Wikipedia talk:Citing sources/Archive 27

From Wikipedia, the free encyclopedia
Jump to: navigation, search


SayWhereYouGotIt question

If you're citing an article that appears on a newspaper's website, should the cite type be "Web" or "News"? Most of the time, I've no idea if the article appeared in the actual print publication. I've always used "Web" for this, and an editor just changed it to news [1]. I'm sorry if this has been covered before, I couldn't find it in the archives. Thanks. --Omarcheeseboro (talk) 12:20, 21 September 2009 (UTC)

We have just had a closely related discussion over at Template_talk:Cite_news#Still_confusion_around_parameters. The short answer to your question, in my opinion, is "cite news". That way, if you put the name of the organ under "Newspaper" or "Work", it will come out in italics, as it should do. In my view, this applies whether or not the article appeared in the printed paper: you should still use "cite news" even if there isn't a printed paper, such as BBC News Online. An online news service is still news, it just happens to be on the web. I think one test of whether something is really a news article or not is whether the item gives a clear publication date (in the case of the BBC, it is the "last updated" date). If an item cannot be precisely dated it is only a website -- in which case the access ("retrieved on") date, a very secondary piece of information when the item has its own clear date, becomes somewhat more significant -- and not a news source.
On your point about not knowing whether the item also appeared in print form, we seem to be in somewhat uncharted territory: I don't think anyone has really thought this through. My own feeling is that if you have seen the item only in web form, maybe you should say so, e.g. by citing the source as rather than The Guardian -- although as it happens the Guardian website is one which does actually tell you whether it also appeared in the paper. I say this because there are increasingly cases where the web content is different from what gets printed. I think the print version is preferable because it is the permanent record which (or the microfilm or digitised image of which) will still be available in libraries decades hence.
In your example, though, since it's a local paper that doesn't look like the kind of big outfit that might run extra online-only news items, I think I might take a bit of a chance and cite it as The Virginian Pilot rather than, given that the byline specifically says that the author is of that paper. Either way, I'm sure it should be "cite news" rather than "cite web", so I agree with the editor who changed your citation. -- Alarics (talk) 13:11, 21 September 2009 (UTC)

Generally speaking it's better to use {{cite news}} for news that is put out by legitimate and reputable news organizations, and not to worry whether it was actually printed. This is by the same reasoning that one should use {{cite journal}} to cite online-only scholarly journals like PLoS Biology. Some judgment is required—many news organizations have blogs by reporters for which {{cite web}} may be more important—but the general rule is pretty clear. Eubulides (talk) 14:48, 21 September 2009 (UTC)

Or sometimes it's best to just use the all-purpose {{Citation}} which accepts all the parameters from cite web, cite news, and cite journal. -- œ 23:23, 21 September 2009 (UTC)
Thanks all, great answers. --Omarcheeseboro (talk) 23:27, 21 September 2009 (UTC)
I've used both {tl|cite web}} and {{cite news}} for news articles. As long as you provide the necessary bibliographic info (and italicize publications), I don't think it's a big deal which you use, although it's best to keep the style consistent throughout an article. Dabomb87 (talk) 03:03, 22 September 2009 (UTC)

Citation standards for use with the Geographic Reference templates {{Template:GR}}

I've seen a number of US City articles that make a lot of use of the {{GR}} template as a reference. This expands in the Reflist to " "American FactFinder". United States Census Bureau. Retrieved 2008-01-31. ". This is a census bureau database, not a specific page for the city in question. Now I see two problems with this: 1) by what means does the reader or editor who happens upon this location later on know that this particular city-related article was actually updated when the 2008 census data came out? 2) if the {{GR}} has been inserted manually by an editor, there is no way to easily do followup and check a citation. This is true particularly for a paragraph that makes many assertions, and one cannot easily tell which ones are, or are not, supported by the US Census Bureau fact finder database. What ought to be the guideline for the use of the GR template. N2e (talk) 22:27, 19 September 2009 (UTC)

Do any of you more citation-savvy editors who frequent this page have any guidance on this? I see many of these citations on paragraphs that make a great many separate claims -- it is very difficult to determine if all the claims are verifiable when the link only take one to a database entry screen. The potentially bogus stuff inserted by vandals and trolls is hard to confirm with this GR-tag method of citation. N2e (talk) 03:52, 25 September 2009 (UTC)
First, note that this template is used in many tens of thousands of articles. (I quit counting at 30,000. Does anyone know how to get the real count? Sometimes I am amazed at how vast Wikipedia really is.) Any changes to the template call is a really big job, requiring the use of a bot and an editor who understands the format of these articles. I quick study of the history page of a few random articles suggests that User:Ram-Man originally created the articles in 2002 using User:Rambot. He may have some insight into how they can be improved as a group.
While it would have been nice if the template accepted the town name and passed this to the census bureau's website, this is not necessary. The website allows a reader to type in the town name and see the information for themselves. So the citation does its job fairly well -- it allows the reader to verify material in the article with just a little typing. ---- CharlesGillingham (talk) 16:50, 25 September 2009 (UTC)

Same übersource, different content, different newspapers

I have a story syndicated through the AP from two different newspapers. Each version includes different details that I want to cite. What's the correct way to source them? Otto4711 (talk) 20:55, 24 September 2009 (UTC)

I think you have to do two separate citations to the two newspapers. They can be put together at the end of your sentence, if that is appropriate, there's no reason why they can't both mention that the agency was AP. -- Alarics (talk) 21:49, 24 September 2009 (UTC)

Referencing YouTube

I've seen youtube videos linked and I believe I have seen them refenced in cases where they were very pertinant such as number of views of the Evolution of Dance video. So, what is the scoop on referencing youtube, or perhaps, some other, more strict video hosting sites, such as how to sites with pro submitted material; what is hte name of that one... (talk) 16:08, 28 September 2009 (UTC) Oh yeah, was the pro submitted one I was thinking of. (talk) 16:09, 28 September 2009 (UTC)

I don't think that You Tube itself is a reliable source, but you should bring that up at Wikipedia talk:Reliable sources.
The issue that concerns this guideline is "how do I write a citation for a You Tube video?" I would cite the original video, not the You Tube website. I would provide the same information that I would if I had rented the video at blockbuster or borrowed it from a library. The title, year, producer, director, production company, and (if appropriate) the exact mins/secs of the material I am citing. See {{Cite video}} for one option. ---- CharlesGillingham (talk) 05:36, 29 September 2009 (UTC)
Be reminded that articles are not to be linked to websites containing material uploaded to the Internet in breach of copyright: WP:YOUTUBE. — Cheers, JackLee talk 04:16, 2 October 2009 (UTC)

Discussion re whether to permit use of all-numeric YYYY-MM-DD format in footnotes

FYI -- there is a discussion at [2] as to whether or not to allow the use of the all-numeric YYYY-MM-DD format in footnotes/references.

I'm raising the point here in the event that you would like to follow it or join in. Thanks.--Epeefleche (talk) 07:29, 29 September 2009 (UTC)

How to deal with duplicates ?

How do one do when use the same reference many times in one article? The reflist list count the exact same reference as independent. So even if only a few references are used one may up with a very large number of duplicates in the reference list ... Benkeboy (talk) 16:19, 5 October 2009 (UTC)

One choice is to use parenthetical referencing. --Jc3s5h (talk) 16:29, 5 October 2009 (UTC)
Depends on the reference system you are using:
---— Gadget850 (Ed) talk 16:43, 5 October 2009 (UTC)

Citation bot, use of et al, and citation template parameters

I just posted a comment at Smith609's talkpage about the citation bot:

I just noticed this edit which added up to nine authors for certain citations. It's standard in scholarly literature to use et al after about three. Adding this many last names is going to make looking at citations more painful. And why not consolidate the first and the last names? In every when we use last names, we will use the first initials with them. Adding a firstname parameter introduces another unnecessary word.

Are these issues being discussed anywhere?

When using citation templates, all authors should be entered as separate parameters so that the metadata is properly preserved. You can limit the number of authors displayed with {{citation}} by using |display-authors=. This is undocumented, but works with at least {{cite book}} and {{cite journal}}. As to how many authors should be displayed before et al, I have seen one and three.
{{citation |title=Book title |last1=Imprimante |last2=Drucker |last3=Impresora |last4=Stampante |display-authors=3}}
Imprimante; Drucker; Impresora; et al., Book title 
---— Gadget850 (Ed) talk 09:10, 8 October 2009 (UTC)
I don't understand why we need to preserve the metadata. Wikipedia is not a repository for reference metadata. Please help me out here? If I choose not to use "complete metadata" when I add citations (partly because I don't like tons of parameters cluttering up the text), I don't think other people should be able to sweep that away because "there needs to be complete metadata". II | (t - c) 23:28, 8 October 2009 (UTC)
More complete and accurate metadata is useful to our readers to find and reference the original source. The additional parameters are only a burden to editors & that burden is small (compared to, say, using templates vs. not using templates). I think the citation bot is performing a useful service, but perhaps you could use the parameter that prevents the citation bot from editing your articles? --Karnesky (talk) 23:58, 8 October 2009 (UTC)
I totally disagree as the metadata canard has hung around the neck of wikipedia for far too long. What the h does it really mean? FWiW Bzuk (talk) 00:02, 9 October 2009 (UTC).
How is it an albatross when there is great utility and very little burden? See Zotero and LibX and other examples as to why this stuff is useful to have. --Karnesky (talk) 01:23, 9 October 2009 (UTC)
Metadata such as the COinS generated by the template allows Wikipedia to play nice with reference management software. Using Zotero or other applications, I can grab a reference from Wikipedia and port it into a paper (and yes, I have done this for college work). You can grab a reference from another site, save it in Zotero and plug itinto an article. This makes Wikipedia a more scholarly work. ---— Gadget850 (Ed) talk 01:33, 9 October 2009 (UTC)
Well, in other places its standard to use "and others." due to the slow removal of academic latin from citations. And in others its standard to ... . Seriously: our citation system is a horrible horrible mess right now. Fifelfoo (talk) 09:16, 8 October 2009 (UTC)
If "and others" is preferred (ten years from now, say), it would be a simple matter to change {{Citation/core}} to produce it. That's the beauty of using templates. The system may be a "horrible horrible mess", but this issue is easily fixed anyway, if consensus exists. ---- CharlesGillingham (talk) 18:01, 8 October 2009 (UTC)
And it would be trivial to add a switch for either use. ---— Gadget850 (Ed) talk 01:33, 9 October 2009 (UTC)
"et al." (note the period as it is an abbreviation for et alibi or et alli which is translated from Latin as "and elsewhere," "and others," respectively) is commonly used to identify the main/lead author of a group that includes three or more authors in at least some bibliographic style guides but not in the APA style that has formed the basis of the citation templates in use in Wikipedia. In APA, all authors are identified in a full bibliographic record which leads to a lengthy list (I have seen as many as 10 authors listed in "References" which makes the entry difficult to read). FWiW Bzuk (talk) 18:19, 8 October 2009 (UTC).
Note that in citations that provide a link to the referenced article or its abstract, a reader or editor thus has easy access to the full author list. So in this case, perhaps the full author list doesn't need to be in Wikipedia in any form, if easy access to the author list is the sole criterion. --Bob K31416 (talk) 02:05, 9 October 2009 (UTC)
I would really love to believe in the promise of permutable citations, but look how biblatex requires extensions to support Turabian, or the issue history articles have in requiring page refs from identical sources. Mechanised promised have not come through particularly well in my field. Fifelfoo (talk) 02:16, 9 October 2009 (UTC)


"Embedded links should not be used to place external links to websites in the body of an article." was added this month, and I've just appended: "(See Wikipedia:Layout#External_links for exceptions.)" The current exceptions are Wiktionary and Wikisource, although a discussion has just started at WT:Layout on whether these exceptions need tweaking. This conversation comes around every now and then, usually at WT:Layout, so it makes sense to me just to refer people there so they can see whatever the current recommendation is. Does this make sense? - Dank (push to talk) 16:32, 28 September 2009 (UTC)

Was that in response to my queston? If so, no it doensn't make sense. (talk) 16:41, 28 September 2009 (UTC)
The only exceptions currently in a guideline or policy that I know of are the ones listed at WP:Layout#External links: Wiktionary and Wikisource. We're talking about external links in the text; most external links that aren't in other endsections belong in External links. - Dank (push to talk) 17:24, 28 September 2009 (UTC)
After all of those discussions, I really ought to have remembered that 'complication'. Perhaps I should have said "external links to non-Wikimedia Foundation websites". That would encompass WP:LAY's "generally" statement while preventing the major problem (which is linking to websites for businesses or organizations named in passing). Or perhaps "(except for Wiktionary and Wikisource)" would be sufficient. WhatamIdoing (talk) 18:18, 28 September 2009 (UTC)
Sure, "(except for Wiktionary and Wikisource)" works for me, and I almost said that, but then I saw that a discussion has just started on the subject at WT:LAYOUT and I didn't want to commit until that discussion is over. - Dank (push to talk) 18:33, 28 September 2009 (UTC)
Do we not use interwiki links rather than external links when linking to non-Wikimedia wikis? Anomie 20:49, 28 September 2009 (UTC)
I do, yes. There's been no response at WT:LAYOUT to Johnbod's question; if nothing happens over there, say, within a week after his question, then "(except for Wiktionary and Wikisource)" here works for me, and we should mention the wikt: and s: markup. - Dank (push to talk) 13:38, 29 September 2009 (UTC)
Okay, there have been no changes at WP:LAYOUT, so I changed my addition to: "InterWikimedia links to Wiktionary and Wikisource are sometimes appropriate in the body of an article; for details, see Wikipedia:Wikimedia sister projects." Feel free to revert and discuss. - Dank (push to talk) 17:32, 9 October 2009 (UTC)
We could probably just eliminate that sentence; it is not really relevant to a citation guideline. Christopher Parham (talk) 17:55, 9 October 2009 (UTC)
Aren't Wikimedia links "embedded"? If so, and if we make a statement about embedded links, shouldn't we clarify? - Dank (push to talk) 01:40, 10 October 2009 (UTC)

Many authors of a cited work

The various templates akin to {{cite journal}} allow for up to 9 authors to be listed, and each may be specified either as |authorn= or as a |lastn=/|firstn= pair (where n is an integer in the range 1-9). If the article to be cited has ten authors, what is the recommended procedure? At Wikipedia:Help desk#When to use "et al" in a list of authors of a citation, User:DoktorMandrake stated "However, all authors must be present in the references section." (his emphasis).

One idea that I had was to specify the first eight authors individually, and put the remainder together as the "ninth", thus:

These theorems were later shown to be true.{{sfn|Angle|Baker|Chart|Droog|2009|p=42}}
*{{Cite journal |title=Confirmation of statements |journal=Journal of Strange Ideas |publisher=Justthefacts |year=2009 |month=October |ref=harv
|last1=Angle |first1=A.A.
|last2=Baker |first2=B.B.
|last3=Chart |first3=C.C.
|last4=Droog |first4=D.D.
|last5=Eagle |first5=E.E.
|last6=Freud |first6=F.F.
|last7=Green |first7=G.G.
|last8=Henry |first8=H.H.
|author9=Ink, I.I.; James, J.J.

which gives:

These theorems were later shown to be true.[1]

  1. ^ Angle et al. 2009, p. 42.
  • Angle, A.A.; Baker, B.B.; Chart, C.C.; Droog, D.D.; Eagle, E.E.; Freud, F.F.; Green, G.G.; Henry, H.H.; Ink, I.I.; James, J.J. (2009). "Confirmation of statements". Journal of Strange Ideas. Justthefacts.  Unknown parameter |month= ignored (help)

But as you can see, the last two authors are replaced by the "et al", so that they don't get credit. However, If I cheat a little further in {{cite journal}} thus:

*{{Cite journal |title=Confirmation of statements |journal=Journal of Strange Ideas |publisher=Justthefacts |year=2009 |month=October |ref=harv
|last1=Angle |first1=A.A.
|last2=Baker |first2=B.B.
|last3=Chart |first3=C.C.
|last4=Droog |first4=D.D.
|last5=Eagle |first5=E.E.
|last6=Freud |first6=F.F.
|last7=Green |first7=G.G.
|author8=Henry, H.H.; Ink, I.I.; James, J.J.

it now shows:

  • Angle, A.A.; Baker, B.B.; Chart, C.C.; Droog, D.D.; Eagle, E.E.; Freud, F.F.; Green, G.G.; Henry, H.H.; Ink, I.I.; James, J.J. (2009). "Confirmation of statements". Journal of Strange Ideas. Justthefacts.  Unknown parameter |month= ignored (help)

so now, I get all ten. Is that permissible, or is there a better, recommended way? User:Gadget850 suggested that I ask here. --Redrose64 (talk) 12:40, 8 October 2009 (UTC)

Note that User:DoktorMandrake is referring to a requirement of the APA style. Other styles (including possibly Wikipedia's various "house styles") may well differ. At the moment, it appears that the "house style" implemented by the citation templates says to use et al after the first 8 authors, but that could change in the future. Anomie 16:38, 8 October 2009 (UTC)
That is definitely not the "house style" of Wikipedia. It is merely the number of authors that got implemented before the implementers of that template got tired. In biomedical articles it's common to put et al. after 6 authors; or after 3 authors if there are more than 6 authors (styles differ). There is no requirement or consensus that all authors must be listed; it's fine if a citation merely says something like "|author= Filipek PA, Accardo PJ, Baranek GT ''et al.''" and there's no need for it to say something lengthy like "|display-authors= 3 |author1= Filipek PA |author2= Accardo PJ |author3= Baranek GT |author-separator= , |last4= Cook Eh |first4= Jr |last5= Dawson |first5= G |last6= Gordon |first6= B |last7= Gravel |first7= JS |last8= Johnson |first8= CP |last9= Kallen |first9= RJ". Eubulides (talk) 16:51, 8 October 2009 (UTC)
As noted above, we can list the authors separately for metadata purposes, but specify the number to show with an et al. If there is no current way to handle more than nine authors, then we should move this to {{citation/core}} and ask for more author fields. ---— Gadget850 (Ed) talk 17:05, 8 October 2009 (UTC)
"et al." (note the period as it is an abbreviation for et alibi or et alli which is translated from Latin as "and elsewhere," "and others," respectively) is commonly used to identify the main/lead author of a group that includes three or more authors in at least some bibliographic style guides but not in the APA style that has formed the basis of the citation templates in use in Wikipedia. In APA, all authors are identified in a full bibliographic record which leads to a lengthy list (I have seen as many as 10 authors listed in "References" which makes the entry difficult to read). FWiW, the MLA (Modern Language Association) style guide that is commonly used in research works for the humanities, uses "et al." in both abbreviated and full entry notations, so that only the first author and et al. is used. Bzuk (talk) 18:19, 8 October 2009 (UTC).
At college we were instructed (not advised) that if our dissertations were to stand any chance of being passed, then et al. was only to be used for the second reference from a given source; the first one had to bear the names of every author - even when it looked like the entire faculty had written the paper. --Redrose64 (talk) 19:09, 8 October 2009 (UTC)
That experience is probably linked to the field or faculty in which you were resident. Every disciple provides the structure demanded in research and publication. FWiW Bzuk (talk) 20:25, 8 October 2009 (UTC).
It sounds like there are some unrealistic expectations here. Would that college have forced your dissertation to list all 314 coauthors of Birney et al. 2007 (PMID 17571346)? or all 580 coauthors of Abazov et al. 2007 (PMID 18233063)? How about Grünewald et al. 2006 (doi:10.1016/j.physrep.2005.12.006)? That paper has has 2,512 coauthors: would your college really have made you list them all? If we're serious about this, shouldn't {{Citation/core}} be expanded to support 2512 coauthors and all the metadata that would go along with it? If so, shouldn't the template support at least 10,000 coauthors, to allow for future expansion? (The answer ought to be obvious: "No".) Eubulides (talk) 06:08, 9 October 2009 (UTC)
I don't think that such high numbers ever came up. Certainly there were several instances of 10+, occasionally 20+, but I personally never had to cite from a paper with more than about 15. The rule given to us was simple: credit every author unless you've already credited them. Maybe our prof. had never encountered papers such as your examples. I'm not saying that you're wrong: just trying to recount my own experiences. --Redrose64 (talk) 10:18, 9 October 2009 (UTC)
Either you or the publisher have miscited "Grünewald" (2006). The publisher shows 7 corporate authors. Fifelfoo (talk) 10:59, 9 October 2009 (UTC)
It's the publisher. Their (software / page layout / whatever) couldn't handle that many authors either. The actual author list of Grünewald et al. 2006 (doi:10.1016/j.physrep.2005.12.006) is in Appendix A of that paper. In contrast, the publisher for Abazov et al. 2007 (doi:10.1103/PhysRevLett.99.191802) did manage to squeeze all 580 author names onto the front. (Pretty cool for an 8-page paper, huh? The list of author names and affiliations consumes 3 of those 8 pages.) Needless to say, there's no reason Wikipedia editors and readers should be forced to wade through that list of authors. Eubulides (talk) 21:47, 9 October 2009 (UTC)

Guidelines and info from Wikipedia re use of et al.

AMA citation guidelines suggest that if there are more than six authors, include only the first three, followed by et al.[1] The Uniform Requirements for Manuscripts Submitted to Biomedical Journals (URM) citation guidelines list up to six authors, followed by et al if there are more than six.[2]

--Bob K31416 (talk) 05:33, 9 October 2009 (UTC)
  • And here's an excerpt from et al.

APA style uses et al. if the work cited was written by more than six authors; MLA style uses et al. for more than three authors.

--Bob K31416 (talk) 14:28, 9 October 2009 (UTC)

Shouldn't the point be this: standards vary. Wikipedia's templates have chosen a standard: a maximum of three authors in a parenthetical reference or shortened footnote, a maximum of nine authors in a full citation. The templates must use some standard and the standard is easily changed, however I can't imagine how this standard is worse than any other. ---- CharlesGillingham (talk) 18:51, 9 October 2009 (UTC)

Retrieval dates: redundant for sources with official publication dates?

