Wikipedia talk:Verifiability/Archive 81

This is an archive of past discussions about Wikipedia:Verifiability. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 75

←

Archive 79

Archive 80

Archive 81

RfC concerning WP:ABOUTSELF

There's an RfC about the applicability of WP:ABOUTSELF to an article on a religious organisation happening at Talk:International Churches of Christ#Request for Comment on About Self sourcing on beliefs section of a religious organization’s article, which might interest watchers of this page. Cordless Larry (talk) 19:55, 15 April 2024 (UTC)

Feedback requested on whether usage examples require sources

Your feedback would be welcome at this discussion about usage examples at linguistic articles, and whether they require sources: Talk:Franglais#Are usage examples "original research"? Thanks, Mathglot (talk) 17:16, 22 April 2024 (UTC)

Responsibility section

Right now, this guide reads Any material lacking an inline citation to a reliable source that directly supports the material may be removed and should not be restored without an inline citation to a reliable source. But the footnote in that quote reflects very different principles: The location of any citation—including whether one is present in the article at all—is unrelated to whether a source directly supports the material.; ie, this footnote belongs on a sentence that says something like "Any material that is not directly supported by a reliable source may be removed." The change to the base text that set this footnote out of wack is old enough that it doesn't show up in the last 50 edits, but I'm not sure if wider consensus was sought for it beforehand. If there was, it's evidently the footnote that needs updating. -- asilvering (talk) 15:02, 29 April 2024 (UTC)

The change was made by WhatamIdoing in June 2021,[1] in a change to standardise the language between BURDEN and CHALLENGE. -- LCU ActivelyDisinterested «@» °∆t° 15:21, 29 April 2024 (UTC)

The footnote is trying to provide a definition of what it means if a source Wikipedia:Directly supports the material.

We have had editors mistakenly think that the location of the little blue clicky number is what determines whether the source directly supports the material. In this error, they mistakenly believe that:

Born in 1927,^[1] Alice Expert became famous for her interest in expertise.

has a citation that "directly supports" the birth year, even if the cited source is all about elephants and does not mention either Alice or 1927 at all.

What we're trying to tell them is that "directly supports" means that if the CHALLENGED fact is about Alice's birth year, then citing a source about elephants is no good, even if the little blue clicky number is actually touching the year. The cited source only "directly supports" the claim if the cited source contains words like "Alice was born in 1927" – and if it does that, then that cited source "directly supports" the claim even if it's at the end of the paragraph (or, technically, if it's not yet cited in the article – though a CHALLENGE requires that it actually be cited). WhatamIdoing (talk) 15:45, 29 April 2024 (UTC)

Ok, but that's not what the footnote is saying. If that's what we want the footnote to say, it should say that. -- asilvering (talk) 02:12, 30 April 2024 (UTC)

@Asilvering,

The footnote says "The location of any citation—including whether one is present in the article at all—is unrelated to whether a source directly supports the material."

I'm telling you that the meaning of the footnote is that the location of the citation is irrelevant to whether the source directly supports the material. What's relevant to the question of whether the citation WP:Directly supports the material in the Wikipedia article is whether the material in the source matches the material in the Wikipedia article.

It seems to me that my explanation is exactly "what the footnote is saying". What do you think the footnote is saying? WhatamIdoing (talk) 21:16, 2 May 2024 (UTC)

What the footnote currently says is, including whether one is present in the article at all. It is telling you that a sentence may indeed be directly supported by a reliable source, even if that source is not cited in the article. That is, an article may have zero references at all, but it is in principle possible to find a source that contains the same information; this is a statement about what makes a fact verifiable or not. That is irrelevant to the sentence the footnote is attached to, since that sentence is saying material should not be restored without an inline citation; it is a statement about what to do once a fact has been challenged and needs to be verified.

Since you've clarified what this is supposed to mean, I can fix it by removing the "including whether one is present in the article at all" bit, so I'll do that. -- asilvering (talk) 02:35, 3 May 2024 (UTC)

I think you've understood the sentence, but you don't believe it.

Fact: A source directly supports the content when the content of the source matches the contents of the article.

Where the citation is placed is irrelevant to whether the source directly supports the content.

Obviously logical conclusions, given a source that says the same thing as the article:

If the citation is placed at the end of the sentence, then the source directly supports the material.
If the citation is placed at the end of the paragraph, then the source directly supports the material.
If the citation is placed at the end of the section, then the source directly supports the material.
If the citation is placed at the end of the article, then the source directly supports the material.
If the citation is placed in the ==External links== section, then the source directly supports the material.
If the citation is placed in an edit summary, then the source directly supports the material.
If the citation is placed on the talk page, then the source directly supports the material.
If the citation hasn't been typed into Wikipedia yet, then the source still directly supports the material.

Naturally, that last case isn't useful to any other editor, and it doesn't meet the requirements of WP:BURDEN (which requires the citation to be formatted as an inline citation, implicitly in the article itself and [for practical reasons, as a means of communicating the relationship between the source and the material to whichever editor complained about the absence of an inline citation] plausibly near that material), but directly supports itself, strictly speaking, is about whether the source matches the article content. It's not about where the citation is located, or even whether the citation has been typed up yet.

