Wikipedia talk:Date formattings

MOSNUM

"Full dates should be linked per MOSNUM" Not any more. See WP:UNLINKDATES. Art LaPella (talk) 18:55, 8 September 2008 (UTC)[reply]

no bots

unlink dates that were unthinkingly linked, yes. Use bots, no. Dates belong unlinked only when an eidtor makes the conscious decision that the date should not be linked in this context. It isn't acceptable to ask editors to add articles to all sorts of whitelists to stop the bots from making unwise edits. If the bots cannot deliver, switch them off. The minimum requirement should be that a bot is able to note and remember when it is reverted, and leave the article alone after that. Otherwise you are effectively using bots to edit-war, which is a practice extremely frowned upon. --dab (𒁳) 09:55, 24 August 2009 (UTC)[reply]

Not at all. If someone has an unusual requirement whitelisting is not unreasonable, we use it elsewhere. {{bots}} is another example, as is {{sic}}. Rich Farmbrough, 09:42, 27 August 2009 (UTC).[reply]

Ambiguous date examples

How is "7/7/1961" an ambiguous date? What else could it be besides July 7, 1961? GoingBatty (talk) 02:36, 11 October 2011 (UTC)[reply]

Er, yes. Bad example. Potentially ambiguous, depending on the numbers. --Ohconfucius ^¡digame! 04:59, 11 October 2011 (UTC)[reply]
- "6/7/1961" is a much better example - thanks for the quick response! GoingBatty (talk) 16:57, 11 October 2011 (UTC)[reply]
  - "7/7/1961" could also be 7 July 1961 . But d=m point noted. Dl2000 (talk) 03:57, 12 October 2011 (UTC)[reply]

ISO 8601?

The project page makes several references to "ISO 8601". However, although this format is an option for timestamps on some system messages, it is has never been adopted for articles. There is no consensus as to whether article dates in the YYYY-MM-DD format are ISO 8601. Jc3s5h (talk) 12:05, 20 September 2012 (UTC)[reply]

outdated scripts section?

Hi, I just tried using the Date.js from User:Plastikspork, but it doesn't seem to be working for me. Is there a newer script that is commonly used? Rachel Helps (BYU) (talk) 15:49, 29 August 2018 (UTC)[reply]

User:Ohconfucius/script/MOSNUM dates.js is the script currently used. Rachel Helps (BYU) (talk) 16:08, 29 August 2018 (UTC)[reply]

Ambiguious dates...

What is the proper way to deal with dates when citing journals. Frequently, they are quarterly or even less regular. This one is a good example: "Asian Music Vol. 18, No. 1 (Autumn - Winter, 1986). To me that means anywhere between August and February. I expect this has come up, but don't know what others have done. Jacqke (talk) 23:45, 1 May 2021 (UTC)[reply]

@Jacqke: There's two issues in your question:

Date ranges
Seasons

1) Date ranges

For date ranges, I usually just pick one. I'm not sure it matters which. The goal is to find the cited reference. Imagine browsing a library or an ordered list of issues. If the citation says "March" and the list says "February, "March/April", "May", you won't have any difficulty selecting the rigth issue.

2) Seasons

First, there's no reason you can't include the season in dates using CS1 templates, eg:

Wrazen, Louise (Winter 1986). "The Early History of the Vina and Bin in South and Southeast Asia". Asian Music. 18 (1): 35-55. doi:10.2307/834157.

In a periodical with volume and issue numbers, specifying the date of publication beyond the year isn't necessary to identify and find the article in question. Most citation style guidelines recommend omitting them. Thus, I would usually write, eg:

Wrazen, Louise (1986). "The Early History of the Vina and Bin in South and Southeast Asia". Asian Music. 18 (1): 35-55. doi:10.2307/834157.

If the periodical doesn't have an issue number and doesn't use continuous pagination, then you really do need to include the date of publication in order to know which issue contains the article.

You can often find a more precise publication date on an article webpage. For example, Tech Journal Winter 2021 doesn't specify a date of publication. However, a quick glance at the articles within suggest that they all show a publication date of 13 December 2021. Daask (talk) 17:55, 26 January 2024 (UTC)[reply]

The goal in a citation is to state the nominal publication date, that is, the date that would appear on the cover, spine, or title page of a printed publication date. The purpose is that when using resources such as web indexes and library catalogs, a reader can find the publication, and once a likely publication has been found, the reader can verify that the correct publication has been found. Identifying the exact day the publication was published (whatever that means) is not the goal of the citation. Jc3s5h (talk) 15:49, 31 January 2024 (UTC)[reply]

Autoformatting dates in CS1 templates

Whenever I create CS1-style citations, I just bung in yyyy-mm-dd dates (because it's quick) in the sure knowledge that the date format template, e.g. {{Use dmy dates}}, at the top of the article will automatically render the dates correctly for the reader (per Template:Use dmy dates#Auto-formatting citation template dates). Other editors seem keen to run scripts to get the dates formatted correctly behind the scenes despite the date format template doing the donkey work for them. Is the template's autoformatting capability deprecated for some reason? I'm a little confused. Maybe a brief note on this in MOS:DATEVAR might clear this up? Rodney Baggins (talk) 18:56, 4 May 2021 (UTC)[reply]

My two cents' worth: The scripts were written long before the latest template was configured to autoformat dates. While what you state is true, I believe that it's still desirable to harmonise dates because it looks tidier when in edit mode. The other benefits are that a script run tidies up all dates on the page, not just those within citation templates. Autoformatting not only chews up processing power of our servers, it does not work when inconsistent dates exist outside the template date parameters, and the algorithm cannot cope with errors that exist with the template parameters, whereas the script will correct a certain number of these (such as access dates that are not: yyyy-dd-mm dates, mmm yyyy dates). In any event, the script run is one mouse click and takes only a few seconds to execute. Removing the script code that changes dates within citation templates would make a marginal difference to its performance, and it's easier than having to rewrite the script . -- Ohc ^¡digame! 21:44, 4 May 2021 (UTC)[reply]

Should all articles have a dmy/mdy template?