This subject keeps coming up. There is an extensive discussion in the archive. --EnOreg (talk) 14:20, 27 April 2009 (UTC)

The article now says, for newspapers, "... and a comment with the date you retrieved it if it is online (invisible to the reader)". But it doesn't tell people how to do this. Anyway, the retrieval date serves no purpose. Either the link still works, or it doesn't. If it doesn't, remove the url and leave the reference to the dead-tree newspaper. Alarics (talk) 18:04, 10 August 2009 (UTC)
User:SlimVirgin deleted this without discussion or consensus. I have now restored it but with an explanation of how to comment out the date. - Alarics (talk) 08:14, 4 September 2009 (UTC)
The advice to delete the dead link, is about the worst I have seen here. Then why do we have the {{Dead link}} template? Or go search for archived versions. Debresser (talk) 23:54, 13 September 2009 (UTC)
I agree that searching for an archived version is the preferred cause of action on dead links. That's the point of having an access date in the absence of a publication date.--EnOreg (talk) 09:26, 14 September 2009 (UTC)
[inserting direct reply to that specific point:] Yes, of course search for another web source to replace the dead link, that goes without saying. I was just making the point that a dead-tree news cite is still a valid cite, whether or not it can be found anywhere on the web. The citation is to the news article; the web link, if there is one, is for convenience. -- Alarics (talk) 13:06, 14 September 2009 (UTC)
Commenting out the date makes no sense to me. If there is a consensus to not show it, then delete the field from the templates and be done with it. ---— Gadget850 (Ed) talk 11:29, 14 September 2009 (UTC)
The point of commenting out the access date is that it is useless information to someone reading the page, whereas it might conceivably be of some use (at least, some people seem to think so) to someone editing the page. We should only show to the reader information that might be of some potential use or relevance to him or her. -- Alarics (talk) 13:57, 15 September 2009 (UTC)

This has all been voluminously discussed before here. Do we really have to go over all this again? Here is a summary of that earlier discussion:

User:Wasted Time R starts off by pointing out several reasons why access dates are often at best superfluous and arguably a positive nuisance. There is then a bit of a diversion while one or two people misunderstand the point (a red herring about "archiving" dates), and another red herring about whether to call it "accessed" or "retrieved"). A total of two people try to supply some justification for access dates in certain circumstances, but even they agree that it is "optional". User:LouScheffer makes the point that much depends on how dependable is the online source, and that there is certainly no point in including an access date when the reference is to a major newspaper. It is at this point that User:EnOreg comes up with the compromise suggestion of commenting out the access date so that it is visible to editors but not to readers, thus avoiding confusion and clutter. There is then some rather technical discussion about how one might apply this to all existing articles (but that is not what I am proposing, I only suggest a sensible policy be adopted from now on. However, the fact that several different people entered into that technical discussion suggests they agree with the idea).

Noted administrator and tech guru Martin Smith (User:Smith609 then enters the fray with the following:

I do not see how an accessdate on sources which do not change - such as journal articles - is beneficial. However, on sources which may change, such as web content, it helps clarify which version of a page is being cited. Therefore I feel it ought to be displayed only in the cite web template. I don't think anyone has disagreed with this feeling here, so I suggest that someone bold goes ahead and proposes or enacts the change at all non-"cite-web" templates. People have had the chance to complain if they feel otherwise!

User:EnOreg sums up thus:

Consensus: It indeed seems we have consensus that access dates for online copies of offline sources, while helpful as a comment in the source, should be hidden from the reader. I have removed the RFC (style) tag and will modify the policy. Anybody who is competent to adapt the citation templates, please do so. Thanks everybody,

User:WhatamIdoing adds: "the access date is really about citing websites that were created as websites, not books that happen to be conveniently available online. I don't cite access dates for news articles that I read online, either. Reuters News or Associated Press stories will be verifiable for many years after the link goes dead." He or she later clarifies: "Since websites do change, it is reasonable to include the access date, just like you'd include a publication date if you were citing a newspaper. If the ref isn't web-only -- and I see no reason to think that this comment is intended to apply to anything else -- then an access date is unimportant." In some further discussion, even those who strongly defend the use of access dates for web-only sources all agree that it's not necessary where there is a real newspaper article with a date.

After seven months of discussion (note: the section furthest down the talk page is not the most recent), there is only one dissident, User:WLU, who wants to revert to all access dates being shown in all circumstances, and is told that he or she does not have a consensus for that, at least a dozen other contributors to the discussion having been either broadly in favour of, or at worst neutral about, what had by then (end 2008) become the status quo at WP:CITE.

Presumably as a result of all that (none of which was anything to do with me, by the way), WP:CITE now reads: "Citations for newspaper articles typically include: (....) the date you retrieved it if it is online, invisible to the reader: <!--accessed: date-->".

It cannot seriously now be claimed that this matter has not been adequately aired. -- Alarics (talk) 13:02, 14 September 2009 (UTC)

I was asked to comment here, although I can't quite tell where this iteration of the discussion sprung from. Since my original remarks on this, I have become even more against using accessdate unless it's for an undated web source. I have successfully defended my "avoid accessdates as much as possible" position to several GA reviewers, so I think the realization that accessdates have been overused in the past is gradually gaining acceptance. I would be happy if some of the cite templates were changed to turn accessdates off or into comments. Wasted Time R (talk) 01:13, 15 September 2009 (UTC)
Sorry guys, but yes, we have to go over all of that again. Furthermore, if you don't put notifications on the talkpages of at least the mayor citation templates, there is no change any editors will implement what you "decide" here. A good case example is Template_talk:Cite_news#Accessdate where a recent discussion ananymously disagrees with all of you guys here.
My opinion is that we should have a single accessdate parameter (no "accessmonthday", or even "accessyear", just "accessdate"), and that it should be visible. Or I could live without an accessdate parameter as well. But having it and hiding it is ridiculous. Such is my opinion, and if asked I'll be happy to explain why I think so. Debresser (talk) 09:33, 15 September 2009 (UTC)
When Debresser says "a recent discussion ananymously disagrees with all of you guys here", he makes it sound as if there has been a long debate with several participants eventually reaching unanimity. What actually happened at Template_talk:Cite_news#Accessdate was that only yesterday he introduced the subject there, at the same time making a change to the text of the project page for which there was no consensus. A grand total of two other editors supported him, one of whom was evidently unaware of the earlier discussion here and the other of whom had actually taken part in that discussion but was in the minority there. That discussion, by the way, lasted six months, whereas the discussion from which Debresser now purports to draw unanimity only began yesterday! The final contributor so far atTemplate_talk:Cite_news#Accessdate, User:EnOreg, does not support Debresser's move (so the three-person unanimity has already disappeared) and points out that here, not there, is the place to be having this discussion. The real tally so far, as far as I can make out, is three editors on Debresser's side of the argument, and seven editors on mine. -- Alarics (talk) 10:08, 15 September 2009 (UTC)
Commenting out accessdates is not a solution. There should be a formal proposal to eliminate them altogether and get this over with. ---— Gadget850 (Ed) talk 03:16, 25 September 2009 (UTC)
|accessdate= can be quite useful even for archival journals, in some cases. For example:
This source was published in an archival journal, where you can buy it; it is also freely available, perhaps temporarily, from the author's personal website. The accessdate applies to the perhaps-temporary free copy, not to the expensive stable copy. This sort of thing is quite common in some high-quality articles; I count five instances of it in Autism. I don't see why this usage should be eliminated. Eubulides (talk) 03:34, 25 September 2009 (UTC)
But what purpose does the accessdate actually serve in such a case? Either the online version of the journal article is still there when you go looking for it, or it isn't. If it's gone, how does it help to know on what date some editor found it in the past? -- Alarics (talk) 09:38, 25 September 2009 (UTC)
Yes, it is helpful. You can search for a mirror of the site made around that date in, e.g., the Internet Archive. If the accessdate is recent, there is also a greater chance that the article might be down only temporarily. --Karnesky (talk) 13:44, 25 September 2009 (UTC)
Agree w Karnesky that accessdate helps in getting the right version from e.g., the Internet Archive, if what you originally cited goes offline. --Philcha (talk) 14:03, 25 September 2009 (UTC)
What you are citing is the original article, not some copy of it. The validity of the reference doesn't depend on the availability of a web link. For the corner case you are describing it is sufficient for an editor trying to recover a dead link to see the commented-out access date in the article source code. --EnOreg (talk) 15:41, 25 September 2009 (UTC)
No, some sources are all-electronic, with not a dead tree in sight - but have editorial boards of reputable people, so comply with WP:RS. --Philcha (talk) 15:54, 25 September 2009 (UTC)
To be clear: that is arguing to maintain the status quo (allow an accessdate, show it when there is not sufficient info to get any other copy & comment it out when a hard or paid copy is available). I am fine with this. User:Alarics wants the accessdate to be removed completely, and Philcha and I were replying to that. --Karnesky (talk) 17:01, 25 September 2009 (UTC)
Yes, the status quo should be maintained. This is not a question of the validity of the citation, as the citation is valid even without URL (or DOI, for that matter). The point that it's sometimes quite helpful to the reader (and to future editors) if the citation displays a relatively-ephemeral URL and accessdate in addition to a permanent (but more-expensive) DOI or PMID. Eubulides (talk) 17:51, 25 September 2009 (UTC)

(outdent) I am rather unclear on how one could summarize all of the discussion above. If this later is consolidated into a consensus proposal, I would be happy to have someone invite me back (via comment on my Talk page) to weigh in on that summarized proposed consensus. However, I will leave my general thoughts on the matter as well: I believe that accessdate is a very useful field in WP citations, as it alone allows a WP subsequent editor to see when a previous editor accessed the material. If one then were to look at the editor who last accessed the material, and find that editor's WP reputation a good one, the subsequent editor could 'let it go' that, perhaps, the electronic source is no longer verifiable online. Thus, accessdate is not redundant on material that has official publication dates. I also agree with User:Debresser who pointed out that a consensus here would not end the issue; a broader consensus would need to be reached on the Talk Pages of the major citation templates if we hope to have any consensus widely implemented. N2e (talk) 17:31, 25 September 2009 (UTC)

Karnesky notes that commenting out the accessdate in certain circumstances is the status quo. Indeed, that is what has WP:CITE has said for over a year, but it was when Debresser took exception to my actually implementing that policy that this argument started up all over again. I am not objecting to the use of accessdate as long as I can comment it out when it clearly isn't of any use to the casual reader. -- Alarics (talk) 19:32, 25 September 2009 (UTC)
Shocking as it may seem, even Major newspapers have been known to change articles on their web sites. Often to save themselves embarrassment over some egregious error. This is why retrieval dates are important where the reference comes from the web site instead of from an actual copy of the paper, book or magazine. filceolaire (talk) 20:20, 12 October 2009 (UTC)
But if it's an "egregious error" that will usually be within hours rather than days. How is that any different from different editions of the printed paper -- which nobody has ever suggested should be indicated separately in citations? -- Alarics (talk) 20:43, 12 October 2009 (UTC)

Archiving Web pages for verifiability

This was discussed first on FL Criteria Talk page, users suggested the move to this guideline page instead. The idea is to make archiving pages and avoiding link rot a guideline.--Diaa abdelmoneim (talk) 21:50, 7 October 2009 (UTC)

Many pages on the web get deleted and become inaccessible. Web Citation provides a way to make sure that dead external links wouldn't occur. The {{Cite web}} template provides archiveurl parameter for the purpose of adding archived links of pages through webcite. sometimes skips pages and the information becomes inaccessible. Shouldn't archiving web references for future verification be a criteria of a featured list?--Diaa abdelmoneim (talk) 21:06, 7 October 2009 (UTC)

And what happens when the Webcite website goes down, as it has in the past? In fact, on one of those occasions it was because Wikipedia (or some bot) was using it too much and crashed it, no? If all the references in all our FLs were cited through Webcite's archives, every single FL would have been unverifiable. I'd rather one or two references in one or two FLs be bad links than every single reference on the project. It's just too risky to rely on Webcite all the time. The only time I use it is if I know the page is definitely going to expire, such as Yahoo news pages, or which changes on a weekly basis. Matthewedwards :  Chat  21:20, 7 October 2009 (UTC)
(EC) Good points, though I don't see any harm in creating archived versions and adding the |archiveurl= and |archivedate= parameters between <!-- and -->. Goodraise 21:32, 7 October 2009 (UTC)
The original link and the webcite link are kept in the reference, so the original links, even if they become dead, are kept in the reference. Webcite links were only broken for a while then returned active again; The site moved to new servers just for the demand of Wikipedia. The bot User:WebCiteBOT still runs, to this day, and archives various newly added links. (It has a backlog though...) --Diaa abdelmoneim (talk) 21:40, 7 October 2009 (UTC)
(EC) I've suffered link rot on some of my lists and it took hours and several editors working together to repair. However, I think as a result we also improved the quality of the sources for the list, or removed items that were never well sourced to begin with. Our citation guidelines (WP:DEADREF) currently don't rise above "consider" wrt the use of WebCite. Looking through the archives, I see issues where the WebCite servers were broken for a period and also concerns that their servers couldn't handle the demand if Wikipedia started routing all external-link-citations through it. I don't think featured content could demand this until using WebCite (or similar) was actually a guideline requirement. Perhaps WT:CITE is the more appropriate venue to discuss making it a formal guideline. Colin°Talk 21:27, 7 October 2009 (UTC)
Agree that we should encourage fixing dead links as necessary, disagree with making explicit mentions of it in the criteria; anyway, it is covered by the citation guidelines that Colin mentions above. Dabomb87 (talk) 21:41, 7 October 2009 (UTC)
I am leery of citing any web source that is no longer on the web. If it's no longer available, there could well have been something wrong with it in the first place. I would not favor requiring the use of Webcite, not only for the performance reasons mentioned above, but also (as Colin mentions) because dead weblinks are a sign that the article needs to be improved rather than fossilized in the past. Eubulides (talk) 03:05, 8 October 2009 (UTC)
I have seen archived weblinks or the use of the Wayback Machine in order to resurrect dead links, so all may not be lost. FWiW Bzuk (talk) 03:09, 8 October 2009 (UTC).
I wouldn't be leery about using archived sites; pages go down for reasons like lack of interest by the owner, by a site taking down expansive archives for server space or when a property is no longer current (as happens on TV-network sites all the time), or when archived material goes into pay archives. Doesn't mean the information is bad.
I see WebCite links as an addition to a particular extant link, not a substitution.
Alarmingly, however, I haven't been able to get to WebCite for awhile, and today got a total "Unable to connect to the server" message today. Is the site gone, or just down? Does anyone know?-- Tenebrae (talk) 14:50, 11 October 2009 (UTC)
Seems to be working fine right now. Must have been a transient problem. — Cheers, JackLee talk 07:44, 12 October 2009 (UTC)


Meaning/Use of either of these terms seems rather ambiguous because:

1) Information that directs me (any author) to use and identify an information source as a "Reference" and when to idetify it as "Citation" does not appear to exist, although there is ample information about the former, and

2) Information about the term "Reference" refers to it using both "Cite" and "Reference" as the base of words used within the definition of "Reference", and

3) There is markup unique to each.

Am I somehow not seeing information that addresses this, or (perhaps) does someone know where I can find it?

Kernel.package (talk) 17:54, 1 October 2009 (UTC)

These words are rather ambiguous, but this guideline attempts to use them in a consistent way, based on their most common or central meaning:
  1. "Citation" --- a line of text in a Wikipedia article, that contains the title and other information about a source.
  2. "Source" --- a website or book, etc. Something outside of Wikipedia that can be used to verify information in Wikipedia.
  3. "Reference" --- can refer to either the line of text or the book, i.e. either the citation or the source, or both together. (Because this word is more ambiguous, this guideline avoids it if one of the other three words can be used.)
  4. "Footnote" --- a line of text at the bottom of the article, associated with a superscripted number in the text of the article, which may contain a citation or an explanatory note.
On talk pages, these words are sometimes used interchangeably, but this guideline attempts to use them consistently with these definitions. ---- CharlesGillingham (talk) 18:16, 1 October 2009 (UTC)]])
In answer to your question about "where are these defined": "citation" is defined in the first line of WP:CITE (this guideline). Perhaps we should define "source" in the first line of WP:Reliable sources and "footnote" in the first line of WP:FOOTNOTE. ---- CharlesGillingham (talk) 18:28, 1 October 2009 (UTC)
Source is defined in a footnote at the Verifiability policy. WhatamIdoing (talk) 19:19, 1 October 2009 (UTC)
Still the same issue, the term "reference" is not a recognizable term in publishing or librarianship. A "reference source" however, describes either a print or non-print resource. A citation is a citation but generally is "folded" into "notes"; a bibliographical record is expressed as a "bibliography", but not in wickywacky world. FWiW, LOL Bzuk (talk) 20:06, 1 October 2009 (UTC).
Yup, you're right: Wikipedia has its own jargon. I think that's ultimately a good thing, but no editor is required to share my opinion. WhatamIdoing (talk) 19:20, 3 October 2009 (UTC)
Huh? FWiW, does anyone, considering your userid? Bzuk (talk) 00:26, 4 October 2009 (UTC).
Does anyone share my opinion on Wikipedia's jargon? Presumably all of the editors that use it adeptly to convey very specific meanings to other editors do. If they didn't, then presumably they'd quit using it. I'm not sure how my userid relates to this. WhatamIdoing (talk) 05:30, 4 October 2009 (UTC)
Trying not to be obtuse, but your userid connotes a certain lack of purpose? (LOL) FWiW, note the many editors that have "adeptly" adopted/purloined/"made-up" language that more clearly expresses the bibliographical terminology. See: citations, endnotes, footnotes, attributed resources, books used, the dreadful "for further reading" and ad nauseum... Bzuk (talk) 13:12, 4 October 2009 (UTC)
Actually, the userid comes from the polar opposite thought: I have several important projects, and I'm limited to the same 24 hours per day as anyone else. Is this what I really want to be doing with my time? There are certainly days when the answer is "no" -- oh, say, like when I'm dealing with judgmental editors that think they can divine my character and my life from my userid, and further think that such unfounded speculations constitute reasoned arguments in favor of their personal POV.
(No, I'm not the least bit angry... but if you actually have a reason for thinking that Wikipedia should confine itself to a vocabulary that was carefully refined for the dead-tree information age, and that furthermore differs by country and profession, then do feel free to comment on that, instead of on other editors.) WhatamIdoing (talk) 01:58, 6 October 2009 (UTC)
Thanks for the clarification as to the reasoning behind your electronic nom-de-plume and believe me, I was merely jesting (perhaps you didn't see the tongue in cheek and LOL tag?). As to the formulation of terminology that is almost entirely made-up, that is my real concern. Many of the terms that I have given as examples have been used and it is puzzling why wiki editors chose to try to remake the wheel when there were perfectly good examples of bibliographic cataloging to use as models. There are already a well-established research and referencing guides that prevail in the "dead-tree" world but they have emerged through a long history of use and acceptance. FWiW Bzuk (talk) 02:16, 6 October 2009 (UTC).
Because the "perfectly good examples" vary by country and profession, and do not provide adequate precision for Wikipedia's purposes. For example, what Wikipedians most commonly title "Works", "References", and "Further reading" could all be given the same name ("Bibliography"). We need a specialized language so that I can say ==References==, instead of "==Bibliography== (oh, I'm using that in the 'what you put the list of works actually cited' under sense, not the other legitimate senses)". WhatamIdoing (talk) 05:33, 6 October 2009 (UTC)
With all due respect, the formalized citation and bibliographic style guides do not vary, merely the exponents of them tend to misappropriate the elements of the formats and apply them inaccurately. All publishing houses apply rigorous standards to their works and proscribe the "house" guide that an author and editor must use. No author/editor creates a format or adapts a current standard, that just doesn't make sense, yet that is exactly what is happening daily in Wiki articles. Ç'est la vie, mais quel dommage... FWiW Bzuk (talk) 06:14, 6 October 2009 (UTC).
If the "correct" use of these terms were absolutely the same in every single country and profession, then publishers would not address this issue in their house style guides, because every single guide would say exactly the same thing, and every author and publisher would automatically do exactly the same thing without being told. House style guides already omit other universal standards (e.g., place a full stop at the end of declarative sentences, and a question mark at the end of interrogatives). Do you understand that? There is no absolute universal standard for which term to use. What you see in English lit is not what you see in history, which in turn is different from what is chosen in the field of physics. WhatamIdoing (talk) 18:03, 7 October 2009 (UTC)
<Hey, Bzuk. You've brought this up before, and I just want to be sure I understand your position. You only object to the way Wikipedians use the word "Reference" in their articles, right? You don't object to the way this guideline uses the words "Citation", "source" and "footnote", do you? You agree that this article should avoid the ambiguous word "reference", unless we're forced to, as when we discuss things like <references/> or ==References==, or the titles of other pages, like Wikipedia:WikiProject Fact and Reference Check, or Parenthetical referencing? <---- CharlesGillingham (talk) 17:33, 7 October 2009 (UTC)
Let me provide some perhaps enlightening information. Here's a list of four common styles heavily used in universities, and what each requires for the name above the list of sources that were used to support content in an academic paper:
  • Chicago: "Center the title Bibliography about one inch from the top of the page"[3] (used by historians)
  • APA: "In APA style, the alphabetical list of works cited, which appears at the end of the paper, is titled 'References.'"[4] (used by sociologists and psychologists)
  • MLA: "Center the title Works Cited about one inch from the top of the page."[5] (used in humanities)
  • CSE: "Center the title References (or Cited References) and then list the works you have cited in the paper; do not include other works you may have read."[6] (used by scientists)
Now, Bzuk, are you still convinced that nobody uses "References" as the title for this section? Are you still absolutely sure that everybody uses exactly the same format in "proper" publications? Do you think that the producers of widely used style guides that have been adopted by entire academic fields throughout most of the English-speaking world are somehow ignorant, undereducated people that have got it wrong, simply on the grounds that you personally don't approve of their choices? WhatamIdoing (talk) 18:29, 7 October 2009 (UTC)
Charles, it's not necessarily the use of the term "references" that is an issue since it is a term widely accepted as referring to the reference source, in either print or nonprint formats. It's when Wiki editors assume that references mean "notes (footnotes or endnotes)" or "citations", that it becomes confusing. I see the use of the title "references" to cover all forms of references, while citation/notes apply to the link to a source, either in full or in abbreviated form (ala Harvard citation) while the overall bibliographic record appears in a "works cited" or preferably, "bibliography." The use of the term "bibliography" becomes problematic in Wiki as it is used to identify an individual's body of work, rather than the listing of the particular use of reference sources in a bibliographic record form. The use of any of the "standards" for cataloguing and bibliographical records, such as the MLA, Chicago Style or others, is what I have been advocating. What is accepted in Wikipedia articles is often an amalgam of styles, made-up styles, or errors in using a proscribed style. Going back to an earlier example, when a work is written and researched for publication (regardless of where it is eventually published), the publishing house dictates a style guide to be used, authors and editors then strenuously observe the dictums of that style. One of the recurrent themes in publishing is that, in order for an academic work to be considered an authoritative work, a comprehensive list of references is provided, in note and bibliographic form. Wiki has made some allowances for the use of a variety of sources which has led to titles such as "external links" and "for further reading" which is much the same as grouping reference sources in the case of the former but becomes an conundrum in the latter's case as is it "works cited" that is actually provided or simply a list of valuable "other" sources? My main concern in referencing a wiki article is the lack of compatibility due to the "many cooks" syndrome (entirely understandable as the effort of creating an encyclopedia by committee inevitably leads to either a collaborative effort or sometimes a chaotic result). More to come, but this is mainly a "stream-of-consciousness" response at this point. FWiW Bzuk (talk) 22:44, 7 October 2009 (UTC)
Wikipedia doesn't wikt:proscribe any method. It also does not wikt:prescribe any method. It explicitly permits the "amalgam of styles, made-up styles" and all styles that are formal styles that you (or any other reviewing editor) just don't happen to recognize as a correctly implemented formal style. This last is one of the reasons for permitting apparently made-up styles: We don't have the resources to determine what all of the True™ styles are. Wikipedia would rather spend its volunteer's time on something that readers can figure out -- even if that means using a "made-up style" or rejecting a "standard" style for any given article. The major impetus behind not requiring every article to use the same style is to avoid a holy war over which style is the One True™ Style. You will never convince a historian that CSE is the correct style for an article about history, or a physicist that he needs to learn and use MLA when writing about physics.
As for your confusion about which works are being cited: WP:LAYOUT and WP:EL prohibit listed reliable sources used to verify article content under either ==External links== or ==Further reading==. No featured article, and remarkably few partially developed articles, breaks that rule. There is even proposal at EL to include this advice in the very first sentence of the guideline.
I would be interested in hearing more about your "lack of compatibility". What exactly is Wikipedia's approach not compatible with? Is this a complaint that physics articles follow different conventions than history articles? Or that most articles don't precisely match the style used in your own professional field (noting that if they did, they'd consequently not match someone else's standard style)? WhatamIdoing (talk) 04:37, 8 October 2009 (UTC)
I agree that Wikipedia does not promote or mandate any individual citation or bibliographical style although the templates appear to only be written in an APA-style. Nonetheless, editors approaching any article from the outset, generally are given precedence as to use of a particular style unless, through consensus, a new rewrite is instituted. In reading WP:LAYOUT, it is apparent that a great deal of movement has occurred in recent times to clarify terminology. The instances where "for further reading" has been misconstrued as a "works cited" is likely to be less and less common in the future. There are numerous instances where an electronic reference source will appear in a citation or note as well as in a bibliography, and then be repeated as an external link. Where incompatibility can be an issue, is when "mixing" of styles, dates, and other information becomes confusing to the reader (as well as the wiki editor). Using ISO dates in a template, but written-out dates, sometimes in two different conventions in the body of the text is one such example. FWiW Bzuk (talk) 04:58, 8 October 2009 (UTC).

