Wikipedia:Date linking request for comment

It is important that to get as many views as possible on the format of a request for comment regarding date linking articles, we get individual views about the process. In your own section, please state your thoughts on how the request for comment should work. A few examples of considerations to make are as follows;

What issues should the RfC deal with? This includes both scope and the detail the process will go into.
What format should the RfC take? Should it ask for outside opinions, or should we develop a set of questions/proposals and put it to a vote?
How should each individual point be put across? i.e. should we list pros and cons of each proposal or should we leave it up to the community to come to their own conclusions?

Please only edit your own section and discussion can take place on the talk page.

View by User:Arthur Rubin[edit]

(copied with some modifications, from my comments on the Arbitration talk page)

This is just an issue list, not dealing with the degree of detail yet. More detail will be needed, and format and presentation is not yet covered.

Autoformatting...

When should dates be autoformatted?
1. Under the current system <seems to be a general consensus that they shouldn't be autoformatted under the current system, but the definition of autoformatting in RFC1#1 and RFC2#1 is unclear.
2. Under the proposed system
  - What should the proposed system be?
  - How should autoformatted dates be tagged as linked/not linked (opposite to the default)
  - When should autoformatted dates be linked?
If dates are/are-not autoformatted according to the above rules, what tools (bots, semi-automated edits which are essentially bots, semi-automated edits which are not essentially bots, etc.) are allowed to correct the issue?

Date fragment links

When should date fragments be linked (DOW (Day of Week), DOY (day of year, such as January 1, month, year, decade, century, millennium, etc.) I tend to agree with DaBomb and Kendrick about the general classification. In order of increasing likelihood of linking, and my opinion as to how often they should be linked:
1. DOW, month (hardly ever linked)
2. DOY (primarily in timeline articles, with rare exceptions)
3. decade, century, millennium (mostly (decade) in (decade in X), etc., but articles about an era which begins or ends in that time interval might be appropriate)
4. year (there is some debate over specific circumstances in which individual years should be linked)
- Perhaps we need a separate positive/negative consensus; in some cases, the MOS should neither make a recommendation for or against linking, but leaving it up to editorial judgment on the specific article or group of articles.
To what extent can a project consensus override the general consensus for or against linking?
If date fragments are / are not linked properly according to the above rules, what tools (bots, etc.) are allowed to correct the issue?
- Does this depend on the strength of consensus as to whether the date fragment should be linked?

In regard the tools, the question is whether a cursory glance (which is all that even a semi-automated edit can provide) or an algorithm (bot) can determine whether the date or date fragment should/should not be linked; and if not, should an error-prone tool be allowed — and, by tool, I include editors who mechanically delink or mechanically revert delinking without thought. — Arthur Rubin (talk) 21:00, 19 February 2009 (UTC)[reply]

reply to User:Kotniski

There were two previous RfCs, but there is little agreement, and certainly no consensus, as to what the results were. We need a carefully constructed RfC so that there can be general agreement as to the presence or absence of consensus. — Arthur Rubin (talk) 15:23, 25 February 2009 (UTC)[reply]

Only the detailed RfC had sufficient, well detail, to determine whether there was consensus for any specific guideline changes, interpretations, or suggested actions, and it doesn't seem definitive, except that:

Use of the current method of autoformatting is discouraged, but I don't see consensus that it's always inappropriate.
There is consensus that some method of autoformatting is appropriate. (Primarily because the "no" !votes didn't give reasons.)
There is consensus that some years, but significantly fewer than are currently linked, should be linked. For day-of-year links, there isn't even consensus that some should be linked.

There is consensus that if there were a bot-detectable method of determining whether a date fragment link should be there, then bots could be used to remove inappropriate ones (although there's a caveat which suggests it should emit non-printing date-indicators so that they could be reversed if later found appropriate.)

