UTC)[reply]
  11.   Support Jsamwrites (talk) 18:43, 1 December 2016 (UTC)[reply]
  12.   Support Such a tool would become more useful every single day. DGtal (talk) 18:54, 1 December 2016 (UTC)[reply]
  13.   Comment See en:Template:Cite doi and en:Template:Cite pmid, both deprecated. —Patrug (talk) 21:20, 1 December 2016 (UTC)[reply]
  14.   Support Miyagawa (talk) 22:41, 1 December 2016 (UTC)[reply]
  15.   Support Citoid already handles this and I wrote a script that turns the bare DOI URL into list-defined refs. I think of this suggestion as just supporting what Citoid is already doing (expands bare URLs using the Zotero translator collection). czar 00:51, 2 December 2016 (UTC)[reply]
  16.   Support Libcub (talk) 02:21, 2 December 2016 (UTC)[reply]
  17.   Support --Continua Evoluzione (talk) 13:52, 2 December 2016 (UTC)[reply]
  18.   Support--Kiril Simeonovski (talk) 13:59, 2 December 2016 (UTC)[reply]
  19.   Support Emir of Wikipedia (talk) 15:13, 2 December 2016 (UTC)[reply]
  20.   Support great idea Oaktree b (talk) 16:25, 2 December 2016 (UTC)[reply]
  21.   Support Agree --Athikhun.suw (talk) 06:53, 3 December 2016 (UTC)[reply]
  22.   Support --Ranjithsiji (talk) 03:29, 4 December 2016 (UTC)[reply]
  23.   Support Pamputt (talk) 11:11, 4 December 2016 (UTC)[reply]
  24.   Support --By erdo can • TLK 17:00, 4 December 2016 (UTC)[reply]
  25.   Support --Hugo (talk) 18:47, 5 December 2016 (UTC)[reply]
  26.   Neutral Like other editors, I see this as something that would be very helpful, but I am having trouble seeing how this process could take into account how the citation format may vary from page to page. --Tryptofish (talk) 22:11, 5 December 2016 (UTC)[reply]
    Naturally this could only work in cases where there is a clearly defined data structure (e.g., articles in a scientific journal or (possibly) in a newspaper, items in a database, etc.). However, this should be the case in many data collections as evident from reference management softwares like EndNote, Reference Manager, Bookends (my favourite), or many others. These can retrieve citations from a large variety of sources and the user can easily add new sources by defining their data structure. --Furfur (talk) 21:48, 8 December 2016 (UTC)[reply]
  27.   Support --Zoeperkoe (talk) 08:13, 6 December 2016 (UTC)[reply]
  28.   Support I still go to off-wiki tools for citation formatting. It would be nice to pull this on-wiki. Blue Rasberry (talk) 18:40, 6 December 2016 (UTC)[reply]
  29.   Support- Useful, if can support different citation styles. DPdH (talk) 19:41, 6 December 2016 (UTC)[reply]
  30.   Support--Afernand74 (talk) 08:21, 7 December 2016 (UTC)[reply]
  31.   Support --Geolina163 (talk) 08:50, 7 December 2016 (UTC)[reply]
  32.   Support H. sapiens (talk) 10:15, 7 December 2016 (UTC)[reply]
  33.   Support --sigbert (talk) 12:28, 7 December 2016 (UTC)[reply]
  34.   Support --Elmidae (talk) 15:08, 7 December 2016 (UTC)[reply]
  35.   SupportNickK (talk) 18:26, 7 December 2016 (UTC)[reply]
  36.   Support — --Ambre Troizat (talk) 10:58, 8 December 2016 (UTC)[reply]
  37.   Support Miniapolis 20:57, 8 December 2016 (UTC)[reply]
  38.   Support --Furfur (talk) 21:08, 8 December 2016 (UTC)[reply]
  39.   Support Using the Citoid service directly from the wikitext editor toolbar (e.g. selecting a citation identifier in the textarea, clicking a button in the toolbar and having the wikitext for the citation automatically added -- no dialogs) would be a massive improvement in the editing process. --Waldir (talk) 10:50, 10 December 2016 (UTC)[reply]
  40.   Support --OrsolyaVirág (talk) 11:42, 10 December 2016 (UTC)[reply]
  41.   Support - Bcharles (talk) 15:38, 10 December 2016 (UTC)[reply]
  42.   Support Anything to reduce the tedium of adding cites will lead to better sourcing.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  17:58, 11 December 2016 (UTC)[reply]
  43.   Support Doug Weller (talk) 18:54, 11 December 2016 (UTC)[reply]
  44.   Support--Mikheil Talk 20:58, 12 December 2016 (UTC)[reply]