I think what you might want is to add a sentence that says something like "Of course, in case of a CHALLENGE, you really do have to have an inline citation to that source" – though I'd consider that to be redundant with the main text of the policy. WhatamIdoing (talk) 05:52, 7 May 2024 (UTC)

Why is this still being discussed after all these years?

"Naturally, that last case isn't useful to any other editor" applies to any placement of the citation that is removed from the content supported by the citation. The citation should be close to the content it supports. (How close can be discussed, but not further than the end of the sentence. You reveal you understand this point. The citations is not about the whole article, but about a specific piece of information in the article, and it should be located adjacent to that piece of information.) Your long repetitions border on a rather silly WP:POINT violation, so we won't take it too seriously. We understand your point, and it is (still) not valid. -- Valjean (talk) (PING me) 06:03, 7 May 2024 (UTC)

The reason it's being discussed is because an editor appears to have misunderstood which words in the sentence provide him with an excuse to object about the location of the citation, and which words provide him with an excuse to object about the contents of the cited source.

The first sentence using this says:

"All quotations, and any material whose verifiability has been challenged or is likely to be challenged, must include an inline citation to a reliable source that directly supports the material."

The words "an inline citation" are the words that say the citation has to be close to the CHALLENGED material.
The words "directly supports the material" are the words that say the contents of the article must match what's in the source.

If your complaint is that the citation is 'physically' in the wrong place, including that it's not in the article at all, then say something like "That's not a proper inline citation, which is required by WP:V."

If your complaint is that the citation is 'physically' in the right place, but the Wikipedia article is talking about when Alice was born and the cited source is talking about elephants, then say something like "That source does not directly support the claim. According to WP:V, a source "directly supports" a given piece of material if the information is present explicitly in the source, so that using this source to support the material is not a violation of Wikipedia:No original research, and no amount of ~~rearranging deckchairs~~ changing the location of the ref tags is going to change the fact that the cited source doesn't directly support the material."

Getting editors to use the right words helps people understand each other. This isn't necessarily popular (for example, a couple of years ago, we had a high-volume editor claiming that all uncited material is automatically a NOR violation, even though the second paragraph of NOR contradicts that), but it is important, especially for editors who are trying to change policies. WP:Policy writing is hard, and it's almost impossible if you don't pay attention to the differences like which words indicate the location of the citation and which words, in the same sentence, tell you something about the nature of the cited source. WhatamIdoing (talk) 06:24, 7 May 2024 (UTC)

! agree with your statement. I have always understood that a cited source needed to directly support content in the article, but when I first started editing WP more than 18 years ago, I did not understand the importance of having the citation as close to the supported content as possible (nor, it would appear, did many other editors at the time). Having seen how often well-intentioned edits move citations away from the content they support, I now support citing individual sentences within paragraphs, when applicable. We have {{Failed verification}} for a source that does not support the content, but if a citation supporting some specific content is not close to that content, then, if an editor does not have the time and patience to search for the displaced citation, {{Citation needed}} may be added. These are distinct problems, and policies and guidelines need to make that distinction clear. Donald Albury 14:18, 7 May 2024 (UTC)

WhatamIdoing and Donald Albury, thanks for the clarifications. I agree. The location of the citation should be close enough to the relevant content that a reader will logically make a connection between the content and its citation. If the citation gets moved or is otherwise located too far away, then placement of {{Failed verification}} or {{Citation needed}} tags is appropriate. That's why I object to the practice of moving all citations to the end of a sentence or paragraph. Some citations need to be placed exactly right after a specific word or phrase and should not be moved, especially for sensitive BLP matters. -- Valjean (talk) (PING me) 15:41, 7 May 2024 (UTC)

I think you've understood the sentence, but you don't believe it. is not true, and I would appreciate it if you stopped trying to speculate. -- asilvering (talk) 16:37, 7 May 2024 (UTC)

Well, do you? Do you actually believe that the source I'm going to cite later today "directly supports" the contents that I'll be citing it for? Or do you believe that since there's no citation in the Wikipedia article yet, that it can't possibly "directly support" that material? WhatamIdoing (talk) 16:45, 7 May 2024 (UTC)

If it isn't placed logically close to the content, it doesn't "directly" support the material. The cite's placement somewhere in the universe does mean it "supports" that material, but that is irrelevant for our purposes. We need it "directly", IOW located "closely", to support the material, hence the need for it to be located "inline". That's pretty close. -- Valjean (talk) (PING me) 17:02, 7 May 2024 (UTC)

No. That's not what's intended.

If it isn't placed logically close to the content, then the content is uncited. Uncited is not the same thing as "directly supports".

We need the source both logically close to the content ("inline" with that content) and for the source to directly support the content ("source and Wikipedia article say the same thing").

Maybe we need to use other words for this. @Valjean, can you imagine a source that indirectly supports content, so that (to reverse the requirement in the policy) "using this source to support the material ~~is not~~ would be a violation of Wikipedia:No original research"?

For example: Imagine a pair of tweets saying "I got married today" and "I'm in London today". It would be a violation of NOR to turn those into "Chris Celebrity got married in London today", right? It's a violation because no source directly says that the wedding was in London – they only indirectly imply it, right?