None of what I see as consensus will help resolve the disputes. If Arbcomm can find a more specific consensus, and if they consider finding that consensus within their jurisdiction, then this RfC would be unnecessary. (It's already been established that ArbComm does not make policy.) If they cannot, it's important to get the outline of the RfC fleshed out so that the RfC can be done and interpreted without reopening the RfAr. — Arthur Rubin (talk) 16:09, 25 February 2009 (UTC)[reply]

View by User:RexxS[edit]

Date linking[edit]

Here, I am not commenting on the issue of autoformatting, but solely on the issue of date linking; that is, linking to years (such as 1809), day+months (such as 12 February) or full dates (such as 12 February 1809). I refer to these as date-links.

Towards the end of the Nov-Dec 2008 RfC's I commented on the question of relevancy. I still believe that the necessary first step is to determine the consensus on the following question:

Is there sufficient value gained by allowing date-links to be an exception to the rule of relevancy?

I believe strongly in the WP:MOSLINK statement: "Internal links add to the cohesion and utility of Wikipedia by allowing readers to deepen their understanding of a topic by conveniently accessing other articles." and hope that that could be assumed common ground before starting any RfC. That is what I've thought of as the principle of 'relevancy' and I suggest that the underlying dispute here rests on whether or not particular categories of date-links provide some value other than direct relevancy to the article. Please remember that our articles such as 18 May are generally no more than "A list of events that happened on 18 May", so I contended, as an example, that the date 18 May is relevant to the article First Hague Conference, but the article 18 May, unfortunately, is not.

A similar argument leads to the question on year links. I do agree, however, that some year articles could qualify as meeting the relevancy principle (1942 in a WWII article is often quoted as an example). Nevertheless the vast majority of year articles could not meet that criterion when considering making a link to them.

So, we need to determine if there is consensus for making date-links when the date article is not directly relevant to the article where the editor is making the link. Some of the reasons may be:

Eventually our date articles will be rich enough in content to provide relevant background;
It costs little to make a hyperlink and it's easier for a user to click it than use the search box;
One of the joys of a paper encyclopedia is that while looking for a particular topic, you often find something else unrelated, but interesting, because it is adjacent alphabetically. The date articles here also provide another kind of adjacency - other topics that share the same year or "birthday".

There are probably others and I would expect an RfC to explore that. In the process, I believe we would see what the consensus is for any exceptions to the relevancy principle and what those exceptions are.

Of course, I also recognise there is a position stating "it is impossible to give guidance on date-links generally, and each one has to be an individual editor's decision", but personally reject that, since I believe that the vast majority of editors prefer some guidance, whilst allowing exceptions when it's in the best interest of the encyclopedia. An RfC should probably test whether there exists a consensus for that position or for mine. --RexxS (talk) 01:02, 21 February 2009 (UTC)[reply]

Date autoformatting[edit]

I believe that date autoformatting is a solution in search of a problem, as WP:ENGVAR deals with identical formatting issues without using markup. Nevertheless, the question "Is some form of date autoformatting desirable" ought to be asked in some way to ensure that the community can reach a consensus.

Personally, I would also seek consensus on the question "Should date autoformatting use markup that could be confused with current wiki-markup?" as well, since the principal problem caused by the current deprecated DA markup is that it uses the wiki-markup for links. If DA is ever to be accepted and useful, I propose that it must use markup that is different from, and can be freely mixed with, wiki-markup for bold, italic, link, template, etc. Given a free choice, I'd suggest something like ##10 March 2009## similar to that used in MS Office to delineate dates. --RexxS (talk) 00:58, 10 March 2009 (UTC)[reply]

View by User:Bzuk[edit]

In respect to datelinking, if there is questionable or limited value in linking dates indiscriminately to a list of events extant for a day, month or year, then it provides no relevance to the user. I would support only linking a date to a specific grouping, such as "year in film" listing but can see no valid reason to link dates for the sake of linking them to generalized lists. FWIW Bzuk (talk) 02:32, 24 February 2009 (UTC).[reply]

View by Masem[edit]

An interesting point has come up throughout the discussion before and during the ArbCom case that might be helpful to focus on, and this is, what is the ultimate goal of the "almanac" pages? Should they remain as, as some see it "random trivia lists", should they be improved to be more prose-y and comprehensive, should we be linking to almanac pages relative to the field of the linked topic, with linkages to larger concepts on those dates/years? Which leads to the question, should there be desire to improve the use of these articles, is it worth it to retain links in the interim to avoid having to create them? Also, are there any potential uses in the metadata of linking dates that can be used by other services through the WP API to reconstruct the data somehow? --MASEM (t) 22:58, 24 February 2009 (UTC)[reply]

View by User:Ohconfucius[edit]

First off, I am slightly confused that there is this new RfC: how does this relate to the earlier one started in RP's userspace (User talk:Ryan Postlethwaite/Date linking RfC)?