Better support for commonly used sources

  • Problem: I have a set of commonly used sources (some books) and it would be useful and much less cumbersome if I could insert them into new articles without the rigmarole of reloading, editing and copying from another wiki page (keeping in mind that I may edit from multiple computers and therefore can't store them outside of wiki)
  • Who would benefit: Editors
  • Proposed solution: Make it easier for editors to store and withdraw sources from a set they already have
  • Phabricator tickets:

Community discussion

  • To maximize the utility value for all editors, perhaps this could be implemented in the form of "editor macros" as follows: Each user could define a number of macros in Preferences where each macro definition would consist of a user-definable name and a replacement string. When a user presses "Send" or "Preview" in the edit window, the Wiki software would scan the edited wiki text for macro names and if found replace them by the replacement string. If the user would define a citation as f.e. @MYCITE1@ this could be used to implant citations into the text, but it could be also used for all kinds of other repeating strings (and therefore it would be more flexible than a specific solution for citations only). --Matthiaspaul (talk) 21:42, 10 November 2016 (UTC)[reply]

Hi LT910001 - just a heads-up, I've moved your proposal to this Citations page. -- DannyH (WMF) (talk) 01:47, 23 November 2016 (UTC)[reply]

Voting – Better support for commonly used sources

  1.   Support per my somewhat more general/flexible suggestion above --Matthiaspaul (talk) 23:46, 29 November 2016 (UTC)[reply]
  2.   Oppose Good problem but not great solution—instead encourage use of clipboard managers for power editors (macros to your heart's content) czar 00:39, 2 December 2016 (UTC)[reply]
  3.   Support The problem is relevant and the proposed solution will surely make improvement. Of course, we may switch to something more straightforward once it comes up as an idea.--Kiril Simeonovski (talk) 14:06, 2 December 2016 (UTC)[reply]
  4.   Support Actually a very useful idea. Storing the data in user subpages and loading the data into the citation machine. I love this feature and use a lot on MS Word.   --Hedwig in Washington (talk) 02:49, 6 December 2016 (UTC)[reply]
  5.   Support - suffer same problem, useful tool. DPdH (talk) 19:43, 6 December 2016 (UTC)[reply]
  6.   Support — I put a section on my user page with common citations in the past, precisely to solve this problem. If it had been built into the interface it would have make life easier. —Ynhockey (talk) 09:32, 7 December 2016 (UTC)[reply]
  7.   Support Miniapolis 21:00, 8 December 2016 (UTC)[reply]
  8.   Oppose. A useful idea, but not a development priority. Kudpung (talk) 02:44, 10 December 2016 (UTC)[reply]
  9.   Support This is necessary. There's a bit of a "war" (at en.wikipedia, anyway) against specific-source citation templates, so something has to replace that. Any loss in ease of sourcing equates to a reduction in citations.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  18:01, 11 December 2016 (UTC)[reply]
  10.   Support Will save me a lot of time. – Uanfala (talk) 23:41, 11 December 2016 (UTC)[reply]

