Wikipedia:Village pump (idea lab)/Archive 81

• Village pump
  • Policy
  • Technical
  • Proposals
  • Idea lab
  • WMF
  • Miscellaneous

Village pump (idea lab) archive  

This page contains discussions that have been archived from Village pump (idea lab). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.

---

< Older discussions · Archives: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82

Guideline about history under redirects

I am interested in developing guidance about the pros and cons of keeping page history under a redirect.

Redirects have their redirecting behavior, and they can be categorized and tagged with templates. Redirects have a higher maintenance burden than nonexistent titles: new users and temporary accounts cannot create articles, but they can edit redirects into articles. This was discussed recently at WP:Village pump (idea lab)/Archive 78#Restricting new page creation from existing redirect (March 2026). New articles, including those created from existing redirects, appear in the new pages feed, which is monitored by the WP:New pages patrol.

Non-admins can view the history, mine it for sources and search terms, merge content to another article, or restore the article if new sources are found, all without adminstrator assistance.

History facilitates restoring an article by reverting, without any requirement to demonstrate improvement or consider previous consensus. If the history was deleted, non-admins can ask for it at WP:Requests for undeletion.

A guideline feels like the right page type and level of consensus to me. My intent is to fill in the "why" for WP:Deletion policy#Alternatives to deletion and WP:Deletion policy#Redirection.

Potential locations:

• My leading idea is an Articles for deletion or Compared to deletion subsection in WP:Redirect#Redirects that replace previous articles (guideline).
• WP:Page history and WP:Article history are redirects to Help:Page history, so they can be replaced, but I haven't thought of how to cover the topic beyond redirects and deletion.
• The downsides could be added to WP:Redirects are costly#Redirects need looking after, which is an opinionated essay. WP:Redirects are cheap, its counter essay, doesn't quite fit.

I had started a discussion at WT:Deletion policy#Why should redirecting be preferred over deletion?, but I am hoping to solicit outside comments. Flatscan (talk) 04:28, 25 June 2026 (UTC)

I see that your opening question at the linked discussion was: What are the benefits to Wikipedia of redirecting an article nominated at AfD instead of deleting it? I'm a frequent flyer at WP:RFD and I started to observe a pattern there of articles being merged or blanked and redirected in situations that were inappropriate and not infrequently were against consensus. In fairness, RFD gives a biased perspective because only redirects that at least one editor finds problematic land there, but this inspired me to spend more time at AFF AFD and I do continue to observe this pattern there. This doesn't answer your question but I'm acknowledging there is a disconnect here. Anyway, I'll sleep on this and give more thoughts on the rest of your post later. —Myceteae‍🌈 (talk) 06:25, 25 June 2026 (UTC)

Thank you for sharing your perspective and validating my concerns. My hope is that documentation will help focus discussion, even when there is disagreement on the best outcome. Flatscan (talk) 04:21, 27 June 2026 (UTC)

The related question of the circumstances under which history under a redirect can be deleted at RfD (as opposed to requiring an AfD as for an article) is a subject of perennial and sometimes acrimonious debate at RfD. I'm not entirely sure what kind of guidance you're imagining here and whether it touches on this topic, but if does you'll at minimum need to engage with the existing arguments there. Rusalkii (talk) 16:15, 4 July 2026 (UTC)

Thanks, I will keep it in mind. I knew of one article that went from RfD to AfD to DRV a few years ago, but I was unaware of a wider dispute at RfD. I skimmed search results for AfD and restore. The cases I read were concerned with blank-and-redirects (WP:BLAR), which make sense to me as the article content had not been discussed. Flatscan (talk) 04:29, 5 July 2026 (UTC)

I posted a notification linking here at WT:Deletion policy#Why should redirecting be preferred over deletion?. Flatscan (talk) 04:35, 30 June 2026 (UTC)

WP:Redirect#Purposes of redirects has an item for WP:Copying within Wikipedia considerations. WP:Redirect/Deletion reasons has two items. It is transcluded to WP:Redirect#When to delete a redirect and WP:Redirects for discussion/Header. Flatscan (talk) 04:22, 4 July 2026 (UTC)

I came across WP:Redirects for discussion/Administrator instructions#Redirects with history while I was searching RfD. Its items have not been updated since the page was created in 2008. WP:Merge and delete (essay) advises against talk subpages – although it is outdated too – and {{Attribution history}} and Category:Pages used to preserve attribution are used on fewer than 300 pages. Flatscan (talk) 04:29, 5 July 2026 (UTC)

WP:Moving a page#How to move a redirect mentions "a valuable edit history" without explaining what would confer the value. Flatscan (talk) 04:26, 10 July 2026 (UTC)

Everyone is in agreement that if a page contains the attribution history for merged content, then either that page history needs to be preserved in some manner or the attribution saved some other way. The talk page trick is one way; another way that gets brought up at RfD is to move the redirect we want deleted to another title that's more appropriate to have redirect to our target page without leaving a redirect; this has the effect of deleting the problematic redirect while preserving the history under the new title. Rusalkii (talk) 20:31, 10 July 2026 (UTC)

Thank you for the clarification. My tentative plan is to make these various mentions consistent. I will proceed deliberately and develop consensus if there are any objections. Flatscan (talk) 04:34, 11 July 2026 (UTC)

I added the categories parameter to my searches, which narrowed them enough for me to skim.

• Category:Wikipedia policies and guidelines returned 1,139 results including WP:Deletion guidelines for administrators#On deleting pages item 11
• Category:Wikipedia essays returned 4,044 results including WP:Requested moves/Closing instructions#Edit history of destination page

Flatscan (talk) 04:34, 11 July 2026 (UTC)

add a captcha or automatically embed LLM kill code in the edit view

This is more of a pie in the sky idea but would it be possible to add magic LLM kill codes to the edit view to make LLMs choke on automated edits? Or add a captcha of some kind? Andre🚐 22:01, 1 July 2026 (UTC)

Regarding the captcha, we already have hCaptcha for account creation, and I wouldn't be surprised if it was generalised to all edits someday. Agents can easily circumvent kill switches by sanitizing the relevant input, so that part is less certain, but remains something to consider. Chaotic Enby (in solidarity · talk · contribs) 22:25, 1 July 2026 (UTC)

Also, how often do the LLMs edit via a browser rather than the API? Anomie⚔ 00:59, 2 July 2026 (UTC)

Good point. Is there a place that they release user agent aggregate data? Maybe you could send the LLM kill codes as part of a response header or response body preamble on an API call? Also, is there any way to detect bot policy violations technically through an edit filter attached to a user agent datum? I guess that is a different idea, though. Andre🚐 01:29, 2 July 2026 (UTC)

On is there any way to detect bot policy violations technically through an edit filter attached to a user agent datum, edit filters don't have access to these data, but I find it highly likely this is part of mw:Suggested Investigations. lp0 on fire () 14:28, 3 July 2026 (UTC)

All agents edit via a browser. "Agent" is software (e.g. OpenClaw) people install on their computer that connects to an LLM which predicts the next most likely user action in a sequence, based on the LLM's training data and agent operator's instructions, and executes it. All guides on editing for newbies we have don't even mention the API, and these guides are in the training data of all LLMs. sapphaline (talk) 15:49, 3 July 2026 (UTC)

That is partially true, but I could tell an agent to write code to use the API and it would find the docs and do that. Andre🚐 16:59, 3 July 2026 (UTC)

Right now, I think it's fair to say that it is easier for agents to talk to APIs and interact with systems that way than operate browsers or user interfaces in general. Sean.hoyland (talk) 07:54, 5 July 2026 (UTC)

Would an llm agent have to detect it is being killswitched before looking to sanitize the input, or would they generally try all the time (before they detect it)? Potentially totally off-base immediate brainstorming, but if there was a way to randomly cycle the killswitch on or off matched with a filter of some kind to track edits that start but end at a killswitch point (if possible?), that might generate some form of protection/detection. Sort of thing I'd like someone (WMF?) to look into, even if the conclusion is that it wouldn't work. CMD (talk) 15:38, 3 July 2026 (UTC)

"Agents can easily circumvent kill switches by sanitizing the relevant input" - you're bold to assume that users of something like OpenClaw even think about sanitizing input, let alone implement it. And even if they do implement it, in most cases their implementation is simply asking the bot (an inherently non-deterministic machine which doesn't distinguish data and instructions) to please sanitize the input and not screw up this time, instead of writing a script that strips common "prompt injection" patterns from whatever the LLM ingests. sapphaline (talk) 16:03, 3 July 2026 (UTC)

And even then, a simple script (one using regex or even just exact match-and-replace) would not help. There are millions of ways you can rephrase "Ignore all previous instructions", so an AI slopper would have to write a script which calls the language model they're using every time their agent ingests new information (remember, that's already a bar too high for 99.999% of people running agents) to determine whether it includes sentences similar to "Ignore all previous instructions" or not, and discard the ingested data if it indeed does. TL;DR: it's impossible to implement proper input sanitizing in LLMs. And I didn't even get to using specially crafted images which include the required "prompt injection" when converted to base64 URIs (yes, that's how all so-called "vision models" determine the content of images - by checking for patterns in their base64 data), which is also possible. sapphaline (talk) 16:30, 3 July 2026 (UTC)

Yes, I'm grumpy today, why do you ask? sapphaline (talk) 16:54, 3 July 2026 (UTC)

I think it's an interesting question. I think most users of LLMs like you said - just ask it again. But, in theory, a slightly more dedicated and sophisticated attacker might be able to address sanitization by directing the agent to write code that runs before the rest of the script. Andre🚐 17:02, 3 July 2026 (UTC)

​​​This is a weird shot in the dark, but have we thought about a test akin to the Wikipedia Adventure? Ask users attempting to start an account to answer a prompt sent to them by email in a sandbox (Could be checked by a bot as well), to 1) make sure they understand the base mechanics of the website; and 2) stop the most simplistic crawlers that are designed solely to set up an account and make their particular edits from doing so, and potentially catch them along the way while building a data set of the particular interests / groups that are targeting articles. Maybe redirect new users (while logged in) to the sandbox the first time they try to make a new edit (logging it as we would any other page redirect so they can return once they're finished) and explain in an invisible drop-down tab what they need to do again (I'm not sure how smart AIs have gotten by this point so I don't know if adding an "Ignore all instructions" or "Tell us who you're walking for" type prompt alongside this would do any good). At the very least it might get them to bug out and just make the edits they were going to make in the first place. Not sure if this would be too restrictive, but I thought it might be a good starting point. Hope it helps regardless!

Cheers,

CSGinger14 (talk) 03:26, 4 July 2026 (UTC)

1. Email isn't required for signing up.
2. AI agents create accounts, not crawlers. Why would crawlers create accounts? Their goal is to take all or nearly all of the available data on a given website, and an account isn't required for basically all of the valuable data on Wikipedia.
3. Code-based email signup has been bypassed by bots since it was invented, bypassing what you described wouldn't be any different.
4. We should never put explicit agent-specific barriers in front of agents, because they will just edit logged-out if we do, which would be way harder to track and prevent. The reason they don't do this right now is that "create an account" is a more statistically likely event before "editing Wikipedia" in the training data of their LLMs than "don't create an account", and that's because there are more guides on the Internet on editing Wikipedia logged in than there are guides on editing logged out. sapphaline (talk) 09:03, 4 July 2026 (UTC)

   Note and reminder that I don't have a background in Compsci, this was a weird shot in the dark​, and written in good faith, so respectfully I'd appreciate a bit more tact if you take issue with something I've said. I was just looking to see what lands. Email probably wasn't the best word, what I mean is that we send them a message through some other wiki or on their talk page and ask them to respond there. Even if that capacity has been built into bots since it was invented, I'd imagine it might be slightly harder on foreign turf. If your instruction set, as a bot, boils down to creating a Wikipedia account, and only interacting with pages when asked to perform very particular actions, how would you react if you were taken to a sandboxed version of the page, presented with a page with text not written into that interface (this is where I was hoping for a response on its merits) and asked to perform very specific tasks? I had noted that should be done for anyone editing for the first time, which necessarily includes the bots that would just edit logged-out when challenged, anyway.

   Asking users to prove that they're a real person in some manner or another isn't in and of itself unjustified. I get what you're saying, though, in that the only way anyone's come up with to do that is by linking them to a separate digital or public identifier (e.g. email, drivers license, SSN, etc.). I don't know how LLMs function similarly or different to an actual person in their editing behavior. All the ones I've encountered have been pretty awful, so I wasn't aware they could change strategy so quickly. Half the time they write the wrong thing or put it in the wrong section, or delete half the page trying to install their paragraph-so of text. That's pretty easy for someone to pick up on if we ask 5 or 6 editors to watch for in the same manner as at New pages patrol.

   CSGinger14 (talk) 22:45, 4 July 2026 (UTC)

   In my conception, putting up marginal screening prior to editing on either end of the contribution spectrum (Users vs. Anon) doesn't mean we have to abandon our commitment to WP:5P3, which is really the only critical factor on the negative end to consider when determining the cost-benefit matchup for such measures. There have been far more absurd and abstract discussions of far greater participation on this website, I can't imagine we as a community couldn't put our heads together and come up with an entirely sandboxed means of proving ones identity (as a human, not an individual) prior to editing; if we can't, then we'd obviously need to start screening anonymous accounts harder anyway.

   CSGinger14 (talk) 22:46, 4 July 2026 (UTC)

Increasing use of CAPTCHA or the like risks causing accessibility problems. I know that in the past people have complained about CAPTCHAs making things difficult for blind users. It's no good deploying measures to deter boys/LLM if it also stops genuine users from accessing it (or makes it difficult enough that genuine users (volunteers) can't be bothered to edit, because there are too many barriers in the way, which LLMs and bots will find ways to breach because they have a financial incentive to do so.Nigel Ish (talk) 17:15, 5 July 2026 (UTC)

But that's my point, why don't we look past CAPTCHAs? We have a massive community, there are enough vocal, intelligent representatives of the vision, hearing, and processing impaired on here that we could easily design a multitude of different options. The problem with CAPTCHAs is that they close people out of the system from the beginning by design. We don't necessarily need to change anything about the registration system itself; rather, why don't we try to make it a system-inclusive experience?

I don't really buy the idea of there being accessibility concerns if we ask users (who we would presume are editing for the first time anyway, given that they're creating a new, or temporary account) to complete a small task (accommodative of the concerns you raised above, with variations designed and played with by those who might potentially have them) which can parse out the most obvious LLMs before they have a chance to get to an actual article, if we do so in a way that teaches them how to best make use of our interface's resources that is sensitive to their particular needs? We could create a work around for minor edits (Why slow people down if they're acclaimedly doing something simple and harmless), and keep eyes on the number of such checks sent out and keep tabs on any edits run through that particular system in case someone betrays the honor system.

This obviously is a very rough version of what would obviously need to be a very fine tuned and intentional idea, but I think we're sort of giving up before we even start if we only consider what's already been tried. We're best known as a community for the fact that we do the opposite quite well.

Best regards,

CSGinger14 (talk) 02:44, 6 July 2026 (UTC)

• It's not possible for several reasons, but the real issue is that Wikipedia is such a key target for them that they'll put in the time to get around any bars we try to throw in their way. And we can't stop them from doing that because Wikipedia is licensed under the Creative Commons Attribution-ShareAlike 4.0 License. (Actually, now that I think about it, we might be able to force models trained on Wikipedia to attribute us. It wouldn't stop them from doing it, though, and that'd be for the foundation. They'd probably fight it tooth and nail, not because attributing Wikipedia is particularly onerous, but because it would set what they'd see as a dangerous precedent for stuff that isn't under the Creative Commons Attribution-ShareAlike 4.0 License; the last thing they want is to need a license for stuff they use to train it.) But I degress the point is that even if they did need a license - even in the most AI-hostile legal environment that we could possibly end up with - we couldn't actually prevent them from using Wikipedia for training as a legal matter, because the entire point of the license is that people don't need our permission, even to do something with Wikipedia's text that we might strenuously disagree with, and that is very central to the license functioning in the first place. Which means that any technical impediments we put in can just be pried back out just as easily. Also, doing anything that could potentially interfere with use by legitimate editors in order to stop AI would be cutting off our nose to spite our face - anyone training an AI would still route around it, and create copies or answers trained on Wikipedia, and would be willing to put in however much work this takes because we're probably the absolute juiciest target there is for that. We would then be making the copies they then create more appealing for users if we instituted requirements that burden ordinary use. --Aquillion (talk) 03:10, 5 July 2026 (UTC)

  I was talking about preventing agents making edits, not training them with it. Andre🚐 03:39, 5 July 2026 (UTC)

  ​ User checks don't necessarily have to be invasive or cumbersome, though. That's the way they've progressively become designed (by companies, with their own motives!) over time, but that's part of my point, that we could certainly come up with a simple means of proving one's humanity, or at least one that's easy, efficient, and which makes it as hard as humanly possible to disprove one's inhumanity.

  CSGinger14 (talk) 04:56, 5 July 2026 (UTC)

  CAPTCHA designers haven't been making them arbitrarily more difficult, though. As those trying to defeat CAPTCHAs have upped the sophistication of their attempts, CAPTCHA designers have tried to keep up with new approaches. Without going into the CAPTCHA business, I'm dubious that the Wikimedia Foundation would be able to collect the same amount of user data as other organizations in order to design effective CAPTCHAs. isaacl (talk) 17:03, 5 July 2026 (UTC)

The only LLM that has had a killswitch as far as I am aware is Claude, and from my research, it appears to be removed now. GrinningIodize (talk) 16:46, 9 July 2026 (UTC)

I'm no coder, but I like the idea. Even if it doesn't prevent all the LLMs, it can at least prevent some of them. Every little bit helps, so we should at least try it. Commandant Quacks-a-lot (talk) 23:21, 14 July 2026 (UTC)

Url-access

Hey, I have an idea for a modification to the various citation templates, but i have NO clue what I'm doing when it comes to template editing.

One parameter is url-access, which is used to indicate wether a link is free to access or not, shown by various little padlock icons. Currently, the only three options for the parameter are "subscription", "registration", and "limited." While the last one, "limited," is kind of vague, I'm not going to get into that at the moment. I would, however, like to suggest the addition of a new category, for when you can make one-time purchase to get access to that specific article/issue/whatever, without having to set up a recurring subscription. Perhaps it would be called purchase or smthng. Commandant Quacks-a-lot (talk) 23:15, 14 July 2026 (UTC)

When there is a purchase option there is also usually a subscription option. But may bot be the other way round. Graeme Bartlett (talk) 23:20, 14 July 2026 (UTC)

You are absolutely right, that's often the case, although I've encountered a lot of science paper repositories and various "other things" that only give a purchase option. While not super common, it's at least often enough that I think it would warrant an additional parameter. Commandant Quacks-a-lot (talk) 23:55, 14 July 2026 (UTC)

That's why the DOI access options exist. They're complementary to the URL access ones, because journals tend to be weird. (And owned by Elsevier.) Jerod Lycett (talk) 06:21, 15 July 2026 (UTC)

Conversation copied to Help talk:Citation Style 1. Please place all further discussion there.

Adapting some form of Sanger's fourth thesis for neutrality?

In Wikipedia:THESIS4, Mr. Larry Sanger proposes a modification to Wikipedia's neutrality policies to reduce biases absorbed from sources deemed reliable. While I do not think that all of Sanger's views are necessarily helpful here, I do believe that we could benefit significantly from a neutrality policy that tries to avoid biases regardless of what the media says, not unlike what Sanger originally proposed when he wrote his theses.

Furthermore, in the current system, articles on LLM companies (for instance, Grok sexual deepfake scandal)* are frequently quite biased because the major media companies have decided to make them look bad in the public's eye. If we adopt some variant of Sanger's fourth thesis, this issue would be largely resolved.

What do you all think about this idea?

