Jump to content

Wikipedia talk:Verifiability/Archive 75

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 70←Archive 73Archive 74Archive 75Archive 76Archive 77→Archive 80

Add "of information" to ONUS

1. I added "of information" to the last sentence of ONUS, explaining "clarify that ONUS relates to information (and not, for example, grammar)." The result was:

The onus to achieve consensus for inclusion of information is on those seeking to include disputed content.

2. Another editor revered with an explanation of "compact."

3. A third editor restored my change, saying "The original wording "inclusion of information" is better. Changing to "its inclusion" causes ambiguity since there is more than one nearby noun."

4. There were some intervening edits - none of which spoke to the substance of the edit summaries at 1 and 3 above - that ended up back at my change (1. above).

5. The reverting editor (2. above) reverted again, saying "Clearly the change to longstanding status quo has destabilized this little sentence. Let's just proceed on talk. I really doubt there was a problem with the longstanding. If there had been, it surely would have come up in the monumental go-rounds on the talk page." This, too, is not a substantive objection to the change. See WP:DRNC.

Is there any substantive objection to adding "of information" for the ''reasons'' given at 1 and 3 above? - Butwhatdoiknow (talk) 19:32, 21 July 2022 (UTC)

We as editors are naturally concerned with exact phrasing, including tone, punctuation, and grammar. Moreover, we often find the compromise that reaches consensus in precise wordsmithing, and sentence and paragraph construction, and should not degrade or disregard that enterprise. We are not just in the work of blurting out information, the work is as much about providing information in a precise, certain, and concrete way. Alanscottwalker (talk) 20:05, 21 July 2022 (UTC)
Better to focus on "content" not "information" Alanscottwalker has brilliantly summarized the operational importance of that distinction. SPECIFICO talk 20:24, 21 July 2022 (UTC)

Move "disputed content" to earlier in the last sentence of ONUS

So you'd both be okay with a switch to: "The onus to achieve consensus for inclusion of disputed content is on those seeking to include it"? - Butwhatdoiknow (talk) 21:03, 21 July 2022 (UTC)
Not sure about that… ”Content” could refer to an entire article, and that isn’t what ONUS is discussing. ONUS is focused on disputes over smaller bits of verifiable but problematic material… and whether that material should be included in or omitted from a specific article. Blueboar (talk) 21:12, 21 July 2022 (UTC)
That is a topic for another day (and one I'd love to discuss). But, for now, I'm just asking about rearranging the sentence in a way that Alanscottwalker and SPECIFICO may find acceptable. - Butwhatdoiknow (talk) 23:49, 21 July 2022 (UTC)
I really think this is not the part of the section that needed clarification. The problem is that this simple section is just trying to quash unsubstantiated claims of WEIGHT based solely and merely on the ground that the content is Verified. And that is a problem, because we have long had editors go off into unrelated tangents about consensus and longstanding and status quo and other unrelated issues. I don't think we should mess with the part that's more or less OK lest we introduce unanticipated problems like the NOCON tangent and other unforseen irrelevancies. SPECIFICO talk 21:24, 21 July 2022 (UTC)
Perhaps the problem is that the section tries to cover two distinct issues in one paragraph. First there is NOTGUARANTEED … which is merely a cautionary statement to tell editors that there is more to inclusion/omission than JUST Verifiability.
Then there is ONUS, which talks about consensus when there is a dispute over inclusion/omission.
Perhaps they should be separated (Even if only by a paragraph break)? Blueboar (talk) 21:55, 21 July 2022 (UTC)
Yeah, but a religion has arisen around this confusion and it's now ~1000 AD into that mess so we can't go back and disrupt the origin story. Folks would just continue the same argument even after the whole section is erased. I think the second part was just anticipating the obstinate refusal of half the editors' tendency to ignore the fist part. SPECIFICO talk 23:04, 21 July 2022 (UTC)
Right that's a problem. But, for now, I'd appreciate your thoughts on my proposed compromise of moving "disputed content" to earlier in the sentence. - Butwhatdoiknow (talk) 23:49, 21 July 2022 (UTC)
I favor no change. Let sleeping dogs lie. SPECIFICO talk 23:58, 21 July 2022 (UTC)
Why do you favor no change? What objection - other than it isn't the status quo - do you have to "The onus to achieve consensus for inclusion of disputed content is on those seeking to include it"? - Butwhatdoiknow (talk) 00:18, 22 July 2022 (UTC)
It's the dogs. SPECIFICO talk 00:30, 22 July 2022 (UTC)
Okay, I'll put you down as "has no non-frivolous objection." - Butwhatdoiknow (talk) 05:34, 22 July 2022 (UTC)
Or, you could just stop asking me the same question over and over and over and note the reason I'd already given at least 3 times above. I don't see a line of editors at the door waiting to tell you how much they like your proposal. SPECIFICO talk 10:34, 22 July 2022 (UTC)
Edits and objections to edits must be supported by substantive "reasons." All I've heard you say over and over is that you oppose my proposed edit because it is a change. That is not a valid reason no matter how many times you repeat it. - Butwhatdoiknow (talk) 15:48, 22 July 2022 (UTC)
@User:Alanscottwalker, as a compromise I propose above changing the ONUS sentence to "The onus to achieve consensus for inclusion of disputed content is on those seeking to include it" (Moving "disputed content" to earlier in the sentence.) Would that be acceptable to you? - Butwhatdoiknow (talk) 05:34, 22 July 2022 (UTC)
Why change it at all, did someone complain they couldn't understand it? Selfstudier (talk) 07:38, 22 July 2022 (UTC)
I am complaining that the subject of the sentence ("disputed content") doesn't appear until the end of the sentence. This is a very minor change that improves readability. - Butwhatdoiknow (talk) 15:48, 22 July 2022 (UTC)
It was OK before but anyway, looks like we are going to have another go at fixing the whole thing as opposed to tweaking. A paragraph with the word information in it 4 times (and nearly 5 times) has gotta be wrong, lol. Selfstudier (talk) 15:56, 22 July 2022 (UTC)
"It was okay before" is not the same as "your proposal doesn't improve it because ..." And I have little doubt that a discussion to fix the whole thing will ultimately fail. I just want to make this small change to make a small improvement. Do you have any reason to believe that it won't accomplish that goal? - Butwhatdoiknow (talk) 01:24, 23 July 2022 (UTC)
I replied to this earlier but for some unknown reason @Fram: decided to delete it. Consult the history if you want to see what I wrote. Selfstudier (talk) 09:48, 23 July 2022 (UTC)
Oh, Go Fish, what a fun game. I found the revert here. User talk:Fram was clearly out of line. I've restored your post below. - Butwhatdoiknow (talk) 15:44, 23 July 2022 (UTC)
Trouble with the mobile, it was not intentional. Selfstudier (talk) 16:52, 23 July 2022 (UTC)
Yeah, but once they knew they had made an error they should have fixed the damage. (I look forward to your responses to my replies to your restored post (see below).) - Butwhatdoiknow (talk) 18:44, 23 July 2022 (UTC)
"Not an improvement" is a standard reason for objecting/reverting. As for failing, I see that it already has so no more tweaking either, seems an RFC will be necessary to make any change. Selfstudier (talk) 09:20, 23 July 2022 (UTC)
@Selfstudier: I look forward to your replies to my July 23 posts (below). - Butwhatdoiknow (talk) 15:21, 25 July 2022 (UTC)
Ah, but it is an improvement. I checked both versions at a readability calculator. The current version has a score of 44.97. Moving "disputed content" to earlier in the sentence raises that to 50.24. - Butwhatdoiknow (talk) 16:05, 23 July 2022 (UTC)
The proposed change just reorganizes the sentence - it does not change the meaning in any way. I fail to see how an RFC that hasn't started prevents us from making this small readability improvement in the interim. - Butwhatdoiknow (talk) 16:05, 23 July 2022 (UTC)
If it makes no substantial difference and there are currently bigger fish to fry, might as well just leave it be, no? Selfstudier (talk) 15:23, 25 July 2022 (UTC)
No, because (1) it makes a significant difference (increasing readability by more than 10%) and (2) frying the small fish does not prevent us from simultaneously frying the big fish. And isn't that what Wikipedia is all about - making improvements when and where you can? Compare Perfect is the enemy of good. - Butwhatdoiknow (talk) 22:07, 25 July 2022 (UTC)
I don't know what you want me to say, it makes no difference to me in either version, I go by comprehension not "readability" according to some criteria Idk or care about. If that's such a great thing, then WP can run it on the entire encyclopedia and we need not worry about it for every little edit. I haven't reverted your edit anyway, ask other editors to agree with you, I'm sticking with fish frying ftb. Selfstudier (talk) 09:26, 26 July 2022 (UTC)
What I hear you saying now is "while I still don't see the point of making this minor change, I do not oppose it." That is good enough for me. - Butwhatdoiknow (talk) 16:30, 26 July 2022 (UTC)
@Butwhatdoiknow: I'm not sure those numbers are reporting what you think. The lower the overall number, the more readable the prose is. Each individual score spit out by each formula represents the "number of years of education" required by an individual to easily understand it. So it would seem that you would want lower numbers, not higher, to make your case. --GoneIn60 (talk) 18:13, 26 July 2022 (UTC)
Well, you made me look. The score for the current sentence is 44.97. For my proposal it's 50.24. That can't be a reference to years of education. So I tried "the cat is red" and got a score of 118.18. Then I tried "the cat is red and likes peanut butter and jelly" and got a score of 78.25. Conclusion: the numbers are reporting what I think and moving "contested matter" to earlier in the sentence would improve readability. - Butwhatdoiknow (talk) 19:20, 26 July 2022 (UTC)
Could you try "The cat is peanut and Red likes jelly" ? How does that score? SPECIFICO talk 19:59, 26 July 2022 (UTC)
Yep, I didn't dig deep enough. At first glance, I thought the overall "Flesch Reading Ease" score was a simple sum of the individual index scores (where lower scores matter, because they represent education years), but a closer look reveals that's not the case. Disregard! --GoneIn60 (talk) 20:12, 26 July 2022 (UTC)
"Red and cat jelly peanut is the like" scores 94.37. ""Jump stain log mole" scores 120. Artificial, no intelligence. SPECIFICO talk 22:15, 26 July 2022 (UTC)
The website tests readability, not coherence. Are you suggesting that "The onus to achieve consensus for inclusion of disputed content is on those seeking to include it" is nonsense? If not, what is your point? - Butwhatdoiknow (talk) 23:12, 26 July 2022 (UTC)
It was an experiment. I concluded that the index is nonsense, or at least not helpful to our efforts here. SPECIFICO talk 21:48, 29 July 2022 (UTC)
We have an article about that: Flesch–Kincaid readability tests#Flesch reading ease
@SPECIFICO, your earlier nonsense sentence produces these stats:
  • Gunning Fog index: 3.20 (years of formal education)
  • Coleman Liau index: 2.54 (US grade level)
  • Flesch Kincaid Grade level: 3.76 (US grade level)
  • ARI (Automated Readability Index): 0.23
  • SMOG: 3.00
  • Flesch Reading Ease: 82.39
Note that some tests require a minimum of 100 words, and others produce consistent results only when given very long passages. None check for grammar, sense, or background knowledge (e.g., whether adults know what a Security deposit is, even though those are "difficult" words).
For comparison, Colorless green ideas sleep furiously produces these numbers: GFI 18, CL 17, FK grade12, ARI 12, SMOG 11, and Flesch Reading Ease 15.64.
@Butwhatdoiknow, the difference of FRE 45 vs 50 is not really significant. Also, you might be interested in https://hemingwayapp.com/ as it does more than just look at readability numbers (e.g., passive voice and overuse of adverbs). WhatamIdoing (talk) 20:39, 29 July 2022 (UTC)
Two questions: (1) Why is 10% "not really significant"? (2) Even if the proposed change would only be a 1% improvement, why not make it if it is an improvement? - Butwhatdoiknow (talk) 01:30, 30 July 2022 (UTC)
Because it's not really 10%. A 10% change in the FRE number does not represent a 10% change in the actual effort needed to understand the text. All that's happened here is that you shortened the sentence by two words ("better" FRE), while slightly increasing the average number of detectable syllables per word ("worse" FRE). Those are the only two things that FRE cares about. You could alphabetize the words in the sentence and get the same FRE score. FRE believes that "I should go" and "I ought to leave" are almost identical in reading difficulty, even though the first has simpler, high-frequency words that a beginning reader can manage, and the second doesn't (unusual sound for ou, silent gh, additional vowel digraph in ea, and a silent e).
If you are interested in using readability scores, then I recommend that you start with https://www.ahrq.gov/talkingquality/resources/writing/tip6.html Please note that, within the literature on readability scores, this is what's considered a supportive viewpoint. If you want to see a more typical viewpoint, then https://www.uxmatters.com/mt/archives/2019/07/readability-formulas-7-reasons-to-avoid-them-and-what-to-do-instead.php has a good, reality-based summary. WhatamIdoing (talk) 01:58, 30 July 2022 (UTC)
My proposal moves the subject of the sentence earlier in the text. What test do you recommend we use to determine whether that is an improvement or not? - Butwhatdoiknow (talk) 03:16, 30 July 2022 (UTC)
I recommend a form of Usability testing, namely that you ask editors on this page which option, if either, is clearer to them. It appears that you have done this and that nobody thinks that it is a material improvement. WhatamIdoing (talk) 16:04, 1 August 2022 (UTC)
Well, I think would be a material improvement. So that's one. And no one is saying it would make the sentence worse. So, is there any reason other than status quo bias to not make the change? - Butwhatdoiknow (talk) 16:47, 1 August 2022 (UTC) @WhatamIdoing: I look forward to your reply. - Butwhatdoiknow (talk) 16:34, 3 August 2022 (UTC)
Status quo bias is frequently considered a sufficient reason to avoid non-substantive changes to the wording of a policy. Some part of the community seems to be happy when the wording of policies is not messed with merely for minor gains in clarity.
Another reason you should consider is whether this is "the hill you want to die on". We collectively believe that the benefit is likely minimal but positive (or at least not negative), but continuing to push for the change could damage your relationship with other editors. Is this minor change worth the risk that your future proposals could be met with a certain amount of "Oh, him again – he can't take a hint, he can't read the room, his proposals are always so painful"?
A third reason is that this change is a bit of "rearranging the deck chairs": Why bother copyediting it now, when North's considering an RFC that might result in it being removed completely?
As usual, note that whenever I give you a list of reasons, it is not a complete or exhaustive list of reasons (unless so labeled). WhatamIdoing (talk) 19:13, 4 August 2022 (UTC)
No need for a change. Moxy- 22:43, 26 July 2022 (UTC)
Do you see any harm arising from the change? - Butwhatdoiknow (talk) 23:12, 26 July 2022 (UTC)

