Wikipedia talk:Manual of Style/Dates and numbers/Archive 80

another inflexibility in the autoformatting system

Discussion on this page appears to have uncovered yet another disadvantage of WikiMedia's date-formatting mechanism: it won't allow the final (parenthetical) comma after North American date formatting without forcing it on other formats (which do not want it). In other words, you can't autoformat "The October 5, 1999, stockmarket correction ...", without forcing the display of "The 5 October 1999, stockmarket correction", which non-Americans will object to.

Here's a list of the dysfunctional aspects of the autoformatting system.

• It's conflated with the linking function, thus forcing the intrusive blue colour and, depending on settings, underlining, for the display. This has all of the disadvantages of overlinking, such as the tendency to dilute the distinctive appearance of high-value links.
• It won't allow the North American final comma.
• It won't allow the economical rendering of date ranges, such as October 5–7, 1999, such as MOSNUM favours for percentage and unit ranges, instead forcing both dates to be spelled out in full hedgehog fashion (October 5, 1999 – October 7, 1999).
• It won't allow the date-slash, specifically mentioned as an option by MOSNUM (the night of May 30/31).
• It works only for Wikipedians who have selected a format (a tiny proportion of readers).

It's little wonder that I resisted the partial backflip on the relaxation of the mandatory use of autoformatting in the recent overhaul of MOSNUM. I want to put it to you that we should not be insisting that dates are "normally autoformatted". I, for one, actively encourage WPians not to use it until it's fixed. Tony 13:11, 16 August 2007 (UTC)

• This, however, seems sensible. Simply not autoformatting date ranges would avoid a lot of this; and unless May 30/31 is followed by a year, it wouldn't be autoformatted anyway. A Bugzilla report should get the autoformatting to deal with the comma, both producing it when necessary and absorbing it when the date is changed from American to British English; although the latter runs the risk that the final comma is marking both the year and the end of phrase. Septentrionalis PMAnderson 16:42, 16 August 2007 (UTC) Septentrionalis PMAnderson 16:39, 16 August 2007 (UTC)

• Saying nothing about the flatly ungrammatical "the attacks of September 11, 2001 are the" is contrary to the purpose of the MOS. The judgment that the automagic system is more important than American grammar is for editors to make; I think it is clear how I would make it, but others may differ. We should at least discuss the issue; silence encourages literate editors to add the comma, without realizing that there is a problem. Septentrionalis PMAnderson 17:55, 16 August 2007 (UTC)
  • What Chris removed was the statement of fact:

    In American English, the month-day-year style of dating ("The attacks of September 11, 2001, were among the most serious.") treats the year as parenthetical; as in the example, the year is divided off from the rest of the sentence by commas. If the year is immediately followed by other puncutation ("on September 11, 2001; the world..."), the trailing comma is omitted.

  Does he dispute this? Septentrionalis PMAnderson 18:36, 16 August 2007 (UTC)

  • I don't think any of us dispute it, but the problem remains that you can't autoformat and add the comma, because it will screw up the date in non-US formats. This, I suggest, is why the phenomenon of not inserting the commain the US format, contrary to what now appears to have been standard practice all along, emerged in the first place on wiki (non-US WPians, of course, come along and remove any comma because it looks odd in their display).

This is yet another reason, IMV, that we need to weaken the "normally" in the guideline for "Autoformatting and linking". Let me warn you that moving the developers over at Bugzilla is like trying to shift an ocean liner. We tried for a second time (with 84 signatures, as well written argument, and no objections) and failed WRT the relatively simple task of producing a parallel script (something like <<date>>, but whatever they felt like choosing) that would not link. I'm not a developer—nothing like it—but something tells me that they'll throw up their hands at being asked to solve all of the problems in the list above (well, the first four). I think the non-linking, the US comma, and the date-range issues are the priorities, in that order. The US comma is not easy to fix, because the function will have to distinguish between instances where there's another punctuation sign in place of the comma. Possible, but not straightforward.

