On Tue, Nov 11, 2008 at 8:00 PM, Charlotte Webb charlottethewebb@gmail.comwrote:
On 11/10/08, Carcharoth carcharothwp@googlemail.com wrote:
The discussion at the talk page of MOSNUM did raise the valid point that single-year linking has been avoided for years, and that discussion
probably
predates the start of date autoformatting.
Actually, it doesn't.
Oh, OK. Do you know:
(a) When autoformatting first started? (b) Whether year or date linking was being done before date autoformatting started?
My view is that the advent of autoformatting meant that the various date linking debates went quiet, and now they are coming back because some people (quite rightly) want to link dates for reasons other than
autoformatting.
Other reasons include but are not limited to: *Lack of an objective way to determine whether a link is "relevant to the context", particularly if one has no familiarity with the subject matter.
Agree. Though human audits using scripts work quite well if the human has clue.
*Lack of a reason to have day/month/year articles, if they will go unread in the absence of incoming links (no traffic).
Not so sure about this. The articles can still be accessed through the calendar templates and the category system. Instead of clicking on the year link, you click one of the date-related categories, and navigate your way to the relevant date page that way. It is a fair point that not everyone will want to do this, so not having the dates linked from the main text of an article as default is a good thing there (I think links in infoboxes and tables and lists are different). I'm actually rather sympathetic to the notion that link-quality in the main body of the text of an article is improved if it is kept to the most relevant links. In other words, the links act as a carefully selected mix of background articles and sister and daughter articles. Overlinking ruins this careful mix of links.
*Lack of an efficient way to update and maintain such pages in the absence of incoming links (no whatlinkshere).
I use whatlinkshere a lot, and I agree with you that it can be very useful. There are other ways to generate the same "metadata" (e.g. articles with a certain year mentioned) without using links. See:
http://en.wikipedia.org/wiki/Wikipedia:METADATA http://en.wikipedia.org/wiki/Wikipedia:Biographical_metadata
If you look at the way the geographical co-ords system is set up, the information is in articles, and machine-readable, but it doesn't interfere with the main body of the text. So all articles with co-ords within a certain range can be identified using a computer, instead of a human. That was one of the main arguments against date delinking, that a lot of potential information that could be generated by analysing the links was being lost. There was a proposal to replace linked dates with a template that would still tag the dates, but not link them. Not sure what happened to that.
Changing subject completely, one of my big gripes is the way that bloated templates stuck at the bottom of an article pollute "whatlinkshere" with irrelevant links. For example, although hardly any of the Nobel Laureates in Chemistry are mentioned in the main text of Marie Curie, they are all in the "whatlinkshere" list because each Nobel Laureate has all the Nobel Laureates listed in a template at the bottom of each article. Indeed, every single Nobel Laureate in Chemistry appears in the "whatlinkshere" list for each Laureate. If instead of the template, there was a link to a list, or a link to a category, then "whatlinkshere" would be much cleaner and could actually be checked carefully to see if all the links are being used correctly. Ditto for listing the outgoing links. If links from transcluded templates could be semantically distinguished from links from the actual articles, that would be great.
In principle, could that be done?
Carcharoth