This is an essay on the Verifiability policy and the Citing Sources guideline.
|This page in a nutshell: When citing material in an article, it is better to cite a couple of great sources than a stack of decent or sub-par ones.|
Wikipedia policy requires all content within articles to be verifiable. While adding inline citations is helpful, adding too many can cause citation clutter, making articles look untidy in read mode and difficult to navigate in markup edit mode. If a page features citations that are mirror pages of others, or which simply parrot the other sources, they contribute nothing to the article's reliability and are detrimental to its readability.
One cause of "citation overkill" is edit warring, which can lead to examples like "Graphism is the study of ...". Extreme cases have seen fifteen or more footnotes after a single word, as an editor desperately tries to shore up one's point or overall notability of the subject with extra citations, in the hope that their opponents will accept that there are reliable sources for their edit. Similar circumstances can also lead to overkill with legitimate sources, when existing sources have been repeatedly removed or disputed on spurious grounds or against consensus. See also this example illustrating an exaggerated case of citation overkill.
Another common cause of citation overkill is simply that people want the source they've seen to be included in the article too, so they just tack it onto the end of existing content without making an effort to actually add any new content.
The purpose of any article is first and foremost to be read – unreadable articles do not give our readers any material worth verifying. It is also important for an article to be verifiable. Without citations, we cannot know that the material isn't just made up, unless it is a case of common sense (see WP:BLUE). A good rule of thumb is to cite at least one inline citation for each section of text that may be challenged or is likely to be challenged, or for direct quotations. Two or three may be preferred for more controversial material or as a way of preventing linkrot for online sources, but more than three should generally be avoided; if four or more are needed, consider bundling (merging) the citations.
Not only does citation overkill impact the readability of an article, it can call the notability of the subject into question by editors. A well-meaning editor may attempt to make a subject which does not meet Wikipedia's notability guidelines appear to be notable through quantity of sources. Ironically, this serves as a red flag to experienced editors that the article needs scrutiny and that each citation needs to be verified carefully to ensure that it was really used to contribute to the article.
Misuse to prove an obvious point
It is possible that an editor who is trying to promote an article to GA-class (good article status) might add citations to basic facts such as "...the sky is blue...". While this might be a good thing in their eyes, the fact that the sky is blue does not usually require a citation. In all cases, editors should use common sense. In particular, remember that Wikipedia is not a dictionary and we do not need citations for the meanings of everyday words and phrases.
Another common form of citation overkill is to load an article up with as many sources as possible without regard to whether they actually support substantive or noteworthy content about the topic. The deceptive goal here is to boost the number of footnotes present in the article as high as possible, in the hope that it will fool other editors into accepting the topic's notability without properly vetting the degree to which any given source is or isn't actually substantive, reliable, and about the subject. This is especially common in articles about people or organizations (including companies), given that they generally have to clear conditional notability standards based on achievement and sourceability, rather than mere verification of existence.
Examples of this type of citation overkill include:
- citations which briefly namecheck the fact that the subject exists, but are not actually about the subject to any non-trivial degree. An example of this is a source which quotes the subject giving a brief soundbite to a reporter in an article about something or someone else.
- citations which don't even namecheck the subject at all, but are present solely to verify a fact that's entirely tangential to the topic's own notability or lack thereof. For example, a statement of where the person was born might be "referenced" to a source which verifies that the named town exists, but completely fails to support the claim that the person was actually born there.
- a series of citations that Gish gallop their way through a rapid-fire list of content that doesn't really help to establish notability at all. For example, an article about a journalist might try to document every individual piece of work they ever produced for their employer, often citing that work's existence to itself; an article about a city councillor might try to document and source their position for or against every individual bylaw or ordinance that came up for council debate at all, regardless of whether or not the person actually played a prominent role in getting that motion passed or defeated; an article about an entertainer or pundit might try to list and source every individual appearance they might have made in media, all the way down to local morning talk shows and interviews on individual radio stations.