<There are several topics under discussion here. The original poster, as I understand them, asked a question about semantics. I'm hoping that we all agree on the basic ontology denoted by various uses of these words and the central meaning of "footnote", "citation" and "source" in this context. If you disagree with my first post above, please let me know right away.

The other (alleged) problems must be discussed directly and made more specific before we can take any action on them:

  1. We call our footnote markup <ref> when it probably should have been <footnote> or <endnote>.
  2. A list of citations to sources in an article is usually titled References.
  3. A list of footnotes containing citations to sources in an article is usually titled References.
  4. Some guidelines refer to "footnotes containing citations to sources" as "references".
  5. People tend to refer to "footnotes containing citations to sources" as "references" on talk pages.
  6. And finally, the deeper issue: Should Wikipedia have a consistent citation method? If so, what?

The fourth is fixed in this guideline (I think) and could be fixed elsewhere. The fifth can't be fixed, of course, but wouldn't be an issue without the first four. As for the first three, they can't be fixed at this point without a truly gigantic effort. Do you, Bzuk, propose that we begin the process of changing any of these? And finally, I think the last issue deserves an organized discussion because there are many arguments and counter arguments to consider. I would recommend starting an WP:Essay, allow other editors to add to it and thereby discuss the issue directly and completely. ---- CharlesGillingham (talk) 17:47, 8 October 2009 (UTC)

Charles, back in Jurassic Times, I did try to involve editors in the forums and emails to look at references and basically was shown the door (trap door, I recall...). However, inexorably, changes have been gradually been incorporated in guidelines and more practically, in practice. The Wiki style that I eventually followed was to create a section titled "References" which refers to all reference sources that houses the following: "Notes" ("Citations", and clarifying Notes/Footnotes/Endnotes) as well as a "Bibliography" ("Works cited", "External links" and any other subdivision of media). Despite the rather fuzzy nature of choosing a particular citation/bibliographic "style", as long as a consistent format was adopted, the articles that I have authored, edited or collaborated with others that have received GA, FA and other categorizations based on peer review, were accepted, since Wiki allows a wide variance in formats. I am still not convinced of the value of having "For further research", "For further reading" as the terms convey the impression that "maybe" there is something worthwhile in these other resources. As for an essay that further elaborates these concepts, a good thought there... FWiW Bzuk (talk) 18:11, 8 October 2009 (UTC).

Some of you may have views that you wish to share here, as to which way we should go.--Epeefleche (talk) 10:20, 14 October 2009 (UTC)

Preventing dead links - idea

Maybe this was discussed before, if you know so, please point out where.

Regarding Wikipedia:Citing_sources#Preventing_dead_links, I think a pretty good way to do it would be to have:

  1. a bot that would look over all the Template:Citation templates and add automatically the complete value on the "archiveurl=" by first checking at and, maybe even google cache, and if not found to request automatically to archive the page then make a note of the archive URL.
  2. a bot to add automatically an empty "quote=" attribute (if it does not already exist), to encourage editors to fill that out. Thus giving precise context on the source of the information, and making sure that if the content of the page is lost, it can be still easily searched for a new URL.
  3. the same could be done with ref's, by first converting them to Citation

Since information is very dynamic on the internet, and loss of valuable information happens I think requesting automatically to arhive the weblink, would be a great solution. What do you think? --HappyInGeneral (talk) 22:20, 10 October 2009 (UTC)

Yup, this has often been brought up. See User:WebCiteBOT. -- œ 05:57, 11 October 2009 (UTC)
Excellent, Thank You! --HappyInGeneral (talk) 09:19, 11 October 2009 (UTC)

Say where you found the material

The old text was:

It is improper to obtain a citation from an intermediate source without making clear that you saw only that intermediate source. For example, you might find some information on a Web page that is attributed to a book. Unless you look at the book yourself, your source is the Web page, which is what you must cite, in turn making clear that the Web page cited the book.

The new text was:

If the cite is a convenience link, make this clear in the contents of the article itself. In such a case, as is frequently found in developing articles, it is improper to neglect to make the content appear as if the citation could be direct, when in actuality, the source is indirect (through an intermediary). For example, you might find some information on a Web page that is attributed to a book. Unless you look at the book yourself, your source is the webpage, and we are temporarily, and conveniently citing it. The clarification occurs in the article text, or in a footnote. As the cite source improves by being more direct, so will the content of the article improve in lucidity, (even if it only improves the lucidity of a footnote), for in place of the need to say "She says he says." type wordings, there will only appear in the article the fact itself (plus a small reference mark), as it should be.

I prefer the old text, as its clear, clean, short, and provides an unambiguous directive. I'm reverting to the older text. Fifelfoo (talk) 21:22, 12 October 2009 (UTC)

I must say I'm rather inclined to agree. In any case, I was under the impression that changes of any substance should be discussed here first, not made unilaterally out of the blue. -- Alarics (talk) 21:37, 12 October 2009 (UTC)
I'm fine with the current wording and don't believe there is any indication there that a "she says he says" type contruction would be necessary. Christopher Parham (talk) 12:32, 13 October 2009 (UTC)
There will be no contesting from me, of the revert. The revert will stay.
I will offer in another discussion (the way I might have proceeded initially), my assertions. Thank you. CpiralCpiral 02:00, 14 October 2009 (UTC)
There may be some misunderstanding of the main point here. The main problem to solve is not "I read the book at, and it's just barely possible that there's a critical typo here", but "Joe's blog said that the news story said... -- oh, but Joe's blog isn't a reliable source, so I'll just say I read the actual newspaper article. I'm sure Joe didn't misrepresent anything."
The newer version appears to be rather more oriented towards convenience links than to the "don't trust: verify" purpose here. WhatamIdoing (talk) 04:05, 14 October 2009 (UTC)
Thank you all. I will not edit any "conceptually unclear" policy page while reading it for the first time. I'm new, and I must apologize for editing a policy page before discussing. (Hereafter the "I'm new." excuse will not be used by me for any reason.) Signed CpiralCpiral 00:58, 15 October 2009 (UTC).

Wikipedia maligns former president obote

I am concerned that the first paragraph in wikipedia about obote carries falsehoods and mudsling material about former Ugandan president Dr. Apollo Milton Obote. As a father of my country and a man under whose leadership Uganda acquired instruments of independence he deserves better description than given here. The authors of this information should have done more research before unleashing political information that offers nothing more than propaganda value. It is common knowledge here that under Obote's leadership infrastructure was built in the country, social services offered to all without discrimination and the economy grew by leaps and bounds. His first government was put to an end by colonialists whose apartheid policy, mainly in southern Africa, the Mulungushi club to which obote was a member had vigorously opposed. He democratically regained power in 1980 despite the machinations by the neocolonialists and their comprado allies who aided selfish men like Museveni to overthrow his second government. These and many other basic facts on the political history of Uganda should have been fully captured by any balanced researcher offering to feed information to the world. Thank you

Moses Nuwagaba ( Tel:+256753462152 —Preceding unsigned comment added by (talk) 16:04, 14 October 2009 (UTC)

This page is completely the wrong place to raise this matter. Nothing stops you from editing the article about Milton Obote provided you can cite reliable sources to support what you write. -- Alarics (talk) 16:18, 14 October 2009 (UTC)
I suggest you discuss the matter on the article's talk page. — Cheers, JackLee talk 16:36, 14 October 2009 (UTC)

Several quotes from one source in one paragraph

If a paragraph summarizes a book review, say, and in doing so uses several literal quotes, duly placed between quotation marks, and it is clear that these are all quotes from the book review under discussion, is it really necessary to add the same citation again and again for each quote? For the purpose of verifiability, it would appear that one citation for the book review is sufficient.  --Lambiam 22:35, 2 November 2009 (UTC)

yes, a single quotation at the end of a paragraph is sufficient indication of sourcing and thus verifiability if only one source is referred to. If two sources you can get away with Fred says, Jane disagrees type comparative style and cite both at the end of para. Any more and you want to use more detailed referencing. If the claims are dubious or large, individual page references would be required rather than large page ranges. Fifelfoo (talk) 23:30, 2 November 2009 (UTC)

Which page number in pdf?

I did a search in the archives, but didn't see this addressed. I'm citing a pdf, specifically this one

The page numbers of the underlying thesis don't match the page numbers of the pdf. For example, page 76 of the thesis is page 83 of the pdf.

Which page should be included in the citation? I can think of arguments for both options; wondering if a convention has been established, or if there is some obviously preferred choice.--SPhilbrickT 15:17, 7 November 2009 (UTC)

Usually, PDFs are also available in paper form, and the Wikipedia citation usually gives enough information for readers to obtain the paper form. The reader of the PDF can use either kind of page number (although the PDF page numbers are easier to navigate with). The reader of the paper document has no way to know how the pages would have been numbered in the PDF. So I favor using the underlying page numbers. --Jc3s5h (talk) 15:48, 7 November 2009 (UTC)
That's what I used, and that's why I used it. (I agree pdf numbers are easier to navigate, especially for a paper using roman numbers or appendix numbering).Thanks--SPhilbrickT 16:20, 7 November 2009 (UTC)
Both. But the primary page numbering should be based on the paper pages, and the PDF numbered pages included as a convenience as in pp. 110-112 [pdf pages 2-4] or pp. iv-vii [pdf pages 4-7]. If you're only going to provide one, provide the paper page numbers. Fifelfoo (talk) 16:39, 7 November 2009 (UTC)

Google Scholar Addon/cite template section incomplete/inaccurate. And can't be editted.

Clicking on the Google Scholar add-on tools takes one to a Google Scholar page. (Not an addon page, as expected). Since I'd already found the paper I wanted to cite, & just wanted a quick and easy tool tool to do the formatting for me, I thought this would be prefect. The page even told me that all I'd have to do, was click on the wikify link, to create the citation. Just what I want!!

Except... it's too good to be true. I cut'n'pasted the title, & got a nice search result. But no wikify link. (I have a screen snapshot of it, but I can't upload the file.) I tried all the Scholar preferences, but none of the citation references even looked like a wiki citation template. (Yes, I'm running Firefox 2: it's not my machine to upgrade, and some addons the owner uses won't work with FF3 (even now!).)

Then I tried looking for the edit link, to edit the project page to indicate it might not work for all users (at very least indicate it's only in Beta), but... no edit link. (Strike 2.) Hence this post here.

Google'ing the Universal Reference Formatter, got me to a page I could fill in manually. (Like.... why bother with that over an editor?) And some user's page indicating that you need Greasemonkey installed and some additional userscript to get the wikify link from Google scholar. (But no indication why this isn't mentioned on the official cite tools page: for all I know, it's a virus lurking....).

Since I wouldn't be able update the main page with the correct instructions even if they do work, I'm giving up on these tools. (And, no, I'm not going to ask that the machine get upgraded.)

But... someone with edit privs should update the main page with the CORRECT information (whatever the heck that is, but it's sure not what's there now). —Preceding unsigned comment added by (talk) 21:50, 11 November 2009 (UTC)

Just tested it myself and it didn't work for me either (and I have FF3). Looks like the tool just isn't working right now, but I'll not remove from the page without some additional comments on the matter. Anon, sorry this problem wasted your time, but sometimes these tools just stop working. Huntster (t @ c) 03:54, 12 November 2009 (UTC)

Question about proper citation format

Suppose you want start out by writing something like: Source challenges the assertion that John Doe travelled to Europe.[1]

After this sentence, suppose the editor continues with: According to Source, "John Doe traveled to South America, not Europe."[1] As proof, Source shows a picture of John Doe meeting his mother at the airport in Buenos Aires.[1]

My question is, assuming all of these citations point to the same page within a book, is the proper style to cite with the [1] after each of the three sentences, or only after the first sentence, or only at the end?

Also, I'd like to know if this is the proper place on Wikipedia to post such questions, and if not, where I should post it. Thanks. Printer999 (talk) 21:22, 13 November 2009 (UTC)

If we were following Univ. of Chicago Press recommendations, a single citation at the end of the paragraph (I presume these sentences form a single paragraph) would be appropriate.
In the dynamic Wiki world, especially in controversial articles, there is an unfortunate (IMHO) tendency to cite each element separately.
I'd recommend a single citation, located at the end of the passage, unless the individual elements are likely to be challenged separately. --SteveMcCluskey (talk) 22:31, 13 November 2009 (UTC)
I generally agree, subject to the point that there appears to be consensus that all quotations need to be separately referenced, even if the reference is to the same citation that appears at the end of the paragraph. — Cheers, JackLee talk 04:47, 14 November 2009 (UTC)



You actually don't need the quotation marks (") in <ref name="name"/> and <ref name="name">. (talk) 10:32, 21 November 2009 (UTC)

That only works if the reference name is a single word. If it is a phrase such as "my name", then quotation marks will need to be used. — Cheers, JackLee talk 11:05, 21 November 2009 (UTC)
heh, I've been meaning to ask about this for ages. Thanks for clearing that up!
V = I * R (talk to Ω) 12:08, 21 November 2009 (UTC)
Tlping request because it has already been handled(explained). BejinhanTalk 13:03, 21 November 2009 (UTC)

Change this page's status

This page is not a style guideline. It basically says that editors can use any reasonable citation style, and gives no guidance at all. The whole purpose of a style guideline is to keep articles consistent, which this fails at, miserably. At best it keeps styles within each article consistent. If anything, this page should be a behavioral guideline as the only real guidance it gives is the sentence forbidding changes to the citation style of an article. Mr.Z-man 01:51, 22 November 2009 (UTC)

I tend to agree, not so much because the page lacks style guidance but because I have always seen it as a subtopic of WP:V, not WP:MOS - especially the top half of this page about when to cite and what citations should include. I might put it under "editing guidelines" rather than "behavioral guidelines". Christopher Parham (talk) 02:28, 22 November 2009 (UTC)
If this page were not the citation style guideline, there would be no citation style guideline at all, and the situation would be the same regarding lack of uniformity. The reason there is no prescriptive guideline mandating a particular style of references is the same as the reason for WP:ENGVAR: people simply don't agree over which style is best. So, just like ENGVAR, we only try to keep articles internally consistent, rather than forcing everyone to adopt a single uniform style. — Carl (CBM · talk) 03:21, 22 November 2009 (UTC)

Regarding The Guardian/

When citing articles released on the above website by the above newspaper - such as this, how should the "publisher" parameter of Cite web be written? I'm sort of split between the following:

etc.? There are a number of variations and I would love if someone could clear this up for me. Thanks, SteelersFanUK06 ReplyOnMine! 01:43, 27 November 2009 (UTC)

This is a style selection issue, which can vary from article to article, but should be constant within an article. Publishers are not normally given for Newspapers in respectable styles; they're very rarely given for Journals; and can often be given for obscure magazines or periodicals. My recommendation, regarding Teh Gridanau, is that being a newspaper it :doesn't need its publisher. Unless it was an obscurely named supplement, say Things about horses ( Fifelfoo (talk) 01:58, 27 November 2009 (UTC)
The "Things about horses" line threw me off a bit. So drop mention of "The Guardian" and just use ""? --SteelersFanUK06 ReplyOnMine! 02:32, 27 November 2009 (UTC)
If it is in The Guardian you shouldn't be using "cite web" in the first place, but "cite news". The "publisher" parameter is left blank, and "The Guardian" goes in the "newspaper" parameter. Preferably also put "London" in the "location" parameter, since there are other Guardians elsewhere. -- Alarics (talk) 08:45, 27 November 2009 (UTC)
Are they no longer the Manchester Guardian for some stupid reason? Fifelfoo (talk) 08:50, 27 November 2009 (UTC) Oh.  :( Fifelfoo (talk) 08:52, 27 November 2009 (UTC)
No, The Guardian dropped that in 1959. Martinvl (talk) 09:42, 27 November 2009 (UTC)
And in 1964 its main editorial office actually moved physically from Manchester to London. -- Alarics (talk) 14:24, 28 November 2009 (UTC)

Contentious statements

A flurry of recent changes, conducted without discussion, made (among other things) the following edit to Qualifying references:

"For statements about which reliable sources are in conflict or that are matters of differing opinion, the text should include sufficient context to qualify the statement or attribute its source. But, for contintious statements, often when references are in conflict, the text should qualify the statement."

