Wikipedia:Administrators' noticeboard

Welcome — post issues of interest to administrators.

• For urgent incidents and chronic, intractable behavioral problems, use Wikipedia:Administrators' noticeboard/Incidents.
• To request review of an administrator's action or other use of advanced permissions, use Wikipedia:Administrative action review.
• If you are new, try the Teahouse instead.
• Do not report breaches of personal information on this highly visible page – instead, follow the instructions on Wikipedia:Requests for oversight.
• For administrative backlogs add {{Admin backlog}} to the backlogged page; post here only if urgent.
• Do not post requests for page protection, deletion requests, or block requests here.
• Just want an admin? Contact a recently active admin directly.
• If you want to challenge the closure of a request for comment, use {{subst:RfC closure review}}

When you start a discussion about an editor, you must leave a notice on their talk page. Pinging is not enough.

You may use {{subst:AN-notice|thread=}} ~~~~ to do so.

Sections inactive for over seven days are archived by Lowercase sigmabot III.(archives, search)

Start a new discussion

Open tasks

XFD backlog

V	Nov	Dec	Jan	Feb	Total
CfD	0	12	160	0	172
TfD	0	0	73	0	73
MfD	0	0	9	0	9
FfD	0	0	48	0	48
RfD	0	0	25	0	25
AfD	0	0	0	0	0

• 7 bot-reported usernames for administrator attention
• 7 user-reported usernames for administrator attention
• 13 bot-generated requests for intervention against vandalism
• 4 user-reported requests for intervention against vandalism
• 88 sockpuppet investigations
• 48 Candidates for speedy deletion
• 3 Fully protected edit requests
• 0 Candidates for history merging
• 3 requests for RD1 redaction
• 33 elapsed requested moves
• 1 Pages at move review
• 61 requested closures
• 70 requests for unblock
• 0 Wikipedians looking for help from administrators
• 13 Copyright problems

Request review of informal COI topic ban on Johann Hari

I previously posted here, on the Admins noticeboard, requesting Wikipedia administrators review and issue formal decisions on three informal bans that were imposed on me on the COI noticeboard.

1) A citation ban that forbids me from citing my own books and newspaper / magazine articles. After discussion on the Admin noticeboard, I consider this matter closed.

2) A topic ban on Louise Vincent. After discussion on the Admins noticeboard, I consider this matter resolved.

3) A topic ban on Johann Hari, which was proposed and then agreed upon by four editors on the COI noticeboard. I oppose this topic ban. The original discussion thread on the Admin noticeboard was closed after issues #1-2 were resolved, but before the Hari matter could be discussed. I am therefore opening this new discussion thread to address the issue of my informal topic ban on Hari. I am requesting a formal decision in the hope that I will be cleared of a COI regarding Johann Hari.

I edit Wikipedia under my real name and have always been transparent about my past professional interactions with Hari. I will describe all facets of those interactions here.

In 2017, I wrote a book called Fighting for Space (FFS). In researching that book, I contacted Hari and requested audio tapes of interviews he conducted with a deceased individual. He sent them. Hari also wrote a short blurb for FFS’s book jacket. I thanked him in the book’s acknowledgements section.

In 2021, I wrote a book called Light Up the Night (LUTN). Hari wrote a short blurb for LUTN’s book jacket. I thanked him in the book’s acknowledgements section.

I’ve also interviewed Hari for two magazine articles. The most-recent such article was published in 2018.

This is the totality of my relationship with Johann Hari.

Regarding allegation of a professional conflict of interest, as detailed on the COI noticeboard, I want to emphasize that I no longer work in journalism and do not consider myself a working author. I have not had a single byline published anywhere in more than five years. Hari and I do not work in the same profession; therefore, I question the extent to which a professional conflict of interest can exist. I’ll also emphasize the highly limited impact that a blurb has on a book’s sales, and note blurbs are routinely made among authors of a shared field with little consideration or impact. In addition, both FFS and LUTN are no longer in print. This was all years ago now. Ergo, regarding the Hari blurbs, even any limited influence on sales that might have once existed can no longer be considered anything but negligible. I therefore declare no professional conflict of interest.

No accusation of a personal conflict of interest was ever made. Regardless, I will emphasize here, I have no personal relationship with Hari. We are not friends, and I have no communication with him. I declare no personal conflict of interest.

I argue that a COI topic ban should not be applied for any past level of acquaintance to a topic regardless of how small the degree. COI topic bans should only be applied when an actual COI exists. I argue that my past contact with Hari does not meet the bar for a COI topic ban.

I request a formal decision on my informal topic ban on Johann Hari. I will tag Wikipedia admins who expressed an interest in this discussion in response to my original post on the Admin noticeboard; @Voorts, @Phil Bridger, @The Bushranger. And I respectfully encourage admins to declare no conflict of interest. Thank you. Tlupick (talk) 00:03, 14 January 2026 (UTC)

I've started two community sanction discussions below. The broader community, not just admins, will decide whether a self citation restriction and/or topic ban are required. voorts (talk/contributions) 00:57, 14 January 2026 (UTC)

Permalink to previous AN discussion, in case it gets archived. voorts (talk/contributions) 00:59, 14 January 2026 (UTC)

Proposal: Self citation editing restriction (Tlupick)

Tlupick (talk · contribs) is indefinitely restricted from citing his own work. voorts (talk/contributions) 00:54, 14 January 2026 (UTC)

Support. I am generally averse to allowing editors cite their own work; I would argue WP:SELFCITE shows that the broader community is too. A topic ban from citing his own work seems like it would resolve most of the problems here. Toadspike [Talk] 18:38, 17 January 2026 (UTC)

@Voorts, @Toadspioke, respectfully... Wikipedia does not have an outright ban on anyone ever citing their own work. It discourages the practice. It does not ban it. Therefore, if this specific restrictive action is going to taken against me -- an indefinite restriction on citing my own work -- there should be a clear reason for it. I do not believe one has been put forth.

I have cited my own work in the past, but not to an egregious degree. In addition, where I have cited my own work in the past, I have learned from those experiences, and I now fully agree with Toadspike where they say, "I am generally averse to allowing editors cite their own work." That said, my books contain substantial original research in the areas of the overdose crisis, harm reduction, and histories of drug-user organizing, and I see value in my reserving the right to cite my own work in rare cases when there is no other public source available for citation.

I do not believe this is unreasonable. And I do not believe there is sufficient justification for a ban that forbids me from doing so. That is, I do not believe there is sufficient justification for Wikipedia admins to single me out and impose a ban on self citation that goes a step further than general Wikipedia guidelines. If there is sufficient reason for a ban, it should be presented here. Otherwise, this action would appear arbitrary. Tlupick (talk) 23:36, 20 January 2026 (UTC)

You haven't been singled out by Wikipedia admins. After the last thread was archived, you asked for an evaluation of the proposed topic bans by admins and a review of other editors' concerns about your use of own sources and edits related to Johann Hari. I've obliged you and started this discussion. Other members of the community will evaluate these proposals, note their position, and the discussion will be closed based on whatever consensus develops. Toadspike's !vote is presumably based on the evidence presented at the prior discussion and COIN, hence why he said that he believes it would resolve most of the problems here. voorts (talk/contributions) 23:48, 20 January 2026 (UTC)

@Voorts forgive me for imprecise language. If a self-citation topic ban is imposed on me by admins without specific justification or explanation, I argue that that would be singling me out. I am simply requesting that if a self-citation ban is imposed, that someone point to specific reason(s) for that ban. Without specific reasons for a ban, I argue that that action would appear arbitrary. Tlupick (talk) 00:39, 21 January 2026 (UTC)

Oppose It seems to me based on the given statement that Tlupick doesn't contest this. "I consider the matter resolved". Implementing something here that no one disagrees with seems unnecessary. Buffs (talk) 18:16, 21 January 2026 (UTC)