I have expressed my desire to see a neutrally and professionally constructed RfC which eliminates the partisan bickering and bias which has crept into the 3 RfCs already conducted on the subject. I would also like to see questions which do not leave too many significant open ends which open the door to further dispute and arguments. To that end, I have found names of some wikipedians who are likely to have those skills. While we should agree on what questions to ask, we should leave it as much as possible to the professionals to deal with 'how' they are asked.

Date-autoformatting, as we know it, is deprecated – nearly dead (although no-one has pulled the plug on the respirator yet). Let's not get ahead of ourselves, and let's start with the fundamental question: Is any sort of autoformatting desirable? Only if that question is answered can we decide

what form that 'son of date-autoformatting' should take. Development must not be haphazard, and proper project management should be put in place, with consensus sought at appropriate junctures of the project (pre-determined prior to commencement). The first such pre-initiation consensus must be one on the scope of software, user (as in reader) requirements and specifications.

View by User:Kotniski[edit]

Rarely am I moved to swear on Wikipedia, but FFS, how many RfCs on the same topic do we need? We've had two; they produced very clear answers, and no new one is going to produce anything clearer. It is quite bizarre that instead of letting people get on with putting the community's decisions into practice, people are trying to keep the dead horse on life-support. The hope, no doubt, is that this time most people will be so bored that they won't be bothered to participate, so those who have climbed the WP:REICHSTAG on this issue will be able to claim a lack of consensus and continue preventing people from implementing the decisions we've made. ArbCom members ought to be perceptive enough to see through this sort of thing. --Kotniski (talk) 10:27, 25 February 2009 (UTC)[reply]

I endorse Kotniski's comment. Tony (talk) 14:47, 25 February 2009 (UTC)[reply]

Reply to Arthur Rubin: you continue this question begging, insisting on consensus about consensus. Editors ultimately don't need to agree on what the consensus is - if there are those who insist on disputing what it is, then ArbCom is surely here to decide that matter. Why do you think there will be any more agreement about the meaning of the results of another RfC? All it will prove is that most editors are tired of being asked the same questions and will probably give increasingly irritated and less-thought-out replies, or just no replies at all.--Kotniski (talk) 15:56, 25 February 2009 (UTC)[reply]

View by User:Deb[edit]

I'd like to turn one of the above questions around and see this question addressed:

Is there sufficient value gained by having a rule of relevancy to reject individual contributors' and users' preference for retaining date links?

In other words, is it appropriate to have a "rule" when opinion is clearly so sharply divided on the issue? Can't we just all agree to differ? Deb (talk) 18:57, 25 February 2009 (UTC)[reply]

View by User Kendrick7[edit]

Per WP:KISS I would propose the following should be the only relevant question under the heading of "Date linking RfC"

1) As a matter of style, should we advise that calendar dates (e.g. 1 January) be linked only for articles concerning holidays, holy days, and dates of commemoration of the subject of the article?

Although I agree that this question has been answered already to everyone's satisfaction. -- Kendrick7^talk 01:14, 26 February 2009 (UTC)[reply]

View by User:LiniShu[edit]

My suggestions below stem both from:

My viewpoint previously expressed in this diff [1] that some of the "date linking/delinking" conflict is due to tensions between pairs of principles or objectives each good in themselves, which may be held in opposition to each other.
My personal concerns as an individual editor.

What issues should the RfC deal with? I (like RexxS) am not commenting on the issue of date autoformatting, but rather on the issue of date (i.e. chronological element) linking as separate from autoformatting. I am not attempting to frame the concerns in terms of RFC proposals (which is not something that I have any experience with); I am hoping that if the ideas have merit then others will be able to adapt them into appropriate proposals.