Some people might try to rest notability on a handful of sources that aren't assisting, while other people might try to build the pile of sources up to 20, 30, 50 or even 100 instead — so this type of citation overkill may require special attention. Either way, however, the principle is the same: sources support notability based on the substance of what they say about the topic, not just the number of footnotes present. An article with just four or five really good sources is considered better referenced than an article that cites 50 weak ones.
Overloading an article with dubious and tangential citations can rebound when the article is nominated for deletion. Reviewing editors may not be prepared to look at all one hundred citations. They may instead choose to look at just a smaller sample. If they find only unreliable sources or sources that do not discuss the subject in depth they could then recommend deletion. The two genuinely supporting sources may be entirely missed.
Material that is repeated multiple times in an article does not require an inline citation for every mention. If you mention the fact that an elephant is a mammal in multiple places in an article, provide a citation after the first one, but you need not follow each and every occurrence of the word mammal with another copy of the citation.
Avoid cluttering text with redundant citations like this:
Elephants are large land mammals ... Elephants' teeth are very different from those of most other mammals. Unlike most mammals, which grow baby teeth and then replace them with a permanent set of adult teeth, elephants have cycles of tooth rotation throughout their entire lives.
- 1. Expert, Alice. (2010) Size of elephants: large.
- 2. Smith, Bob. (2009) Land-based animals, Chapter 2: The Elephant.
- 3. Christenson, Chris. (2010) An exhausting list of mammals
- 4. Maizy, Daisy. (2009) All about the elephants' teeth, p. 23–29
In addition, as per WP:PAIC, citations should be placed at the end of the passage that they support. If one source alone supports consecutive sentences in the same paragraph, one citation of it at the end of the final sentence is sufficient. It is not necessary to include a citation for each individual consecutive sentence, as this is overkill. This does not apply to lists or tables, nor does it apply when multiple sources support different parts of a paragraph or passage.
This is correct:
In the first collected volume, Marder explains that his work is "about the affinity of life", wherein the characters "understand that ultimately they depend on each other for survival". Wiater and Bissette see this relationship as a wider metaphor for the interdependency of the comics industry. Indeed, addressing the potential underlying complexity, Marder suggests that "it's harder to describe it than it is to read it". He also calls it "an ecological romance ... a self-contained fairy tale about a group of beings who live in the center of their perfect world [and are] obsessed with maintaining its food chain", a self-described "really low concept!" Equally, he says, "the reader has to invest a certain amount of mental energy to follow the book", which includes "maps and a rather long glossary". Despite these potentially conflicting comments, Wiater and Bissette reiterate that "there is no simpler or more iconographic comic book in existence".<ref name="Rebels">[[Stanley Wiater|Wiater, Stanley]] and [[Stephen R. Bissette|Bissette, Stephen R.]] (eds.) "Larry Marder Building Bridges" in ''Comic Book Rebels: Conversations with the Creators of the New Comics'' (Donald I. Fine, Inc. 1993) ISBN 1-55611-355-2 pp. 17–27</ref>
This is also correct, but is an example of overkill:
In the first collected volume, Marder explains that his work is "about the affinity of life", wherein the characters "understand that ultimately they depend on each other for survival".<ref name="Rebels" /> Wiater and Bissette see this relationship as a wider metaphor for the interdependency of the comics industry.<ref name="Rebels" /> Indeed, addressing the potential underlying complexity, Marder suggests that "it's harder to describe it than it is to read it".<ref name="Rebels" /> He also calls it "an ecological romance ... a self-contained fairy tale about a group of beings who live in the center of their perfect world [and are] obsessed with maintaining its food chain", a self-described "really low concept!"<ref name="Rebels" /> Equally, he says, "the reader has to invest a certain amount of mental energy to follow the book", which includes "maps and a rather long glossary".<ref name="Rebels" /> Despite these potentially conflicting comments, Wiater and Bissette reiterate that "there is no simpler or more iconographic comic book in existence".<ref name="Rebels">[[Stanley Wiater|Wiater, Stanley]] and [[Stephen R. Bissette|Bissette, Stephen R.]] (ed.s) "Larry Marder Building Bridges" in ''Comic Book Rebels: Conversations with the Creators of the New Comics'' (Donald I. Fine, Inc. 1993) ISBN 1-55611-355-2 pp. 17–27</ref>
If consecutive sentences are supported by the same reference, and that reference's inline citation is placed at the end of the paragraph as described at WP:CITETYPE, an editor may want to consider using Wikipedia's hidden text syntax
<!-- --> to place hidden ref name tags at the end of each sentence. Doing so may benefit others adding material to that paragraph in the future. If that happens, they can uncomment the hidden citations and switch to citing references after every sentence. Having hidden citations could cause confusion however, especially among inexperienced editors, so the approach is strictly optional and should be used cautiously.