There are templates like {{Use dmy dates}} or {{Use mdy dates}}, which are often placed at the top of an article setting the preference for the date format.

Should they be used on all articles? Including articles whose dates are already correctly formatted and where the template won't result in a visible change to the article text? Or on articles that don't currently have any dates in them (either in the prose or in the references)? – Uanfala (talk) 01:13, 21 April 2022 (UTC)[reply]

There is no hard requirement to add them, although generally there is no harm to add them. I usually only add them when the article has a mix of date formats and it needs consistency - especially in the references where the cite templates obey the use xxx dates templates. Stepho talk 04:11, 21 April 2022 (UTC)[reply]

I always add them to new articles I create, because in my experience if I don't do this then other editors will come around with scripts to auto-add them and change the format from my preference to something else. (Most recent offender: User:BrownHairedGirl [1] [2] [3] [4] [5] [6], but she's far from the first.) The editors who do this aren't generally insistent on one format over another, but the scripts they are using to do this aren't smart enough to notice that they are changing one consistent format to another and the editors who use them, when asked not to do this, generally say that they will keep on doing it because they like what their script does and don't have the expertise to make it smarter. —David Eppstein (talk) 05:18, 21 April 2022 (UTC)[reply]

Yeah, part of the reason I asked here was to find out whether mass additions like this are OK. Currently an editor (that same one) has started mass adding the dmy template to the tens of thousands of Britain-related articles. – Uanfala (talk) 12:25, 21 April 2022 (UTC)[reply]

It seems to me that an answer that applies to editors taking an action one article at a time, and taking a good look at each article, does not necessarily apply to bots. Also, certain rapid editing techniques that are not technically bot edits nevertheless fall under the bot rules. Jc3s5h (talk) 15:59, 21 April 2022 (UTC)[reply]

Not all articles have dates. An example would be namespace disambiguation; many scientific articles also don't have dates. The template would not be needed in these. However, most articles on UK or Commonwealth subjects ought to display dates in dmy, and most US (except US military) articles would display in mdy per WP:TIES, so inserting the relevant template would be entirely appropriate in those cases, even if done en masse by automated or semi-automated editing. BTW, I don't think it would be collegial to use terms such as "offender" to refer to editors who do this work merely because you dislike it.

-- Ohc ^{revolution of our times} 16:12, 21 April 2022 (UTC)[reply]

It is difficult to determine, en masse, whether an article is US non-military or UK. As an example, an article could be about a person who was born in the British Isles but didn't become prominent until moving to an American British colony which later joined the US. The features associated with the article, such as birthplace, categories, or WikiProjects could be difficult for automation to interpret.

Of course Canada is a Commonwealth country but there is no consensus on which date format to use for articles related to Canada. Jc3s5h (talk) 16:39, 21 April 2022 (UTC)[reply]

We don't even know what measurement units to use, or how we're supposed to spell things anymore. _signed,Willondon (talk) 17:42, 21 April 2022 (UTC)[reply]

@Jc3s5h: in my list-making for adding {{Use dmy dates}}, one of the steps I take is to try to exclude all emigrants. That's particularly important for Irish biogs, where emigration was huge for the 1840s to 1990s, but I also apply it to UK biogs.

I use many other filtering techniques, but that's what I do on that one. BrownHairedGirl (talk) • (contribs) 00:07, 23 April 2022 (UTC)[reply]

Thanks for the ping, @David Eppstein. It is deplorable that the same courtesy was not extended to me by @{{u|Unfala}, who opened this discussion about an issue they were discussing on my talk, without notifying me, and without notifying an ANI thread they opened about me.

I have checked the first 4 of the diffs you posted above, and am surprised by your objections:

[7] Sarah B. Hart is described in the article as a British mathematician whose academic career has been entirely at British universities
[8] Alice Rogers is described in the article as a British mathematician whose academic career has been entirely at British universities
[9] Nancy K. Nichols: nationality not stated, but her only stated academic affiliation is to a British university
[10] Joyce Aylard is English, and appears to have spent all of her career in England

So AFAICS, per MOS:TIES, they should all use the DMY date format, and my edits were correct. I see no policy basis for your personal preference for another date format, or for your decision to spend some of your time modifying the {{Use dmy dates}} templates to use the ANI date format in the citations. What is that all about? BrownHairedGirl (talk) • (contribs) 18:03, 21 April 2022 (UTC)[reply]

The change you made was from spelled-out dates for publication dates and numeric dates for access dates and archive dates (my preferred format, and an acceptable format according to the MOS) to spelled-out dates for everything including access dates and archive dates. See the immediately following edits to each of the diffs above, in which I corrected the dmy tag to specify this format. —David Eppstein (talk) 18:06, 21 April 2022 (UTC)[reply]

@David Eppstein: see MOS:DATEUNIFY: "When a citation style does not expect differing date formats, it is permissible to normalize publication dates to the article body text date format, and/or access/archive dates to either, with date consistency being preferred".

