Wikipedia talk:Manual of Style (dates and numbers)/Archive 90

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Archive 85 Archive 88 Archive 89 Archive 90 Archive 91 Archive 92 Archive 95

Inconsistent use of binary prefix by MediaWiki

Just recently noticed that MediaWiki is using the IEC prefixes: Search

Inconsistently, though... Image

Isn't that cute... -- mattb 23:49, 5 November 2007 (UTC)

Date linking wording

Currently, the section reads:

Full dates, and days and months, are normally autoformatted, by inserting double square-brackets, as for linking. This instructs the WikiMedia software to format the item according to the date preferences chosen by registered users. It works only for users who are registered, and for all others (i.e. most) will be displayed as entered.

This is misleading. Someone not familiar with the situation can read it as meaning that dates are autoformatted automatically, with no action by an editor required. Since this page gives guidance towards editors, it should be rephrased to do just that. My suggestion is:

For full dates, and days and months, autoformatting should be facilitated, by inserting double square-brackets when editing, similar to normal Wikipedia linking. This instructs the WikiMedia software to format the item according to the date preferences chosen by registered users. It works only for users who are registered, and for all others (i.e. most) will be displayed as entered.

The autoformat is not something which is enabled by the editor. Each user sets their own autoformat preferences. However, proper editing facilitates the autoformat, while omitting the square brackets means that autoformat will not work. Neier 12:34, 6 November 2007 (UTC)

I had no problem with the original wording. The first rewrite by Neier removed the preference for linked dates. I restored that with a modified wording. In my vocabulary, the linking "enables" autoformat, because without it, autoformat cannot work. A user can select a preference or not. The word "facilitate" is not fully correct, since it means "to make easier". So far no reliable method has been found to autoformat without hints from an editor. −Woodstone 12:51, 6 November 2007 (UTC)
How about: Full dates, and days and months, are autoformatted when double square-brackets are inserted by the editor, similar to normal Wikipedia linking. as the opening line? I definitely didn't intend to remove the preference for linked dates in the first revision, as I regularly correct articles that I notice which don't have the brackets. The current version still seems like it is too lenient, as it is possible to read it and think that no action is required when editing the article because the brackets are inserted by the software. We know that this is not the case, but, since the MoS is written as instructions to the editors, I think the ambiguity should be removed. Neier 13:07, 6 November 2007 (UTC)
That wording again does not state explicitly that linking is preferred. It just says that "if" linked, the preferences will work. The opening line should be more like:
"Full dates, and days and months, should preferably be prepared for autoformatting by inserting double square-brackets, similar to linking articles in Wikipedia."
Woodstone 15:44, 6 November 2007 (UTC)
Stating that it is done "for autoformatting" is really not needed when the next sentence explains that. I would suggest something like:
Full dates, and days and months, should be wrapped in double square-brackets ([[ ]]), similar to internal links. This instructs the MediaWiki software to format the date according to the date formatting preferences of registered users. For unregistered users, and registered users that have not set a date format preference, the date will be displayed as entered.
This also makes it more clear what the actual effect of auto-date-formatting is and tweaks some other wording. "Should generally" may be preferable to "should". Also, I am going to go ahead and correct "WikiMedia" to "MediaWiki". — Aluvus t/c 23:26, 6 November 2007 (UTC)
I'm OK with either Woodstone's or Aluvus's version, as they both solve the problem that I see. Neier 23:32, 6 November 2007 (UTC)
Sorry, "Full dates ... should be wrapped in double square-brackets" is not clear enough; putting double square-brackets around "February 14, 2007" does not work. The month-and-day pair needs to be linked separately from the year. Chris the speller 03:20, 7 November 2007 (UTC)
I agree that it is somewhat ambiguous, but I couldn't come up with anything better. Do you have any ideas? Merely stating that square brackets should be "inserted" is actually slightly less clear (suppose someone interpreted that as suggesting "Janu]]ary 5, 20]]07", or something similarly odd). The examples of exactly what to do follow shortly after, which substantially reduces any confusion that might result either way. The examples should probably be moved up right after the first paragraph in the section, now that I think about it. — Aluvus t/c 05:24, 7 November 2007 (UTC)
But, it is neither better nor worse in the ambiguity than the current version; so, that should not distract us from fixing the first problem first. The details for exactly how to place the brackets are spelled out later. Neier 00:06, 8 November 2007 (UTC)
  • As someone who actively discourages the use of the dysfunctional autoformatting system (see archives here for a fuller discussion), I'm uncomfortable with wording that implies that it "should" be used. That is why we came to the idea of using "is normally", as a compromise with a few editors who think it should be mandatory. Tony (talk) 05:14, 7 November 2007 (UTC)
  • Substituting "are normally" for "should be" in Aluvus's version should alleviate that concern; but, I guess I still disagree with the principle that until the software is modified (if ever) to untangle linkage from date prefs, that we should abstain from encouraging editors to try and make qualified dates (full dates, or day/month pairs) appear as the viewers would want them to. Neier 00:06, 8 November 2007 (UTC)
The silly entanglement of linking and autoformatting is one thing, but there are a host of other issues. See archives here. On this matter, I haven't received a satisfactory answer from anyone to my question of why the dates on our signatures are autoformatted without being a blue splotch. Surely it's just a matter of nominating a simple code (<< date >>, or the like) that can hook into that signature-date process? Tony (talk) 02:07, 8 November 2007 (UTC)
I think the signature dates are static, and don't follow the user prefs. My prefs are set for "month day, year", but, signatures on this entire talk page show up as "day month year". The code that writes to the database is separate from the code which formats when the page is viewed; if the dates are written as static text, then there is no formatting to occur. Neier 02:21, 8 November 2007 (UTC)
So it's because I've got day/month/year settings that I presumed it autoformatted them. Fancy that: I'd expect the US formatting to be used as a default, given the technical infrastructure is in the US. Tony (talk) 06:15, 8 November 2007 (UTC)