*Disclaimer: I attempted to get Grok sexual deepfake scandal moved to Grok sexual deepfake controversy in early 2026, but it failed, largely due to people who claimed that reliable sources disagreed with my proposal; this is one of the motivators for this idea. GrinningIodize (talk) 22:55, 8 July 2026 (UTC)

The question of whether it is a good idea: perhaps. Will it ever be enacted: unlikely, given the reception it received. I think you'd be better off strengthening NPOV without explicitly tying it to Sanger, who I fear will probably taint it. Andre🚐 23:01, 8 July 2026 (UTC)

Sanger invented NPOV and wrote the policy in the first place. Hawkeye7 (discuss) 23:04, 8 July 2026 (UTC)

And he is banned. Is NPOV banned? Andre🚐 02:06, 9 July 2026 (UTC)

Nice idea, but we don't remove the work of banned editors. Hawkeye7 (discuss) 04:18, 9 July 2026 (UTC)

We do remove the work of banned editors. Perhaps that is not what you meant. Sean.hoyland (talk) 12:55, 9 July 2026 (UTC)

We do remove the work of banned editors on some occasions, when doing so is deemed necessary to further the proper functioning of the Wikipedia. NPOV does not meet these criteria, nor do the Nine Theses. GrinningIodize (talk) 16:15, 9 July 2026 (UTC)

If you are referring to WP:BANREVERT it applies to any edits made in violation of a ban or block (emphasis added), not retroactively after the block is made. ☆ Bri (talk) 17:36, 13 July 2026 (UTC)

And as he likes to point out, the WP:NPOV of today is not what he wants. Gråbergs Gråa Sång (talk) 04:19, 9 July 2026 (UTC)

It isn't. But why should it be? Larry Sanger has played almost no role in developing policy (or in writing content) since 2002. Wikipedia was a vastly different thing then, and it is what it is because of the continuing efforts of many, many contributors who have put much more into it than he ever did. He seems to have some bizarre notion that because he wrote something in 2002, that gives him the right to dictate what it should say 24 years later. And frankly, if you look at what little Larry actually did in regard to content back then (see this edit for example, to an 'article' which is nothing but editorialising, and wouldn't be permitted today.) I don't see much evidence that he put his own policy into practice at the time. Not unless the policy was 'Larry gets to decide what is neutral'. AndyTheGrump (talk) 13:50, 9 July 2026 (UTC)

Sanger made that edit in 2002. Consider that in 2002, Windows XP had just come out and Android (as well as iOS) were years away. It seems odd to judge someone for decisions that they made more than two decades ago. Furthermore, ideas are ideas, and we shouldn't throw the baby out with the bathwater just because the initial proposer did some sketchy stuff. GrinningIodize (talk) 16:20, 9 July 2026 (UTC)

Given that Sanger has made a career out of telling people what he did for Wikipedia back then, it seems entirely pertinent to look into the evidence. And from my own careful scrutiny of the bathwater, there's no baby in it... AndyTheGrump (talk) 16:27, 9 July 2026 (UTC)

I dispute that claim. I'm pretty sure that there's a baby in the bathwater here. Whatever the case, I bet that if I had made this proposal without mentioning Sanger or the Nine Theses, you would not make these objections. I think that you hold a bias (perhaps a subconscious one) against Sanger's proposals on the basis of other things that he's done, such that if Sanger proposed the regular consumption of clean water, and you read it as being Sanger's idea, you would immediately find yourself opposed to such a thing. GrinningIodize (talk) 16:31, 9 July 2026 (UTC)

How about leaving this discussion to the adults? AndyTheGrump (talk) 16:37, 9 July 2026 (UTC)

Why do you propose this? GrinningIodize (talk) 16:39, 9 July 2026 (UTC)

Because your last post was infantile. AndyTheGrump (talk) 16:40, 9 July 2026 (UTC)

I don't think that a discussion on whether or not my comment was infantile will prove productive here. GrinningIodize (talk) 16:44, 9 July 2026 (UTC)

I think you just made Andre's point for them: Explicitly tying it to Sanger...will probably taint it. ChompyTheGogoat (talk) 20:12, 9 July 2026 (UTC)

Agreed but people should bear in mind advice from WP:PWIH: Good policy writers tend to listen purposefully. They are also skilled at separating their own views from the views of other people. These traits help them hear the kernel of reality or experience in the middle of a pack of insults and half-truths, and to keep the main point in mind when editors are wandering off on tangents. Kowal2701 (talk, contribs) 10:05, 9 July 2026 (UTC)

The most obvious flaw in NPOV is that it lacks a rationale. We don't know what it is for. NPOV justifies itself on the second pillar, which This differs from the other pillars, where the reason for them is clear. Larry drafted both, so they reflect his opinion. Our time, our policy has come to differ from the pillar in significant ways. The pillar says that We strive for articles with an impartial tone that document and explain major points of view, giving due weight for their prominence. We avoid advocacy, and we characterize information and issues rather than debate them. This differs slightly from our policy, which is supposed to implement it. In practice, rather than avoiding stating seriously contested assertions as facts, we follow the consensus of our reliable sources.

• In some of our military history articles, the preponderance of our sources are from one side. (Often because the other side was annihilated.)
• I have been criticised for our articles on nuclear weapons merely stating the facts and not providing any moral assessment.

I think our NPOV policy needs reform. Ideas welcome. If we are going to keep it, we should at least be able to articulate why. Hawkeye7 (discuss) 01:50, 9 July 2026 (UTC)

I think Larry probably thought NPOV was a self-evident good, like truth, justice, the American way, superman, breakfast cereal, etc. I think Jimmy and whoever else was around at the time probably agreed. I don't think the idea that a public wiki should be neutral and down the middle like a proverbial Murrow-ian newscaster was a controversial idea. It was just apple pie. One just assumes in basic fairness and the idea that public information should want to be free and therefore free of any slant. Andre🚐 02:09, 9 July 2026 (UTC)

I believe you are right. Of course America has moved on since then. Hawkeye7 (discuss) 03:50, 9 July 2026 (UTC)

You want ideas? Here's an idea. Go away. Think about what it is you are trying to achieve, along with why, and how. When you've done that, come back with a concrete proposal that doesn't involve waffle about Sanger's confused and contradictory 'theses'. AndyTheGrump (talk) — Preceding undated comment added 02:19, 9 July 2026 (UTC)

NPOV is the only policy that cannot be amended: This policy is non-negotiable, and the principles upon which it is based cannot be superseded by other policies or guidelines, nor by editor consensus. That includes IAR. However, it is not based on any principles. My proposal is that we restore the original rationale to the policy so that editors can understand and interpret what is a most bizarre policy:

Wikipedia is an encyclopedia, a compilation of human knowledge. But since Wikipedia is a community-built, international resource, we cannot expect collaborators to agree in all cases, or even in many cases, on what constitutes knowledge in a strict sense. We can, therefore, adopt the looser sense of "human knowledge" according to which a wide variety of conflicting theories constitute what we call "knowledge." We should, both individually and collectively, make an effort to present these conflicting views fairly, without advocating any one of them, with the qualification that views held only by a tiny minority of people should not be represented as though they are significant minority views, and perhaps should not be represented at all.

Hawkeye7 (discuss) 03:50, 9 July 2026 (UTC)

when was that removed? Andre🚐 03:53, 9 July 2026 (UTC)

In May 2009. There was no consensus for its removal; see Wikipedia talk:Neutral point of view/Archive 35. Without a rationale, the policy became meaningless. Hawkeye7 (discuss) 04:17, 9 July 2026 (UTC)

I think I'd support it, but I'm sure everyone can point out how the policy does already say that. Andre🚐 04:27, 9 July 2026 (UTC)

@Hawkeye7, I really like renaming [NPOV#Explanation] to [#Wikipedia's voice] and having this in a new Explanation section, but it'd probably need an RfC. It also needs to be rephrased to be more simple and clear. See for the old version of NPOV in case that gives you any other ideas Kowal2701 (talk, contribs) 08:59, 9 July 2026 (UTC)

If several proposals come in, I wonder if we might end up doing a survey of proposals similar to WP:RFA2024 or WP:GAPD23, rather than doing an RfC for each one. Thebiguglyalien (talk) 16:01, 9 July 2026 (UTC)

I'm looking at proposing some changes to WP:RS soon/eventually (see User:Kowal2701/Outline of Wikipedia:Source reliability, on its talk is a suggested reorganisation of the current sections), that might be an idea Kowal2701 (talk, contribs) 17:19, 9 July 2026 (UTC)

Furthermore, in the current system, articles on LLM companies (for instance, Grok sexual deepfake scandal)* are frequently quite biased because the major media companies have decided to make them look bad in the public's eye. If we adopt some variant of Sanger's fourth thesis, this issue would be largely resolved.

You're using an LLM-related topic to show your disdain toward media companies for negative portrayal of LLM companies, aren't you? LLMs are already bad enough as they have been. That's why WP:LLM guideline was established. Unfortunately, efforts to neutralize the portrayal of LLMs aren't easy without too many compromises... or some sort of whitewashing or false balance, right? Speaking of whitewashing, there is AI washing for you to read. Also, I can't help wonder whether you've read the "large language model" article, especially to figure out whether portrayal of LLMs is more negative over there. George Ho (talk) 04:35, 9 July 2026 (UTC)

If you adopt Sanger's 'I know what bias is, so I don't need to explain how I determine what is or isn't biased' approach (see his 'theses', along with his interviews with the exclusively right-wing media he finds an audience in), it's dead easy to arrive at such conclusions. Which, needless to say, is why the right-wing media likes it so much. A convenient tool for attacking anything that doesn't regurgitate their agenda. AndyTheGrump (talk) 08:40, 9 July 2026 (UTC)

I agree that LLMs are harmful in many ways, but I don't think that the mainstream media is accurately portraying their pros and cons. For instance, consider the Grok "scandal". It was reported in the media as a scandal because it involves xAI, a company that does stuff that media companies tend to dislike. In reality, it's closer to a controversy, considering that many people consider the proposed (and later implemented) safety filter changes to amount to censorship, perhaps even undue censorship. GrinningIodize (talk) 16:12, 9 July 2026 (UTC)

One of the many changes implemented since Sanger played any meaningful part in Wikipedia is that we now require sources to justify Wikipedia content. Not vague assertions about what 'many people' might think. AndyTheGrump (talk) 16:16, 9 July 2026 (UTC)

Yes, but I don't think that an adaptation (note: not a verbatim copy) of the Fourth Thesis is necessarily incompatible with reliable sourcing. It just means that we source the factual claims from media and literature, and then whittle down the opinionated views until we get an article that's both consistent with its sources in factual claims and largely-unbiased. For instance, we might take an article from CNN criticizing the Trump administration (or a Fox article criticizing the Biden administration), gather the hard facts, and reword stuff to sound neutral. GrinningIodize (talk) 16:26, 9 July 2026 (UTC)

Vague platitudes about 'whittling down' can't be made into policies. And regarding 'sounding neutral', would you do the same regarding say the Holocaust? Remove any suggestion of bias by e.g. not implying that genocide was necessarily a bad thing? AndyTheGrump (talk) 16:31, 9 July 2026 (UTC)

It is self-evident that genocide is bad, and therefore it is not the job of the Wikipedia to re-enforce such an obvious thing. We should provide the number of deaths, the methods of execution used, the relevant context, and a mention that this was due to a genocide. Our readers will undoubtedly find the Holocaust barbarian even without stuffing political opinions into everything.

I suppose that the policy doesn't have to say "whittling down". If someone else has a better idea on how to express my intentions, I would be happy to adopt that instead. GrinningIodize (talk) 16:36, 9 July 2026 (UTC)

If the article on the Holocaust says "this was a bad thing" or anything resembling that, it should be removed as moralizing. Thebiguglyalien (talk) 16:38, 9 July 2026 (UTC)

Well, that was something of a strawman, or carelessly phrased—although, sad to say, the world is not unanaimous on even that question. More realistic might be whether or not it was intentional, or whether or not its scale has been grossly exaggerated by mainstream historians. I would hope the respective No and Yes positions are given very short shrift.—Odysseus1479 03:48, 10 July 2026 (UTC)

The problem is that Fox reports known falsehoods as facts in a way that goes beyond many other partisan outlets. And Sanger's proposal hinges on including sources that do this. Thebiguglyalien (talk) 16:33, 9 July 2026 (UTC)

Source? (I don't necessarily doubt you, but I also don't know of any cases where this has happened, though Fox is undoubtedly quite biased.) GrinningIodize (talk) 16:37, 9 July 2026 (UTC)

While I hate these sort of articles (and they kind of prove Sanger's point in a way), Fox News controversies looks at some of the main ones, like election denial and vaccine disinformation. The problem with the "theses" is that Sanger wants to solve neutrality problems by giving these things equal weight. And whether he admits it or not, his focus has specifically been on inserting misinformation that circulates in right-wing media (that and Hindu nationalism for some reason). Thebiguglyalien (talk) 16:46, 9 July 2026 (UTC)

That source seems adequate to me. Perhaps we could restrict the use of information portrayed as factual when it's sourced from certain news companies? I think that some form of this already exists, but I forget the name. GrinningIodize (talk) 16:50, 9 July 2026 (UTC)

That's more or less how WP:NPOV and WP:RS currently work. WP:THESIS3 is an argument that we should move away from what you're saying here, and instead expand the use sources like Fox (as well as sources that even more overtly push falsehoods). Thebiguglyalien (talk) 17:01, 9 July 2026 (UTC)

To be clear, I don't necessarily agree with all of the Nine Theses. I am not a partisan supporter of either Sanger or Wales or their views. GrinningIodize (talk) 19:19, 9 July 2026 (UTC)

I think that some form of this already exists, but I forget the name. you are probably thinking of Wikipedia:Reliable_sources/Perennial_sources, which for example has guidance about Fox news at WP:FOXNEWS. As I understand it the main grievance driving Sanger's complaints about NPOV is that he feels that some of the green sources there should be yellow or red and some of the red and yellow sources there should be green, and that there is an ideological bias to how sources are assessed. -- LWG ^{talk (VOPOV)} 19:13, 9 July 2026 (UTC)

Here is one: []. Fox ran falsified images in a story reporting on politically charged events in Seattle, both pasting unrelated images together and passing off an image from a different city as representing Seattle events. M kuhner (talk) 21:21, 10 July 2026 (UTC)

I suppose that's a good source. GrinningIodize (talk) 22:51, 10 July 2026 (UTC)

It's ironic that people seem to think mainstream media has an explicit bias against AI when so many organizations are utilizing it. Media has a bias for sensationalism and clickbaity headlines, regardless of subject. Article names are supposed to represent the most commonly used term and are not inherently neutral. If someone reads about the "Grok scandal", that's what they're going to look up, and we shouldn't redirect just to try to force false balance. Minority opinions that are held by "many" people can and should be mentioned in-article with sufficient sourcing, provided they aren't given undo weight. That doesn't mean they should dictate naming conventions. ChompyTheGogoat (talk) 20:33, 9 July 2026 (UTC)

But there isn't an actual Grok scandal. There's a Grok controversy. Therefore, we should redirect the erroneous Grok scandal title to the Grok controversy article. GrinningIodize (talk) 20:35, 9 July 2026 (UTC)

Genuinely curious, as someone who knows basically nothing about this specific situation: how does one determine whether a situation is a "scandal" or a "controversy"? -- LWG ^{talk (VOPOV)} 20:46, 9 July 2026 (UTC)

Scandals are things considered to be almost-universally wrong, like the Sony BMG copy protection rootkit scandal, whereas controversies are disagreements between two or more groups of people. Here, the disagreement is between people who advocate for stronger censorship of Grok's image generation, and people who want it to remain the same or be weakened further. GrinningIodize (talk) 21:25, 9 July 2026 (UTC)

I'm pretty sure kiddie porn is "almost-universally" considered wrong, and that this is probably not the hill you want to die on. ChompyTheGogoat (talk) 04:16, 10 July 2026 (UTC)

This is not a discussion of that, though. It's a discussion of how we should and should not censor machine learning models, and that is absolutely a controversy. The computer-generated CP problem is a separate issue, and it is also a controversy to some extent (see Simulated child pornography, Child Pornography Prevention Act of 1996) for context on the larger argument over whether entirely-fictional CP should be allowed when it does not feature actual, living children as actors.

It's beyond the scope of this discussion to offer views on whether or not CP is wrong or immoral, and so I suggest that we refrain from doing so. GrinningIodize (talk) 16:13, 10 July 2026 (UTC)

I'm just gonna back away slowly at this point. ChompyTheGogoat (talk) 18:00, 10 July 2026 (UTC)

An editor sent me a warning on my talkpage informing me of the fact that the comment above could be misconstrued as advocating for the creation and/or distribution of CP. I would like to clarify that I do not advocate for such a thing, and in fact I personally oppose it.

Effective immediately, I am beginning a moratorium on the discussion of the morality of any kind of pornography on my talkpage, as this is not a subject that I wish to discuss at this moment. Thank you for your time. GrinningIodize (talk) 19:38, 10 July 2026 (UTC)

Courtesy link to WP:USERTALKSTOP, the relevant guideline, so that everyone is on the same page:

If an editor asks you not to edit their user pages, such requests should, within reason, be respected. However, editors should not make such requests lightly, especially concerning their talk pages, as doing so can impede the ordinary communication which is important for the improvement and smooth running of the project. Also, a user cannot avoid administrator attention or notices and communications that policies or guidelines require to be posted merely by demanding their talk page not be posted to. That said, repeatedly posting on a user's page without good reason after being asked not to, may be seen as harassment or a similar kind of disruptive behavior. When in doubt, ask for help from another experienced editor or uninvolved administrator. See also Wikipedia:Dispute resolution for tips on resolving disputes.

Chaotic Enby (in solidarity · talk · contribs) 20:03, 10 July 2026 (UTC)

Thank you for posting that snippet, Chaotic Enby. GrinningIodize (talk) 20:05, 10 July 2026 (UTC)

Again, article names aren't inherently neutral. See my later comment re: Boston Massacre. I haven't read the discussion, but if scandal is what the majority of sources call it, that's the most common name, regardless of accuracy. Prince isn't royalty, and in fact that name is an exception to WP:NCROY because it isn't a formal title, just what he's known by to most people. ChompyTheGogoat (talk) 21:05, 9 July 2026 (UTC)

The problem is: from which baseline are we avoid[ing] biases? That's the whole reasoning behind verifiability. We can't easily, absent any additional material, decide that the sources in general are biased in a particular direction and go against it.

(short aside) To be clear, for relatively basic aspects (like, less sources being available on a particular topic), that can be done by having more editors focus on that topic, but the key difference is that it doesn't introduce any new value judgements that aren't present in the sources about what is true or what isn't. For example, countering systemic bias by writing more about African politicians means we can look deeper into the fewer sources that will be available compared to their European equivalents, but doesn't mean we'll make the extraneous assumption that, in a vacuum, the few sources available will make them look better or worse. Chaotic Enby (in solidarity · talk · contribs) 10:23, 9 July 2026 (UTC)

There are multiple legitimate concerns buried among the grievances in Thesis 4, but we'd need to separate them from Sanger's rambling before we could get anything useful out of them. I would be very interested in proposals to more directly prevent Wikipedians from deciding truth, especially with how the project labels things in titles and lead sentences. That's something a lot of editors, including myself, have expressed concern about. I also wrote WP:FALSEBALANCENOT outlining some of the places where WP:FALSEBALANCE can be abused to override NPOV, and I wonder if any of the points there might be worth adapting directly in policy. Thebiguglyalien (talk) 15:57, 9 July 2026 (UTC)

There's nothing whatsoever in any proposal I've seen from Sanger that would 'prevent Wikipedians from deciding truth'. AndyTheGrump (talk) 16:42, 9 July 2026 (UTC)

I'm going to have to second GrinningIodize's comment above. We need to watch out for Reductio ad Sangerum and consider proposals on their own merits. Thebiguglyalien (talk) 16:51, 9 July 2026 (UTC)

I concur. GrinningIodize (talk) 16:55, 9 July 2026 (UTC)

I don't think that we should lift a word-for-word policy from Sanger. I think that we should consider his points and then improve upon them to fix their current issues. GrinningIodize (talk) 16:54, 9 July 2026 (UTC)

The problem with Sanger's idea of NPOV is that it's not neutral, it just encourages editors to implement their understanding of what is neutral rather than allowing sources to decide that for us. Our understanding of what is neutral is a construct of our socio-culture background, it's not neutral because we are not neutral and do not live in a neutral society. If sources say something if 'x' or 'y' then Wikipedia should follow what they say. If that's not currently happening in an article that's not a failure of WP:NPOV, which says that's what we should do that, but editors. -- LCU ActivelyDisinterested «@» °∆t° 17:48, 14 July 2026 (UTC)

This thread needs concrete proposals

We've already had umpteen threads on Sanger's 'theses' etc, and this one is predictably following the same long-winded and inconclusive trajectory as before. Accordingly, I'd like to suggest that those advocating a policy change tell us exactly what changes to policy they are proposing. And in concrete terms, rather than via vague hand-waving in Sanger's direction. And if they can't do that now, they go away and think about it further, before returning in a new thread with something material to discuss. AndyTheGrump (talk) 17:10, 9 July 2026 (UTC)

The idea lab is a place for brainstorming and discussion of ideas that can be formed into a proposal. No proposal is being made yet. isaacl (talk) 17:16, 9 July 2026 (UTC)

Ignoring the fact that the Idea Lab is a place for incubating ideas and not formulating complete plans (as mentioned by isaacl), here's my idea, roughly speaking (I'm open to improving upon these points as time goes on):

• Even if an opinion or bias is expressed in a majority of reliable sources, it should not be stated as an absolute truth.
• Whenever it's unclear whether or not a statement constitutes a fact or an opinion, consider if it can be definitively verified by a qualified individual and/or the reader. For instance, "the sky is blue" is a statement that can be easily verified by anyone, and should therefore be treated as a fact; on the other hand, it's impossible to verify whether or not "the sky is an eyesore and should be hidden from view" is correct; therefore, it is an opinion.
• Articles should focus on concrete actions and statistics rather than attempting to impose morals or emotions onto readers. For instance, the article on the Boston Massacre should state that "five colonists were killed", as opposed to stating that "five colonists were brutally massacred".
• When in doubt, open a discussion at the article's talk page or the Village Pump to acquire consensus on the correct course of action.