This change goes too far. In-text aqualification is not required for any contentious statement: otherwise, in-text attribution would be required for claims such as "Humans evolved from other animals" that are not controversial among reliable sources on evolution, but which are controversial in some religious circles. Wikipedia articles don't (and don't need to) use in-text attribution for claims opposed only by unreliable sources. (The change also contains a misspelling, but that's a minor problem.) Eubulides (talk) 20:37, 3 December 2009 (UTC)

I concur with your reversion although in the context of the revisions I don't believe any change in meaning or practice was intended. I think you provide a good rationale for the additional wording in the original version. Christopher Parham (talk) 20:39, 3 December 2009 (UTC)
Thanks. Come to think of it, why does the longstanding wording say "or that are matters of differing opinion"? If reliable sources agree on a topic, surely there's no need to use in-text attribution. Otherwise, a contrarian editor could say "It's just the establishment's opinion that humans evolved from other animals, so any such claim in Wikipedia about evolution must use in-text attribution." Eubulides (talk) 20:40, 3 December 2009 (UTC)
As with the case you point out, I think that the advice needs to be taken with a grain of common sense, but I think the idea is that it applies on "matters of opinion" - I wouldn't usually put issues of scientific consensus under this category. Especially where there are very few reliable sources commenting on a topic, I might want to qualify statements even where they agree, if it is a matter of subjective opinion. Christopher Parham (talk) 21:39, 3 December 2009 (UTC)

Citing BLP-controversial stories from British newspaper websites

FYI there's a potential problem—I don't know how often it happens—if BLP articles use British newspaper websites to support controversial claims that haven't yet appeared in print. The risk is that we cite their story and they then amend it so we no longer have cover for a libellous allegation. The potential problem is that some sites (e.g.,,,, now flag their content as "NOARCHIVE", so the publisher can amend a story at any time and there's no official way an editor can snapshot their page as proof that a claim appeared there. Given that Google seems to track our changes within minutes (example) this might be important in future. Possible solutions are (A) use – and archive – publishers like, or who don't currently "NOARCHIVE" their content; and/or (B) instead of {{Cite news}} use {{Cite web}} with a full accessdate=yyyy-mm-dd hh:mm:ss parameter so there is some way to trace back from the citation to the web version cited. This problem will only get worse now that Rupert Murdoch is arguing (BBC report) that those he calls "aggregators and plagiarists" should "have to pay a price for the co-opting of [news providers'] content". - Pointillist (talk) 00:00, 28 November 2009 (UTC) BTW, if this should have been posted to Wikipedia:Reliable sources/Noticeboard or Wikipedia talk:Reliable sources please accept my apologies and move the discussion without consulting me. Thanks - Pointillist (talk) 00:08, 28 November 2009 (UTC)

Citing fee-sources: I want to evolve around the troubles mentioned by Pointillist, i.e. using technically unreliable and unavailable sources that must be controlled but who provide technical obstacles. My troubles occurred when browsing Formation and evolution of the Solar System specifically Timeline of Solar System evolution. I wanted to control the numbers but found a "Douglas N. C. Lin (May 2008). "The Genesis of Planets" (fee required). Scientific American 298 (5): 50–59., which made me wish to add a second source that wasn't requiring fees. The troubles with using fee-required-sources are obviously
  • only those having a fee-account at the site can control the source, which practically makes the source unavailable to less financially able (poor) folks, or people having less general interest in the payment site,
  • it borders to kind of marketing a payment site, if no free backing sites are available, for anyone to read and learn, if Wikipedia live on charity, why should profit sites make a lot of money from free marketing?
I don't say that we shall avoid payment sites entirely, just that they should preferrably be backed by free sources. If we use too many payment sites, it negatively affects the general culture and philosophy of Wikipedia: freely available to everyone, all editors, irrespective of background, nationality and social class, attain integrity and "status" by Wikipedia achievements only, not by external achievements. (The random Burkinabé noneducated farmer guy should count equal to any random university level Swede).
Now, my troubles and Pointillist's troubles can partially be addressed by the fact that we are referring to secondary academic sources, not primary research reports, but the primary research reports are not seldomly hidden behind a payment site too, so I propose a few extra guide lines, such as:
  • a payment site requiring login, should if possible, be backed by free sites,
  • it is then desirable that the editor adding the payment reference verifies that the free sites tell us the same story as regards to the article text using the reference,
  • the backing sites should be marked as backing sites, so that the cleanup-guy doesn't in good faith remove the backing site, while keeping the payment site; the backing site should preferrably be contained within the same <ref></ref> in order to achieve this,
  • it would be a good thing to have templates displaying [reference requires fee — please add a backing site]
  • and for Pointilist's case [reference provides NOARCHIVE content — please control the reference against the article] or [reference has variable content — please replace with a stable contents reference]
(Except of course that those messages must be much shorter).
Opinions?! ... said: Rursus (mbork³) 11:14, 28 November 2009 (UTC)
I would hope the libel problem that Pointilist raises is pretty theoretical. Is it likely in practice that Wikipedia might get sued for good-faith quoting of something that appeared on a respectable newspaper website, even if the latter subsequently disappears? Has such a thing ever actually happened? If it is really a problem, why not just take a screen grab of the web page (free software that does this very easily includes "Siteshoter") and upload it to Google docs, thus bypassing the NOARCHIVE problem. But it might also be a good idea not to quote on WP anything that seems potentially libellous until you have definitely seen it in print.
On pay sites, one can always cite a newspaper source without it being a web link. If Murdoch starts requiring payment for online access to articles in The Times, as long as the article can be assumed to have appeared in the printed paper one can still cite it, just as one already does with pre-1985 Times articles, which have always been fee-paying (unless you have the right library card, i.e. it is your library that is paying the fee). Of course in many cases one can find the same thing reported in some other source (e.g. The Guardian) which is free, and that would be preferable. -- Alarics (talk) 14:52, 28 November 2009 (UTC)
For this reason, anytime that an editor uses a web source, they should provide an access date in the citation. As for pay sites, I don't think there needs to be any restriction in that regard, so long as the sources are reliable. Christopher Parham (talk) 15:28, 30 November 2009 (UTC)
  •  ?. I just successfully archived a Times [7][8] and Telegraph [9][10] page with WebCite. I wouldn't bother worrying about tabloids, they are unlikely to ever be considered RS for BLPs. MickMacNee (talk) 12:54, 5 December 2009 (UTC)

Dealing with citation problems

I have noticed a number of articles that have a citation needed flag which at times I think is "over the top". If a statement is made that can easily be verified by an independent third party, is a citation really needed? I recently removed a citation needed flag from an article where I was able to verify the fact inside five minutes using Google Earth, but I would not know where to start if I were to try and find a real citation for this particlar item. (Or should I have replaced the Citation Flag with coordinate references and a suggestion to use Google Earth?) For the record, I had no prior knowledge of this particular fact nor had I ever visited the country concerned. —Preceding unsigned comment added by Martinvl (talkcontribs) 21:12, 29 November 2009 (UTC)

Do what you can by adding a citation to the coordinates. It would be preferable to look around on Google Books, GScholar, or GNews, for an actual citation which might stand up to more than "oh, this is there", as yours is likely to be removed for that reason, but it's better than nothing. You can leave the T:citation needed as well (though I think there's another in that family of templates which suggests that a stronger citation should be found). --Izno (talk) 17:36, 7 December 2009 (UTC)

Replacing duplicate footnotes with named footnotes

Note: Discussion was started by User:Hegvald at the Village Pump (Policy) page on 17:43, 14 November 2009. User:CBM started a new discussion at Wikipedia talk:Citing sources later that same day. The two discussionw went on in parallel for a few days until CBM moved the original discussion to this page. It is now in the section in the blue box below. --Hegvald (talk) 11:05, 22 November 2009 (UTC)

Link to discussion from July 2009

Btw, the initial discussion on list-defined refs, with good points made on both sides, is at Wikipedia_talk:Citing_sources/Archive_26#Improving_.3Cref.3E. - Dank (push to talk) 17:16, 15 November 2009 (UTC)

Refname footnotes [start of discussion, from the Village Pump (Policy) page]

Moved from Village pump (policy) to consolidate discussion — Carl (CBM · talk) 17:50, 19 November 2009 (UTC)
Moved to top of section and put in coloured box in order to clarify chronology. --Hegvald (talk) 10:58, 22 November 2009 (UTC)

I don't like these. At all. So far I have had to revert changes to the Charles Boit article three times, after clueless people or bots changed the referencing, apparently taking for granted that this system is somehow inherently and obviously superior. It is not. It introduces unnecessary complexity to the wiki "code" and makes it more difficult to edit and source articles.

I just have to ask: is there really a general consensus that these are good and useful, or is it just that a bunch of computer geeks like them and use their bots and scripts to force them on everyone else? --Hegvald (talk) 17:43, 14 November 2009 (UTC)

I don't like them. It makes the footnote numbers out of order within the article. It prevents people from combining multiple supporting references for the same sentence into a single footnote. And it prevents (or at least discourages) annotations from being added to each footnote to explain or elaborate upon how the reference supports the statement in the article. I think its only benefit is labor-saving, which should never trump other concerns. It's one thing for an article's editors to agree to use them in a given context (though I would prefer a guideline discouraging it). But we certainly shouldn't have bots running around imposing them on every article. postdlf (talk) 17:53, 14 November 2009 (UTC)
personally, I completely disagree with the criticisms of list defined references here. I mean... I just find that everything said in criticism of them is simply not true, in my view. That being said opinion in this area is decidedly split. I think that we're going to have to "impose" a rule on everyone not to change the existing system on pages, due to the fact that opinions are so split. Those of us who like them can add them to articles that are created, and those of you who don't like them will simply have to live with it when you do come across them.
AWB and Ssmackbot (along with all of the other AWB bots and users) are a fairly serious problem, in relation to LDR's, however. I've been tempted for a while now to bring the issue up and if necessary to temporarily de-authorize AWB's use site wide until they get up to speed with the change. That a tool is not up to date with the site software is a completely unacceptable situation, to me.
V = I * R (talk to Ω) 18:19, 14 November 2009 (UTC)
  • Hey, wait... you're just talking about using the name parameter? what in the world is the problem with that? Using the name parameter doesn't do any of the things that you guys are talking about... I don't get it.
    V = I * R (talk to Ω) 18:24, 14 November 2009 (UTC)

Regardless whether these new footnotes are better or not, no editors should be going around in an automated way switching articles from one referencing style to another. As WP:CITE says, "You should follow the style already established in an article, if it has one; where there is disagreement, the style used by the first editor to use one should be respected." It is extremely uncommon that there is a need to change an article from one referencing style to another. — Carl (CBM · talk) 18:27, 14 November 2009 (UTC)

It turns out that no one is. I'm not 100% clear on what Postdlf is complaining about, but Hegvald at least is simply talking about using the name parameter with references. I don't really understand his complaint, since the name parameter is often essential to the proper organization of references in this sort of instance, so I'm hoping that he'll follow up with more info on his position here.
V = I * R (talk to Ω) 18:29, 14 November 2009 (UTC)
Actually SmackBot seems to be doing this [11]. I want to wait a little while, but I may block the bot temporarily if it starts a new task that includes this non-feature. — Carl (CBM · talk) 18:34, 14 November 2009 (UTC)
Well right, but it's doing that because that's what WP:CITE says to do. That's the way that references are supposed to be formed, if they are able to be.
V = I * R (talk to Ω) 18:37, 14 November 2009 (UTC)
WP:CITE says to use the style already established. If the established style does not use this sort of named footnotes, then switching to it is a change. Later, WP:CITE says, "Optionally, one may add the name attribute by using <ref name="name">details of the citation</ref>. Thereafter, the same footnote may be used multiple times by adding <ref name="name"/>." This makes it more clear that using named references is an option that a particular page might or might not adopt. Automated processes such as bots, in particular, should not be changing references from one optional style to another. — Carl (CBM · talk) 18:42, 14 November 2009 (UTC)
"Optionally" here refers to the fact that a name paramter can be added to any ref, including non-repeating ones. This we did at Fort Hood shootings for example, where we waanted a clear scheme for managing the references. AWB does not add names to non-repeating refernces. Rich Farmbrough 19:13 14 November 2009 (UTC).
Since one can simply repeat the same reference twice, using the name parameter even for that sort of reference is also optional. In general people should not be making widespread changes to referencing styles, whether with AWB, by bot, or by hand. — Carl (CBM · talk) 19:19, 14 November 2009 (UTC)
we're not talking about style here though, we're talking about correct usage of cite.php. WP:CITE is talking about changing {{ref}}, {{note}}, etc... template use to use cite.php references instead. There is clear and widespread consensus that organizing references to eliminate duplicates, with the added benefit of reducing the size and adding organization to reference lists, is a maintenance task which we should all accomplish whenever we are able to. AWB added the ability to add the name parameter to references more than a year ago, with broad support to do so, and that feature is now a part of the "general fixes" that it will run all the time. All of this is documented on the AWB pages, in the bot working group, and someplace on cite (at least, it used to be. Maybe someone has changed it recently?)
V = I * R (talk to Ω) 18:50, 14 November 2009 (UTC)
If the editors of an article feel that the citation style of the article includes not using the "name" parameter, then nobody should be adding the "name" parameter, automatically or otherwise. Apparently Hegvald and postdif feel this way. This is the "clear and widespread consensus" about not making changes from one optional style to another that is described both at WP:CITE and at WP:MOS. As I pointed out on user talk:SmackBot, I didn't find any bot approval for reformatting references. Editors using AWB manually are a separate problem; I often see them make changes that they should not be making, although I only revert them on articles that I already follow. — Carl (CBM · talk) 19:00, 14 November 2009 (UTC)

Well Hegvald does have a point becasue the refs in that particular articlee are things like "Smith p.34" so replacing them with name=Smith p.34 is of marginal if any benefit. I will drop a feature request for AWB to ignore refs that wil be equivalent to theeir name. Rich Farmbrough 19:25 14 November 2009 (UTC).

I have to side with User:Ohms law on this one, it's not changing the style, it's just replacing duplicate refs with a named ref. I would think that'd be a good thing because if a ref needs updated you don't have to go through and update 20 odd references. Q T C 19:31, 14 November 2009 (UTC)

That is a change in style nevertheless, as repeating each reference whenever it is used is a plausible choice that can be made by the editors of an article (and was, apparently, made by the first two editors in this thread). Can you point to any bot approval for SmackBot to change references? That seems like exactly the sort of thing that requires human discretion and would not be given approval as a bot task. — Carl (CBM · talk) 19:37, 14 November 2009 (UTC)
{e/c}While I generally agree that the existing style should be maintained throughout the article, I take exception when the exact same footnote is being used in multiple places. The name parameter should be used if using the same source more than once. This is not only for whatever minor performance issues there may be (duplicating the reference adds unnecessary length and file size to the page), but for maintainance reasons as well. Dealing with dead links, archived versions, malformed text or other problems in a reference should not need to be done multiple times on the same article. Although in this particular case it's seems almost moot because of the use of shortened notes+references system. But this is not a change in style, simply adding a parameter that is used in the current style.Jim Miller See me | Touch me 19:40, 14 November 2009 (UTC)
If this is correct, the wording of WP:CITE should be changed to require, or at least encourage, named footnotes when there are duplicates. At the moment the matter is treated as completely optional. The fundamental reason that we avoid changing from one optional style to another is to avoid having people discuss back and forth about them. If replacing duplicates is clearly worthwhile, WP:CITE should reflect that. However, WP:CITE has historically been written to give editors an extremely broad amount of freedom about exactly how to handle references in each article. — Carl (CBM · talk) 19:44, 14 November 2009 (UTC)
I'll start the discussion at WT:CITE. — Carl (CBM · talk) 19:45, 14 November 2009 (UTC)
I don't think this should be discussed anywhere else. I suspect that whatever imagined consensus people may have thought existed is based in the fact that this has previously mostly been aired among people who are already mostly interested in the technical aspects of referencing. It is likely reach a wider group of people right here.--Hegvald (talk) 20:08, 14 November 2009 (UTC)

The issues Jim Miller mention are better served by using an alphabetic list of references at the end of the article for complete bibliographic data on the sources. Then, if an URL or something changes, you only have to change this once. In any case, an article published in 1925 or 1933 is not usually affected by things like dead links. Even if a link to an on-line facsimile goes dead, the Wikipedia article should contain enough bibliographic information to find the original print publication in a library. The text remains the same.

That said, I will allow for different needs in different types of articles. An article on a recent event may well need a different technical solution from an article on an 18th century painter. I also think this may be a science vs. humanities issue. The only place I have ever seen this type of non-consecutive footnoting outside Wikipedia was in a science journal. In publications in the humanities I have never seen it. --Hegvald (talk) 20:08, 14 November 2009 (UTC)

The only slightly unusual aspect to the article in question is that it uses Harvard style referencing. APA/Chicago style references are much more common on Wikipedia, but Harvard style referencing is hardly unheard of. In reality, with a bit of familiarization with the citation system I think that you'll realize that the group and name parameters are there to help you. You can create nice and compact reference lists using groups and names, and they then behave in a manner that you seem to be touching on in your description of an "ideal". Smackbot and Drilbot weren't making the optimal change, but they were trying to help at least. For a live example of grouped and named Harvard style referencing, take a look at the bibliographical portion of the references section in Moon landing conspiracy theories.
V = I * R (talk to Ω) 20:18, 14 November 2009 (UTC)
Sorry, I don't follow you. Which article are you referring to (the one with a "slightly unusual aspect")?
As for the Moon landing conspiracy theories article - are you serious? That article has seven different sets of footnotes, each with its own numbering! And it uses those hideous source templates in the middle of the text! Can you imagine how difficult most people will find it to make changes to an article like that? Perhaps is a good ting to make it difficult to edit an article on conspiracy theories, but in most cases this would not be the case. The average retired professor of art history wanting to make a small referenced correction to an article would just give up if he opened the edit window and encountered something looking like this.
I am well acquainted with footnotes and how they are used outside Wikipedia. I like them just the way I am used to everywhere else. I don't want Wikipedia-specific referencing systems. Don't try to tell me that I really should like your preferred system. There are good reasons why footnotes in a normal academic publication do not look like those in Moon landing conspiracy theories, and why adding footnotes in a normal word processor is as simple as it usually is. --Hegvald (talk) 20:41, 14 November 2009 (UTC)
Are the "hideous source templates" that you mentioned {{cite web}}, {{cite news}}, etc...?
As for the later comment, I think that you'll find that putting aside your personal preferences in favor of working with the Wikipedia community norms will lead to a happier time here. You're far from the only person with strong personal feelings on issues, especially when it comes to referencing, but we all have to work together here. The community norms on Wikipedia have developed over years, and tons of people have collectively decided to use the rather unique style that tends to predominate on Wikipedia. You'll likely be able to impose your own preference on an out of the way article such as the Charles Boit article for quite some time, as long as you're actually active, but eventually the wider community standard will wash over it.
V = I * R (talk to Ω) 20:52, 14 November 2009 (UTC)
OK, I realize that it is fairly pointless to discuss with someone setting themselves up as an incarnation of the "Wikipedia community norms". A small number of bots and their keepers do not represent the "Wikipedia community norms", and neither do you. --Hegvald (talk) 21:05, 14 November 2009 (UTC)
The WP:CITE guideline probably does more to make articles harder to edit than any one thing on Wikipedia. By encouraging the use of half a dozen different reference formats and not allowing changes, our articles are a complete and utter mess when it comes to references. We've developed ways to make references less distracting when trying to edit (named refs, list-defined refs), but our ability to use them is hampered by poor policy. The whole purpose of the MoS is to keep article styles consistent, but for some reason, we prefer to do the exact opposite when it comes to references. Mr.Z-man 22:40, 14 November 2009 (UTC)

Using named references is not a change in citation style; it is how you are supposed to use the software. Duplicated references should be transformed to uses of a named reference at every opportunity. The reasons why are detailed extensively in a variety of places. OrangeDog (τε) 16:11, 16 November 2009 (UTC)

Use of named references to consolidate duplicates is described as optional in the guideline, and this is intentional. Naming references changes the final appearance of the article and there is difference of opinion as to whether it is an improvement. In articles that use short footnotes, there is also usually little saving of space (often AWB will add to the length of such articles). Christopher Parham (talk) 14:25, 18 November 2009 (UTC)

    • "I like them just the way I am used to everywhere else." only makes sense if you are used to them in a single subject field with well-defined and universally-followed conventions. In most of the subject fields I am familiar with, the way they are used differs widely from journal to journal, and the only Wikipedia style conforming even roughly to the way they are used elsewhere is Harvard, which in the RW is very rarely found outside the humanities & which most Wikipedia editors would find very cumbersome. Most librarians are prepared to teach several different of them, and the first question we ask a student who wants help about this is "who is your professor--did he give any handouts on what he wants?" DGG ( talk ) 00:11, 19 November 2009 (UTC)
DGG, I already said above: "I also think this may be a science vs. humanities issue. The only place I have ever seen this type of non-consecutive footnoting outside Wikipedia was in a science journal. In publications in the humanities I have never seen it." There may be good reasons for this, such as the way footnotes are used in different fields, or even the average length and scope of publications in the sciences vs the humanities. (And I am not talking about a single subject field, but a wide range of fields within the humanities: history, art history, literature etc.) In this case, we seem to see science/engineering people using bots and scripts to impose their own preferred style on everyone else.
In any case, I have asked User:CBM who started the new discussion at the CITE page to move this entire discussion over there. We shouldn't have two parallel discussions and be forced to repeat the same arguments in both places. --Hegvald (talk) 16:45, 19 November 2009 (UTC)

Named references [Start of discussion at Wikipedia talk:Citing sources]

It has been suggested that WP:CITE should have more firm guidance about replacing duplicate footnotes with the same content with multiple references to the same, named footnote (as in this edit). Right now, WP:CITE says,

"Optionally, one may add the name attribute by using <ref name="name">details of the citation</ref>. Thereafter, the same footnote may be used multiple times by adding <ref name="name"/>."

Apparently, AWB automatically implements this "optional" aspect, thus changing the referencing style in articles where duplicate footnotes were intentionally kept separate. Should the policy here say that named footnotes are recommended? Required? — Carl (CBM · talk)

As I noted on WP:VPP, the inconsistency encouraged by this "style guideline" is frankly, terrible. I would strongly support this, and any change that helps us move toward making articles easier to edit (as refs are one of the most confusing aspect, and having half a dozen or so different options just makes it worse). Mr.Z-man 22:43, 14 November 2009 (UTC)
When refs need to be repeated multiple times in an article, it is largely in the interest of editors to use named refs. This should be encouraged. Dragons flight (talk) 00:16, 15 November 2009 (UTC)
Clearly not in the interest of all editors. --Hegvald (talk) 01:51, 15 November 2009 (UTC)

My problem with list-defined references is that they're meant to make the editing screen easier to read, but Harvard refs, aka shortened refs, (<ref>Harris 2006, p. 4.</ref>) have less distracting code than even the list-defined refs, so I prefer Harvard refs. OTOH, most of my references are books that I cite more than once ... people who tend to use only one citation per reference sometimes say that Harvard references and list-defined references are extra work. Maybe there's some script that could save keystrokes, I don't know. - Dank (push to talk) 14:29, 15 November 2009 (UTC)

Dank this isn't about LDR. Just names. Rich Farmbrough, 13:58, 18 November 2009 (UTC).
I use shortened footnotes a lot, and often choose ref names which work for article editors like Harvard refs, e.g., <ref name=smith1996pp5-6 />. I haven't been using LDRs to move the ref declarations out of the article prose, though. I'm currently working up a new article here; perhaps I'll give LDRs a try there and see how they "feel" style-wise using this ref naming style. Wtmitchell (talk) (earlier Boracay Bill) 00:43, 23 November 2009 (UTC)

Discussion location

Is it really a good idea to have this discussion in two places? Wouldn't it be better to keep it at the village pump, where it is more likely to be seen by more people? --Hegvald (talk) 01:51, 15 November 2009 (UTC)

Generally, I prefer a pointer from VPP to whichever is the appropriate guideline or policy page, because people will actually be able to find previous conversations about citations if you keep them at WT:CITE. - Dank (push to talk) 14:29, 15 November 2009 (UTC)

Postdlf's three arguments against named references

I'm strongly opposed to any mandatory imposition of named refs. The only virtue I can see in them is that they condense the footnotes section of an article, but that certainly shouldn't come at the expense of clarity within the body of an article, or at the expense of more information being provided. I think ref names have at least three main drawbacks.