I know we agree that this problem can't be solved by moving the refs around. Even if you format it as "Chris^[1]^[2] Celebrity^[1]^[2] got married^[1]^[2] in London^[1]^[2] today^[1]^[2]", that pair of tweets is still a NOR violation, because they only indirectly support the claim. This would be a violation of the requirement for the source to WP:Directly support the content, even though those ref tags are literally touching every single significant word in the sentence. WhatamIdoing (talk) 17:55, 7 May 2024 (UTC)

Wouldn't it be a SYNTH violation to use them that way? -- Valjean (talk) (PING me) 17:59, 7 May 2024 (UTC)

It absolutely would violate SYNTH.

And SYNTH, as every experienced editor knows, is part of NOR.

And this footnote says that if the only way your source (NB: not "location of ref tags") could be said to "support" your content is a NOR violation, then your source (NB: not "location of ref tags") does not "directly support" the content. WhatamIdoing (talk) 18:04, 7 May 2024 (UTC)

That wording really confuses me. There are too many different elements for me to be sure how to parse it. Is there a word missing somewhere in all that? -- Valjean (talk) (PING me) 18:32, 7 May 2024 (UTC)

I don't think there's a missing word.

What's difficult about saying that sources that don't say Chris got married in London don't support a claim that Chris got married in London? WhatamIdoing (talk) 18:37, 7 May 2024 (UTC)

There is nothing difficult about that at all. This is about SYNTH violations. Is that really your main point in this thread? -- Valjean (talk) (PING me) 18:41, 7 May 2024 (UTC)

That is the main point, not "in this thread", but "in the policy itself". A source does not "directly support" content if it's a NOR violation for that content. The definition of "directly supports" is given in that footnote. The definition is: When you compare the source to the article's contents, it wouldn't be a NOR violation to claim that the source and the article are saying the same thing.

I think the main point from other editors is "I want a sentence in an actual policy that will let me insist that the location of the citations be changed". Some editors have gotten into the habit of claiming (and genuinely believing) that "directly supports" is about the location of the citation, rather than the contents of the source. I've no objection to them having such a sentence. I only object to them using words about source–text integrity ("a reliable source that directly supports") when talking about formatting (e.g., the location of the ref tags). WhatamIdoing (talk) 18:53, 7 May 2024 (UTC)

What I believe about the statement in question, ie the note, is irrelevant to the question I raised in my initial post, which is that the two statements we have (the sentence in the main text, and the sentences in the note) introduce an unnecessarily confusing discrepancy. I understand both statements perfectly fine. What I am saying is that it is unhelpful to add "including whether one is present in the article at all" in the note when we are talking about what to do when "material lacking an inline citation to a reliable source that directly supports the material" has been challenged in the main text. Indeed, the location of the inline source is irrelevant when it comes to whether information is verifiable or not. However, once material is challenged, it is in fact very relevant that sources are present in the article.

I posted on this talk page about it because it appeared likely to me that this discrepancy was the result of someone changing the main text and forgetting to change the footnote. If that was the case, it was possible that the change was done without wider consensus. I received my answer; this was not the case. Great! So, I edited the footnote to suit its context. I'm not sure why you reverted that, since my changes did not change the meaning of any of the text. -- asilvering (talk) 19:07, 7 May 2024 (UTC)

The whole note:

A source "directly supports" a given piece of material if the information is present explicitly in the source, so that using this source to support the material is not a violation of Wikipedia:No original research. The location of any citation—including whether one is present in the article at all—is unrelated to whether a source directly supports the material. For questions about where and how to place citations, see Wikipedia:Citing sources, Wikipedia:Manual of Style/Lead section § Citations, etc.

The phrase The location of any citation—including whether one is present in the article at all—is unrelated to whether a source directly supports the material. should be removed. It just creates unnecessary confusion. -- Valjean (talk) (PING me) 18:32, 7 May 2024 (UTC)

That depends on whether your goal for the footnote is to make people understand what the words WP:Directly supports mean in this policy, or whether your goal is to explain how to properly place an Wikipedia:Inline citation. WhatamIdoing (talk) 18:36, 7 May 2024 (UTC)

I think the problem is one of grammar; people sometimes incorrectly read it as "must include an inline citation that directly supports (that means 'touches', right?) the material (and by the way, this is WP:V so of course we're only talking about citations to reliable sources, but that's not really relevant here)". Instead, what the policy says is:

Under specified circumstances, the article "must include an inline citation", and
that the citation must point "to a reliable source", and
that the cited reliable source (NB: not the citation itself) "must directly support (i.e., match) the material".

Here's the first sentence about this in the policy and the footnote, with a few adjustments that might make it easier to understand:

The footnote would have to be adjusted to match the new language:

Alternatively, we could add a sentence such as {{xt|"Even if the Cite error: A <ref> tag is missing the closing </ref> (see the help page).