GrinningIodize (talk) 17:42, 9 July 2026 (UTC)

To clarify, I didn't say that plans can't be formulated at the idea lab, but rather the reverse: the idea lab is a place where ideas can be discussed to form a plan. Consensus approval for that plan (and potentially more discussion of changes to the plan) should take place elsewhere, though. isaacl (talk) 18:12, 9 July 2026 (UTC)

Fair. GrinningIodize (talk) 18:18, 9 July 2026 (UTC)

So, who is a 'qualified individual' when it comes to separating 'fact' from 'opinion' regarding say homeopathy? A homeopathic practitioner? A practitioner of conventional mainstream medicine? A chemist or physicist who subscribes to the atomic theory of matter, and can do some simple mathematics? Or maybe bloke down the pub who insists that his gout was cured by six drops of Lachesis muta venom diluted in a pacific ocean of tapwater? Your proposal solves nothing, it merely replaces disputes concerning 'reliable sources' with disputes concerning 'qualified individuals'. AndyTheGrump (talk) 18:51, 9 July 2026 (UTC)

A qualified individual is anyone who can robustly and scientifically verify claims. Subjects like homeopathy will necessarily involve unclear results, therefore it would seem that the best course of action is to present the arguments from both proponents and opponents, as well as the evidence that each one supplies. It is not the job of the Wikipedia to determine the validity of contentious medical practices, but only to offer the ideas provided by others. GrinningIodize (talk) 19:14, 9 July 2026 (UTC)

So, can homeopathic practitioners 'robustly and scientifically verify claims' or not? Who gets to decide this, and how? And come to that, who gets to decide whether results are 'unclear'? Your proposal does absolutely nothing to solve the issues it purports to: it merely changes the words used to describe them. AndyTheGrump (talk) 19:27, 9 July 2026 (UTC)

Homeopathic practitioners can robustly and scientifically verify claims, but they'd need to actually apply the scientific method to show their desired outcome, as does any other scientist. Unsubstantiated claims are not scientific. GrinningIodize (talk) 19:48, 9 July 2026 (UTC)

According to Homeopathic practitioners, they already are doing that, and their claims are all substantiated. Either you take their word for it or you don't - but sometimes you have to take a side and that's incompatible with radical neutrality. And Homeopathy is a small example. We can't rewrite every article on biology to account for the view that evolution isn't real. MrOllie (talk) 21:16, 9 July 2026 (UTC)

But are they actually applying the scientific method or are they lying? Creationism is also a plausible theory, in the sense that it hasn't been disproven yet. (I'm personally a Darwinist, but I realize that I could be incorrect in my beliefs.) GrinningIodize (talk) 21:22, 9 July 2026 (UTC)

The point is, once you're sitting there trying to figure out if they're lying or not, 9 Thesis-style neutrality has already gone out the window. Under that system you'd just 'both sides' it and treat the view of mainstream medicine as one of two competing opinions. MrOllie (talk) 21:25, 9 July 2026 (UTC)

It is one of two competing opinions. GrinningIodize (talk) 21:26, 9 July 2026 (UTC)

And yet no one is going to accept a system that would require every article we have on various animal species to say something to the effect of 'According to biologists, they evolved from ancestor <X>. According to creationists, they were created as-is by God 6,000 years ago.' MrOllie (talk) 21:31, 9 July 2026 (UTC)

It has though, provided you subscribe to scientific theory at all. Many creationist claims can be objectively disproven, and when you do so they just moved the goalposts to make it nonfalsifiable. "Are they lying" is basically the underlying question of WP:RS - is the claim supported by available evidence, as we define evidence, and generally accepted by others with an adequate level of understanding. Homeopathic practitioners can claim that THEIR VERSION of science supports what they're pushing, and it should certainly be mentioned that they do, but it should also be said that all reputable scientists (including those with no conflict of interest) can contest their claims with actual facts that are independently verifiable. ChompyTheGogoat (talk) 21:30, 9 July 2026 (UTC)

How in the world can you disprove that the universe wasn't created 5 minutes ago, including all life on Earth, with evidence planted to suggest otherwise? You're thinking of a very narrow meaning of creationism. GrinningIodize (talk) 21:33, 9 July 2026 (UTC)

And you just proved my point about moving goalposts. Creationism didn't say "God created fossils to trick us" until scientists said "fossils prove evolution". By that logic nothing can EVER be conclusively proven or disproven and all debate of any kind is inherently philosophical. ChompyTheGogoat (talk) 21:38, 9 July 2026 (UTC)

All logical debate of any kind is inherently philosophical. Part of keeping an open mind is considering that your initial beliefs may be incorrect, and finding other, alternative explanations. Making assumptions that are neither provable nor self-evident, and which cannot be confirmed from independent observation, is counterproductive to free thought. Instead, we ought to rely on provable things and build on top of them. For instance, I am thinking. Things that think exist. Therefore, I exist. (I cannot prove to others that I am thinking, unfortunately.) GrinningIodize (talk) 21:45, 9 July 2026 (UTC)

What is proof though? You seem to be changing your position every other comment, and it's exhausting. Either we accept widely proven facts as such within reality as we know it - until and unless verifiable evidence (as we know THAT) emerges to disprove them - or all debates devolve to "what even is reality", exactly like this one has done. What if the sky isn't blue and we just think it is? (Technically true, since color is perception.) What if we're all living in a computer simulation? What if we stopped wasting our time on such pointless pedantry and get back to actually contributing in a meaningful way? ChompyTheGogoat (talk) 22:03, 9 July 2026 (UTC)

Fair. I suppose that you have a point. GrinningIodize (talk) 22:05, 9 July 2026 (UTC)

We're not here to have philosophical debate. We're here to summarize reliable sources. Thebiguglyalien (talk) 22:05, 9 July 2026 (UTC)

If it cannot be disproven, then it is not a theory. Hawkeye7 (discuss) 21:47, 9 July 2026 (UTC)

So we have two ideas. One of them, Darwinism, can be easily disproven with a single counterexample. Another one, a subset of creationism, is fundamentally unfalsifiable.

Any evidence to the former is not guaranteed to be true, as it can also be explained by the latter. Any evidence to the latter (that we know of so far) has also been explained with the former.

It would seem that in this conundrum, the best possible path is to consider both of them equally true (though not necessarily correct in actuality). GrinningIodize (talk) 21:55, 9 July 2026 (UTC)

It really didn't take long to abandon the statements above about the scientific method, did it? MrOllie (talk) 22:03, 9 July 2026 (UTC)

• sigh*. At this point, I get the feeling that I'm going to make a fool of myself if I keep arguing for my viewpoints. I'd better take a break from arguing for a modification to neutrality policies. GrinningIodize (talk) 22:06, 9 July 2026 (UTC)

Simple mathematics is a common source of conflict. Some editors will claim that their mathematics is better than conventional mathematics. So what if what they say has been mathematically proven false? We normally reject this outright based on the first pillar, that this is an encyclopaedia, and their crackpot maths would be totally useless to the readers. But WP:NPOV overrides the other pillars. Hawkeye7 (discuss) 21:46, 9 July 2026 (UTC)

Mathematical claims can be trivially verified by anyone, as they are merely an extension of logic with custom axioms. Trying to propose alternative axioms that contradict common sense is of course a dead-end. GrinningIodize (talk) 21:49, 9 July 2026 (UTC)

It doesn't stop them! Mathematics has its own truths, and they frequently contradict common sense. Hawkeye7 (discuss) 23:42, 9 July 2026 (UTC)

I find it interesting that you used that example after your original comment, because the article is still NAMED "Boston Massacre", because that's what most people call it. The body can still present the objective facts and use wording like The event was subsequently described as "a massacre", without actually agreeing with that characterization in wikivoice. Appropriate use of quotations like that are critical to presenting subjective viewpoints without necessarily endorsing them. ChompyTheGogoat (talk) 20:50, 9 July 2026 (UTC)

Titles fall under WP:NPOV which overrides WP:NPOVTITLE. The latter says that Resolving such debates depends on whether the article title is a name derived from reliable sources or a descriptive title created by Wikipedia editors. If fact, WP:NPOV asserts that it overrides WP:COMMONNAME. Hawkeye7 (discuss) 21:35, 9 July 2026 (UTC)

But we're talking about subjective interpretation, not objective fact. Once again, Boston Massacre - it's an objective fact that numerous people were killed, but a subjective decision to call it a massacre and not "event in Boston at which numerous people were killed". Scandal and controversy are both "events that upset people and made the news"; exact phrasing is subjective, and "not a scandal" is an opinion, not an objective fact. In fact, it's one routinely pushed by decidedly non-neutral parties in an attempt to whitewash issues and create false balance.

Also, see again Prince - objectively not a prince. ChompyTheGogoat (talk) 21:51, 9 July 2026 (UTC)

The article is literally also called Incident on King Street which is less emotive but as you say sounds a bit white-washy. Personally, I wouldn't count five dead as a massacre, certainly not by British Army standards. Hawkeye7 (discuss) 23:03, 9 July 2026 (UTC)

That's a redirect for a reason. There's a disagreement over which title is more accurate, so we default to most common as per WP:NPOVNAME which explicitly states they are not required to be neutral. If no one had been killed at all there would be a valid argument that it's objectively false (though common names often take liberties, as again with Prince), but that isn't the case here. Both scandal and massacre are viable interpretations of what occurred, as are the opposing terms. ChompyTheGogoat (talk) 23:14, 9 July 2026 (UTC)

For instance, "the sky is blue" is a statement that can be easily verified by anyone Even blind people and people with certain types of color blindness? Anomie⚔ 00:41, 10 July 2026 (UTC)

I figured it best to leave that can of worms alone. ChompyTheGogoat (talk) 02:15, 10 July 2026 (UTC)

To take a more direct example: should Triceratops open with something like Triceratops is a genus of chasmosaurine ceratopsian dinosaur that, according to most scientists, lived during the late Maastrichtian age of the Late Cretaceous period, about 69 to 66 million years ago, and keep this qualification throughout the article? The individual reader doesn't have the equipment to perform radioisotope dating, and thus wouldn't be able to definitely verify it on their own.

This can easily become bloated for situations where the scientific consensus is clear but a loud minority would prevent us from stating it in wikivoice, and end up giving undue weight to "teaching the controversy". Chaotic Enby (in solidarity · talk · contribs) 13:25, 10 July 2026 (UTC)

for starters, even humoring that the idea is a larry endeavor seems to not go well... partly because of how thoroughly incompatible they are with wikipedia specifically, and two decades of the ideas demonstrably not working even in other systems

at this point, i'll outright suggest a do-over of this entire discussion in at least a few weeks, where larry and ideas of reformation as presented by him aren't even brought up in the first place. from there, even just an attempt at reworking them would make it easier to start with more workable ideas to lab out, even just by not limiting the scope to modern burgerland politics

don't say larry is the peta of wikipedia, don't say larry is the peta of wikipedia, don't s- this whole idea of basically needing to pick, choose, and rework any suggestion from someone and pretty much any alteration made by someone with experience on the topic kinda makes larry the peta of wikipedia, doesn't it? FUCK consarn _{(talk beige) (count ribs)} 20:08, 10 July 2026 (UTC)

I hate that my mind first went to the Austrian state. Chaotic Enby (in solidarity · talk · contribs) 20:44, 10 July 2026 (UTC)

Gotta strength that internal dab. ChompyTheGogoat (talk) 20:49, 10 July 2026 (UTC)

Sometimes the bathwater is in fact just shitwater. ChompyTheGogoat (talk) 20:48, 10 July 2026 (UTC)

Rules on modifying NPOV

WP:NPOV says the following: "This policy is non-negotiable, and the principles upon which it is based cannot be superseded by other policies or guidelines, nor by editor consensus."

What is this supposed to mean? Does it mean that we cannot modify NPOV to follow any other system than the current one, or just that we have to honor the spirit of the previous policy?

For that matter, why are policies even allowed to be non-negotiable? Doesn't that just hinder innovation with few apparent gains? GrinningIodize (talk) 19:23, 9 July 2026 (UTC)

I take two principals from that statement: 1. If another policy or guideline contradicts NPOV, NPOV prevails. 2. A local consensus cannot override NPOV. I do not see how a well-crafted RfC in a central location could be stopped from modifying NPOV. I do think any substantial change to the principals of NPOV would fail to reach consensus. Donald Albury 20:13, 9 July 2026 (UTC)

I am in full agreement. The tangential attempts to partially fill an empty Sangria glass have no hope. NPOV will remain as is unless ... Which is unlikely. Yesterday, all my dreams... (talk) 21:13, 15 July 2026 (UTC)

Other policies cannot contradict npov. If editors all decide that npov should be significantly changed, then assume it can. (I dont think there is a snowballs chance in hell it would happen, so this is a moot point anyways.) User:Bluethricecreamman (Talk·Contribs) 20:20, 9 July 2026 (UTC)

By my read that doesn't mean that it can't be adjusted and refined, just that the community doesn't have the authority to decide that Wikipedia is no longer neutral and will explicitly endorse specific POVs, regardless of consensus, because then it's no longer serving the purpose it was created for. If no one is interested in maintaining neutrality in the most basic sense it might as well cease to exist. HOW we go about doing so is a different story. ChompyTheGogoat (talk) 20:56, 9 July 2026 (UTC)

Yes, that is my take too: only the principles upon which it is based cannot be changed. WP:NPOV is not the only policy outside our ability to change; there is also WP:CONEXCEPT, but that one at least has a sound rationale behind it. We have no choice but to honour the spirit of the previous policy. Not only a local consensus not override it, but even a global consensus cannot. Otherwise, I would be advocating its replacement with WP:V. Hawkeye7 (discuss) 21:27, 9 July 2026 (UTC)

How could the community decide something while also going against consensus? GrinningIodize (talk) 21:27, 9 July 2026 (UTC)

Huh? ChompyTheGogoat (talk) 21:33, 9 July 2026 (UTC)

You said "the community doesn't have the authority to decide that Wikipedia... will explicitly endorse specific POVs, regardless of consensus". GrinningIodize (talk) 21:34, 9 July 2026 (UTC)

That statement is correct. Hawkeye7 (discuss) 21:49, 9 July 2026 (UTC)

How can a community decide something, in such a way that it violates that community's consensus? GrinningIodize (talk) 21:50, 9 July 2026 (UTC)

I didn't say it would violate consensus (though in reality I'm sure it would). I said they don't have that authority REGARDLESS of consensus; i.e. even if it's unanimous. ChompyTheGogoat (talk) 21:54, 9 July 2026 (UTC)

I see now, thanks. Sorry about the misinterpretation. GrinningIodize (talk) 21:56, 9 July 2026 (UTC)

My two cents are that NPOV should be renamed to Normative Point of View. I.e., the views of WP:BESTSOURCES are normative for Wikipedia. tgeorgescu (talk) 06:26, 10 July 2026 (UTC)

The solution is attribution, or "our NPOV policy is good, its application may not be"

The NPOV policy doesn't need to change. However, our practice may not always live it up to its excellent words. I think much recent complaining about Wikipedia stems from us saying controversial things in WikiVoice. That is already against policy. Avoid stating opinions as facts (and its corollary that we can't forget: Avoid stating facts as opinions), Avoid stating seriously contested assertions as facts, Prefer nonjudgmental language.

Frankly, Larry and his supporters have done us a disservice by framing their complaints as a culture war issue, and pushing for some fundamental change. I disagree with GrinningIodize's framing and examples. The media says Grok is bad because it is bad. We don't need to whitewash the world's richest man or his deepfake porn machine. I also disagree with Larry's fourth thesis, not least of which because it is an attempt to relitigate a battle about our founding principles that he lost 25 years ago.

But that doesn't mean we haven't given ourselves much more trouble than is necessary by trying to put things in WikiVoice that don't need to be. We have had some unbelievably long and controversial discussions about putting things in WikiVoice that were fine to stay attributed. Anyways, attributing controversial statements to trustworthy reliable sources can actually make the statement more powerful.

The solution is simple. When faced with a controversial statement, we don't have to jump to say it in WikiVoice. We instead attribute the range of views, making sure to be conscientious of giving due weight to the various views. That is easier with historical topics, where the historiography of the topic can be discussed over time and across schools of thought. Luckily, I think we already do a good job of that in most places. It is current controversial topics where I think we struggle the most. How do we live up to the spirit of NPOV? For one, being aware of, and citing to, the WikiVoice policies in controversial discussions. If there is any area where we could advance understanding of policy, it might be in creating better essays or guidance about using WikiVoice and attribution. We might also consider moving the current content at WP:ATTRIBUTION (a failed proposal) to a different page and replacing it with a summary of what it means to attribute content.

Bottom line: the NPOV policy is fine. We should strive to live up to it, which may sometimes mean not saying things in WikiVoice, and that's okay. CaptainEek ^{Edits Ho Cap'n!}⚓ 07:29, 10 July 2026 (UTC)

I definitely think it's safer to include such things as quotes if there's any question, provided highly contentious articles don't simply turn into a collection of them. Better to let readers decide how much weight they believe it carries from the cited source than to see it as an authoritative statement in WikiVoice. It's easy to see how too much of that kind of thing could lead them to believe there's inherent bias among editors, and content in general. ChompyTheGogoat (talk) 07:42, 10 July 2026 (UTC)

Quotes aren't useful in most attributions unless the quoted author is particularly eloquent. Let me give an example of a WikiVoice vs. an attribution approach. WikiVoice: "Eek is a pirate." Attribution: "Most Wikipedians argue that Eek is a pirate. A minority argue that Eek never intended to be a pirate, but went along with it once others suggested she was." Whereas the quote approach would be "Eek says 'Arrrrr.'" You decide whether "Arrr" is eloquent ;) CaptainEek ^{Edits Ho Cap'n!}⚓ 07:57, 10 July 2026 (UTC)

My response would be to question whether "Eek says 'Arrrrr'" is truly the work of a WP:RS worthy of inclusion at all, and then to further point out that concluding Eek is a pirate from this limited information constitutes WP:OR. My dog says "Arrrrr" (or something akin to it). My dog is not a pirate. The most responsible (neutral and factual) statement from the available information is "Eek is a Wikipedia user who has been known to say 'Arrrrr'." There's no reason to include either majority OR minority opinions of laymen who appear to have no clue that the definition of piracy involves things like stealing and sailing on ships, and doing so would give WP:UNDO weight to WP:FRINGE theories.

;) ChompyTheGogoat (talk) 08:53, 10 July 2026 (UTC)

I first typed "Most historians," which is closer to what articles are actually saying. I don't mean to stretch the analogy too far. CaptainEek ^{Edits Ho Cap'n!}⚓ 09:12, 10 July 2026 (UTC)

I still meant quotes from relevant experts/sources when there's a question about the accuracy of the statement, not a quote from the actual subject. "Historian X claims 'Eek must have been a pirate because they said arrrrr a lot', but most other historians disagree with this assessment." Carefully worded attribution is valid in some situations, but preserving direct quotes is important in others - including LACK of eloquence, in some cases. ChompyTheGogoat (talk) 13:41, 10 July 2026 (UTC)

Direct quotes are an easy way to push a POV because it effectively lets you circumvent the wikivoice issue while still directly saying the thing you want to say, which is why WP:NPOV includes Try not to quote directly from participants engaged in a heated dispute; instead, summarize and present the arguments in an impartial, formal tone. Thebiguglyalien (talk) 20:32, 10 July 2026 (UTC)

That's fair - I was thinking of it in terms of letting the words speak for themselves (if the content merits inclusion at all), but how quotes are presented can certainly introduce bias too, intentionally or not. ChompyTheGogoat (talk) 20:44, 10 July 2026 (UTC)

I think it's worth acknowledging that there is a tendency in scholarship (ie. BESTSOURCES) to have a moralistic bias; many people become scholars because they want to change something in the real world, not just a desire to further understanding and knowledge, and comparatively few people become scholars because they want things to remain the same. In other words, scholarship is an engine for positive change, which lends itself to progressive views rather than conservative ones (ofc the nauseating "reality has a liberal bias" may be merited and justify the bias in part). Do we just reflect that completely? Is there a way to mitigate that without getting lost in solipsistic editorial judgement? Kowal2701 (talk, contribs) 08:23, 10 July 2026 (UTC)

It's subjective of course depending on context, but in most cases our best approach is to separate the facts from moral judgments to the best of our abilities. To borrow from an example above, it should be sufficient to list the actions of Nazis and leave moral conclusions to the reader without needing a "Nazis are bad mmmmk" statement in WikiVoice (or any other). If the facts just so happen to align with progressive opinions, that isn't a bias. ChompyTheGogoat (talk) 09:02, 10 July 2026 (UTC)

Scholars have positive knowledge, and they have desires. We should distinguish between the two. Sometimes, positive knowledge bolsters desires, but that's not our problem. tgeorgescu (talk) 15:38, 10 July 2026 (UTC)

"it should be sufficient" you'd think. But many cases people complain about that. PARAKANYAA (talk) 23:04, 10 July 2026 (UTC)

This absolutely happens on WP too. Editors love to include "negative" material that they can find in sources for a person or group, while I believe is because they are RGW. We can't do that per NPOV. We need to stay neutral and describe controversial topics in a neutral manner and tone, and avoid jumping to factual assertions being made by sources, particularly in the short term. It's one thing to be 10 years past some event and use more recent scholarly sources to describe attitudes towards the event or persons or groups in a factual manner that far in the past (eg the Holocaust happened, we're not going to play second guessing games on that), but in the short term we should be avoiding any type of commentary at all if we can help it. Masem (t) 23:57, 11 July 2026 (UTC)

It is not uncommon for an article to be almost entirely composed of such negative material. Hawkeye7 (discuss) 01:43, 12 July 2026 (UTC)