1. Because all cites to the same ref name will display the same numbered superscript link (i.e., footnote) within the body of the article, the use of ref notes will cause the numbering of those footnotes within the article body to be confusingly out of order.
2. The use of ref name for all duplicate references discourages editors from combining multiple cites for the same statement of fact into one footnote; this is commonly seen in many articles, where one single sentence will end with half a dozen or more footnotes. This causes distracting and confusing clutter within the article body (particularly given the inconsistent numbering that results from ref name usage), all apparently for no other purpose than to condense the footnotes section. Which is not a sensible trade-off.
3. The use of ref name for all duplicate references discourages editors from adding explanative annotations (or often even specific page cites) to footnotes for separate citations to the same reference, where those citations are in support of different statements of fact within the article.

The first drawback is a simple fact, a consequence of how ref name works. The second and third drawbacks are easily observed in practice, even though ref name doesn't absolutely prohibit the individualization of some cites. But it definitely discourages it, whether because editors think they can't deviate from using a ref name for all instances of a particular cite once that's begun within an article, or because editors are simply more likely to lean on the crutch that ref name provides of relieving the need to retype or copy a full ref in a new place. And certainly editors are less likely to expand upon the content within individual footnotes when they aren't already maintained in separate form. postdlf (talk) 15:02, 15 November 2009 (UTC)

Discussion of Postdlf's three arguments against named references

1. Because all cites to the same ref name will display the same numbered superscript link (i.e., footnote) within the body of the article, the use of ref notes will cause the numbering of those footnotes within the article body to be confusingly out of order.
Not so, AWB can and does order cites so they appear not as[13],[3],[8],[1] but [1],[3],[8],[13].Rich Farmbrough, 13:58, 18 November 2009 (UTC).
I think the point is, rather, they will appear as [1],[2],[3],[1],[3],[2],[4]. That's an odd way to count to 7. Christopher Parham (talk) 14:19, 18 November 2009 (UTC)
Sorry, I don't follow the last comment. Why would an article ever cite exactly the same source twice at the same place? I've never seen a Wikipedia article that said anything like "It is possible that regressive autism is a specific subtype.[3][5][3]" and if I did see one, I'd change that trailing "[3][5][3]" to "[3][5]" without a second thought. Eubulides (talk) 20:02, 18 November 2009 (UTC)
Sorry, I assumed it was clear that there were intervening sentences each being cited; that's not one block of citations for a single fact. I mean A.[3] B.[5] C.[3] Christopher Parham (talk) 20:08, 18 November 2009 (UTC)
AWB, and all other automated tools that I am aware of, don't change the layout of references with are separated by text at all. Meaning, in the A.[3] B.[5] C.[3] example which you gave, AWB wouldn't change that layout at all. It may convert the second [3] reference to a named ref, which is important and a good change because a)it's shorter, and b)doing so prevents later editors from changing one ref without changing the others (it helps maintenance).
V = I * R (talk to Ω) 20:34, 19 November 2009 (UTC)
Yes, the problem is that it changes the second reference to a named reference. It's not necessarily shorter (oftentimes longer when the change is made by AWB). However, it is easy to revert AWB editors so I don't see this as a serious issue. Christopher Parham (talk) 22:27, 19 November 2009 (UTC)
I guess that I really don't understand the criticism of using the name parameter. It's not as though the parameter is a recent addition either, which actually starts me scratching my head when it comes to this discussion. The increase in maintainability and readability (both of the page source and the list of references) easily justifies the general use of the name parameter, though.
V = I * R (talk to Ω) 22:56, 19 November 2009 (UTC)
To be clear, this discussion as I understand it is about whether to make use of the name parameter for duplicate references mandated or encouraged by this guideline. I've not problem with it as an option for editors, where it is indeed quite longstanding. I don't really see much improvement in maintainability or readability that comes from use of this feature, and it has a number of features that can make editing more difficult, including separating the content of a citation from the text it references in the edit box (getting them together was one of the main innovations behind cite.php). I can see that if you are using citation templates, it might be a helpful feature to reduce the size of references. Sometimes it's useful, others not; I think maintaining its optional status is the best choice. Christopher Parham (talk) 14:30, 20 November 2009 (UTC)
2. The use of ref name for all duplicate references discourages editors from combining multiple cites for the same statement of fact into one footnote; this is commonly seen in many articles, where one single sentence will end with half a dozen or more footnotes. This causes distracting and confusing clutter within the article body (particularly given the inconsistent numbering that results from ref name usage), all apparently for no other purpose than to condense the footnotes section. Which is not a sensible trade-off.
I don't see what this has to do with the WP:CITE document and/or named references, though. I agree that having a single sentence (or even a paragraph) referenced to 8 different news stories get's to be ridiculous, but that problem isn't something which should be addressed through the reference name parameter at all really. An actual solution to that problem is to editorially select 2 or maybe 3 of the more important references and get rid of the others.
V = I * R (talk to Ω) 20:38, 19 November 2009 (UTC)
3. The use of ref name for all duplicate references discourages editors from adding explanative annotations (or often even specific page cites) to footnotes for separate citations to the same reference, where those citations are in support of different statements of fact within the article.
I understand this criticism, but I don't find it very convincing. That sort of practice really requires larger editorial planning and work in general, so I think that it falls a bit outside of the scope of this discussion. However, just to take this criticism at face value, I am generally willing to accept this as a tradeoff to using the name parameter in the interests of reducing lists of references to more managable numbers. What may be acceptable in small article with up to a dozen references isn't as good of an idea in an article with 120 references (which often could be more like 300 references if all instances of use of the name parameter were deprecated).
V = I * R (talk to Ω) 20:44, 19 November 2009 (UTC)
There is no reason for this to happen if combined cites are appropriate. In fact there is no reason a footnote cannot read:
{1] See for example Bloggs [12], chapter 4, LeBowski [13] especially the section on pigeons, and Bond - Annals of Defenstration, July 1999 pp228.
Doubtless the much requested "page number" fix would provide relief from these uncertainties too.
Anyway no-one is talking about mandatory, we are talking about names being provided for duplicate refs.
Rich Farmbrough, 13:58, 18 November 2009 (UTC).
I agree with Postdlf on all three points; the implementation of named references, particularly the way they are displayed in a non-intuitive manner that defies the conventions found in most printed and online matter (sequentially numbered footnotes), is not ideal. I support use of named references being optional and regularly revert AWB edits to add them when they pop up on my watchlist. Christopher Parham (talk) 14:27, 18 November 2009 (UTC)
You know, your "non-intuitive manner" is the standard convention in many scientific publications, including both Science and Nature. It's a choice of style and not a particularly unusual one. Dragons flight (talk) 20:14, 18 November 2009 (UTC)
I'm aware this is a standard in some fields. The use of the style for articles on scientific topics would seem eminently reasonable, if that is the style with which writers and readers in that field would be most familiar. Nobody is suggesting it not be an option. Christopher Parham (talk) 20:25, 18 November 2009 (UTC)
This is another "Oh I'm a professor, they re-used ref-1 after ref-4 my head is going to explode." type effect? Because the people who read these learned journals can only cope with increasing numbers, while we expect Little Johnny to grasp abbreviations like pp. ed. Ann. Phy. and so forth intuitively? Or is it the "ZOMG the people who write in APA style will never take WP seriously if we do thus and so?" Because if that's the problem - news flash - "Those that mind don't matter and those that matter don't mind." Storm meet teacup. Rich Farmbrough, 15:03, 19 November 2009 (UTC).
I am quite happy to call this a storm in a teacup and keep the current language in which the use of named references to consolidate duplicate citations is optional. Christopher Parham (talk) 16:28, 19 November 2009 (UTC)
I came late to this discussion and would want to be counted among those who want to keep the optional nature of named notes. I much prefer consolidating several references in a single note (sometimes with comments like Jones , p. 123 and Smith, p. 45 but Brown, p. 67 disagrees.) to a string of numbers at the end of a sentence. But then I'm a humanist (who also writes in the sciences and social sciences sometimes). --SteveMcCluskey (talk) 16:45, 19 November 2009 (UTC)
Personally I think it would make it easier for editors to move towards one (or a small number of) "recommended" styles rather than allowing editors to do whatever they want. Having an encyclopedia broken up into a dozen different styles makes it harder for editors working in one area to move into other areas. I have a personal opinion about what should be recommended, but I'd say having recommendations is more important than exactly what they are. Dragons flight (talk) 17:00, 19 November 2009 (UTC)

Basically, I think that we should essentially "do nothing" here. I agree that it would make things easier if we could all agree on one (or a small number of) "recommended" styles, but as is probably evident to most here already I do see any firm consensus developing on this topic (for some reason). The curious fact (to me) is that we already use our own house reference style, so it's not as though we're forcing people to choose some standard over another. The real issue is that our house referencing style is ad hoc I guess, but it seems to me that is at least somewhat intentional.
V = I * R (talk to Ω) 20:58, 19 November 2009 (UTC)

I personally am in favor of the style using names to group references, since it makes articles shorter. Not doing that makes sense in a multi-page printed edition, but not on a one-page-a-time website. That be as it may, I can understand the arguments against it, and therefore agree that automated tools should not interfere in this matter (unless it would be decided to use one style only). Debresser (talk) 23:38, 14 December 2009 (UTC)

[Why not?]

Wait, why wouldnt we want named references. Consider this:

Content Section

Lorem ipsum dolor sit amet,[3] consectetur adipiscing elit. Quisque et eros lorem. Nullam mi dui, euismod id ultrices ut, posuere id ante. Maecenas metus felis, bibendum in iaculis eget, pretium vitae turpis. Donec vel condimentum sem.[4] Quisque sit amet pharetra libero. Aliquam dolor urna, sagittis et bibendum sit amet, pulvinar sed sem. PellentesqueCite error: A <ref> tag is missing the closing </ref> (see the help page). nec porta nibh risus eu massa. Aliquam imperdiet fringilla metus eu aliquet. Maecenas convallis nulla id sem euismod suscipit. Etiam fringilla lorem placerat augue convallis cursus. Nullam quis lectus id risus volutpat sagittis. Integer quis turpis lacus. Praesent adipiscing nisl porttitor lacus gravida lacinia. Fusce lectus justo, rutrum [5]vitae consectetur sit amet, faucibus vitae lacus.


  1. ^ Delaney, Robert (November 8, 2006). "AMA Citation Style, American Medical Association Manual of Style, 9th edition". Long Island University C.W. Post Campus, B. Davis Schwartz Memorial Library. Retrieved 2008-04-16. 
  2. ^ "International Committee of Medical Journal Editors (ICMJE) Uniform Requirements for Manuscripts Submitted to Biomedical Journals: Sample References". United States National Library of Medicine work=MEDLINE/Pubmed Resources. Retrieved 2009-10-08. 
  3. ^
  4. ^
  5. ^

Content Section

Lorem ipsum dolor sit amet,[1] consectetur adipiscing elit. Quisque et eros lorem. Nullam mi dui, euismod id ultrices ut, posuere id ante. Maecenas metus felis, bibendum in iaculis eget, pretium vitae turpis. Donec vel condimentum sem.[1] Quisque sit amet pharetra libero. Aliquam dolor urna, sagittis et bibendum sit amet, pulvinar sed sem. Pellentesque imperdiet, dui at ornare rutrum, urna est vulputate enim,[1] nec porta nibh risus eu massa. Aliquam imperdiet fringilla metus eu aliquet. Maecenas convallis nulla id sem euismod suscipit. Etiam fringilla lorem placerat augue convallis cursus. Nullam quis lectus id risus volutpat sagittis. Integer quis turpis lacus. Praesent adipiscing nisl porttitor lacus gravida lacinia. Fusce lectus justo, rutrum [1]vitae consectetur sit amet, faucibus vitae lacus.


  1. ^ a b c d
I like the second one more, because the same source is grouped into one reference, instead of having duplicates. Tim1357 (talk) 16:27, 21 November 2009 (UTC)
Evidently you would wish to employ this optional feature in articles you are writing. Others prefer not to use them for the reasons cited above. Christopher Parham (talk) 00:40, 22 November 2009 (UTC)

SmackBot is at it again

SmackBot is again changing reference system. I am taking this to the administrators' page now. --Hegvald (talk) 10:02, 5 December 2009 (UTC)

The discussion from Wikipedia:Administrators' noticeboard/Incidents is now archived at Wikipedia:Administrators' noticeboard/IncidentArchive584#SmackBot changing referencing style, again (dearchived). --Hegvald (talk) 14:30, 13 December 2009 (UTC)

DOI usage

Where a citation of a journal article includes a DOI, it seems to me that some other citation elements become redundant and should be avoided (I am really thinking about citation template parameters, but the same principle applies to manually entered citations).

  • The article's URL is redundant, because the DOI links to the same target web page. That can frustrate the reader who clicks both.
  • An ISSN number is redundant, because the DOI gives the same information and much more.
  • A PMID number contributes no useful information. Since the DOI links to the article or abstract, a reader who also clicks the PMID gains nothing.
  • An access date is irrelevant, because a DOI is supposed to be permanent.

Comments?—Finell 09:25, 11 December 2009 (UTC)

  • The URL is redundant if it links to the same page as the DOI. However, often it links to a different page (the author's own copy, or some other institution's copy), and in this case it can be quite useful, as often the URL version is free but the DOI version is not).
  • Yes, the ISSN number is almost always unnecessary. The exception might be if the publisher goes out of business or sells the journal and the DOI no longer works and it's an obscure journal.
  • The PMID points to metadata that may not be accessible via the DOI, e.g., commentary, or a PMC link. I generally include the PMID even if it's redunant, as the PubMed servers are typically more reliable than various publisher's servers. In my experience DOIs are not nearly as reliable as PMIDs (that's why there is a |doi_brokendate= but no |pmid_brokendate=.
  • Completely agree about access date: they're not useful for archival journals.
Eubulides (talk) 18:02, 11 December 2009 (UTC)

Bot from changing multi line refs to one line refs

Currently many refs are spread over 10 - 15 lines of text. Is there a bot that could put the ref all on one line to make editing easier? For example this:

{{Cite journal
  | last = 
  | first = 
  | authorlink = 
  | coauthors = 
  | title = 
  | journal = 
  | volume = 
  | issue = 
  | pages = 
  | publisher = 
  | location = 
  | date = 
  | url = 
  | issn = 
  | doi = 
  | id = 
  | accessdate = }}

to this

{{cite journal| author = | title = | journal =  | volume = | issue = | pages = | year = | pmid = | doi = | month =| issn =}}

On highly referenced pages it is hard to find the text between the citations in the first example and much easier in the second. Thanks.Doc James (talk · contribs · email) 15:12, 4 December 2009 (UTC)

Have posted over at Wikipedia:Bot_requests#Bot_from_changing_multi_line_refs_to_one_line_refs to find consensus there.Doc James (talk · contribs · email) 16:03, 4 December 2009 (UTC)
I suspect that the bot will have a poor reception as many people prefer the expanded format - that is why you see it used so often. =) Christopher Parham (talk) 21:14, 4 December 2009 (UTC)
I have been told that a simple script can be created to give everyone the appearance that they wish. Thus all shall be happy.Doc James (talk · contribs · email) 21:47, 4 December 2009 (UTC)
"put the ref all on one line to make editing easier" -- huh??? It is the expanded version that is much easier to edit, surely. -- Alarics (talk) 23:03, 6 December 2009 (UTC)
Indeed. When it's broken up into multiple lines, it's easy to find the end of the {{cite}} tag -- when it's all collapsed into one line, the content and references blur together. --SarekOfVulcan (talk) 18:27, 7 December 2009 (UTC)
I find advantages for both forms. Maybe a compromise? Allow the compression to one line and leave the trailing
on a line by itself. That would allow finding the end of the citation more easily. Vegaswikian (talk) 00:12, 15 December 2009 (UTC)
That's fine if you are editing only the text and not the footnote, but if you are editing the footnote itself (which I often am, because I constantly find them being done incorrectly) it doesn't help at all. -- Alarics (talk) 08:23, 15 December 2009 (UTC)
(outdent) I don't think this is the sort of thing which a bot should be let loose to enforce wikipedia-wide. Wtmitchell (talk) (earlier Boracay Bill) 06:27, 15 December 2009 (UTC)
I usually believe that the vertical style is more readable, but neither style is nearly as readable as the new list-defined references feature. List-defined references allow you to stick all of the messy templates in the ==References== section, and use only <ref name=blah /> throughout the text. WhatamIdoing (talk) 20:25, 15 December 2009 (UTC)

Making pages faster to load

SlimVirgin recently pointed out in Wikipedia talk:Featured article candidates #Citation templates that citation templates make pages very slow to load, and that this is hurting the editing of featured articles. This is a problem I've noticed as well. It's pretty bad: when I'm editing, citation template formatting consumes the vast majority of the time spent waiting for pages to load.

  • I have developed experimental versions of {{cite journal}} and {{cite book}} that attack this problem. In my tests with Autism they have reduced page generation times by over a factor of three (from about 30 seconds to about 8 seconds for that very long article); this is a dramatic improvement to the editing experience.
  • These templates also reduce the size of the HTML generated, by about 35% in my test case (this is counting all the HTML generated for the entire article, including all the article text); this is a noticeable improvement in the reading experience, especially for readers who are not on broadband (this is still the majority of the Internet).
  • These templates use Vancouver system format rather than the hybrid style currently used by {{cite journal}} etc., partly because the Vancouver system is a bit more efficient textually, and partly because the Vancouver system is standardized in the biomedical field, which is where I spend most of my editing time.
  • I'm currently calling these templates {{cit journal}} and {{cit book}}; the name is just a character shorter so that it's easy to use the Diberri tool to generate calls to these templates. You can see an example of their use in my sandbox, which is a copy of this version of Autism, modified only by replacing "{{cite" with "{{cit".
  • The templates need documentation and some more testing. Also, substitutes for the other cite XXX templates are still needed: currently {{cit web}} is just a redirect to {{cit journal}} (I haven't verified whether this satisfies the Vancouver rules, but I hope it does), and less-commonly-used templates like {{cite video}} haven't been rewritten yet.
  • Feedback is welcome.

Eubulides (talk) 21:29, 6 December 2009 (UTC)

    • You made an edit to Template:Cite journal/doc that belongs on the doc for your own template; I think I fixed it but check. Otherwise, I think it is a welcome change. One problem with templates like these is feature creep and there's obviously a major problem when this bloat, multiplied hundreds of times, is significantly slowing page load times. Christopher Parham (talk) 03:06, 7 December 2009 (UTC)
    • This is OK, although I won't use it because of where it puts the date. I don't know really understand COInS and don't see how that information is omitted (could try to explain that in a less technical way or provide an example). The Vancouver style of putting the date at the end beside the journal volume was poorly thought out, and I hope this doesn't spur a major shift to that (I will fight it); author-date citation styles are common in the sciences, extremely common in Wikipedia (including medicine articles) despite footnotes, and put a very important piece of information where it's easy to see rather than hard to see. Differences will be jarring to readers, who are not always scientific elites and typically expect author-date. The omission of the wikilinking PMID and doi is something you should do to the regular templates. II | (t - c) 04:01, 7 December 2009 (UTC)
    • There has been recent discussion at {{cite journal}} about standardising Wikipedia "cite X" style in a document outside of the implementation, and describing it as an optional "in house" style. Additionally, I would prefer the existence of a Turabian to allow history articles the full breadth of citation requirements they often need. There is agitation regarding the quality of citations in general. Fifelfoo (talk) 04:13, 7 December 2009 (UTC)
It would make sense to support Turabian style better, too, for fields that prefer it; that would probably need to be a different template, as the details differ so much from the Vancouver system. Turabian style also puts dates after the journal title and volume, so it's not like the Vancouver system is odd in this regard, and it's certainly not an issue of "scientific elites". The vast majority of readers won't care or be affected by the location of the date, and an advantage of using a standard style such as Vancouver or Turabian is that it should lessen the number of disputes over trivia like date placement. Eubulides (talk) 17:28, 7 December 2009 (UTC)
But will it increase the number of disputes over which style should be used? Anomie 19:11, 7 December 2009 (UTC)
  • No answer on the COinS question? Chicago "humanities style notes and bibliography" puts the date in parentheses after the journal volume and edition [12], as does Turabian (they're related) [13]. Vancouver puts the date before the journal title and volume. You may want to fix your error (you said "Turabian style also puts dates after the journal title and volume, so it's not like the Vancouver system is odd in this regard", which is incorrect). Chicago scientific author-date (generally parenthetical) style puts the date after the author, and the vast majority of Wikipedia uses this style with footnotes.Whether or not readers will care is impossible to say, unless you're a mindreader. However, it is clear that this will just introduce a lot more complexity into citations for uncertain benefit. I know plenty about citation styles, but if we move towards "standard styles" I will have probably to spend a significant amount of my time referring to them. The placement of dates for books, newspapers, and other things gets more complicated. We'll have to have several different autogeneration software options, newbies (and people like me) will continually have to be corrected. Turabian, "Like the Chicago Manual of Style on which it is based ... offers those in the natural and social sciences the option of using an author-date system with notes and parenthetical references" [14]. I also suspect that Chicago allows the hybrid style currently used in Wikipedia articles. Anyway, I can't stop you from doing this, but you'll be needing consensus to change articles with a current standard to a Vancouver or Chicago/Turabian humanities style. And I think it's a very bad idea, and a bit sad that you're going to campaign for greater complexity in citations. For history and maybe humanities articles which rely on less clearly published works, sure, a different style may be needed, but not for scientific articles. II | (t - c) 20:02, 7 December 2009 (UTC)
  • COinS is a type of metadata; the citation templates use the info users input into the template to create these metadata tags and attach them to Wikipedia pages. Eubulides's versions omit this feature; whether people have actually found this feature useful in the existing templates is unknown. Christopher Parham (talk) 20:09, 7 December 2009 (UTC)
  • In practice the COinS data significantly hurts all readers and editors, for the benefit of a very few readers whose needs can easily be satisfied by some other method. Eubulides (talk) 20:47, 7 December 2009 (UTC)
What in the template produces COinS? The cit journal template has these parameters: {{cit journal |author= |title= |journal= |date= |volume= |pages= |url= |doi= |pmid= |pmc= }}. The cite journal template: {{cite journal |author= |title= |journal= |volume= |issue=|pages= |year= |doi= }} What's producing the COinS? II | (t - c) 00:03, 8 December 2009 (UTC)
The {{cit journal}} template does not produce COinS, for efficiency reasons. The {{cite journal}} template does, and this generates a COinS tag that contains article title, author names, date, series, volume, issue, edition, etc., etc., all derived from the template argument. This contributes significantly to article size and to page load time. Eubulides (talk) 00:40, 8 December 2009 (UTC)
What exactly in the cite journal template example above is producing the COinS, which does not exist in cit journal? There should be something textual producing it. I don't see what it is. II | (t - c) 01:04, 8 December 2009 (UTC)
There is nothing special in the arguments to {{cite journal}} that causes the template to produce COinS. {{cite journal}} always produces COinS, no matter what, just as it always produces a textual citation in HTML, no matter what. Eubulides (talk) 02:21, 8 December 2009 (UTC)
Regardless of what format we use, I agree there is definitely a need to improve the efficiency of the citation templates. Perhaps some of the more common ones could be split into two templates. One would include only the basic fields which would be sufficient for 90% of usage, the other would include all the less-common fields as well. Mr.Z-man 20:36, 7 December 2009 (UTC)
Unfortunately the main problem is the sheer number of subtemplates invoked, and these templates would still be invoked even if there were separate templates for common patterns. By itself I don't think this sort of factoring will buy much. Eubulides (talk) 20:47, 7 December 2009 (UTC)
The problem is not really the number of templates, its the complexity of the templates - the number of parser functions and variable substitutions it has to do for each template. Even when you're not using the trans_chapter field in {{cite book}} the parser still has to set it to an empty string and evaluate it in a bunch of parser functions. If simplified versions were made (using a simplified version of Template:Citation/core, or not using a single meta-template for all types), it could speed things up significantly. Its mainly feature bloat that's causing them to be so slow. Mr.Z-man 00:14, 8 December 2009 (UTC)
OK, well then, please consider {{cit journal}} to be a first cut at that. It rarely uses parser functions and it tries to substitute each variable at most twice, once to see whether it's empty and once to use it. (It doesn't always succeed in that.) But even if this technique were applied to {{cite journal}}, it would still be significantly worse than {{cit journal}} due to the COinS data. Eubulides (talk) 00:40, 8 December 2009 (UTC)
Outputting citation metadata is one of the fundamental purpose of citation templates; ignoring metadata is not a very good solution to template bloat. It looks to me like the better solution would be to implement the core template in a Mediawiki extension, and get rid of all the parserfunctions entirely. — Carl (CBM · talk) 03:11, 8 December 2009 (UTC)
As long as the extension is fast, well written, and provides services for a wide variety of citation styles, media types, and various "tricks" of the trade. We've been locked into an in-house style by happenstance, and one which has a confusing variety of terms for the title of the object cited versus the work contained in (as one example of poor interface). The in-house style we've picked has a number of crudities due to its accretion of features, which is a result mostly of poor social decision-making as a community rather than any criticism of the template writers. Fifelfoo (talk) 03:16, 8 December 2009 (UTC)
If such an extension could be built, that would be a good thing. One other point: the extension needs to generate COinS data conditionally. That is, it should be possible to turn off COinS, and the default (when reading an ordinary Wikipedia page) should have COinS disabled, so that the vast majority of readers who don't need COinS aren't penalized by its large overhead (over a 50% 20% overhead for Autism, in terms of number of bytes that go across the Internet). Eubulides (talk) 03:28, 8 December 2009 (UTC)
The 50% figure seems high; the HTML is 425kb and the data in the COinS title fields is 92kb. — Carl (CBM · talk) 03:43, 8 December 2009 (UTC)
Sorry, you're right, it was a 23% overhead by my measurements (22% in yours, no doubt because the article mutated between measurements). The 58% figure includes all the bloat generated by the citation templates; only about half of that is COinS. I struck the wrong figure in my previous comment, and inserted the correct one. Eubulides (talk) 18:42, 8 December 2009 (UTC)