Oppose struck for now as I have not assessed this matter. Will update when able. Buffs (talk) 19:58, 21 January 2026 (UTC)

Returning to opposition. Tlupick has clarified and stated he generally will not and I see no reason to doubt him. I would caution him that if he uses such sources, he should clearly cite that in the edit summary with some sort of stipulation such as "I am the author of this published work. If you feel it is in breach of WP:V feel free to revert and we can discuss" or something to that effect...an author like you should be able to come up with something more concise. Buffs (talk) 22:58, 22 January 2026 (UTC)

@Buffs wrote, "I would caution him that if he uses such sources, he should clearly cite that in the edit summary with some sort of stipulation such as "I am the author of this published work. If you feel it is in breach of WP:V feel free to revert and we can discuss" or something to that effect." -- I am happy to abide by this suggestion. Tlupick (talk) 23:07, 22 January 2026 (UTC)

Tlupick does disagree. He considered the matter resolved after asking that we sequentially discuss these issues. We never got through the issue of self citations at the prior AN thread, so it is not, in fact, resolved. voorts (talk/contributions) 18:31, 21 January 2026 (UTC)

@Tlupick: could you clarify? Buffs (talk) 19:11, 21 January 2026 (UTC)

@Buffs On the Wikipedia:Conflict of interest/Noticeboard, it was proposed and subsequently agreed by other editors that I receive a citation ban that forbids me from citing my own books and newspaper / magazine articles. It is my understanding that decisions on the COI noticeboard are informal. I have therefore sought clarification and a formal ruling on the Admin's noticeboard (here) that I be cleared of a self-citation ban. I do not believe that anyone has ever has put forward specific reasons for a self-citation ban to be imposed, and so one should not be applied. Therefore, in order to remove confusion or ambiguity caused by an informal ban, I kindly request that admins decide here that one is not justified and not imposed. Thank you. Tlupick (talk) 19:25, 21 January 2026 (UTC)

There's no such thing as an informal ban. COIN isn't the forum for imposing TBANs by the community. AN is. I appreciate that you want this addressed. Hopefully a consensus one way or the other will be reached here. voorts (talk/contributions) 19:55, 21 January 2026 (UTC)

I would ask you to additionally clarify that above then. Buffs (talk) 19:57, 21 January 2026 (UTC)

I am confused. You stated at the WP:COIN that I have no objection to a ban on citing my own work [1]. Are you changing you position from there?

As for the edits in question at the WP:COIN, I'm not sure that they really add any bright-line policy-violating material, but I would suggest refraining from citing your own work for a number of other reasons. This edit [2], for example, adds material about political consequences that are verified by other WP:RS (see e.g. [3]; Owen’s approach to drug users was at odds with many of his colleagues and his political party. In 2002, the NPA selected a different person to run as mayoral candidate in the city’s municipal election.) The anecdote about friends and his wife Brita not supporting him isn't supported by other sources, e.g. [4], which states After leaving the mayor’s office, Mr. Owen continued to advocate for harm reduction and supervised-injection sites, with his wife, Brita, staunchly supporting him, but this could also be the bias of obituaries rearing its head. This article [5] does seem to confirm those claims about his friends; Mayor Owen faced a great deal of criticism, not only from his constituents, but even his own colleagues and friends. Katzrockso (talk) 03:41, 22 January 2026 (UTC)

You stated at the WP:COIN that I have no objection to a ban on citing my own work [13]. Are you changing you position from there? No, he changed his mind and then opened the prior AN thread. voorts (talk/contributions) 03:48, 22 January 2026 (UTC)

I do remember that thread now, thanks for reminding me. I read it at the time, but never commented. Katzrockso (talk) 04:19, 22 January 2026 (UTC)

@Buffs and @Katzrockso, as @Voots clarified, I changed my mind about a self-imposed self-citation ban and then opened the prior AN thread. I understand this caused confusion. My apologies for that.

I'll now emphasize two points:

1) I do not believe that anyone has put forward sufficient examples of where I have engaged in excessive self-citation to warrant a ban. (I agree that @Katzrockso’s example detailed above from the Philip Owen page comes close. That was a long time ago, and I have learned from that.)  Therefore, I suggest a ban is not justified.

2) I have come to wholeheartedly agree that I should refrain from citing my own work on Wikipedia, as multiple admins here have suggested. That is how I intend to edit going forward. I only oppose a total self-citation ban for the very rare possible instance when I may find that no other public source is available. If admins here vote to oppose a ban, I promise to exercise this option very rarely, if ever. Tlupick (talk) 22:27, 22 January 2026 (UTC)

I can accept that. Oppose restored Buffs (talk) 22:58, 22 January 2026 (UTC)

Oppose. I do not believe that anyone has put forward sufficient examples of where I have engaged in excessive self-citation to warrant a ban. Therefore, I suggest a ban is not justified. In addition, I agree that I should refrain from citing my own work on Wikipedia, as multiple admins here have suggested. I pledge here, that is how I intend to edit going forward. I only oppose a total self-citation ban for the very rare possible instance when I may find that no other public source is available. If admins here vote to oppose a ban, I promise to exercise this option very rarely, if ever. Thank you. Tlupick (talk) 19:28, 29 January 2026 (UTC)

Proposal: Topic ban from Johann Hari (Tlupick)

Tlupick (talk · contribs) is indefinitely topic banned from making edits related to Johann Hari, broadly construed. voorts (talk/contributions) 00:55, 14 January 2026 (UTC)

Despite the well-written appeal by Tlupick, I have to support this topic ban as well. My support for this topic ban does not hinge on whether Tlupick has a conflict of interest with Hari "on paper". Taking the background given above at face value, I would generally say that an editor who has had minimal contact with a BLP subject and none in the past ~5 years probably doesn't have a conflict of interest. However, Tlupick's edit history shows that he is unable to edit neutrally on the subject of Johann Hari.

Tlupick's most recent edit to the Johann Hari article is [6], which has the summary "Removed salacious and unnecessary sentence about Hari's personal life as a gay man." The source cited for the sentence in question is an article by Johann Hari. Clearly Hari himself had no problem sharing (proudly flaunting?) this information publicly. I don't see why we should then refrain from including it, and I especially take issue with calling it "salacious". Most of Tlupick's previous dozen or so edits to the page consist of revert-warring with several other editors. An example is [7], which was removed as "WP:WEASELish promotional language" – I agree that "popular programs" is promotional, while "frequently participates" and "regularly appears" are also very subjective. All of the sources cited are primary (the programs he appeared on), which is completely insufficient to support such wording. Stuff like [8] is clearly promotional – on its own it would not be concerning, but in the broader context it's a clear example of a pattern of biased editing. Toadspike [Talk] 18:59, 17 January 2026 (UTC)

@Toadspike, thank you for your thoughtful reply. You have written, "My support for this topic ban does not hinge on whether Tlupick has a conflict of interest with Hari ‘on paper’. Taking the background given above at face value, I would generally say that an editor who has had minimal contact with a BLP subject and none in the past ~5 years probably doesn't have a conflict of interest. However, Tlupick's edit history shows that he is unable to edit neutrally on the subject of Johann Hari."

I question whether a COI topic ban should take into account an editor's contributions to a page. Does the editor have a COI or not? Based on pre-existing facts, if they do not, they do not. And that would seem to be your opinion of me and the suggestion of a COI topic ban on Johann Hari.

My edits of the Hari page have been far from perfect. Specifically, they have been more positive than negative, and I agree this could be interpreted as biased. I have also used promotional language, although I believe most instances of promotional language occurred earlier in my editing of the Hari page and can be attribute to a Wikipedia learning curve, and one from which I have learned and since applied those lessons. If my infraction has been to make mistakes and contribute to Wikipedia without first understanding best practices, I should be tutored, not punished with a topic ban.