References should be as specific as possible, so other editors can easily verify that the information is correct. So the reference is for the 7 June update, rather than the general page for the Energy and Climate Change Ministerial Council. -- LCU ActivelyDisinterested «@» °∆t° 11:32, 29 August 2024 (UTC)

Thank you! You are so kind. Purplewhalethunder (talk) 06:22, 2 September 2024 (UTC)

References

Purplewhalethunder's most recent attempt didn't include external links and did include citations. There is no verifiability issue here. While there were problems with the previous edits, this one looks like an improvement to me. Yes, it relies on primary sources, but the old version is the one that included commercial sites as sources. @Purplewhalethunder: Wikipedia has a preference for the high quality independent sourcing. So, for example, a newspaper, magazine, journal, or book which talks about those certifications rather than the website of the certification itself. IMO MrOllie should self-rv the most recent edit, as it's overall an improvement that could be fixed up from there (the section heading, for example, which shouldn't include the title of the article). Finally, this isn't a good page to ask for help with content. I'd recommend visiting the WP:TEAHOUSE for that. — Rhododendrites ^talk \\ 12:02, 29 August 2024 (UTC)

Thank you. I will keep this in mind. Purplewhalethunder (talk) 06:51, 2 September 2024 (UTC)

FAQ item

I have added this information to the FAQ:

Are sources required in all articles?: Adding sources is the best practice, but prior efforts to officially require at least one source have been rejected by the community. See, e.g., discussions in January 2024 and March 2024.

My main goal is to have a place to store links to these discussions. I don't really want to put it in Wikipedia:Perennial proposals, because I hope that the answer will someday change (i.e., officially. It's already the actual practice by NPP and AFC). I have accordingly tried to write this so that it encourages the addition of sources but also admits that we have so far been unable to get this adopted as an official requirement. WhatamIdoing (talk) 02:01, 7 September 2024 (UTC)

Print only newspapers

I can understand allowing paywalled sources, since it's possible for others to check and verify that information. Same thing with offline books. While I don't go into libraries, others do and often check these books, and that's enough for me. Also somewhat applies to the "rare museum sources" that the page mentions.

But isn't the whole point of print-only news stories that they're only published once and never again? How can that be verifiable? I know many people have copies of these, but these people will at some point quit Wikipedia, and when that happens, the source will essentially become Lost media. And it becomes even more complicated if the source was carried over from a page translation.

I don't want them to be banned or anything like that, but I'm curious on why there aren't any restrictions or guidelines on using them.

Needless to say, my comment does not apply to archived newspapers such as the ones on Newspapers.com. I'm talking about ones like this ^[1]^[2] Bonus Person (talk) 01:53, 14 August 2024 (UTC)

References

^ Fictional News Place "Person accidentally buys yacht" 7 November 1987
^ The Gallifreyan Post "Tardis stolen at 4 AM" 5 August 1980

Newspapers.com and other sources like that exist because copies persisted for decades, long enough to be digitized. Some GLAMs collect newspapers in some non-web format, such as microform or microfiche (or the print itself); for that matter, so do some news publishers. Nikkimaria (talk) 01:59, 14 August 2024 (UTC)

Yup, the New York Public Library (for example) has an entire section devoted to periodicals and newspapers. Blueboar (talk) 21:58, 14 August 2024 (UTC)

The entire run of my hometown's defunct paper is sitting on microfilm in our town's historical society, and the main newspaper for the county is sitting on microfilm in several libraries here in the county. The local university is digitizing those microfilms, but even if they weren't, the paper is still verifiable by visiting a library with the microfilms. I believe the Library of Michigan also contains microfilms for all of the major papers in the state, so someone wouldn't even have to travel to my remote corner of the country to access the issues not yet digitized.

In sum, I don't see a problem with citing archived newspapers, since they're just that, archived. They may be even be more available than the rare sources we also allow because they're archived in multiple locations, as per my anecdote above. Imzadi 1979 → 23:47, 14 August 2024 (UTC)

Libraries hold newspapers too. The British Library has every issue of every newspaper printed in the British Isles since 1840, for example, and provides digital copies online. Support your local library, Wikipedia couldn't exist without it! – Joe (talk) 10:58, 16 August 2024 (UTC)

If the newspaper isn't defunct, then you can usually go to the building and ask to see the old newspapers. A copy of every single newspaper usually ends up bound into a large book and stored in the "morgue", for the newspaper's own reference purposes. In pre-digital days, these were traditionally used by the staff (e.g., to remind themselves which restaurants ran ads for this holiday last year, or what the editorial said about the mayor during the previous election), but they are generally open to the public. In the US, the morgue for a defunct newspaper tends to end up either at the local library or sometimes the local historical society. WhatamIdoing (talk) 19:42, 18 August 2024 (UTC)

That may no longer be true everywhere. A few years ago when I checked with the local daily, I was told they no longer kept old issues for more than a year. Previous copies had been microfilmed and sent to the library district. The district in turn sends microfilm rolls that are more than a few years old to the local historical museum, which keeps them in an unstaffed archives/library building. Access requires staff being available to go over to the unstaffed building to let you in and sit there while you search the microfilm (Disclaimer: I used to work at the library and volunteer at the museum, but do neither now). I did find old issues (1970s) of the paper online, but I don't remember where. Donald Albury 20:01, 18 August 2024 (UTC)

