Wikipedia talk:Manual of Style/Writing about fiction

It's probably time to revisit this criterion given the huge amounts of LLM-generated content we are being bombarded with, including on fiction articles. Right now, there is no requirement to cite plot sections given that the book is its own source. The implication is that people will have actually read the book, or at least sources about the book, and will not be hallucinating inaccurate text based on vectors and vibes.

In practice LLM plot summaries do tend to add a lot of synthesis as well (which is what makes them easily findable) but what often happens is that editors will rembove the synthesis slop paragraph and leave the rest alone, not realizing that the more "objective" plot summary is also suspect. Gnomingstuff (talk) 15:04, 15 August 2025 (UTC)

If a plot summary is suspected to have had some parts of AI/LLM contributions, and a version without that can't be found in the recent history, its probably best to remove it until such a time that a editor-created version can be made. Masem (t) 15:50, 15 August 2025 (UTC)

I agree that there's a high risk of LLMs getting things wrong. We need to provide human oversight. Shooterwalker (talk) 16:49, 17 December 2025 (UTC)

I've tried as an experiment many times. Except for very well-known and popular books, LLM summaries are worse than useless. For books that are part of a series they will often very confidently transpose characters from one book into another. MichaelMaggs (talk) 17:23, 17 December 2025 (UTC)

Part of the danger is it's so grammatically, and confidently wrong. It gives the illusion of good content to anyone who lacks the time or experience to verify the facts. It's why reliable secondary sources should only become more important in the era of post-intelligent information. Shooterwalker (talk) 17:39, 17 December 2025 (UTC)

Here Scribolt (talk) 09:26, 4 February 2026 (UTC)

I would like to contest a particular sentence added to the guideline in January 2025: "This content should not be included outside of the plot summary section of an article, unless also covered in reliable secondary sources" and I have boldly added the words "or a section on characters" to it. Please correct me if I have overlooked anything, but from my understanding, the updates to the guidelines in January 2025 were based on this discussion. While most of the amendments were thoroughly and repeatedly discussed and reasonably incorporated into the guideline, I did not see substantial support for the addition of the aforementioned sentence. In fact, an IP even contested it, and although they were blocked for sockpuppetry, I actually agree with the IP's proposal. Based on what Scribolt wrote, I believe his concern is that allowing primary sources could lead to abuse and the addition of fancruft or large chunks of unsourced text, especially in character articles discussing in-universe biographies and fictional settings. I fully agree with Scribolt on this point that primary source usage should be minimal. However, there are also occasions where PLOTSOURCE must be used outside the plot summary section, for instance, character descriptions in the Cast/Characters sections of films, books, video games, etc. The guideline currently seems to prohibit the use of fictional works as primary sources outside the plot summary, implying that the Cast/Characters sections in an article about a fictional work (not character articles) should be entirely supported by secondary sources. However, this contradicts reality. Many FAs and GAs rely on PLOTSOURCE to provide brief descriptions of characters. For example, in Raiders of the Lost Ark, Anthony Higgins is cited with the British Film Institute, which only mentions his character name as Gobler, while his military rank and role as "Dietrich's right-hand man" are not included; George Harris is cited with Digital Spy, which only mentions his character's name and does not specify that he is the "captain of the Bantu Wind tramp steamer"; Alfred Molina is cited with The New York Times, which only notes that this film was his debut, without mentioning that he plays "Jones's traitorous guide Satipo". If one example is not sufficient, we can consider another FA, The Shawshank Redemption. Scott Mann is cited with the Mansfield News Journal, which only states his character's name as Glenn Quentin and does not mention his role as "golf instructor" to Andy's wife; Bill Bolender is cited with Cineworld, which also only mentions his character's name without noting that he is "a convict implied to be responsible for the crimes for which Andy is imprisoned". These articles were promoted to FA status in 2021 and 2019 respectively, which is quite recent. The use of PLOTSOURCE is even more lenient in some older articles, such as the 2015 FA Characters of The Last of Us, where all biographies of secondary characters are unsourced and derived from the video game itself, same as the 2009 GA Harry Potter and the Philosopher's Stone#Characters. What I want to convey is that character descriptions in articles about fictional works often need to rely on the work itself as a primary source as well. The reason we initially allowed primary sources in plot summaries was that it is difficult to find secondary sources that fully recount the entire plot. The same applies to character descriptions. Even big films like Indiana Jones or books like Harry Potter often lack comprehensive secondary sources about supporting characters, let alone the vast majority of fictional works. Of course, I do not advocate for completely unsourced content in the Cast/Characters sections, WP:V remains one of our five pillars after all. That is why I agree with the approach taken in the FA articles mentioned above. Real-world content, such as actors' credits or the conception and creation of characters, should definitely be supported by secondary RS. However, for character descriptions that describe them within the context of the plot, I believe PLOTSOURCE should suffice. Regarding concerns about potential abuse, the current MOS:FICTIONPLOT already instructs that these statements must be "straightforward, descriptive", and "can be verified by any educated person with access". This means the allowance for using primary sources in writing character descriptions is already kept to a minimum, such as only stating their occupation ("a police officer") or relationships ("John's mother"), which already discourages editors from inserting unsourced fancruft. I think the current guideline already has sufficient preventive measures against potential abuse and this sentence seems unnecessary to me. If we retain the current wording, I am afraid that few FAs or GAs about fictional works would be able to maintain their status, as there would be "unsourced" content that is nearly impossible to find secondary sources for replacement, and it would magnify the complications in writing articles about fictional works. Therefore, I propose the change, or if any editors still find it to be overly specific, I suggest deleting the sentence entirely. —👑PRINCE of EREBOR📜 18:58, 22 February 2026 (UTC)

I think I would favor deleting that sentence entirely. I'm not quite sure what I think about anything now, partly because the 04:01, 4 August 2025 edit was revdelled, and I don't quite remember everything that was in it. There was a lot of thought and logic put into that, and it was revdelled for reasons that had nothing to do with actual content. If an admin would unrevdel it, I promise I won't make any actual edits to the page. But it could be useful for others to consider. ~2026-13018-71 (talk) 20:46, 27 February 2026 (UTC)