Furthermore, I argue that any contributions that I have made to the Hari page that are perceived as biased toward the positive have primarily occurred in response to a small number of editors who have brought their own biases to the Hari page in the form of overly negative contributions. These editors have acted as unfair gate keepers and exercised their negative biases by persistently refusing to allow anything positive on the Hari page. They have created an overall picture that significantly departs from a neutral presentation of reality.

Another editor recently summarized one of my primary issues with the Hari page on the Hari Talk page: "I was struck by how prominently past misconduct is foregrounded, while much of his subsequent work, reception, and ongoing professional activity appears either minimized or framed primarily through critique. The plagiarism issue is real and well-documented, and it belongs in the article. But it does not seem to be the entirety of his career or how his work is generally treated in most reputable sources today."

If my contributions to the Hari page are to be weighed in an evaluation of whether a COI Hari topic ban is imposed, I respectfully ask that I first be given the opportunity and time to explain what I believe are significant problems with the Hari page. These problems are what prompted my initial interest in the Hari page, and so are important in understanding why I have attempted to make the edits that I have.

I will explain the allegations I’ve made here in greater detail by Jan 31 (as a self-imposed deadline). Thank you for understanding time constraints, and thank you for your patience and consideration. Tlupick (talk) 00:43, 21 January 2026 (UTC)

Tlupick, a topic ban is one of the tools we use to prevent disruptive editing. The fact that you do or do not have a COI with Hari is one consideration in imposing a topic ban, but not the only one. voorts (talk/contributions) 01:02, 21 January 2026 (UTC)

Furthermore, I argue that any contributions that I have made to the Hari page that are perceived as biased toward the positive have primarily occurred in response to a small number of editors who have brought their own biases to the Hari page in the form of overly negative contributions. These editors have acted as unfair gate keepers and exercised their negative biases by persistently refusing to allow anything positive on the Hari page. They have created an overall picture that significantly departs from a neutral presentation of reality. Please read casting aspersions. I recommend you retract these comments about the motives of other editors. voorts (talk/contributions) 01:04, 21 January 2026 (UTC)

@Voorts I retract them. And I apologize. I did not intend for those remains to cast aspersions. I was only trying to provide a very brief summary of the issues I intend to explain further. I intend to keep the subject of this thread on content, not editors, and I apologize if my preliminary summary of these related issues fell short of that goal. I retract them and I apologize. Tlupick (talk) 19:28, 21 January 2026 (UTC)

Might I recommend you strike the comments then? Buffs (talk) 19:55, 21 January 2026 (UTC)

@Buffs, please forgive my ignorance, but what's the correct procedure to do that? Do I use the "edit" function at the top of this thread and add a strikethrough font over that text? Thanks for your guidance. Tlupick (talk) 20:25, 21 January 2026 (UTC)

yes Buffs (talk) 22:34, 21 January 2026 (UTC)

The sentence that was removed is obviously not WP:DUE for inclusion, it's a trivial factoid that does not belong in a WP:BLP, whether or not the statements were WP:ABOUTSELF or written by a third-party. We should commend @Tlupick for removing such unencyclopedic material from the article rather than impose sanctions on such a basis. As for the term "salacious", the first definition I located stated having or conveying undue or inappropriate interest in sexual matters. I have a hard time coming to the conclusion that the sentence in question In a 2002 article, he stated that he had had sex with men who were members of homophobic far-right and Islamist groups is not salacious. Indeed, a later editor (Drmies) removed all mention of Hari's sexuality from the article.

Otherwise no comment on the merit of a topic ban, but if this is the basis for such a topic ban, I am not convinced. Katzrockso (talk) 03:20, 22 January 2026 (UTC)

At this point, it's been about a week. Care to weigh in with a !vote? Buffs (talk) 20:44, 28 January 2026 (UTC)

• Oppose While a COI might have existed at some point, I concur with the general topical inputs of Tlupick, though not necessarily the specific verbiage. The fascination with publicizing Hari's sexuality (which isn't notable) as well as sexual activities (again, where's the notability of this?) is unnecessary and appropriate to delete. I think Tlupick's assessment that it was a "salacious and unnecessary sentence about Hari's personal life as a gay man." was apropos, if not at least a reasonable edit upon which reasonable people could disagree. Adding what he's done in 2025 as a social commentator on multiple high-profile shows is a reasonable addition, even if the exact verbiage wasn't ideal. I do not concur that this edit is "clearly promotional" when the previous version of the sentence was not. I would also caution Tlupick to avoid casting aspersions. Just present the facts and let wikipedians decide for themselves/let the chips fall where they may.

Addendum re: salacious content: On the talk page, there was unanimous agreement to remove that sentence by all parties [9] [10], albeit for different reasons. Salacious is defined as "arousing or appealing to sexual desire or imagination". It was clearly an attempt to add sexual content that wasn't necessary nor notable. Literally no one at the time (2 months ago) felt it was an appropriate addition. Therefore, I find the claim I don't see why we should then refrain from including it, and I especially take issue with calling it 'salacious' both misleading and void of reasoning. Buffs (talk) 02:20, 22 January 2026 (UTC)

I've struck the sentence about that diff. You are correct that it's bad evidence that I completely misinterpreted. I have now struck that part.

A valid reason to remove the "salacious" line (mentioned on the talk page) is that it was cited to a primary source and thus undue. But it being "salacious" on its own is not our concern. I was not aware of the talk page discussion when I posted my above comment and thus only saw the rationale in the edit summary, which I still hold to be insufficient.

Now, if we are applying the rules, we must apply them consistently, and that makes the wording added in the remaining diff, supported solely by primary sources, especially concerning.

That said, I'm now much less convinced this tban is needed. My initial response was based on a cursory review of Tlupick's edits to the page, and I thought I had seen enough. Evidently I have not, so I will take a closer look in the coming days and either strike my !vote or present further evidence to support it. Toadspike [Talk] 05:45, 24 January 2026 (UTC)

Anything further to add? Buffs (talk) 20:46, 28 January 2026 (UTC)

I do not find the links and arguments presented here compelling enough to warrant a topic ban. Buffs (talk) 19:08, 21 January 2026 (UTC)

Furthermore, I find the quality of Tlupick's edits to be mostly benign and or reasonable. Reasonable people can disagree on what "appropriate" is. I can chalk up much of what Tlupick did inappropriately as noob errors. I have no dog in the Hari fight. I know nothing of significance about the person. My interest is little more than a passive one where I just stumbled onto it via WP:AN. I believe that, while Tlupick may have gone overboard, I think he is also generally correct about editing the article to present a neutral view of the man. COI concerns seem to be largely based on winning the argument via exclusion, not a genuine conflict of interest. Buffs (talk) 02:30, 22 January 2026 (UTC)

In response to the proposal that I be topic-banned from Johann Hari, I am writing to explain my interest in the Hari page, and to explain issues I found there that led to my sustained interest in the page. I ask that this information be taken into account by admins and editors when deciding to support or oppose a topic ban on Hari.

I did not begin editing Wikipedia with any particular interest in Hari. I wanted to write about significant figures in the North American harm reduction movement, as it the subject of two books I wrote in my former career as a journalist, and because I found it to be an area lacking coverage on Wikipedia. The first Wikipedia page I created was Liz Evans. I also wrote about Philip Owen, Bud Osborn, and made small contributions to a few other individuals and related topics. Then I found the Hari page. Hari’s work, Chasing the Scream, is a significant book on the subjects of harm reduction and the war on drugs, and so Hari is a person of interest related to harm reduction.

Here is a November 2024 version of the Johann Hari page, as it existed when I first found it.