Our policies presume a dichotomy/ binary flow chart.....if it's accessible with great effort, hours, $$ and difficulty it counts (in this respect) the same as something that anybody can verify on line in seconds. The reality is is that there are degrees of verifiability. I'm not advocating for any structural changes for that except to acknowledge that in discussions that this be recognized. Verifiability which requires great effort and investment to verify is weaker than something that can be easily verified. North8000 (talk) 20:33, 18 August 2024 (UTC)

They also assume that someone falsifying verifications will eventually be caught, that if one was able to access it another will be able to check it and that editors who repeatedly falsify verifications will be permanently banned. Yes arguably its an easy system to abuse once or twice, but the more times someone falsified obscure verifications the greater the chance their whole house of cards would come down. Horse Eye's Back (talk) 20:44, 18 August 2024 (UTC)

There's a time element to consider there too. We use a lot of ephemeral internet sources that are highly available now but frequently disappear. This means that in practice we are dangerously reliant on a single archive for the verifiability of huge swathes of coverage. A traditional, physical archive is less easy to access now but far less likely to simply disappear overnight, and if anything liable to become more accessible over time (as librarians and archivists continue to work on indexing and digitisation projects etc.) – Joe (talk) 04:07, 19 August 2024 (UTC)

I'll also note that the importance of a newspaper citation tends to vary with its accessibility. Some podunk place has an article on a now-celebrity's performance in Academic decathlon or their high school musical from 20+ years ago? Not the most earth shattering bit of trivia to be sourced, nor generally the most controversial or damaging if falsified. Jclemens (talk) 04:47, 19 August 2024 (UTC)

Our policies are also explicit (not presumptive) that they value editorial judgement on evaluating sources. If all you have to verify an article is print RS that are difficult to re-verify, that's great, grumble-grumble-but-nobodys-gonna-verify-for-years (so even if the article was originally written in pristine quality, there will be citation drift within months). If you can verify content with a comparably good online source, online excerpt of the print source, or whatever, then you can append those citations inline in parallel with the print source; if you find online-accessible sources that are even better (such as new reviews/retrospectives years later), so much the better.

And of course since an old newspaper is a wp:primary source, you'll almost certainly be looking to cite it within the context of a secondary source that makes reference to that article or coverage. You'd keep the primary citation (from which you may still be drawing a direct quotation or something), but you have the secondary citation that gives the main structure to that section that is easily verifiable (a primary historical source should probably not be the only thing supporting the structure of an entire section in an article -- but I'd be interested to see what article you're working on so this is no longer an abstraction). SamuelRiv (talk) 14:00, 16 September 2024 (UTC)

No mention of Template:Image reference needed

We describe the use of {{cn}} but not {{Image reference needed}}. Should we include that? We do have quite a variety of maps and infographics. NebY (talk) 20:25, 9 September 2024 (UTC)

I hope the community will recognize the enormous problem on WP (and on publications sourcing from Commons) of unsourced charts, maps, and diagrams, (as well as inadequately sourced photos), and that taking the first step of simply asking for sources will not break WP as we know it. In my last attempts at raising the issue I was quickly shouted down.

Making a recommendation here might help. I would support adding {{image reference needed}} and variants as a recommended tag on V (along with contacting the image author directly). SamuelRiv (talk) 20:44, 9 September 2024 (UTC)

I didn't know this existed. I can see the need for tightening up sourcing for images. If an image is consistent with cited content in the article, how much documentation should we require for the image? If an image is described as an "artist's concept", what would we want supporting the accuracy of the depiction? I am sure there are other questions we can ask. Donald Albury 21:37, 9 September 2024 (UTC)

Two examples of how much a creator can put in an image (and in these cases, maybe quite rightly - I'm not suggesting they include falsehoods):

File:Europe 180ad roman trade map.png maps which commodities were carried from where to where in the Roman Empire c. 180 AD. It also has a table of the Roman Empire's annual costs. It doesn't matter much whether the bends in the routes are correctly placed, but how is it verifiable that those are the main commodities; that they were moved from X to Y; and that those were the annual costs? It's used at Roman commerce but not supported by it.

File:Constitution of Rome.jpg has capsule descriptions of Roman institutions and elected officials, and diagrams some interrelationships. The illustrations are decorative; it's the text that matters and that would be subject to normal editing and refinement, in accord with WP:V, except that it's been turned into a jpg. NebY (talk) 22:49, 9 September 2024 (UTC)

If those two images are doing their job of illustrating the article, then their contents ought to already be in the text of the article (in some form, not word for word). WhatamIdoing (talk) 00:23, 10 September 2024 (UTC)

That's beautifully straightforward and I'll feed it back to the discussion that first set me to looking around for policy. Thank you. I fear it may be a rather high bar for our many historical maps, that one included. NebY (talk) 18:08, 17 September 2024 (UTC)

So... I'm a bit nervous about this. There are times when this is warranted (though an ordinary fact-tag on the caption should frequently have the same effect, so this isn't always necessary). Here are three stories:

Once upon a time, I ran across a POV pusher of the virus-denialist type on wiki. It seems that it's a bit more challenging to convince people of your belief that the virus doesn't exist, when there's an actual photo of the virus right there in the Wikipedia article. So he tried to get the photo removed. One of his ideas was to say that the photo wasn't a reliable source. The thing is, one enveloped virus looks much like another. They all basically look like blobs, and while it's a technical triumph to get the photo, if you made a blurry, low-resolution, black-and-white photo of a little blob of used chewing gum, it'd look about the same. But: the purpose of an image is to "illustrate", not to "prove", and that micrograph does a fine job of illustrating that it's a boring enveloped virus instead of a glorious structure like a bacteriophage. In that sense, it doesn't actually matter whether it's the purported virus. What actually matters is that it looks like the purported virus. This dispute is the basis for one of the examples in Wikipedia:Manual of Style/Images#Pertinence and encyclopedic nature.
Another day, there was a real-world dispute about whether a politically disfavored person's African farm was actually in use, or if the ruling party could declare it abandoned, legally seize it, and give it to one of their cronies. Someone (presumably connected somehow to the event) uploaded an image to Commons, showing a very large cow, with a building in the background (the farm house?). Editors decided to omit the image, because they thought that even if it was "true" (i.e., someone really did take a photo of a cow at that farm) it might be "false" in another way (e.g., maybe that cow was trucked over there for the purpose of taking the photo, and then trucked right back home again). If there hadn't been a dispute over the farm, they probably would have accepted it, and if memory serves, they did believe that the photo was taken of a live cow, actually standing in the purported location. (In this part of Africa, "there's a cow there, so it's a working farm" is culturally reasonable. In other places, they'd have shown some sheep, or someone driving a tractor.)
More recently, I sent a batch of images for deletion at Commons. We determined that they were unmarked AI-generated "portraits" of 19th-century politicians. (Commons might accept some AI-generated images, because supporting images for articles such as Artificial intelligence art is within their scope, but having them unmarked means they're potential hoaxes.) We deleted them.

The first story shows the danger of rejecting images just because someone wants citations, especially for low-risk things. We actually don't want people to demand that a photo allegedly of a cupcake, that looks like a cupcake, get certified as being a cupcake before we can put it in Cupcake. The second shows that even if you have citations, you may not want the photo, because it might imply things that aren't appropriate for the article to be implying. The third shows the problem of not having reliable sources.

I don't think there is a single answer that works in all cases. WhatamIdoing (talk) 23:00, 9 September 2024 (UTC)

If it's not the virus, but it looks like the virus (because it has the same type of capsid, say), then that is a fine kind of photograph to put in an article that has no other photos (because capsids are very relevant to a virus). But what kind of virus is it? Where does the photo come from? (And e.g. if there's no provenance and if we can't ask the uploader, how do we verify copyright?) If the virus is a different virus, the caption and photo description should say what virus it is! This is not a question of what content is or is not appopriate for an article, but a question of WP:V!
A Wikipedian's uploaded photo of a cow on a farm is supposed to be used to lend support to a political claim one way or another in an article? That's blatant WP:OR, regards the content of the article, and is not necessarily a question about the verifiability of the image.

Nobody said anything there being about a single answer (nobody anywhere ever honestly suggests there being a single answer, outside religion -- that statement felt anticonstructive). We just want people to apply WP:V to images. SamuelRiv (talk) 02:35, 10 September 2024 (UTC)

What exactly does it mean "to apply WP:V to images"? Some POV pushers will interpret that in the most maximalist way possible. Other editors will call you tendentious if you insist that a proper reliable source be produced for simple, obvious illustrations, like "This is a cupcake", even if they're familiar with things like Is It Cake?

Wikipedia talk:Verifiability/Archive 40#RfC: Do images need to be verifiable? may be interesting if you don't remember it from back in the day. WhatamIdoing (talk) 06:53, 10 September 2024 (UTC)

From what I remember of journalism class, a photograph for which you cannot assert the veracity -- that is, that what is inside it is what it is, in a genuine portrayal of the subject of the article -- becomes an "illustration". (That is, you can use it, but you call it an 'photoillustration' instead of a 'photograph' in your caption.) Typical classroom examples would be moving objects in a news scene foreground to better frame a shot, or in your example linked, a syringe of one thing in one place that's instead implied or claimed to represent another. SamuelRiv (talk) 20:16, 15 September 2024 (UTC)

That is reasonable, and on wiki, it's the standard practice for drawings of dinosaurs.

But sometimes you need a caption that says, e.g., "Intramuscular injections are injected at a 90° angle", in which case what matters is that it looks like it's approximately 90°, and not whether it's "real".