That was what my edits achieved, and which your edits undid. Please revert your edits. BrownHairedGirl (talk) • (contribs) 18:20, 21 April 2022 (UTC)[reply]

Incorrect. Spelled-out pubs and numeric access IS A CONSISTENT FORMAT. It is a format supported by the MOS, it is a format supported by the dmy tags, and it is a format supported by the citation templates, which automatically convert citations to that format when the dmy tags specify that format. It is a format I chose intentionally, because when encyclopedia readers look at reference dates I want their attention to be drawn towards the important dates, the publication dates, and away from the dates that are more relevant only for editors, the access dates. You are changing that consistent format to a different format. You should not be doing that. —David Eppstein (talk) 18:33, 21 April 2022 (UTC)[reply]

I'd given roughly zero thought to the matter prior to this, but I do like that formatting. It draws the reader's attention to the appropriate place, keeps the overall length down a bit, and allows for situations where the publication date is something like "Spring 1970". Putting on my wiki-lawyering wig, I interpret "consistent" to mean "applied in the same way across all the references in an article", so as long as all the citations use spelled-out publication dates and numerical access dates, it's fine. XOR'easter (talk) 15:08, 22 April 2022 (UTC)[reply]

David Eppstein's edit was in accordance with the discussion at Wikipedia talk:Manual of Style/Dates and numbers/Archive 158#Proposal: End "date-forking" into different styles for publication and access/archive in same cite. Jc3s5h (talk) 18:38, 21 April 2022 (UTC)[reply]

That is a perverse decision. If YYYY-MM-DD dates are to be used for accessdate and archivedate in conflict with article style, then the citation templates should apply that style consistently.

It makes no sense at all to have a situation where an editor can set their own personal preference, creating a random jumble of styles across a set of similar articles. BrownHairedGirl (talk) • (contribs) 19:03, 21 April 2022 (UTC)[reply]

You are not using the more advanced version of the use xxx dates templates. See Help:Citation Style 1#Auto-formatting citation template dates. As an example, if you wanted dmy for dates in the article and publication dates, but YYYY-MM-DD for the access and archive dates, you could add

Of course, that would complicate your list-making process. Jc3s5h (talk) 19:15, 21 April 2022 (UTC)[reply]

I would prefer a solution where formatting of dates is a reader preference as discussed below rather than an editor preference, but we don't have that available currently. As it is, MOS:DATERET applies. —David Eppstein (talk) 19:30, 21 April 2022 (UTC)[reply]

Thanks, @Jc3s5h, but I absolutely will not even attempt to apply that sort of parameter.

If an editor wishes to complicate things by asserting their own personal preference for date format in refs (and until there is a consensus to end this absurdity), then they can add the relevant markup themselves to clarify that the yyyy-mm-dd dates are a stylistic choice rather than the default addition of a bot or tool.

My additions of {{use dmy dates}} skip any page with an existing dmy/mdy template, so any such preference will be undisturbed. BrownHairedGirl (talk) • (contribs) 20:34, 21 April 2022 (UTC)[reply]

If you add {{use dmy dates}} to a page which consistently uses dmy for everything except access dates and archive dates, the last two being formatted YYYY-MM-DD, you will change the date format of the displayed page, which seems to not be in accord with the consensus of the thread I referenced in my post of 18:38, 21 April 2022 (UTC). I would count that as disturbing the page. Jc3s5h (talk) 20:46, 21 April 2022 (UTC)[reply]

@Jc3s5h: in the vast majority of such cases any YYYY-MM-DD dates are simply the default output of various tools which have not determined the page's style.

For that reason, if an editor wishes to assert that YYYY-MM-DD is an actual style choice, they need to state that explicitly. BrownHairedGirl (talk) • (contribs) 21:32, 21 April 2022 (UTC)[reply]

It should be entirely possible to distinguish articles where the publication dates and access dates are consistently formatted, from tool-formatted citations where the date formats are not consistent. It just takes a little effort rather than blind scripting. What I am hearing from you is that you are unwilling to put any effort into respecting other editors' choices. —David Eppstein (talk) 04:49, 22 April 2022 (UTC)[reply]

Quite the contrary, @David Eppstein: I have put that effort in from the outset of my {{Use dmy dates}} task, long before your objections.

If you declare your choice by applying the template, then I guarantee that unless there is some malfunction of AWB, my tools will respect your choice by skipping the page. They are set to skip any page where the markup includes {{Use dmy dates}} and/or {{Use mdy dates}}, or their aliases, using this regex:

\{\{ *(Template *: *)?(Use[_ ]+dmy[_ ]+dates|Dmy|DMY|Use[_ ]+DMY[_ ]+dates|Usedmy|Use[_ ]+dmy|Use[_ ]+DMY|Use[_ ]+mdy[_ ]+dates|Usedmydates|Mdy|MDY|Use[_ ]+MDY[_ ]+dates|Usemdy|Usemdydates|Use[_ ]+MDY|Use[_ ]+mdy)

So, if a page contains e.g. {{Use dmy dates|cs1-dates=ly|date=April 2022}}, then my AWB jobs will skip it.

However, if that template has not been applied, then I will not try to determine whether any YYYY-MM-DD dates in the template have been added as a matter of choice by an editor, as the default output of a tool or bot.

The only way that I can see to make such a determination is by examining in detail each revision and checking both its author and its edit summary. That is an onerous job even for a human, and I cannot even begin to think how a bot could do it. I guess that a more skilled programmer might be able to accomplish some of it, but even then I think that it would be unreliable.

Now, will you please stop with the aggressive edit summaries when you belatedly add to articles the templates declaring your preference? There has been a flurry of them this morning, like this one[11]

If you declare the pref, I will respect it. If you choose not to declare the pref, don't complain about the consequences of your omission: it is thoroughly unreasonable of you to demand that other editors and their tools try read your mind. And you have been here long enough to know better than to abuse edit summaries as a tool of attack. BrownHairedGirl (talk) • (contribs) 08:52, 22 April 2022 (UTC)[reply]

I do not require that BrownHairGirl read minds. It seems to me BrownHairGirl is demanding that other editors be clairvoyant. The citation template feature that allows publication dates to be automatically formatted differently from access dates was first documented in April 2019 by Trappist the Monk. Editors who cleaned up citations and made the format consistent before then could not have known to add the advanced syntax of the use xxx dates template. Indeed, doing so would have formatted all the dates the same, which, for some articles, would have violated WP:CITEVAR.