I share Tony's view. I too would discourage the use of the dysfunctional autoformatting system. In the continuum of forbidding to mandating, the mos is very close to mandating. I support any move back along the continuum away from mandating it. Lightmouse 08:52, 8 November 2007 (UTC)

Forgive me if I've missed a previous part of this discussion, but telling people to not link (and therefore not autoformat) dates has the first impression upon me of saying to European users "tough! you've got to use US date format" Is that the case or is the request to not link/format part of a campaign to get the WP programmers to actually come up with a solution? - X201 09:12, 8 November 2007 (UTC)
It's nothing to do with enforcing national date-formats; I'd rather put up with (within-article-consistent) US formats, which I dislike, than a system that spatters bright blue everywhere and is inflexible in ... four, is it? ... ways. Believe me, we've tried to get the WikiMedia developers to do something about it, but it's like trying to move a stationary ocean liner. Tony (talk) 09:58, 8 November 2007 (UTC)
I'm with you on the "bright blue" and I don't like every date linking to a "date x in history" type page. The code to format the dates obviously exists so it only needs to be disconnected from the linking side of things. What was the developers reason for not changing/fixing it? - X201 10:08, 8 November 2007 (UTC)
Despite its name, 'autoformatting' does not format automatically. It does not work for unregistered readers (there are lots of those). It does not work for registered readers that do not set a preference (such as myself). Thus many readers see whatever format the original editor used, whether US or non-US.
Like many good ideas, autoformatting is unsatisfactory because of basic design flaws and inadequate implementation. This brutal cure is worse than the disease. Lightmouse 10:34, 8 November 2007 (UTC)
Surely date formatting for both registered and unregistered users could be made available with a simple cookie? - X201 10:57, 8 November 2007 (UTC)

(Outdent) The response included:

Lack of substantive response means that none of us feels like writing the code, generally. If any of the 70 Wikipedians who were willing to spend a few seconds to sign a petition are willing and able to spend a few days or more writing the appropriate code and revising it in response to criticism, it might (but might not, admittedly) get approved. I'm not willing to commit (no pun intended) to review the patch, but another developer might be.

Gee, thanks. On the WP page that led to this push, there are now 85 signatories. I'm sure more would sign on now. But what's the use?

Full discourse is at

BTW, other problems with the system are that, for example, you can't slash ("The air-raids on the night of 30/31 May"), you can't use date ranges ("20–27 August"); there are others I can't recall right now. Tony (talk) 13:54, 8 November 2007 (UTC)

And the archives here. Tony (talk) 13:56, 8 November 2007 (UTC)