Citation quality assessment

  • Problem: In the Wikipedias we currently have no metrics or even a general grasp as to the reliability or suitability of citations used throughout the wiki. This hinders editors' and readers' ability to judge reliability of particular sources or the overall quality of sourcing in an article (or group of articles).
  • Who would benefit: Wikipedia readers and editors; all media and their consumers
  • Proposed solution: Provide a way for users with well-rounded citing experience (perhaps having a special user right) to qualitatively assess, via a simple survey, citations with respect to various criteria (see "Examples of evaluation criteria" below). These assessments would aggregate for individual articles and various groups of articles (subject areas, WikiProjects, time periods, etc.) to determine the overall quality of sourcing. There would also be assessment aggregation for media organizations (e.g., newspapers and book publishers), and perhaps even authors, broken down in various ways (esp. time periods).
    • Examples of evaluation criteria:
      • Does the citation adequately support the preceding or otherwise associated content? (Strongly Agree/Somewhat Agree/Unsure/Somewhat Disagree/Strongly Disagree)
        • Does it support all content back to the previous citation or start of the section (whichever applies first)? (Y/N)
      • Does the source cover this content in a neutral manner? (i.e., does it give voice to all reasonable relevant points of view, not just culturally dominant ones?) (Strongly Agree->Strongly Disagree)
      • Does the source have financial interests of any kind (advertising, university funding, etc.) that may present a conflict of interest with respect to the content the source covers? (Strongly Agree->Strongly Disagree) (Note: I wouldn't expect many evaluators to be able to answer this, but I think it's a fair question that is compatible with this proposal)
  • More comments:
    • This would be a mechanism to help improve the overall quality of sourcing in Wikipedia while potentially holding media organizations accountable (a good side effect).
    • If a citation is assessed to not have the highest degree of quality for the content it is covering, this may spur the search for a more appropriate source with which to cite. The more this is done, the more reliable an article or group of articles become, assuming subsequent citation evaluations.
    • Wikidata:WikiProject Source MetaData, given it is the future of ID'ing and tying together citation uses across wikis, would be a prerequisite and thus the foundation of this proposal.
    • Ultimately, perhaps a "Citation Quality" badge based on the aggregated metrics for the page could be displayed for readers. This is a signal to the reader to "how well backed up" the article is.
    • Please anyone help me improve this idea, as there may be many blanks to fill in.
  • Phabricator tickets:

Community discussion

  • I imagine there would be some difficulty in tracking assessments to individual citations. Perhaps this assessment would have to take place on a revision of an article rather than the "current" version. At least then the cites would be in a steady state and therefore could be ID'd. If these assessments would migrate to subsequent revisions, there would have be some minor AI that decides that the citation is the same as before, or similar enough, whereby the assessment stands. Stevie is the man! TalkWork 18:44, 20 November 2016 (UTC)[reply]
  • This isn't quite as simple as one would believe, as the tool must identify which part of the text is in fact covered through a citation. There is also the problem of identifying the factoids in the citation, and how the cited source use those factoids. Existence of an identified factoid does not imply that the cited source actually confirm them, it can make a negative claim. The simplest solution is to identify a citation and assume it cover the previous period, then a metric would be number of citations divided by number of periods. That would give a coarse number, but it could be usable for some (g)estimations.
Yes I have made some efforts on solving this problem, and we can at least verify correct citations. That would imply storage of cited text fragments, but that is legal in most jurisdictions. If we want to do inference from the cited source, then the project would be a lot more difficult. — Jeblad 21:18, 20 November 2016 (UTC)[reply]
I am thinking of this as a qualitative survey that has the assessor make judgment calls perhaps by picking a Strongly Agree -> Strongly Disagree (or similar) rank on a few questions. Perhaps this could also include a free-form note where the assessor says how far back the cite covers, or what factoids are covered and which or not, and where to look on the cited material page. I'm not looking for anywhere near absolute precision. Of course, implementing all this wouldn't be easy. I'm basically looking for a way to test the cites and test the journalism simultaneously, for Wikipedia's and the media's mutual interest. 21:52, 20 November 2016 (UTC)
This would be pretty difficult to implement without Wikidata:WikiProject Source MetaData. For practical purposes, that will probably be a prerequisite. Ryan Kaldari (WMF) (talk) 23:22, 21 November 2016 (UTC)[reply]
  • I think you have confused "citation" with source. The latter is where we find material, a citation is what describes or points to the source. Assuming that you want evaluations of cited sources: this proposal is not well thought-out. The reliability, suitability, applicability, etc. of a source is strongly dependent on its use, and for that we have to rely on the judgment of the editor making the cite. To improve that it would probably be more fruitful to work up a guideline for edtiors. ~ J. Johnson (talk) 23:01, 27 November 2016 (UTC)[reply]
    • As a 12-year editor of Wikipedia, I don't think I can confuse these things, but perhaps the presentation of the proposal does so, and I will be happy to take steps to amend. What I am seeking is a qualitative evaluation of the source as used for the citation in backing up the content that is presented. Of course we rely on the judgment of the editor in creating the cite, but this proposal calls for an assessment of these cites so as to ascertain the reliability or suitability of citations, with assessments being aggregated. If there's a better way to present this, I would appreciate some example changes of text in the proposal. Stevie is the man! TalkWork 23:19, 27 November 2016 (UTC)[reply]
      • For what you seem to want I suppose each project could be encouraged to form teams of editors knowledgable in that area who could review each article in that project. But this proposal doesn't present any definite concept of what is wanted, which I think is a failure of concept. I don't know how to remedy that. ~ J. Johnson (talk) 22:58, 7 December 2016 (UTC)[reply]
  • Question What about citations to websites that become dead links? The citation may have been in order when it was made, but over time the web address has changed or there is some other cause. What would happen in a case like this under your proposal? Michael Glass (talk) 05:34, 2 December 2016 (UTC)[reply]
  • Question Just how is this different from the typical & expected process of debating sources on the Talk pages? Based on what I read so far, this proposal is simply reinventing the wheel, a wheel that works better than an unproven proposal. -- Llywrch (talk) 06:41, 2 December 2016 (UTC)[reply]
  • General comment: Given the aim of the proposal, if this move moves forward I would hope to see the evaluation criteria provide suitable accommodation for citation resources that are generated by independent or community based outlets/authors, particularly those from racialized or non-Western demographics. The nature of the verifiability process, for example, can discredit smaller publication outlets because they aren't deemed reputable enough, which in some cases is fair, while in others is a way to discredit legitimate news sources generated from underrepresented communities that sit outside mainstream (and increasingly homogenized) media outlets, etc.--Dnllnd (talk) 14:19, 7 December 2016 (UTC)[reply]