Complexity in citations

  • "Vancouver puts the date before the journal title and volume." This seems inconsistent with an earlier comment by the same editor: "The Vancouver style of putting the date at the end beside the journal volume". Anyway, let me try to resolve the confusion. The Vancouver system as specified by the NLM, which is what I've based the template on (see Citation Rules with Examples for Journal Articles) puts the date after the journal title and edition, just as Turabian does. Because Vancouver and Turabian are similar in this respect, I'm not sure what the date-format fuss is about.
  • "And I think it's a very bad idea, and a bit sad that you're going to campaign for greater complexity in citations." No matter what change is made, it's going to increase complexity in the short run. I have asked for the bloat to be removed from {{cite journal}} on multiple occasions, and each time the answer has been no, for various reasons. I see no improvement on the horizon: for good or ill, these templates seem to have become ossified. The result is an environment in which it is becoming more and more intolerable to edit. If I want to fix this, apparently I have to work outside the {{cite journal}} system; so I might as well use a well-specified format rather than the hodgepodge format that {{cite journal}} uses now: after all, that hodgepodge is part of the reason {{cite journal}} is so slow.

Eubulides (talk) 20:47, 7 December 2009 (UTC)

Sample citations:
  • Vancouver from the template page: Baird G, Cass H, Slonims V. Diagnosis of autism. BMJ. 2003;327:488–93. doi:10.1136/bmj.327.7413.488. PMID 12946972.
  • Chicago [15]: John Maynard Smith, “The Origin of Altruism,” Nature 393 (1998): 639.
  • Turabian[16]: Christopher Policano, "Dueling Colas," Public Relations Journal 41, no. 11 (1985): 16.
You can see what I'm referring to: Vancouver puts the date before the volume, Chicago/Turabian put the date after. It's not the end of the world, but it illustrates the irritating inconsistencies among citation systems. Earlier Eubulides said "Turabian style also puts dates after the journal title and volume" - yes, it does, but Vancouver does not. I'll admit I got confused and used the word title when I meant volume. I'll admit that I like Vancouver better than Chicago. However, the NLM book does not have documentation on how to cite an online newspaper. It's designed for medicine, and complex articles with medical and social aspects will probably be tricky to deal with. The NLM example on citing an internet source is annoyingly crowded and complicated [17]: Hooper JF. Psychiatry & the Law: Forensic Psychiatric Resource Page [Internet]. Tuscaloosa (AL): University of Alabama, Department of Psychiatry and Neurology; 1999 Jan 1 [updated 2006 Jul 8; cited 2007 Feb 23]. Available from:
As Christopher said, how many people benefit from COinS is unknown. Thus far there hasn't been a grounded, concrete explanation of COinS, how it is produced from the template HTML, and who uses it (and for what). II | (t - c) 00:03, 8 December 2009 (UTC)
  • Ah, sorry, I see that I contributed to the confusion. When I read your first comment about author-date styles being common, and Vancouver being bad because it puts the date at the end, I thought the key point was whether the date immediately followed the author name (as in {{cite journal}}) or followed somewhere after the article and journal titles (as in Vancouver, Turabian, etc., although the details differ in relatively minor ways, and I agree this is irritating but I cannot fix that I'm afraid).
  • "However, the NLM book does not have documentation on how to cite an online newspaper." That topic is covered in Chapter 8; there is a section 23 Newspaper article on the Internet. I don't expect editors to routinely use all the NLM-required fields such as "[Internet]", though they may if they wish to. Currently {{cit news}} is a crude redirect to {{cit journal}}, so it doesn't implement all the stuff in that chapter on newspapers; it shouldn't be that hard to fix this but I haven't gotten around to it, as I wanted to get journals right first.
  • I, like most readers and editors, don't use COinS and don't benefit from it. If it didn't cost much I wouldn't mind, but it's slowing down page edits by a factor of three or more, and it's bloating pages by 50%, so it needs to go back to the drawing board. Sorry, that's about all I know about it; I don't really want to know about COinS internals.
Eubulides (talk) 00:37, 8 December 2009 (UTC)
For what it is worth, I also think COinS template support is a poor solution to a problem nearly nobody has. It is intended for computers so Wikipedia should provide a separate computer-readable (e.g., XML) interface for that purpose if it is felt important. For many citations, the ISBN or PMID is the only key field a computer program needs; the rest is baggage. Colin°Talk 09:13, 8 December 2009 (UTC)
Turabian and Chicago require Journal titles in ital so changed the examples. Turabian also has at least 5 presentation formats, refs / reflist; footnote / shortfoot / bib. Fifelfoo (talk) 01:05, 8 December 2009 (UTC)
Yes, isn't it wonderful that these standards have so many options? Anyway, it's OK to put journal titles in italics in the Vancouver system, as it says nothing about fonts and does not care about fonts. My template uses italics for such titles because that's the Wikipedia style for such works (regardless of whether they occur in citations). Eubulides (talk) 02:21, 8 December 2009 (UTC)
Its wonderfully necessary for certain fields; for other fields its a useless bell or whistle. Fifelfoo (talk) 02:32, 8 December 2009 (UTC)

Writing a citation formatter as a Mediawiki extension

It seems to me that the better solution would be to implement a Mediawiki extension that formats citations. This could have parameters for many styles, and would eliminate the slowness of parserfunctions entirely. — Carl (CBM · talk) 03:14, 8 December 2009 (UTC)

That would no-doubt help with the page-load problem, though it would raise a new problem of its own: it might well make it harder for editors to develop their own citation formatting styles, or to improve on the hardwired styles. I assume that the extension would let one turn off COinS selectively, so that an ordinary page read would not have COinS data at all, but a user could extract COinS data using a gadget or button or suchlike. Eubulides (talk) 03:28, 8 December 2009 (UTC)
I am not sure why you believe COinS is a significant factor in page loading times, apart from the complexity of templates that produce it. Do you have any data that suggests that the volume of COinS data is large enough to make a difference? For example, Autism, with 171 references, is only a 425kb HTML file, and this is at the extreme end of the spectrum. Of that 425kb, about 92kb seems to be in the "title" fields corresponding to COinS, which comes to about 500 bytes per reference for the COinS data itself, plus the tag overhead. — Carl (CBM · talk) 03:32, 8 December 2009 (UTC)
  • Sure: I've created sandbox versions of {{cite journal}} etc. that differed only in not generating COinS data. They showed that COinS imposed a 23% overhead in the total size of the HTML for Autism. (This is not the total size of the references section; it's the total size of the whole page.) I measured the same thing for Virus, and COinS imposed a 26% overhead for that article. Please see Template talk:Cite journal/Archive 2009 October #COiNS bloat for details.
  • Not all the bloat in the output of {{cite journal}} is COinS bloat. Some of it is other stuff. I recently took Autism and generated one version with the experimental templates (which omit the bloat), and one with the standard templates (which generate COinS data and the other bloat). The former contains 265,716 bytes of HTML, the latter 418,969 bytes. The page's HTML is thus 58% larger than it has to be, for no particularly good reason: there's no more information in the bloated version than in the trimmed-down version.
Eubulides (talk) 04:01, 8 December 2009 (UTC)
The COinS data you are omitting is more information, in a somewhat trivial way. — Carl (CBM · talk) 13:42, 8 December 2009 (UTC)
OK, but I was talking about information in the sense of information theory; the COinS data is completely redundant with the information already presented to the human reader in textual form, and so it conveys no additional information in that sense. Eubulides (talk) 17:58, 8 December 2009 (UTC)
Even if the added cost is rather small, say 5-10%, this is still significant for a feature that virtually no users benefit from. There must be a solution that can deliver this information to users who require it as promoted, not to every user every time a page is loaded. Should usage of this feature gain more widespread traction, it would be easy to turn back on. Christopher Parham (talk) 13:34, 8 December 2009 (UTC)
One main point of using citation templates is to add the citation meta information, so that we don't have to type it by hand into articles. A 10% increase in page length would not be particularly significant. — Carl (CBM · talk) 13:42, 8 December 2009 (UTC)
The problem is that the benefit, so far as I can tell, is not significant at all, and for users with slow connections, the added time is meaningful. If it could be explained who uses this feature that could not easily manage it another way (say writing a tool to read the source and create appropriate meta tags when needed), that would be useful in understanding why anyone should care. Christopher Parham (talk) 13:52, 8 December 2009 (UTC)
Plus, we're not talking about a 10% increase in page length; we're talking about increases that are around 25% for the two featured articles that I checked. I think many Wikipedia editors are spoiled by being on broadband connections, where it's not too bad. We tend to forget that most of the Internet has low-bandwidth connections and is not on broadband (my source is the 2009-09-24 Economist report on telecoms in emerging markets). Many of these people disable images, for example, because the page load time is too long otherwise. A 20% to 25% bloat hurts these guys significantly, and discourages them from reading Wikipedia's best-sourced articles. Eubulides (talk) 17:58, 8 December 2009 (UTC)
You're aware that HTML page content is sent over the network compressed? The current version of Autism takes 80kb to send over the network, which would be about 20 seconds on 28.8kb modem. That's with all the COinS data, and this is already an exceptional article in terms of number of references. — Carl (CBM · talk) 18:55, 8 December 2009 (UTC)
Yes, I'm aware of that. In my measurements the COinS data compressed slightly worse than the rest of the HTML, which means that the 23% and 26% bloat I measured for the uncompressed text of Autism and Vaccine articles actually underestimates the amount of actual bloat over the Internet. It's true that shorter articles don't suffer as much, but Wikipedia technology should support our high-quality and well-sourced featured articles; it shouldn't be getting in their way. For what it's worth, my impression is that African Internet connections are typically in the 9.6 Kb/s to 28 Kb/s range nominal, with actual rates somewhat slower. Eubulides (talk) 19:27, 8 December 2009 (UTC)
It seems to me that you are stretching the justification here. If you actually have statistics on the number of Wikipedia readers with a 9.6 kbaud connection that indicate that page size (rather than latency, etc) is a problem, I would be interested to see them. But we expect things to be somewhat bad for such users anyway, just as we expect things to be somewhat bad for users with an 800x600 display. If we really wanted to minimize average page size, I don't think that COinS data would be the place to start, since few articles have that many references.
Do you have any statistics on the average number of citation templates over the pages that use citation templates? I would guess that articles such as Autism are far from the median in that regard. Without statistics like that, I don't see any reason to think that COinS tags actually make a noticeable difference when averaged over multiple page views, especially given that most readers are logged-out and will benefit from caching of the compiled pages. — Carl (CBM · talk) 19:43, 8 December 2009 (UTC)
Please see #Real-world effects of citation bloat below.

Real-world effects of citation bloat

  • "If you actually have statistics on the number of Wikipedia readers with a 9.6 kbaud connection" Nobody has readership statistics like that, and it's not clear how relevant they would be anyway. If article bloat discourages people with slow connections, and they are therefore a small percentage of Wikipedia readers, that does not mean that article bloat should be of no concern; on the contrary. We don't need to commission a scientific research study to know that we have a very real problem here.
  • "Do you have any statistics on the average number of citation templates over the pages that use citation templates? I would guess that articles such as Autism are far from the median in that regard." Sorry, but I'm afraid that intuition is incorrect in an important practical sense. What matters is articles that are commonly read. And here, statistics show that Autism and Vaccine are typical; they're not unusual. Lots of very highly-accessed pages in the same ballpark as, or even more citations than, Autism (171 citations) and Vaccine (36 citations). Here is a list of the ten most-accessed pages on English Wikipedia in July (the most recent month available) according to Wikistics.[18] I am omitting self-referential pages such as Main Page and Wikipedia:
  1. The Beatles (112,000 hits/day, 128 citations)
  2. Michael Jackson (80,000 hits/day, 267 citations)
  3. YouTube (72,000 hits/day, 97 citations)
  4. Barack Obama (49,000 hits/day, 240 citations)
  5. Deaths in 2009 (49,000 hits/day, 0 citations)
  6. United States (46,000 hits/day, 204 citations)
  7. Facebook (43,000 hits/day, 215 citations)
  8. Swine influenza (40,000 hits/day, 106 citations)
  9. Eminem (33,000 hits/day, 108 citations)
  10. Lost (TV series) (33,000 hits/day, 162 citations)
  • Except for the outlier Deaths in 2009, these pages all take waaaaayyy to long to load on my browser (unless they happen to be cached, which in my ad hoc tests is not as often as I thought it would be). And they're often the first articles that new readers see. This is a terrible advertisement for Wikipedia.
  • For uncached articles it appears that most of the delay is due to COinS data and other citation bloat. I do not have the detailed statistics, but from what we've seen so far I would guess that more than half of the CPU cycles consumed by Wikipedia page-rendering servers are squandered on citation bloat, and that more than 10% of all the bytes shipped to browsers are wasted due to citation bloat.
  • The citation bloat might be worth it if it was useful. But it's not. Almost nobody uses COinS data, and the people who do use it would be satisfied with another button that shipped COinS data only on request. And the non-COinS citation bloat provides almost zero benefit for a relatively high cost.

Eubulides (talk) 21:37, 8 December 2009 (UTC)

    • "but from what we've seen so far I would guess that more than half of the CPU cycles consumed by Wikipedia page-rendering servers are squandered on citation bloat," — do you have any evidence to back that up? We can all describe our personal opinions; I'm looking for something more scientific than numbers pulled out of the air, or short lists of articles chosen at random. As to readership statistics, we do have (at least) data on geographical distribution of readers. That is, it's in the server logs; I don't know if anyone has analyzed it. — Carl (CBM · talk) 21:59, 8 December 2009 (UTC)
      • A list of the most accessed articles is hardly random. At any rate, trying to understand the portion of readers negatively impacted here is not useful unless we also have an understanding of the benefits derived from these features. Are readers using COinS data to improve their Wikipedia experience? I suspect that the rare readers who use it would be techncially sophisticated enough to employ an alternate method of extracting this metadata, which wouldn't interfere with our general reader experience. In short, whether the impact of this bloat is great or small it probably justifies eliminating a feature whose value is minimal. Christopher Parham (talk) 22:35, 8 December 2009 (UTC)
        • My limited data obviously do not come from an exhaustive survey of all the articles on Wikipedia: far from it! But it is good evidence that we have a significant performance problem. Here are two more data points. I just put into my sandbox copies of The Beatles and of Michael Jackson. The bloated versions required 21 and 22 seconds to load (uncached), respectively; the trimmed-down ones 12 and 13 seconds. The bloated versions consume 92,355 and 98,482 bytes of compressed HTML; the trimmed-down ones 81,976 and 85,655 bytes. This indicates that for The Beatles, the citation bloat consumes 75% more CPU time and 13% more network traffic, and for Michael Jackson it's 69% more CPU time and 15% more network traffic. I will concede that my CPU-time estimate of 100% was too high, but a 70% bloat is still quite bad. And my 10% network-traffic estimate seems to have been too low, if these two high-volume samples are any indication. I agree with Christopher Parham that we need to compare costs to benefits here. Eubulides (talk) 23:19, 8 December 2009 (UTC)
          • Again, it looks to me like you are cherry-picking specific articles where you know the use of citation templates will slow things down. We have thousands of commonly-read articles, most of which are not FAs. In particular, we have 17,000 articles that average over 1000 hits per day. So picking 5 or 10 articles off the top of your head is not any sort of fair sample, and isn't evidence that there is any widespread performance problem.
               Moreover, your "trimmed-down" templates do more than just remove COinS, they also rewrite the template structure, and change the output format. In particular, does your template support all the parameters that the original ones do? A fair test would use a "lean" template that accepts all the same parameters and produces the same output as the current templates. At the very least, you could see whether adding COinS to your templates actually causes the performance to degrade to its prior levels. Lacking that sort of testing, there doesn't seem to be evidence that COinS is what is slowing things down.
               In case it might help you generate actual data, I have put a list of the top 1000 articles by average daily hitcount at [19]. — Carl (CBM · talk) 01:00, 9 December 2009 (UTC)
          • P.S. Also, because of caching, compile time for pages is not likely to be a significant factor for not-logged-in users. Once a revision is compiled once, it can be served to additional logged-out editors with no parsing overhead. This casts more doubt on the argument that parsing time is relevant to how quickly most of our readers are served pages. — Carl (CBM · talk) 01:09, 9 December 2009 (UTC)
            • The charge of cherry-picking is completely unfounded. I chose the top 10 regular articles on Wikipedia, and had no idea ahead of time how many citations they had. I reported the results even though one of these articles had no citations whatsoever, and obviously did not favor my case. I am just a single editor editing by hand, and do not have the resources to benchmarks thousands of revisions of thousands of pages. Nor is it realistic to expect me to. The evidence I've presented so far is compelling evidence that there's a significant overhead to the citation templates, and that a sizeable fraction of that is due to COinS data. If you don't choose to believe the numbers, and think that the fraction is smaller than what's stated, that's fine, and you're free to generate your own number. But there's no way that the fraction is insignificant: it's a noticeable performance degradation on the articles that I regularly edit. In contrast, there's zero evidence that COinS is providing benefit to Wikipedia readers; certainly zero evidence that the current approach, where everybody pays to get COinS data that almost nobody wants, is any better than a button that would generate COinS only on demand. The cost–benefit tradeoff here is clear. Eubulides (talk) 01:19, 9 December 2009 (UTC)
              • Actually, I do think it is realistic to expect that somebody who is arguing that a certain template has performance problems has actually done some benchmarking of a significant number of pages to back up their claims. Otherwise, there's no way to distinguish between reality and perception. The problem of limited samples is exacerbated by the fact that caching works much better for non-logged in readers than it does for logged-in editors. — Carl (CBM · talk) 01:23, 9 December 2009 (UTC)
                • Was that same performance case made when COinS data was put in? If not, then why have a higher burden for proof for removing a feature that has obvious costs (even if we may disagree with how high they are) and no demonstrated benefits? Anyway, I can do some more benchmarking when I find the time, but it's unrealistic to expect me to do tests involving thousands of pages. Eubulides (talk) 02:16, 9 December 2009 (UTC)

The median article, in a traffic-weighted sense, gets about 300 hits per day. Meaning that 50% of traffic goes to pages with more hits per day and 50% goes to pages with fewer. (Some 15% of total traffic goes to the 85% of articles averaging less than one hit per hour.) Also, please keep in mind that render times can easily vary by a factor of two or more simply depending on which server gets asked to do the rendering. Any conclusions you attempt to draw based on these served by numbers should be repeated many times. Dragons flight (talk) 01:42, 9 December 2009 (UTC)