The page is overwhelmingly negative, and the vast majority of the entire page relates to events during one period of 2008-2011. During that period, Hari was involved in multiple professional scandals, including plagiarism and the malicious editing of Wikipedia. So negative coverage of this 2008-2011 period is justified. But since 2015, Hari has published four books, all of them best-sellers in many countries. He become a celebrated public speaker with one of the most-viewed TED Talks ever, was made a repeat guest on high-profile shows such as Real Time with Bill Maher, and received public praise from the likes of Oprah Winfrey and Hillary Clinton. (The significance of any of this can be debated. I list these events here simply to illustrate that Hari’s career continued beyond 2011.) Before 2011, Hari was a young newspaper reporter, albeit one the subject of high-profile scandals. Sine 2015, his career and influence have grown in many ways. But this post-2015 period (which happened to be more positive for Hari) barely received a mention on his Wikipedia page, thus creating an overall negative picture that differed from reality and was also simply out of date.

Here are four examples of specific content on the Hari page as it existed in November 2024 for which I had concerns.

1) Page intro lacking balance and citation

The intro section for Hari concluded, “He has since written books on the topics of depression, the war on drugs, the effect of technology on attention span, and anti-obesity medication, which have attracted criticism for inaccuracies and misrepresentation.” I argue that while Hari’s books contain errors, they are minor errors, and the sort of criticism that many high-profile nonfiction books attract, in part because it is very hard to write 400 pages without a few mistakes slipping through. (This is my take; you are free to have your own.) Yet this significant statement is included in Hari’s Wikipedia intro (and therefore the very top of his Google results) without any balance, such as noting his books have also attracted widespread acclaim, and even without citation. I was not able to see this section improved, and I believe it remains that way today.

2) Inappropriate content

The page included multiple instances of salacious content that simply had no business on anyone’s Wikipedia page. One example: “Hari is gay.” Another example: “In a 2002 article, he stated that he had had sex with men who were members of homophobic far-right and Islamist groups.” A third example: “David Allen Green noticed that an 'methuselahproductions' email address associated with the David Rose identity [Hari] had also been used to post incest erotica.” I was eventually successful in seeing this content removed, but it took me significant effort and multiple debates with other editors who argued it should be kept on the page.

3) Overly detailed accounts of minor negative events

Where Hari’s post-2015 books did receive coverage on his Wikipedia page, it was entirely negative, even when the events described were minor in nature. For example, the page read: “Restaurant critic Jay Rayner criticised Hari for incorrectly stating, in Magic Pill, that Rayner had taken Ozempic.” Another edit that was deleted before November 2024 stated, “Hari apologised on X, saying that he had confused the article by Rayner with an article by Leila Latif in the same paper.” So this was a simple mistake, and one that Hari apologized for publicly and immediately. But the description of the mistake was kept on his Wikipedia page while Hari’s mea culpa was deleted from the page. I argue such a minor mistake does not justify inclusion on a Wikipedia page at all, but if it does, surely a related apology should be included, too. I was not able to see this section improved, and I believe it remains that way today.

4) Problematic editor conduct

Without assigning motive, I will note two examples. A “neutrality of this article is disputed” banner was removed from the Hari page without any agreement or even discussion. In addition, in late-2025, editors used the “revert” function to restore the entire Hari page to a version from early-2025, and repeatedly did so without describing their concerns related to specific edits, and without explaining their reasoning for taking such a drastic action. I believe this was in contravention of Wikipedia revert guidelines.

There were many other problems with the Hari page as it existed when I found it in November 2024, and there are many problems with how it exists today. I will limit this post to those four examples for the sake of brevity, and because it is not the goal of this thread to correct every issue with the Hari page.

I share these four examples to illustrate my initial interest in the Hari page, and to explain that these issues were the reason for my sustained interest in the page, not any conflict of interest.

All four examples above can be debated. That’s fair. But I think we can agree that they are issues that a reasonable person could find problematic, and therefore can be considered understandable reasons for an editor’s interest in a specific Wikipedia page.

My edits of the Hari page should be viewed in this context.

I edited the Hari page because I wanted to correct something I viewed as unfair and inaccurate. The overall picture of Hari’s career on his Wikipedia page is simply not an accurate reflection of reality.

It can be argued I eventually become over-zealous in my editing of the Hari page. And at times, I became frustrated with opposition to my edits on the Talk page (although I believe I always maintained civility). But I suggest these imperfections in my conduct do not warrant a topic ban.

In addition, I encourage Wikipedia admins to review the Hari page and evaluate why it has been so difficult to add what I maintain is only balance and fairness to the page. After one year, I remain unsuccessful in improving the page. Whether or not I am issued a topic ban on Hari, I hope that admins and/or other editors will pay attention to this matter, agree with some of my concerns, and work to correct them.

Thank you very much for your time and consideration. Tlupick (talk) 19:04, 23 January 2026 (UTC)

Oppose. I previously described my past relationship with Johann Hari, and I argued it does not meet the bar for a COI, let alone a COI topic ban. I also previously posted a lengthy explanation of my interest in the Hari page, which I believe speaks directly to additional reasons for voting against a topic ban on Hari. For the reasons described at these two links, I oppose. Thank you. Tlupick (talk) 19:33, 29 January 2026 (UTC)

Proposal: Topic ban from Louise Vincent (Tlupick)

Since no one has seen fit to add this and it was brought up in the initial statement, I'll go ahead and propose this with the intent to shoot it down so we have a place for evidence (if any is provided): Tlupick (talk · contribs) is indefinitely topic banned from making edits related to Louise Vincent, broadly construed.

Oppose I don't see any evidence presented here to warrant such a topic ban. I'm open to being convinced otherwise. Buffs (talk) 23:03, 22 January 2026 (UTC)

I do not oppose a topic ban on Louise Vincent.

Background: I wrote a book about Louise Vincent and one other woman (Jessica Tilley). After the book was published, Ms. Vincent (now deceased) wrote an article for an online publication that was highly critical of the book. In journalism (my former profession), a subject's criticism of a journalist does not qualify as a conflict of interest, and certainly not one severe enough for a newspaper to ban that journalist from writing about that subject in the future. (Imagine if Justin Trudeau could stop a journalist from writing future articles about him by criticizing that journalist. Journalism wouldn't work.)

All that said, I understand that standards for Wikipedia are different from those for journalism. And though I hold no ill will toward Ms. Vincent whatsoever, I can appreciate how the perception of a COI exists. I therefore do not oppose a topic ban on Louise Vincent. Tlupick (talk) 23:21, 22 January 2026 (UTC)

Self-restraint is all that is needed here. If you aren't going to edit because you feel a COI could exist, then it's simple enough to state so and a topic ban is not needed. Starting this section here is more procedural than anything. If no one is going to put forward evidence of malfeasance/inappropriate edits/actions, a topic ban is not warranted. If you say "I'm not going to" and then you do...that would be grounds to bring you back here for a re-evaluation. Buffs (talk) 02:27, 23 January 2026 (UTC)

@Buffs: It was closed by @The Bushranger. I don't see how he's involved. voorts (talk/contributions) 16:35, 23 January 2026 (UTC)

I swear I read that twice to make sure who closed it, but you're correct and this is an error on my part. I still believe it to be ill-advised for this relatively new user to accept such a TBAN and it seems to be that the purpose of this discussion is to see if such restrictions are warranted. But it appears I am alone in my opinion...so be it. I'll restore it. Buffs (talk) 16:40, 23 January 2026 (UTC)

Because he didn't sign it...that explains what I missed. Thought it was you. Apologies. Buffs (talk) 16:43, 23 January 2026 (UTC)

Wait...what? You were the one who hatted this discussion, not Bushranger. No, I'm re-opening this. You're still involved here. If someone else feels I'm in error, please feel free to close. Buffs (talk) 16:49, 23 January 2026 (UTC)

I'm talking about the prior AN discussion. The Bushranger closed it and logged the restriction. I also don't see how I'm involved. I've only engaged in an administrative capacity here. I haven't !voted in this discussion and my last discussion at the prior AN thread with Tlupick was talking things through with him. voorts (talk/contributions) 16:58, 23 January 2026 (UTC)