Here is a reminder: it can't cope with "5th November", "September 11th" and "4th of July". Currently, our editors are obliged to use "5 November", "September 11" and "4 July". Human formats are currently subservient to the machine formats. Lightmouse 18:51, 8 November 2007 (UTC)
But "5th November" is not a format that Wikipedia guidelines permit, so I don't see that last argument as pertaining to this discussion. I understand that Tony and Lightmouse dislike and personally discourage use of the autoformatting, but the consensus has been and still is to link dates that contain month and day. The vast majority of the dates in WP articles are not slashed overnighters or date ranges, and the autoformatting works well in most cases. Chris the speller 21:48, 8 November 2007 (UTC)
I agree. I notice that neither Tony nor Lightmouse have been editing long enough to recall the date format edit wars of the summer of 2003 which led to the present solution, but I have to strongly disagree that "This brutal cure is worse than the disease", and advocate that mandatory date-linking be encouraged notwithstanding any aesthetic disagreements over the blue links. -- Arwel (talk) 22:09, 8 November 2007 (UTC)
I disagree. Links are a universal internet mechanism for jumping to a different page that elaborates, or ties together, a concept on the current page. We are now developing the situation where a click on part of a date is leading to a page that has nothing to do with the page just left. For example, this DVD release section has a linked date of January 15, 2008 (among others). Clicking on 2008 leads to a page that has nothing to do with the television show in question. And, I can just imagine how long the edit that puts a DVD release on the January 15 page would last (in an attempt to tie the two pages together so as to reduce the user's confusion as to where they have actually landed themselves). The cure is worse than the disease, and is aesthetically displeasing. 00:16, 9 November 2007 (UTC)
That page also contains: [[September 13]], [[1974 in television|1974]]. Wikipedia is full of such errors due to the inability of users to understand autoformatting. If there were high speed bots cleaning up the multiple and messy side-effects of the cure, it would not be so bad. Furthermore, to clarify for Chris, Wikipedia bans the "5th November" format because human formats are not permitted to supercede autoformatting formats. Lightmouse 13:26, 9 November 2007 (UTC)
Lightmouse 13:26, 9 November 2007 (UTC)
  • I'd rather put up with foreign date formats, just as 99% of readers do, since they don't log in and thus have no autoformatting display. Just as long as they're consistent within each article. Or fix it technically! Not willing to wait for that, however, since I suspect it won't happen for years. Tony (talk) 13:33, 9 November 2007 (UTC)
    • I wouldn't typically have an opinion about this at all, and I'm late to the conversation anyways, but I'd like to note for the record that the links wouldn't be nearly so painful if the year pages weren't such a chaotic mess. --JayHenry (talk) 07:54, 29 November 2007 (UTC)
True, but to me, that's not the point. Tony (talk) 10:45, 29 November 2007 (UTC)

Spelling out numbers with scientific units

I think this item should be added to "exceptions" in the section "Spelling out numbers":

  • Numbers followed by an abbreviated unit of measurement are not spelled out, for example 8 MB rather than eight MB and 4 km rather than four km. However, "eight megabytes" and "four kilometers" are also correct.

This convention is actually used in the section about units of measurement, without being mentioned explicitly. Han-Kwang (t) 22:37, 8 November 2007 (UTC)

Sounds good to me. Tony (talk) 13:29, 9 November 2007 (UTC)
I added this to the page: Numbers followed by an abbreviated unit of measurement, such as km for kilometer, are not spelled out. Examples: 4 km, 8 MB, . However, if the unit is spelled out, then the number is also spelled out, for example: four kilometers, eight megabytes. In a less technical context, spelling out both the number and the unit may be preferred. I added the last phrase because it is natural to use '4 km', '7 m/s' in scientific articles, but it would look strange in other contexts. Han-Kwang (t) 12:42, 12 November 2007 (UTC)

Well Tony1, if you don't like the wording, how about commenting on the wording here? Han-Kwang (t) 11:33, 13 November 2007 (UTC)

In prose, the unit of measurement should always be spelled out, so this is only relevant to infoboxes, tables, parentheses, and the like, where the numbers are never spelled out anyway. —Centrxtalk • 04:40, 15 November 2007 (UTC)

In scientific texts it is standard practice to abbreviate units of measurements. You will never find this: "The current density was less than five milliamperes per square centimeter." Rather, it would be written as "The current density was less than 5 mA/cm2." Now within the context of Wikipedia it depends on the article. For a technical article, the conventional scientific style is natural (liquid helium, pressure), while a biographical article describing how the person lived two kilometers from school would spell out both the number and the unit. By the way, WP:MOSNUM#Unit symbols and abbreviations gives more correct examples as well. Han-Kwang (t) 13:07, 15 November 2007 (UTC)
But not if the number is larger: 12 kilometres, 120 kilometres, 12 km, 120 km. Tony (talk) 13:34, 15 November 2007 (UTC)