It also seems to me that two forces are working, from opposite directions, against editors who just like to write in English, not computer code. One crowd wants to use the YYYY-MM-DD dates because they want to make everything easy for computer coding. BrownHairGirl wants to run her automation rather than actually examining the articles. Jc3s5h (talk) 12:55, 22 April 2022 (UTC)[reply]

I looked at some of the diffs in at the top of this discussion, eg [12] The earlier version had no template, didn't use any dates beyond years in the prose but consistently used yyyy-mm-dd in the references. BrownHairedGirl applied the dmy template without options and the article then displayed references as dmy. That is a change to the reference format - a change done without consensus. Which is against MOS:RETAIN and WP:DATERETAIN. WP:CITESTYLE allows yyyy-mm-dd (even when the subject is purely British), therefore a valid reason or consensus must be given before a change in reference format can be accepted (or rather, you can be bold but any editor can challenge your change, flip it back and start a discussion, as per WP:BRD). These changes are not just a case of adding an administrative template - these are changes to the reference style. Adding a dmy or mdy template with options to keep using the same reference format that the article previously used is perfectly acceptable but that is not what happened in these cases. Stepho talk 13:12, 22 April 2022 (UTC)[reply]

@Jc3s5h & @Stepho-wrs: you both seem to miss my core point.

So I will restate it: without very detailed and painstaking examination of the page's history, there is no way for an editor or tool to determine whether any YYYY-MM-DD dates in citation templates are

the choice of the original major editor
the output of a tool or bot which uses YYYY-MM-DD dates as a neutral default.

The overwhelming majority of cases are of Type2: the output of a tool or bot.

The cases of Type1 (choice of the original major editor) can be reliably determined only if they are tagged. So this is not a bot or script issue; it is a failure-to-declare-style issue. BrownHairedGirl (talk) • (contribs) 14:34, 22 April 2022 (UTC)[reply]

I agree that many of the style retention guidelines cannot be followed without a detailed look at the talk page, and if nothing is found there, the edit history. It is indeed burdensome. You seem to feel that because this burden interferes with your automated procedures, you can just ignore the current consensus.

The "correct" procedure would be to propose a change to the consensus. But that would bring up a long discussion, with all the ideas put forward since the founding of the project being thrown into the discussion. For example, I would certainly propose banning YYYY-MM-DD dates altogether except in articles about that format. So I see no good way forward. Jc3s5h (talk) 14:45, 22 April 2022 (UTC)[reply]

@Jc3s5h: an undeclared and hard-to-detect style like this does not just interfere with my current work. It interferes with the work of any editor, regardless of whether they are using any tools, 'unless it is explicitly declared.

Whatever system we have for using or not-using YYYY-MM-DD dates should be consistent and rules-based. Making this a matter of editor preference just creates an entirely avoidable tangle such as what we have here.

But so long as we are stuck with the mess created by editor-preference, the only remedy is for those who impose that style to declare it by using {{Use dmy dates|cs1-dates=ly|date=April 2022}} ... or accept that their preference looks like an oversight, and is likely to be overridden. BrownHairedGirl (talk) • (contribs) 17:25, 22 April 2022 (UTC)[reply]

The style you are changing is undeclared because it was established in these articles long before the ability to declare them properly existed. You are asking for an impossibility and using its impossibility as a poor excuse to run roughshod over MOS:DATERET. If you are irritated by edit summaries documenting your misbehavior, it is only the least of the consequences you should face for this misbehavior. —David Eppstein (talk) 18:02, 22 April 2022 (UTC)[reply]

@all: AFAIK, the vast majority of editors and readers aren't bothered if there are YYYY-MM-DD or not. But when they do, they do so with a vengeance. I have made hundreds of thousands of such script-assisted edits that render date formats uniform, and see placing of the template as imposing generally desirable uniformity of date formats in a non-uniform Wikipedia. However, I make it a point not to contest if any of my edits are reversed because of upset sensitivities. -- Ohc ^{revolution of our times} 19:44, 22 April 2022 (UTC)[reply]

David Eppstein seems determined to adopt a WP:BATTLEGROUND approach, so I see no point in continuing the conversation. BrownHairedGirl (talk) • (contribs) 22:47, 22 April 2022 (UTC)[reply]

I would be perfectly happy if you would stop changing the established date format of articles, especially those on which I was responsible for establishing the date format. As it is, you are making a lot of work for me checking all of your many edits that show up on my watchlist and fixing the ones where you have changed the established date format and where I was the one to establish it. That is all I have been doing. If my continuing to defending the integrity of the content I have created is battleground behavior in your mind, and your continuing to pooh-pooh any suggestion that you change your MOS-violating behavior is not, then the problem is in your mind, not in my behavior. —David Eppstein (talk) 07:41, 24 April 2022 (UTC)[reply]

We should all be using machine-readable YYYYMMDD format in the wikitext, and then let the reader's client render that date in their local format. We have the technology to do this (we already do it for talk page comment timestamps), but we just don't do it in mainspace, because... well, because Wikipedia will always choose to have an argument over something instead of implementing a technological solution that makes everyone happy. Levivich 17:30, 21 April 2022 (UTC)[reply]

@User:BrownHairedGirl, I don't believe I have missed the point but rather that the point is irrelevant. It doesn't matter whether a person or a bot set the date format - only that it is the existing, established format and that you have changed it. You don't have to painstakingly check the history and/or talk pages - you only have to check the current status of the page. If the references are yyyy-mm-dd before your edit then you should make sure that they are yyyy-mm-dd after your edit. You are of course free to start a discussion for consensus to change it. Or even be bold and just change it - but other editors are then free to revert your bold changes, as per WP:BRD. Stepho talk 23:02, 22 April 2022 (UTC)[reply]

@Stepho-wrs: that approach assumes that YYYY-MM-DD dates added by a bot or tool represent a conscious style choice. I disagree with that assumption. BrownHairedGirl (talk) • (contribs) 23:04, 22 April 2022 (UTC)[reply]

@User:BrownHairedGirl, I'm making no assumptions about an article's history. The article is what it is, regardless of whether it was a person choosing wisely, a person choosing arbitrailiy or a cold, heartless bot. It seems that you have your own assumption - that YMD dates were all (or mostly) generated by bots and therefore changing away from YMD is automatically ok. No - you must respect the state of the article as it is when you start editing. That is the point of WP:RETAIN and WP:DATERETAIN. Stepho talk 23:34, 22 April 2022 (UTC)[reply]

Stephen, I apply {{Use dmy dates}} per MOS:TIES, and put a lot of time into selecting articles which have close ties to the UK or Ireland.