Rewrite entire ONUS section discussion

If you look at the context of the paragraph that onus is in, it looks like a first attempt of putting what should have actually been in there which is "Verifiability is a requirement for inclusion, not a reason for inclusion." And then wikilawyers mis-applied it elsewhere. And the fuzzy Wikipedian system did it's best to incorporate it, including it's otherwise-conflicts with wp:consensus), in the way that I described it previously. We should just replace onus with that and solve this whole big complicated mess (plus other problems) with one fell swoop. North8000 (talk) 12:41, 22 July 2022 (UTC)

BTW, "Verifiability is a requirement for inclusion, not a reason for inclusion." that is not a change, it is a summary of of current policy and reality. North8000 (talk) 13:51, 22 July 2022 (UTC)
Deal.Selfstudier (talk) 14:10, 22 July 2022 (UTC)
  • I would favor a complete rewrite. The current version tends to derail disputes into arguments over process rather than attempts to reach consensus. --Aquillion (talk) 18:59, 22 July 2022 (UTC)

OK, if it must be tweaked, how about

While all article content must be verifiable, not all verifiable information should be included. Content that is WP:UNDUE does not improve an article. Such information should be omitted or presented instead in a different article.
The onus to achieve consensus for inclusion is on those who seek to include disputed content.

SPECIFICO talk 14:30, 22 July 2022 (UTC)

The second and third sentences are implied or follow from the first. So it's just that last one, which I'd say is pretty much how consensus works anyway. Selfstudier (talk) 14:36, 22 July 2022 (UTC)
Yes the last sentence really could be omitted. That would also get the word "consensus" out of the way. Everything's done by consensus so nothing new there. That would also address North's objection below. This section is really just a pointer from V to NPOV, saying content needs both, and not just V. SPECIFICO talk 16:16, 22 July 2022 (UTC)
Shall I strike the ONUS sentence? That would be funny. SPECIFICO talk 16:17, 22 July 2022 (UTC)
That makes the main problems (confusion and wikilawyering mis-use) even worse, including now entangling 3 policies instead of two. North8000 (talk) 14:59, 22 July 2022 (UTC)

Better version

Per the discussion below, the last sentence adds no meaning beyond sitewide editing policy and protocols. Hence, new proposal:

While all article content must be verifiable, not all verifiable information should be included. Content that is WP:UNDUE does not improve an article. Such information should be omitted or presented instead in a different article.

SPECIFICO talk 19:39, 22 July 2022 (UTC)

No, our policies and guidelines do not come down to just DUE. And of course, it adds meaning, we need to construct pieces of work that satisfy the WP:Content policies and in accord with WP:Guidelines to reach a solution "that incorporate[s] all editors' legitimate concerns, while respecting Wikipedia's policies and guidelines." (WP:CONSENSUS). The proponent of text edits and images necessarily must do that, to the satisfaction of editorial process. Alanscottwalker (talk) 20:05, 22 July 2022 (UTC)
I'm not seeing anything in that wording that deprecates any of our other content policies and guidelines. But as others have said, since editing decisions (with BLP COPYVIO and a few other exceptions) are all by consensus, there's really nothing added by referring to consensus or behavior on a page that's about content and verification. In fact, this seems to have led to lots of confusion, the last sentence about behavior and consensus in this location. SPECIFICO talk 20:17, 22 July 2022 (UTC)
Thanks, I meant to add, at least 3 of the Legal Policies (the ones dealing with article text and images) need to be satisfied. But I see, you are deprecating the other policies because a true consensus respects all of them, not just DUE. Alanscottwalker (talk) 20:29, 22 July 2022 (UTC)
I'm more or less back at my original feeling that we should let sleeping dogs lie. These policy pages are basically codifying existing practice -- like a compilation of case law or common law principles. It's not as if what we change here is going to change the course of editing history. It would at best just create a more helpful page to which we could refer when the inevitable edit wars and arguments erupt over cherrypicked UNDUE sourced content. SPECIFICO talk 20:34, 22 July 2022 (UTC)
But part of my problem is that at least one reasonably-common interpretation of ONUS does not reflect actual practice. If you look above you'll see some people saying that a single sentence in ONUS effectively negates all of WP:IMPLICITCONSENSUS, WP:EDITCONSENSUS, and large parts of WP:CONSENSUS as a whole by making it so only explicit consensus reached through discussion counts when there's a dispute or a WP:NOCON situation - meaning that in any circumstance where there's no clear previous discussion to point to, they think we should default to removal. That is clearly not and has never been anything remotely close to actual practice. RFCs with no consensus generally keep the last stable version; it is possible to get text removed in that situation if you can show that it never had consensus, if it violates BLP, or if there's some other issue with it, but unless text was recently added the onus there is generally on the people arguing that it never had consensus to convince the closer that that is the case, not on the people who want to retain it. I think our current practice works well but that that interpretation of WP:ONUS frequently derails discussions into debates over process. I also think it's a problem in the sense that editors who have no valid objection to a particular piece of text will sometimes just cite ONUS, which I feel is inappropriate. Once text passes WP:V objections must be specific; they don't necessarily need to be good, that's something to work out via discussions and RFCs if necessary, but citing purely procedural reasons to remove something, with no underlying objection, is often a sign of stonewalling because it effectively just derails the consensus-building process without providing an actual way forwards. (Also, I wonder if this might be worth a general essay - "don't derail content discussions into debates solely over process." Consensus-building requires stating objections and hashing out arguments over the text and sources, not using red tape to argue who wins by default.)--Aquillion (talk) 09:48, 23 July 2022 (UTC)
Only reached through discussion? Very little is originally added to the pedia through discussion, that does not relieve the author, the proponent, of adding decent content (improved content). What's decent? What's decent is content that does not raise legitimate concerns. Content, is decent when it "incorporate[s] [the] legitimate concerns, while respecting Wikipedia's policies and guidelines." (WP:CONSENSUS) And, of course, it is sometimes work to tease out the legitimate concern, that is why we have discussion, discussion is not just a process, it the only hope of solution, and when a legitimate concern never materializes, it does not defeat WP:CONSENSUS. -- Alanscottwalker (talk) 13:17, 23 July 2022 (UTC)
I agree that it's unworkable to only rely on consensus that is reached through discussion; but see the conversations above, where a significant number of people argued that ONUS requires exactly that. That's a core part of the dispute we're trying to resolve here. Likewise, my point is that ONUS, by its nature, discourages discussion because people are citing it to argue that they do not need to participate in discussions to produce and workshop a legitimate concern - they believe they can simply say "I object, and I have no further responsibilities to say anything else; the ONUS is on you to do everything from here." --Aquillion (talk) 22:53, 23 July 2022 (UTC)
@Aquillion, IMPLICITCONSENSUS and EDITCONSENSUS are the same thing. And they evaporate into the mists as soon as someone challenges the results. The second sentence of that section ends with the words "until it is disputed or reverted."
So you start off with IMPLICITCONSENSUS, and then "it is disputed or reverted", at which point, IMPLICITCONSENSUS does not exist. We have never had a rule that says "But nobody else objected for a long time, so your objection is invalid" rule. We have always had a "As soon as someone objects, you have to find consensus" rule. WhatamIdoing (talk) 22:50, 24 July 2022 (UTC)
I also don't think that ONUS requires consensus to be reached through discussion. (How could it? It doesn't even mention discussion.) ONUS says who has to do the work, but it doesn't prescribe the method by which the work is done. WhatamIdoing (talk) 22:54, 24 July 2022 (UTC)
  • Something to remember: Consensus is not permanent… because “consensus can change”. Something may have had consensus when it was added … but that does NOT mean it has consensus NOW. This is true whether that consensus was implicit or determined through discussion. Any time someone questions content, the previous consensus has to be reassessed, and is either reaffirmed or overturned. Consensus is not an end result, but an ongoing (never ending) process. Blueboar (talk) 23:11, 24 July 2022 (UTC)
The trick is to know, which way the wind is blowing. GoodDay (talk) 23:44, 24 July 2022 (UTC)

The history of WP:ONUS

Ok… it may help to look back and explore how (and when) the current language developed:

  • the phrasing “Verifiability is not a guarantee of inclusion” was first added to the policy on 28 Oct. 2011. As part of implementing an RFC. It was originally part of the lead paragraph of the policy.
  • In February of 2013, it was moved to its own sub-section, and expanded. At that point it just said that inclusion/omission was determined by consensus (no onus).
  • The word “onus” was added on 13 August 2014… with a shortcut added soon after that.

I would suggest that everyone look through the archives at the talk page discussions around these dates - to better understand how we got to the language we have now. Then we can better decide if we need a change and, if so, what language to use going forward. Blueboar (talk) 15:11, 22 July 2022 (UTC)