Another common form of citation overkill is to cite multiple reprintings of the same content in different publications — such as several different newspapers reprinting the same wire service article, or a newspaper or magazine article getting picked up by a news aggregator — as if they constituted distinct citations. Such duplicated citations may be piled up as multiple references for the same fact or they may be split up as distinct footnotes for different pieces of content, so watching out for this type of overkill may sometimes require special attention.
This type of overkill should be resolved by merging all of the citations into a single one and stripping unhelpful repetitions — when possible, the retained citation should be the originator of the content rather than a reprinter or aggregator, but if this is not possible (e.g. some wire service articles) then retain the most reliable and widely distributed available reprinter (for example, if the same article has been linked to both The New York Times and The Palookaville Herald, then The New York Times should be retained as the citation link.)
A similar case is redundant citation of an article that got its information from an article we have already cited. An exception, to many scientific and technical editors, is when we cite a peer-reviewed literature review and also cite some of the original research papers the review covers. This is often felt to provide better utility for academic and university-student users of Wikipedia, and improved verifiability of details, especially in medical topics. Similar concerns about the biographies of living people may sometimes result in "back-up" citations to original reportage of statements or allegations that are later repeated by secondary sources that provide an overview.
In controversial topics, sometimes editors will stack citations that do not add additional facts or really improve article reliability, in an attempt to "outweigh" an opposing view when the article covers multiple sides of an issue or there are competing claims. This is something like a PoV fork and edit war at once, happening inside the article's very content itself, and is an example of the fallacy of proof by assertion: "According to scholars in My School of Thought, Claim 1. However, experts at The Other Camp suggest that Claim 2."
If this is primarily an inter-editor dispute over a core content policy matter (point of view, source interpretation, or verifiability of a claim), talk page discussion needs to proceed toward resolving the matter and balancing the article. If the dispute seems intractable among the regular editors of the article, try the requests for comments process; the applicable NPOV, NOR or RS noticeboard; or formal dispute resolution.
If the matter is the subject of real-world dispute in reliable sources, our readers actually need to know the conflict exists and what its parameters are (unless one of the conflicting views is a fringe viewpoint). Competing assertions with no context are not encyclopedic. Instead, the material should be rewritten to outline the nature of the controversy, ideally beginning with secondary sources that independently describe the conflicting viewpoints or data, with additional, less independent sources cited only where pertinent, for verification of more nuanced claims made about the views or facts as represented by the conflicting sources. Sources that are opinional in nature – op-eds, advocacy materials, and other primary sources – can usually simply be dropped unless necessary to verify quotations that are necessary for reader understanding of the controversy.
Other views and solutions
Contrary views (and approaches to addressing their concerns) include:
- A cited source usually contains further relevant information than the particular bit(s) it was cited for, and its removal may be thought to "deprive" the reader of those additional resources. However, Wikipedia is not a Web index, and our readers know how to use online search engines. In most cases, if a source would be somewhat or entirely redundant to cite for a particular fact, but has important additional information, it is better to use it to add these facts to the article. Or, if the additional material is not quite encyclopedically pertinent to the article but provides useful background information, add it to the "Further reading" or "External links" section instead of citing it inline in a way that does not actually improve verifiability.