Voting – Citation quality assessment

  1.   Support as proposer. Stevie is the man! TalkWork 01:51, 28 November 2016 (UTC)[reply]
  2.   Support--Shizhao (talk) 02:34, 28 November 2016 (UTC)[reply]
  3.   Support would be a great way to expand the applications of the machine learning tools as well. This is important for reader literacy, Sadads (talk) 14:53, 28 November 2016 (UTC)[reply]
  4.   SupportJeblad 01:33, 29 November 2016 (UTC)[reply]
  5.   Support --Tarjeimo (talk) 21:54, 29 November 2016 (UTC)[reply]
  6.   Support – and best of luck! AWhiteC (talk) 18:27, 1 December 2016 (UTC)[reply]
  7.   Oppose To assess the quality of a news provider, one needs to know about many things: the company's history, the lives of its authors, the politics surrounding its subject matter and the frequency with which they are refuted by other commentators. All of these things change with time and even with millions more editors, there would be no way for us to objectively catalog every source in the world as being good or bad. Users with well-rounded citing experience already have a mechanism for improving the quality of sourcing. That mechanism is reading an article and applying whatever tags may be warranted (citation needed, unreliable source, dubious, clarify, not in citation given, non-primary source needed). If that sounds time consuming, you bet it is. The people doing this would probably appreciate your help more than a vague, McCarthy-esque layer of micromanagement. Connor Behan (talk) 19:02, 1 December 2016 (UTC)[reply]
    Thank you for expressing your concerns. To clarify one aspect, this is about assessing the quality of the citation with respect to the actual content in the source. The person doing this isn't making a full judgment of the news provider -- just a judgment about the specific material as used for the cite. If there's any potential "judgement" of the news provider, it comes in aggregates of assessments. Also, certainly the process of citing should get all the assistance it needs. This proposal is a quality assessment not just for the sake of editors, though. We're here for the readers too. They deserve to be able to get an idea of "how well backed up" an article's content is. The "McCarthy-esque" part doesn't really need a response, as QA and the expectation of quality is anything but that. Stevie is the man! TalkWork 22:22, 1 December 2016 (UTC)[reply]
  8.   Oppose Completely unrealistic. As Connor described above, editors already use considerable judgment to do these things whenever we want to, using template tags & article reviews & quality assessments & policy guidelines & Talk page comments, and I see nothing in this proposal that's likely to improve the situation. Sorry. —Patrug (talk) 21:04, 1 December 2016 (UTC)[reply]
    These are all important activities, no doubt, but I don't see how they address the Problem as I stated in the proposal. Stevie is the man! TalkWork 18:56, 5 December 2016 (UTC)[reply]
  9.   Oppose Verification is, I believe, the paramount concern of Wikipedia readers and perhaps even the editors. I agree that WP doesn't have any solid mechanism for showing (1) how well the citation verifies the source, and (2) the veracity of the statement itself, but I also agree that the above as proposed isn't the solution. Based on the wiki format, readers/editors already have the tools to verify content and propose changes, even if the accessibility isn't there. The idea of a survey to assess quality reminds me of the reader feedback that used to appear at the bottom of the pages. I liked that, personally, but I also know it was removed because more people disagreed. czar 00:46, 2 December 2016 (UTC)[reply]
    Note that this would not be general public feedback like the reader feedback. This would be done only by those editors deemed to understand citations/sourcing. Stevie is the man! TalkWork 06:23, 2 December 2016 (UTC)[reply]
  10.   Oppose I am concerned that there is a great imbalance between citing facts from 1) a scarce set of sources and 2) an abundant set of sources. The problem here is that readers usually do not know if the citation they are assessing belongs in 1) or 2), so they will most likely treat all citations the same, assuming that there is an equal amount of sources available. This information asymmetry will inevitably end up in scores for different citations that are not comparable one to another, which may be easily converted into inappropriate recommendations to work with the limited resources for the lower quality citations (the quality assessment will become a market for lemons). To explain this more concisely, here is an example:
    Suppose that fact A is cited with an average source out of 5 available and fact B is cited with an average source out of 500 available. Given the vastly larger size of the set of sources for fact B compared to the one for fact A, it is more likely that the average quality of sources is higher and the best source is with higher quality for fact B than for fact A. A reader assessing the citations quality for fact A and fact B normally assumes that they were both taken from sets with the same size, say, 100 each. The outcome of the quality assessment will be higher score for the citation of fact B than for fact A, indicating higher priority to work on improving the quality citation of fact A. Since the set of sources for fact A is small, the quality improvement is limited and the overall result will be insignificant.
    In fact, there are many Wikipedia articles on topics that have small sets of sources available (with average quality that barely exceeds the minimum threshold of reliability), where any improvement of the citations quality will not result in a significant positive change.--Kiril Simeonovski (talk) 15:40, 2 December 2016 (UTC)[reply]
    These are certainly very useful concerns, so I would hope they would be considered carefully when devising a citation assessment approach. It seems a bit too complex for me to answer very well at this stage. Stevie is the man! TalkWork 19:01, 5 December 2016 (UTC)[reply]
  11.   Oppose For example: is Fox News a credible source? A lot of people I know would say no, other would say yes. Is The Huffington Post a credible source? A lot of people I know would say no, other would say yes. Fundamentally, citations are there so that if a user wants more information, or if they disagree with what they've read, they can go and look at the source for themselves. Then they can determine if the source is credible or not.
    Those doing these assessments would have to work by a guideline that is testing the actual source content against the wiki content per the citation. They wouldn't be grading the overall media company. We certainly should have a guideline against doing that (in case it's not obvious). Also, this assessment would not be open to all editors -- just ones who have proven knowledgeable with respect to citations/sourcing. Stevie is the man! TalkWork 18:54, 5 December 2016 (UTC)[reply]
  12.   Oppose The use(fulness) of sources depends on the circumstances, they need to be assessed individually and not automatically. --Andropov (talk) 17:44, 5 December 2016 (UTC)[reply]
    This proposal is for individual assessment by human editors. There is no call for automatic assessments. The only thing that might be considered automatic would be the aggregation of assessments done by these human editors. Stevie is the man! TalkWork 18:48, 5 December 2016 (UTC)[reply]
  13.   Support - What can't be measured can't be managed, assessment metrics can help assessors in their job. DPdH (talk) 19:48, 6 December 2016 (UTC)[reply]
  14.   Neutral Good in theory, but nigh impossible to achieve. ArgonSim (talk) 18:16, 7 December 2016 (UTC)[reply]
  15.   Oppose: No definite concept of just what is wanted. (As to providing a metric: how? on what basis? Not explained.) ~ J. Johnson (talk) 23:03, 7 December 2016 (UTC)[reply]
  16.   Oppose Completely unrealistic, the difficulty is on par with teaching robots to write articles. Better direct team efforts to less fantastical endeavors like suggesting sources from other language versions. Ivan Pozdeev (talk) 12:00, 8 December 2016 (UTC)[reply]
  17.   Support Miniapolis 21:02, 8 December 2016 (UTC)[reply]
  18.   Oppose a) I can't see a compelling need. b) Techially disproportinally complex to develop. c) A bot to automatically detect and flag dead links and disallowed external links would be a better idea, but it hasn't been proposed.Kudpung (talk) 02:51, 10 December 2016 (UTC)[reply]
  19.   Oppose Too much room for bias and although splitting it up by project seems like a good idea - a WP:FRINGE reference used erroneously elsewhere would end up with a low score which would be harmful when attempting to put it in proper context. Panyd (talk) 15:06, 11 December 2016 (UTC)[reply]
  20.   Support FoCuSandLeArN (talk) 22:56, 11 December 2016 (UTC)[reply]