You have already looked/were responsible for some of it, so is North's 1-liner OK from where you're sitting? Selfstudier (talk) 15:58, 22 July 2022 (UTC)
I don't actually see any discussion in the archives about the "onus" line at the time it was added by JzG. I do see HighInBC's edit summary when it was agreed to: "I suppose any change challenged needs consensus."[1] Kolya Butternut (talk) 17:26, 22 July 2022 (UTC)
I tend to approach content disputes with the view that the majority decides on what's added or deleted from a page. I'm aware of WP:VOTE, but it's extremely rare, that the minority side wins. GoodDay (talk) 17:29, 22 July 2022 (UTC)
@GoodDay, what should happen in the rare circumstance that the vote is a true tie (by whatever method you choose to use)? WhatamIdoing (talk) 19:18, 22 July 2022 (UTC)
If it's an RFC? The closer decides. But if it's a local consensus attempt? A tie would normally come out as - 'keep it there (if it was already there) or leave it out (if it wasn't there). A tie says maintain the status quo. GoodDay (talk) 01:00, 23 July 2022 (UTC)
@GoodDay, people who write closing statements aren't allowed to pick the outcome when it's a true tie. Their job is to say "no consensus", not to record a WP:Supervote so that there will be a decision made. WhatamIdoing (talk) 22:40, 24 July 2022 (UTC)
Also: What if there isn't a "status quo version"? For example, imagine a dispute over an article that was just a day old at the time the dispute arose. WhatamIdoing (talk) 22:43, 24 July 2022 (UTC)
There are times when editors have challenged RFC closings. GoodDay (talk) 22:47, 24 July 2022 (UTC)
Wow.....so one word in a core policy which has steered tens of millions of words of disputes and wikilawyering was added with no discussion !?!?!?!?! North8000 (talk) 18:13, 22 July 2022 (UTC)
Wow is right. Typical.... Huggums537 (talk) 01:32, 1 September 2022 (UTC)
The key takeaway here, at least for me, is that there has never been a consensus supporting the final sentence of WP:ONUS, at least under the sweeping interpretations we sometimes see - it appears to have slipped in under the assumption that it was a mere clarification of existing policies (ie. nothing would change if it was removed) and then repeatedly cited out of context until it became treated as a significant part of our dispute-resolution process in its own right. But it is, itself, something whose inclusion never had a consensus behind it. --Aquillion (talk) 19:02, 22 July 2022 (UTC)
Unfortunately it's been there so long that I suspect any attempt to remove or clarify it will be continue to be contentious, at least going by the above "meaning of ONUS" subsections. Sideswipe9th (talk) 19:23, 22 July 2022 (UTC)
I would say, there has always been a consensus that an author of our articles must edit in accord with a true consensus, meaning the work respects our policies and guidelines in the judgement of themselves and, as importantly, other editors. -- Alanscottwalker (talk) 20:51, 22 July 2022 (UTC)
Ultimately we want everything edit to have a consensus backing it; but we also need WP:NOCON for what to do when consensus-building mechanisms break down. Part of the problem (as you can see in the discussions above) is that editors don't always agree on what constitutes consensus, or what constitutes a proposed change from an existing consensus reached via editing vs. a challenge to something that never had consensus. But another part of the problem is that stridently-worded dispute-resolution policies can sometimes make it harder to reach a consensus because they encourage discussions over content to derail into disputes over process rather than substance (ie. "rv, get consensus per WP:ONUS"; or even editors simply refusing to engage in discussions at all beyond demanding that someone else produce a consensus, which makes consensus-building hard.) There's no perfect solution to this, but at the very least ONUS lacks the depth and nuance that WP:CONSENSUS has.--Aquillion (talk) 09:34, 23 July 2022 (UTC)
When consensus finding breaks down you move to dispute resolution, those who don't participate in an RfC, for example, obviously get no say. But there is a more pervasive problem than citing ONUS, it's, despite the double stringent warnings in V, the, "I gotta source, so no matter what this piece of content goes in."
The thing is, the onus, or responsibility, does not first arise when there is a dispute, it arose at the time an author's first article addition, including at article creation. The principle for adding content is that it improves the pedia, we judge what improves the pedia by whether it accords with editing policies and guidelines: an irrelevancy does not improve the pedia, a mangled context does not improve the pedia, a misplaced context does not improve the pedia, a lack of context does not improve the pedia, an overstated or misunderstood source does not improve the pedia, a trivia does not improve the pedia, a POV does not improve the pedia, a CVIO does not improve the pedia, an original research does not improve the pedia, etc. The onus, the responsibility, never moves, it is always on the author. No author is perfect or can be perfect but they do have to work to improve the pedia, and show it to the world, in their work.-- Alanscottwalker (talk) 12:21, 23 July 2022 (UTC)
it arose at the time an author's first article addition - no, this is flatly incorrect (or at least misleading.) Per WP:EDITCONSENSUS most edits are presumed to enjoy consensus until someone objects; as a result, there is no requirement to demonstrate consensus before someone objects - it is presumed. This is why "rv, get consensus" with no other objections is not a valid thing to do. Beyond that, the problem is that ONUS isn't just being cited as an abstract principle but as if it were a defining part of dispute resolution, so we need more than those general statements - we need to discuss how it is going to work in practice. Let's say a paragraph was added years ago, has been heavily edited since then by countless editors with no objections; someone then sweeps in and deletes the whole thing without explanation and reverts any attempt to restore it. There's discussion about it on talk (the first time the paragraph was ever actually discussed) and there's no clear consensus. There's then an RFC, which is also closed with no consensus. Do you believe 1. this RFC outcome will result in the paragraph being deleted, and 2. that this RFC outcome should result in the paragraph being deleted? Current practice is unambiguously no on both accounts; through editing and sufficient people reading it without objecting, text eventually achieves sufficient weight that removal becomes a significant edit that requires consensus, even in the absence of prior discussions. --Aquillion (talk) 22:42, 23 July 2022 (UTC)
You just proved that the consensus forms at first edit as a matter of presumption, thus, "it arose at the time an author's first article addition" is completely correct. The existence of article content is an implied consensus, meaning an implied assertion that the material satisfies legitimate editorial concerns, while respecting Wikipedia's policies and guidelines (the meaning of consensus), and not just to other editors, to the world. You go on to say, that as time goes by, as editing continues, that implication becomes stronger and stronger. And at some point, when and if the presumption is tested, it is either sustained or not. -- Alanscottwalker (talk) 08:54, 24 July 2022 (UTC)
This is an extremely widespread problem in the PAGs. (I think we have a bizarrely upside-down essay on the subject, although I can't seem to find it; I thought it was called "SETINSTONE" but apparently not. The upshot as I recall was "just because these PAGs were created by a bunch of randos editing randomly doesn't mean you don't have to treat them like the word of God, you peasant". Maybe that was just a bad dream I had -- I hope so.)
The problem is partly due to a historical shift in how policies and guidelines have been treated -- from "just another part of the wiki that should be edited boldly and ignored freely" to the current, uh, situation. But most of that shift had taken place by 2014, so I'm guessing ONUS has more to do with the other reason -- that flaws in Wikipedia' implementation of consensus make it easy for determined editors to push the project toward their desired policy positions by sneaking their preferred positions into policy and then declaring that, since policy represents consensus, there is therefore a consensus for their position and the burden is on anyone who thinks otherwise to demonstrate consensus against it. Like edit consensus, but for wikilawyering. (On a side note, the original creator of WP:N was a true master of this craft, and although I worked closely enough with them to be confident that they truly believed they were acting in the project's best interests, I would have to say that history has not been kind to this approach.) I suspect that closer analysis of whatever else was happening on-wiki around the time that change was made, and especially where "onus" was first used in a content dispute, might be revelatory.
Given that the vast majority of provisions of our sprawling PAGs have never had either positive consensus or the backing of valid fiat authority, I think the project would be much healthier if we removed everything but the nutshell and lede from every policy, especially this one, and gave "Guidelines" the more valid name of "Best practices". Maybe someone should just do that, and then declare that those changes now represent consensus. Sauce for the goose...-- Visviva (talk) 23:47, 22 July 2022 (UTC)
Re the essay: there's one named WP:STONE but it seems more pro-peasant. Peter Gulutzan (talk) 14:54, 23 July 2022 (UTC)
I think a broad discussion on this, at the village pump, to determine a policy on how to handle these bold additions, would be more productive than having a debate every time an individual boldly added PAG becomes contentious. BilledMammal (talk) 03:46, 23 July 2022 (UTC)
I doubt that such a discussion would be productive, and I'm convinced that it is a bad idea to suggest that "long standing undiscussed" and "informally discussed" advice is somehow less-than, non-binding, or otherwise worse than "formally discussed" changes. Very little of our policies and guidelines was written on the basis of RFCs. I doubt, for example, that you will find any RFCs officially voting that the core content policies should be called policies instead of guidelines.
I suggest that a more functional way to understand the changes is that advice that worked back in the day does not always work today. In other words, the prior/current advice doesn't have to be "wrong" or "false" or "added without following the proper bureaucratic procedures" for it to be "not what we need today". The practical reason to change a policy or a guideline is that we don't want it to say that any longer, and not, e.g., that MOS:ISLAM didn't follow the WP:PROPOSAL process – which it didn't, because MOS:ISLAM was written several years before I wrote PROPOSAL. WhatamIdoing (talk) 05:59, 25 July 2022 (UTC)
I agree that these guidelines and policies are as binding as those formally discussed, but I disagree that such a discussion won't be productive, as coming to a consensus that we can point at to settle such debates in the future will be very useful. In addition, it may give us an idea of when they become binding. BilledMammal (talk) 09:12, 25 July 2022 (UTC)
But you can't settle debates for the future, because Wikipedia:Consensus can change. We can agree now that the current consensus supports (or doesn't) something, but we cannot agree that it's binding in the future. If you point to "Well, back in the day, we had an RFC and everything, so that's consensus", then they can point to WP:CCC and say "that was then, and this is now – where's the evidence of this alleged consensus now?" WhatamIdoing (talk) 21:28, 25 July 2022 (UTC)
No, but you can require editors to go to a certain level for consensus to change. If we agree now that the current consensus supports something, then we can add it to a suitable policy (probably WP:CONSENSUS) and editors who believe consensus can change will need to propose that the line is removed from CONSENSUS; until they do that, debating it at lower levels will be pointless. BilledMammal (talk) 13:22, 27 July 2022 (UTC)
So the problem you present with MOS:ISLAM is that we agreed back then that the current consensus (back then) supported something, and we added that to a suitable guideline, namely MOS:ISLAM. And now that editors (now) believe that consensus has changed, you're complaining here that they have to go through some sort of bureaucratic process to remove those lines from MOS:ISLAM, because debating it at lower levels is pointless (even though those "lower level" debates are how we usually find out that the guideline no longer represents the current consensus....). But you recommend that exact process for changes to WP:Consensus. Do I have that right? WhatamIdoing (talk) 20:44, 29 July 2022 (UTC)
  • While there are many reasonable interpretations of ONUS, I think one thing is for certain as a result of discovery: Long-standing content cannot be removed from an article solely due to "no consensus for inclusion". Because to do so would be hypocritical as it rejects the very foundation by which ONUS even exists as a policy. -- King of ♥ ♦ ♣ ♠ 06:11, 23 July 2022 (UTC)
    • Can we at least agree that "There IS a consensus for exclusion" is a valid rational for removing material? That is what the VNOT sentences are about. Blueboar (talk) 21:04, 29 July 2022 (UTC)
I feel that we need a better definition of "long-standing", which is one of the points of contention above. Text that has been heavily-edited on a high-traffic article, with nobody objecting to it over a long period of time, has a stronger consensus behind it than text that has sat on some stub which has only had one editor ever, even if it sat there for ten years. That is what really has to be hashed out. --Aquillion (talk) 09:34, 23 July 2022 (UTC)
Agreed. There are different levels of implicit consensus for sure, but it seems there may be some disagreement on how to define them and whether we should. Even in high-traffic articles, there's no guarantee the content was ever seen, let alone evaluated or modified, unless you take the time to dig through diffs. Time and traffic alone cannot be the only factors to consider in support of implicit consensus, especially in a lengthier article. ONUS is a special case, of course, since it has been used in countless discussions both here and elsewhere since its inclusion. --GoneIn60 (talk) 15:53, 23 July 2022 (UTC)
More often, the problem arises in a thinly edited article where some gossip or thinly sourced news flash was added to the article -- maybe even disparaging BLP content -- and then it is called "longstanding consensus" as grounds for edit warring it back into the article when it's discovered. SPECIFICO talk 16:52, 23 July 2022 (UTC)
IMPLICITCONSENSUS does not say that anything "has a stronger consensus behind it", no matter how heavily edited or high-traffic the article is. IMPLICITCONSENSUS says that there is a presumption of consensus. This is a very important difference.
"Has consensus" means that when you go to a high-traffic article, like Diabetic neuropathy (8,000+ page views a month, 487 registered editors total, consistently has dozens of edits per year, WP:ORES rating of B-class), and you see something that's been in it for multiple years, then there are editors who will actually support the content you are looking at.
"Presumption of consensus" means that when you go to that article, you can just remove the hoax statement in it, even though it was there for 4.4 years, because a presumption of consensus is not the same as actual consensus, just like a presumption of innocence is not the same thing as actual innocence. WhatamIdoing (talk) 18:59, 25 July 2022 (UTC)

Hope you all can come up with a solution. I'm moving one. GoodDay (talk) 23:45, 24 July 2022 (UTC)

The history of WP:ONUS: the prequel

I'm still of the opinion that the onus is on those supporting WP:ONUS to reaffirm that there is a consensus for this policy section. – wbm1058 (talk) 03:35, 3 September 2022 (UTC)

The successful followup to the "no consensus" 2011 RfC was this 2012 RfC which established a consensus for the lead section of the policy but did not address "onus". That word only appears one time in this RfC: "The onus is on us to be good teachers." Not relevant here. The version chosen by consensus was "version D" which only mentioned "inclusion" in a footnote, which remains today as footnote #1. "Version A" said "verifiability does not guarantee inclusion" but there was no consensus for this version. – wbm1058 (talk) 17:49, 3 September 2022 (UTC)

  • In February of 2013, it was moved to its own sub-section, and expanded. At that point it just said that inclusion/omission was determined by consensus (no onus).

Have I met the burden/onus yet of demonstrating that there has been NO CONSENSUS established for the repeatedly boldly added text "The onus is on those seeking to include disputed content, to achieve consensus for its inclusion."? This section was added because of concerns about irrelevant or trivial information. A fan adding irrelevant trivia about one of their favorite topics is "POV pushing"? – wbm1058 (talk) 21:33, 3 September 2022 (UTC)

It seems possible that any consensus is implicit/silent rather than active, certainly this sequence should be considered in the upcoming RFC. Selfstudier (talk) 22:00, 3 September 2022 (UTC)
@Selfstudier: Wikipedia:Silence and consensus § Scope of application says "Apply the rule of silence and consensus only when a weak consensus would suffice. Silence and consensus does not apply when either a strong consensus or a mandatory discussion is required." Should only a weak consensus be suffice for the content of policy pages? I think not; WP:ONUS should redirect to a user essay until consensus for it as a policy is demonstrated. Consider the essay Wikipedia:Silence does not imply consent when drafting new policies. – wbm1058 (talk) 12:47, 5 September 2022 (UTC)
Editing policy (WP:EPTALK) says "Bold editing does not excuse... edits designed to create a fait accompli, where actions are justified by the fact they have already been carried out." I think that's the sort of policy editing I've documented above. – wbm1058 (talk) 13:23, 5 September 2022 (UTC)
  • Ultimately, it does not matter whether ONUS had consensus (back then)… what matters is whether it has consensus NOW. Remember that “consensus can change”. The RFC will establish what the current consensus is, and we can then act accordingly. Blueboar (talk) 13:03, 5 September 2022 (UTC)
    We can't assume that any RfC will establish a consensus; it could result in NO CONSENSUS. – wbm1058 (talk) 13:23, 5 September 2022 (UTC)
    It has been sitting there for some time so while theoretically we can invoke ONUS (lol) and say it has to go out, it's been there this long, might as well wait for the RFC, see what happens with that first. Selfstudier (talk) 13:39, 5 September 2022 (UTC)

Recent reverts

Agreed on "inline" wording change—no more to do here. Altanner1991 (talk) 07:50, 6 September 2022 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

@Jc3s5h: @Zero0000:

Could someone explain to me why the last edits were reverted? Altanner1991 (talk) 12:52, 18 August 2022 (UTC)

If your major policy change were allowed to stand, the policy would have read

Any material that does not have a source must be removed.

Why this change is a bad idea is explained in detail in the essay Wikipedia:You don't need to cite that the sky is blue.
In particular, the consensus policy contains the statements

In the English Wikipedia, verifiability means other people using the encyclopedia can check that the information comes from a reliable source

and

All content must be verifiable.