I do realize that the render times can vary, and explain part of the noise in the figures I've generated so far. The preliminary numbers I've done provide compelling evidence. And they are easily reproducible (you can visit the links I've provided yourself). Eubulides (talk) 02:16, 9 December 2009 (UTC)
Render times (10 trials)
Autism (current version) 26.152 23.118 16.281 16.235 22.776 16.468 28.366 28.318 16.391 16.437
Autism (Eubulide's bloated version) 19.261 19.33 18.962 16.278 23.622 16.239 39.378 16.309 16.492 23.775
Autism (Eubulide's slim version) 12.08 8.341 8.335 8.255 8.531 14.862 8.373 8.369 8.356 8.513

These are essentially indistinguishable in median (19.622 vs. 19.1115) and mean (21.0542 vs. 20.9646). There is no meaningful difference between your approach and the original as far as I can tell. Eubulides, I think you are fooling yourself in to believing there is a significant improvement when there isn't. Dragons flight (talk) 02:29, 9 December 2009 (UTC)

That's because you measured the wrong page! The version you labeled "Autism (Eubulide's version)" is the bloated version: it is a clone of Autism except that stuff that should not be in user pages (such as {{featured article}} and categories) are commented out to avoid trouble. The bloated version generates all the COinS data and other unnecessary gorp. The version that you should have measured is the non-bloated version; and, to be fair, you should compare it to the the bloated version rather than to Autism directly. I just now did this, and got the following timings (measured by wall-clock time between pressing "shift-reload" and seeing the "Done" message on Firefox):
Total turnaround times (10 trials)
the non-bloated version 11 16 16 11 17 17 12 12 13 14
the bloated version 32 32 27 27 22 27 22 27 22 22
That averages out to about 14 seconds for the non-bloated version, and 26 seconds for the bloated version. My timings for the bloated version are higher than yours most likely because I'm measuring the total turnaround times (including network latency, and browser rendering on an older desktop) rather than just the CPU on the server side. Eubulides (talk) 06:24, 9 December 2009 (UTC)
Okay, so I can be dumb some times. I've updated the table above with the correct versions. I agree that there does appear to be a significant difference, median (19.1115 vs. 8.371), mean (20.965 vs. 9.402). Dragons flight (talk) 07:44, 9 December 2009 (UTC)
Running a series of 10 with just COinS suppressed, suggests that roughly 1/3 of the CPU difference is caused by COinS while presumably the other 2/3 comes from the other changes you made to the templates. Dragons flight (talk) 08:36, 9 December 2009 (UTC)

Basically a (lack of) caching issue

The basic problem is that we don't cache template calls, so every time the page changes every template has to be rerun even though most changes affect very little of the page. At the expense of more disk space, one could cache the results of all template calls (or more likely just the large and/or slow ones) in order to avoid rerending all templates on every edit. I can think of a couple different approaches to such a cache, though it's not immediately obvious what would provide the best balance of performance gains without ridiculous resource requirements. People have also argued for caching <ref> render results which would be another approach if caching all templates is too aggressive. Dragons flight (talk) 03:38, 8 December 2009 (UTC)

Possibly a solution would be section caching. This is the same data that is already being cached, just handled slightly differently. Clearly there are dependencies, such as ref numbers changing that might mean an intermediate level of caching was needed, but again there may be ways to deal with that. Rich Farmbrough, 20:06, 8 December 2009 (UTC).
So the problem is really one for editors rather than readers. Rich Farmbrough, 20:06, 8 December 2009 (UTC).
Apparently this is done on Wikia - it helps with the pre-render buffer limit too. Rich Farmbrough, 16:03, 22 December 2009 (UTC).

Size increases due to COinS vs non-COinS bloat

At CBM's request I separated out the citation bloat into two parts: one part due to COinS, and the other part due to the other unnecessary gorp that {{cite journal}} etc. put into templates. I measured three pages, Autism (which is what started this thread), and The Beatles and Michael Jackson, the two most popular regular articles in July. I've been asked to compile similar numbers for thousands of articles, but I don't have the resources to do that, so I did it for just these three, as this should at least give us a clue as to how much bloat is going on. I measured the compressed and uncompressed HTML of the resulting pages, in my sandbox (see wikilinks below, in each compressed entry). Here are the byte counts and bloating factors that I observed:

Citation bloat (compressed and uncompressed HTML byte counts)
Article current templates remove COinS remove other bloat too COinS bloat other bloat total bloat
Autism 78,212 A1 63,110 A2 61,214 A3 24% 3% 28%
(uncompressed) 418,969 311,383 265,716 35% 17% 58%
The Beatles 92,355 B1 85,118 B2 81,976 B3 9% 4% 13%
(uncompressed) 439,035 376,052 346,839 17% 8% 27%
Michael Jackson 98,482 C1 89,164 C2 85,655 C3 10% 4% 15%
(uncompressed) 473,333 398,305 358,031 19% 11% 32%

All byte counts are total HTML bytes; for each page, the first row is for compressed byte counts, and the second row for uncompressed. For example, in Autism adding COinS increases the number of compressed HTML bytes from 63,110 to 78,212, which is an increase by a factor of 1.239..., so it counts as 24% bloat. Adding just the other bloat increases the number of compressed HTML bytes from 61,214 to 63,110, a factor of 1.030..., so it counts as a 3% bloat. Bloat factors multiply, so the total bloat is 78,212 / 61,214, or a factor of 1.27768..., so it counts as a 28% bloat. And so forth.

Looking at these numbers, my earlier comments about COinS not compressing as well as the rest of the article seem to be incorrect; the reverse is true for these articles. Sorry about that: I don't know how I got the wrong figures there. Overall, for these (large, well-referenced) articles, the overall bloat, after compression, is in the range of 13% to 28%, with COinS alone responsible for a bloat of from 9% to 24%.

Eubulides (talk) 08:24, 9 December 2009 (UTC)

Full citations obsolete?

I confess that I like full, well formatted bibliographic citations. However, I am coming to the conclusion that some traditional citation elements are obsolete. If a citation of a book includes the ISBN number, is it useful to include the name and location of the publisher or the edition number? I would still retain citation elements that convey information on sight: The author(s)' name(s), title, and year of publication. But is the rest of it helpful? In citing the Encyclopædia Britannica, does it really help the reader to add Chicago: Encyclopædia Britannica, Inc.—especially if Encyclopædia Britannica is wikilinked?—Finell 09:48, 11 December 2009 (UTC)

EB has always been so well known that it hasn't needed a publication location. I find publishers to be more or less essential in judging article quality in the humanities, obscure publishers require locations, and then once you've got locations for a couple. Fifelfoo (talk) 09:58, 11 December 2009 (UTC)
One very good reason for including the the name, publisher location and edition number or year of publication is that some books are published simultaneously in two or more countries under different ISBNs. If the reader is in a different country to the writer, then the reader might choose to look at the version that was published in their own country rather than the one that was published in the writer's country. —Preceding unsigned comment added by Martinvl (talkcontribs) 12:45, 11 December 2009 (UTC)
I find it helpful to give the name of the publisher even if the ISBN is available, to give the reader a hint as to the establishedness of a book without having to click on the link. The location is useful only for obscure publishers. If the publisher is the same as the book name (as in the Encyclopædia Britannica) there's no point to listing the publisher separately. The question of book editions is separate: some books come out in different editions even with the same publisher, and the edition needs to be cited regardless of the publisher. Eubulides (talk) 18:02, 11 December 2009 (UTC)
I agree that location is the least important element of a "full" cite these days. On many new books it is hard to even pick a location, because it may have been simultaneously published in several places in identical editions, with all of them listed on the title page. Publisher, on the other hand, can be important as a clue to the nature of the source. Some publishers are known as vanity presses or for ideological bias, so seeing their name (instead of, say, a major university press) can trigger further investigation of the source. You could find this out by following an ISBN, but that is extra steps. The odds of typos and other mistakes is also more likely with number strings than it is with ordinary names. --RL0919 (talk) 16:00, 15 December 2009 (UTC)

Disappearing website

I've got a problem with sources in an article I wrote today. For the Bronzewing Gold Mine article, used a lot of references from the View Resources company website, a former owner, which is in administration. Apparently, once the administration process is over, the company will be liquidated and, I would think, its website will disappear. That will make all my references dead links. Any ideas what can be done about that? I could save all the company announcements I used on my computer but that won't make them accessable to others. Calistemon (talk) 10:07, 15 December 2009 (UTC)

You can archive individual pages using WebCite, but I don't know whether this will work for PDFs. I've queued the 6 Feb 08 Bronzewing update PDF for archiving. More news when I have it. - Pointillist (talk) 12:14, 15 December 2009 (UTC)
Yes, WebCite works for PDF documents available on websites. Calistemon could also check if web pages have already been automatically archived by — Cheers, JackLee talk 12:22, 15 December 2009 (UTC)
Alternatively, upload your PDFs to Google docs and link to them there. -- Alarics (talk) 15:32, 15 December 2009 (UTC)
In theory the resource is now available at, but I can't make it work. Perhaps Jacklee could have a go? The Google docs suggestion is novel: at first glance it seems that this technique would breach the View Resources copyright and fails to provide an audit trail from the source website to the archive. I can also see potential problems with Google's terms (e.g. 8.3, 9.6, 11.1-11.4). - Pointillist (talk) 22:31, 15 December 2009 (UTC)
I have no problem viewing the archived file at What problem are you having with it? By the way, Calistemon, note the use of the |archiverurl and |archivedate= parameters to refer to archived web pages if you are using citation templates such as {{citation}} in your Wikipedia article. — Cheers, JackLee talk 07:20, 16 December 2009 (UTC)

Thanks for all the advice, guys! I'm having problems, too, to view the file, my computer tries to open it with Adope and then tells me it can't find the file. I will try some more and see whether I can get it to work. Calistemon (talk) 10:13, 16 December 2009 (UTC) works fine for me. -- Alarics (talk) 10:31, 16 December 2009 (UTC)
Strange! I tried archiving another one, is now under and it seemed happy to do so but I can't access this one either! Any ideas whats locking me, and possibly also Pointillist out? Calistemon (talk) 10:34, 16 December 2009 (UTC)
As a last try, I've archived a third source, this time not a PDF file and I can access this one fine, without trouble. I would say, its me than having trouble with the PDF's. Thanks for all your help, guys, you put me in the right direction! Calistemon (talk) 11:14, 16 December 2009 (UTC)
Hi Calistemon, what PDF reader/writer application do you use? My problems only seem to happen on machines that have Adobe Acrobat (writer) installed. For example: opens OK using FF3.5 with Adobe Reader 8 on Vista Home, OK using FF3 with Adobe Reader 8 on XP Pro, and OK using IE6 with Adobe Reader 8 on Windows 2000 Server. It fails using either FF3.0 or IE8 with Acrobat 9.2 Standard on XP Pro and fails using IE8 with Acrobat 9.2 Pro on Windows 7 x64 Pro, Hmmm. - Pointillist (talk) 13:13, 16 December 2009 (UTC).
However, it opens OK using Chrome with Acrobat 9.2 Pro on Windows 7 x64 Pro, Hmmm again. Pointillist (talk) 13:57, 16 December 2009 (UTC) also works fine for me. I use Firefox browser with Foxit PDF reader. I do not have Acrobat installed. It sounds as if your problem is with Adobe. -- Alarics (talk) 14:28, 16 December 2009 (UTC)

Google Books

I've gotten into a disagreement over the interpretation of WP:SAYWHEREYOUGOTIT with regards to citations to books and articles found via the above methods. I feel that since I am reading the original work, albeit reproduced online, I can cite the original work. The editor who raised this issue instead footnotes a raw URL for the Google Books result in question, and puts the original work in a "related reading" or some such section. Opinions? --Orange Mike | Talk 18:17, 7 January 2010 (UTC)

I always cite the actual work, and then also supply a URL to the Google Books version. --Golbez (talk) 18:22, 7 January 2010 (UTC)
I treat google books the same as I would treat a photocopy, and just cite the original source. In other words, I treat it like I would treat a journal article obtained through a document delivery service at the library, which is another situation where I would never see the true original.
Of course there is nothing wrong with providing a courtesy link to google books, although these are so opaque I have never been certain they will work as expected for other readers.
My main concern with google books is that you have to go out of your way to check the context in which the cited material is being used. For example, making a citation based only on a "snippet view" of the original would be absurd. — Carl (CBM · talk) 18:30, 7 January 2010 (UTC)
I agree that the snippet view is almost always inadequate, but for books with "limited preview" it is often possible to see all the relevant context, and of course some books are available for full viewing. Of course there is nothing that says an editor can't go to the actual book for verification, and challenge the original citation if it proves to be inaccurate. --RL0919 (talk) 22:56, 7 January 2010 (UTC)
I don't think you need to indicate that you only saw the work on Google Books; we can trust their copy to be accurate just as much as we could trust, say, a library's copy to be accurate. Either way, there should be citation text in the article containing the title, author, date, etc. of the book in question. A raw footnote together with such a citation in the references section would seem to meet the basic requirements of this guideline, but this isn't the most elegant format, and putting the citation text in a "related reading" section, or any other place where it's not clear the work is actually cited in the article, should be discouraged. Your method sounds more in line with what's looked upon as best practice, but the other editor's method sounds close to Wikipedia:Embedded citations. Christopher Parham (talk) 18:32, 7 January 2010 (UTC)

I am the editor in question. The original discussion follows:

Please don't post the raw URL of a Google Books result; that is inconvenient to the reader and deprives the actual writer of the work "cited" of any genuine credit. --Orange Mike | Talk 05:02, 7 January 2010 (UTC)

"It is improper to include a reference from an intermediate source in a Wikipedia page without stating so. For example, on a Web page, you might find some information that is attributed to a book. Unless you examine the book yourself, your reference is then the Web page, not the book. You should, in turn, make it clear in the reference that the Web page cited the book."

This is a direct quote from As a writer, I cannot say I am happy with this, but it is the situation that prevails. If you can get it changed, I will willingly change. In the meantime, I list every book so cited in the References section of the article.

Georgejdorner (talk) 17:04, 7 January 2010 (UTC)

Interesting reasoning (and thanks for explaining); but I don't think the language you cite applies in the case of a Google Books result. These results give us direct access to an accurate reproduction of the original work, and I always footnote the text of the original work based on that access. --Orange Mike | Talk 17:10, 7 January 2010 (UTC)

"Unless you examine the book yourself, your reference is then the Web page, not the book." That's not my reasoning; that is the reasoning of the administrators. It's pretty clear. If you should manage to get this policy changed, please let me know.

In the meantime, when I Google a webpage of a book, I feel that the above policy applies. BTW, what policy can you quote to back your approach? (Which I might mention, I have used in the past, and prefer to use.) Georgejdorner (talk) 17:19, 7 January 2010 (UTC)

I am making a distinction between a random Google and the use of Google Books (or Google Scholar) to find a reference source. I don't feel that the policy you follow applies in the latter case. --Orange Mike | Talk 17:28, 7 January 2010 (UTC)

If your distinction is based on a Wikipedia policy instead of Mike's opinion, I will be glad to follow it. If the distinction is based only on Mike's opinion, then I face the prospect of following the next administrator's opinion/whim/interpretation when the same questioning reoccurs. And it does. Boy, does it. Unlike about 90% of the articles on Wikipedia, mine are always fully cited; yet, I have dealt with a long line of complaints about my citations. It seems that the more effort I put into it, the more criticism I draw.

Georgejdorner (talk) 18:08, 7 January 2010 (UTC)

I would say that looking at the usual view of Google books does qualify as looking at the book itself, just as looking at a photocopy of the book would qualify as looking at the book itself. The text you quoted is about when the text has been retyped onto some webpage, not about when we are looking at reliable duplicates of the actual book. — Carl (CBM · talk) 20:26, 7 January 2010 (UTC)
It seems to me obvious that Carl is right here. Looking at Google Books' image of the book must surely be regarded as the same thing as looking at the book itself. If that's not clear from the WP guideline then the guideline needs rewriting. -- Alarics (talk) 20:41, 7 January 2010 (UTC)
Yes, I agree that Carl and Mike's interpretation is the usual and useful one, though once in a while it is questioned, with similar explication. Google books, or any archival reproduction is not considered an "intermediate source" for the purpose of the guideline. Saywhereyougotit was put into the guideline by User:Zero0000, I am pretty sure the intent was as Carl and Mike suggest. So I think George should feel free to use this interpretation.John Z (talk) 21:29, 7 January 2010 (UTC)
Agreed - Google books is not an intermediate source as described in this guideline, anymore than the university guideline is an intermediate source when the librarian fetches you a book from the depository. Christopher Parham (talk) 22:23, 7 January 2010 (UTC)

Three comments:

  • The quoted advice is irrelevant (although this is unfortunately not obvious). The quoted advice is about this situation: I read [insert name of polemical blogger here], who says "This book says that ____." We would accept a claim that "Blogger says that book says...", but not a claim that "The book says..." -- because you don't know that the blogger got it right.
  • Reading a book on paper, by fax machine, as an ebook, online, in Braille, or with assistive technologies like a screen reader is still reading the same book. We care about the editor actually reading what was written, and not relying on what some other person says was written. It doesn't matter if you're reading the original words in their original font or format so long as you're reasonably confident that you're reading the true text. Requiring a distinction between "the book" and "the book that was scanned into a computer" is like requiring the footnotes to distinguish between the hardback copy, the softcover copy, and the copy that my grandfather had rebound by hand in leather.
  • I generally don't link to (or any other source) if there's an ISBN for the book. I'm not absolutist about this practice, but I do think that it's normally better to let the ISBN magic word take readers to their preferred source instead of promoting my own preferred source. WhatamIdoing (talk) 22:33, 7 January 2010 (UTC)
I entirely agree with all of those points, excepting that distinguishing editions is sometimes helpful because pagination may differ or there can be subtle variations in content (such as typos or corrections of mistakes from an earlier edition). But since Google Books is doing scans, you just need to specify which edition they scanned, which is usually clear from the ISBN when one is present. --RL0919 (talk) 22:56, 7 January 2010 (UTC)
True enough, if the citation includes page numbers. There are also, on occasion, differences between editions (particularly in forewords and appendices) that might need to be called out. However, I believe such considerations are normally rare, particularly in modern books, and that most editors will be able to resolve such problems without being explicitly told.
I have boldly updated this section. If anyone can think of a clearer method of communicating our actual intent, please feel free to improve on what I've done. WhatamIdoing (talk) 23:06, 7 January 2010 (UTC)
I slightly tweaked WhatamIdoing and SlimVirgin's change to indicate reading a computer image is equivalent to reading the original. This distinguishes situations that might come up, such as a book (e.g. Kindle) that is originally distributed in digital form, and books that have been transcribed (e.g. Project Gutenberg) with typographical errors and changes in pagination. --Jc3s5h (talk) 00:14, 8 January 2010 (UTC)

Son of a gun, a quick balanced rational solution to a citation problem. It is enough to restore (some) belief in me concerning the citation processs.

Many thanks, folks.

Georgejdorner (talk) 19:02, 8 January 2010 (UTC)

Christopher Hibbert - Tower of London

To Whom It May Concern:

I wish to add another title to the Bibliography of Christopher Hibbert. He wrote a book in a series of book for Newsweek called Wonders of Man, which was first published in 1971. Below is all the information. Since I am new to Wikipedia, I am wondering if someone can please add the following information to Mr. Hibbert's page for me? I love this book and feel that it should be included in his literary works.

Tower of London-A History of England from the Norman Conquest

by Christopher Hibbert and the Editors of the Newsweek Book Division

Newsweek, New York

Editor: Joseph L. Gardiner Associate Editor: Edwin D. Bayrd,Jr. Picture Editor: Laurie P. Phillips Assistant Editor: Eva Galan Copy Editor: Lynne H. Brown European Correspondent: Russell Ash Publisher: Alvin Garfin

Wonders of Man (Series)

Consulting Editor: Milton Gendel

1st Printing, 1971, 2nd Printing, 1972, 3rd Printing 1973, 4th Printing, 1974, 5th Printing, 1975, 6th Printing, 1978.

ISBN: Clothbound Edition 0-88225-002-7 ISBN: Deluxe Edition 0-88225-003-5

Library of Congress Catalog Card No. 70-136436

Copyright 1971 - Arnoldo Mondadori Editors S.p.A.

Thank you,

Roseofsharon444 (talk) 16:18, 8 January 2010 (UTC) RoseofSharon444

All rights reserved, Printed and bound in Italy by A. Mondadori, Verona

I've added a simple bibliographic citation. If you edit Christopher Hibbert#Publications, you'll see the template that I used. The same format can be copied and pasted (just change all the 'answers' to the correct information for any other book). With the ISBN, you can ask this webpage to do it for you, based on the official ISBN database. Template:Cite book lists more options if there's something special you want to include (like editors).
Note, as well, that using this template is never required on Wikipedia, so if you don't like it, you can change it to make the style match the rest of the list. WhatamIdoing (talk) 20:19, 8 January 2010 (UTC)

Cit and C citation templates at TfD

The following citation templates have been nominated for deletion for lack of use and redundancy to the existing templates:

All are listed at Wikipedia:Templates for discussion/Log/2010 January 14. -- AnmaFinotera (talk · contribs) 07:17, 14 January 2010 (UTC)

Odd two sentences

I removed the following, but it was restored saying there had been consensus for it here:

"It is not necessary to say how you obtained a book or other source. A book's contents are assumed to be the same whether you buy it, borrow it from a library, or read an image of it on a computer screen."