The default formatting for dates in the cases is DD Monthname YYYY. If editors wish to assert an exception, the means exist for them to do so, either before or after I add {{Use dmy dates}}.

A above, I do not believe that the existence of possibly bot-formatted YYYY-MM-DD dates necessarily indicates a style choice.

Please re-read MOS:DATERET, and its stress on "strong national ties". BrownHairedGirl (talk) • (contribs) 23:46, 22 April 2022 (UTC)[reply]

@User:BrownHairedGirl, my name is not Stephen but I do go by the nickname Stepho. I double checked at WP:TIES and it is about language, not dates. Assuming you really meant WP:DATETIES, that talks about dates in the article text. Higher up on the MOS page in the WP:DATEFORMAT section it says "Special rules apply to citations; see Wikipedia:Citing sources § Citation style" - ie, references do not necessarily follow the guidelines given on the rest of the page DATEFORMAT/DATETIES page. Following the citing sources link, it says that YYYY-MM-DD is an allowed reference style. I see nothing that says DATETIES overrules DATERET for references. I do see WP:CITEVAR that says don't change the reference style. If DATETIES did overrule DATERET and CITEVAR then that is effectively saying that YYYY-MM-DD is banned on all UK and US articles and also for many other countries that officially use DMY or MDY. In which case, why bother to allow YYYY-MM-DD at all? Stepho talk 10:30, 23 April 2022 (UTC)[reply]

@Stepho-wrs:Of course, the devil is always in the detail. There are often diffferent date formats used in any given article, and the collection intra se may be inconsistent even with WP:CITEVAR in mind. Acceptance of YYYY-MM-DD date formats within citations, whether within |date=, |archivedate= or |accessdate= does not make them obligatory in any way. The overriding concern is uniformity, and not what User:EditorX or User:EditorY prefers as being the most [or insert your preferred superlative].

Many en.wp articles pre-date the widespread use of citation templates, or the current best practice of applying citations or indeed access dates or archive dates. The only way would be an investigation into the history of date formats used in each and every article on WP before applying a tag, and this bearing in mind the marginality of the YYYY-MM-DD format in the overall scheme of things, would IMHO not be conducive to the advancement of the project. -- Ohc ^{revolution of our times} 11:19, 23 April 2022 (UTC)[reply]

@Ohconfucius:, what you appear to be saying is that YYYY-MM-DD are allowed in references but not obligatory - which I agree with. But then you say that because they are not obligatory it is therefore ok to change them at will. As we both agree, YYYY-MM-DD is allowed in references but not text. By your chain of thought for applying WP:DATEUNIFY to cover both text and references indiscriminately, then YYYY-MM-DD is not allowed in references either because it will always be different to the date format used in the text (where we agree that it is not allowed). If it is not allowed in the text and effectively not allowed in references, then why on Earth does WP:DATEFORMAT say that it is allowed at all?

By the way, you mentioned applying personal preferences. I have a preference for YYYY-MM-DD in references when creating articles (as allowed by MOS). But I never change an article's existing reference date format. I will revert when someone else bulk changes them in any direction. I will also change the occasional outlier reference to confirm with the majority of references on that page. But I don't change a page's reference date format merely to match my own preferences. Stepho talk 01:41, 24 April 2022 (UTC)[reply]

@Stepho-wrs: Like you, I actually think the YYYY-MM-DD date format is good for |accessdate= and |archivedate=. What I am saying is that it is extremely difficult at times to work with the multiplicity of formats that may adorn a given article. It's work that I do on a daily basis, but I often find it frustrating for that reason. You see, you sometimes have dmy, mdy and yyyymmdd formats all mixed up within citations in an article, (sometimes with no dominant) and we gnomes are left with the choice of either leaving the article with inconsistent formats (not good), or changing them "arbitrarily" into the format given by WP:DATETIES (not good). The arbitration between dmy and mdy is usually easy enough. It's when enter yyyymmdd that problems arise.

Sometimes, what at first appearance may be the "dominant" in the reference section is yyyymmdd because a bot or a johnny-come-lately editor has populated several citations with it over the first citation, which may originally have been a dmy or mdy citation, but this seems to have been your chosen path probably because it's easier to change one or two dates manually (even if it may have been the original) than to and change the majority as WP:RETAIN would have dictated (by the "first major contributor rule"). Unfortunately, we have the situation where we all agree to be tolerant of the diversity, and we are left with a rather unmanageable mess (not good); add to that, there are editors who go around reverting such attempts at harmonisation (not good). All this makes the job of harmonising date formats a bleeding complicated job. The only sure-fire was to avoid falling foul of WP:RETAIN is to investigate the placement of the very first citation, but it's an approach that few are willing to spend time doing, preferring a certain expediency or efficiency. -- Ohc ^{revolution of our times} 09:35, 24 April 2022 (UTC)[reply]

Support (the tech solution, not the choosing to have an argument over something instead part) _signed,Willondon (talk) 17:42, 21 April 2022 (UTC)[reply]
Support Levivich's "coffee maker" solution (see thread below[13]). We could even apply a similar approach to MOS:ERA. –Austronesier (talk) 14:59, 22 April 2022 (UTC)[reply]