  • Problem: None of the common automated, full-page, bare-link citation completion tools (ReFill, Reflinks, etc.) seem to go very far in completing a citation, and often leave too much manual work for the editor.
  • Who would benefit: Wikipedia editors
  • Proposed solution: Put some development resources into improving these tools. Find out ways where they are lacking (like not usually adding author/work/date or using the correct citation templates). Make these cite completion tools more, er, complete.
  • More comments: If these tools can't be improved much further in terms of completing the citations, they should at least make it easier to manually complete the cites. Sending us to a preview/diff page doesn't really help here. Each improved cite should be presented directly for filling in missing information.
  • Phabricator tickets:

Community discussion

  • Use citoid? It should be possible to complete existing templates with the tool, now the template can only be manually edited. — Jeblad 18:55, 21 November 2016 (UTC)[reply]
  • @Stevietheman: It sounds like you want a tool that does what Citoid does, but acts on each bare link on the entire page. However one could edit using VisualEditor, and selectively use Citoid, where they'd be able to fix any errors that may happen. If you are not familiar with VisualEditor and/or Citoid, would mind trying it out and see if this fulfills your needs? Otherwise maybe we clarify this proposal to be an implementation of Citoid that acts on the entire page (like ReFill and Reflinks), rather than having to go through each reference one by one. This would ensure voters know exactly what they're supporting/opposing during the voting phase. Thanks for your participation! MusikAnimal (WMF) (talk) 19:16, 21 November 2016 (UTC)[reply]
  • Jeblad and MusikAnimal, this is about tools that fix bare-link citations on the whole page, pulling information from the URLs in the cites. This shouldn't have anything to do with the editing tool. Stevie is the man! TalkWork 21:32, 21 November 2016 (UTC)[reply]