You don't want an outright dishonest caption ("This dinosaur was definitely this exact shade of green"), but you also don't want to add a bunch of irrelevant disclaimers ("The point is to illustrate the concept of a 90° angle, for people who don't remember much about geometry. We do not guarantee that this image is real, or that the person whose skin is shown is actually receiving an injection at the time, or that there's a needle on the hidden end, or that the liquid in the syringe is a medication instead of water, or that any of the clear liquid is being injected. No cute animals were harmed in the making of this illustration, unless it turns out to be AI-generated, in which case there's a possibility that the computing power contributed planet-warming emissions equivalent, possibly causing a tiny increase in heat-related stress for all living beings"). WhatamIdoing (talk) 20:53, 15 September 2024 (UTC)

("The point is to illustrate the concept of a 90° angle, for people who don't remember much about geometry. We do not guarantee that this image is real, or that the person whose skin is shown is actually receiving an injection at the time, or that there's a needle on the hidden end, or that the liquid in the syringe is a medication instead of water, or that any of the clear liquid is being injected. No cute animals were harmed in the making of this illustration, unless it turns out to be AI-generated, in which case there's a possibility that the computing power contributed planet-warming emissions equivalent, possibly causing a tiny increase in heat-related stress for all living beings").
Thanks for the laugh! Schazjmd (talk) 20:55, 15 September 2024 (UTC)

Apologies for such a slow response. Thank you for the detailed examples and the links. Clearly there's much to think about and lots of history here (so glad I didn't just boldly edit WP:V) and I don't have the relevant experience to really engage with it (or time RN). Mildly, I am surprised we seem to be stuck in a situation where editors can't come to WP:V to find policy or guidance on image verifiability, or even links to such. If hard cases do make bad law, is this a situation where the risk of hard cases is blocking the making of good guidelines? NebY (talk) 18:22, 17 September 2024 (UTC)

Aright, let's get this out of the abstract. Some notable failures of a variety of image types:

c:File:Battle_of_aine_jalut.png: I forget where the issue was first brought to my attention, but about the only thing this drawing has in common with the Battle of Ain Jalut is that we can confirm there were humans there. (Contacting the uploader/author gave me no response, which is typical for these types of images on Commons; without even info basic info like how it was drawn, or what source material was used to draw it, if the author is even the author as they claim, there's no way to check where another version would be to check against copyright.)
File Talk:Arabic_Dialects.svg: the discussion is extensive on every reason a map can fail, and why it was continued to be pushed despite knowing it was junk.
c:File_talk:Map_of_Archaic_Greece_(English).jpg is an example of a map that fortunately gives a verifiable source (kudos to the author), and thus the work and reasoning can be checked, but is factually inaccurate and misleading. It is still used on WP articles.
Turkish vocabulary pie chart (edit request): how could a pie chart be wrong, when it's just numbers, in print?

Academics have been known to uncritically republish Commons images, even in print books. For images without provenance given anywhere, this critically increases the risk of wp:citogenesis. There was also an incident recently (within 2 years), (which I can't find, reported on VP iirc), where an academic book reprinted a (somewhat inaccurate and inadequately sourced) map from Commons with zero attribution. These are not trivial issues. Images are essential to quality articles, and readers take images seriously. SamuelRiv (talk) 22:53, 15 September 2024 (UTC)

There are two types of images…

Images that are used to illustrate information that is stated in the article text.
Images that are used to present information in an article.

In the first situation, the important thing is that information in the article is reliably sourced. Consensus can determine whether the illustration accurately depicts that verifiable information, and is captioned appropriately.

In the second situation, the caption needs to include a citation to a reliable source, to establish that the information being presented in the image is verifiable. Blueboar (talk) 18:41, 17 September 2024 (UTC)

For example, if you have text in the article that says "14% of Turkish words are of foreign origin", then that pie chart is probably fine. But if you don't, then it's possible that you fix the image by changing the caption to say something like "According to the Turkish Language Association, 14% of words in modern Turkish are of foreign origin".

And then maybe you add another pie chart, saying something like "According to Global Language Experts Association, 14% of words in modern Turkish come from foreign languages, 27% come from Ancient Turkish, and the rest come from Proto Turkic" – all depending on the facts at hand, and what points need to be made in the article.

Also, thank you for mentioning the "battle" drawing; I've sent it to CSD as a probable copyvio. If you will go to c:Special:Preferences#mw-prefsection-gadgets-gadget-section-Maintenance and scroll down a little, you should find an item called "Google Images & Tineye", which can be useful for discovering that the supposedly "own work" images uploaded recently have been kicking around the internet for a decade. See c:Help:Image searching. WhatamIdoing (talk) 02:56, 18 September 2024 (UTC)

I think people are talking past each other if the word "caption" is being used to describe different things: there's the caption on the Commons page, which is confusingly a separate field from the image description field, but it nonetheless appears on the image preview when clicked from an article; then there's the caption text on the wp article page in the box below the downscaled image where it is displayed. (There is also the image alt text, but that's generally too brief to be problematic, and nobody is discussing it.) The latter in-article caption is relatively frequently edited and sourced somewhere to conform to the article text. The caption and image description on Commons (or else the WP File upload) is rarely if ever updated or checked over for accuracy, even if the main article text is significantly rewritten. (This also ignores that there are usually several articles across several wikis using any given halfway-decent illustration.)

In the pie chart example, you fix the image by changing the caption is absolutely wrong, for example even if the numbers are accurate, if the image was made using a different source from that in the article. So if the Commons caption doesn't match the article caption source (even if facts match), then that takes away from the credibility of the article (again, people see the Commons caption when they click to zoom in on the image, which is relatively common behavior). If the pie chart has extra information not in the source you cite in the article, then that's misattribution. (I get that a pie chart is easy for anyone to remake, but maps, scatterplots, and diagrams are a lot more difficult -- that's why you get situations like the dialects map, that people continued to use despite a consensus that it was misleading at best.)