Procedural comment: Please note that this discussion is part of a wider campaign against me by @Uanfala, which now extends to a total of 4 venues: my user-talk, ANI, this page, and WT:BOTPOL#Is_MEATBOT_not_relevant_any_more?. I have raised the issue at WP:ANI#Sneaky_forum-shopping_by_Uanafla. BrownHairedGirl (talk) • (contribs) 17:41, 21 April 2022 (UTC)[reply]
- It should also be noted that this discussion here hasn't deterred BrownHairedGirl from continuing to add the basic template to thousands of articles on a daily basis. – Uanfala (talk) 13:34, 22 April 2022 (UTC)[reply]
  @Uanfala: as requested in several venues by several other editors, please WP:DROPTHESTICK. BrownHairedGirl (talk) • (contribs) 14:35, 22 April 2022 (UTC)[reply]
@Uanfala:Placing of the {{use xxx dates}} template is not considered a non-consequential edit, notwithstanding the dispute. The original intention of the template is as a maintenance tag indicating that the article has undergone a date audit that would ensure consistency of all dates, but the new autoformatting allows most citation dates are rendered uniformly, which is one step in the right direction. -- Ohc ^{revolution of our times} 19:44, 22 April 2022 (UTC)[reply]
A big portion of the articles the template is being added on don't have any dates in them. And given the nature of the sources typically cited in some of those topic areas, dates aren't likely to start appearing in them in the future. The template doesn't do anything at all for these articles. It is consequential only to the extent that it takes up space in the wikicode (therefore creating clutter and occasionally confusing newbies), and that its removal (say, as part of sensible copyediting) will likely trigger its angry re-insertion by the people who watch over the tracking category. – Uanfala (talk) 22:30, 22 April 2022 (UTC)[reply]
Can you prove that "A big portion of the articles the template is being added on don't have any dates in them"? Whilst there are indeed a number of those articles without dates, it would still make sense to place the maintenance template on it if it has undergone a date audit, which is what I do. I usually don't need to do such work on scientific articles for that reason, so I select target articles carefully (OK I admit to the occasionally drive-by edit, but that's usually if I see a date that needs fixing). Maybe there are other editors who also do date audits, but to deliberately target articles not likely ever to have dates doesn't seem like efficient pratice. While Wikicode is abundant and often confusing, the problem lies with the nature of the beast – It's just how the MediaWiki software needs to be, so we shouldn't be blaming it on one or two maintenance templates placed at the top of an article – that act like referees in a game – that don't interfere with the play. And it certainly doesn't warrant systematic removal even as part of sensible copyediting because, as you say, it could provoke an edit war of sorts. Continuing to pick on editors who place tags you happen not to find appealing doesn't sound like very productive practice either. -- Ohc ^{revolution of our times} 06:44, 23 April 2022 (UTC)[reply]
I don't have statistics, but this was the overwhelming impression I got from the articles whose tagging washed up on my watchlist in the last week (no, it's not just British overseas territories). This is not a surprising statement. Articles won't need to mention specific dates in their prose unless they need to document events (or the lives of people). Such articles will often have dates in the references – either as publication dates for newspaper articles, or as access dates for web content. In many topic areas (culture and society, the sciences...), you may sometimes have such sources, but only temporarily, until you find something more reliable and academic (for books or papers you won't normally cite exact publication dates, and access dates are rendered unnecessary by the use of identifiers like dois.). – Uanfala (talk) 14:54, 23 April 2022 (UTC)[reply]
Of course there are types of articles that are not likely ever to have dates. It depends on how scholarly the article or topic is already, but article evolution isn't always linear or predictable; knowledge may also evolve, so the end point may not immediately be within one's sights. The mere fact that an article has, or is likely to have, a date at one point of its existence may be sufficient to warrant a tag, IMHO. Wikipedia isn't a crystal ball. Then, there's productivity, and efficiency for goming editors. But then, WP is a large community, with many hands making light work. Some work forwards, some work leftwards, some rightwards. I'm cool with any effort so long as it's not working backwards. -- Ohc ^{revolution of our times} 15:58, 23 April 2022 (UTC)[reply]

Uanafala claim to base their comments on the overwhelming impression I got from the articles whose tagging washed up on my watchlist in the last week.

But on my talk, Uanfala said[14] I only keep track of a few dozen England-related articles. I have added about 180,000 of these templates, so Uanfala's sample of less than 0.05% would be too small to be statistically relevant even if it was a random sample. But those articles are not a random sample; they are the topics which interest Uanfala.

So Uanfala's claim about a big portion is just an unevidenced declaration of a false assumption. BrownHairedGirl (talk) • (contribs) 06:20, 24 April 2022 (UTC)[reply]

@Ohconfucius: The claim by Uanfala that A big portion of the articles the template is being added on don't have any dates in them is false.

What happened was that one of my batches of articles was to tag was UK&Ireland-related stubs. Uanfala looked at my edits when I was tagging British Overseas Territories geography stubs, in which the Falklands set included a lot of South Georgia sub-stubs. Uanfala wrongly assumed that this small set was representative of the whole body of over 100k articles.

The basis of my tagging lists is very simple: articles relating predominantly to Ireland & the UK which have neither {{use dmy dates}} nor {{use mdy dates}}.

I aim to include all topic areas, but not all at once. I build my lists using a small set of topic areas at a time, because that it the most efficient way to avoid false positives. So at various points my lists have concentrated topics such as players of a particular sport, university-related lists, engineers, buildings of various types, crimes, statue law, writers, albums, aircraft crashes, ships, mountains, musical groups, plays, novels, political office-holders, streets, railway stations, football club seasons, military history of English counties, invasions of England, medieval Scotland, elections, and most recently the history of London.

The hard bit is excluding the false positives. Here are some examples (among many many hundreds) of the issues:
Irish biographical articles include a lot of emigrants to the USA; in most cases, those articles should use mdy dates

British military topics include conflicts outside the UK, foreign equipment, American military based in the UK

Categories for medieval and earlier topics, esp those related to Scotland+Ireland, include a lot of works by American writers, and some huge sets of speculative fiction franchises

I have a bundle of techniques for cleaning up my lists, some of which can be seen in the Petscan I used for the history of London batch: https://petscan.wmflabs.org/?psid=21939245 . Note that this batch is done, so the search now just returns a few false positives.