This does not mean every article must contain a citation for even the most trivial and obvious fact. For these trivial and obvious facts, it is left to the reader who doesn't already know the fact to look it up in some commonly available reference work, such as a world atlas. The consensus version of the article allows well-known easily verified facts to be present without a citation; your version would require a citation within the article for every single fact. Jc3s5h (talk) 13:11, 18 August 2022 (UTC)
I have to digress because "source", "reference", and "citation" are words in gradating order of proximity to the body's text, with source implying verifiability at its most flexible interpretation, and citation meaning the superscript placed next to the paragraph content. To say that sources are questionable can directly contradict the principle of verifiability. That was the distinguishing I was trying to make. I think it would be useful if Wikipedia readers/editors would become accustomed to this language, so that policies can be developed and communicated in easier/more precise ways. Altanner1991 (talk) 12:32, 19 August 2022 (UTC)
The phrase Any material that needs a source directly contradicts the preceding statement: everything on Wikipedia needs to be verifiable; there are no exceptions. The current policy page therefore implies either this contradiction, or it is confusing the concept of a "source" with the concept of a "citation"; the most likely explanation is the latter, but the former explanation can also used unfortunately as an excuse to have unverifiable information. Altanner1991 (talk) 16:25, 21 August 2022 (UTC)
You're right. That could be confusing, especially if someone reads it out of context (i.e., without reading the immediately preceding sentence, which defines when inline citations are needed). I've therefore substituted the word "inline citation" to be clearer. WhatamIdoing (talk) 22:42, 23 August 2022 (UTC)
The grammar had been ambiguous; thank you for clarifying. Altanner1991 (talk) 20:12, 24 August 2022 (UTC)
There is still more though that should be done to distinguish "sourcing" from "citing"/"referencing". These are necessary distinctions.The distinctions were actually sufficient. Altanner1991 (talk) 16:15, 25 August 2022 (UTC); edited 07:50, 6 September 2022 (UTC)
I would just like to endorse Jc3s5h's response here. I would have reverted this change, as well. Happy (Slap me) 13:13, 18 August 2022 (UTC)
In addition, it is not true that material that doesn't have a source but needs one must be removed. There are alternatives such as (a) add a citation-needed tag, (b) add a source, (c) put a note about it on the talk page, (d) let someone else handle it. The original wording "may be removed" is correct. Zerotalk 13:17, 18 August 2022 (UTC)
Indeed you are correct. Thank you. Altanner1991 (talk) 12:29, 19 August 2022 (UTC)
My guess is, the change was reverted because there was no established consensus for it. GoodDay (talk) 15:58, 18 August 2022 (UTC)
Agreed. Altanner1991 (talk) 12:29, 19 August 2022 (UTC)
If so, that rationale would have been insufficient. See wp:DRNC. - Butwhatdoiknow (talk) 14:45, 19 August 2022 (UTC)
WP:DRNC does not apply to policy pages. See the first box at the top of the page: Changes made to it should reflect consensus. Generalrelative (talk) 19:31, 19 August 2022 (UTC)
Generalrelative, the box links to wp:CONSENSUS, which allows bold editing. For more specific support, see WP:PGBOLD, which says "directly editing [policies and guidelines] is permitted by Wikipedia's policies." - Butwhatdoiknow (talk) 16:01, 21 August 2022 (UTC)
A continued search for explicit truths in WP is doomed to failure, thought that was already clear in the ONUS thing. Wiggle room and vagueness is built in, I suspect deliberately. Anyway, this talk is V not CON. Selfstudier (talk) 16:09, 21 August 2022 (UTC)
Selfstudier: Agreed, perfection cannot be achieved, but I encourage you to not give up on efforts to lessen the size of the wiggle room and reduce the level of vagueness. - Butwhatdoiknow (talk) 16:39, 21 August 2022 (UTC)
Agree with ONUS and the box at the top of the page, but the edit was indeed only reverted for being too bold. Altanner1991 (talk) 20:58, 30 August 2022 (UTC)
About the other change:
It is true that any material that does not have a source may be removed on grounds of verifiability. (On grounds of WP:DUE or WP:VANDAL or WP:NOT, etc., you can remove even well-sourced information.)
This rule is not limited to "material that needs a source", because the moment someone removes unsourced material and claims that it needs a source (or that it can't be sourced, or that it ought to be sourced, or any variation on such a complaint), that material automatically becomes material that needs a source. This policy does not recognize the existence of material that some editor claimed needs a source but is somehow exempt from removal just because someone else doesn't agree that it really needs an inline citation. WhatamIdoing (talk) 22:50, 23 August 2022 (UTC)
Yes, that is the usual sourcing policy, thank you. Altanner1991 (talk) 20:08, 24 August 2022 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Must alternatives be equivalent?

An editor made the following change:

Any material that needs an inline citation but does not have one may be fixed, tagged, or removed.

The edit summary says "not equivalent alternatives." I agree that, for example, fixing is better than removing. But I wonder: Do we have an "alternatives must be equivalent" rule? - Butwhatdoiknow (talk) 05:46, 24 August 2022 (UTC) (Bolding added. - Butwhatdoiknow (talk) 15:56, 24 August 2022 (UTC))

Whether to add alternatives to "remove"

To my mind, the question of whether to mention the alternatives depends on what we intend when we use the word “may”… are we warning editors that unsourced information might be removed, or are we giving editors permission to remove? If the former, then there is no need to mention the alternatives. If the latter, then mentioning alternatives is helpful. Blueboar (talk) 12:29, 24 August 2022 (UTC)

Given this statement follows the line of things that require cites (quotes. Contentious material), this line shouldn't use "may" but "should". But to that end we should say "should be removed unless proper sourcing can be cited inline". Masem (t) 13:02, 24 August 2022 (UTC)
Best practice for material that might be right is to tag cn and if one not forthcoming in a reasonable time, remove. If material is thought to be crap/wrong, remove it without the cn step. Fixing is not compulsory, nor should it be. (I'm an AI editor so every statement requires a cite). Selfstudier (talk) 13:20, 24 August 2022 (UTC)
Again, for this specific part of policy, it 7s clearly in reference to material that must be cited, like quotes. This is where fixing is compulsory, but whether that is removing the uncited material or adding the right ref is the question. Tagging an unsourved quote with a cn tag is not an appropriate fix. Masem (t) 13:59, 24 August 2022 (UTC)
No it clearly is not. The whole paragraph is about all material on Wikipedia. The only part about anything that should be removed is the last sentence. Huggums537 (talk) 06:35, 30 August 2022 (UTC)
Actually, now that I look at it, I do see how the thing is sort of stuck in between the last part about removing contentious BLP, and some advice above it saying quotes and potentially challenged material need cites, but then there would be no point for the guidance to say anything about potentially challenged material if it was going to just tell us it "should" be removed anyway. I think material that is "likely to be challenged" is not required to have a cite like quotes are unless it actually gets challenged so the guidance kind of goofed up by lumping these two together since quotes are required to have something done about about them as you say. I just wanted to point out that it isn't clearly all about material that must be cited. Huggums537 (talk) 07:18, 30 August 2022 (UTC)
I believe it was giving guidance over both "potentially challenged material" and "should be removed" because those are two common scenarios. And material that is "likely to be challenged" is actually to most in need of a cite. Not sure what doesn't need a cite; the paragraph clearly states that all material must be verifiable. Altanner1991 (talk) 01:04, 31 August 2022 (UTC)
"should be removed unless proper sourcing can be cited inline" looks like a good start to me. Maybe put in a link to WP:PRESERVE. - Butwhatdoiknow (talk) 16:00, 24 August 2022 (UTC)
This is not true.
When you encounter something that is required to have an inline citation, you have many options, including:
  • remove it
  • cite it
  • change it (to something that doesn't require an inline citation)
  • tag it
  • ask for help
  • do nothing
It is not true that it should be removed; it is not always even true that you (personally and individually) "should" do anything at all. You might, for example, leave unsourced material alone because you're sending the whole page off for deletion.
It is especially not a good idea to write that it "should be removed unless proper sourcing can be cited inline", because we will end up with disputes: "Why are you so mad at me? WP:V says I ought to remove absolutely everything that's unsourced unless I personally can cite a proper source, and I just don't have the time to find sources for anything and my dad won't pay $39 per journal article, so I actually can't cite any of this. Wikipedia:Use common sense isn't a policy, and policies always outrank using your brain, so I'm mass-blanking everything in sight."
@Blueboar, I'm not sure whether this sentence is meant to provide warning to the lazy editors or permission to the article-blankers, but a statement of permission to the latter group would also have the effect of warning the former group. I kind of agree with @Altanner1991, though: BURDEN already says "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", so it's not clear to me that we actually need to repeat that rule ("Any material that needs an inline citation but does not have one may be removed") in the lead. WhatamIdoing (talk) 22:02, 24 August 2022 (UTC)
"May" is ambiguous, and so we have to assume either interpretation, because readers cannot be expected to know otherwise. So, at this point, the message has included the possibility of simply removing material. Altanner1991 (talk) 23:38, 30 August 2022 (UTC)

Another question is: who are we addressing - an editor wishing to include material, or an editor following up on material that has already been added? If the former, we want to stress citing the source when the material is added. No need to mention alternatives. If the latter, then we want to stress FIXING… and there are multiple ways to do that, including adding a citation yourself, tagging (so someone else will add a citation), or removal. Blueboar (talk) 14:21, 24 August 2022 (UTC)

It seems clear that both types of editors are being addressed since all material on Wikipedia is mentioned including editors who wish to add stuff that's "likely to be challenged" as well as material that is clearly a contentious BLP violation, and has already been included, but should be removed. Both are covered in the paragraph. Since both are covered, we want to stress fixing for the benefit of those it applies to. Huggums537 (talk) 11:40, 30 August 2022 (UTC)
Well, I agree with @Blueboar in that it would be beneficial to cater to both new material and old/uncited material, because these two different scenarios might demand slightly different instructional wording, to avoid confusion. But @Huggums537 does bring up the good point of the fact that both views are perhaps sufficiently already in the introduction. Altanner1991 (talk) 23:43, 30 August 2022 (UTC)

I should note that Zero0000 (talk · contribs) had raised in the prior discussion: the material can be fixed, tagged, or raised on the talk page. Altanner1991 (talk) 20:11, 24 August 2022 (UTC)

I would remove Any material that needs an inline citation but does not have one may be removed. because it is too overbearing to resummarize basic Wikipedia principles in that sentence's location in the article, unless others think we have been crystallizing Wikipedia policies enough to do that. Altanner1991 (talk) 20:19, 24 August 2022 (UTC)
If material needs citation, yes, it can be removed. I oppose de-emphasizing this. That someone can simply add a citation is obvious and would be absurd to repeat. And "tagging" shouldn't be emphasized. I've seen far too many erroneous claims stick around for years that someone tagged, but nonetheless stuck around and was not cited. People need to know be directly told that bad unsourced text can be removed. Crossroads -talk- 06:17, 25 August 2022 (UTC)
What is the basis for your argument that the option to remove is less obvious than the option to fix? - Butwhatdoiknow (talk) 06:21, 25 August 2022 (UTC)
Might or might not be, depends the editor and why they are there. As whatamIdoing said, lots of possibilities in theory. Selfstudier (talk) 09:36, 25 August 2022 (UTC)
I would say the responsibility of editors in their edits speaks to the quality of their edits, and that if material isn't cited, it probably wasn't the best quality work. Altanner1991 (talk) 20:34, 30 August 2022 (UTC)
I believe @Crossroads argued that the option to remove is more obvious than the option to tag/fix. Altanner1991 (talk) 00:13, 31 August 2022 (UTC)
Yes, that option is the most ideal. I was speaking to the Wikipedia "crowd"with inexperienced editors in mind, but I, too, am very strict about reliable sources. Altanner1991 (talk) 16:08, 25 August 2022 (UTC); edited 20:36, 30 August 2022 (UTC)
I actually prefer that options for how to deal with uncited information be as broad as possible. I imagine someone is in this situation:

"I have come across this statement that does not have a citation. What is the correct way to handle this?"