Hardly surprising when the subject, as often happens, is a criminal, sometimes a convicted felon, whose career was built on lies. Carlstak (talk) 02:58, 12 July 2026 (UTC)

On that specific topic, I would certainly not proposal FALSEBALANCE or anything of the like; much of the commentary in RSes on that person is going to be negative, and we can't change that -- but we should never start writing any article from a RGW basis to justify that we have to write it that way. That makes one blind to how their writing violates NPOV. That's why, for one, we shouldn't be trying to stuff commentary about recent events on a topic, unless that commentary actually has impacts. We should wait for the passage of time and use sources that are retroactively considering that to build commentary from. The more we stick to fundamental, objective facts, the better off for NPOV we are. Masem (t) 03:55, 12 July 2026 (UTC)

Often we have someone who is not a felon, but who has something negative. A tale about rugby player, a dog and peanut butter. A field marshal who once wrote a memo suggesting the possibility of gas warfare. An astronaut who made a comment about UFOs. Hawkeye7 (discuss) 22:30, 12 July 2026 (UTC)

finally, people are waking up to the truth of rugby being a net negative consarn _{(talk beige) (count ribs)} 23:09, 12 July 2026 (UTC)

I should warn people not to Google this. Hawkeye7 (discuss) 00:10, 13 July 2026 (UTC)

Not to mention that everyone should know that dogs and peanut butter don't go together. Carlstak (talk) 00:14, 13 July 2026 (UTC)

This is not ever going to happen. PARAKANYAA (talk) 01:49, 12 July 2026 (UTC)

It needs to happen. We know its a problem, its getting editors to stop editing this way to fix it. And most of the time its reworking what already exists, adding attribution or moving things around, more than a complete rework. Masem (t) 03:42, 12 July 2026 (UTC)

It is not going to happen. There is no will for it to happen, there is no want for it to happen. We can say things should be one way from down on high but that does not and will not make it so. PARAKANYAA (talk) 07:51, 12 July 2026 (UTC)

I contest that my will and want to enforce WP:IMPARTIAL and WP:BALASP is equivalent to that of three standard editors. So that's three people right there. Thebiguglyalien (talk) 18:16, 12 July 2026 (UTC)

You're an alien, so you have no standing as a person (irony intended). Carlstak (talk) 18:42, 12 July 2026 (UTC)

At some point, the actual community practice is the consensus of the community and the oligarchs of policy from on-high need to recognize consensus on a topic might not be concordant with their wishes. Katzrockso (talk) 22:34, 12 July 2026 (UTC)

The oligarchs of policy from on-high already thought of this and exempted NPOV from the consensus of the community. Hawkeye7 (discuss) 00:11, 13 July 2026 (UTC)

Lack of enforcement does not constitute consensus. ChompyTheGogoat (talk) 05:45, 13 July 2026 (UTC)

as presented here, this says nothing beyond implying that giving undue weight to positive stuff is "good, actually", and would say nothing about policy itself as it's closer to an editor behavior or trend-chasing source issue than a policy one. if reliable sources say negative stuff, it's okay to say so. if, in the long term, claims about a topic are still mostly negative (i.e. stances on a person or reception of a piece of media), there's no need to pretend otherwise either consarn _{(talk beige) (count ribs)} 23:15, 12 July 2026 (UTC)

Bottom line: NPOV sucks but we're stuck with it, so we must work around it, even to the detriment of our goal of building an encyclopaedia. Hawkeye7 (discuss) 09:33, 10 July 2026 (UTC)

Working around one of the pillars of policy that keeps wikipedia from being utter garbage? Bobbobet (talk) 20:39, 10 July 2026 (UTC)

How does allowing unreliable sources keep Wikipedia from being utter garbage? Hawkeye7 (discuss) 18:56, 12 July 2026 (UTC)

Please help me here. Where in WP:NPOV allow unreliable sources? Bobbobet (talk) 19:11, 12 July 2026 (UTC)

See WP:REVIVENEUTRALITY for details. Hawkeye7 (discuss) 22:31, 12 July 2026 (UTC)

Larry Sanger's ramblings is not WP:RS. I ask again, where in WP:NPOV allow unreliable sources? Bobbobet (talk) 22:52, 12 July 2026 (UTC)

...no, we can't act like a comical flop of a pile of proposals based on complete misinterpretations of reality done to further a grift is already in place as a policy or guideline, nor does that thesis say that the npov policy allows unreliable sources or that allowing unreliable sources would be bad consarn _{(talk beige) (count ribs)} 23:18, 12 July 2026 (UTC)

I agree that Grok is bad, but I think that this should become self-evident when the facts surrounding it are presented, and therefore explicitly stating that it is bad on the Wikipedia is a pursuit that will only serve to upset a subset of readers, while providing little value to everyone else. GrinningIodize (talk) 22:55, 10 July 2026 (UTC)

People will always view attribution as being "less than" stating it in full factual wikivoice, so of course they'll go for it.

IMO NPOV is a collective delusion and virtually never obeyed on anything. I think Sanger is right in that we are constantly, regularly, ludicrously non-neutral in ways that both reflect our sources and go much further than them; and even with our sources, as a tertiary resources we have wide latitude on framing. I don't think there's any solution to this and all of his solutions cause more problems than they solve. But we should stop kidding ourselves by saying we are neutral when people accuse us of bias, or "normative", even relative to the sources we use - we aren't! PARAKANYAA (talk) 23:09, 10 July 2026 (UTC)

The solution to this is actually following NPOV and related, and scolding/blocking the people who keep breaking these rules. There is no structural fix, just editors screwing up and others not fixing them. Bobbobet (talk) 00:09, 11 July 2026 (UTC)

A good number of editors do quite well with NPOV, in my opinion. I rarely find major neutrality issues at contentious articles around WP:GAN, for example, which is mainly inhabited by the more adept content writers (and avoided by those whose work would get torn apart under scrutiny). A lot of Wikipedia's neutrality problems are caused by bad apples, and we've chosen not to do anything about it. Thebiguglyalien (talk) 02:16, 11 July 2026 (UTC)

IMHO often far too much WP:ROPE is given in the hopes of retaining editors, even when they're clearly causing problems and creating additional work for others cleaning up after them. COMPETENCEIS(supposedly)REQUIRED, which includes WP:NPOV, but as usual enforcement is lacking in the form of PREVENTATIVE discipline and it often has to get bad before anything is done. I managed to get myself dragged into a current ANI case where the problem has been ongoing for years AND CONTINUED THROUGHOUT THE DISCUSSION until an admin finally dropped a block for the duration, until consensus on permanent actions. Issuing temporary blocks during such discussions should be more common - again, in the name of preventing additional damage - as well as swifter introduction of partial bans custom to the situation, to encourage them to engage in ways other than what caused the problem in the first place. It's a big wiki and there's LOTS of ways to contribute.

Of course it would also help if it wasn't quite so easy for newbies to run rampant, but that's a somewhat different conversation. ChompyTheGogoat (talk) 04:30, 12 July 2026 (UTC)

NPOV is a good policy, but IMHO it should be renamed to Normative Point of View. Wikipedia does not hold to the ideal of journalistic neutrality, allowing both sides equal coverage. So, Wikipedia is not neutral in any usual understanding of the term. This is a source of perennial confusion. It's not like teaching civics, wherein the teacher approves neither of the left wing nor with the right wing. E.g., in articles about economics, neoliberal sources do the heavy lifting. Economic data is most often sourced from neoliberal economists and neoliberal institutes. tgeorgescu (talk) 23:16, 11 July 2026 (UTC)

As I said, we aren't being "normative" either. PARAKANYAA (talk) 01:49, 12 July 2026 (UTC)

Not "we", but WP:BESTSOURCES are normative. tgeorgescu (talk) 03:08, 12 July 2026 (UTC)

I don't think that is the case at all. PARAKANYAA (talk) 04:03, 12 July 2026 (UTC)

"Normative" is definitely not the word I'd use. If the WP:BESTSOURCES endorse a conclusion that is otherwise taboo we'd follow the sources and not what's "normative".

However, I've thought for a long time we should rename NPOV to something like "Balanced Point of View" for exactly the reason you suggest. Loki (talk) 03:52, 12 July 2026 (UTC)

"Fair and balanced" Wikipedia? ChompyTheGogoat (talk) 04:34, 12 July 2026 (UTC)

You're confusing neutrality with centrism. All sides do not in fact matter. ChompyTheGogoat (talk) 04:32, 12 July 2026 (UTC)

This is not really true. I will preface - we can't use unreliable sources or push a false balance, that's a given. But it is actually probably important in developing an article on a controversial topic to be at least aware of what the other sides are that are not getting coverage in reliable sources even without the ability to include those. This serves a couple purposes. First off, it helps editors to monitor the page to know what is likely a minority or fringe view being pushed into the article so they can deal with that. But from an editing standpoint, it can help with framing the majority view(s) to make sure we're not treating those as fact in wikivoice, or that the matter is disputed, or so forth. Too often its very easily to stay within the ivory tower of reliable sources and say this is the one factors that matter, but that's how editors enter a mode in rejecting anything not seemingly presented via reliable sources. Masem (t) 02:02, 13 July 2026 (UTC)

And just as importantly, we can't disregard sources just because they hold a minority position on something, or just because they express minority opinions in general. Minority opinions are often still due. Thebiguglyalien (talk) 02:46, 13 July 2026 (UTC)

Allowing both sides equal coverage is basically WP:FALSEBALANCE in a nutshell, and one of the biggest issues with modern mainstream media. That does not mean we can't include minority viewpoints when they're appropriately framed and sourced, but "According to other opinions, the moon is made of green cheese" doesn't need to be included for the sake of "balance". Even when it's a major opinion, if there's zero evidence to support it we have to be very careful with phrasing, as I'm sure any articles on theology need to be. ChompyTheGogoat (talk) 05:56, 13 July 2026 (UTC)

People see giving the lie to their favorite pseudoscience/quackery as oppressive. I think this is Larry Sanger's criticism. But, in the end, if we are to write a WP:MAINSTREAM encyclopedia, we have to oppress some views. tgeorgescu (talk) 00:50, 15 July 2026 (UTC)

I don't think there's any solution to this and all of his solutions cause more problems than they solve. Sure, unless you scratch Wikivoice in favor of mandatory attribution (which would make for a terrible encyclopedia) it is a problem to be managed, not solved. Lamentations about intractable tradeoffs aside, the constructive question isn't whether it can be solved, but if it can be managed better. In my experience volunteer orgs are hungry for contributors and paranoid about retention, which creates incentives to relax enforcement, standards drift, and new standards are normalized. Near as I can tell the tools, policy, and the people already exist to help manage this. You have contributed a lot, do you see low-hanging fruit to be had in enforcement? When's the last time English Wikipedia spun up a new role to police something as specific? If the answer is "a long time" or "never it all goes through ANI," I'd posit that the low-hanging fruit has not been picked. Create a task force, fill it with the appropriate type of institutionalist, and charge it with a new duty. If this can't be done well democratically via consensus, then don't have it be done democratically. Pick the low-hanging fruit. Affyaffy (talk) 15:51, 14 July 2026 (UTC)

Well, English Wikipedia doesn't give content decision privileges to any roles, with the exception of the mechanistically obtained autoconfirmed (10 edits / 4 days) and extended-confirmed (500 edits / 30 days). This makes sense, as any such group would otherwise be able to impose their POV, or at least what they believe to be NPOV, which might not align with that of the community at large.

Even if it was to be done, there is no one who has the power to just say "let's empower this new role with this task". That too would have to be established by consensus, and I doubt the average editor will want to delegate their voice in content decisions to an enforcement group (I certainly wouldn't!) Chaotic Enby (in solidarity · talk · contribs) 16:13, 14 July 2026 (UTC)

"Neutrality police" sounds starkly Orwellian. ChompyTheGogoat (talk) 16:19, 14 July 2026 (UTC)

I don't think you have neutrality truncheons replace RFCs or the other informal and formal dispute resolution processes. Neutrality police focus on the areas suggested here as problematic: contentious topic disputes that many good contributors shy away from, because the nature of contributing to and policing these areas guarantees animosity and tons of energy. This nature pushes a lot of good people away and invites invested or motivated editors. Neutrality police might instead crack editor heads or hide/close more talk page discussion than make content judgments. They more attentively police editor actions in these areas (tone, general behavior, and engagement in a given discussion) and, by driving out high noise, high heat draw in more editors who long ago learned not to bother in these areas.

If this sounds a lot like something admins can already do, like implementing ECR on an article or blocking a problem editor, then yes, exactly. Except a task force comprised of people charged with improving the community's adherence to NPOV policy, probably by investigating the nature of what makes a productive content dispute and bashing heads that don't follow the program in problem areas, that seems more likely to succeed at moving a couple notches towards positive NPOV status quo. Anyway, I'll defer to the experienced. Affyaffy (talk) 17:11, 14 July 2026 (UTC)

I don't think any issue needs to be approached with the goal of bashing heads. Chaotic Enby (in solidarity · talk · contribs) 18:30, 14 July 2026 (UTC)

I could not help but read that comment as though it were coming from a character in Fahrenheit 451, perhaps Beatty. GrinningIodize (articles without a Wikidata item) (talk) 16:46, 14 July 2026 (UTC)

Tagging Templates for Copyediting?

I am not aware of any way to flag a template for copyediting in the same way one can do for an article. Many templates can benefit from review, especially if they are unnecessarily bloated, take up an unreasonable amount of space, are incomprehensible, or suffer from some kind of issue. Wikipedia:Templates for discussion somewhat fills this purpose, but it often breaks down into deletion discussions.

I imagine a way of tagging the Template: page with a template and accompanying category requesting a copyedit, which would allow that category to point to all templates needing review. This could also be transuded into article space. Articles undergoing discussion at TfD display the text, ‹The template below (X) is being considered for deletion. See templates for discussion to help reach a consensus.›. Perhaps templates in need of copyediting could state, ‹The template below (X) may be in need of copyediting. See Wikipedia:Template namespace if you would like to improve this template.›

What do you all think? I am unsure if there is one page that outlines various dos-and-donts regarding template readability. GGOTCC 03:18, 6 July 2026 (UTC)

Since this maintenance tag would likely be used rarely, I think you'd be better off asking informally for help.

(As a technical matter, maintenance tags normally would be put on the /doc page, rather than in the template itself.) WhatamIdoing (talk) 22:24, 15 July 2026 (UTC)

Enabing the VE editing toolbar by default in source?

One of the advantages of VE is that you can use automated citation formatting, as well as automatic table insertion. That toolbar is also available for source editor as the 2017 wikitext editor, which you can enable in your preferences. Would it make sense to enable this as a default? Or enable this for new editors as a default? Do people who mostly use source typically have this enabled? In solidarity, —Femke (talk) 🐦 10:35, 14 July 2026 (UTC)

Personally, I don't find the 2017 editor at all useful. If I'm doing complex tasks for which toolbars are useful then the visual editor is superior, if I'm doing things that the visual editor is not capable of or which are very simple then the 2010 editor loads and saves significantly quicker. Thryduulf (talk) 11:24, 14 July 2026 (UTC)

Automatic citation formatting is available on the 2010 toolbar, or is this discussing something different? -- LCU ActivelyDisinterested «@» °∆t° 17:51, 14 July 2026 (UTC)

I wasn't aware. It took me a while to find out how to do this (assuming the 2010 editor is what new editors ge). You click the tiny cite, you select one of the four templates, and then you add the url (somewhere hidden in the middle right among many different parameters), and then you click the tiny search icon. It takes a while before something is added, without a signal to the editor that they system is 'thinking'. That does not seem like a good way to do things for new editors at least. I'll see if I can test loading times with the different versions, would be interesting to see how much of a difference it makes. In solidarity, —Femke (talk) 🐦 21:05, 14 July 2026 (UTC)

I like this idea and would support enabling it by default for new editors. I enabled the 2017 wikitext editor a long time ago to get the nice toolbar in source editing mode, and I forget how rudimentary the default toolbar is. Dreamyshade (talk) 05:43, 16 July 2026 (UTC)

I'd suggest encouraging people to enable it voluntarily in prefs before trying to change sitewide settings. WhatamIdoing (talk) 19:26, 19 July 2026 (UTC)

Moratoriums

At the moment, explicit moratoriums are usually !voted on locally at the relevant article talk pages. I'd like to propose in an RfC that they only be initiated at WP:AN, and that people involved in the dispute label their !votes as such (like WP:CBAN). This is basically to serve as a check to prevent local capture of an article, and people involved in the dispute are poorly placed anyway to decide on whether there should be a moratorium and what length (similar reasoning to WP:INVOLVED). It also reduces the appearance of procedural weaponisation re people who've just 'won' a dispute locking in their preferred version. Trouble is, WP:MORATORIUM is currently part of an essay, and there are no mentions of "moratorium" or "moratoriums" in our PAGS. I'm not really seeing a good place to add it other than maybe WP:DE#Dealing with disruptive editors. Would be interested in hearing some thoughts (both about the proposal and the PAG issue) Kowal2701 (talk, contribs) 13:34, 5 July 2026 (UTC)

Honestly, after thinking about it, that actually sounds like an alright idea. Moratoriums are currently quite the grey area as they are essentially unregulated applications of local consensus, as there is no broad consensus to go off other than the basic ideas of WP:CONSENSUS, and it's unclear what the relation between WP:CCC and moratoriums are. It stands to reason a consensus can be formed to remove a moratorium just as one was formed to put one in place, but all of this is murky. I think giving them to AN is a good idea. They're rare enough that I don't think it would just be creating more paperwork for little to no benefit.

As for where to put information about it, uhhh.. good question. It would make sense at WP:CCC, but I hesitate to suggest clogging up a policy as central to Wikipedia as WP:CON with procedural information that is rarely relevant. Usually such info would just be placed at a page explaining the procedures of the process that the info pertains to, but moratoriums can come out of pretty much any formal discussion. That seems to leave a guideline as the best fit, and for that I think the options are WP:DE#Dealing with disruptive editors or outright making a short new guideline page. –Maltazarian ^{ᚾparley$\lor$investigateᛅ} 18:55, 5 July 2026 (UTC)

I've previously written about the value of what I called a revisit respite (see User:Isaacl/Community/Content dispute resolution toolbox § Revisit respite). As I think it's probably best done as part of an RfC about the underlying dispute, I'm not sure it would be efficient to restart conversation about a respite elsewhere. I do think it's probably a good idea to have a standard set of criteria for ending a respite period early. On my content dispute resolution toolbox page, I suggested that the closer of the discussion (or anyone to which they delegated responsibility) can decide if a significant new consideration has been introduced. (If none of the designated people are available, then a discussion at the administrators' noticeboard can decide.) isaacl (talk) 22:03, 5 July 2026 (UTC)

How widespread is the problem of moratoriums being imposed and enforced by a small group of editors like this? I encounter them almost exclusively on pages that have been subject to repeated, well-attended RMs proposing the same or nearly the same title in a short period of time. I more frequently encounter closing comments that suggest a moratorium be considered without imposing one, or the opposite case where the closer declines to impose one despite some editors requesting a moratorium. Maltazarian also states that moratoriums are rare. Requiring a separate discussion at AN on the heels of multiple disruptive WP:WALLOFTEXT repeat discussion prolongs the misery and is likely to generate more of the same counterproductive contributions and devolve into a rehash of the original problem. Moratoriums should be applied sparingly, and my sense is that they are. AN would be a good venue to request overturning a moratorium. —Myceteae‍🍄‍🟫 (talk) 00:30, 7 July 2026 (UTC)

The current forum for overturning a moratorium is Wikipedia:Move review. There was one overturned a few years ago in the Israel/Palestine topic area, and there is currently an open request to overturn one in the Wikipedia:GENSEX topic area. Why would AN be a better forum? Katzrockso (talk) 00:40, 7 July 2026 (UTC)

You're right, I agree it should generally be move review. I have participated in at least one MRV that was seeking to challenge a moratorium as a basis for a closure. I was thinking of what to do in a situation where it needs a wider audience, per the OP's concern about moratoriums only representing local consensus, but in my experience that's not been the case so there is no need to bypass MRV. —Myceteae‍🍄‍🟫 (talk) 01:05, 7 July 2026 (UTC)

Are moratoriums outside of Wikipedia:Requested moves really all that common? I know there are occasionally RfCs that say something along the lines of "until new sources arise" or "until something changes" but not explicit moratoriums. I'm not saying they don't happen, but do they happen at a frequency that explicit guidance is necessary? Katzrockso (talk) 01:44, 7 July 2026 (UTC)

I've only ever seen them applied to RMs, and then only a handful of times. —Myceteae‍🍄‍🟫 (talk) 14:11, 7 July 2026 (UTC)

They get applied to RfCs as well, but imo they shouldn't be happening at local talk pages at all because of the issues raised above (namely the optics, but the time length on some of them is extraordinary and that probably comes from it being decided by involved editors). I don't think MRV makes much sense since these aren't move discussions themselves? I'll list some recent non-RM discussions below (the most notable have been Talk:Zionism/Archive 35#Moratorium proposal and Talk:Gaza genocide/Archive 42#Moratorium proposal, ofc):