I have made no attempt whatsoever to determine whether an article currently has any dates, or whether it is likely to include any dates in the future. I simply assume that if expanded sufficiently, every article will probably eventually include one or more dates, either in body text or in a ref.

I would have been happy to explain this to Uanfala, but sadly their approach throughout has been to assume bad faith and make accusations, so our discussions consisted of me replying to the nonsense. If Uanfala had instead asked questions in a collaborative spirit they would have learnt a lot more. BrownHairedGirl (talk) • (contribs) 12:48, 23 April 2022 (UTC)[reply]

@BrownHairedGirl:This approach seems sound and mirrors my own when I build work lists. False positives are indeed a potential problem, and I have been known to go back and revert the change if/when I realise one of my edits applied the wrong tag; but military articles and conflicts tend to be dmy by a very large majority. The only significant way my approach differs fro yours is that I limit the article size to those in excess of 6kB, which allows me to avoid stubs. But it is my personal choice. Of course, it isn't reasonable to assume that stubs will remain stubs forever, as WP is a work in progress. -- Ohc ^{revolution of our times} 13:27, 23 April 2022 (UTC)[reply]
Thanks, @Ohconfucius. I also do some self-reverts.

My view on stubs is that per WP:STUB, the whole point of a stub is that is capable of expansion. (If not, it should probably be merged to another article). So I reckon that adding {{use dmy dates}} is a small help in its expansion. But no prob if you target your energies differently. BrownHairedGirl (talk) • (contribs) 13:46, 23 April 2022 (UTC)[reply]

Incorrect statement about YYYYMMDD

At 17:30, 21 April 2022 (UTC) Levivich stated

We should all be using machine-readable YYYYMMDD format in the wikitext, and then let the reader's client render that date in their local format. We have the technology to do this (we already do it for talk page comment timestamps), but we just don't do it in mainspace, because... well, because Wikipedia will always choose to have an argument over something instead of implementing a technological solution that makes everyone happy.

This statement is disproven at great length at Wikipedia:Mosnum/proposal on YYYY-MM-DD numerical dates. A specific example of why the statement is false is the YYYY-MM-DD format has widely been understood to only be for Gregorian dates, but many Wikipedia articles use the Julian calendar because the events described in the article took place when the Julian calendar was in force. Even people who think YYYY-MM-DD could be used for a Julian date will have to acknowledge there is no standard or convention on how a Julian YYYY-MM-DD date could be distinguished from a Gregorian YYYY-MM-DD date in a machine readable way. Jc3s5h (talk) 17:53, 21 April 2022 (UTC)[reply]

OK 1. that was 13 years ago. 2. We already do this for talk pages signature timestamps -- they can be rendered in the local time zone and in any date/time format. 3. The whole world, like every goddamn piece of software, already does this. Hell my coffee maker does this. 4. Julian dates? What? We should be using dates in.. you know.. the modern calendar system, and not an obsolete calendar system. Levivich 18:49, 21 April 2022 (UTC)[reply]

The link I gave was specifically limited to certain dates in citations. There is an older, but broader, discussion which rejected the automatic formatting of dates in articles at Wikipedia talk:Manual of Style (dates and numbers)/Archive D6# Again calling for date linking to be deprecated. As far as Julian dates go, don't worry about them. Just make sure you never enter any date earlier than 1 March 1923. To be safe, you should avoid reading articles about events that happened before 1 March 1923, or people born before that date. Jc3s5h (talk) 19:00, 21 April 2022 (UTC)[reply]

A computer can automatically show both old style and new style dates for historical dates. A computer can show you a date in any and every known calendar system, if that's what you want it to do. And this is a lark for so many reasons. First, with the exception of some British folks, the world doesn't generally change the dates for historical dates, despite changes in calendar system. For example, Bastille Day was July 14, 1790, still celebrated on July 14; Independence Day (United States) was July 4, 1776 (actually, July 2, but...), still celebrated on July 4, etc. Second, the date format is not the same thing as the calendar system. We can show Old Style (Julian) and New Style (Gregorian) dates in multiple date formats: e.g. 14 July 1790 [1 July 1790 O.S.] can also be shown as 1790-07-14 [1790-07-01 O.S.] or as July 14, 1790 [July 1, 1790 O.S.]. If we did as I suggest, and include just the YYYYMMDD format in the wikitext, for both Old Style and New Style dates, the computer can automatically render that as MMDDYYYY or DDMMYYYY or anything else, and we don't have to worry about "use MDY dates" at all. Levivich 19:21, 21 April 2022 (UTC)[reply]

By the way, there is already a gadget that does this for talk page timestamps (WP:LOCO, lulz), a user script (User:Mxn/CommentsInLocalTime), a template ({{date}}), and a phabricator task (T21992, from 2009... and as was pointed out in the thread, this has been doable since 2014). This is one of the many problems that are very easily fixable. Levivich 19:32, 21 April 2022 (UTC)[reply]

I'll reply to Austronesier's post in the previous thread because it is more germane to this thread. In the archived threads I referenced earlier in this thread, it came up that none of the automation could handle a passage like this:

The Metre Convention was signed on 20 May 1875; it was based on earlier informal cooperation among several nations.