If all we advise such a user to do is "you may remove it", it gives them no other options. If we say "Here's a list of things you may do" then it allows the user some flexibility and I find it is also better, when there actually are a multitude of ways to act, that we present those options and not narrow our advice. Certainly, removal is an option, but it does not need to be the only option we present; nor should it necessarily even be the best option, nuance needs to be applied in every case, and when all we say is "you may remove it..." we aren't taking that nuance into account. --Jayron32 16:17, 25 August 2022 (UTC)
And how would that change the wording on the policy page? Altanner1991 (talk) 16:22, 25 August 2022 (UTC)
I would make it say exactly what it said above: "Any material that needs an inline citation but does not have one may be fixed, tagged, or removed." That is sufficient. --Jayron32 13:08, 26 August 2022 (UTC)
Yes it was my suggestion but Crossroads presented I think the more ideal option: we don't want to encourage uncited material. Altanner1991 (talk) 16:56, 28 August 2022 (UTC)
I agree with the simple suggestion of Jayron32. Huggums537 (talk) 11:55, 30 August 2022 (UTC)
Again, the act of "tagging" has become overly broad for unsourced material to linger for years. It's like a sorry loophole! (pardon my British) Altanner1991 (talk) 20:24, 30 August 2022 (UTC) Changing my position per WP:Lede: intro should say the same as (and not contradict with) the body of the article. Altanner1991 (talk) 00:15, 31 August 2022 (UTC)
Yes, @Jayron32 and others have made the same suggestions and I agree, however I think that talk pages should also be included as an option, per an earlier suggestion from @Zero0000. Altanner1991 (talk) 00:41, 31 August 2022 (UTC)
Point to WP:EDITING, WP:FIXTHEPROBLEM etc Selfstudier (talk) 16:24, 25 August 2022 (UTC)
Of course. Altanner1991 (talk) 16:25, 25 August 2022 (UTC)
  • I just want to point out that the policy already contains an entire section outlining alternatives to simple removal. Are we focusing on individual trees instead of the forest? Blueboar (talk) 16:58, 25 August 2022 (UTC)
    Maybe. Perhaps the question we should ask is whether the second paragraph of the policy's lead should contain:
    1. an incomplete statement about what "may" happen to unsourced text;
    2. a (reasonably) complete statement about what "may" happen to unsourced text; or
    3. nothing in this specific paragraph about what "may" happen to unsourced text, because (a) it's explained in detail in the first section and (b) we should move straight on to the point removing contentious BLP matter.
    IMO all of these are reasonable choices, and it's just a matter of picking one. I currently prefer #3 myself. WhatamIdoing (talk) 18:23, 25 August 2022 (UTC)
    I see nothing wrong with a short, and simple summary statement about options for what may happen with unsourced text. If it is covered elsewhere, then that would only reinforce it, and the consequence of not making the statement is potential conflicts with what is covered elsewhere. Huggums537 (talk) 12:13, 30 August 2022 (UTC)
    I would like to throw my support for all of the above suggestions. (Changing my position given what good ideas you have said here. Essentially, WP:Lede would entail not having the contradiction, as you call it.) Of note: ""may" be removed" is unfortunately an ambiguous wording, as it could mean "required", or "potentially". This is especially of issue if no alternatives to removal are shown in the introductory paragraphs. Altanner1991 (talk) 23:14, 30 August 2022 (UTC); edited 00:27, 31 August 2022 (UTC)
    I'm not sure why you think that "may be removed" might indicate that removal is "required".
    If someone says "I may go to the store tomorrow", do you think they are required to do so? WhatamIdoing (talk) 03:29, 31 August 2022 (UTC)
    "Required" as in "essentially expected". Altanner1991 (talk) 05:02, 31 August 2022 (UTC)
    Does "essentially expected" mean "you shouldn't be surprised if it happens"? I would agree with that.
    If you interpret it as "you should be surprised if it doesn't happen promptly", then I would not agree with that. Perhaps 30% of the English Wikipedia's sentences do not have any inline citations. I think it's been more or less stable for about five years. If you are "essentially expecting" the indiscriminate blanking of 30% of article content, you will be disappointed. WhatamIdoing (talk) 05:26, 31 August 2022 (UTC)
    Indiscriminate blanking? No, but future-forward tightening of policy? Absolutely. Altanner1991 (talk) 05:44, 31 August 2022 (UTC)

Let's think logically and try to analyse possible actions (listed above): "remove it", "cite it", "change it to something that doesn't require an inline citation", "tag it", "ask for help", and "do nothing".

  • Do nothing. By doing that, you abstain from any action, i.e. you are acting not as an editor, but as an ordinary reader. The policy that we are discussing does not regulate reader's behaviour. Therefore, this "alternative" does not belong to this list.
  • Ask for help. IMO, the difference between that option and tagging the statement is only in the amount of users to whom you are addressing. That does not change the fact that you ask for help from others in both cases. And, what is more important, by asking for help of by tagging the statement you just initiate the process that may have two outcomes: the issue shall be fixed by adding a reference, or the unreferenced statement shall be removed. Therefore, asking/tagging is not a true alternative to removal.
  • change it to something that doesn't require an inline citation in other words, by doing that, we remove a non-obvious information that requires an inline citation. That is tantamount to the next option, namely
  • remove it. Indeed, if some sentence combines some totally obvious information and some information that requires an inline citation, one may remove it completely, or rephrase it in such a way that only obvious information remained in it. Therefore, I see not much difference between "rephrase" and "remove".
  • cite it - no comments are needed to this option.

Now, when we take a look at the proposed list of the actions again, and combine doppelgangers together, we have:

  • Ask for help either by posting information on some talk page or by tagging the statement, which gives a start to a process that may have two different outcomes: (i) a reference is added, and the content stays, or (ii) no reference is provided in a reasonable time, and the statement goes.
  • Remove an unsourced statement.
  • Fix it by adding a necessary reference.

I think it is clear from my analysis that "tag" and similar options are just temporary, and they are not equivalent to removal.--Paul Siebert (talk) 21:00, 29 August 2022 (UTC)

So we add "Fix It" but not "Tag It"? I'm good with that. - Butwhatdoiknow (talk) 00:55, 30 August 2022 (UTC)
I am against being too prescriptive, in general uncited material that needs a cite either has to get one or be removed. In other words, it may be removed. Selfstudier (talk) 10:29, 30 August 2022 (UTC)
And it may also get fixed (i.e., "get one"). Why state only one of the two alternatives? - Butwhatdoiknow (talk) 15:16, 30 August 2022 (UTC)
After the words "Instead of removing content from an article, consider:", WP:PRESERVE already lists a bunch of possible alternatives, just point to that. May be removed, here at V, is simply reinforcing the need for citations (verifiability). Selfstudier (talk) 15:39, 30 August 2022 (UTC)
Still not clear to me why the need for citations can't be reinforced at V by listing both alternatives. - Butwhatdoiknow (talk) 21:56, 30 August 2022 (UTC)
Of course you are entitled to your view, there are not just two alternatives and in my view more to do with WP:EDITING than with verifiability. Selfstudier (talk) 22:18, 30 August 2022 (UTC)
If some user believes that some statement must be supported by an inline citation, and the reference is missing, there are three ultimate outcomes:
  • Somebody adds the missing reference, e.g., instead of
"The Einstein–Podolsky–Rosen paradox (EPR paradox) is a thought experiment proposed by physicists Albert Einstein, Boris Podolsky and Nathan Rosen (EPR), with which they argued that the description of physical reality provided by quantum mechanics was incomplete."
we get:
"The Einstein–Podolsky–Rosen paradox (EPR paradox) is a thought experiment proposed by physicists Albert Einstein, Boris Podolsky and Nathan Rosen (EPR), with which they argued that the description of physical reality provided by quantum mechanics was incomplete.[1]"
  • Somebody removes the unsourced statement (which may include tagging as a temporary measure), so, instead of:
"The Einstein–Podolsky–Rosen paradox (EPR paradox) is a thought experiment proposed by physicists Albert Einstein, Boris Podolsky and Nathan Rosen (EPR), and, independently, by Lev Landau, with which they argued that the description of physical reality provided by quantum mechanics was incomplete.[1]"
we get:
"::"The Einstein–Podolsky–Rosen paradox (EPR paradox) is a thought experiment proposed by physicists Albert Einstein, Boris Podolsky and Nathan Rosen (EPR), and, independently, by Lev Landau,[citation needed] with which they argued that the description of physical reality provided by quantum mechanics was incomplete.[1]"
and then:
"The Einstein–Podolsky–Rosen paradox (EPR paradox) is a thought experiment proposed by physicists Albert Einstein, Boris Podolsky and Nathan Rosen (EPR), with which they argued that the description of physical reality provided by quantum mechanics was incomplete.[1]"
  • Users achieve a consensus that no refrence is needed (for example, because the statement is too obvious), so instead of
"London is a capital of Great Britain[citation needed]"
we get:
"London is a capital of Great Britain"
IMO, that is an exhaustive list of possible outcomes. Tagging is just a provisional measure. Paul Siebert (talk) 01:56, 31 August 2022 (UTC)
So, when consensus says a citation is required, you're good with the alternatives being "fix it" and "remove it"? - Butwhatdoiknow (talk) 03:38, 31 August 2022 (UTC)
Repeating what has been already said once doesn't improve the case. Selfstudier (talk) 09:25, 31 August 2022 (UTC)
The sentence in question appears in the second paragraph of the lede. Where above that does V suggest "fix" as an alternative? - Butwhatdoiknow (talk) 15:10, 31 August 2022 (UTC)
I would say, unless consensus says no citation is required, any statement that was tagged (or challenged in some other way) should be either removed or supplemented with a reference. Paul Siebert (talk) 14:38, 31 August 2022 (UTC)
The question we are discussing in this section is whether to remove or change "Any material that needs an inline citation but does not have one may be removed." I gather you don't want to add "tag" to that sentence, but what about adding "fix"? - Butwhatdoiknow (talk) 15:08, 31 August 2022 (UTC)
Paul Siebert, I don't find that your reply (below at 15:32, 31 August 2022) clearly answers my question. Whether it's "may" or "must," should we add "cite" as an alternative to "remove"? - Butwhatdoiknow (talk) 14:54, 1 September 2022 (UTC)

References

  1. ^ a b c d Einstein, A; B Podolsky; N Rosen (1935-05-15). "Can Quantum-Mechanical Description of Physical Reality be Considered Complete?" (PDF). Physical Review. 47 (10): 777–780. Bibcode:1935PhRv...47..777E. doi:10.1103/PhysRev.47.777.

Whether to change "may" to "must"

See my comment in the next subsection. I prefer "must (not "may") be removed if the problem cannot be fixed in a reasonable time". If we see that some unreferenced statement needs a reference, we should try to find it, but if our good faith efforts are unsuccessful, this statement must be removed. Paul Siebert (talk) 15:32, 31 August 2022 (UTC)

We can't use must in this context because it is incompatible with our WP:VOLUNTEER status. Nobody can require you to add sources or blank content.
For example, World War II#Start and end dates contains this uncited sentence: "Others follow the British historian A. J. P. Taylor, who held that the Sino-Japanese War and war in Europe and its colonies occurred simultaneously, and the two wars became World War II in 1941." That's the kind of thing we would normally expect to find an inline citation after.
How would you apply your preferred rule to that sentence? There's one page view every few seconds. That's thousands of potential editors each day. Which one of them are we going to blame for "violating" your rule that they "must" remove or fix it? WhatamIdoing (talk) 16:19, 31 August 2022 (UTC)
The phrase "the statement must be removed" has somewhat different meaning than the phrase "editors must remove the statement". The former means that if you remove some unsourced statement, noone can restore it without providing a citation. The latter makes removal an obligation.
I propose "must" instead of "may", because the latter makes removal non-mandatory (which is wrong).
With regard to the example provided by you, I added a "cn" tag to this statement. It seems the WWII article already cited Taylor, so the citation is present, but it was misplaced. I have no time to re-read Taylor right now, if other users will not fix the problem by the end of the year, I'll try to do that. Paul Siebert (talk) 01:25, 1 September 2022 (UTC)
The former means that someone has to remove the uncited sentence. It says nothing about whether people (including that same someone) is allowed to put it right back.
Removal is non-mandatory. There are other options, ranging from fixing it to benign neglect; therefore, removal is not actually mandatory. WhatamIdoing (talk) 06:33, 1 September 2022 (UTC)
  • What about "should", following the MoSCoW method? " Should" is stronger than "may" but not as strong as "must". Masem (t) 15:29, 1 September 2022 (UTC)
    I'm perfectly ok with saying "should" as long as we include the alternatives, but I agree with WAID that we absolutely can't say "must" because if we do include the alternatives, then there is no way to make it imperative for an editor to choose between fixing, tagging, or removing, and we can't force an editor to be obligated to choose to one since it is just simply a morally bad option. Huggums537 (talk) 23:18, 1 September 2022 (UTC)

The policy page is, at this point, out-of-order

At this point, the policy communicates removing anything uncited and this is completely unacceptable.

Yes, "may" can mean "may not", but unfortunately readers are not required to take this view and so we have a problem. I suggest restoring the "alternatives" regardless of whether or not they are "equivalent", as an emergency repair to the policy's introduction. Altanner1991 (talk) 01:15, 31 August 2022 (UTC)

Another option would be to write "any unsourced material may be removed" because this means the material is definitively invalid, and so ultimately would never be able to be on Wikipedia. At least the logic of that policy would be sound. The question of sourcing would merely be "in delay", say, discussion or otherwise. Altanner1991 (talk) 01:19, 31 August 2022 (UTC)
you need to read that sentence in context of the preceding one: 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. Any material that needs an inline citation but does not have one may be removed. the only things that needs an inline citation are quotes or contentious material. Masem (t) 01:38, 31 August 2022 (UTC)
Ah okay, thank you. Altanner1991 (talk) 01:46, 31 August 2022 (UTC)
I agree with @Masem, not all material needs an inline citation, only the four types listed at WP:BURDEN. Also, I agree with you that restoring the alternatives don't have to be "equivalent", and really don't see what that has to do with anything at all. Huggums537 (talk) 01:58, 1 September 2022 (UTC)
Sorry, I didn't realize inline citations weren't always required. And so at this point I do think that it might be too quick a summary to present each alternative as equivalent, because editors should know the differences for when to use each alternative. Altanner1991 (talk) 03:20, 1 September 2022 (UTC)

Frankly, this "likely to be challenged" is redundant. If I see some statement that, in my opinion, is likely to be challenged, I can either take no action (which means, I am acting as an ordinary reader, so this policy in not applicable to me), or I may write on the article's talk page: "Hey, the statement X it is likely to be challenged, let's do something with that.", or I may put a "cn" tag. However, by doing that, I de facto challenge the statement X. In other words, we have no way let other users know that some statement X "is likely to be challenged" without challenging that statement. Therefore, "is likely to be challenged" is redundant, and it should be removed.--Paul Siebert (talk) 02:08, 31 August 2022 (UTC)

