User:RexxS/NoCiteBar

There really needs to be no bar on improving citations in Wikipedia. At present far too many articles are fossilised into using unsuitable citation styles because an early author chose (or made up) a format that worked at the time, with one editor, but which has since proven to be a poor choice when other editors have expanded the article. Just to be clear upfront, the solution is not to stop all the other editors from editing the article. This essay attempts to explore the problems and suggest some – sometimes painful – solutions.

What ENGVAR is (and CITEVAR is not)
The reason why MOS:ENGVAR is such a useful guideline to settle issues of which spelling style to use is: The parallel guideline MOS:DATEVAR has exactly the same purpose and decision scheme and is completely uncontroversial in deciding whether to use the completely equivalent DMY or MDY date formats.
 * 1) that it has an agreed purpose: to prevent back-and-forth edit wars between two editors over two completely equivalent styles;
 * 2) that it is amenable to an agreed algorithmic scheme of deciding the outcome: opportunities for commonality; strong national ties; talk page consensus; retaining the existing variety; first post-stub revision that introduced an identifiable variety.

However, WP:CITEVAR is not comparable because it does not compare a limited number of equivalent styles. Not all styles are created equal, and the debate cannot be framed as a choice between two equally valid options. Wikipedia editors might use one of two standard templates, but there's no guidance on why that might be a good idea (e.g. once one has been chosen, subsequent editors can follow that precisely). In fact, Wikipedia editors are free to make up their own personal style of citation format, with rules that only they themselves are aware of. The purpose of CITEVAR then becomes solely that of giving authority to one editor to impose their personal preference on an article, no matter how poor a choice that may be. Not only that but it becomes a centre-point of the efforts of small cadres to impose a walled-garden around the articles they wish to establish ownership over.

What's MLA
Modern Language Association format (MLA) is a style used for written papers that are not subject to update. It comprises two parts: a short citation and a long citation. According to Dennis G. Jerz and also The University of North Carolina, MLA format requires parenthetical inline short citations. To complete the picture, Purdue University defines the style of the long citation as "These are the general pieces of information that MLA suggests including in each Works Cited entry. In your citation, the elements should be listed in the following order: <> Each element should be followed by the punctuation mark shown here."

Compare that with Wikipedia's CS1 default style for cite book which gives the full citations the format <> CS1 templates consistently impose that order on citations, removing the chance for inadvertent variations. It's difficult to understand why a minor rearrangement, for the sake of improving consistency, should be the source of outrage. For what it's worth, MLA dictates "1-inch margins all around; double-spaced; no extra spacing after paragraphs; 12-point typeface (usually Times New Roman)" – none of which are in place in our articles either.

Multiple books
If multiple books by the same author are cited, MLA uses <> for the short citation.

Compare that with sfn and harvnb which consistently use <> in every single case.

Purpose of a citation
MLA is unsuitable for Wikipedia. The first purpose of a citation is to allow a reader to verify the source (book, journal, etc.) that supports the Wikipedia text under the scope of the citation, and the format of the citation should allow that to be done easily and unambiguously. It is perfectly acceptable to use (author page) as the short citation style where you have a brief written paper with fixed text, but that is not suitable on a collaborative project where a different editor may later come along and add another citation – especially in the field of literature where an expert author will write more than one book – because there is then no way for the reader to work out from <> whether they have to consult page 270 in <> or <> when they look at the long citations. It is precisely this flaw in many of the older references that led to the mess that many of our older articles have become.

Does MLA fit our purpose
Unfortunately, MLA – and many other "made-up" schemes, purportedly based on it – does not measure up to what we need in a reference scheme suitable for use by many editors across many articles with content always subject to update.

MLA is unsuitable for use in Wikipedia, and so is any custom style that uses short citations with (author page) as default. This is because our article content, unlike a written essay, is not fixed. Using an "MLA-type" scheme, one editor may cite <> and accurately use <> as the short cite for page 2012. Another editor may later cite <> and inaccurately use the short cite <> as the short cite for page 2003 of that work. Then nobody can tell which work supports which piece of text in the article, without having to consult both works and doing the detective work. That's just unsuitable for collaborative work. Let's say the second editor realises that it's a "multiple-works-by-the-same-author" situation when they add the long citation. They then have to make up a short title for their short cites – perhaps <> (an interesting new use of the adjective "short" that I hadn't come across before). You now need good eyesight to see that 1984-94 is part of the "short" title, not another page range. But worse is that all the original <> short cites have now become incorrect. The second editor has to find all of the <> cites and convert them to <>. Good luck with that. It also makes importing text from other articles problematical because different editors can make up different short titles like <> which don't fit the convention used in the other article.