- An additional citation may allay concerns of some editors that the text constitutes a copyright violation. This is usually a short-term issue, and is often better handled by discussing the evidence on the talk page, if the additional citation does not really increase verifiability (e.g., because the original citation, with which the added one would be redundant, is to a clearly reliable source, and there are no disputes about its accuracy or about the neutrality or nature of its use).
- As alluded to above, an additional citation may allay concerns as to whether the other citation(s) are sufficient, for WP:RS or other reasons. While this is often a legitimate rationale to add an additional source that some editors might consider not strictly necessary, it is sometimes more practical to replace weak sources with more reliable ones, or to add material outlining the nature of a disagreement between reliable sources. How to approach this is best settled on a case-by-case basis on the article's talk page, with an RfC if necessary, especially if the alleged fact, topic, or source is controversial. Adding competing stacks of citations is not how to address WP content disputes or real-world lack of expert consensus.
How to trim excessive citations
If there are six citations on a point of information, and the first three are highly reputable sources (e.g., books published by university presses), and the last three citations are less reputable or less widely circulated (e.g., local newsletters), then trim out those less-reputable sources.
If all of the citations are to highly reputable sources, another way to trim their number is to make sure that there is a good mix of types of sources. For example, if the six citations include two books, two journal articles, and two encyclopedia articles, the citations could be trimmed down to one citation from each type of source. Comprehensive works on a topic often include many of the same points. Not all such works on a topic need be cited – choose the one or ones that seem to be the best combination of eminent, balanced, and current.
In some cases, such as articles related to technology or computing or other fields that are changing very rapidly, it may be desirable to have the sources be as up-to-date as possible. In these cases, a few of the older citations could be removed.
For many subjects, some sources are official or otherwise authoritative, while others are only interpretative, summarizing, or opinionated. If the authoritative sources are not controversial, they should generally be preferred. For example, a company's own website is probably authoritative for an uncontroversial fact like where its headquarters is located, so newspaper articles need not be cited on that point. The World Wide Web Consortium's specifications are, by definition, more authoritative about HTML and CSS standards than third-party Web development tutorials.
Try to construct passages so that an entire sentence or more can be cited to a particular source, instead of having sentences that each require multiple sources.
Sometimes it may also be possible to salvage sources from a citekill pileup by simply moving them to other places in the article. Sometimes, a source which has been stacked on top of another source may also support other content in the article that is presently unreferenced, or may support additional content that isn't in the article at all yet, and can thus be saved by simply moving it to the other fact or adding new content to the article.
If there is a good reason to keep multiple citations, for example, to avoid perennial edit warring or because the sources offer a range of beneficial information, clutter may be avoided by merging the citations into a single footnote. This can be done by putting, inside the reference, bullet points before each source, as in this example, which produces all of the sources under a single footnote number. Within a simple text citation, semicolons can be used to separate multiple sources.
Each of these articles has been corrected. Links here are to previous versions where a citation problem existed.
- Way too many to count
- – 83 citations for one sentence, part of 139 citations in one paragraph
- – 17 citations for one sentence
- – 54 citations to verify one statement (all but one from the same domain)
- – 18 citations for one sentence
- – 20 citations for one statement, after the phrase "advanced technology"
- – 65 citations in opening paragraph
- – many unnecessary citations
- – 14 citations for one statement
- – 16 citations
- Ora Golan - 18 citations for notable alumna
- – 6 consecutive citations of the same source in one paragraph
- – 33 citations for one sentence
- – 29 citations for one sentence
- – 20 citations for one sentence, majority of which were YouTube.
- – 77 citations for an article not even 70 words in length.
- WP:Citation underkill - An essay with a contrary viewpoint suggesting to cite every sentence/statement
- Wikipedia:Wisps' Law
- Wikipedia:You don't need to cite that the sky is blue
- Wikipedia:Masking the lack of notability
- Wikipedia:Overlink crisis
- mw:Extension:HarvardReferences – extension to improve references into Harvard style
- Wikipedia:Why most sentences should be cited
- Category:Citation overkill, for Wikipedia articles that display a case of citation overkill