I may put a "cn" tag. However, by doing that, I de facto challenge the statement X. I think that is only your interpretation saying, "I hereby challenge this material.", and it sounds ok on paper, except there is no direct accountability for the WP:BURDEN of the person who added it, just a general challenge issued to the community at large. This is very much at odds with BURDEN since it is not designed to be a war with the general community, but a responsibility of the person who added, and is now trying to restore. Another different interpretation of the CN tag is an editor sending a general message to the community, "hey somebody might want to help find sources for this because it looks like maybe is likely it could get challenged.", and "this needs a source, but I can't find one please help.", or any number of other interpretations that are not "de facto" challenges to the editor adding or trying to restore the material. In other words, simply leaving a tag doesn't notify the editor who added and might want to restore, it only notifies the general community so a "challenge" has not been issued. You can't demand the BURDEN be on an editor, and then turn right around and deprive them of the very opportunity to meet that responsibility by keeping them in the dark when their material is challenged. (BTW, don't ask me how you are supposed to dig up the history, and find the editor who added the material, or the proper procedure for how to notify them about your "official challenge" because I didn't write this messed up policy, I just interpret as I read, but it only goes to show you that even the most heavily bolded emphasis on policy is just a little bit off enough to prove maybe that wasn't so "de facto" after all) Huggums537 (talk) 00:23, 2 September 2022 (UTC)
Nevermind all that. I just now saw you agreed down below that "likely to be challenged" is more aimed at editors adding material, and I also agree with that. Huggums537 (talk) 00:44, 2 September 2022 (UTC)
I also don't like the fluff in that policy wording. It breeds potential for abuse. Altanner1991 (talk) 02:39, 31 August 2022 (UTC)
I think we should keep it. The language is also present in other places—at the very least it's in WP:BLP and MOS:LEADCITE. LEADCITE is the place I see this mentioned the most, and editors are having productive conversations about what content in the lead needs a citation based on the likelihood of a challenge. Firefangledfeathers (talk / contribs) 02:55, 31 August 2022 (UTC)
I think the "likely to be challenged" bit is aimed at the editor who is adding the material, not people looking at it later. We don't want scenarios like this:
  • 2010: I add a sentence asserting that when B cells internalize antigens for processing, membrane-anchored IgD is systematically excluded from the resulting vesicle. Nobody's challenged my new sentence yet, and policy doesn't require a citation for material unless it's a direct quotation or already challenged. I can easily predict someone will want a citation for this, but policy says I don't have to bother until the challenge appears.
  • 2011: I quit editing.
  • 2012: Someone really wonders whether I just made up all of that.
WhatamIdoing (talk) 03:43, 31 August 2022 (UTC)
Indeed, I had likewise figured that it is for adding material. Altanner1991 (talk) 05:47, 31 August 2022 (UTC)

Thank you for explaining the policy to me. There is no issue. I did not know that material could exist without some form of citation. Altanner1991 (talk) 03:20, 31 August 2022 (UTC)

Would you find it useful if the policy said somewhere, perhaps in a footnote, something like "While all material must be verifiable, not all material is require to have an inline citation"? WhatamIdoing (talk) 03:44, 31 August 2022 (UTC)
Not necessarily, because I would say that editors should strive to include citations as much as possible, but thank you. Altanner1991 (talk) 05:06, 31 August 2022 (UTC)
You said "I did not know that material could exist without some form of citation." I imagine that other editors also do not know this. How do you think that we could we help people to learn this fact? WhatamIdoing (talk) 05:28, 31 August 2022 (UTC)
The "explicit policy" approach: yes, that's good too. I suppose a footnote would suffice, but I would support either that or even a sentence without footnote. "While all material must be verifiable, not all material is required to have an inline citation" is nicely clear. Altanner1991 (talk) 05:33, 31 August 2022 (UTC)
For future reference, this has since been raised as a proposal at Wikipedia talk:Verifiability#Verifiable, not verified. Altanner1991 (talk) 08:13, 6 September 2022 (UTC)
It seems that you are right, and "likely to be challenged" is an instruction to those who add some new material. I would say that these two sentences ("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. Any material that needs an inline citation but does not have one may be removed.") should be made more clear. The first should describe what we should do with the information that we add, whereas the second one should tell what should we do when we see some unreferenced statement. Obviously, when I am adding some information that is likely to be challenged, I am expected to provide an inline citation. That is what the first sentence should say. ("All quotations, and any material whose verifiability is likely to be challenged must include an inline citation to a reliable source that directly supports the material.") The second sentence explains what we should do when we see some information that was not supported by a reference: "Any material that needs an inline citation, but does not have one should be removed if the problem is impossible to fix in a reasonable time."
In other words,
  • if you add some information and you feel that information is not obvious, add an inline citation. That is what the first sentence should say.
  • if you see some information that was added by somebody else, and this information needs a reference, that must be fixed (either by adding a ref or by removal). That is what the second sentence should say.--Paul Siebert (talk) 15:28, 31 August 2022 (UTC)
Honestly, I think all this rewording is too much, and only confuses things more. It was much more simple when we were just talking about adding "fix"/"tag", and choosing between "may" or something else. Huggums537 (talk) 01:08, 2 September 2022 (UTC)
Having said that, I think Paul S. is right that the guidance is not clear. Everything in the paragraph that talks about inline citations references the exact same four things mentioned in WP:BURDEN. Three of these things are very easily identifiable as obvious reasons for either adding a citation when you add the material, or removing/fixing the material when you encounter it. This fourth "likely to be challenged" is not so obvious because I think most editors don't realize that it belongs to a group of items which are very easily identifiable as obviously needing a citation meaning that since it belongs to that group, this is how it is intended for us to interpret it. I think sometimes editors see it as meaning "maybe might need a citation, maybe not" or "questionable enough so go ahead and add" and I have even seen such sage advice as "when in doubt just add", and this all seems like reasonable advice on the face of it, but how can you really quantify what lingering "doubts" there might be as to what qualifies as "questionable" in the mind of any particular editor? The answer is that it is far easier to remove the doubts, and quantify what qualifies as likely to be challenged by including it within the list of things that are required to have inline citations. Those are the things likely to be challenged, and even those things don't always get challenged. Many times they are just tagged, or fixed. The point there being that just because the guidance tells us it is only the obvious stuff that is "likely to be challenged" still might not mean it would ever be actually challenged. Although it "likely"(80%-90%) would. Huggums537 (talk) 11:20, 2 September 2022 (UTC)
I have often thought that the phrasing should be reversed; from: "... is challenged or likely to be challenged" to "... is likely to be challenged or actually is challenged". This would better follow the chain of events:
1) you want to add material...
2) You should ask: is it the sort of material that is likely to be challenged?
3) if so, you should include a citation... if not, you can skip it... unless/until
4) it turns out that someone else actually does challenge it. At that point, you are required to cite it, even if you originally thought a challenge unlikely. Blueboar (talk) 12:43, 2 September 2022 (UTC)
Another option for 4) is "If another editor thinks your added material is correct but is likely to be challenged then the other editor may provide the citation." - Butwhatdoiknow (talk) 14:38, 2 September 2022 (UTC)
In which case, the material was not challenged… it was edited. Blueboar (talk) 14:48, 2 September 2022 (UTC)
These sound like reasonable ideas here, but I forgot to mention in my comment above about the opposite end of the spectrum. I think that since the guidance is telling editors who want to add that the most obvious things that are likely to be challenged are required to have inline citations, then the opposite is also true that only the most obvious things should ever be included without an inline citation. In other words, always add some kind of attribution, (which must be an inline citation in some cases), unless it is just obvious one is not needed. The problem is that it is not always obvious to everyone when an inline citation is not needed. Even if you look at the sky is blue examples about the capital city Paris, France that @WhatamIdoing always argues, you will see that attribution, and sources likely exist within those articles (perhaps a map with a legend for example) to verify it. This eliminates the need for an inline citation. I think the argument that the sources are "out there somewhere" is a valid one, but the point can equally be proven without even having to say that. However, the funny thing is that after you have proven the point that a citation is not needed right there at that spot, you kind of do get to say that now. I don't think there are very many editors who get the concept that general references and attribution in the article body are allowed even though they are not the best. Huggums537 (talk) 16:19, 2 September 2022 (UTC)
"At that point, you are required to cite it" assumes that "you" are still editing (that article; at all), which is not something to be relied upon.
Also, I think that editors would read that line and think '"you" means "not me", so I don't have to be helpful or collaborative. All I have to do is say "Simon says I challenge this!" and then I get to blank everything unless you do it my way.' This leads to pointy-headed behavior, like challenging material that you are certain is verifiable just because you don't like adding citations yourself, or tagging all uncited sentences in an article and coming back 30 days later to blank everything, unless a whole bunch of other volunteers drop everything to provide all the citations by your arbitrary deadline. See also "This article is a mess, so I'm going to send it to AFD, because the threat of deletion motivates people to clean up articles." WhatamIdoing (talk) 16:27, 2 September 2022 (UTC)
Blueboar, you are almost right, except " At that point, you are required to cite it". It is Wikipedia, and nobody is required to do anything. It would be more correct to say:
"At that point, neither you nor anybody else can prevent your material from being removed, even if you originally thought nothing was wrong with the text. The only way to preserve it is to add some reliable source that directly supports your statement."
--Paul Siebert (talk) 03:39, 6 September 2022 (UTC)
Paul's got it. That's our real process. WhatamIdoing (talk) 04:05, 6 September 2022 (UTC)

Sources must have been "published"

This edit was reverted[3] by Jc3s5h, but AFAIK, this is the current policy, and I was not changing it, simply explaining it better. Andre🚐 02:15, 29 August 2022 (UTC)

The means of gaining access to a source mentioned include online access that requires paying a fee, a print source in a library, and rare historical sources in special museum collections and archives. More often than not, providing access to a source through these means is publication. By adding the statement "though they still must have been published at one time" it creates an implication that making documents available through a public archive, or a web site that charges fees, is not publication.
An example of a document that is not published is a document in an archive, or similar entity, that is not accessible to the public. An example of an unpublished document would be a death certificate held by a US state health department that is not made available to the public, because it has been less than 100 years since the person died. An example of a published document would be a death certificate from the very same state health department that is available to anyone who pays the required fee because the person died more than 100 years ago. User:Jc3s5h 02:28, 29 August 2022‎
An unpublished document is not verifiable because it's not published. Verifiability is about published sources. Archives where you have to make an appointment to dig through them, that's OR. See also discussion at Wikipedia_talk:Reliable_sources#Objection_to_"Definition_of_published" Andre🚐 02:30, 29 August 2022 (UTC)
Documents available to the general public, especially in an organized way, are published. Lots of documents are hard to use, that doesn't make them unpublished. Jc3s5h (talk) 03:00, 29 August 2022 (UTC)
I feel like we're talking past each other. I agree that some documents are for-pay, or hard to obtain, whether journals needing a subscription, or something that's out of print. These sources are acceptable to use. I am trying to clarify that unpublished archival sources aren't usable. Andre🚐 03:04, 29 August 2022 (UTC)
If the archive is open to the public (or the archive makes some documents available to the public and the document being cited is available) then it's published. Jc3s5h (talk) 03:14, 29 August 2022 (UTC)
(edit conflict) An archive is not a publication. "creates an implication that ... a web site that charges fees, is not publication" I don't see that implication, at all. Per WP:NOR as a dilettante, we should not be looking at primary source archival records. Leave that research to legitimate historians as we can use the secondary sources they publish. Again, the point here is that publication is the bedrock of verifiability, regardless that a book is out of print or a website is behind a paywall. That a primary source document exists somewhere does not verify facts in the way we need. And, might I say, the way many editors on this website shamefully ignore checking sources, I'm surprised you want to quibble over this point. Chris Troutman (talk) 03:17, 29 August 2022 (UTC)
I agree with Chris, for a change ;-) Andre🚐 03:18, 29 August 2022 (UTC)
You're trying to say that they are 1) unpublished, 2) not usable, and 3) to use them is OR. All three are incorrect interpretations of WP's definitions of "published", acceptable RS, and OR (what does WP:OR have to do with you having to pay to access a document? Where is that kind of thing mentioned, or hinted at, anywhere in that policy, or in that policy's history?) SamuelRiv (talk) 03:14, 29 August 2022 (UTC)
Huh? I just said, that, for-pay sources: "These sources are acceptable to use" Maybe you don't know about this, but there is such a thing as an archive, it's a physical place where you can go and look through records. These records often were never published and that is why in the past, and as far as I know, continuing to today, the community has considered it original research and unverifiable to go digging through specialized records that were never published, as in made available to the public (not, just, to some people like academic researchers by appointment) Andre🚐 03:17, 29 August 2022 (UTC)
That sort of thing would very much depend on the context and I don't think a policy can reliably specify when an archive would be an acceptable source. For example, an archive of presidential papers that permitted general access might be acceptable as verification for some text (and it might not). If a business has an archive box of papers in their warehouse, it would probably not be acceptable as a source. Johnuniq (talk) 03:56, 29 August 2022 (UTC)
There's whiplash between here and the RS thread -- I suppose I had the expectation that you could take combined context from the conversation previously, within hours between posts, remember them, and conclude that I probably made a typo, and then address the substantive point: that OR says nothing about "digging through records that were never published" (and whether or not "the community has considered it original research and unverifiable" (two different policies) (to claim the community has concluded this requires a link to past discussion) is not really helpful if the stable policy does not in any way state this. OR is about writing articles, not about how you go about "digging through records".
The broader issue seems that you want to expand WP's definition of 'published'. It's short and simple and easy to understand as is. It is then a tool which helps write clearer policies which are then used for RS, V, etc., that might conform to what you might colloquially prefer to think of as 'published'. Don't then go around and say that a foundational tool made to be as simple as possible is now inadequate because it doesn't fit your preferred colloquial definition -- that's not its purpose. SamuelRiv (talk) 03:56, 29 August 2022 (UTC)
Hang on, it's a well-established community consensus that unpublished sources, which does include some archival records, are not acceptable per WP:V. You are incorrect in your understanding of policy. Since you and I have had this difficulty in the past of not really operating in the same consensus-based factual reality, let's agree to disagree and someone else will be around presently to correct you (or me, but I'm pretty sure it's you) Andre🚐 04:05, 29 August 2022 (UTC)
You misunderstood entirely what I said. I don't think me restating it will help. SamuelRiv (talk) 04:24, 29 August 2022 (UTC)

If you read the whole policy you will find

Source material must have been published, the definition of which for the purposes of Wikipedia is made available to the public in some form.