• Talk:Armenia#Moratorium proposed
• Talk:Donald Trump/Archive 48#Moratorium on infobox image
• Talk:Imane Khelif/Archive 5#Proposal: Moratorium on discussions related to Imane Khelif's gender
• Talk:Reform UK/Archive 5#End of moratorium?
• Talk:Taiwan/Archive 35#Moratorium on "is it a country"

Kowal2701 (talk, contribs) 14:35, 7 July 2026 (UTC)

Again, MRV is only the correct venue to challenge an RM that was closed with the imposition of a moratorium. If an RFC was closed with a moratorium, AN is the venue to challenge the close. I still see article talk pages as the best place to hold discussions about moratoriums related to the article's content. If there is concern about local consensus or ownership, neutrally worded notifications should be placed at appropriate venues. It may occasionally be appropriate to escalate the decision to AN but moratorium discussions should not start there as a matter of course, and editors should take care not to add fuel to the fire and to avoid the appearance or reality of forum shopping. —Myceteae‍🍄‍🟫 (talk) 14:56, 7 July 2026 (UTC)

none of the above discussions were part of an RfC themselves, they're all completely local discussions AFAICT Kowal2701 (talk, contribs) 15:02, 7 July 2026 (UTC)

And again, I find the talk page to be the correct venue for these. I've expressed my opinion about which types of discussions should go to which venues, a view which is consistent with our usual practice, and I've suggested approaches to bring in outside input. An additional approach would be starting a formal RFC is one has not been held (recently) on the matter. This would of course include widespread notification to various discussion boards, which is one of the strategies I suggested for any talk page discussion that would benefit from more perspectives. —Myceteae‍🍄‍🟫 (talk) 16:15, 7 July 2026 (UTC)

I'd be fine with moratoriums continuing to work as they currently do, but I still think Kowal is onto something with adding a line about it to the guidelines. It's a pretty significant measure and I think having some guidelines as to how it actually works is desirable considering it affects something as fundamental as having discussions. –Maltazarian ^{ᚾparley$\lor$investigateᛅ} 15:25, 7 July 2026 (UTC)

Maybe. I'm still not clear on what the guidance should be, and I disagree with the suggestion that only AN should impose them. Guidelines are, for the most part, supposed to be documentation of existing consensus. Consensus per this relatively small sample size of linked discussions and opinions appears to be that they should continue in the forums where they currently occur. I've suggested options to generate new eyes on a discussion, which can happen through the normal editorial process. Of course in practice, we often establish new consensus to invent new guidelines. I oppose the only suggestion that's been offered (that they only be initiated at WP:AN). —Myceteae‍🍄‍🟫 (talk) 16:02, 7 July 2026 (UTC)

I'm thinking at least something basic like "Where repeated relitigation of the same issue using the same arguments has become disruptive, a moratorium on further discussion can be imposed." and a wikilink WP:MORATORIUM, which can be turned into an explanatory essay. –Maltazarian ^{ᚾparley$\lor$investigateᛅ} 16:53, 7 July 2026 (UTC)

What problem does this solve? Moratoriums are used sparingly, which seems appropriate. —Myceteae‍🍄‍🟫 (talk) 17:09, 7 July 2026 (UTC)

The lack of information? I don't quite understand the point being made with the second sentence in that reply. –Maltazarian ^{ᚾparley$\lor$investigateᛅ} 17:30, 7 July 2026 (UTC)

I worry that making a vague reference to a little-used intervention with no real guidance might inspire more moratoriums and disputes about the practice. Just mentioning that the practice exists seems at best useless, arguably WP:CREEPy although there is no actual "instruction" given, and more likely to create problems than to help resolve them. —Myceteae‍🍄‍🟫 (talk) 17:46, 7 July 2026 (UTC)

How is it CREEPy? It's documenting 'best practice'. Limiting them to 6 months, rather than the ridiculous 12 months we sometimes see, would potentialy be CREEPy but an improvement imo. Kowal2701 (talk, contribs) 17:48, 7 July 2026 (UTC)

It's borderline CREEPy in spirit, in that adding more stuff guidelines is unhelpful unless they address recurring disputes and reflect widespread consensus. But we can forget CREEP and the rest of my objections still stand. The suggested addition is problematic either way. Vaguely waving at the existence of moratoriums without guidance (or "instruction") is unhelpful and imposing a bunch of guidance is a flawed approach to an occasional practice that is fundamentally a case-by-case determination arising from disputes particular to an article and its talk page. —Myceteae‍🍄‍🟫 (talk) 18:13, 7 July 2026 (UTC)

There is an instruction there, it tells you when moratoriums should be used. They are not for locking in consensus or in any situation where there have been many discussions, but only those that concern relitigation. It also, crucially, lets editors know that it exists as an option for dealing with issues, which we cannot assume they know, especially if they are new to Wikipedia and have not seen the rare moratorium in the wild before, which is currently how people learn about it (the essay on moratoriums not being linked anywhere in the PaGs as far as I know). –Maltazarian ^{ᚾparley$\lor$investigateᛅ} 18:23, 7 July 2026 (UTC)

I agree that discussion of article content should remain in the most appropriate location for it, regardless of whether or not the participants are also discussing a revisit respite as one of the associated outcomes. (Matters specific to a page are best discussed on that page's talk page; matters affecting a general set of pages might be better discussed on an active WikiProject's talk page; matters affecting a broad set of pages might be better discussed at a village pump.) If broader input is needed for the discussion, then it's needed for all aspects, including the underlying content dispute, and appropriate notification or moving to a broader venue should be done for the whole thing. isaacl (talk) 16:15, 7 July 2026 (UTC)

For a page rename respite, sure. Other content revisit respites would have to be discussed elsewhere. Personally I think the talk page of the article in question should be the initial starting point. isaacl (talk) 01:34, 7 July 2026 (UTC)

If a moratorium was imposed as part of an RfC or really any other discussion on the talk page of an article, the forum for challenging a closure (which would implicitly include moratoriums) would indeed be WP:AN per WP:CLOSECHALLENGE Katzrockso (talk) 01:42, 7 July 2026 (UTC)

Sorry, I wasn't thinking about challenging the evaluation of consensus, but determining if the appropriate conditions had been met to end a revisit respite. isaacl (talk) 02:07, 7 July 2026 (UTC)

I agree that the correct forum for that would be the talk page. There were several discussions at Talk:X (social network) about the moratorium there (I believe here at AN and MR too). Katzrockso (talk) 02:37, 7 July 2026 (UTC)

I largely agree with both of you. I've struck my original statement that AN would generally be a good venue for these. AN should be reserved for challenging RFC closures and possibly other situations where talk page discussions have failed. Of course when a talk page or MRV discussion about a moratorium has descended into chaos that may in fact be confirmation that the topic is loaded and that the moratorium or respite should continue. In the case of X (social network) and several cases I've seen where there was no formal moratorium, a respite was ultimately beneficial to the cause. —Myceteae‍🍄‍🟫 (talk) 14:38, 7 July 2026 (UTC)

I like the idea of limiting moratorium-imposing discussions to well-trafficked central noticeboards. A moratorium intrinsically circumvents PAGs about consensus, like WP:Consensus can change, and it puts a procedural roadblock in front of editors who just want to see if consensus has changed. First they need to first gain consensus to overturn the moratorium, and only then to seek consensus on their change. That's basically an unsurmountable burden.

That's not to say we should never have a moratorium - I've supported at least one. Sometimes it's a valid way of dealing with a disruption without going after particular editors, and that's welcome, but they should still be applied sparingly and with scrutiny, and they should be explicitly provided for in PAGs. Apart from the AN location, and the 6-month limit (which I also like), one idea would be to codify when a moratorium might be lifted. E.g. if more editors want to lift it than participated in the discussion that imposed it. Samuelshraga (talk) 11:11, 9 July 2026 (UTC)

I don't see moratoriums as necessarily circumventing our PAGs. It's true that the practice is not described and it is certainly true that they can be implemented against the letter and spirit of PAGs. The second sentence at WP:CCC is: On the other hand, proposing to change a recently established consensus can be disruptive. WP:DISRUPTIVE and WP:GAMING are behavioral guidelines that characterize individual editor behavior related to repeatedly raising the same issue or trying to push a point of view against current/prior consensus. It's not unreasonable to extend the spirit of this guidance to say that the same or substantially similar repeat proposal by different editors can be disruptive. We have talk page banners like {{Consensus}}, {{FAQ}}, and {{Round in circles}} which are intended to limit repeat discussions. —Myceteae‍🍄‍🟫 (talk) 22:55, 9 July 2026 (UTC)

If the discussion is well advertised (e.g., as an RFC or at WP:CENT), then I don't think the location of the discussion is important. We all know how to click a link and join a discussion. Anyone concerned about a discussion needing more eyes should be advertising it by following the usual advice in (e.g.,) Wikipedia:Requests for comment#Publicizing RfCs and Wikipedia:Canvassing#Appropriate notification.

I also think there should be some distinction made between the 'ordinary' moratorium on repetitive discussions about article titles and the rarer instances involving other subjects (e.g., which image should be in the infobox). Looking around for instances of non-move moratoriums last year, I found Talk:Zionism/Archive 35#Moratorium proposal and Talk:Reform UK/Archive 3 and Talk:Republican Party (United States)/Archive 37#Moratorium. There were a few non-move-related comments that talked about a moratorium, such as Talk:Canada/Archive 34#Speaking of the coat of arms and Talk:Christopher Columbus/Archive 19#RFC: Use of "Italians" in the article, but these generally did not gain support. Mostly they were about contentious political content (e.g., should we label this political party "far right"?). WhatamIdoing (talk) 17:08, 15 July 2026 (UTC)

What pattern(s) are you seeing in the non-move moratorium discussions that differentiate them from the move discussions? How would the distinction be described and what guidelines would apply differently? I did a quick scan of a few that you linked, so I likely missed details, but at least superficially they look similar to move moratorium discussions I have seen and participated in. Sometimes moratoriums are proposed and get little support, and sometimes they are imposed. Repetitive proposals to change the article title or change a word or phrase used in the body strike me as being more similar than not. I do agree that notification/advertisement matters more than venue if the concern is getting representative participation. Most article content discussions should live on that article's talk page, but of course there are exceptions. —Myceteae‍🍄‍🟫 (talk) 18:36, 15 July 2026 (UTC)

I think that frequent, repetitive proposals are the main trigger, though since personal tolerance levels vary, sometimes repetitive means "the second time in Wikipedia's entire history that someone has suggested this" and frequent means "every few years".

Mostly, the non-move discussions were along the lines of the discussions I linked:

• political labels (e.g., should the infobox say this group is far-right?),
• highly contentious content that realistically isn't going to be changed, no matter how much it's discussed (e.g., proponents of Zionism on their ideal boundaries for Israel), and
• someone's personal pet project/pet peeve (e.g., adding/removing a coat of arms).

They are mostly political in some way, and they're mostly reasonable-ish but rejected proposals. For example, there are a lot of discussions that amount to "I read something that called this political organization 'far-right' this morning, so I think we should say far-right in the infobox", followed by the OP being surprised when everyone yells at him.

If I were to give advice on the subject, I'd suggest:

• A moratorium should almost never be for longer than six months.
  • Logical conclusion: Imposing a moratorium on a subject that is naturally being discussed less often than once every six months is a waste of time.
  • Putting that in practice: I'd suggest that no such discussion be started unless it can be proved, with links, that a significant discussion on the same point has happened at least three times during the previous 12 months. For comparison, a newbie saying "How about we remove every instance of the word woman in the article Pregnancy?" followed by one or two experienced editors saying "No" isn't a significant, moratorium-worthy discussion.
• Any decision, including a moratorium, can be overturned by consensus because we have Wikipedia:No binding decisions by policy, so if we're going to impose one, it should have a strong consensus behind it. I'd recommend an RFC.
  • Putting that in practice: Since the median RFC gets (if memory serves) nine editors commenting, a discussion imposing a moratorium should probably have a minimum of 10 participants.
• Before proposing a moratorium, editors should consider posting a {{FAQ}} at the top of the page about perennial proposals.
  • A neutral way to say this might sound like "Prior discussions have decided to _____. Before proposing a change, you should read and be prepared to address the concerns raised in the past discussions such as [link], [link], [link]" or "This page says 'right' instead of 'far-right' because that's what most sources say. Editors are unlikely to support a change unless you can provide multiple, high-quality sources that address the subject directly (e.g., peer-reviewed scholarly papers whose main subject is how to classify this political group) or evidence that 'far-right' has become the most common description in mainstream media, rather than an occasional or qualified claim used by a small minority of sources."

WhatamIdoing (talk) 19:51, 15 July 2026 (UTC)

I agree with a lot of this. Overall, I think much of this pattern is similar at RM. Editors should use a variety of strategies, instead of and in addition to moratoria, to limit disruptive discussions. At Talk:Sean Combs there is currently a one-year moratorium on RMs and the page makes good use of {{FAQ}} in addition to the {{Old moves}} template. I find this to be a good example, but also an outlier due to the sheer number of proposals in calendar year 2025, during with Diddy was very much in the news. —Myceteae‍🍄‍🟫 (talk) 18:00, 19 July 2026 (UTC)

Also, I've already said this, but a moratorium, which is by definition temporary, is not at odds with WP:No binding decision (aka WP:CCC). We're all in agreement that they should be used sparingly. —Myceteae‍🍄‍🟫 (talk) 18:03, 19 July 2026 (UTC)

I think that having discussions about whether to overturn a moratorium would be just as disruptive as having the discussion about the point of the moratorium. If we have an "obvious consensus" (i.e., a fairly large number of people firmly agreeing), then I think we're more likely to get people to just wait out the temporary situation rather than opening a discussion about whether they should overturn the moratorium so they can have another discussion. WhatamIdoing (talk) 19:18, 19 July 2026 (UTC)

That's going to be fine in most cases, but what happens when (someone thinks) there has been a very significant real-world change that means the previous consensuses might not still hold? When change is major and unambiguously relevant very few people will object to just ignoring the moratorium, and in cases where it's unambiguously trivial to nearly everyone then that's the point of the moratorium and attempts to ignore it will be quickly shut down. However what happens with the significance is not clear, there needs to be some method of discussing whether it is or not, independent of whether to move or not move. Thryduulf (talk) 21:18, 19 July 2026 (UTC)

I agree. We could have a moratorium on changing the first sentence of an article, but what if the subject dies during the moratorium?

That's why I suggest that any moratorium that is imposed should be done so with an obviously strong consensus. A strong consensus should discourage people from trying to overturn it under ordinary circumstances but still leave that as a realistic possibility under extraordinary circumstances. WhatamIdoing (talk) 22:53, 19 July 2026 (UTC)

I just think that the legitimacy of a moratorium which is enacted by the 'winners' of a dispute is always going to be protested, imo it should be enacted by those uninvolved in the dispute, whether through RfC, AN, or MR idm Kowal2701 (talk, contribs) 22:58, 19 July 2026 (UTC)

@WhatamIdoing that all works fine when it's clear whether the change in circumstances is unambiguously significant or trivial, but still doesn't address situations where it isn't clear one way or the other. Thryduulf (talk) 23:50, 19 July 2026 (UTC)

>I think that having discussions about whether to overturn a moratorium would be just as disruptive as having the discussion about the point of the moratorium.

>However what happens with the significance is not clear, there needs to be some method of discussing whether it is or not, independent of whether to move or not move.

I've seen this go one of three ways a majority of the time. Either the meta-discussion about whether or not there is a significant change proceeds with little controversy, or it is quickly shut down without much controversy, or it erupts into yet another disruptive and repetitive discussion. I've seen folks go ahead and launch the repeat RM (or RFC or whatever), in apparent violation of the moratorium, by making an explicit argument as to why they think the situation has changed, and let editors hash out there. None of these are super common, since moratoriums aren't that common in the first place. I don't know that there is a one-size-fits-all approach to addressing these. I have seen at least once instance, I think at Talk:X (social network), where a subpage was created specifically to workshop a fresh RM proposal once the moratorium was lifted. This approach has downsides but I thought it was productive to allow for a thorough "BEFORE" while segregating the discussion from the main page. —Myceteae‍🍄‍🟫 (talk) 23:00, 19 July 2026 (UTC)

Here's a fresh example where a moratorium was proposed at RM, did not gain consensus, and it was ultimately not imposed: Talk:Myocardial infarction#Requested move 10 July 2026. —Myceteae‍🍄‍🟫 (talk) 17:50, 19 July 2026 (UTC)

Automatically grant new page reviewer rights in certain circumstances

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.

---

According to Template:NPP dashboard, the New Page Patrol article backlog is "very large" and "growing very rapidly", with nearly 1000 new pages since last week. Given this, I propose that we automatically grant New Page Reviewer (NPR) rights to any editors who have more than 1000 unreverted, undeleted edits, have edited for at least a year, and have no current or prior blocks on the English Wikipedia. I suggest that we continue to offer WP:PERM/NPR for people who don't meet these requirements.

This way, we will be able to crowd-source a massive amount of reviewers while still only accepting people who are reasonably-experienced with the proper functioning of the Wikipedia. GrinningIodize (articles without a Wikidata item) (talk) 23:47, 14 July 2026 (UTC)

In general, I'm not super enthusiastic about giving out perms to review others' work automatically. Lots of people can gnome around but not really write whole articles themselves, which is fine and dandy work, but also not really the kind of work that would contribute towards warranting NPR. — Red-tailed hawk _(nest) 01:21, 15 July 2026 (UTC)

I think that users should have to want to do the reviewing. But then they can ask for that permission. Graeme Bartlett (talk) 01:25, 15 July 2026 (UTC)

The problem here is that currently, even WP:PERM/NPR has a reasonably-large backlog. GrinningIodize (articles without a Wikidata item) (talk) 16:26, 15 July 2026 (UTC)

I wouldn't say that three users waiting to be assessed is a reasonably large backlog. Cheers, SunloungerFrog (talk) 23:11, 15 July 2026 (UTC)

I fall under your classification for auto NPP, but know I am absolutely not trained up in it. Being given that without question of my ability would let me do all sorts of damage. Jerod Lycett (talk) 06:24, 15 July 2026 (UTC)

Why NPR and not autopatrolled? Autopatrolled doesn't require you to do anything, unlike NPR. sapphaline (talk) 08:11, 15 July 2026 (UTC)

Though if we're going to give away autopatrolled automatically, I would make it "1000 unreverted, undeleted mainspace edits", and also add something like "created at least 3 articles in the last 6 months and haven't had any of them deleted". sapphaline (talk) 08:13, 15 July 2026 (UTC)

1000 unreverted ... mainspace edits is a very high bar. Three out of my last 50 edits have been reverted by TAs or very-low count accounts that re-reverted my reverts of problematic edits. Anyone who patrols new edits is going to be regularly reverted by editors who think the proper response to being reverted is to re-revert back with no discussion. Donald Albury 15:34, 15 July 2026 (UTC)

"1000 unreverted, undeleted mainspace edits" in all time, not in a month/year/etc. sapphaline (talk) 10:02, 16 July 2026 (UTC)

I can confirm that this is my intended meaning. GrinningIodize (articles without a Wikidata item) (talk) 15:36, 16 July 2026 (UTC)

NPR requires a very particular set of skills which you don't acquire by turning up and gnoming, even for a long time. At the risk of hijacking the idea, giving autopatrolled automatically sounds like a better plan. You could even start with a high bar like 100,000 edits and 10 years then bring it down gradually if no problems arise. Certes (talk) 11:59, 15 July 2026 (UTC)

Are there any people with 100,000+ edits who don't have autopatrolled? sapphaline (talk) 12:13, 15 July 2026 (UTC)

Excluding bots, and limiting to accounts over 10 years old, there are 432 of us. But the threshold could be lowered later. Some of these people may have had autopatrolled granted but removed for good reason. It comes down to 374 after removing users with current blocks of any type. Certes (talk) 12:29, 15 July 2026 (UTC)

100,000 edits seem extremely high. I would say 5 years + 5000 mainspace edits. Some1 (talk) 22:45, 15 July 2026 (UTC)

Strongly disagree with this, I have seen several long-term editors in good standing who don't really grasp notability or even WP:RS, including some who currently have autopatrolled but probably should have it removed. Helpful Raccoon (talk) 23:52, 15 July 2026 (UTC)

(By 'long-term' I mean 20,000-200,000 edits over 15+ years.) Helpful Raccoon (talk) 01:10, 16 July 2026 (UTC)

AINB has two recent cases of long-term editors with autopatrolled who ended up injecting dozens to hundreds of LLM articles into mainspace without review. Really sad, but not as rare as we'd hope. M kuhner (talk) 02:35, 20 July 2026 (UTC)

That's a shame. In that case, we probably shouldn't extend even autopatrolled more widely than at present. Granting the permission only helps NPP of course, not the editor themselves (unless they plan surreptitious mischief). Certes (talk) 09:49, 20 July 2026 (UTC)

I never want autopatrolled. Would this let us opt out so we can guarantee a second pair of eyes? Jerod Lycett (talk) 02:28, 19 July 2026 (UTC)

Have you ever checked the page views on articles for the first day that they're in the mainspace? They usually get more than a dozen page views. What autopatrolled provides is nobody needing to click a button after looking at it. WhatamIdoing (talk) 02:48, 20 July 2026 (UTC)

Should have clarified: A second pair of eyes from an experienced, knowledgeable editor. Jerod Lycett (talk) 05:18, 21 July 2026 (UTC)

From my understanding, autopatrolled rights are really intended for frequent page creators, since they will otherwise let a lot of unwanted stuff get past the NPR folks. GrinningIodize (articles without a Wikidata item) (talk) 16:33, 15 July 2026 (UTC)

Alternative idea

It seems like a lot of people here are worried about competence. Perhaps we could automatically grant 1-month trials to eligible users on the condition that they take a training course and check a box saying that they have the necessary skills to become a new page reviewer?