I don't know the conventions of how amendments to MoS pages are supposed to be done. I made a suggestion here, that I think makes reasonable sense, although the exact wording and its dealing with boundary cases might need to be polished. So my fellow Wikipedians, if you generally agree with my proposition, please add it (rephrased or not) to the MoS page. If not, say so, and leave the page as it is. But the ambiguous way the proposal has been dealt with so far looks like a waste of time for me, so I'm removing this page from my watchlist. Han-Kwang (t) 18:04, 18 November 2007 (UTC)

I can understand your frustration. It's an issue I raised long ago, when the MoS wording probably wasn't as infected with instruction creep as it is now.
NIST's Guide for the Use of the International System of Units (SI), 1995, NIST Special Publication 811 puts it quite succinctly in section 7.6:[1]
7.6 Symbols for numbers and units versus spelled-out names of numbers and units
This Guide takes the position that the key elements of a scientific or technical paper, particularly the results of measurements and the values of quantities that influence the measurements, should be presented in a way that is as independent of language as possible. This will allow the paper to be understood by as broad an audience as possible, including readers with limited knowledge of English. Thus, to promote the comprehension of quantitative information in general and its broad understandability in particular, values of quantities should be expressed in acceptable units using
  • the Arabic symbols for numbers, that is, the Arabic numerals, not the spelled-out names of the Arabic numerals; and
  • the symbols for the units, not the spelled-out names of the units.
Example:   the length of the laser is 5 m   but not:   the length of the laser is five meters
the sample was annealed at a temperature of 955 K for 12 h   but not:  the sample was annealed at a temperature of 955 kelvins for 12 hours
1. If the intended audience for a publication is unlikely to be familiar with a particular unit symbol, it should be defined when first used.
2. Because the use of the spelled-out name of an Arabic numeral with a unit symbol can cause confusion, such combinations must strictly be avoided. For example, one should never write "the length of the laser is five m."
3. Occasionally, a value is used in a descriptive or literary manner and it is fitting to use the spelled-out name of the unit rather than its symbol. Thus this Guide considers acceptable statements such as "the reading lamp was designed to take two 60-watt light bulbs," or "the rocket journeyed uneventfully across 380 000 kilometers of space," or "they bought a roll of 35-millimeter film for their camera."
4. The United States Government Printing Office Style Manual (Ref. [4], pp. 165-171) gives the rule that symbols for numbers are always to be used when one expresses (a) the value of a quantity in terms of a unit of measurement, (b) time (including dates), and (c) an amount of money. This publication should be consulted for the rules governing the choice between the use of symbols for numbers and the spelled-out names of numbers when numbers are dealt with in general.
[end of quote]
Note the citation to the GPO Style Manual rules as well.
The same rationale for the NIST rule, that they "should be presented in a way that is as independent of language as possible" is especially applicable to the English Wikipedia.
Han-Kwang was trying to make a point about scientific usage which the flip side of the coin of what NIST talks about in discussing use "in a descriptive or literary manner".
The measured quantities aren't really "whole" numbers. No measured quantity is ever exact, no measured quantity can ever be exact. That doesn't mean that a definition using units of measurement cannot be exact, but how well any measurement fits that definition is never exact. Part of the problem is the slight vagueness in what is meant by a "whole number". You and I probably have a understanding of whole numbers counting numbers and integers which is different from what many editors will think of when they see "whole numbers". Talking about "five sheep" involves a different kind of number than the one used when talking about "5 km".
Another part of the problem is, of course, overgeneralization and being prescriptive when it should be permissive. Han-Kwang's proposal suffered from that as much as what was here before. Han Kwang said, "However, if the unit is spelled out, then the number is also spelled out, for example: four kilometers, eight megabyte" but as Tony pointed out, "But not if the number is larger: 12 kilometres, 120 kilometres, 12 km, 120 km." Note also that while the NIST rule says not to use a spelled-out number with a unit symbol (don't use "five m"), it does not say that the converse is unacceptable: Arabic numerals may be used with the spelled out words in the cases where spelling the units out are acceptable (Tony's "120 kilometres"), something that NIST just did not bother addressing.
My advice to Han-Kwang is to do like everyone else does—argue the MoS when it agrees with you; ignore it when it does not. Fortunately, most of the people who argue for ever-increasing spelled-out numbers don't know beans about science anyway and pretty much stay away from the number-intensive scientific articles and from anything where the units involved are more complex than "miles per hour". If you run into someone insisting on applying these particular rules inappropriately, go look for help in one of the science Wikipedia:WikiProjects; don't come here. Gene Nygaard (talk) 01:02, 19 November 2007 (UTC)