This is followed by a footnote which states

This includes material such as documents in publicly accessible archives as well as inscriptions in plain sight, e.g. tombstones.

The edit which prompted this thread made no distinction between special museum collections or archives which are public and those which are private. Allowing the edit to remain makes the policy self-contradictory. Jc3s5h (talk) 04:24, 29 August 2022 (UTC)

I am not particular about the text, but I am concerned if users think it's OK to add material from archival sources that haven't been published at any time. I maintain that a private museum collection or an unpublished record in an archive, is not published. Verifiability and reliable sources talk about the concept of publication for a reason. AFAIK this hasn't changed. I've pinged a bunch of people from the prior discussion who are still around to see if any will drop by and shed some light. Andre🚐 04:36, 29 August 2022 (UTC)
  • They talk about publication, but the actual Wikipedia policy on verifiability (Wikipedia:Verifiability), which you reference, explicitly states as noted above, that this (the term "published", which the policy explicitly states "the definition of which for the purposes of Wikipedia is made available to the public in some form") "includes material such as documents in publicly accessible archives as well as inscriptions in plain sight, e.g. tombstones.". Thus, respectfully, you are simply wrong. The policy is quite explicit that publication does not mean "published as in a journal/book/magazine/etc", it simply means "made available to the public in some form". Indeed, WP:NOR also explicitly recognizes that, although they should be less-used than secondary sources, primary sources are also ok. Many primary sources are not "published" in the manner you suggest, but are absolutely "available to the public in some form" (such as in an archive) and are, therefore, both "published" and "verifiable" for WP purposes. A tombstone (an example the policy explicitly references as acceptable) is not "published" in the traditional sense you reference, but is available to the public in some form, so is 'published' in a Wikipedia sense. An article that overly relies on primary source material (whether archival or 'published' somewhere, in your parlance, like in a collection of letters) may run afoul of WP:NOR's guideline that primary sources should be used "to a lesser extent", but un-"published" (by your definition) archival materials are no more or less reliable (and certainly no more or less verifiable) than, say, some random self-published book or website online. tl;dr: As I read it, WP's definition of "publication" turns on availability to the public in any form and not availability via a book/journal/website, as you seem to be calling for. Best. Staxringold talkcontribs 12:59, 29 August 2022 (UTC)
    • As an add-on afterthought, I agree with Jc3s5h that your edit makes the policy self-contradictory. The entire policy says (to paraphrase) "these materials do not have to be 'published' in the traditional sense, they must simply be available to the public" and you added a footnote that said (again, to paraphrase) "but they totally have to have been 'published' in the traditional sense at some point." Staxringold talkcontribs 13:06, 29 August 2022 (UTC)
      Within the framework of WP:V, "published" means that the material is in a fixed or tangible record that can be cataloged or indexed, and accessed (perhaps at cost) by others not closely tied with author. This means that any citation should be able to be located and reviewed by any reader, albeit with some factors related to cost and time...the material cannot be wholly inaccessible to any person. For example, we would not be able to use a phone call made between a editor and an expert as there's no record of it. However, a phone interview transcribed by a reliabke source in a journal would be published. We'd not be able to use the personal notes of an expert stored by their spouse in the home, but we would be able to use those notes if stored in a university archive. Masem (t) 13:59, 29 August 2022 (UTC)
I don't really understand the problem here. The text in WP:V says we can use "historical sources" if they're in an archive, and it discusses that in the context of "reliable sources". And WP:RS, in turn, says in the opening sentence, "Wikipedia articles should be based on reliable, published sources". So there we go. Published sources that are found in an archive are in. That doesn't mean that unpublished sources in an archive, for instance non-peer reviewed self-published books (of which there are many in my local state archive) or unpublished manuscripts are therefore reliable sources. It's not the location that matters--it's the oversight that comes with publishing, which is why not all published books are equal: some publishing houses do more checking than others. Drmies (talk) 14:21, 29 August 2022 (UTC)
Wikipedia:Reliable_sources#Definition_of_published defines published as being made available to the public, not solely commercially published. Self-published materials may be problematic for other reasons, but technically they are considered published by this definition. Nikkimaria (talk) 14:50, 29 August 2022 (UTC)
Sure but that is missing the forest for the trees. "Commercially published" is already kind of a misnomer, since much academic publishing isn't really commercial, and self-published material isn't the crux of my argument. Much of what we find in archives isn't "self-published"--it isn't published in the first place, and there's tons of different kinds of material. Let's say you find measurements and counts of fish in Alabama, written down by a grad student or state worker during field work, and it's deposited in an archive. It's being in an archive doesn't make it any more reliable, and I think those of us who work in science and humanities, and who deal with archives and scholarly publishing, know this very well. I'm not even talking, really, about the edit and its revert--it's just that I sense some misunderstanding here, and not necessarily on your part, Nikkimaria. Drmies (talk) 15:56, 29 August 2022 (UTC)
I agree, which is why I'm not going to invest too much time in debating the definition of published - I think the definition at RS would include that archived paper, but I don't think that matters because reliability (and potentially OR) are much stronger arguments for excluding it. The published piece of "reliable, published sources" is more useful for excluding personal communication citations. Nikkimaria (talk) 16:03, 29 August 2022 (UTC)
Drmies attempts to give an overall view of the policy, but the discussion in this thread is what the policy is and how to express it. And Drmies's understanding of policy is wrong. A non-peer reviewed self-published book may be a reliable source if the author met the criteria spelled out in WP:SELFPUB. Jc3s5h (talk) 15:03, 29 August 2022 (UTC)
Yes, sure--except that putting it that way obfuscates the many precautions we have to take, that self-published material needs to meet very stringent criteria, and that it can't be used for a BLP. Drmies (talk) 15:56, 29 August 2022 (UTC)
Published does not mean reliable. Reliable does not mean published. English-speaking people have developed separate words for these two separate concepts for very good reasons, and Wikipedia should respect the distinct meanings of these words. Jc3s5h (talk) 17:12, 29 August 2022 (UTC)
FWIW, I agree with Drmies. The oversight of publishing is valid and important. Andre🚐 03:55, 31 August 2022 (UTC)
Oversight is hugely important – for reliability. But for merely determining whether it's "published" or "unpublished", then it's pretty much irrelevant. Oversight isn't inherent in the publication process. (If it were, then self-publishing would be an oxymoron.) WhatamIdoing (talk) 04:03, 31 August 2022 (UTC)
That's true. Oversight is for reliability. Self-published and unreliably published sources do exist, and are published. But a reputably published source is a thing too. I think we need to be able to distinguish the different levels and values of different kinds of publication. Academically published, or published by a major newspaper, website, journal, publishing house, with a good reputation, can can confer reliability. Totally unpublished should be bottom of the ladder or close to it. I think the semantic difficulty in the wiki definition of published is that it seems to open the door for unreliable stuff to come in more easily. That's the subtext and why we keep relating verifiable and reliable to published. Andre🚐 04:07, 31 August 2022 (UTC)
I will throw in here my opinion that "reliable sources" are a subset of "published sources". We determine reliability by reputation, that is, by whether the publisher, author, or the source itself have a reputation for being reliable. We use a lot of shortcuts, such as distinguishing academic publishers, newspapers of record, etc., but when it comes down to deciding whether a particular source is reliable, we, as a community, are making a judgment based on reputation, however badly we perceive it. How, then, can a source have acquired a reputation for reliability if it has not been published, and thus has not been subject to review by multiple, unrelated readers who have indicated their view of the source publicly (i.e., commenting on or using the source). Donald Albury 17:48, 31 August 2022 (UTC)
Agree 1000% with Donald. Andre🚐 17:50, 31 August 2022 (UTC)
How, then, can a source have acquired a reputation for reliability if it has not been published, and thus has not been subject to review by multiple, unrelated readers who have indicated their view of the source publicly (i.e., commenting on or using the source)? Approximately the same way that anything published five minutes ago does? Approximately the same way that a press release does? Approximately the same way that a social media post does?
Do you remember the BLP whose birthday was wrong in the Wikipedia article? It took him several attempts to get a message on social media that Wikipedia editors would accept. (If memory serves, the first post didn't have the full date.) He ended up writing something like "Dear Wikipedia, my birthday is January 1, 1965" on a piece of paper, taking a selfie, and posting it to one of his verified social media accounts.
But I ask you: How has his self-taken, self-published selfie been subjected to review by multiple, unrelated readers who commented on or used the source? And yet, we accepted it, because it was a reliable source for statements like "Bob posted on social media that his birthday was January 1, 1965". WhatamIdoing (talk) 00:38, 1 September 2022 (UTC)
The self-published birthday info is a special case exception that is specifically spelled out in the policy. Which I assume you well know.
Let's talk about where reliability actually matters. Reporting on contentious and controversial stuff. Got anything cut from that sort of jib? Andre🚐 00:41, 1 September 2022 (UTC)
Wikipedia ideally doesn't "report"; ideally, it "summarizes".
For controversial content, one ideally wants sources that are published and secondary and independent. That means "available to the public", "combining ideas from multiple sources to produce a new thing" (e.g., a summary or a compare-and-contrast analysis), and also "not paid for or beholden to any of the actors in the controversy".
The section heading here is ==Sources must have been "published"==. Whether they're reliable is a much more complex question. WhatamIdoing (talk) 00:46, 1 September 2022 (UTC)
I was referring to the sources doing the reporting, but either way, you're quibbling semantics.
What I think we're missing is that sources get more reliable depending on how they were published. Andre🚐 00:47, 1 September 2022 (UTC)
  • This is interesting. I was called in here, I think, because I participated in the discussion of the unsuccessful 2006 proposal to accept publicly-available archived sources. Interestingly, while that proposal failed, the policy has over time evolved to be, if anything, considerably broader. Nor is this a matter of unchecked BOLDness: the meaning of "published" for WP:V appears to have been discussed pretty extensively over many years. The discussion most responsible for the current wording appears to be this one from 2013.
    FWIW, I think this shift in policy is correct, for substantially the same reasons that I supported the failed 2006 proposal. If we say that a published book that can no longer be found in any library is a valid source, but that an old organization report stored in a public archive (that anyone can view on request) is not, we aren't really doing the concept of "verifiability" any favors. Obviously we should avoid using such problematic sources whenever possible even if they are reliable. But they may be the best source available for some particular factual questions. If we exclude them, we will inevitably end up with articles that contain demonstrably false information that cannot be corrected because the correct information is only available in an archive, on a gravestone, etc. That wouldn't serve the project, or the reader, very well.
    (I am intrigued that the policy drift here seems to be an example of the community gradually course-correcting away from excessive exclusion, which I suspect was possible because excluding sources does not operate to exclude voices from the discussion in the same way that excluding content does.) -- Visviva (talk) 15:19, 29 August 2022 (UTC)
    @Andrevan, could you explain how your specific concern (I am concerned if users think it's OK to add material from archival sources that haven't been published at any time) would work in practice? Specifically, if the archived material isn't available to the public, then how exactly do you imagine that the Wikipedia editor – who is presumably a member of that same "public" – accessed it? WhatamIdoing (talk) 15:22, 29 August 2022 (UTC)
    There is all kinds of stuff in an archive, like private letters, internal memoranda and so on, and archives can be opened selectively. I am concerned that it's original research to be looking through primary sources that haven't been published, not reliable since the publication lacks oversight, not verifiable since many archives are only open to academic researchers by invitation, or for those with the proper government or journalistic connections or so on. As I said, I'm not attached to the language I added or the edit I made, but I don't see that the 2013 discussion considered the question of archival sources explicitly, so I think it's worth having a broad discussion about it. I can't find a more recent discussion than the 2006 one on that specific question. I am still under the impression that the community doesn't want published to be so broad as to include a private letter that happened to end up in an archive box. There might be some very narrow usage of such a thing for simple facts only, per WP:PRIMARY. This discussion has only confirmed my thought that we should clarify the "definition of published" Andre🚐 16:23, 29 August 2022 (UTC)
    I don't feel like you've answered my question. Here's the story:
    A Wikipedia editor wants to write an article about Joe Film. We know that there is an archive: the famous Joe Film Collection at the Big University Archive. We even know that there is a pile of letters from Joe Film's mother in there. Unfortunately, the archive only allows specifically invited researchers to look at anything in the archive, and they're all treating it like fragments of the Dead Sea Scrolls, so nobody knows anything about what his mother wrote in her letters.
    So: The letters are not available to the public, and, per the English Wikipedia's definition, the letters are therefore "not published". Every one of those letters is what you called material from archival sources that haven't been published at any time
    Now for the part that I hope you can fill it for me: How would the Wikipedia editor be able to add material from archival sources that haven't been published at any time? The Wikipedia editor has no access to the archive. How exactly does the Wikipedia editor read the source?
    And if it is impossible through any reasonable, non-magical means for an editor to add material from archival sources that haven't been published at any time, then why are we worrying about whether any "users think it's OK to add material from archival sources that haven't been published at any time"? What's the point of worrying that people might think it's okay to do something that they physically can't do? Isn't that like worrying that they'll start a nuclear war by whistling? WhatamIdoing (talk) 05:46, 31 August 2022 (UTC)