I know that SMcCandlish is keen to make another push, and has hinted that this time we need to start higher up the food chain. Tony 23:38, 16 August 2007 (UTC)

I don't think that non-implementation will bother many American readers, anyway. The comma after the year is more commonly dropped than used. Askari Mark (Talk) 03:04, 17 August 2007 (UTC)

Anderson, months-days are explicitly enabled by autoformatting: May 30 responds to the order of the setting. The slashed date renders a broken link. Tony 06:43, 17 August 2007 (UTC)

Note that month-days are only autoformatted for some of the date settings: May 30 and 30 May do not generate the same thing with an ISO date setting. --PEJL 08:15, 17 August 2007 (UTC)

So get it fixed. — The Storm Surfer 17:14, 19 August 2007 (UTC)

I support any initiative to move off of the wikilink method of date formatting. Any support to the assertion that a "tiny proportion of readers" don't have user preferences set? (Or are we assuming that more visitors are just readers than editors?) Perhaps a developer would consider making trailing punctuation work the same as [[plural]]s pull in nonpunctuation; or maybe the punctuation mark can sit inside whatever enclosure a date uses. I disagree with the rationale that trailing commas don't "bother American readers". While that may be true, wrong is still wrong.—Twigboy 04:30, 25 August 2007 (UTC)

Precision?

Tony1 is revertwarring for the following text:

• The month-day-year style of dating ("The attacks of September 11, 2001, were among the most serious.") treats the year as parenthetical; commas are inserted before the year and, unless it is followed by another form of punctuation, after it.

to replace

The month-day-year style of dating ("The attacks of September 11, 2001, were among the most serious...") treats the year as parenthetical; it should be separated off by commas at both ends, unless already bounded by other punctuation.'

Tell me, what circumstance would give these sentences different meanings?

It may be possible to replace bounded with followed; but the clumsy repetition of it in the top text is no improvement. Septentrionalis PMAnderson 16:57, 16 August 2007 (UTC)

• Another unfortunate effort at mindless precision appears to have provoked this edit. The text was

• Yearless dates (5 March, March 5) are inappropriate unless the year is obvious from the context. If there would be any doubt, include the year.

As phrased, this would prohibit Guy Fawkes Day is celebrated on 5 November; and Kandolin attempted to clarify that this was not intended — as I trust it is not. The proper response is to recast, so it only prohibits ambiguity. Septentrionalis PMAnderson 17:10, 16 August 2007 (UTC)

In the second example, the antecedent of "it" is "style", which therefore makes the wording wrong. Using "it" a grand total of twice in one sentence is no crime, and is probably preferable to the awkward phrase "separated off". — Aluvus t/c 01:13, 17 August 2007 (UTC)

Modern times

Quite often in modern times or is now considered should not be replaced with "as of 2007". When they refer to the present state of scholarship, they can approximate "since 1950", "since 1895", or "since the Renaissance". To state that scholarship holds X as of 2007 implies that we have a source written, or at least published, this year; often we don't. We could require "since [time]", but that would be very difficult to source, and an invitation to original research. Septentrionalis PMAnderson 17:21, 16 August 2007 (UTC)

in modern times is just too vague. Remove now from the second one and it might be OK in some contexts. Tony 06:25, 17 August 2007 (UTC)

If you wish to include some other discussion of them, do so; but they don't belong where they were. Septentrionalis PMAnderson 15:36, 17 August 2007 (UTC)

Yearless dates

The recent (reverted) changes in the "yearless dates" subsection concerned the referencing of holidays. Under the prevailing wording, the following sentence violates the guideline:

• January 1 is New Year's Day.

Since annual holidays do not predicate on any particular year, why not make an exception for it? —Kanodin 18:52, 16 August 2007 (UTC)

Reworded to Yearless dates (5 March, March 5) can be ambiguous. Either ensure that the context makes obvious what year is meant, or include the year.

I think this removes the implication that the sentence violates MOS; do we need more? Septentrionalis PMAnderson 19:10, 16 August 2007 (UTC)

I think the Jan 1 example still violates the MOS. The revised wording requires an editor to include a definite year, like January 1, 2007, or to ensure that January 1 refers to a definite year by virtue of context. A holiday implies no year and there is no yearly context (except historically). It happens every year; implying a year or writing a year would be inappropriate.

Simple alternative:

* Yearless dates (5 March, March 5) can be ambiguous. Either ensure that the context makes obvious what year is meant (if any), or include the year.