If it were converted to MDY format in a mechanical way, without complex analysis of the context, it would probably come out like

The Metre Convention was signed on May 20, 1875,; it was based on earlier informal cooperation among several nations. (boldface for emphasis of error)

It's possible, with careful coding, to represent dates according to the preference of the reader when the computer program is writing the entire passage around the date. But no one has put forward code that can convert date formats properly in the midst of editor-written text. Jc3s5h (talk) 16:11, 22 April 2022 (UTC)[reply]

I agree with @Jc3s5h's caution. This is a lot more complex than it first appears. BrownHairedGirl (talk) • (contribs) 10:28, 23 April 2022 (UTC)[reply]

The servers of WMF are already under very heavy load with computing how each template of each article is rendered for the reader to give a global appearance for each article. Adding code that would convert date formats according to reader preferences would necessitate wrapping all dates within a template in the first instance, and would also significantly increase that server load. -- Ohc ^{revolution of our times} 11:31, 23 April 2022 (UTC)[reply]

Don't worry, the server doesn't make the calculation, the client does. Levivich 13:24, 23 April 2022 (UTC)[reply]

But no one has put forward code that can convert date formats properly in the midst of editor-written text. MediaWiki already does this. Go to preferences and change your date/time format to anything you want (MDY, DMY). You can also set your local time zone. Go to gadgets and there's one to show signature time stamps in local time or UTC. I believe we even have user scripts that'll do things like "XX hours ago". This technology exists and you're using it right now! We just don't have it turned on for mainspace articles. (And no, it doesn't create that error with the comma, lol.) Levivich 13:27, 23 April 2022 (UTC)[reply]

The timestamp is separated from the post it is associated with. It is not in the middle of a passage written by an editor. You do not read what I write. Jc3s5h (talk) 02:43, 24 April 2022 (UTC)[reply]

It doesn't matter if the time stamp is in the middle (13:45, 4 May 2022 (UTC)) or at the end, the computer can still format it in your local preference. The time stamp is just a string of text, it doesn't matter where the string appears. The technology to do this exists and we're already using it. Arguing there is some technical impediment is just false. Did you know we have a remote control helicopter flying on Mars? We can also format time stamps according to local preferences. If anyone doubts this, in desktop mode, go to Preferences->Gadgets and turn on the one for showing time in local time ("Change UTC-based times and dates, such as those used in signatures, to be relative to local time", aka WP:LOCO), and then come back here and look at whether MediaWiki changed the time stamp that I just put in the middle of this comment. Amazing isn't it! MediaWiki already does this! Levivich 13:45, 4 May 2022 (UTC)[reply]

The fact that my script can successfully convert dates from one format to and from another with few false positives and false negatives means that there is great certitude conversions can be done with even greater sophistication if it was coded into the MW software. -- Ohc ^{revolution of our times} 16:47, 4 May 2022 (UTC)[reply]

There is a vast difference between a script (which I too use sometimes while editing) that ***suggests*** changes to an editor, and software that changes date formats for readers who have no Wikimedia account and hence no ability to change their preferences. As an example of the inadequacy of Ohconfucius scripts for the latter purpose, consider a sentence.

On May 20,1875, the Metre Convention was signed.

The sentence contains commas either side of the year soley to set off the year, as is customary with mdy dates. When written in dmy format no comma is called for, but the script incorrectly leaves a comma.

On 20 May 1875, the Metre Convention was signed.

Jc3s5h (talk) 18:34, 4 May 2022 (UTC)[reply]

I agree that the script isn't perfect, but there's nothing inherently incorrect with the second sentence. A comma isn't absolutely necessary but it's not out of place. Although the script was written that way, I am pretty certain I can adjust the regex to remove the trailing comma in such sentences should it be necessary. But it's not, as I said. -- Ohc ^{revolution of our times} 18:51, 4 May 2022 (UTC)[reply]

How to keep an article in ISO format?

I'm with Levivich. YYYY-MM-DD is the ISO format, and it was selected as the international standard for good reasons; see ISO_8601#General principles for some of them. I recently started an article which mainly cited German sources using this format, and had someone (not checking who) come and tag it with a "use dmy" template. They also fixed errors in my citation formatting, which was nice, but the errors seeme dto be the citation templates not recognising the ISO-format dates provided by the sources thru Citoid, which is not good at all. As the ISO format was what both I and the sources preferred, should it have been left? There is no "Use ymd" template; what should I have done to keep the article in ISO format? I'm not actually interested in putting work into changing date formats, it is not the sort of task I like, but it would be nice if I didn't inflict such work on other people when it's utterly needless. HLHJ (talk) 21:58, 15 October 2022 (UTC)[reply]

The first question is which guideline applies. You seem to be interested in citations, rather than the rest of articles, so the appropriate guideline would be Wikipedia:Citing sources. But if you used citation templates, specifically those described at Help:Citation Style 1 (CS1) or Help:Citation Style 2 (CS2), those styles have adopted the date styles from Wikipedia:Manual of Style/Dates and numbers.

Unless an article adopts a citation style, other than CS1 or CS2, that adopts some special format for dates in citation, MOS:DATEUNIFY calls for publication dates to be in the format used in the article body text, or an abbreviated version. For example, if the article body dates are similar to "October 15, 2022", then acceptable publication dates in the citations could be "October 15, 2022", or "Oct 15, 2022". Dates in the YYYY-MM-DD format could be used for access dates or archive dates.

ISO_8601#General principles describes the general principles of how ISO 8601 dates are written, but does not explain the merits of that standard. In Wikipedia, ISO 8601 is unfit for general use because it forbids dates from being written in the Julian calendar. Many articles about the United Kingdom, Europe, and British colonies in North America contain such dates. Jc3s5h (talk) 23:28, 15 October 2022 (UTC)[reply]

There is no {{use ymd dates}} template because ymd is not allowed in prose and those date templates are meant to mainly cover prose. However, ymd dates in references (both pub date and access date) are allowed to be ymd, as long as the style is consistent among the article's references. We do this by {{use dmy dates |cs1-dates=y}} or {{use mdy dates |cs1-dates=y}}. See {{use dmy dates}} and {{use mdy dates}} for the full list of options. Since you created the article, your choice of date style for both prose and/or references is the one that other editors are suppose to respect, as per WP:DATERETAIN and WP:CITEVAR, unless a consensus has been formed on the talk page to change it. Stepho talk 23:57, 15 October 2022 (UTC)[reply]