    WhatamIdoin's comment seemed to be addressed to Andrevan, and I'm not Andrevan, but I'll answer anyway. Famous Director is an alumna of Big University, and donated money to build the building that houses the School of Cinema at Big University. She is also a Wikipedia editor. She has access to the collection and cited a letter from the collection in a Wikipedia article. But ordinary readers, even those who are willing to make an appointment and travel to Big University, don't have access. So the citation and the statement it supports should be removed, unless another suitable source can be found.
    At several points in this discussion I have seen presumptions that would unnecessarily prevent qualified researchers from using the methods they normally use. For example, using primary sources to fill in details that secondary sources omit. Some methods used by qualified researchers are not suitable for Wikipedia (for example, citing personal communications they received) but we should not take an approach that unnecessarily alienates qualified researchers. Jc3s5h (talk) 14:10, 31 August 2022 (UTC)
    Jc3s5h, but we do take an approach against original research. WP:NOR. Primary sources are usable for only very basic facts, and should be supported by reliable secondary sourcing if possible. If you're using primary sources to fill in significant details, that's likely original research. This is a feature, not a bug.
    As to WhatamIdoing's question, in the scenario where the film collection is hard to access, there is still a scenario where someone knows Joe Film's cousin and gets in to the hard-to-access archive on a partial basis. Or a more realistic example would be the State Archive of the Russian Federation. It doesn't appear to be covered in our article, but if you're studied Russian history at the university level, you would know that the archives were closed throughout the Cold War to Western researchers. They then opened up in 1992 or so, prompting a ton of revisionism and new studies, and new books and papers revisiting the facts and analysis of the Soviet era. (Goldman, "Terror and Democracy in the Age of Stalin" is the book I'd recommend if you're interested, and she is a fabulous teacher as well. Let History Judge by Roy Medvedev was the text for her class.) The archives have started closing again more recently since Russia is a global pariah now thanks to Putin. Maybe an extreme case, but I have concerns about the reliance on physical archives for a source. I feel much better when there's at least an archived photo or document record. Andre🚐 14:24, 31 August 2022 (UTC)
    Sure: But the policy already says that if the public – not blood relatives, not people with special connections, not members of the club – don't have access, then we say "Sorry, not available to the public, not WikiJargonPublished, not allowed in articles, end of story". Why are you worrying about this scenario, when 99.9% of editors can't do it, and the other 0.1% of editors are prohibited from doing so by policy?
    The scenario that makes sense to worry about (a little) is the one in which the hand-written letter in the archive *is* WikiJargonPublished, and anyone who makes an appointment can have a squint at it. But we need not worry much, because the most important restrictions for that scenario are found at WP:PRIMARY. We don't need to ban overreliance on the letter in WP:Published because overreliance on that source is already banned elsewhere. WhatamIdoing (talk) 15:31, 31 August 2022 (UTC)
    It's a good point Andre🚐 15:34, 31 August 2022 (UTC)
  • It seems my position has not changed since 2006. I think, however, that maybe we should focus less on "published", and instead consider how accessible a source is. I think "published" was used all those years ago because it meant that identical copies of the work were distributed, and were thus available to multiple editors and readers. Thus, and I remember this question coming up years ago, a book that was published, but which was now so rare that the only known copy was locked away in a private library, accessible only to a few select scholars, would not meet the spirit of "published". So, that's my understanding of "published" in the context of verifiability: a work has been "published" if it is available to multiple users at multiple locations, either on the Internet, or in libraries, bookstores, etc. - Donald Albury 18:13, 29 August 2022 (UTC) (Edited 18:14, 29 August 2022 (UTC))
    I agree, maybe that is the key as you say, broad public audience distribution and availability. Andre🚐 19:40, 29 August 2022 (UTC)
    I don't agree; that would rule out all primary materials held in one location with access provided to multiple users, but with duplication restricted. If that is really meant to be excluded, that case should be explicitly called out. Mathglot (talk) 01:02, 31 August 2022 (UTC)
    Out of curiosity, are there any sources like that in articles that you know of? Andre🚐 01:04, 31 August 2022 (UTC)
    Interstate 8#cite note-sign-26 cites a sign on a road. St James' Church, Sydney#cite note-Kennedy-139 cites a plaque on a church wall. If you want to know whether those are real, then someone needs to physically go to those places. WhatamIdoing (talk) 04:01, 31 August 2022 (UTC)
    Thanks so much for providing an example and a very good faith engagement in this discussion. However, both seem to have a link to an archived photo that's accessible on Commons, so easily verified. Andre🚐 04:08, 31 August 2022 (UTC)
    Photos on Commons are WP:USERGENERATED and therefore unreliable sources. You have no way of knowing whether those items exist in the alleged locations, or even if they exist at all. The only way to actually verify that those signs exist is for someone to go there. WhatamIdoing (talk) 05:32, 31 August 2022 (UTC)
    See also James Wood Bush#cite note-FatherGrave-5, St. Michael's Cathedral, Qingdao#cite note-Weig Tomb-46, M-553 (Michigan highway)#cite note-sign2-68, and Interstate 355#cite note-61. None of those have links to photos, and they're all Featured Articles. WhatamIdoing (talk) 05:55, 31 August 2022 (UTC)
    (Incidentally, road signs on major roads can be verified with Google Street View, which is definitely a published source.) Zerotalk 11:08, 31 August 2022 (UTC)
    Indeed, I forgot about Google Street View, and, per good faith, if we have a photo on Commons under a free license from a contributor in good standing, that's not the same as not having any verification, and makes me much more confident the place and the citation are accurate. Andre🚐 13:30, 31 August 2022 (UTC)
    I see these two statements:
    • WP:RS: Content from websites whose content is largely user-generated is generally unacceptable.
    • Andrevan: photo that's accessible on Commons, so easily verified
    One of These Things is not like the other. WhatamIdoing (talk) 15:36, 31 August 2022 (UTC)
    Well, that's another good point, but if anything is making me doubt whether such "cite grave" and "cite sign" should be usable at all. I like the Google Street view thing much better. We don't cite from inside of Wikipedia or inside of Commons uploads. Andre🚐 15:40, 31 August 2022 (UTC)
    That would require a change to the policies, and I think that convincing people to make that change would require a good explanation for why we can trust editors to post information from expensive paywalled sources (which, realistically, almost nobody is going to double-check), or from archived materials that are open to the public (which, realistically, nobody at all is going to bother double-checking), but we can't trust those same editors to post information from giant signs that a million people a month drive past. WhatamIdoing (talk) 16:10, 31 August 2022 (UTC)
    I'm not sure I have a smoking gun of a bad thing coming in due to an unverifiable cite to a tombstone or a road sign. I think it gets back to Donald's point that published sources, are "subject to review by multiple, unrelated readers who have indicated their view of the source publicly." Everyone can go check out a road sign, and probably will. Not everyone will be able to get an invitation to the archive or museum collection that. It's not open all the time, and they can potentially turn people away. Again, it's a bit abstract because I don't have an example of people being turned away from a museum or archive. And I think it's a good argument that maybe I'm worrying about nothing. On the other hand, look at all the hoaxes and bad unverifiable things that have been snuck into Wikipedia. We have to be vigilant and somewhat paranoid when it comes to verifiability, while still assuming good faith. Andre🚐 17:55, 31 August 2022 (UTC)
    Nope. Published sources are available to the public, even if nobody ever reviews them. WhatamIdoing (talk) 00:39, 1 September 2022 (UTC)
    Re-stating pure contradiction in the form of the "current wiki definition of published means anything available," but not addressing the main issue: verifiability. Andre🚐 00:49, 1 September 2022 (UTC)
    Verifiability means that people (not counting the editor who added the information originally) "can" find this information in reliable sources. It does not mean that they "will". Content is verifiable if it's possible, regardless of whether it's done. WhatamIdoing (talk) 04:14, 6 September 2022 (UTC)
    You said it's verifiable if someone can verify it, even if it nobody ever does. But someone should, and if they can't, it can be removed. There are plenty of hoaxes and failed verification cites. That is the concern. Andre🚐 04:39, 6 September 2022 (UTC)
    We usually don't know if someone has checked out a cited source and found it acceptable. We also don't know if someone has tried to verify a cited source, and failed, unless they tag it as 'failed verification', bring it up on the talk page, or otherwise publicly indicate a problem. I do try to check sources occassionally, and will often try to find a replacement source when a cited source has become a dead-link, does not look like a reliable source, or actually fails to support the material in the article. Off-line sources are more of a problem, calling for AGF, although there, if I suspect something about the citation, I will try to see if I can find something in my own library or on-line that either supports or refutes the material. Of course, all of this depends on some combinatiion of time available and my interest in the subject. Donald Albury 16:53, 6 September 2022 (UTC)
    @Andrevan, let's stipulate that someone "should" check all the sources in articles. I estimate that are about 200 articles per autoconfirmed editor who made any edit at all during the last month, and 2,000 articles per active high-volume editor (like you). How many of articles are you going to check all the sources on?
    Of course, it would be nice if everything were triple checked, but I'm really not sure that's the highest priority, and calls for such work often seems to be Somebody else's problem. WhatamIdoing (talk) 02:37, 7 September 2022 (UTC)
    It's a good question, but I think the answer gets back to how Wikipedia works. If we can set up a system that makes it easy to write verifiable articles and verify the sources, the long tail of contributors and proper incentives will help us. People do check sources and tag them as failed verification, or talk on the talk page, I can't quantify the extent to which it occurs, but it anecdotally does. Need I remind us all of Wikipedia:List of hoaxes on Wikipedia? There are multiple 10+ year, some 15+ year hoaxes that were discovered in 2022. And I should add, there's a chance I may check citations that are available on the public internet or through Wikipedia Library, but there's very little chance I can check an archive in London. Andre🚐 02:49, 7 September 2022 (UTC)
    Not just hoaxes. I few years ago I was reading through an article I created 10 years earlier, and saw a mistake, the wrong name for a city. I got curious and looked for who had introduced the error and when. I eventually found that I had made the error when I first wrote the article. (I blame it on a brain fart.) I already knew I am terrible at proof-reading anything I've written in the past six months, but I'm pretty sure I had read over that article more than once over the years. Donald Albury 14:41, 7 September 2022 (UTC)
    It's not difficult to check most sources right now, since most cited sources are already freely readable websites. And yet, it rarely happens in any sort of systematic way outside of FA and GA. My experience of those occasional {{failed verification}} tags is that they either involve a POV pusher trying to get the "wrong" POV out of an article, or someone with subject-matter knowledge who reads something, suspects that it's wrong, and checks the cited source to see whether the unexpected content was just a mistake in their own memory.
    What would the "proper incentives" be?
    Also, I suspect that you would be interested in Meta/Facebook's independent research project: https://verifier.sideeditor.com/ WhatamIdoing (talk) 16:08, 7 September 2022 (UTC)
    @Donald Albury, if the community agreed with the "multiple locations" idea, then we'd have deleted Template:Cite sign long ago. We have a long-standing definition of published. It says "made available to the public", not "made available to the public in multiple locations". WhatamIdoing (talk) 03:52, 31 August 2022 (UTC)
    Actually, I think not all signs are reliable sources. I have seen photos of historical markers used as sources in articles (and I have done so in the past), but I have become aware that historical markers, at least in the State of Florida, are no better than self-published sources, as a private individual or organization submits the text for a historical marker to the state for approval, and then pays for the marker itself and its installation. I assume the state makes sure the sign is coherent and doesn't violate any laws, but I strongly suspect it doesn't insure the historical accuracy, as the text is often based on unpublished material. Other signs I have seen used in WP, either as sources or for illustration, are more or less promotional, and certainly should not be used as sources. Donald Albury 18:03, 31 August 2022 (UTC)
    Hmm, do you think you could dig up a few examples or point me in the right direction to look for this? Because that's a really good example. New York has tons of random historical markers, and I think I read something a few years ago that they are very old in some cases, but they don't want to take them down or replace them because they're landmarked and preserved, and therefore some are not so accurate, or contain errors. Andre🚐 18:09, 31 August 2022 (UTC)
    It's not a good example of a source being "published" or not. Self-published is still published. A self-published historical marker is not only published, but also reliable for certain statements, such as "There's a historical marker about a fishmonger on this building". One might hesitate to use some of them for statements of fact in Wikivoice, like "George Washington slept here in the 18th century", but it's usually fine to use them for statements of fact about the source itself ("There's a historical marker claiming that George Washington slept here in the 18th century"). WhatamIdoing (talk) 00:42, 1 September 2022 (UTC)
    Missing the point here, that one might presume a road sign or a marker placed by a local municipality or the parks department to be reliable, because it was published by those entities, but is probably much less reliable than their other publications. Where is a statement in Wikipedia that you'd need to say, "there's a historical marker about X at location Y"? What article would ever say something like that? Andre🚐 00:50, 1 September 2022 (UTC)
    You (presumably) looked at an FA earlier that says "The only signage present along the route to indicate the highway number was the street signs erected by the City of Marquette Department of Public Works" and another that says "control cities for I-355 while on I-55 are "West Suburbs" and "Southwest Suburbs"." "There's a sign on this highway that says West Suburbs" is not very different from "There a historical marker about X at location Y".
    If that's not close enough for your tastes, then the exact quoted phrase "There is a historical marker" appears in 74 articles today. WhatamIdoing (talk) 06:37, 1 September 2022 (UTC)
    That's fair, but on a cursory review most of them seem to source the existence of the historical marker to normal secondary sourcing. I'll look through and see if there are any examples that are germane to our scenario. I think the highway example is slightly different than the historical marker one, but I concede at least the general point. Andre🚐 14:34, 1 September 2022 (UTC)
  • It should be noted that the original thread of this thread was WT:RS, I had started one here since I feel this is really a verifiability topic, but reliability is of course intertwined in many ways as is WP:NOR. Andre🚐 23:34, 29 August 2022 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:Verifiability/Archive_75&oldid=1116722856"