What's available
We want robust references that: To achieve that, we should use a scheme for the short citations that identifies the source by author/authors and year - consistently. There needs to be a link, preferably an automatic one, from the short citation to the long citation. There should be assistance scripts available to highlight errors and inconsistencies. The obvious candidate meeting all of those is sfn, shortened footnotes. They have the added advantage that should an editor add a duplicate citation, it is automatically aggregated with its duplicate, removing the need for an editor to search and then use named references.
 * 1) ensure that there can be no ambiguity about the source document;
 * 2) encourage consistent presentation;
 * 3) facilitate maintenance;
 * 4) are portable between articles.

Should we use p. and pp. before page numbers
Yes. Use p/pp in front of the page numbers. It then helps maintainers to work out whether a citation added as <<Bennett and Elliott 2005>> means <<Bennett and Elliott (2005) page number missing>> or <<Bennett and Elliott (year missing) page 2005>>. We're not so short of space that two or three extra characters make any significant difference and they improve the readability for the average reader.

Can we leave out the year
No. The short citation should contain the year in every case. The year is often a necessary piece of information when disambiguating the source, and it should not be optional because consistency in display is more important than saving a few characters in half of the short citations.

Can we have one long citation for a collection
No. When you cite several chapters from the same collection written by different authors, it is best to create a separate long citation for each author. If you cite a chapter written by Sutherland in a collection edited by Todd, the short citation should still be <<author year, p.page>> (i.e. <<Sutherland 2005 p. 21>>) without any need to mention Todd, who will be credited as the editor of the collection in the long citation <<>>. This incidentally correctly attributes Kathryn Sutherland as the author, which is not done in made-up schemes like <<Todd, "Chronology of composition and publication", 21>> and its ilk. Incidentally, unless the collection restarts page numbering with each chapter (and examples of those are about as common as rocking horse shit), there's no need to mention the name of the chapter as the page number is perfectly sufficient. Following the urge to name chapters as a means of showing your erudition is just intellectual masturbation.

How do we match short and long citations
Shortened footnotes will automatically create a link if the long citation includes harv (although that can be customised if necessary by using SfnRef). Ucucha's script will automatically identify short cites without a corresponding long cite, and vice-versa. It works with sfn and its close cousin, harvnb, and shows errors comprehensively on the longest of pages.

Anyone who wants to suggest a different viable scheme will need to demonstrate that it offers advantages comparable to sfn.

Bundling citations
Bundling is a poor idea for a Wikipedia article. Once again, it is fine for written text that does not change, but causes problems when text may be altered, moved around, or exported by other editors. An electronic encyclopedia is best served by a separate citation for each source used. A direct correspondence between the in-line attribution and each single short citation, linked to the long citation, is easier to read and simpler for later editors to mimic the style. How do you verify when there are three bundled sources citing two claims? The first source and last source are clear; but it's not obvious which claim the second source supports. Look at how many permutations there would be for five bundled cites supporting three claims.

To take an example, "The sky is blue and the sun is yellow.<sup style="color:blue;">[51][52] " is easy to convert to "The sky is blue.<sup style="color:blue;">[51] ... [lots more stuff in between] ... The sun is yellow.<sup style="color:blue;">[52] " or "The sky is blue, the sun is yellow and the sea is green.<sup style="color:blue;">[51][52][53] ". I can also easily use "The sky is blue.<sup style="color:blue;">[51] " in another article. Now try that with "The sky is blue and the sun is yellow.[51]" Editors separate and re-arrange statements all the time, and there's no good reason to deliberately make life far more difficult for them, when the simplest solution is the most useful.

Experienced editors will tell you they use bundling all the time and have no problems. But not every editor has the experience to know when bundling can be done sensibly, and they don't adhere to the restrictions more experienced editors place on themselves. New editors arrive at an article, see several citations bundled together, and then blindly copy the idea themselves with exactly the results outlined above. The only sure-fire way to avoid the issues is not to use bundling at all. It is unnecessary complication for little result. It also bloats the section containing the short citations because you create a different short cite for each bundle, which prevents the use of named references (or sfn) for duplicates.