...you asked for an evaluation... I've obliged you and started this discussion. As you started this discussion to include the topics at hand, I find it to be at least inconsistent if you also close discussion as well. I also don't find it to be with malice or ill-will (I hold no grudge towards you), I just find it to be somewhat improper and introduces a COI (being both the opener and closer) that isn't necessary. Buffs (talk) 19:28, 23 January 2026 (UTC)

Yes, I started the discussion in my capacity as an uninvolved admin. Admins starting sanction discussions does not make them per se involved. Additionally, there's already a voluntary TBAN in place. Tlupick can appeal it if he wants, but even if no consensus is found in this discussion to impose a community TBAN, that won't vacate the voluntary one. voorts (talk/contributions) 19:34, 23 January 2026 (UTC)

Then we can agree to disagree on that point, but if consensus is found to LIFT that ban, then it WOULD negate it. Buffs (talk) 17:43, 24 January 2026 (UTC)

He would need to appeal. I don't know why you're arguing to lift a TBAN that the banned user himself has agreed to. It's not your place to appeal on Travis's behalf. He is more than capable of advocating for himself. voorts (talk/contributions) 21:38, 24 January 2026 (UTC)

Disagreements are resolved through consensus-based discussion, not by tightly sticking to rules and procedures. We can look at this and determine via consensus an appropriate avenue for meaningful contributions by this user. We don't need a formal appeal.

As indicated below, that doesn't seem to be what he thinks he agreed to. Buffs (talk) 03:21, 27 January 2026 (UTC)

I state that I will not edit the Louise Vincent page because I feel a COI could exist. Tlupick (talk) 18:42, 23 January 2026 (UTC)

And, to me, that seems sufficient. However, this is a topic ban and those are generally considered to be "broadly construed". So if you edit a page about something she was heavily involved in or her coworkers or any of a number of other topics, that could be considered a violation of that ban. I don't think you were apprised of that and it's worth clarifying the extent of such a ban/voluntary refrainment. Buffs (talk) 19:23, 23 January 2026 (UTC)

One example of this would be to clarify which Louise Vincent we're referring to. Apparently there are at least 2 people of note that share that name. One is an British actor who was noted in A Saturday on Earth, Chanel Solitaire, and Witch Way Love (among others not listed on WP) and the other an American activist. People have mistakenly put this actor's link to the activist in error. If you were to (quite reasonably) go and either create a disambiguation page or fix the links in these articles, you could be subject to a block...I don't think that's your intent by accepting such a restriction. Buffs (talk) 19:41, 23 January 2026 (UTC)

I am seeking clarity on one aspect of the proposed topic ban on Lousie Vincent (American harm reduction advocate). @Buffs wrote, "So if you edit a page about something she was heavily involved in or her coworkers or any of a number of other topics, that could be considered a violation of that ban." -- I did not understand this, and I strongly oppose any ban that prohibits me from editing the pages of organizations or individuals to which Ms. Vincent was related.

I've become confused by the back and forth between admins in the thread above. Can someone please tell me what specific action (if any) I need to take to secure the ability to edit such pages in the future?

Perhaps the simplest answer is for admins here to take my word that I will refrain from editing the Louise Vincent Wikipedia page, and leave it at that?

The totality of all edits I've ever made to the Vincent Wikipedia page are minimal and largely insignificant. And as soon as another editor requested I stop, on account of a perceived COI, I immediately did stop editing that page.

Can we simply record here that I've promised to refrain from editing the page in the future, and clear me of any topic ban on Vincent?

Thank you. Tlupick (talk) 23:37, 26 January 2026 (UTC)

Personally, I am fine with that. Based on your reply above, I do not believe you accepted a Topic Ban with the full knowledge of what it entailed. As such, clarification here would be a reasonable discussion for the community to have and for you to receive guidance. If there is consensus that way, I think rescinding the TBAN is the most appropriate course of action. If not, consensus the other direction would rule the day and the TBAN would be confirmed. I would like for other members of the community to weigh in.

I would also advise you to keep your responses as brief as possible...that's something I struggle with too. Buffs (talk) 03:30, 27 January 2026 (UTC)

I'll add that you're allowed to make a statement (a.k.a. !vote or Not-vote) as well in any of these matters. Simply put oppose or support and your rationale. Buffs (talk) 20:49, 28 January 2026 (UTC)

Oppose. I don't believe that anyone has put forward evidence or examples of misconduct to justify a topic ban on Louise Vincent. As noted previously, the totality of all edits I've ever made to the Vincent Wikipedia page are minimal and largely insignificant. And as soon as another editor requested I stop, on account of a perceived COI, I immediately did stop editing the page. In addition, I have voluntarily promised to refrain from editing the page in the future. Thank you. Tlupick (talk) 19:17, 29 January 2026 (UTC)

OKA: problematic paid translation and lead rewrites via LLMs across thousands of articles.

Context

• OKA is a Swiss non-profit that provides monthly stipends to full-time contributors and translators and leverage AI (Large Language Models) to automate most of the work. OKA editors are required to make a PAID declaration.
• OKA’s metawiki Instructions for editors describe extensive machine/LLM-based translation workflows including to Leverage Grok as a starting point. They're explicitly encouraging use of general-purpose LLMs for translation.
• Their public Google Sheet tracker includes a tab Simplify lead with instructions directing editors to Instructions: pick an article, copy the lead section into Gemini or chatGPT, then review if some of the suggestions are an improvement to readability. Make edits to the Wiki articles only if the suggestions are an improvement and don't change the meaning of the lead. Do not change the content unless you have checked that what Gemini says is correct!