This strikes me as WP:BEANS. The number of editors going around doing "eight km" are near zero; there is no consensus I'm aware of that "8 kilometers" is offensive to anyone who is not so obsessed with grammatical nitpicks that even the MOS regulars think they are insane. I.e., where is the actual problem? That said, I don't mind a one-liner addition to MOSNUM that "eight km" is off-limits; I just don't think it's really necessary. — SMcCandlish [talk] [cont] ‹(-¿-)› 15:24, 24 November 2007 (UTC)


Unresolved: Proposal still open (re-re-proposed replacement language)

This is merger of several related discussions and proposals. — SMcCandlish [talk] [cont] ‹(-¿-)› 00:07, 19 December 2007 (UTC)

Clarification about when to link to dates? (according to MOS)

I'm looking for some clarification about when dates should be linked. In the Autoformatting and linking section of the MOS, the last bullet reads:

"Wikipedia has articles on days of the year, years, decades, centuries and millennia. Link to one of these pages only if it is likely to deepen readers' understanding of a topic. Piped links to pages that are more focused on a topic are possible ([[1997 in South African sport|1997]]), but cannot be used in full dates, where they break the date-linking function."

Does the sentence in boldface (my emphasis) mean:
1. That dates (using any combiations of days, months, and year) should only be linked if the linked date will add to the reader's understanding.
2. That only full dates (i.e. November 1, 2007) should be linked, but not dates like November 2007.

Another Wiki user has been trying to tell me that the latter is the case, while I have been trying to indicate that the first meaning is correct. (The discussion can be viewed here: User_talk:NatureBoyMD.) Which is correct? -NatureBoyMD 21:26, 11 October 2007 (UTC)

The second is close to correct. Full dates, like November 1 2007, as well as just months and days, like October 14, should always be linked. This allows a user's date display preferences to work. TomTheHand 21:37, 11 October 2007 (UTC)
"Should always be linked"? No, the wording is "are normally linked", after a lot of tension several months ago about the dysfunctional state of the dateformatting script. See archives here for a discussion of the ?five key disadvantages in using it. As I've loudly trumpeted here and elsewhere, I now actively discourage the use of the autoformatting function, and will continue to do so until it's fixed.
Number 2 above, second point is a different issue: autoformatting applies only to full dates, not years alone and years and months. Please avoid linking those items unless there's a COMPELLING reason to do so. Tony (talk) 03:46, 14 October 2007 (UTC)
Agreed with Tony. Wikipedia is awash in pointlessly linked dates. Linking a date for no particular reason is no better than linking to Wiktionary for definitions of every single word in an article. — SMcCandlish [talk] [cont] ‹(-¿-)› 19:35, 14 October 2007 (UTC)

Proposal: Do not use surprise links

The albums project contains the following:

  • Quote: Do not use piped links to "years in music" e.g. [[1991 in music|1991]], instead add (see 1991 in music) where you feel it is appropriate.

I propose that we make that generic and include it. Lightmouse 11:09, 15 November 2007 (UTC)