Complex alternative:

* Yearless dates (5 March, March 5) can be ambiguous. If the date refers to a recurring event, then specify how the date recurs. Otherwise, either ensure that the context makes obvious what year is meant, or include the year.

This sounds anal, but you know someone will hold other editors to the letter of the MOS, regardless of ignore all rules. —Kanodin 06:08, 17 August 2007 (UTC)

It says can be ambiguous. The wording is fine, except that "ambiguous" is wrong; ambi- means both, so "unclear" would be better. Tony 06:24, 17 August 2007 (UTC)

We are writing English, not Latin. The OED definition is Admitting more than one interpretation, or explanation; of double meaning, or of several possible meanings; equivocal. This explicitly permits "several possible meanings", and English usage clearly makes no such distinction. Since we use disambiguation freely, and as a quasi-technical term (see any disambiguation page), ambiguous is actually better here.

I suggest, and will include: Note: This caution does not apply to describing recurring events, such as "January 1 is New Year's Day", as there is no ambiguity to be resolved. It would be artful to quote our usage below: "dates other than 1 January were used as the start of the year"; but this is probably too long. Septentrionalis PMAnderson 15:02, 17 August 2007 (UTC)

No, you won't include it, or I'll revert. It's bloat, the whole thing. Why does can not provide sufficient lattitude for editors to work out for themselves that "January 1 is NYD" doesn't need a year attached? It's just too silly. Tony 15:33, 17 August 2007 (UTC)

I wrote can; but Kanodin wants more. Kanodin's complaint, which he has made several times, is clearly a genuine worry. He may be right to worry; I have seen enough semiliterate uses of the MOS by the half-educated not to include this pre-emptively, since he asks. Septentrionalis PMAnderson 15:40, 17 August 2007 (UTC)

• The principle will need to be stated first; then the example; that was one reason I reverted. But I'm unconvinced that your "can" isn't sufficient to solve the problem. Tony 16:20, 17 August 2007 (UTC)
  • When they consistitute a single sentence? Harmless, I suppose, but that is "silly". Septentrionalis PMAnderson 17:44, 17 August 2007 (UTC)

Surely the "(if any)" revision is small enough? The verbose versions are admittedly more cumbersome, and I would like to keep things simple. The ambiguity sentence does not inform the reader what needs clarity. The sentence following it, in prescribing how to remedy the ambiguity, mandates one of two options: (1) implying a specific year, or (2) writing a particular year. The "(if any)" qualification allows an editor to imply that a yearless date refers to no particular year. —Kanodin 19:09, 17 August 2007 (UTC)

I considered (if any). The problem with only can is, if I understand you correctly, that it doesn't make clear in which cases yearless dates should be left alone, but forces the reader to deduce that we mean holidays and other recurrent events; if we are content to trust that every reader will follow this implication, why add anything? I think (if any) has the same problem. Septentrionalis PMAnderson 20:08, 18 August 2007 (UTC)

I like the current wording:

• Yearless dates (5 March, March 5) can be ambiguous. Include the year if the meaning is unclear. There is no such ambiguity with recurring events, such as "January 1 is New Year's Day".

—Kanodin 07:40, 19 August 2007 (UTC)

Anderson, please cooperate and collaborate

There are too many reverts happening, because you act unilaterally, without talking things through here first. It's not a good look for the page. None of us owns MOSNUM; while there's often a bit of push and pull, when I see seven or eight of your edits in a row, some of them changing policy, I draw breath. Can you please cooperate? Tony 23:45, 16 August 2007 (UTC)

There are too many reverts happening because you revert even patently correct statements of English grammar, such as Michael Hardy's, below. Please stop being disruptive. Septentrionalis PMAnderson 14:46, 17 August 2007 (UTC)

It's a serious problem when you try to deflect such a critical comment. Tony 15:34, 17 August 2007 (UTC)

It's a fact that August has 31 days, that dates higher than 31 are incorrect, make WP look unprofessional, and should not be used, and that August 491 is laughably incorrect. But the manual should not have instructions added except by consensus, so it would be wrong for me to take it upon myself to insert those observations into the manual. And their being correct does not make them immune from reversion. Chris the speller 16:40, 17 August 2007 (UTC)