Existing discussions/reports

• October 2022 at WP:VPM: OKA Founder User:7804j announces the formation of the non-profit. Early responses showed machine-translation errors, poor English competency, duplicate articles, overproducing low-quality pages that volunteers then had to clean up.
  • many of the results are very problematic, with a large number of WP:CFORKs, editors who clearly have very poor English, don't read through their work (or are incapable of seeing problems) and don't add links and so on. As the OKA instructions linked above make clear, these are machine translations, with all the problems that brings. link
  • the translations include duplicates of topics we already have link
  • My understanding is that professional translation usually requires natives of the target language for the best result and so the OKA might be going against the natural flow. Apart from command of the language, there are also social factors to consider. The English language Wikipedia is rife with conflict and even native English speakers are often discouraged or driven off. link
• November 2022 at WP:VPI: Suggested tighten controls on translations with flagging of low-quality translations by OKA editors with inadequate quality control and translator English proficiency.
  • In recent months I'm seeing a number of translations, from Spanish and Portuguese, of long and important articles on art history that might or not be machine translations (probably they are) but are certainly not checked by anyone competent in English link
• November 2023 at WP:COIN: OKA editors producing new articles containing lavish tangential detail that might function as subtle promotion, with machine-translation artefacts.
  • The church article is a way too faithful translation of the German one, without a look at that English church articles should look different ... I would not know where to start to bring the product in line with the English Wikipedia ... articles like that add work to reviewers, perhaps more than they provide facts to readers. link
  • The trouble is, they are machine translations by people often with a really poor level of English. They are not adding new content, just translating the whole of the German or whatever article they've chosen. If they bother to do any checking through, which mostly they don't seem to, their English is too poor to spot even glaring and basic mistakes. The linking is always terrible. link
• May 2024 at OKA Founder User:7804j: OKA editors not following COI or PAID disclosures.
  • I have looked through some of the editors from this, and they are not following the conflict of interest or paid editing requirements.
• May 2024 at WP:Intertranswiki: OKA Founder User:7804j queried why OKA editors had to go via AfC to create new articles.
  • OKA creates several thousand new articles per year (99% of which are translations), so I have concerns that if all of these were to go through AFC, it would completely clog the process and makes us unable to operate. We will test it but, given the circumstances, there is a real risk we may need to shut down OKA entirely on EN WP and re-assign our ~15 full-time translators to other Wikipedia languages with less stringent paid editing policies link
• June 2024 at WP:VPP: A discussion on if OKA editors should go through AfC. Cited recurring problems like minimally edited machine-translation prose, awkward titles, poor linking/categorization, no aftercare, and content forks/duplication with AfC justified to protect enwiki from extra cleanup burden.
  • Our core principle is that our translators are free to work on anything that interests them. We provide them with a monthly stipend, some training on how Wikipedia works, but we then see them as volunteer contributors on whom we impose some process to ensure they do not abuse the grant and provide overall value (eg, quality checks, quantity checks). To help them find articles to translate, we curate an optional backlog ... Articles of this tracker primarily consist of "Featured" and "Good" quality articles from other Wikis, as well as red links from these articles
  • We're not paying per quantity, but per hour of work and instructing that people should focus on quality. Our translators are also paid when they work on improvements of existing articles.
  • They have caused a lot of work; mostly these are machine translations by people whose English is rather poor.
  • I've seen no evidence that OKA translators are creating unedited machine translations.
  • 7804j has never denied that these are machine translations, and they normally appear on en:wp in a single edit, & are not edited further except for a couple of tidies. There is no evidence that they are edited machine translations when OKA bow out, and they should be treated as "unedited machine translations"
• September 24 at Talk:Lycée Molière (Paris): An article translated into English by an OKA editor using LLM, with quality issues.
  • The school's significance gets lost among the minutiae of historical enrolment numbers, timetables, uniforms, and report cards – details that might just be acceptable in an institutional history published by and for the school, but have no place in an encyclopaedia which is supposed to summarise sources, not reproduce trivia from the school's newsletter archive.
• 01 January 2026 at WP:ANI: An OKA editor repeatedly broke infoboxes, large translation overwrites that introduced MOS issues, and failure to communicate. Temporarily blocked, then unblocked and resolved after editor promised to be more careful.
  • I have also twice reverted their edits with a summary explaining that communication is required and that they are breaking the page
  • article edits all appear to be expansions by overlaying all or most of the article with a translation
• 02 January 2026 at WP:AINB: Unreviewed / poorly reviewed AI translations by an OKA editor. Unresolved.
  • it is clear that the rich in preserve a rich and varied folklore from Acadian folklore isn't in the original frwiki text (and neither is the *relevailles* markdown). Not looking good
• 07 January 2026 at WP:AINB: Unreviewed / poorly reviewed AI translations with synthesis and promo-language by an OKA editor. Unresolved.
  • There is AI text that does not have any counterpart in it, most of which are synthesis tacked on to paragraphs

The ongoing VPP thread which prompted me to create this AN report

• January 2026 at WP:VPP: The thread starts with OKA Founder User:7804j asking whether the new WP:NEWLLM policy also effectively bans LLM-assisted translations that are, which would disrupt the OKA model. Also includes an example of a OKA translated article containing info not verified by the source.
  • AI translations do not tend to be faithful to the original, and the text that deviates tends to contain the usual problems with standard AI-generated text. In practice the level of review being done is far short of that. For example, this "translation" inserted an entire chunk of text under "River transport" when the corresponding section in the French article was completely empty.
  • The level of human review to AI additions is, quite often, very insufficient
  • However, looking at the provided page number, 147, it doesn't talk about the La Bourdonnaye family at all – instead, the families listed on that page range from Cadelac to Cado.The actual mention of Bourdonnaye, on page 114, does not talk at all about them originating from a lordship in Trégomar, Côtes-d'Armor. I am worried that you might not have actually checked the sources that you have been adding, and I would be happy to see evidence of the contrary.

Main concerns

• Edit quality: OKA editors are making large-scale changes that introduce errors and LLM artefacts.
• Overwrites: repeated overwrites of existing articles during LLM-assisted translation which breaks templates/infoboxes, and makes it hard to review because they're not incremental changes.
• Coordinated mass-editing: OKA trackers indicate various planned campaigns for mass changes (e.g. the Simplify lead campaign) affecting large numbers of pages without on-wiki discussion or consensus.
• QA: from the outside, it’s not clear what OKA’s actual QA looks like (vetting, onboarding, supervision), given the above outcomes.

Suggested remedies

• Edit quality & Overwrites: OKA editors using LLMs to translate must check for quality and accuracy (also pending an LLM translation policy RFC). OKA editors performing translation expansions on existing articles must use incremental edits rather than text overwrites.
• Coordinated mass-editing: Editing campaigns should either have 1) on-Wiki consensus on the affected articles before edits go ahead or 2) required to submit their proposed changes via Edit Requests on the article Talk page.
• QA: OKA to provide documentation regarding their hiring, onboarding, and supervision processes. How do they assess English competency? How they check that LLM output hasn’t added unsupported content? etc

_{Courtesy ping: @7804j as OKA Founder.} qcne (talk) 14:54, 19 January 2026 (UTC)

This is a long-running and serious problem (some of the ye olde quotes above are by me) & while the suggested remedies would be fine if they were made to work, it has long been clear that 7804j is only intermittently interested in engaging with the community, and puts very little effort at all into quality control over the people he pays. He is not himself an en:wp editor with any real experience, and simply does not understand many of the issues. One key issue is identifying OKA-paid material. There used to be a speadsheet that did this, but I don't know if this is still accessible. Most, but I think not all, the editors he pays identify themselves on their user or talk pages, but the articles they create are very often not given any sort of tag. Many of them have pretty poor English, so must use machine-translation of articles from other WPs, or I suppose now freshly-brewed AI texts. A remedy like "OKA editors using LLMs to translate must check for quality and accuracy" just won't fly with the existing set-up - the editors just won't know how to do this. I haven't noticed that the situation common in the early days persists, where an OKA editor set up a FORK for a major article evidently without checking whether en:wp already had one (normally we did, with a rather more idiomatic title). He was warned about that, & it may have stopped - the solution was just to redirect the OKA title to the existing one, so his money was wasted. Johnbod (talk) 15:26, 19 January 2026 (UTC)

To me this comment "[...] we may need to shut down OKA entirely on EN WP and re-assign our ~15 full-time translators to other Wikipedia languages [...]" shows that they are indeed using machine translation. I don't see how a translator could translate competently into English, and then just switch at the drop of a hat to translating into Finnish or Farsi instead.

I have accepted a few of these at AfC, and they seemed quite okay to me, although I admit I didn't specifically look for possible MT problems. I also feel, rightly or wrongly, like investigating the translation quality or provenance would be beyond the scope of AfC's (or for that matter, NPP's) remit. -- DoubleGrazing (talk) 15:55, 19 January 2026 (UTC)

If he meant any random language, sure. But a lot of translators can do bi-directional translation, so if you're doing es→en, it might be easy to take assignments that go en→es. Many translators know multiple languages, too. (For that matter, my plumber speaks seven languages. In some parts of the world, being multilingual isn't remarkable.) WhatamIdoing (talk) 23:56, 22 January 2026 (UTC)

Fluency is also a spectrum. I know enough French to watch movies and read books, which is useful when there's an annoying lack of English sources for something. But my French conversational skills are severely lacking and my writing is iffy on a grammatical level. There's a reason I don't hang out on fr.wiki even if I read their articles sometimes. Clovermoss🍀 (talk) 21:11, 23 January 2026 (UTC)

Yes; similar for me. And sometimes the limited fluency leads to one-way translation: I'm confident translating from French to English, but less so from English to French. Cremastra (talk · contribs) 23:09, 23 January 2026 (UTC)

A significant concern that occurred during the NEWLLM discussion was this edit, where an obviously llm-generated text with fabricated sources was restoreddetagged by 7804j. While it was later removed, I do note checking that a fabricated source added elsewhere in the article was not removed. I have removed it now, but clearly the issue extended beyond just that additional portion and I doubt anyone has actually checked the other sources to see if they were fabricated. I don't know if we need an OKA specific remedy, but hopefully OKA participants are made aware of the potential issues they might run into on en.wiki. CMD (talk) 17:13, 19 January 2026 (UTC)

I appreciate the community bringing these concerns forward. I recognize that as OKA grows, our activity becomes more visible. I want to address the specific points raised and clarify how we manage quality.