It's a good idea, since most readers will assume that piped year-links are of the trivial type (which shouldn't be linked at all) and won't bother to follow them. Tony (talk) 11:43, 15 November 2007 (UTC)
Any suggestions for generic wording? Lightmouse (talk) 12:35, 18 November 2007 (UTC)
We could expand further; surprise links masked as single words are as bad as ones masked as years. But this should be a recommendation; [[1991 in music|1991]] can make perfect sense in a table, for example. Septentrionalis PMAnderson 06:32, 23 November 2007 (UTC)
It depends on the word that represents the pipe. The problem with years is that because they have been commonly linked but unpiped, readers tend to ignore the piped ones because they can't be bothered to check each blue-spattered year on the off-chance that it's piped to somewhere useful. The same does not generally apply to other words that are piped, although I concede that in some cases the principle is similar ("restaurant" --> "Chinese cuisine" in a recent FA, I seem to remember). Tony (talk) 12:07, 23 November 2007 (UTC)
If you would like to modify the scope that is fine by me, as long as this date problem is dealt with succinctly and strongly enough to have a real effect on editors. My proposal is to replace the following wording:
  • Piped links to pages that are more focused on a topic are possible ([[1997 in South African sport|1997]]), but cannot be used in full dates, where they break the date-linking function.
  • Do not pipe a link from "year" to "year <something>" or "<something> year" e.g. [[1991 in music|1991]]. Use "(see [[1991 in music]])" if appropriate. Note that piped links break the date-linking function if used in full dates.
Lightmouse (talk) 09:40, 23 November 2007 (UTC)
I have to strenuously disagree. While I'm aware that somewhere in one of these guidelines there is an old recommendation to never use so-called "surprise links", this advice, which I surmise dates to around 2003 or so, has been completely abandoned in actual practice by the community. E.g. there is no longer any consensus at all that there is anything wrong with "composed of silica and aloxite". Piped links exist for a reason. I don't see any rationale at all for enshrining a weird date-specific exception, and (Tony are you listening? ;-) I think it would be detrimental to our goals of ending the wikilinking of random bare-year dates, by removing any reader expectation that linked years every go anywhere vaguely useful. Rather, I think that we need to do the exact opposite and recommend not only that bare year dates never be linked as 1996, but also that the only time that such years should be linked is when the do go to a topically-relevant and -specific year article (1996 in baseball, or whatever). For WikiProject Sports I've been been formulating a guideline on when and when not to use such links (e.g.: "won third place in the 1996 World Championship" vs. "won the 1996 World Championshiop"). Agree that mentioning that piped links break the date-linking function if used in full dates is good to mention. Really, I think that proponents of this move simply do not frequently edit articles in which these sorts of links are useful (especially sports and entertainment stats tables come to mind, despite the music project's own advice). If I get shouted down on this, I might be able to compromise on their being permitted in tables, since I don't think they particularly are (or aren't) useful in general article prose, but do find them particularly useful in tables. Cf. WP:MOSFLAG - pretty much the same dividing line (except that the majority of editors, myself included, find flagicon use in main prose genuinely detrimental while sometimes useful in tables.) — SMcCandlish [talk] [cont] ‹(-¿-)› 15:19, 24 November 2007 (UTC)
PS: Don't mistake me for an overlinker of dates; see my comments at the end of #Clarification about when to link to dates? (according to MOS), above. I don't advocate that "Year in X" date linking be done very much, only when it is truly germane. — SMcCandlish [talk] [cont] ‹(-¿-)› 16:08, 24 November 2007 (UTC)
OK, I'm all ears; I hear you! Tony (talk) 07:11, 28 November 2007 (UTC)
I am also listening. My proposed wording was just a suggestion, I would be interested to see other suggestions. Lightmouse (talk) 20:40, 28 November 2007 (UTC)
SMcCandlish, I think your (compromise) proposal that they be permitted in tables is a good one. I disagree that the existence of piped links justifies these types of links in body copy, as I consider that too much of a "surprise". What if we changed the wording to discourage their use in copy (as above) but encourage their use in infobox and other tables where it makes sense? -- Renesis (talk) 00:49, 29 November 2007 (UTC)
OK, how about:
  • Do not pipe a link from "year" to "year <something>" or "<something> year" e.g. [[1991 in music|1991]] in copy text. Use "(see [[1991 in music]])" if appropriate. In tables, piped links are permitted if they make the table compact. Piped links must never be used in full dates because they break the date formatting function.
Lightmouse (talk) 09:11, 29 November 2007 (UTC)
What about links in football articles such as 1987–88? Woodym555 (talk) 11:33, 29 November 2007 (UTC)
Alt. version:
  • Avoid piping links from "year" to "year <something>" or "<something> year" (e.g., [[1991 in music|1991]]) in the main prose of an article. Use "(see [[1991 in music]])", if it is appropriate to link a year to such an article at all. In places where compact presetation is important (some tables, infoboxes and lists), piped links may be useful. Piped links must never be used in full dates (e.g. [[January 1]], [[1991 in music|1991]]) because they break the date-formatting function. Piped topical year links should only be made when the event in question is genuinely notable in the context of the year article to which would link.
SMcCandlish [talk] [cont] ‹(-¿-)› 11:54, 29 November 2007 (UTC)
  • I don't think my question regarding football dates such as 1987–88 has been answered. If we (Royal "we" of mere mortal WP:FOOTY editors) were to follow the new text, then we would have (See 1987-88 in English football) at the end of every other sentence. My recent WP:DASH sweep through Premier League as part of the FAR highlighted the fact that there are many season links in the article. All of them provide context that would be missed if they were to be "delinked". Woody (talk) 00:56, 12 December 2007 (UTC)
  • I think that would indeed be an undesirable result. The point seems to be that such links shouldn't be used at all the main article prose. I'm honestly not sure I can agree with that, so it looks like we are right back to the drawing board. Some are opposed to "surprise links", period, others like me and I think you see them as very useful ways of avoiding redundant clutter wording in ever other sentence. Others think they should only be used in tables/lists of tabular data. I accidentally supported that viewpoint, but don't actually share it. I do think that "surprise" links should be used in such cases, but not that those should be the only cases. Anyway, it strikes me that we don't have consensus on this one way or the other yet. The proposal, and the language in MOSNUM presently both do not adequately deal with the situations that arise. — SMcCandlish [talk] [cont] ‹(-¿-)› 20:48, 12 December 2007 (UTC)
You read my feelings perfectly. I am currently of the opinion that they do no harm. I don't think that consensus will be very forthcoming, given that peoples views vary differently. Trying to accomadate everyone's views leads to watered down text that actually helps no-one.