Hi Stevietheman - just a heads-up, I've moved two of your proposals to this Citations page. -- DannyH (WMF) (talk) 01:46, 23 November 2016 (UTC)[reply]

  1.   Support as proposer. Stevie is the man! TalkWork 01:51, 28 November 2016 (UTC)[reply]
  2.   Support or an option to edit a bare link citation with citoid!! Ninovolador (talk) 11:30, 28 November 2016 (UTC)[reply]
  3.   Oppose Already exists in Citoid, and will be integrated into the next generation Wikitext editor, Sadads (talk) 14:51, 28 November 2016 (UTC)[reply]
    Does Citoid automatically run across a full page, converting all bare links into full cites? I just want to make sure we're not comparing apples to oranges here. Stevie is the man! TalkWork 14:53, 28 November 2016 (UTC)[reply]
    No, but it greatly simplifies targeted footnote expansion. The problem with doing batch fixes, is that most folks won't take the time to gut check the citation to make sure it's thorough. The current situation forces systematic review of the citation itself and examining the citation in context of the verifiable context, which is much preferable IMO. Sadads (talk) 13:28, 29 November 2016 (UTC)[reply]
    I appreciate your concern here and it is generally best for editors to pay attention to specific citations to get them just right. However, I come across many articles with many bare links, and making them all much better than before with an automatic tool is a welcome stop-gap. These cites can always be improved further at a later time. I just want the automatic full-page tools to be as complete in their work as possible. Stevie is the man! TalkWork 15:42, 29 November 2016 (UTC)[reply]
  4.   Support Would like to see redoubled efforts to auto-complete at least "work", "publisher" and "date" for a wider range of citationsNoyster (talk) 13:58, 1 December 2016 (UTC)[reply]
  5.   Support Connor Behan (talk) 18:37, 1 December 2016 (UTC)[reply]
    I have a script that does this—just need to clean it up and do more testing for use cases that aren't my own czar 00:47, 2 December 2016 (UTC)[reply]
  6.   Support --JB82 (talk) 01:12, 2 December 2016 (UTC)[reply]
  7.   Support Libcub (talk) 02:16, 2 December 2016 (UTC)[reply]
  8.   Support--Kiril Simeonovski (talk) 13:57, 2 December 2016 (UTC)[reply]
  9.   Support Please at least localize them. We lack such tools in smaller Wikis. 4nn1l2 (talk) 15:50, 2 December 2016 (UTC)[reply]
  10.   Support -- SBaker43 (talk) 23:52, 2 December 2016 (UTC)[reply]
  11.   Support--Ranjithsiji (talk) 03:28, 4 December 2016 (UTC)[reply]
  12.   Support--Thunderforge (talk) 15:10, 6 December 2016 (UTC)[reply]
  13.   Support--Afernand74 (talk) 08:21, 7 December 2016 (UTC)[reply]
  14.   Support Iadmc (talk) 04:46, 8 December 2016 (UTC)[reply]
  15.   Support Miniapolis 20:51, 8 December 2016 (UTC)[reply]
  16.   Support Should also trigger and link to an archive capture of the url target, be that WikiWix, or Archive-it, or something local. There's no excuse anymore for allowing content based on subsequently-dead links to become unsupported. LeadSongDog (talk) 18:03, 9 December 2016 (UTC)[reply]
  17.   Support Interesting comments/possibilities above. Might be especially useful for non-top-10 wikis, to have those gadgets made localizable. Quiddity (talk) 04:45, 12 December 2016 (UTC)[reply]