My most pressing concern is with "the relevancy question"; the statement from WP:MOSLINK: "Internal links add to the cohesion and utility of Wikipedia by allowing readers to deepen their understanding of a topic by conveniently accessing other articles." I am concerned that this understanding of the purpose of hyperlinks in articles narrows the function of links to be uni-directional; requiring that the utility of the link must be for the enhancement of the article, with a lack of focus on the possibilities of the link as a gateway to an expanded context; i.e. a focus on the "building of the web" (hyperlinks as creating encyclopedia infrastructure.)

Even within the concept of a relevancy test, I see further narrowing of application of the term "relevancy" with respect to date links (chronological items):

(a) For a link to a "bare year" article to be considered relevant, every event (or nearly every event) in the list must be relevant.

(b) There have been differences of opinion among editors whether a link to a particular "bare year" article was relevant to an article from which it was delinked (I am thinking here, for example, of the historical articles that Hex and Kendrick7 work on); For a type of link to pass a relevancy test, do only some editors need to find it relevant or does every editor need to agree that it is relevant?

Other topics from MOS and/or recent RFC's regarding "date linking" which cause concern to me specifically:

That MOS could proscribe piped linking to "Year in" articles, either requiring an awkward interruption of the flow of the text or relegating such links to a "See Also" section.
That MOS proscribes the use of links to "Year in" articles in lists of works.

What format should the RfC take? Should it ask for outside opinions, or should we develop a set of questions/proposals and put it to a vote? If we truly are faced with choices between focus on good but opposing ideals (e.g. consistency between, and polish of, individual articles vs. building up the infrastructure with bi-directional linking), then it seems to me that the decision has to be made by a true community consensus regarding which of the two "goods" is valued most. It could be that it is a case of personal preference, and that it would have to come down to a vote. I personally, would "bow" to a vote resulting in the defeat of my preferred position, given that a neutral party determined the following: that the questions/proposals were framed to clearly show which options participants are choosing between (therefore I believe that we should have pros and cons of the proposals listed); that a significant number of people participated; that there was a clear majority for the "winning side", or, if no clear majority for either side, a neutral party determined a "default position" for how to proceed to settle the questions in conflict.)

As a closing comment, I'd say that I thought RexxS presented above a commendable statement of his belief in the relevancy principle, along with a very fair view of suggested possible exceptions to it for which there could be consensus, and, although I think he and I stand on different sides of the question regarding date linking, I think we may be in agreement about the types of questions that we hope this RFC will demonstrate consensus on.

Thanks, Lini (talk) 05:52, 26 February 2009 (UTC)[reply]

View by User Gerry Ashton[edit]

It is accepted in previous RfCs that the current system of autoformatting is depricated. Questions related to this that some still dispute are

Is it fair to suppose that most linked dates are associated with the current autoformatting system, and it is appropriate to use fully automated means to remove them?
How can legitimate date links be protected from fully automated date delinking bots?
Is there a desire to pursue a future system that provides preference-based formatting for logged-in users and, for any given article, provides a fixed presentation of marked-up dates to non-logged-in readers?
Is there a desire to pursue a future system, the technical feasibility of which is uncertain, that provides a preference-based formatting for all users, and if non-logged-in users cannot register a preference, provides formatting based on the user's browser language setting?
If the answer to number 3 or 4 turns out to be yes, should the existing markup of dates for the current system of date autoformatting be preserved, despite the overlinking thus created, in the hope that one of the future date formatting systems will actually be implemented?

View by User Sapphic[edit]

As long as there's some kind of markup around dates (we currently use square brackets like with categories and images, but it could be anything) we can build an autoformatting/autolinking system (to replace the current one) that makes all the other issues irrelevant. Or at least of much less importance. There's less reason to have disputes if everybody can just set their own preference and see things how they want to be: all dates linked vs. some vs. none at all, all dates reformatted vs. per-page formatting vs. no reformatting, etc. Some people would still argue over what the defaults should be for anons (who can't set individual preferences) but that'll be a lot less people and (hopefully) a lot less impassioned "debate" on MOSNUM and elsewhere. I'm willing to argue until I'm blue in the face to keep my current ability to view dates in the format I want and (until recently) mostly all linked, but it doesn't bother me as much if that's not the default for anons.