I can't think of a situation in which it might be necessay to say where a book came from. I therefore can't see why we're saying it isn't necessary (it isn't necessary to say whether you bought your book secondhand; it isn't necessary to say whether your dog tried to chew it as you were reading). Am I missing something? :) SlimVirgin TALK contribs 00:03, 16 January 2010 (UTC)

See the #Google books thread. The point is that if a source that can be trusted not to deliberately falsify the contents of publications, such as Google books or the Astrophysics Data System, provides images of the book pages, those book pages can be considered to be interchangeable with the original book and it is not necessary to specify that it was the images that were viewed, not the original pages. Of course, if the images are still online, one would normally provide a hyperlink to the online source. But if the pages had been taken offline for whatever reason, there might be no point in mentioning where the pages had been viewed.
Even if you don't agree with the reasoning, it should be plain that the version in which you deleted the paragraph in question does not adequately explain how readers should cite images of a publication, and something other than simple deletion is in order. --Jc3s5h (talk) 00:15, 16 January 2010 (UTC)
Okay, first, what you wrote has nothing to do with whether you buy a book or borrow it. Adding that kind of thing to the guideline looks very odd. As for books made available by Google, that's a separate matter. I don't see why we would need to cite "images" of a publication differently, nor why we would need to say that we don't need to do it. It seems so obvious. If someone holds a book page up on television, and as luck would have it, it's the very page you need for your latest Wikipedia article, there's no need to say, "I saw it on the telly," and no need for us to tell people, "You don't need to add 'I saw it on the telly'." All that matters is name, title, publisher, year of publication, page number. Readers who track down the book will find that the material is there or not there. Whether the Wikipedian who added it says he first saw it on Google will not change that fact. SlimVirgin TALK contribs 01:25, 16 January 2010 (UTC)
I did not write the statement, I just restored it. I feel it is too diffuse, and too distant from the dispute that lead to its inclusion. But since there was at least one dispute, described in the Google books section, where an editor insisted that Google books be cited rather than the original text, some statement is necessary. --Jc3s5h (talk) 01:48, 16 January 2010 (UTC)
I'm going to remove it again. If this has been added just because of one editor insisting on something, then I'd say we haven't yet seen a need for it, and the thing about buying/borrowing looks too strange. SlimVirgin TALK contribs 02:02, 16 January 2010 (UTC)

Village pump discussion

Please see this discussion about a standard citation style on the Village pump talk page. — John Cardinal (talk) 18:24, 17 January 2010 (UTC)

make criticism work for you

Literary criticism is the study, discussion, evaluation, and interpretation of literature. Modern literary criticism is often informed by literary theory, which is the philosophical discussion of its methods and goals. Though the two activities are closely related, literary critics are not always, and have not always been, theorists.

Whether or not literary criticism should be considered a separate field of inquiry from literary theory, or conversely from book reviewing, is a matter of some controversy. For example, the Johns Hopkins Guide to Literary Theory and Criticism draws no distinction between literary theory and literary criticism, and almost always uses the terms together to describe the same concept. Some critics consider literary criticism a practical application of literary theory, because criticism always deals directly with particular literary works, while theory may be more general or abstract.

Literary criticism is often published in essay or book form. Academic literary critics teach in literature departments and publish in academic journals, and more popular critics publish their criticism in broadly circulating periodicals such as the Times Literary Supplement, the New York Times Book Review, the New York Review of Books, the London Review of Books, The Nation, and The New Yorker. —Preceding unsigned comment added by Balajnoor (talkcontribs) 15:25, 18 January 2010 (UTC)

What exactly does all that have to do with the price of fish? -- Alarics (talk) 16:13, 18 January 2010 (UTC)

Related Discussion

There is a discussion occurring at Wikipedia:Centralized discussion/Wikipedia Citation Style. Your participation would be appreciated.
V = I * R (talk to Ohms law) 02:06, 19 January 2010 (UTC)

Links within citation

If one cites a famous learned journal, should there be a link to the Wikipedia article on that journal? In an article that I follow, an editor changed, in all citations to well-known learned journals, from simple mentions to links to the articles on those journals - journals such as Nature (journal). In the case of a single reference source, let's say a book which is notable enough to have an article, making that a link seems OK to me, and a link to a famous author seems to be encouraged by the cite template. Should I, in my future editing, always make a link to Nature, Science, and so on? Should I retroactively change my citations? Or should I go to the other extreme and revert the changes that my fellow editor introduced? Or just forget about it, and let personal preferences rule? I have no strong feelings, one way or the other. Or have I missed somewhere where there is a policy on this? Just asking. TomS TDotO (talk) 14:32, 16 January 2010 (UTC)

There's no policy. I tend to link to authors, only to show that I'm using sources who are known, but I don't link to newspapers or journals, because I don't like a sea of low-value blue. Other editors link to anything that stands still. It's not worth reverting over, unless it gets to the point where the text is hard to read because of too much blue. SlimVirgin TALK contribs 14:43, 16 January 2010 (UTC)
I think it is quite superfluous, as in wp:overlink, to link the names of publications when they really are famous. The great majority of readers presumably don't need to consult the WP articles about Nature or The New York Times or CNN or the BBC to find out that they are well-known reliable sources. The few readers who do need to do that can easily find the relevant articles for themselves. -- Alarics (talk) 15:55, 16 January 2010 (UTC)
That reflects an American-centric view. The majority of the English world may not know who or what Nature is, or The New York Times. -- AnmaFinotera (talk · contribs) 16:28, 16 January 2010 (UTC)
It is generally appropriate to link the work of the journal/newspaper/publisher if they have an article or are likely to. I have found it is best to link them consistently, and constantly within the citations. I disagree that it makes the text hard to read, and its very presumptive to presume that any one journal or newspaper is so famous that most English readers know it. One must always keep in mind that English != American, and that most English readers from other countries may not know those sources. -- AnmaFinotera (talk · contribs) 16:28, 16 January 2010 (UTC)
Myself, I had never thought of linking on the name of a journal when citing an article in the journal. That is to say, I had never thought of it, not that I had rejected it out of hand. I have linked to a book, when the book is notable enough to have its own article, but I had never reflected on my reason for doing so. In retrospect, I'd justify the practice as it lets the reader see that the cited book is notable, and therefore might be more than usually worth reading for further information; and it might indicate greater than average authority behind the citation. By the way, Nature (journal) is a UK-based publication, so I wouldn't say that it is being USA-centric to assume familiarity with it. I'd guess that the upshot of this is that, as far as I'm concerned, I will at least give consideration to linking on the name of a journal when it might be helpful to the reader. Thanks for the thoughts. TomS TDotO (talk) 18:08, 16 January 2010 (UTC)
The purpose of a link is not just to let readers know a publisher is well-enough known to have a Wikipedia article; it also lets the reader now that the article exists. It might not have occurred to the reader that such an article would exist. That said, I would try to follow the principle that there need be only one link per article to another article. --Jc3s5h (talk) 18:37, 16 January 2010 (UTC)
It's inappropriate to link to the The New York Times Wikipedia article when citing a source published in The New York Times, and similarly for most other journals and newspapers. Not only is this WP:OVERLINKING, the wikilink to the journal distracts the reader from the purpose of the citation, which is to cite the article, and makes it more likely for the reader who is trying to click on the article to mistakenly click on The New York Times. I had a bad experience with an editor coming through an article and wikilinking every blasted journal title in its sea of 100 references, creating a ton of red links. I reverted and then the other editor reverted; rather than edit-war I created some stubs for the red links and then these stubs got removed by other editors as being non-notable topics (deletions that I agreed with); what a mess and a waste of time! Obscure or no-longer-published journals may benefit from wikilinks, but if you have a URL or a PMID or a DOI to an article, please don't also wikilink to the journal title, as this makes things harder on the reader. Eubulides (talk) 00:36, 17 January 2010 (UTC)
I think that red links in refs are highly undesirable, but a change to the Wikimedia software that automatically linked to Medical Hypotheses#Coverage and controversy would be a valuable service to readers. Overall, I wouldn't make a general rule one way or the other: Editors should use their best judgment. WhatamIdoing (talk) 01:17, 17 January 2010 (UTC)
I generally link to all newspapers/journals which have Wikipedia articles, as it helps the reader understand the quality of the source, if they wish to click on the link. There's no guarantee that a reader will actually be familiar with The New York Times - even if they are American. Jayjg (talk) 04:46, 17 January 2010 (UTC)
I agree with Jayjg. I often provide citations thusly:

Lowenstein, Roger. “Walk Away From Your Mortgage!The New York Times Magazine. January 7, 2010. (Retrieved 2009-01-17.)

The reader is provided with a direct link to the article, and a link to the wikiarticle about the publication should s/he be unfamiliar with it. — SpikeToronto 05:39, 17 January 2010 (UTC)
That's the way I do it too, but I don't include the "Retrieved" statements for printed material. If something has been printed, then it exists independent of the internet, and the details about when someone retrieved an online copy of that printed material are irrelevant. Jayjg (talk) 20:16, 17 January 2010 (UTC)

Rather than speaking abstractly, to give a specific example of the sort of thing that gave rise to my inquiry, it was in the article List of sequenced eukaryotic genomes which now has references like this: Douglas S, Zauner S, Fraunholz M; et al. (2001). "The highly reduced genome of an enslaved algal nucleus". Nature. 410 (6832): 1091–6. PMID 11323671. doi:10.1038/35074092.  Unknown parameter |month= ignored (help) As you see, it has both PMID and DOI, and I think that the only people who would be the least bit interested would already know about the journal Nature. And I'd like to make it clear that I inquired about this, not so much as a complaint about this particular change to this article, as to get guidance as to future action. If I should happen to make an editorial addition of a reference to Nature, I would ordinarily not make a link to Nature (journal), but I would feel uncomfortable in that not doing that would disturb the uniform appearance of this particular article, and perhaps even be considered a provocation to my fellow editor. TomS TDotO (talk) 17:12, 17 January 2010 (UTC)

Perhaps on an article that specific, maybe most readers would know Nature, but we shouldn't make the presumption and the same can't be said for every article that might reference it. Its also important to remain both neutral and consistent in such wikilinking. Why only link the ones we think readers won't know? Its presuming to determine who is reading the article and that they are of the same knowledge as ourselves. I think its far better to err on the side of caution and link all journals/newspapers/etc if they have articles or are likely too. I just don't see any's title as being a "common" term. -- AnmaFinotera (talk · contribs) 20:28, 17 January 2010 (UTC)

The normal practice in the body of an article is to only wikilink a word or phrase the first time it occurs. I don't see why the same shouldn't apply to the reference list. --Jc3s5h (talk) 17:20, 17 January 2010 (UTC)

Agreed. When I wikilink an author's name in the References list, I do it only on first reference. This policy should be consistent with Wikipedia:Linking, part of the MoS, and it cautions against overlinking. SlimVirgin TALK contribs 21:44, 17 January 2010 (UTC)
Absolutely; a journal name should be linked at most once in the references list. But it's not helpful to wikilink to a journal name like Nature; the wikilink's very small utility is completely outweighed by its disadvantages. In the above example, the Wikipedia article on The New York Times Magazine provides zero useful information about the reliability of the Lowenstein source. A style that has such a low bar on wikilinks might as well wikilink the other terms in the source, too, no? Something like this:

Lowenstein, Roger. “Walk Away From Your Mortgage!The New York Times Magazine. January 7, 2010. (Retrieved 2010-01-17.)

The wikilink to 2010 may not sound useful to you, but it's surely useful to some readers (and it helped me to find a typo in the original example!). There's just as much argument for these other wikilinks as there is for the wikilink to the article on the magazine. Eubulides (talk) 23:33, 17 January 2010 (UTC)
I don't understand. Was that sarcasm, or are you saying that we should actually link to 2010 in references? Dabomb87 (talk) 23:18, 18 January 2010 (UTC)
It was sarcasm. Apparently it didn't go far enough: perhaps I should have wikilinked the parentheses to Bracket #Parentheses ( )? Eubulides (talk) 00:19, 19 January 2010 (UTC)
No, I'm just dense. Dabomb87 (talk) 01:29, 19 January 2010 (UTC)
I agree with Eubulides. One of the most annoying things for me about reading WP has always been inadvertently clicking on links that seem to promise relevant information, only to find it's a link to an ordinary word, or the name of a newspaper I would have looked up myself if I'd wanted to. SlimVirgin TALK contribs 23:40, 18 January 2010 (UTC)
The purpose of a Wikilink is not necessarily to show that a certain citation is reliable; it is to show that Wikipedia has an article that is some way related to the article being read. Perhaps it is not useful to wikilink The New York Times Magazine since that is a general coverage magazine, so has little relationship to any article one might be reading. But it would be sensible to wikilink Sky & Telescope in any astronomy article that cites it, just because it is a magazine about astronomy and Wikipedia has an article about it. --Jc3s5h (talk) 23:47, 17 January 2010 (UTC)
I would link a journal title every time it appears in the list of references. Unlike article text, references are usually not read in a linear way from top to bottom, so the link should be available in whatever reference the reader happens to come to. — Carl (CBM · talk) 03:31, 18 January 2010 (UTC)
That's clear WP:OVERLINKING. As for Sky & Telescope, no, that shouldn't be wikilinked to either, not in a citation. Otherwise we get monstrosities like Atmosphere of Jupiter #Cited sources, in which the wikilinks to the journals wrongly dominate the citations. This is bad practice. Citations are for pointing readers to sources, not for dazzling them with complexity and leading them off into wild-goose chases. Eubulides (talk) 03:42, 18 January 2010 (UTC)
I don't think it's overlinking. We worry about overlinking in the text of the article so that it does not become a sea of blue. Linking inside references is different, for several reasons:
  • References are not read linearly, they are usually read in response to a citation. So having lots of blue isn't so much of an issue there.
  • References are more like a form of metadata than like text. If I don't recognize a journal name in a reference, it's nice to be able to click it to get the journal. Same with author names. I don't think this is a wild-goose chase: if I expect to get the article about the journal, and I do, then all is as expected.
  • Many articles use the Cite.php system, which does not sort footnotes by author. So if an author is linked only one time in the references, it will be very difficult to find that one link in among all the other unsorted footnotes. In that setting, linking the author's name every time makes sense.
— Carl (CBM · talk) 13:40, 18 January 2010 (UTC)
Atmosphere of Jupiter looks bad because there are red links to journals. Just remove those, and I think it would look much better. — Miym (talk) 15:47, 18 January 2010 (UTC)
Yes, it would look better, but it would still be a sea of blue and would still look bad. But my primary objection to this overlinking is not the visual appearance: it's that the wikilinks get in the way of the primary purpose of a citation. They make it more likely for the reader to mistakenly click on the link to Sky & Telescope rather than to the link to the cited work (which is what is important). One other point, and perhaps this is the thing that bugs me the most about it: these distracting wikilinks often impose editorial opinion and raise POV issues. If some citations' journals are in blue text and others in black (or red!) text, it's a clear indication that Wikipedia editors consider some journals to be more "important" than others. Citations should just do their job—namely cite sources–and should not be larded with these distracting and problematic glorifications of sources. Eubulides (talk) 16:51, 18 January 2010 (UTC)
In the particular case that caused me to raise this issue, there is no question of POV. The article is basically a reference list for species which have had their DNA completely described, and the references are always to highly-rated journals (I think that all of them have their own Wikipedia articles linked to). If, on the other hand, it becomes something of a standard to link whenever possible then this issue may arise.
However, I do agree that it gets in the way of the primary purpose of a citation; and that it is distracting; and that it may mislead the reader into thinking that there is something that will enhance the understanding of the topic by following the link. For readers who are "sophisticated" enough to know about the journal, it is no help; but for the readers who are "not sophisticated", it is an enticement to waste time on reading about the journal, rather than the subject. Not so much that one may accidentally click on the link, but that one might think that there is something more to be learned about the subject by clicking on the link.
I have a couple of half-baked ideas, probably not very good, but I'm trying to think of compromises. How about at the end of the footnote citations to give a "reminder" which says, "Note that many of these journals may have their own Wikipedia articles which may give information which may help the reader to understand the reliability of the citations blah blah blah".
Thank you all for a lively discussion. TomS TDotO (talk) 19:10, 18 January 2010 (UTC)
I don't think that reading about a journal is "wasting time". Also, the blue links help me see which journals have an article on wikipedia. Like I was saying, I look at it more as metadata, and less as article text. — Carl (CBM · talk) 19:17, 18 January 2010 (UTC)
A section at the end of the references would be OK; the section could even contain the wikilinks to the respective journals. Such a section would address the main objections to wikilinks, as it would avoid overlinking, would be unlikely to be clicked on by accident, and would significantly reduce the POV concerns. It'd be more work to maintain if the section listed the wikilinks, but the maintenance burden wouldn't be much. Eubulides (talk) 20:11, 18 January 2010 (UTC)
  • I agree with Slim Virgin. This sea of blue-spatter in citations needs to stop right now. How many times have I seen a URL to a NY Times or BBC News site, then a link to the organisation as whole. Why? No one clicks on these wikilinks; they click on the external URL. It is a waste of blue, which should be rationed so that wikilinking is not treated by our readers as a nuisance, as an inevitable part of every sentence on WP. We owe it to our readers to highlight the most useful links in an article, whether in the main text, the infobox, or the list of citations. As well, it's plain uglly and gives the text an unprofessional look. Tony (talk) 01:47, 19 January 2010 (UTC)
    • I'd say that the way that references are read is very different than the way that ordinary sentences are read. A sea of blue in the references section is not the same as a sea of blue in the prose. — Carl (CBM · talk) 02:02, 19 January 2010 (UTC)
      • I'm with Carl on this one. References are read one at a time, in relation to a specific point in the article, not as long paragraphs of text. The links to the authors, newspapers, journals, publishers, etc., are all useful when reading that single line. Jayjg (talk) 05:30, 19 January 2010 (UTC)

A separate issue that I don't think has been raised is that many of the links will be added automatically by citation templates, which can be copied and pasted from one article to another. It would be silly to remove information that had been added to the template when you copy it somewhere else. Over time, I would expect that the various link fields in templates will all get filled in, either by hand or by automated processes. The result will be that that more things are linked. — Carl (CBM · talk) 02:00, 19 January 2010 (UTC)

It's not silly to remove distracting and not-that-relevant information from citations, just as it's not silly to remove such information from article prose. I do it all the time. Eubulides (talk) 05:33, 19 January 2010 (UTC)
Carl, that's yet another good reason to oppose citation templates, if one were needed. SlimVirgin TALK contribs 05:36, 19 January 2010 (UTC)
One of the reasons why I like links to authors, journals, etc., is the "What links here" feature. For example, when I'm reading about a scientist in Wikipedia, I can click "What links here" and find Wikipedia articles that cite this researcher. If people don't like the blue colour, perhaps we could have a different link style that could be used in these cases? — Miym (talk) 08:56, 19 January 2010 (UTC)
Miym, it is much more efficient to use the search function, or even a WP-constrained google search. On SV's point about citation templates: yes, I'd be delighted to get rid of this cancer on WP. I never saw the original rationale for them ... Tony (talk) 10:22, 19 January 2010 (UTC)
They facilitate banking references, as in that has a large collection of references used in math articles. In this way, they serve the same sort of purpose here that BibTeX does for professional publishing. The idea of typing a reference completely by hand is quite strange when this has not been necessary in any of my published papers. — Carl (CBM · talk) 11:06, 19 January 2010 (UTC)
Miym, you may find that useful but how many readers will click on a link to see what else links to it? Links should serve one purpose: direct readers to articles that increase their understanding of the article they were reading. We should not be justifying links solely for metadata, "highlighting" important information, or autoformatting dates to the benefit of the small fraction of the percent of readers who have registered accounts and set date preferences. I don't think publisher links are nearly important enough to warrant a "different link style", either. Dabomb87 (talk) 13:47, 19 January 2010 (UTC)

I came here to comment after pondering this issue overnight, only to find that Carl had eloquently made the main points I wished to make. Though I'm generally with Tony in preferring a reduction in the number of wikilinks in Wikipedia, I don't see that the rationale behind WP:OVERLINK is applicable to references. Regarding the POV issue that Eubulides' raised, I think the solution is to link the author or journal whenever Wikipedia has an article. Doing so would ensure consistency and reduce editorial disputes. I would not assume that journals such as Nature are so well-known that they needn't be linked—I didn't know what Nature was until after I began uni, and I was a top science student in high school. I think Eubulides' point about readers clicking a wikilink when they mean to click an external link is a valid one, but I don't think it justifies omitting links that may help readers evaluate the authority of an article's sources. It shouldn't take the average reader long to realise that this -> symbol identifies the external link. Adrian J. Hunter(talkcontribs) 12:47, 19 January 2010 (UTC)

  • Omitting just the red links doesn't solve the POV problem: the presence or absence of blue links clearly indicates Wikipedia editorial position as to which journals are "important" and which are not. I've had journal article stubs removed because the journals were not considered notable enough for an article, and honestly I had to agree with their removal.
  • The "What links here?" point actually argues against the wikilinking in question: when I visit Special:WhatLinksHere/Nature (journal) I want to see other articles that are directly relevant to Nature, but what I get is a listing of over 3,000 links, almost all of them useless.
  • Automated reference generation such as zeteo is useful to a few readers, but Zeteo doesn't benefit from this wikilinking; on the contrary, since wikilinking to the journal screws up the bibliographic info that Zeteo uses, Zeteo is an argument against this wikilinking.
Eubulides (talk) 19:42, 19 January 2010 (UTC)

My general view on this issue is in line with Carl's. These wikilinks seems useful to readers who wish to learn more about the author or publication being cited. At the same time, they have little downside: their targets are clear, they don't clutter distract from higher value surrounding links (since, within the reference section, all links are probably of similar purpose and value). In general, I think we greatly overrate our ability to gauge whether links will be "useful" to readers. Christopher Parham (talk) 20:03, 19 January 2010 (UTC)

I disagree with everything Christopher Parham has said. The upside is highly questionable; the downsides are a the blue mess (whether in main text or ref list doesn't matter); the greater difficulty in picking out the external links; the edit-mode clutter that makes it harder to read and edit citations; and the huge rate of link repetition. Tony (talk) 22:38, 19 January 2010 (UTC)
We are talking about passages of text that are one to two lines long. Who could have difficulty in picking out the external link in such a small volume, regardless of the surrounding links? As for edit mode clutter, links add four characters in edit mode; this is minimal. Christopher Parham (talk) 23:40, 19 January 2010 (UTC)
Every useless artefact of computer syntax we pile onto edit mode moves Wikipedia further from the notion of "the project everyone can edit". It all adds up. We should be making it more, not less accessible. I don't mind syntax when it's useful and has little or no downside. This is not the case here. Exactly the same applies for those dreadful citations templates that have grown like a cancer on the face of WP. Tony (talk) 02:09, 20 January 2010 (UTC)