Wikipedia:Citation templates are evil

This is an essay. It contains the advice or opinions of one or more Wikipedia contributors. This page is not an encyclopedia article, nor is it one of Wikipedia's policies or guidelines, as it has not been thoroughly vetted by the community. Some essays represent widespread norms; others only represent minority viewpoints.

This page in a nutshell: Not using citation templates is fine. Don't mess with a problem that doesn't need fixing.

Templates are evil. Well, not all templates are evil, but many are, and in particular, citation templates. Why? Because they introduce unnecessary clutter in the edit window. Wikimarkup was originally very clean and simple. It was easy to understand and hardly interfered at all with the readability of the text. Citation templates, in contrast, introduce a lengthy list of parameters, many of which are hard to understand how to use without some research, and only fulfil simple functions that can easily be achieved with plain text.

No they don't. It is remarkably easy to mess up a citation template. Category:Pages with citation errors usually has thousands of entries, and that's just the ones that the software has managed to detect and no one has got round to fixing yet. The formatting of a traditional citation in wikimarkup is very simple and straightforward: italics for the journal name, and quotation marks for the article name. Italics is a double apostrophes, and quotation marks are, well, quotation marks. That is not hard. It is childishly easy.

Yes, our first priority should be readers, next content creators, and lastly, other editors doing structural work. Let's examine the claim—how exactly are templates supposed to help readers? For the vast majority of readers, they will never look at the references at all. We are talking about a tiny minority of the readership who will even have an interest. For most readers, they are only ever likely to come across references if they spot a simple typo and try and correct it. For them, the reference text is just going to be a source of confusion, so the less of it the better.

For the small number of readers who want to follow up from the Wikipedia article with some research of their own, the references are indeed useful. But it hardly matters to them how the formatting was done internally. The hard work for them will be in obtaining copies of the sources for themselves.

{{sfn}} (shortened footnotes) keeps most of the citation text out of the body of the article. It does, but so does a plaintext shortened reference. So does the Harvard referencing which has the added advantage of making the source visible to the reader without them needing to follow any link. Harvard referencing is a good system, but it is little used on Wikipedia because the "little blue number disease" (see next section) would make the text too cluttered.

Use of sfn templates requires that the full reference list uses citation templates. The linking does not work otherwise. So to use sfn templates, we are back to messy citation templates, albeit only in the reference section. There are certain advantages to a plaintext reference list that citation templates break. One of these is that with plaintext one can ensure that the author surname appears first on the line, making alphabetical sorting easy.

First of all, this argument is throwing out the window the principle of serving readers first and content creators second. It is arguing for priority to be given to non-content creators, who we agreed above should be priority #3. Secondly, no it doesn't do what is claimed. Yes, one can click on the little blue number and whizz down to the reference list. That must save tens of milliseconds over scrolling down manually. That is not verification, that is just scrolling down, whether it was done automatically or manually. The time saved is utterly immaterial compared with the time taken for the real task of verification.

Thinking that verification consists of checking that every sentence has a little blue number and every little blue number leads to a source is one of the biggest misconceptions on Wikipedia. Demanding more and more blue numbers for ever smaller passages of text is the Wikipedia little blue number disease. Only on Wikipedia is that level of citation density demanded, and it makes writing articles ten times harder than it needs to be. That is not verification. Every well constructed hoax we have ever had here has been littered with little blue numbers, and every one has led to a reliable source. No, the real work of verification involves actually reading the source which takes many orders of magnitude longer than the time it took to scroll down the page. Just obtaining the source, when it is not online (and the best sources usually aren't), can take a significant amount of time, effort, and sometimes money.

Let's just remind ourselves on our agreed priorities. Readers #1, content creators #2, other editors #3. We didn't put bots on the priority list at all. Well, if we put them on now, they can't be higher than #4 and a bunch of stuff we haven't thought of yet will probably push them down to about #10. The idea that an automated process could not be written to extract semantic meaning from a plaintext formatted reference is just plain untenable. Just for a start, a traditional bibliography is very similar to a CSV file. Dropping it straight into Excel as a CSV will separate it into fields without any trouble. The type of information can be deduced from the position and formatting; the first field(s) will be author(s), quotation marks and double ampersands identify the first non-author fields and so on. With the technology now available it is a fairly simple matter to do such processing. It does not justify distorting the referencing system to benefit a few backroom processes. Bots are meant to assist the creators, not the other way round. In short, the argument simply does not hold water; glug-glug-glug, it just sunk.

No. WP:CITEVAR explicitly proscribes changing citation styles (including adding templates) without the consensus of the page editors. Those who believe that not using templates is "non-standard", "against consensus", "virtually equivalent to bare urls", etc, should first get this so-called consensus agreed at a policy discussion venue and then have it written into the guidelines. Good luck with that, but be warned, this has been discussed to death numerous times in the past and the policy remains that Wikipedia does not have a house style for citations.

Numerous articles completely bereft of citation templates have become Featured Articles and appeared on the main page. This includes some recent articles, not just dated pages before our policies were so well developed. So even for our very best content, the consensus is that citation templates are not a requirement. Featured content is heavily reviewed and is our yardstick for what makes a high quality article. When stylistic elements have reached Featured status, that is a good sign that they should not casually be changed by drive-by editors.

No, not in general. If there are specific aspects of a particular article that militate towards a particular citation style, then that is a legitimate topic for discussion at that article. It is not reasonable to ask an editor to repeatedly defend their choice of citation style on article after article because the questioner prefers something else. If it is desired to prescribe that a certain style be used, or proscribe a certain style from being used, then the correct venue for that is a policy discussion page such as WP:VP/P. However, anyone considering that would be wise to read through Wikipedia:Perennial proposals first. Such proposals have been repeatedly rejected in the past.

WP:Bundling is the combining of multiple citations to the same source into a single line in the references section. It requires a name to be added to each ref tag, which of course, is more clutter for new editors to have to try and work through. Bundling may be helpful in some cases where the citation style is to place the full reference in the body of the article rather than the bibliography. However, it does create a mess when the citations being bundled are to different page numbers in the same source. This is a common problem with bundling, and the solutions are either to unbundle or use the {{rp}} template. This latter solution not only introduces code clutter in the edit window, it introduces visible clutter in the article.

When applied to an article using shortened footnotes and a separate bibliography, bundling is positively evil. With shortened references, the space saved is quite minimal, and it is space at the bottom of the article where nobody cares about it. It puts the references badly out of sequential order. Sequential order is a great boon if one want to check, say, where a whole paragraph is cited to. All the references to that paragraph are consecutive, not all over the place in the references list.