We already have a demo system that was the built as the result of community feedback and UC Bill's programming efforts, and it does almost everything everybody asked for. The rest shouldn't be hard to develop with more community effort, which can hopefully take place when (and if) the RFC participants decide that we should keep some form of autformatting/autolinking system for dates.

So maybe we should have a preliminary RFC to ask about date autoformatting/autolinking, since if that's what people want, the other issues (when to link dates or not, determining relevancy, etc.) aren't as big a deal and probably don't even need an RFC. Similarly, if people reject a software solution outright, then the proponents of that approach will — hopefully (I know I would..) — accept that it's not going to happen, and focus our energies elsewhere. There still might need to be a second RFC in the latter case (since there's much dissension amongst the opponents of a software solution) but it would be significantly less complicated since it could assume there'd be no autoformatting/autolinking to worry about. --Sapphic (talk) 02:28, 6 March 2009 (UTC)[reply]

View by User JonH[edit]

The RfC should have a clear option for date markup that can do something different to what the current markup does. In the discussion about date linking, I have seen hints about this, but no real debate.

My view is that the current system is unhelpful (ever since I saw something like "the facilty was opened on May 6, 1972" and I was disappointed that clicking on the link did not give a detailed account of the opening ceremony) but I can imagine a system in which a page, popup or tooltip gives potentially relevant information such as the day of the week, the phase of moon, the date in different calendars, etc, etc.

Wikipedia already has two examples of special markup that lead to more information: ISBN and {{coord}}. As well as the links created in the articles, Google Maps and Multimap are able to use the {{coord}} markup to provide links to Wikipedia articles from their maps. In the same way, third parties could use date markup to create links from a calendar system.

{{coord}} has taken years to get to its current state (and there are still some problems to fix), so it is likely to take some time for a new date markup system to be developed. But it would be a waste for bots to go on the rampage changing "[[May 6]], [[1972]]" to "May 6, 1972" (thus losing the semantic knowledge) when they could instead change it to something like "{{date|May 6, 1972}}". For the time being, the {{date}} template could just generate "May 6, 1972". Later a more useful system could be agreed on and implemented. I am assuming that the {{date]} template would accept all the formats understood by the current autoformatting system, and that human editors could position a trailing comma inside or outside the template as appropriate.

JonH (talk) 16:01, 8 March 2009 (UTC)[reply]

View by Locke Cole[edit]

Any discussion of "Dynamic Dates" (also called auto formatting) should be in general, not about a specific implementation (except, of course, MediaWiki's current system which is broken in some regards). The RFC should instead propose a "fixed" version (not necessarily one written by anyone in particular) that resolves the issues with the current system and providing some additional enhancements to address other concerns. In particular, a "fixed" Dynamic Dates system should consist of (in no particular order, and likely incomplete):

Consistent date output for anonymous users, including the ability for anonymous users to choose their own date format.
Dates and years to not be automatically linked, but rather to be optionally linked via a setting (perhaps even for anonymous users).
Robust error handling for malformed dates or date fragments (could conceivably handle [[2nd November]] [[2008]], for example, or a full date, [[2 November 2008]]).
Better date range handling.
Support for date (sans year) formatting: [[2 November]] should auto format to 11-02 if date preference is set to YYYY-MM-DD.

The community should not be misled in to believing such a system is complicated, a "waste of time", or impossible to implement (this is clearly and patently false given UC Bills quick progress on his proposed fix). The community should also be aware that such a system has existed on Wikipedia since 2003, and that editors are largely already aware of the effects of wikilinking date/year pairs.

Perhaps most important to any serious discussion of Dynamic Dates is that fixes to the system may be incremental. Specifically, not all things may be fixed at the same time, nor to the level of satisfaction to some editors.

View by User X[edit]