Regarding OKA’s output, it is essential to look at the data in context. With over 13,000 articles created globally (including 4,800 on English Wikipedia) by more than 80 OKA editors, a small number of errors are statistically inevitable in any high-volume project, whether volunteer or funded. While we treat every report of a problematic edit as an opportunity to refine our internal guidance, these instances remain outliers. They do not represent the vast majority of our work, which provides a significant net benefit by filling content gaps and expanding articles. Our goal is not to "clog the process" but to continue optimizing our internal checks so that our positive impact remains high while minimizing the burden on other editors.

LLM usage and accuracy

The concerns regarding fabricated sources (hallucinations) are the most serious.

• Source verification: We are reinforcing to all grant recipients that LLMs are tools for phrasing and translation, not for facts. Any editor found adding unsupported content or fabricated citations receives a formal warning.
• Policy alignment: I initiated the recent discussion that led to the current RFC because I want OKA to work within the community’s consensus on WP:NEWLLM. We are also finalizing a study on AI effectiveness to share with the community.

Process and vetting

• Overwrites: I fully agree with the feedback here. Overwriting existing content is not ok, unless that content is poor. This is clearly reflected in our instructions. Any editor who ignores this receives a formal warning; two warnings result in the immediate suspension of their OKA grant.
• English proficiency: Our process already requires editors to demonstrate their ability to produce idiomatic, high-quality English that meets WP:MOS before they are eligible for a stipend. We aren't "hiring" staff, but rather providing grants to independent contributors who meet these standards.
• Transparency: Our tracker is public at oka.wiki/tracker (and on-wiki). Every article we work on is tagged as such in the talk page too. We are open to suggestions on how to make OKA-funded edits more easily identifiable (e.g., specific edit summaries) to assist reviewers.

Clarifications on specific comments

• On my experience: Johnbod mentioned I am not myself "an en:wp editor with any real experience". This is not true -- I have been a Wikipedia editor for over 10 years. Like many long-term contributors, my focus has been primarily on the content side rather than policy discussions or noticeboards, so it is true I have less experience on these aspects. I am engaging here to ensure the community has the facts about our workflow.
• On "switching languages": To clarify: our editors only work in languages they are fluent in. My previous comment about "re-assigning to other languages" referred to moving our funding focus to different language editions of Wikipedia, not asking individual translators to work in languages they don't know.
• The "fabricated source" incident: To be clear, I did not "restore" fabricated text. I removed a maintenance tag based on a misunderstanding of whether the section was a translation or an expansion. Once it became clear the editor hadn't verified the sources properly, we took action. I'll take responsibility for the mistake in that specific review, but it was an isolated error.

Moving forward

We don't want OKA editors to be held to a "different" standard, but we recognize that as a funded organization, we have a responsibility to ensure we aren't creating extra work for the community. We believe the current model provides a substantial net gain for Wikipedia. 7804j (talk) 19:41, 19 January 2026 (UTC)

Cmon man, the training guide instructs translators to create multiple email accounts to get around LLM usage caps… ExtantRotations (talk) 20:12, 19 January 2026 (UTC)

...yes, and? 7804j (talk) 20:16, 19 January 2026 (UTC)

Did you write this with the usage of an LLM, @7804j? Three AI detectors suggest it's AI generated. qcne (talk) 21:08, 19 January 2026 (UTC)

Whether or not I used an LLM, spell-checking tool, or the help of someone to review and provide feedback on my response is irrelevant. I would like us to focus on the substance of this discussion.

Also keep in mind that tools that "detect LLMs" look for signs such as the presence of bullet points, headers, bolding, etc. which are also best practices for communication. Structuring arguments as lists and highlighting core ideas is always how I communicate when needing to convey complex ideas to a large audience, when I can spend sufficient time to copyedit and review my response. For example, the message I am currently writing is less structured (no bullets or headers), mostly because I am writing while I am on the go, thus having less time to iterate on it. 7804j (talk) 08:55, 20 January 2026 (UTC)

The tools used while editing Wikipedia are directly relevant to this discussion. They are the very topic of this discussion. Being deliberately evasive on very simple questions is not a helpful way to develop trust. CMD (talk) 10:06, 20 January 2026 (UTC)

@7804j It is absolutely relevant, and this response is disappointing. qcne (talk) 11:00, 20 January 2026 (UTC)

We have ended up here at AN largely because of some OKA editors' careless use of LLMs, so yes, whether the founder and main driver of the project uses LLMs to communicate with us is very relevant. And having spent months working alongside Qcne at AFC, I am confident that he understands the nuances amd pitfalls of AI detection and isn't making this accusation lightly. ClaudineChionh (she/her · talk · email · global) 09:14, 20 January 2026 (UTC)

Yes, I use LLMs to help me draft and structure my thoughts, especially for long responses where I want to be clear and address multiple points. I don’t believe there is anything inherently wrong with using modern tools to improve the quality and readability of a discussion, provided the person posting takes full responsibility for the content.

I’m sorry if my previous response seemed evasive; that wasn’t my intent. My point was that the substance of the arguments should matter more than the software used to assist in writing them. To me, using an LLM to refine a post is a matter of personal preference and productivity, similar to using a sophisticated spell-checker or a translation tool.

Trust is a two-way street. Being questioned on the mechanics of how I write feels like a distraction from the actual policy issues we are here to solve. Many professional environments now encourage the use of these tools specifically because they can improve the clarity of communication.

The core issue here is whether OKA's output on Wikipedia is accurate and valuable. I am here to take responsibility for that output and to work on the process improvements I mentioned above. I would prefer we focus on those specific quality controls rather than the tools I use to draft my replies. 7804j (talk) 11:14, 20 January 2026 (UTC)

I mean technically you aren’t taking responsibility for the output though; you have made it clear that all errors are the fault of your (quite low paid) employees, and that they will receive all the punishment. Nowhere in your list of changes do you discuss introducing additional training or help for translators. ExtantRotations (talk) 12:20, 20 January 2026 (UTC)

I want to be clear about what I mean by responsibility. As the founder, I am responsible for the system and instructions our editors follow. If there is a pattern of errors, it means our training or oversight needs adjustment, and that is on me.

When I say editors are responsible for their output, I’m simply stating the Wikipedia rule: every person is accountable for the edits they make. However, we don't just "wait for mistakes to punish." Our responsibility is in the support we provide:

• Training: We don't just hand out money; we provide onboarding and direct mentorship from experienced editors.
• Systemic Fixes: When the community flags an issue—like overwriting or infobox errors—we don't just blame an individual. We update our global guidelines and hold group sessions to ensure every OKA editor understands the fix.
• Quality Control: The "warning" system is a safeguard for Wikipedia. It’s there to ensure that if someone consistently fails to meet community standards despite our training, they stop receiving funding.

I am not "passing the buck." I am here to ensure that OKA as a whole is a net positive, which means I am constantly iterating on our training to prevent these issues from happening in the first place. 7804j (talk) 12:43, 20 January 2026 (UTC)

We don't just hand out money; we provide onboarding

Are the editors compensated/paid for their time during onboarding?