For the "Battle", it's been deleted now, so I can't re-search, but of course a reverse image search was the very search thing I did however many years ago that was. Not sure where you found it now, if authorship was definitive, but anyway the first thing I did after searching was message the uploader and remove it from articles. SamuelRiv (talk) 00:09, 19 September 2024 (UTC)

Btw, maybe I should go over the main points you made in the 2010 RfC you linked:

The first is whether we're willing to trust image uploaders -- yes! The problem we're trying to convey is when uploaders don't give adequate provenance (what where and when) -- just say what the image you upload is when you upload it and I'll take your word for it! -- and when editors use unverifiable uploads uncritically. Your quote about published academic sources is a strawman -- nobody is asking the uploaders for RS to show that your virus example fits the description of that species. Just tell us when where and of what they took the photo. (Of course, for maps, graphs, diagrams (these are by some standards considered primarily conveyers of fact, rather than creative expression), and historical svg reproductions like flags (derivative works), you need to give a specific source to verify the actual information.) But just give the information -- browsing Commons and WP articles, many such uploads have bare descriptions.

Interesting you suggest using WP:IUP (and that policy be descriptive over prescriptive). First, IUP is great in principle -- IUP#RI has what I'm asking for in photos for required information, and on diagrams it says it is required to include verification of the source(s) of the original data when uploading. (Of course, it uses as an example the featured image File:Conventional_18-wheeler_truck_diagram.svg, which in no version gives the source of the original data -- literally any Michelin manual in any auto shop would do, but that's beside the point.) So we have a policy that is not being followed even in its own examples; it's ineffective in practice because it only allows enforcement on wikipedia uploads, while most images we use (including the aforementioned truck diagram) are uploaded to and called from Commons, which has no such policies and for many reasons has no intent on implementing such. It also is a policy for enforcement at the image pages, whereas the eyes for verifiability in practice happens at the articles. Checking the WP deletion log for the File namespace, I see nothing.

We really do care about what the images look like, not whether they're "real". If we know (or suspect is very likely with good reason and no claim to the contrary) that an image is of A, but we present it to the reader uncritically in an article about B as if it were B, then we are deceiving the reader. I don't know how else to put it, but in an encyclopedia meant to educate with verifiable factual information, this act is simply a lie. You may think this is some theoretical journalism or academia thing, but it's just what it is, and people get into serious trouble for knowingly doing exactly what you've suggested is ok. SamuelRiv (talk) 01:08, 19 September 2024 (UTC)

> Your quote about published academic sources is a strawman -- nobody is asking the uploaders for RS – except, you know, for the editors who actually said things like "To me it's perfectly obvious that the image has no reliable source...What is the reason I doubt this pictures is HCV? Its lack of a verifiable source, that's what" and "it doesn't satisfy WP:RELY at all. The issue isn't merely one of quality. We simply don't know what it is of. It's all over the Internet with no reliable source". Another(?) editor said, of the same image, "It was uploaded in November 2007 by user < PhD Dre >, but he or she did not indicate where the image came from. was it his or her own scientific research? was it published? if so, where is the citation for a scientific publication?" POV pushers, in particular, actually do ask for sources, and they sometimes specifically ask for academic sources.

> on diagrams it says "it is required to include verification of the source(s) of the original data when uploading" appears to be an undiscussed bold addition by @Masem. If (i.e., Since) it does not reflect the community's practices, it should be re-worded. For example, "It is required to include verification of the source(s) of the original data" (The passive voice was used / Responsibility was shirked) could become "Please include information about where you found the original data".

> we are deceiving the reader – I think it's more complicated than that. Consider the bird in this photo. It's either a black-capped or a Carolina chickadee. The two species are notoriously difficult to tell apart by sight. Although they are very common birds and were first described scientifically 200–250 years ago, researchers weren't even sure they were separate species until about 20 years ago. But if the point you are trying to illustrate in the article is "This chickadee sometimes eats small insects that it finds in trees", and this is obviously a photo of a chickadee that looks like the right kind (even though they're hard to tell apart visually) and looks like it is sitting in a tree (which could be a realistic-looking but plastic Christmas tree) and looks like it's looking for a tasty snack between the needles (but for all we know, it's a stuffed specimen), does it actually matter if DNA testing or capturing it and taking it to an expert for identification would prove that it's the right/wrong species? What the article needs is an illustration, not photographic evidence.

We're not saying "This is a certified photo, following strict Chain of custody rules for evidence that this common type of bird exists". We're just saying: Look, sometimes you'll find these looking for insects in a pine tree. WhatamIdoing (talk) 03:57, 20 September 2024 (UTC)

We need to be cautious here. What's done with images is often the same thing that we do with text.....summarize copyrighted material where the material itself can't practically be put in Wikipedia. IMO something that reserves action for the most problematic cases would be best. Where there is a specific challenge that the statement made in the caption or by image are false, misleading or baseless. North8000 (talk) 18:19, 17 September 2024 (UTC)

[1] Fictional News Place "Person accidentally buys yacht" 7 November 1987

[2] The Gallifreyan Post "Tardis stolen at 4 AM" 5 August 1980

[1]

[2]