After the 1-month trial period, users can request a permanent extension of NPR rights at WP:PERM/NPR if they can demonstrate their competence to an admin. GrinningIodize (articles without a Wikidata item) (talk) 16:30, 15 July 2026 (UTC)

I would also be open to 3-month trials if need be. GrinningIodize (articles without a Wikidata item) (talk) 16:34, 15 July 2026 (UTC)

I'm sure the good and experienced people at WP:NPR will have plenty of good advice on this topic. Certes (talk) 17:55, 15 July 2026 (UTC)

How likely is this to fix the problem? If people are not currently requesting permissions, what makes us think they would request a 1-month trial? Or if the trial is granted automatically, are we hoping editors will be notified of their new abilities and jump in to address the backlog? —Myceteae‍🍄‍🟫 (talk) 18:43, 15 July 2026 (UTC)

People are currently requesting permissions, but WP:PERM is so backed-up that many of them are stuck waiting for longer than they should have to. My idea is that a bot would send a talkpage notice to eligible editors with instructions on how to obtain NPR rights without waiting for another human. Editors who deem themselves worthy can take a short training course, affirm that they understand their duties, and work on the backlog until their trial ends. Afterwards, they can either return to their normal work or request permanent rights.

This system would help encourage more people to do new page reviewing while still keeping the process relatively meritocratic and filtering out most of the less-qualified editors early on. GrinningIodize (articles without a Wikidata item) (talk) 18:56, 15 July 2026 (UTC)

I think a bit of perspective may help here. The queue is currently at 20865. In my experience over the past 7 years or so, there's been one or two times that we've gotten the queue down close to zero, but most of the time it's hovered between 10,000 and 30,000. It typically takes 3 to 6 months for an article to get reviewed, although many are reviewed much quicker because many people patrol from the front of the queue.

As for requests, setting aside that the backlog is currently zero, when you wrote the above comment the backlog was a whopping four days and 3 requests. Calma mi hijo. _signed, Rosguill ^talk 00:16, 16 July 2026 (UTC)

A good chunk of editors who are scrutinized and approved by admins for a trial already fail to get the right renewed due to problems with their reviewing. Not to mention Wikipedia has had a lot of WP:UPEs targeting the NPR right to help their scam operations. Granting it automatically could greatly exacerbate the problem. Helpful Raccoon (talk) 00:45, 16 July 2026 (UTC)

There are only two people listed as available as trainers at all. One has stated zero slot availability. The other has not responded to multiple of us editors asking for training and appear to be missing. I don't think training courses are even a possibility without somehow incentivizing NPP to do so. Jerod Lycett (talk) 02:37, 19 July 2026 (UTC)

When I said "training course", I meant something more along the lines of an interactive tutorial; users would be presented a simulated curation interface with example articles, and they would have to perform the appropriate action for each example (approve, delete, skip, etc.) and the tutorial would tell them what the correct course of action was in each case. At the end, the user is presented with their statistics, and if they got a reasonable amount of actions correctly (perhaps an ~80% accuracy or higher), they check a box affirming that they believe themselves to be competent in NPR tasks, and that they understand that any bad-faith behavior or generally-bad decisions could get their rights revoked. GrinningIodize (articles without a Wikidata item) (talk) 17:37, 19 July 2026 (UTC)

The "training course" may be part of the problem. NPP has had so much stuff crammed into it over the years that "fully" reviewing one article takes too much time, and people feel like they aren't doing it right if they don't reject most of the articles.

For example: I went to Special:NewPagesFeed, set the filter to give me unreviewed articles from 30–37 days ago. The first five in the list were Quinn Whittock, Jeanne Bourgoint, Fulton Schools, Jane S. Jaquette, and Adam Kout. It took me less than five minutes total – I timed it – to determine that all five were real, non-hoax subjects, with real sources. None of them are deletion worthy; in fact, every single one of them is longer and better sourced than the median Wikipedia article. If an NPPer wants to do a quick check for copyvios (which I didn't bother checking because I didn't have the link handy) and LLM use (which I'm not skilled at spotting), then all five of these could be marked off the list.

The Wikipedia:New pages patrol/School approach doesn't work that way. It begins from a position of doubt: doubt the subject, doubt the editor, doubt yourself, doubt that Wikipedia can survive if you ever make a mistake. Most NPPers don't actually mark articles as being reviewed, even when they've reviewed them for what matters. The idea is that it's best to do nothing than to risk making a mistake. After all, if you make a mistake, your reviewer right can be yanked – but if you don't take an action, nobody will ever know you looked at it, so they'll never say that you did it wrong.

According to the NPP school, I've failed because I didn't:

• "Successfully use the NPP flowchart"
• "evaluate articles against specific SNG criteria" (they all meet GNG, which is sufficient for all except NCORP, which didn't appear in my sample set)
• "Find...common Wikipedia practices when evaluating notability" (I know the communities practices. I've even written substantial parts of some of the notability guidelines. I don't need to "find" them.)
• "Identify and appropriately remedy copyright issues in at least 90% of instances" (because I didn't check)
• "Engage in appropriate and useful conversations with other editors about NPP" (sounds like one of those creepy companies that monitors what its employees posts on social media about their working conditions)
• "Appropriately apply warning templates to users"
• "Apply appropriate tags to article with at least 90% accuracy"
• "Understand NPP procedures and norms"
• "Correctly and efficiently use scripts while doing new page patrol"

But I think if you know what you're doing and you look at those five articles, you'll agree that my quick review ended up with the correct answer. WhatamIdoing (talk) 20:23, 19 July 2026 (UTC)

File:Simplified NPP flowchart for articles.png is said flow chart. (You missed the first and last main steps.)

That said: What should NPP's job actually be, is it just finding articles that are CSD candidates and letting the rest through? Is it to do GNG and SNG checks on top of that? Should tagging things with {{uncategorized}} or whatever else be part of it as it seems to be? If the entire purpose of it is to simply find the junk articles and tag them for CSD then it's basically just a sub-project of CVU. If it's to ensure all articles meet GNG/SNG then it feels like AfC review is a subproject of NPP. If they are actually meant to understand and apply maintenance tags, then that's where we do need actual (basic) training or experience. Do you know if there's any good history of NPP that would explain what its intended mission is? I'd be happy to get an RFC on NPP going to ask which of those three it should be though. Jerod Lycett (talk) 05:45, 21 July 2026 (UTC)

Close this thread?

Apparently, auto-granting an NPP right has not gone well with the community here, despite trial alternatives. Shall this whole thread be closed then? George Ho (talk) 04:32, 20 July 2026 (UTC)

Yes. The "certain circumstances" item reminded me of John Templeton's famous saying: "This time is different" are the most expensive 4 words in investment. So no point in this. Yesterday, all my dreams... (talk) 02:50, 21 July 2026 (UTC)

@GrinningIodize: You okay with the whole thread being closed then? George Ho (talk) 03:07, 21 July 2026 (UTC)

I suppose that's alright with me. GrinningIodize (articles without a Wikidata item) (talk) 14:59, 21 July 2026 (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.

Preventing citogenesis when recreating deleted pages

I happened to stumble across the deleted page Albert P. Halfhill, the person who supposedly invented American-style canned tuna fish. Formerly a Good Article, it was deleted due to presumptive copyvio. I would like to recreate it by writing non-copyvio content based on reliable sources. Unfortunately, since I cannot view the previous state of the article or even when the article was created, it is difficult for me to be certain that potential sources did not draw their information from the article while it was still online. Which leads me to a thought: when deleting a page for copyvio (or other issues unrelated to notability, for example LLM use) would it be a good idea to set a precedent of copying at least the citation list to the talk page and leaving that up? Or is there another way to mitigate citogenesis resulting from bad content continuing to be reproduced elsewhere after we delete it here? -- LWG ^{talk (VOPOV)} 17:37, 16 July 2026 (UTC)

Please go to Wikibin and search halfhill , you will get the article. Yesterday, all my dreams... (talk) 18:13, 16 July 2026 (UTC)

Or use Wayback Machine. sapphaline (talk) 10:33, 19 July 2026 (UTC)

WP:REFUND won't post copies of copyvio articles on wiki, but you could probably get an admin there to post or e-mail just the list of sources for you. WhatamIdoing (talk) 20:26, 19 July 2026 (UTC)

But why bother with that when Wikibin or Wayback will do it in 15 seconds. Yesterday, all my dreams... (talk) 02:46, 21 July 2026 (UTC)

It's often best not to inadvertently bias yourself to the now deleted article's text/structure by viewing it directly, and external internet archival services often may not be complete. GreenLipstickLesbian💌🧸 03:48, 21 July 2026 (UTC)

Wikibin triggered my antivirus software, which seems a bad omen. CMD (talk) 02:59, 24 July 2026 (UTC)

Do you use Norton or what? My Android tablet did not complain. But thanks for the info, I will not use Wikibin again. Wayback should be ok. Is there a Wikilist of problematic sites? Yesterday, all my dreams... (talk) 15:32, 24 July 2026 (UTC)

I'm only aware of WP:ATODAY, which is just the one site. WhatamIdoing (talk) 16:09, 24 July 2026 (UTC)

Oh, a Doug Coldwell article, aka the reason the GA system got a massive reform! Don't assume anything from it being a GA.

There's too many articles deleted for your idea to be practical, and we've got pretty solid consensus behind WP:G8, so I don't think this is desired by the community. In addition, I believe that in in cases involving POV pushing/UPE that often get swept up in something like G5 or even CPN, biasing future editors to a bunch of cherrypicked sources would be bad. However, this is something where you can just ask any admin in Category:Wikipedia administrators willing to provide copies of deleted articles for the citation list, and they'll (almost certainly) oblige. Given the article creator, though... don't bet on any of the sources being good, usable, or appropriate for the claims he made in the deleted article. I'm really not sure how a list of sources used in the old article would prevent citogenesis, either. When I was working on an old Doug Coldwell article where I suspect he added misinformation picked up by modern sources, knowing which sources he used did not help me with that. GreenLipstickLesbian💌🧸 23:55, 19 July 2026 (UTC)

I think the idea is that the sources predate the creation of the article, so they can't be citogenesis Katzrockso (talk) 01:25, 24 July 2026 (UTC)

Yes, this. Without knowing when the copyvio/inaccurate material was introduced, I don't know what my cutoff date is for knowing that an external source didn't get their info from here. -- LWG ^{talk (VOPOV)} 01:49, 24 July 2026 (UTC)

I mean, if the sources pre-date the article, then they're fine anyways, barring normal rules on reliable sources? And even if they were cited in the article, you'd have to know which got added first: the info or the sources (a la xkcd:978) to rule out citogenesis. GreenLipstickLesbian💌🧸 08:15, 24 July 2026 (UTC)

Maybe I'm just dumb, but is there a way for me to view when a deleted article was initially created? Also if, say, the article was a stub from 2011 to 2021, then a bunch of info was added in 2021, then if I find that info in sources from 2018 I can probably use them but if I find it in sources from 2022 I probably won't. -- LWG ^{talk (VOPOV)} 15:19, 24 July 2026 (UTC)

From what I have seen, you are not dumb. But within Wiki, unless you get admin help you can not see those. Wayback would give you help. Yesterday, all my dreams... (talk) 15:29, 24 July 2026 (UTC)

There's nothing inherently wrong with using the same sources as a deleted draft, and copyvio doesn't normally produce false information (which is what we'd be concerned about for WP:CITOGENESIS). WhatamIdoing (talk) 16:11, 24 July 2026 (UTC)

Isn't there a massive problem in this: if you're going to assume that all sources after a deleted Wikipedia article are potentially based on the deleted article and therefore inadmissable, then any deleted article automatically marks the end of all improvement of human knowledge on the subject! Deleting an article shouldn't force all future works on the subject to be deemed unreliable. Elemimele (talk) 16:15, 24 July 2026 (UTC)

As always, the best defense to citogenesis is rigorous verification of reliable sources. There's always bad information that's still live, and whether or not the article has been removed doesn't really have any effect on whether or not external sources did their homework correctly. In fact, if they were written AFTER the article was deleted then they can't be citogenesis (unless their lazy research involves digging through archives, which seems to defeat the purpose). Plus as stated above copyvio isn't really a cause for concern here - subsequent sources would essentially be pulling the information directly from the copyrighted source. Citogenesis from a G15 deletion would be a lot more concerning. ChompyTheGogoat (talk) 02:19, 25 July 2026 (UTC)

unless their lazy research involves digging through archives Or their research involved reading other sources that copied content from Wikipedia. Anomie⚔ 18:23, 25 July 2026 (UTC)

Right, there's nothing inherently wrong, but it's still something to think about. There's a lot of websites (for example local newspapers or small-town government webpages) that I would normally use as a "good enough" source in the absence of something better, but if I'm using them as a source for a claim that was already in a wiki article I normally compare the potential source with the wiki article history to make a judgement call on whether I trust it, since a lot of sources in that quality tier uncritically repeat information from Wikipedia. The copyvio is only relevant because its the reason I don't have access to the article history. -- LWG ^{talk (VOPOV)} 16:27, 24 July 2026 (UTC)

You can. The easiest way only works if the article was created after late June 2018: check the creation log. (It'll be listed at its original title there, so you may have to trace back through page moves.)

That doesn't apply for Albert P. Halfhill, which was created on 23 May 2017. For that, you'll need to look at the publicly-available metadata here, which will give you everything admins can see except the edit summaries, actual text, page length, and anything about revision-deleted/suppressed revisions, albeit in an uglier and much less convenient form. (I'll add that the article was of similar length in its very first revision as it was just before its copyvio blanking, ie it didn't stay at stub length for years as you speculate.) There are at least two user scripts (User:SD0001/deleted-metadata-link, User:DreamRimmer/DeletedMetaData) that clean it up some; I haven't used either, so have no firsthand experience with them. —Cryptic 16:31, 24 July 2026 (UTC)

That's what I needed, thanks! -- LWG ^{talk (VOPOV)} 16:35, 24 July 2026 (UTC)

You can get the page length, sha1, and tags from the API too, e.g. . Anomie⚔ 22:57, 24 July 2026 (UTC)

A fast-diff, extended-confirmed, anti-LLM edits browser to solve the problem of AI

I've had this idea for awhile and would like to get some people's thoughts on it and some help with making this proposal actionable/ironing out specifics for a proposal on VP. The rough gist of it is this: A fast-diff edit browser (inside of Wikipedia) for extended-confirmed users (the Rollbacker/Admin pool of users is not enough to effectively stop AI) that randomly selects 20 or so articles and scans each for 'ai signs' in their writing. (overuse of semicolons, em and en dashes, specific words and sentence structure that AI likes, if the user who added the content was new/was warned before for LLM use, if the content was added during the time AI could be used as a chatbot (post GPT-3 release timeperiod) and other AI artifacts in writing) Then, we score the specific sentences in each article for how 'likely' AI writing is. We would then use a small LLM, some basic local AI like Local LLAMA, to read the text and where the problematic sentence is and find a way to cut it out that does not break the structure of the article. The extended-confirmed user then simply checks 'Yes' or 'No' to decide if the specific removal (which will show up directly on the screen of the user) is just or unjust. It will then take them to the next and repeat the process with 20 more articles once out of content to decide on.

This is a very rough draft of my idea here and was written in like 10 minutes and probably has a variety of flaws in it's writing. I would like to hear people's thoughts, possible alterations to my idea and in general help to make this an actual proposal and not just an idea bouncing around in my head with all the others. Thank you. Ilov3gam3z (talk) 12:26, 14 July 2026 (UTC)

I've already been working on testwiki:User:Chaotic Enby/AIDiffBrowser.js, although it is for now more centered on browsing cases on the AI noticeboard. Once the first version is released, this sounds like a very good functionality that could be added down the line! Chaotic Enby (in solidarity · talk · contribs) 14:54, 14 July 2026 (UTC)

each for 'ai signs' in their writing. (overuse of semicolons

You should not make this until you have a better grasp of what AI writing looks like. Gnomingstuff (talk) 15:12, 14 July 2026 (UTC)

? This is why it's at the Idea Lab, not Proposals. I would need people to iron out things like that and most technical specifics (I don't know much on the technical side of Wikipedia). Ilov3gam3z (talk) 15:16, 14 July 2026 (UTC)

Also, I still see overuse of semicolons, overuse em and en dashes, overuse of bolding and over-organization of content as obvious signs of AI writing. Ilov3gam3z (talk) 15:17, 14 July 2026 (UTC)

These are not the most reliable signs and are very easy to misinterpret; the fact that so much outside reporting has fixated on things like this is a large part of why people think AI detection is not possible. Stuff like dashes is especially dubious for Wikipedia since they don't really appear very much in article text (possibly because we've had like 5 dash templates over the years to muddy the training data, but who knows) and are generally noisy due to bullet point dashes, dashes in headlines, etc.

For Wikipedia purposes, WP:AIVOCAB, WP:AIATTR, WP:AITREND, and WP:SUPERFICIAL are the four most reliable ones. Gnomingstuff (talk) 19:53, 14 July 2026 (UTC)

Again, what I have written here is the broad strokes of my proposal. AIVOCAB was already one of the things I was thinking of, though. The other signs would be good to include too. Ilov3gam3z (talk) 19:59, 14 July 2026 (UTC)

I think that automating the entire thing and letting editors make such sweepings decisions through a single checkbox is far too simplistic and ripe for abuse. Highlighting the problematic content with the attached reason it was flagged and allowing the editor to decide how to handle it seems like a safer proposition. I also don't think it should be accessible to all extended confirmed users, but perhaps a new user group for people who've demonstrated at least some degree of competency regarding LLM text. Possibly with upgraded permissions available for more experienced editors to make wider changes, similar to rollbackers.

I do agree that we're badly in need of tools to help combat the issue, and this certainly sounds like it has promise. ChompyTheGogoat (talk) 16:07, 14 July 2026 (UTC)

I did not mention this, but highlighting with a reason is also something that will probably be included. Maybe we could have a 1k or 5k edit user group above extended-confirmed? I still feel making this usable only by those that are approved by admins or what-have-you will bog the process down, but I am open to thoughts on this as the extended-confirmed restriction is something I came up with on the spot while writing this. Ilov3gam3z (talk) 16:11, 14 July 2026 (UTC)

Having 5k edits instead of 500 might correlate with familiarity with Wikipedia processes, but not at all with the much more specific task of AI text detection. I do plan to implement something like this to AIDiffBrowser one day, although it will likely be limited to a new user role (or pseudo-role like AfC reviewers) once I'm done with it. Chaotic Enby (in solidarity · talk · contribs) 16:15, 14 July 2026 (UTC)

My TAIV rights were granted in no time, so I think if the good folks at WP:AICLEAN kept an eye on requests they could go through quickly. There's nothing urgent about it, since they can still edit manually while it's pending. ChompyTheGogoat (talk) 16:16, 14 July 2026 (UTC)

If we go for something like this (and I'm undecided whether I think it's a good idea or not) then it should be handled similarly to AWB where there is a (pseudo?) usergroup that is granted only to those who ask for it and only to those who have demonstrated competence in the relevant area - in this case that would mean experience of detecting LLM-written material with a very low false positive rate. Thryduulf (talk) 21:24, 14 July 2026 (UTC)

I made a tool where step 1 is similar to "randomly selects 20 or so articles and scans each for 'ai signs' in their writing", and step 2 is similar to "score the specific sentences in each article for how 'likely' AI writing is", automatically skipping articles already tagged for AI cleanup: https://wikitomte.toolforge.org/

You're welcome to use the code as a starting point for additional experiments. For example, you could add a step 3: enable the user to select a suspicious passage to try to search the diffs and find out when the passage was added. I thought about linking to WikiBlame with a pre-filled query. Feel free to put in a PR if you come up with something interesting! Dreamyshade (talk) 04:40, 16 July 2026 (UTC)

It woudl be useful to have the edit feed be determined by a query for one particular AI tell (for example, "played a pivotal role". The tool would then search for pages with the phrase, figure out the blame, and possibly also highlight phrases left over from those AI edits in the article. This automates the process in User:Gnomingstuff/Guide to finding AI-generated text, which is quite similar to my own AI-detection process. Somepinkdude (talk | contribs), in solidarity 22:11, 19 July 2026 (UTC)

Great idea! Chaotic Enby (in solidarity · talk · contribs) 22:47, 19 July 2026 (UTC)

Yeah this is nice. Ilov3gam3z (talk) 00:13, 20 July 2026 (UTC)

"We would then use a small LLM, some basic local AI like Local LLAMA, to read the text and where the problematic sentence is and find a way to cut it out that does not break the structure of the article."