New version to address issue raised by Woody:

*Avoid piping links from "year" to "year something" or "something year" (e.g., [[1991 in music|1991]]) in the main prose of an article. Use an explicit cross-reference, e.g. ''(see [[1991 in music]])'', if it is appropriate to link a year to such an article at all. In places where compact presentation is important (some tables, infoboxes and lists), piped links may be useful. Another exception is the main prose of articles in which such links are used heavily, as if often the case with sportsperson biographies that link to numerous separate season articles, in which the ''(see ...)'' phrasing would rapidly become repetitive and cluttering; in such a case it is best to make it clear that the link is to such a topical date article, e.g. in [[2007/2008 snooker season|the 2007/2008 season]], rather than in [[2007/2008 snooker season|2007/2008]]. Piped links must never be used in full dates (e.g. [[January 1]], [[1991 in music|1991]]) because they break the date-formatting function. Topical date links should only be used when the event in question is genuinely notable in the context of the year (or month, etc.) article to which would link.

Revised version below.

I think this will adequately address the cases that need to be addressed so far, and would actually slightly help advance the growing consensus that a replacement for the date formatting function is needed so that dates are only linked when there is a contextual/informative reason for linking them.

PS: If we wanted to address piped links other than dates, I believe that is a way bigger matter, and should be discussed at WT:MOS, probably with an RfC, since it is bound to be highly controversial.

SMcCandlish [talk] [cont] ‹(-¿-)› 21:00, 12 December 2007 (UTC)

I agree with almost all of those points. That addresses all cases brought up so far, but this is a very insular forum for something like this. If the discussion were to be broadcast, I think you would find a wide variety of subjects where these links are useful. The date formatting function is a whole different kettle of fish. (metaphor about barge pole comes to mind). Also agree that "hidden links" is a topic that needs wider discussion. Your text seems fine for what it is, yet to me it seems complicated and too wordy. There are too many exceptions for it to be a useful editing guideline. (IMO of course). Woody (talk) 21:13, 12 December 2007 (UTC)
It can be made more readable with indented bullet points. The insularity is intended, because we're only trying to address over- and inappropriate date linking without opening a larger can of "hidden links" worms. I'm sure there are other cases where such date links arise, but isn't mentioning both music and sports good enough to get it across that we're speaking in generalities? I wouldn't want to see us list 15 "different" exceptions that aren't really different in a meaningful way. :-) Here's a revised version:
  • Avoid piping links from "year" to "year something" or "something year" (e.g., [[1991 in music|1991]]) in the main prose of an article in most cases. Use an explicit cross-reference, e.g. ''(see [[1991 in music]])'', if it is appropriate to link a year to such an article at all. Exceptions:
  • Piped links may be useful in places where compact presentation is important (some tables, infoboxes and lists).
  • Piped links may also be useful in the main prose of articles in which such links are used heavily, as is often the case with sportsperson biographies that link to numerous separate season articles.
Piped links must never be used in full dates (e.g. [[January 1]], [[1991 in music|1991]]) because they break the date-formatting function. When using piped links, it is best to clearly indicate that the link is to such a topical date article, e.g. in [[2007/2008 snooker season|the 2007/2008 season]], rather than in [[2007/2008 snooker season|2007/2008]]
A topical date link should only be used when the event in question is genuinely notable in the context of the year (or month, etc.) article to which it would link.
How's that? — SMcCandlish [talk] [cont] ‹(-¿-)› 00:44, 13 December 2007 (UTC)
Looks good to me. Lightmouse (talk) 13:20, 15 December 2007 (UTC)
As good as it is going to get I think. Well done. (The insular was referring to the limited number of people who look at, and comment on this particular forum.) Woody (talk) 13:41, 15 December 2007 (UTC)

