Wikipedia:Why most sentences should be cited

Verifiability policy requires that the material must be attributable to reliable sources. This means others must be able to check that the material comes from a reliable source. When the material is attributed to a source by using an inline citation, anyone can check to verify the statement by reading the source.

The material must also be neutrally written and not contain original research. It is better to place each citation closely where it verifies each material than use one citation at the end of a paragraph because the content may be rearranged or anyone may believe the uncited content is unsourced. Without placing each citation where it verifies each material, it could lead to concerns of original research and eventually deleted content.

Maintaining text–source integrity
One of our policies is on "text-source integrity". That policy rightfully states (as of this stable revision): "The point of an inline citation is to allow readers and other editors to check that the material is sourced; that point is lost if the citation is not clearly placed. The distance between material and its source is a matter of editorial judgment, but adding text without placing its source clearly can lead to allegations of original research, violations of the sourcing policy, and even plagiarism."

So far, so good, and in fact I have no problem with that policy. However, my understanding of it differs from that of some others; namely, I see this policy as sufficient to warrant the requirement of citing every sentence, even in cases where all sentences in question use the same reference. After all, on wiki, people can split paragraphs, move content, insert new text, referenced or unreferenced, and unreferenced sentences can lead to text-source integrity problems.

Unreferenced material
Consider the following example:

 The sun is pretty big. The moon is not so big. The sun is also quite hot.[1]

Notes

^ Miller, Edward. The Sun. Academic Press, 2005, p. 1.

It seems fine at first glance. We can assume that all three sentences come from the same cited source, right?

Wrong. How can we make such assumption in a project which "anyone can edit"? Without looking at the history of the article, one cannot be certain that all three sentences were inserted by the same author, using the same source. What if they were written by editor A, and editor B came along later, adding a new, relevant, referenced sentence, but with the ref for the third sentence only? What if the first and third sentences were written by editor A, with a reference, but the second, unreferenced sentence was inserted by editor B later?

Let's however assume that all three sentences, with the one, end-of-para reference, were written by a single editor. Let's see what can happen later, in Case Study 1:

 The sun is pretty big. The moon is not so big.

The sun is also quite hot.[1]

Notes

^ Miller, Edward. The Sun. Academic Press, 2005, p. 1. 

Here, we have editor B coming and deciding that the paragraph would be better split into two, or perhaps moving the third sentence elsewhere. And what happens? We have a new, unreferenced paragraph. Now, can readers trust unreferenced paragraphs? Heck, WP:V clearly allows for their removal...

Case Study 2

 The sun is also quite hot.[1] The sun is pretty big. The moon is not so big.

Notes

^ Miller, Edward. The Sun. Academic Press, 2005, p. 1. 

What if editor B simply decides to change the order of sentences, but keeps the citation with the intended sentence? Now the later sentences look simply unreferenced.

Case Study 3

 The sun is pretty big. The moon is not so big.[2] The sun is also quite hot.<sup id="nbFoot01a" class="reference">[1]

Notes

<li id="noteFoot01a" >^ Miller, Edward. The Sun. Academic Press, 2005, p. 1.</li> <li id="noteFoot02a" >^ Brown, Rebecca. "Size of the Moon," Scientific American, 51(78):46.</li>

</ol>

What if our editor B decides that the second sentence needs a reference, and inserts one that is good for this sentence, but not for the first one? If the original reference supported all three sentences, it no longer does – and the logic of consecutive sentences supported by the last reference suggests that the first sentence (unreferenced) is supported by the ref in the second sentence (incorrect).

Case Study 4

<blockquote style="background:white; padding:1em; border:1px solid #999;"> The sun is pretty big. The moon is not so big. The sun is yellow and the moon is gray. The sun is also quite hot.<sup id="nbFoot01a" class="reference">[1]

Notes

<li id="noteFoot01a" >^ Miller, Edward. The Sun. Academic Press, 2005, p. 1.</li>

What if editor B inserts an unreferenced sentence into our paragraph? How is the reader supposed to know which sentence comes from the end-of-para reference, and which has been inserted later?

Wikipedia articles should be trustworthy as "seen". They should not require readers to check edit history. Even then, it's somewhat understandable if one wants to verify that the article is stable; that just takes a few seconds of looking at the recent history, with no need to compare revisions. But with insufficient citations, an editor has to track down the exact diff adding content, to see if it came from the same editor who added end-of-para reference. And frankly, even then you have to assume good faith or verify all but the referenced sentences with a source to be sure that that editor did not add any sentences that are not covered by that source.

Yet my proposals to clearly state that every sentence should be referenced, cases of You don't need to cite that the sky is blue exempt, have been rejected by the community not once but twice: uno, due.

Your confused editor, signing off.

Examples
In case anyone thinks my example concern is theoretical, here's a perfect case study:
 * new paragraph, with citations, added (all sentences referenced to the same source: Kościuszko left for the United States, through Stockholm, Sweden, and London, departing from Bristol on 17 June, to arrive in Philadelphia once more on 18 August. Despite being welcomed by the American populace, he was viewed with suspicion by the United States government, at that time controlled by the Federalists, who disliked Kościuszko for his previous association with the Democratic-Republican faction. )
 * paragraph merged with the subsequent one, also using the same citations
 * citations removed, no explanation provided – presumably cleaning "excess references" because someone doesn't think all sentences need to be referenced to the same source
 * paragraph broken, resulting in an unreferenced paragraph
 * unreferenced paragraph complained about at FAC
 * citations restored (time finding them wasted: 10 minutes; time to compile this example: 20 minutes; net loss for Wikipedia: one DYK I could've written instead)

I am not going to waste more time citing further examples, but I've seen this happen enough times. Someone (me...) references sentences. Someone else copyedits the article, removing what they think of are excess citations. AND/OR Someone else breaks paragraphs, creating unreferenced ones. People complain, article is seen as unreferenced, even though someone either took the time to properly reference every sentence (or didn't...). Teeth grinding ensues.