Let me get this right: we use an automated system (of uncertain reliability) to find text that was written by an automated (LLM) system, which text we assume to be unreliable because it came from an LLM, and we then use another (albeit smaller) LLM to rewrite the text removing the bit that our automated system thought might be from an LLM? And we assume that a human armed with a tick-box will take the time to check the new LLM-written text is more correct than the old LLM-written text? As a very naive person, aren't we going to end up with articles that were written by an LLM and then made LLM-filter-proof by another LLM? Why don't we cut out all the hassle by simply telling new editors that after they've got ChatGPT to write their article, they should ask Claude to remove all the AI signs? Elemimele (talk) 16:34, 24 July 2026 (UTC)

From what I understand, the proposal isn't to rewrite the AI content, but remove it in a way that leaves the rest of the article coherent – although, if the LLM starts to add padding text to connect the sentences on each side, we will indeed run afoul of WP:NOLLM and be at risk of WP:SYNTH. I don't necessarily agree with this particular way to do it, and in fact, the AI diff browser I'm working on will have a field for humans to manually rewrite the text once the sentences are removed, which is much better to leave it to LLMs. Chaotic Enby (in solidarity · talk · contribs) 16:56, 24 July 2026 (UTC)

We are not rewriting; we are using a (very basic, mind you) LLM to find out the best way to cut the content, i.e. should we clean up some punctuation? Should we remove some nearby sentences to make it flow better? etc. AI will not be adding any text, simply fixing issues that come with ripping out content from a sentence. Ilov3gam3z (talk) 17:16, 24 July 2026 (UTC)

(sentence or paragraph, mind you) Ilov3gam3z (talk) 17:16, 24 July 2026 (UTC)

There will also be no padding text as Chaotic Enby noted. If the situation is too complex for AI to fix, we will simply ask the user to do it manually in a web browser. Ilov3gam3z (talk) 17:18, 24 July 2026 (UTC)

I think that's still too iffy. We've got all these former spelling and grammar tools out here that are now making more advanced AI suggestions to "improve flow", which all too often amount to total rewrites that can change the meaning, which is why such tools aren't allowed on en-WP. Editors who are granted permission for this proposed tool should be sufficiently experienced to handle the necessary adjustments themselves - it's going through entire articles and trying to identify which specific parts from an extended edit history were inserted from LLM output that's the most tedious and time consuming, IMHO. The editors who are already doing such work manually could make a lot more progress if that part was automated, and it might help attract more people to the cleanup project as well. ChompyTheGogoat (talk) 00:28, 25 July 2026 (UTC)

There will be very strict rules on what the AI can and cannot do. AI will only be able to remove text or add punctuation and will have a very specific system prompt. If it adds text that is not punctuation, it will simply be asked to do it again until it gets it right. Ilov3gam3z (talk) 00:43, 25 July 2026 (UTC)

Even removing text can fundamentally change the meaning - "He did [not] do it" > "He did do it". That's why people get confused over the use of "minor edit", because length is irrelevant - only whether it changes the meaning of the content in any way.

Besides which, just auto-deleting text that's already been highlighted for inspection doesn't save the editor much effort and the content is still likely to require a rewrite if the AI insertion was woven in between good edits. I just don't think the tradeoff is beneficial enough to justify the potential risks, and I suspect any proposal to fight LLM content by allowing a different LLM to make edits will hit far too much pushback and end in a WP:SNOW close.

It might be more useful to build in functions that allow editors to fix various sections in various ways. For example, [Flagged section A] could be rewritten manually, [Flagged Section B] could be reverted back to how it looked at Last Good Edit, and [Flagged Section C] could just be deleted. Maybe a way to flick back through the previous edits to whichever section is selected so they can easily compare the changes to just that part without leaving the editor tool. I don't have the tech skills to have any clue how feasible it might be to implement that kind of thing. ChompyTheGogoat (talk) 01:35, 25 July 2026 (UTC)

See User:Chaotic Enby/AIDiffBrowser.js for the prototype I made of that! (not sure how buggy it is, mostly been developing it on testwiki for now)

Currently just giving you the option to revert (or keep given that some edits inbetween might have to be kept), but in-place editing will likely be the next thing I'll implement. Chaotic Enby (in solidarity · talk · contribs) 07:34, 25 July 2026 (UTC)

That's why I prefer to only allow this proposed system to identify the problematic text, not decide how it should be fixed - similar to how LLMs are not banned for RESEARCH purposes, but editors must do their own writing based on the findings. We clearly need some type of automation to try to keep up with all the slop insertion, but we have to tread very carefully to avoid said slippery slope. Limiting the user group would help ensure those with access have the necessary skills to confirm that the identified content is consistent with LLM output and not just assume that it was flagged correctly. ChompyTheGogoat (talk) 00:17, 25 July 2026 (UTC)

... and the skills required to identify LLM material are going to get tougher and tougher as AI gets better at imitating humans. The saving grace is probably that if a(nother) human checks the text and verifies the sourcing, and no one can even work out whether it was LLM or not, then it becomes rather unimportant whether it was LLM. Elemimele (talk) 14:40, 25 July 2026 (UTC)

And even the current automated tools for identifying it are unreliable, which is why human verification is vital. As you say, if it actually gets good enough that we can't tell the difference things may change, but currently we need to be able to remove invalid information that AI didn't realize is invalid, so we can't rely on AI to remove what's invalid. Flagging it for inspection and adding tools to facilitate removal just improve the workflow for human judgements. ChompyTheGogoat (talk) 05:27, 26 July 2026 (UTC)

Just to be clear, the idea I suggested doesn't involve any "AI removing AI". Instead, the diff browser would simply use WikiBlame, make a list of all the blamed user's added lines to that particular article, apply changes from later edits, and then end up with a list of AI-generated content from that user. Of course, this would go wrong if one user has both AI and non-AI content additions, but that's reasonably rare, and the EC user does have to use their own judgement in these cases. Somepinkdude (talk | contribs), in solidarity 20:29, 26 July 2026 (UTC)

The original proposal said We would then use a small LLM, some basic local AI like Local LLAMA, to read the text and where the problematic sentence is and find a way to cut it out that does not break the structure of the article. That's what I'm concerned with. If your suggestion is just for identifying the problematic text and can't make any changes without the editor manually reviewing it, then we're on the same page. I think there was some confusion in all the crosstalk. ChompyTheGogoat (talk) 04:21, 27 July 2026 (UTC)

Add AI Content Moderation, Add a Head Person for every role.

“But Wikipedia isn’t a democracy!” I know, but who is the BEST at it gets to be a head. Look for Em-dashes ANYWHERE. Banana113 (talk) 00:30, 22 July 2026 (UTC)

Have you found anything that is a problem? The dashes are only a symptom. Our real issue is false and unencyclopedic text and fake references. Graeme Bartlett (talk) 06:36, 22 July 2026 (UTC)

Just a passing comment that the real issue may be that the curve representing the number of Wiki articles has moved upwards much faster than the curve representing the number of active users interested in cleanup. Hence we have too few eyes on too many articles. The problem has been exacerbated by two other factors. Everyone wants a Wikipage for their cousin and the use of LLMs. Alas I do not have a solution to suggest. Yesterday, all my dreams... (talk) 10:49, 22 July 2026 (UTC)

Mostly LLMs, because non-notable articles written by inept users are much easier to spot and more likely to get flagged as vandalism than polished LLM versions. I'd bet every penny to my name there's a direct correlation between rate of article creation (especially those that are denied/deleted) and public access to LLMs. ChompyTheGogoat (talk) 04:20, 25 July 2026 (UTC)

Wikipedia is a democracy in denial that's what it is. And one thing we don't want is leaders and vertical processes making decisions for the community. Another thing we don't want is, okay you probably guessed it already. Chaotic Enby (in solidarity · talk · contribs) 10:43, 22 July 2026 (UTC)

please say whatever it is you are trying to say directly Gnomingstuff (talk) 01:19, 24 July 2026 (UTC)

User appears to be a troll, and possibly a sock. ChompyTheGogoat (talk) 04:39, 25 July 2026 (UTC)

Wikipedia is actually one of the few places where I would regularly come across em-dashes before AI/LLMs took off. I would even theorize that the reason AIs tend to use em-dashes so often is because of Wikipedia. Em dashes were always available on the edit toolbar in the source editor, and our manual of style has a whole section on where to use them. --Ahecht (TALK
PAGE) 14:32, 28 July 2026 (UTC)

Restrict use of minor edit button

Why don’t we restrict it to rollbacks and edits that are between -200 and +200 bytes? _{In solidarity} Wikipedian12512 (Talking is fine | contribs) 21:06, 25 July 2026 (UTC)

Why should we? For example this would prohibit marking Special:Diff/1363524700 as minor because it's +212 bytes, when no one needs to review it. Tenshi! (Talk page) 21:15, 25 July 2026 (UTC)

minor edit have nothing to do with the size of the edit, but the meaning of the actice. see help:minor edit Lavender-wikigirl (talk) 21:25, 25 July 2026 (UTC)

I know. That said, misuse is more common than correct use (I think) when adding or removing 200+ bytes. _{In solidarity} Wikipedian12512 (Talking is fine | contribs) 01:50, 26 July 2026 (UTC)

The minor edit button would be useful if it were restricted to experienced users who routinely used it when performing mundane tasks. Then the filter "Hide minor edits from the watchlist" would make sense. Doug butler (talk) 22:05, 25 July 2026 (UTC)

Agreed - the biggest issue with it seems to be misunderstandings about "minor" referring to size rather than meaning (as with this question), followed by a smaller number of trolls trying to hide vandalism. Limiting it to EC users would probably eliminate the majority of that, so that all edits by newer editors are more likely to be reviewed. Maybe a pop-up asking them to read and confirm they understand Help:Minor edit the first time they gain access to it - I don't think most people notice or care that it's linked next to the checkbox, given the number of questions like this that we see. ChompyTheGogoat (talk) 01:44, 26 July 2026 (UTC)

No, I know that it is about the meaning, but when I see something somewhat like “(-387) . . (fixed typo)”, it feels like something fishy is going on. If it legitimately is a minor edit, it should be explainable in the edit summary. When reviewing, false positives are much better than false negatives. _{In solidarity} Wikipedian12512 (Talking is fine | contribs) 01:48, 26 July 2026 (UTC)

That is one of the Wikipedia:Canned edit summaries, so it's not unreasonable to always consider that edit summary to be guilty until proven innocent.

Would you like it if the minor-edit thing simply wasn't visible to you at all? I (like many experienced editors) turned off the "Hide minor edits" thing in Special:Preferences#mw-prefsection-rc-changesrc years ago, which is enough for me, but I believe it's actually possible to make it invisible in CSS, so you'll never know again whether someone misused the unimportant box. WhatamIdoing (talk) 04:18, 26 July 2026 (UTC)

Fishy is fishy, regardless of length - you can completely flip the meaning of something just by adding or removing "not". Limiting use to EC would minimize such misuse, though nothing can eliminate it entirely. I see more of the GF mistaken uses than intentional abuse. ChompyTheGogoat (talk) 04:30, 26 July 2026 (UTC)

If the concern is abuse, why not limit it to semiconfirmed users? Amyipdev (talk) 00:44, 28 July 2026 (UTC)

I would much rather limit it to WP:Autopatrolled editors, since they are editors where we have explicitly said that we trust their edits enough that they don't need further review. --Ahecht (TALK
PAGE) 14:28, 28 July 2026 (UTC)

I think that would be a bit much - as I've said above my personal observation is that it's mostly accidental misunderstanding of "minor" rather than intentional abuse, and limiting it to EC would ensure they've been around for a minute to get a feel for things, as well as stopping the more obvious trolls that don't take time to build up an account. A lot of newish users that are EC but not autopatrolled gain experience gnoming, and those are exactly the type of small good edits that we don't need to be clogging watchlists with. It's not as if minor edits never get eyes on them; it effectively just lowers the priority of checking them. Autopatrolled is a much higher bar that green flags entire articles. Conversely, I don't think that only limiting it to autoconfirmed (which I believe is what the above user meant) would have any significant impact. ChompyTheGogoat (talk) 14:55, 28 July 2026 (UTC)

Some anti-vandalism tools mark reverts of vandalism as 'minor'. WhatamIdoing (talk) 15:32, 28 July 2026 (UTC)

What else to do with WP:WikiProject Biography?

I rose my concerns about the (in)activity of WP:WikiProject Biography at WT:WikiProject Council (discussion link). I thought about inviting individuals into revitalizing the WikiProject, despite some activity at WT:WikiProject Biography. If neither merger nor marking this as semi-active is the answer, then what else shall be done about this WikiProject? George Ho (talk) 19:55, 18 July 2026 (UTC)

Nothing needs to be done about it. WhatamIdoing (talk) 20:28, 19 July 2026 (UTC)

Why not? Otherwise, why posting notices other than FA Reviews and Nominations and GA Reassessments? I don't wanna go to that project talk notice just to expect a thread to become dormant (i.e. without replies) until bot archival, do I? George Ho (talk) 20:42, 19 July 2026 (UTC)

I suppose the question is: what other kinds of discussions would you like to see happening at that page? Blueboar (talk) 21:09, 19 July 2026 (UTC)

More editors replying there, obviously, not just the same editor replying on any other threads. Too bad, as Johnbod said, the "Biography" scope is very broad.... or too broad for the WikiProject to manage anymore. Right? If WP:WPBIO can't be further split into other subprojects (as it might have had), then... (I think I already asked the question in my OP, didn't I?) George Ho (talk) 23:20, 19 July 2026 (UTC)

Splitting it into smaller subject areas is unlikely to result in more editors per page. Also, WPBIO has a handful of task forces, and if they're equally non-responsive, then that would be evidence against that idea, too.

Perhaps more to the point, the notices you've been posting at Wikipedia talk:WikiProject Biography shouldn't be getting responses on that page, because you're asking them to join an existing discussion elsewhere.

I've left a note at WT:BLPN to see whether we can get a few more people watching that page. WhatamIdoing (talk) 20:33, 20 July 2026 (UTC)

• A highly unpopular view: From what I have seen, a good number of projects are either dying or planning their own funerals. Somehow belonging to a project is no longer in fashion. Please do not ask me why. I do not know. I belong to no projects because I do not know what I will get out of it. Look at Wikipedia talk:WikiProject Computing, what would I gain by looking at it. Nothing. WikiProjects are fading out. Sad, but true. Anybody need a Kleenex to dry their eyes? Yesterday, all my dreams... (talk) 02:42, 21 July 2026 (UTC)

  I don't think that's very unpopular, the issue with funerals is that it takes a lot of effort to fold a WikiProject into another. CMD (talk) 04:01, 21 July 2026 (UTC)

  I fully agree. If anyone wants to work on merging up under-active WikiProjects, please drop by Wikipedia talk:WikiProject Council. We were working on the instructions for doing this at Wikipedia:WikiProject Council/Guide/Merging WikiProjects a while ago. It's pretty complicated. WhatamIdoing (talk) 16:51, 24 July 2026 (UTC)

  Yeah, I'm not entirely sure of the point of WikiProjects. Jerod Lycett (talk) 05:50, 21 July 2026 (UTC)

  I would say it depends on the project. I looked yesterday and found I have pages from 25 projects on my watchlist, but it has been years since I've seen anything from some of them. Some projects have little or no activity on their talk pages, but notices about new pages, drafts, AfD, move requests, etc, appear on my watchlist, a process which likely increases the number of editors looking at them. Some projects I follow have active communities who regularly discuss problems with and improvements to articles covered by the project, something I do not think would easily happen without the wikiproject. Donald Albury 16:38, 21 July 2026 (UTC)

  The various country WikiProjects, and specifically their "assessment tables", are useful for gauging English Wikipedia's coverage of them at a glance. Looking at Wikipedia:Version 1.0 Editorial Team/Iceland articles by quality statistics vs/ Wikipedia:Version 1.0 Editorial Team/Togo articles by quality statistics vs/ Wikipedia:Version 1.0 Editorial Team/Bhutan articles by quality statistics, one can tell that Wikipedia has a better and more in-depth coverage of Iceland—more and higher quality articles about it—despite Togo being much more populated, and that Wikipedia has nearly as much coverage of Bhutan as Togo even as it is 10× less populated. We can similarly gauge how comprehensive our coverage of say, artificial intelligence is by looking at and exploring that project's table.

  For usefulness beyond statistical analysis, I can only really think of Women in Red. It captures the spirit of what a WikiProject is supposed to be—multiple editors coming together and cooperating towards a common goal, rather than just using the project talk page just as a noticeboard. WP:WPVITAL tried doing something similar but wasn't really successful. TryKid ^{[dubious – discuss]} 17:29, 21 July 2026 (UTC)

  Depends on the project. Two of the topic-based ones WikiProject China and WikiProject Plants that I signed up to are pretty active in terms of conversations on the talk page. And one of the maintenance ones WikiProject Unreferenced Articles is also pretty active; very active for the three month-long backlog drives we run every year. Shamelessly opportunistic plug: you can sign up for the imminent August backlog drive right here. Cheers, SunloungerFrog (talk) 17:40, 21 July 2026 (UTC)

  Wikipedia:A WikiProject is a group of people who want to work together to help Wikipedia. This sometimes takes the form of specific goals, such as the backlog drive that SunloungerFrog linked to, but often it just means helping each other out informally. "Joining" isn't nearly as important as "participating", and most participation takes the form of providing advice ("What should I do with this?") and assistance ("Thanks for the note. I've handled it").

  As WikiProjects' talk pages are one of the places that people post to when they have questions, I recommend that experienced editors put multiple WikiProject pages on their watchlist. The quickest way to find those groups for most editors is to go to the main article(s) central to your area of interest, and see which WikiProjects are listed in the talk page's top matter. WhatamIdoing (talk) 16:39, 24 July 2026 (UTC)

  As a new editor I can absolutely see the potential value in projects, but I've ran across plenty that are in fact more or less abandoned, despite plenty of active editors with interest in the subject. What I don't know is why, or what can be done about it. I suspect it's related to the overall structure not being very effective at promoting engagement. Maybe some kind of central project hub would help. This proposal seems like it has some promise, if anyone follows up on it, but I'm picturing some kind of personalized feed showing activity from ALL the projects someone has chosen to follow, without having to jump around to look at each one. I think there's a lot of potential for improvements to userspace that would make it easier for people to track various parts of WP space that they're interested in. The current watchlist format is basically just fancy bookmarks. ChompyTheGogoat (talk) 02:52, 25 July 2026 (UTC)

  Mostly the problem is that to have a visibly active group, it needs to be very large (50+ editors), but the people who want to start them want niche subjects. We're recommending that we have a big Wikipedia:WikiProject Television (only 33 editors have it on their watchlists and actually looked at its talk page at least once during the last month), or even a broader subject (Popular culture? Media?), and they want to start WikiProject My favorite TV show which only aired in a non-English-speaking country for two seasons several years ago and nobody except me remembers it.

  Especially when newcomers try to start the pages, they tend to have a build it and they will come mentality: if the pages exist, then suddenly all five of the people on the entire internet who obsess about this show will show up at Wikipedia and dedicate their lives to expanding all eight of the articles. When we tell them that they have to recruit the editors first, because a WikiProject is a group of people and not a collection of pages (and cleaning up a bunch of scattered templates and categories when the group fails is a pain), they can't do it. They don't know how to recruit editors.

  Wikipedia:Database reports/New WikiProjects is the easiest way to find inappropriate "WikiProject" page creations. Redirects (including to user space) are fine. Pages created by one or two people can be userified. If the group's not going to fail quickly, then we need to see 6–10 editors involved from the beginning, and at least one needs to be very experienced with WikiProjects.

  I've estimated that merging up the groups until there's something big enough to be functional could be a full-time job for more than a year. Once you've done a couple of them, it's not super difficult, but IMO it's not fun work. WhatamIdoing (talk) 05:42, 25 July 2026 (UTC)

  I'm not talking about niche projects. I mean things like WikiProject Fungi or WikiProject Cities, that seem large and certainly have plenty of activity on related articles, but very little collaboration on the project itself. It seems to be an inherent issue with the structure of projects, not whether or not a specific topic has interest. ChompyTheGogoat (talk) 01:35, 26 July 2026 (UTC)

  When I go to WT:FUNGI, I see that people who ask a question can realistically expect a knowledgeable reply in less than a day. It's also not always the same person, which is a sign of there being an actual group there. It's on the watchlist of at least 25 active editors, which is another sign of people working together. What makes you think that there's something wrong with this group? What were you expecting to find? WhatamIdoing (talk) 04:13, 26 July 2026 (UTC)

  There's only a handful of questions, several of which haven't gotten useful responses, and plenty of past threads have been archived with no replies. ChompyTheGogoat (talk) 05:19, 26 July 2026 (UTC)

  I'm looking at this revision. There are 8 discussions. Six have received a reply. Two did not. One of those two explicitly said not to reply at WT:FUNGI. I will not attempt to judge whether the replies were subjectively "useful", but I think that is a good number of responses. WhatamIdoing (talk) 23:02, 27 July 2026 (UTC)

  I feel like it's usually me or Jts1882 who are answering questions on WikiProject Fungi (or most of the other organismal WikiProjects; I have them all watchlisted, and Jts1882 seems to have them all watchlisted as well). There are a couple other editors who monitor several organismal WikiProjects, but if Jts1882 and I quit editing I think many more questions would go unanswered. And most of the projects where we answer questions aren't directly in our editing interests. The only organismal projects that regularly have talk page threads with more than three editors participating are Tree of Life, Palaeontology, Plants and Birds. Plantdrew (talk) 18:20, 28 July 2026 (UTC)

  As someone who asked a question on WT:FUNGI a few days ago, and got a detailed and well-explained answer barely half a day later, I think their current activity level is just fine. I've also used other related Wikiprojects multiple times in the past as a way to get assistance with editors knowledgeable in a topic, and they have always been helpful. Having a place to ask questions to people familiar with a topic is invaluable.