I think input from more editors is required before removing "surprise links" from every article. It should be discussed at the village pump. --Pixelface (talk) 12:33, 16 December 2007 (UTC)

I would much rather see a piped link than have a random (see....) after some sentences. Piped links are extremely common on the English wikipedia, and (see...) is not. In my opinion, navigation should either be in a See Also section or through wikilinks embedded in the text (whether they are piped or not). Karanacs (talk) 22:31, 17 December 2007 (UTC)
To be honest, that is also my opinion. It is also why I have argued for the inclusion of "escape clauses" and some ambiguity in the text. The word heavy is always open to interpretation. Within the prose of an article "hidden links" are useful and help to keep the prose flowing. The links do provide context as well. Woody (talk) 22:45, 17 December 2007 (UTC)
Right. That is pretty much the largest point of this exercise. I know that some guideline somewhere, probably a MoS page (I can't actually find it yet, so it may have already been deleted), argues against any and all so-called "surprise links" generally, and it is my contention that it does not in fact represent consensus at all, since the utility of such links is obvious, as is the plain fact of their use in hundreds of thousands of WP articles with no sign of them being deprecated by the general editorship. The other point, of course, is to lay out some sane guidelines on usage of such links within the scope of WP:MOSNUM, at least with regard to years. We might later need to cover other cases, but for now I think that the years case is sufficient, per WP:CREEP and WP:BEANS (i.e., do not attempt to create a guideline about something on which the editing community doesn't genuinely need guidance.) — SMcCandlish [talk] [cont] ‹(-¿-)› 23:17, 18 December 2007 (UTC)
Yep, completely agree. In that vein, lets add the text in, barring any further objections. Woody (talk) 23:29, 18 December 2007 (UTC)
Pixelface, I honestly don't understand what you are getting at. There is no proposal to remove "surprise links" from every article. This proposal is actually quite opposite that, and sanctions the use of such links where they are genuinely useful and explains somewhat what the criteria of that usefulness might be. There is already, somewhere else that I seem unable to locate right now, a more general condemnation of piped links, which I contend is nonsense and hope to change into something more reasonable when I actually find it (and of course after another but more general discussion of this sort). — SMcCandlish [talk] [cont] ‹(-¿-)› 23:22, 18 December 2007 (UTC)


If you do this, then I think that, if a date appears within parentheses, it should automatically be replaced with "(see XXXX in music)" (or "video gaming" or whatever). SharkD (talk) 03:49, 17 December 2007 (UTC)

That would in way, way too many cases auto-violate the the principle that such "YEAR in TOPIC" links should only be made when the linked-from item is genuinely significant within the context of that year. For example: SOME_RANDOM_FILM_NAME (1993), on one hand, versus SOME_VERY_SIGNIFICANT_FILM_NAME took the 1993 "Best Picture" Oscar on the other. And that probably isn't even a good example, really. It is hard to come up with a generalized case for when one should make such links; it is much easier to say that most of them should simply be deleted, as blue-link "noise". — SMcCandlish [talk] [cont] ‹(-¿-)› 23:09, 18 December 2007 (UTC)

Any reason the remaining dispute tag can't be removed?

It's on "Autoformatting and linking". Can't see the point of the continued presence of the tag. Tony (talk) 08:34, 24 September 2007 (UTC)

The overall discussion topic still seems to be open on it, and someone even cross-listed the discuss at WT:MOS. Better to let it stand until that discussion settles out, I think.


Just for the record I want to call for any last discussion/debate that anyone feels is needed. The #Re-proposal above seems to not be truly objectionable to anyone, even if it not a bright line in the sand, if I may mix a few metaphors. A possible exception is SharkD, but I think the concerns raised by that editor are addressed (and I speak only for myself on that observation). I think that the re-proposal (as re-re-proposed; follow the boldface) does represent actual consensus on the usage, and also feel that consensus may change on the matter to make the line brighter, one way or another, some time in the future. It is better to have some advice on the issue than either no advice, or, worse yet, advice so obsolete that everyone ignores it (since the latter case simply weakens MoS as a whole - the more it is treated as optional or theoretical, the more wildly inconsistent WP articles will become). — SMcCandlish [talk] [cont] ‹(-¿-)› 00:10, 19 December 2007 (UTC)