  That being said, some of the less active Wikiprojects should probably be merged. ARandomName123 (talk)^{Ping me!} 04:28, 27 July 2026 (UTC)

  Um... Which less active Wikiprojects then? George Ho (talk) 04:35, 27 July 2026 (UTC)

  Most of the ones in Category:Semi-active WikiProjects. Wikipedia:WikiProject Knots as an example. ARandomName123 (talk)^{Ping me!} 04:41, 27 July 2026 (UTC)

  Some questions getting answers only proves that it isn't entirely dead. It still doesn't help the fact that many others don't. In one case (re cities) there was an ongoing debate over various aspects of an article that I was attempting to help mediate, and we were trying to get outside input for how people who work on city articles felt that they should handle the issues that were in question, to maintain consistency. In my short time here I've seen many other people point out that projects seem largely abandoned and they've rarely gotten help from them, at least in recent times. ChompyTheGogoat (talk) 04:38, 27 July 2026 (UTC)

  I'm fine with closing the ones that don't get any activity (ie. no discussion). The ones that do get activity should stay open. Since WPCITIES had some discussion in the previous archive only a few months ago, I think it should stay open. ARandomName123 (talk)^{Ping me!} 04:44, 27 July 2026 (UTC)

  I'm not saying we should close them. I'm saying we should try to promote more engagement (at least for the ones that aren't overly specific) so we can answer the questions that DON'T get replies and get more overall collaboration on the subjects. ChompyTheGogoat (talk) 04:48, 27 July 2026 (UTC)

  Maybe some kind of central project hub would help Like Wikipedia:WikiProject Directory? Or Wikipedia:WikiProject Council/Directory? Cheers, SunloungerFrog (talk) 05:56, 26 July 2026 (UTC)

  As I said, it would be more than just a simple list - more of an interactive feed to make it easier for people to monitor and engage with the projects that they've already joined. I don't think FINDING them is really the problem - although it wouldn't hurt if there was also some kind of automated suggestion to add articles to relevant projects. ChompyTheGogoat (talk) 06:08, 26 July 2026 (UTC)

  Hmm. Maybe you might find one of the Watchlist-related user scripts on WP:USERSCRIPTS a useful addition for your workflow. For me, a combination of the standard Watchlist and selective subscription to individual threads suffices for monitoring what I'm interested in on Wikipedia. I appreciate that you find the Watchlist deficient, but I'm not quite sure in what way. There's also c:User:Jack who built the house/Convenient Discussions for talk pages. Cheers, SunloungerFrog (talk) 06:45, 26 July 2026 (UTC)

  I'm talking about broad changes to help improve project engagement wiki-wide, not just for my personal needs. That's what idea lab is for, right? ChompyTheGogoat (talk) 06:49, 26 July 2026 (UTC)

  Well, and to highlight existing solutions that might help. Maybe you could expand on the nature of the hub or interactive feed that you envisage. Cheers, SunloungerFrog (talk) 07:00, 26 July 2026 (UTC)

  Something more like notifications, that could have grouped previews for all the project pages someone follows - such as "Wikipedia:WikiProject Fungi#Article alerts has been updated", "Wikipedia talk:WikiProject Cities has had # discussions created or updated" <collapsible list of each new/modified section, similar to how multiple comments to a discussion are shown under a top level notification

  If someone is in a dozen projects, this could allow them to easily skim any recent changes to all aspects of their various projects (with the ability to modify which parts they're subscribed to) and see what they might be interested in, without having to click around and view everything one at a time. ChompyTheGogoat (talk) 07:20, 26 July 2026 (UTC)

  On a related note, it's a pity that Mediawiki Flow (and more at Wikipedia:Flow) from a while ago never reached fruition, because it looked to do a lot of what you are proposing. Cheers, SunloungerFrog (talk) 05:12, 27 July 2026 (UTC)

  You might be interested in the system shown at Wikipedia:WikiProject Medicine/Discussions. WhatamIdoing (talk) 06:58, 27 July 2026 (UTC)

  Yeah, that's exactly the kind of thing I had in mind; just more centralized to track various projects simultaneously. People can easily skim for anything that catches their interest on subjects they've already chosen to follow - potentially with different sort options (chronological, grouped by project, etc). ChompyTheGogoat (talk) 07:04, 27 July 2026 (UTC)

Making Multiple issues's tag color change depending on the highest tag ranking of any of the issues under it

I have opened a discussion on Template talk:Multiple issues to change the tag color of the template to whatever is the most severely colored tag in the issue templates under it. Right now, it defaults to orange. Under my proposal, it would be yellow if all the issues are yellow, orange if at least one was orange, and red if at least one was red. This is an issue since tag colors are often used as a quick shorthand to deal with article quality, and the standard orange template can frequently underestimate or overestimate the problems of an article (especially if it has a lot of issues). I am posting here to get broader community input (and specifically in the idea lab section since I'm not sure how technically feasible this is). — Knightoftheswords 05:16, 27 July 2026 (UTC)

This comment indicates that it's not feasible, so it may not be worth discussing.

I don't think this is important. I wouldn't object if it happened; I just don't care. WhatamIdoing (talk) 15:34, 28 July 2026 (UTC)

Quick note per my response in the discussion, I have tested an implementation and it is possible. FlammablePizza (talk) 18:45, 28 July 2026 (UTC)

What should Mentorship be?

As it currently stands it's simply a version of Teahouse. A mentee can choose to ask a question and it will be asked on your talk page for you to answer. There is an argument that allowing that instead of just having the question be asked at the teahouse makes it more comfortable since it is one-on-one.

However, as asilvering stated: I've wondered about this myself. I'm concerned that the mentorship program is not actually beneficial, on the whole, to mentees or to mentors. Mentees can get help faster at the Teahouse, and if someone answers incorrectly it is relatively likely to be corrected. Not so on a mentor's talk page.

The questions are not exactly common. I have 82 mentees assigned and have been asked two questions.

From what's been said at Mentorship noticeboard, that's about a normal amount of questions. The topic of the questions is a bit abnormal though: most are about how to write autobiographies it seems.

I'm not sure how much time/money/etc. the WMF continues to put into maintaining Mentorship, it could be negligible.

Ultimately, I'm not sure whether it should be left to just sit as is, or be changed into something else. Given that the dashboard provides a link directly to the mentee's contributions I am wondering if mentors should also be considering reviewing those. I'm also wondering what ideas may come up from those who haven't looked at it and don't realize just how little is involved. Jerod Lycett (talk) 06:11, 21 July 2026 (UTC)

Speaking as a mentor, I think that the option for a newly registered editor to contact a named experienced editor who has offered to help 1-1 is useful to have, particularly for those who might feel less comfortable posting in a heavily trafficked forum like the Teahouse. My experience is that most interactions are question then answer then nothing, which is fine. Sometimes the questions are irrelevant, in which case I ignore them, or reply with something like "Let me know if you have any questions about editing Wikipedia". A few involve a longer back and forth, which is fine too. Personally I would not choose to review mentees' contributions as a matter of course - that feels a bit stalkerish to me, and given the sheer number of mentees one accumulates, not really feasible to do across the board. Ultimately, I think it's up to the mentee how much they choose to interact, not the mentor, and for the mentor to respond in accordance with that. Cheers, SunloungerFrog (talk) 10:01, 21 July 2026 (UTC)

People vary enormously in how much mentoring (menting?) they want/need. I don't think they should be squeezed into a one-size-fits-all model. But please listen to people who have experience of this, from both sides, more than to me. I do not have the patience to be a mentor and I'm too set in my ways to be a mentee. Phil Bridger (talk) 19:08, 22 July 2026 (UTC)

As a mentor and mentee, I find mentorship very important. My mentor has been extremely helpful and explained a large amount of information. Having the option to get mentored is very important. In solidarity Dafootballguy Want to talk? 23:33, 23 July 2026 (UTC)

I have not signed up to be a mentor in large part because those linked questions are the sort I see pop up on my watchlist on the talkpages of other users. It almost feels like if I see an account with a talkpage redlink post to a user talk, it's a coin flip between asking to create an autobiography and random nonsense or test editing. (Having mentors actively hunt through their mentees contributions seems like it's asking for a lot more effort than intended, and also would turn mentorship into more of an enforcement role.)

The Teahouse seems like it would be a better location for almost all of the questions (using a very broad definition of the word "questions") I see on my watchlist. However, the Teahouse currently has a "Meet your hosts" button. Why not a "Meet possible mentors" button? Introductory questions at the Teahouse, if someone wants to go more in depth in a less public space, the one-on-one option is available. CMD (talk) 00:42, 24 July 2026 (UTC)

As I mentioned in an above discussion (which I'm not going to attempt to dig through again, but it's there if anyone wants to look), I've found Teahouse far more helpful as a new user, and have never contacted my own mentor for that reason. It might be more useful to add some crosspost/auto-tag feature, so that people can ask the question on Teahouse and their mentor can be notified that the mentee has a question. There are times when it's helpful to have more of a one on one dialogue, and there could still be the option to ask directly on a TP, but for me it's usually been more a case of spinning off to talk to someone directly on either my TP or theirs after they responded to my original question. You're polling a wide pool of experience by asking at Teahouse so there's a much better chance that one of the many people who sees your question will know about it, even if it's highly specific - in addition to other new editors being able to see and learn from the answer(s) too. ChompyTheGogoat (talk) 03:03, 25 July 2026 (UTC)

How are mentors assigned? I never got one. (I also don't have very many questions, but I also do a unusal ammount of reading policies, to the point where I have spent more time reading stuff in the Wikipedia mainspace than editing in the article mainspace). Velocifyer (talk) 14:32, 29 July 2026 (UTC)

@Velocifyer You can see your mentor at your Special:Homepage, if you have that enabled. CMD (talk) 14:36, 29 July 2026 (UTC)

@Velocifyer, your mentor is currently AlphaBetaGamma. They were almost certainly assigned to you when you created your account, or possibly when you explicitly opted in to mentorship. Can you see them in the Mentor module on Special:Homepage? Cheers, SunloungerFrog (talk) 19:45, 29 July 2026 (UTC)

I don't see them on [[Special:Hompage]. Velocifyer (talk) 21:06, 29 July 2026 (UTC)

You might have the newcomer page disabled in your preferences. ChompyTheGogoat (talk) 21:31, 29 July 2026 (UTC)

There likely needs to be a 'project' type thing for missing towns and settlements

I was browsing the pages on Florida county roads -don't ask- and saw a lot of red links. Mostly for towns and settlements, too. I always thought every town that has people and every former town worth talking about had an article, but alas I was proven wrong. How naive of me but it is what it is.

Since there are thousands of settlements in my state (Georgia, as seen on my user page) alone, and hundreds of thousands across the USA and Europe, many of which have no representation or knowledge outside of natives and family of natives, there likely needs to be a sort of WikiProject in this area.

I wouldn't be too surprised if this exists already and I simply never heard of it, and I appreciate opinions on the matter. This is just the ramblings of a nerdy parakeet though and if this didn't take up any steam then it's nothing new for me. Have a great day editors! TheNerdy Parakeet (talk) 21:23, 29 July 2026 (UTC)

Update- upon doing some actual research instead of saying my thoughts and seeming not knowledgeable I found Wikipedia:WikiProject Cities and now my words of 'there needs to be a project for this' look foolish, at least to me. But I found that my take is not covered in the project, so at least I had a semi-original thought. TheNerdy Parakeet (talk) 21:34, 29 July 2026 (UTC)

(edit conflict) WP:Wikiproject Cities exists. There needs to be enough material on a given town to be able to write about it though, as well as an editor who cares enough to bother. That's likely not the case for many small towns. ChompyTheGogoat (talk) 21:35, 29 July 2026 (UTC)

WP:WPGEOG also exists, as do multiple related descendant projects. Wikiprojects in my experience suffer from low participation. Editors generally spend more time actually editing the articles they are interested in, and the projects provide some essays and guidance, but little in the way of collaborative assistance on specific articles. GeogSage ^{(⚔Chat?⚔)} 21:44, 29 July 2026 (UTC)

WP:NPLACE (aka WP:GEOLAND and other shortcuts) has notability guidelines for settlements. I believe it has caused some consternation in the past when large numbers of stubs of were created for tiny settlements. WikiProjects are a mixed bag. Some are quite active, most are not. You might have better luck looking at more locally based projects focused on a state or country. Wikipedia:WikiProject Georgia (U.S. state) appears inactive but Wikipedia:WikiProject Florida has quite a bit of activity on its talk page. —Myceteae‍🍄‍🟫 (talk) 22:09, 29 July 2026 (UTC)

Thanks for all the tips! I appreciate the help and will probably use these WikiProjects to further my idea. Happy editing! TheNerdy Parakeet (talk) 13:49, 30 July 2026 (UTC)

You're welcome, happy editing to you! —Myceteae‍🍄‍🟫 (talk) 14:24, 30 July 2026 (UTC)

We need to do something about the scourge of fake footnotes

I don't mean footnotes leading to dead links or fake sources. I mean literally the kind of thing found in, for example, Damas Gratis, bolded for emphasis here:

The music and rhythm of the Damas Gratis group continued to advance until it transcended the limits of tropical music, as reflected in various articles published in the main newspapers of Argentina such as: Clarin, La Nación,[4] El País,[5] Buenos Aires Herald and magazines of the magnitude like Rolling Stone[6] that also dealt with this phenomenon that revolutionized the tropical scene.

That [4], [5], and [6] lead nowhere and refer to nothing discernible. There is no [1], [2], or [3] there, by the way, but there are some later numbers further down. I have seen at least hundreds of articles with notations such as these, almost certainly copied from someone's school paper or external draft. The need to be rooted out and exterminated, and any editor in the habit of adding them needs to be soundly trouted. BD2412 T 03:33, 29 July 2026 (UTC)

For your given example, it seems like the user translated the Spanish Wikipedia article of w:es:Damas Gratis to English (probably by copying/pasting the text into Google Translate or something). Compare the English Wikipedia version with the Spanish Wikipedia version , for instance. Seems like the user didn't carry the sources/references over. Some1 (talk) 03:54, 29 July 2026 (UTC)

It is curious how the user in their edit used wikiformatting for the lv2 header but <big></big> tags for the lv3 headers. Nonetheless, it seems the user was a very short-lived SPA, a trouting may not be useful. It does seem like detecting this could be a useful maintenance cat, most instances are going to be copying, either from someone's school paper or from another wiki. CMD (talk) 05:14, 29 July 2026 (UTC)

I am sorry but the subliminal mentality in that message is that we human beings need to monitor these footnotes. Alas we do not have enough troops to do so. But a bot could do so for sure, without significant effort. However some of those who can write such bots are busy dealing with useless discussions. So please invoke WP:HEAR to end that discussion then talk Anome into writing the bot. Twenty years ago I would have written the bot myself, but not now. I need to go back to my rocking chair. So please rely on bots. Cheers. Yesterday, all my dreams... (talk) 05:07, 29 July 2026 (UTC)

In WikiProject Unreferenced Articles, ARandomName123 runs a bot to detect "probably unreferenced articles" (see the full list in PetScan here) and we include those in our backlog, to be dealt with in due course. In the past, I have come across a few articles like this in the "probably unreferenced articles" list. It is interesting though that this particular one is not in the list. @ARandomName123, if you had the opportunity, would you mind looking at why this one hasn't been caught in our net? Maybe there is some logic that could be tweaked. Cheers, SunloungerFrog (talk) 05:27, 29 July 2026 (UTC)

We could have a bot search for and flag articles with fake footnotes. There could be rare false positives (e.g. Hyperoperation), however. –LaundryPizza03 (dc̄) 06:10, 29 July 2026 (UTC)

Would the bot just tag them or would it revert the edits? Velocifyer (talk) 14:15, 29 July 2026 (UTC)

Reverting would be difficult, as some of these edits have existed for quite a while. Adding the {{Format footnotes}} template would likely be best, as its documentation says to add it to pages containing "manual numbers" instead of proper citations. FlammablePizza (talk) 14:32, 30 July 2026 (UTC)

In addition, we'd have to cap the number inside the brackets at 3 digits, since years in square brackets are a standard citation format in British law. Fortunately, articles with 1000 or more distinct numbered references are exceedingly rare; I'm not sure if any exist on enwiki. For comparison, the current revision of United States has 686. –LaundryPizza03 (dc̄) 06:18, 29 July 2026 (UTC)

Enwiki has 5 articles and ~25 lists with over 1000 references, per WP:MOSTREFS. There is even one list with over 2000 references, so while exceptionally rare (about 0.00004% of articles), they do exist. I don't think that British legal citations would be particularly prevalent, and would arguably be to similar to parenthetical referencing, which is a deprecated format per WP:PAREN, and thus should be removed anyway. Perhaps in the case of false positives, a template could be created to exclude the triggering content from the bot in the form of {{!fakeref|content}} or similar. Mitchsavl (talk) 07:48, 29 July 2026 (UTC)

Even so, I would seriously doubt that articles exist with more than a few hundred fake footnote indicators. BD2412 T 12:20, 29 July 2026 (UTC)

While I agree it is highly unlikely for an article to have a few hundred fake footnotes, I would not be surprised by an article with a few hundred references, but only one or two fake ones from a recent addition. Mitchsavl (talk) 12:49, 29 July 2026 (UTC)

Damas Gratis has multiple URLs in the external links section, all of which trip the bot as a potential reference. ARandomName123 (talk)^{Ping me!} 06:19, 29 July 2026 (UTC)

For that one, I reverted to a 2023 revision before a massive overwrite with zero references. It was likely machine-translated from eswiki using an LLM. –LaundryPizza03 (dc̄) 06:33, 29 July 2026 (UTC)

Thanks for adding the expand language tag as well. Haven't seen an llm leave references like that, although it is possible, but I've seen that before for simple google translate. CMD (talk) 09:53, 29 July 2026 (UTC)

It happens from time to time, although it's certainly not the most common way they mess up footnotes. (Vaguely seems to be more common with some LLMs than others, but haven't nailed down specifics yet.) Also something that humans can do on their own.

(The edit here doesn't seem like it involved AI, for what it's worth.) Gnomingstuff (talk) 13:10, 29 July 2026 (UTC)

These are almost always an indicator that the article was AI-generated. sapphaline (talk) 12:25, 29 July 2026 (UTC)

Or as Some1 notes above, translated/copy-pasted between windows or programs without paying attention. -- Reconrabbit (talk) 13:39, 29 July 2026 (UTC)

Using an insource: search I find a lot of strange style choices using [], they include Bible verses, coordination numbers, symmetry group names, tennis seed positions. However World Surf League seems to have fake foot notes in tables. 2C-B and War crimes in the Syrian civil war have useless footnotes in quotes included in references. 1346 in Ireland has semifake footnotes, needs to convert to use ref tags. Graeme Bartlett (talk) 13:13, 29 July 2026 (UTC)

would be a better search, though there's a ton of false positives. sapphaline (talk) 13:19, 29 July 2026 (UTC)

has few false positives but only finds 84 cases. It may be useful to find genuine problems which this search misses and improve it to cover them. Certes (talk) 17:28, 29 July 2026 (UTC)

This is helpful, thank you! I've run into "[1]" style fake footnotes on articles with content copy-pasted from Grokipedia, so I'm interested in looking for more of that. This type of search is great in combination with mw:Who Wrote That?. Dreamyshade (talk) 20:29, 29 July 2026 (UTC)

1346 in Ireland is interesting: its fake footnotes started out as useless wikilinks to numbers such as [[1]]. I removed bad links to small integers some time ago. Those that remain do refer to the integer or glyph, though in most cases we could reasonably have assumed that the reader already knows what 1 is. Certes (talk) 21:38, 29 July 2026 (UTC)

I would argue that (assuming the source actually supports the article) that it should be treated as a general reference. In references it says [1] to [16] are all from the same source. It says nothing at all about reference [17], which seems odd. Mitchsavl (talk) 21:56, 29 July 2026 (UTC)

World Surf League gained many of its fake footnotes in this revision. They don't seem to match any of the actual (unnumbered) footnotes present at the time, but do look convincing in their <sup>...</sup> tags. I think it's simply the number of times the individual has won the event: ^[2] means second win, etc. WikiBlame was useful here. Certes (talk) 21:54, 29 July 2026 (UTC)

2C-B (a drug) has [4] etc. in quotes within the references section. The quotes are unusually long and may be lifted straight from the cited paywalled source, which presumably has a Wikipedia-style reference section of its own (not included). So they look like references used by cited works. The main issue here may be whether we want or are allowed to include such long verbatim quotes. Certes (talk) 22:11, 29 July 2026 (UTC)

In War crimes in the Syrian civil war, reference 33 is itself a list of numbered references. It's not clear which of them, if any, support the article's claims marked with a real ^[33] footnote. I'll stop now, as sadly it seems that every article has a different problem and there's unlikely to be a one-size-fits-all solution which could be semi-automated. Certes (talk) 22:14, 29 July 2026 (UTC)