As a personal request; if you reply to me, can you please use only your own words? I do not like talking to someone through a computer-generated layer, it actually feels disrespectful to me. --Gurkubondinn (talk) 12:57, 20 January 2026 (UTC)

Yes, they are compensated during onboarding regardless of the outcome of the trial period 7804j (talk) 16:06, 20 January 2026 (UTC)

Thanks a lot! I presume that this refers to a broader program than the initial 40 hour trial, as, regarding the latter, the job posting template states that:

Note that payment for the trial requires that you publish at least 5 articles successfully without quality issues and share your progress daily within the established trial period.

If you do refer to a longer paid onboarding, I would be happy to hear details about how it is being managed by the senior editors, and what oversight is present regarding this process. This could really help convince the editing community that your translators receive appropriate supervision and training. Chaotic Enby (talk · contribs) 16:18, 20 January 2026 (UTC)

This sounds like a contraction to compensated during onboarding regardless of the outcome of the trial period. --Gurkubondinn (talk) 16:25, 20 January 2026 (UTC)

The difference is that the job posting refers to the payment as requiring articles without quality issues, which isn't the same as regardless of the outcome, which is why I believe they must be referring to different processes. Chaotic Enby (talk · contribs) 16:38, 20 January 2026 (UTC)

My reading of regardless of the outcome was that they would be paid even if they did not meet the publish at least 5 articles successfully requirement. But it's not entirely clear to me, maybe there's another process like you said. I just don't like the idea that someone could spend their time on this and not even get paid for it, because this seems a little bit exploitative to me. --Gurkubondinn (talk) 16:46, 20 January 2026 (UTC)

Yes, that's what confused me too – reading the job posting, it seemed like they wouldn't be paid for it, which is more worrying. Chaotic Enby (talk · contribs) 17:24, 20 January 2026 (UTC)

I understand the confusion. The job posting does indeed say 5 articles (which is a mistake, actually it should be saying 2), but this is more a "protection" against abuse of ill-intentioned actors, not something that we really have to enforce in practice except from really egregious cases. What we want to avoid is someone signing up, not doing any work at all, and asking for money. If the person has actually tried to onboard but just didn't succeed, and there is evidence of this (e.g., conversations, drafts, etc.) then we've always paid the entirety of the 40h. 7804j (talk) 18:48, 20 January 2026 (UTC)

Setting asides concerns that this comment was LLM-written or generated in some part, I don't see how holding OKA editors to basic standards of competence and verifiability is holding them to a different standard. The raw volume of errors found and possibly introduced but not yet found is of great concern. We cannot, as a community, keep up with an organized and fast flow of misinformation. This is exacerbated by the pattern of avoiding communication; I was in the 1/1 thread, and it took an outright block to enforce basic communication.

Also, the general rule in translation is to translate from a foreign language into your native language, with which you are hopefully more familiar. I don't see any indication that this is the case. Iseult ^Δx talk to me 03:10, 20 January 2026 (UTC)

@7804j, I'd appreciate seeing more details about the onboarding process.

1. Do you have an example of a position description or call for expressions of interest from potential new translators? Where do you advertise?
2. How are new translators selected and what are the selection criteria?
3. What training is provided to new translators, especially around the different policies snd guidelines of the different language Wikipedias? Links to training materials would be especially helpful for this discussion.
4. When English Wikipedia editors bring up quality control issues with you, what specific steps do you take to address these with the translator? e.g. do you show them the relevant noticeboard/talk page discussion, do you identify the specific parts of the text that have problems, do you read and discuss the relevant policy together?
5. Is anyone other than yourself involved in the selection, training, or supervision of translators?

ClaudineChionh (she/her · talk · email · global) 03:49, 20 January 2026 (UTC)

Translators are selected by other, more senior, translators. I do not personally get involved in the process. This happens through a mix of personal referrals, and postings on Linkedin. We have tested different platforms, and found that it was the most suitable to get a high-volume of diverse profiles without cost. One of our goals is to onboard non-Wikipedians to Wikipedia, so that we grow the total pool of editors, so we intentionally do not recruit directly within Wikipedia. Of course, this comes at the cost of more complex onboarding and more mistakes being done in the beginning, but we see it as part of the value of OKA to train people on the use of Wikipedia. You can find the post template here.

All our onboarding materials are published on-wiki, such this quick-start guide and the main instructions. We have these additional instructions for non-en Wikipedias. Senior editors conduct the initial resume screening, interview, and also have several video sessions with new editors to walk them through it and answer their questions. During the trial period, their contributions are reviewed in detail by more senior editors, and the degree of review decreases over time as they gain experience. We continue to monitor things such as: spot-checking of articles, AfC rejections, issue tags, mentions of translators into any Wikipedia discussions. We also organize monthly optional round-tables, and have several internal chats where editors can ask questions or share best practices.

Our core guidelines is that each OKA editor is responsible for directly interacting with the community, and responding to and resolving issues. It's hard to always be aware of every mention of an OKA translator, but whenever this happens we look into it and try to establish the root cause and solutions. This typically results in a mix of changes to our processes or guidelines, reminders to the broader group, and if the translator was at fault, a formal warning (or dismissal for very significant issues or deliberate violations). A second warning typically leads to the termination of the grant. We aim to be as specific as possible, pointing to the specific text or edit, and to the relevant Wikipedia and OKA guidelines or policies.

Most of these processes are "self-managed and decentralized" to the grant-recipients, where senior editors take on these review and management activities. We then have three volunteers (myself and two other people) overseeing the effort, doing an additional degree of spot-checking, designing and building the tools and processes, and acting as escalation layer.

Generally speaking, we try to be as transparent as possible with the community about the inner workings of OKA, and have been actively seeking suggestions for process improvements. As long as it remains within the spirit of OKA's mission, I welcome any proposed changes to our processes. 7804j (talk) 07:01, 20 January 2026 (UTC)

In the interest of transparency, would you be willing to identify who else (besides you) has a supervisory or oversight role? At the moment, I am only aware of you as founder, and the lists of translators supported by OKA, which do not indicate who else is involved in supervision and quality control. ClaudineChionh (she/her · talk · email · global) 09:22, 20 January 2026 (UTC)

Additionally, it would be helpful to clarify the responsibilities of the senior translators, regarding selection, management and review of their junior colleagues. As this isn't outlined in the job posting, is there a different position/contract involving these aspects, or is it an expected part of the duties that all translators sign up for? It could be helpful to make this information available on-wiki, especially regarding who is responsible for whom. That way, we know who to ask for help if there is an issue with a newer translator, and OKA's internal review process can work jointly with Wikipedia's. Chaotic Enby (talk · contribs) 09:48, 20 January 2026 (UTC)

From what I understand – although please correct me if I am wrong – OKA editors are independent contractors, with a very low salary of 397 USD per month for a 40 hour week, which I believe corresponds to 2.29 USD hourly. Given this precarious status, I am worried that more uncertainty in the translator duties may lead to an overloading of responsibilities, which is worrying as independent contractors do not necessarily have the same protections as paid employees. Having more clarity regarding their exact responsibilities and the internal structure through which they work may be necessary to reassure some of us in that regards. Chaotic Enby (talk · contribs) 10:07, 20 January 2026 (UTC)

There are three of us at the core organization level: myself, @Nickoka (based in Winterthur, Switzerland), and a software developer. Nick handles our internal operations and finances, while the developer is building an open-source stats platform to improve our tracking. Neither are active on-wiki editors, which is why I am the primary point of contact for the community.

I have updated our list of editors to list the three currently active OKA senior editors who assist with peer-review and onboarding.

The relationship between senior and junior editors is one of mentorship rather than a strict hierarchy. We don't have rigid "roles descriptions" for them; instead, senior editors are long-term grant recipients who have demonstrated a high level of proficiency. They are tasked with:

• Onboarding new grant recipients.
• Conducting peer-reviews during trial periods.
• Coaching on OKA, Wikipedia and local project policies. They are given significant freedom to manage their "mentee" groups as they see fit, provided the output meets Wikipedia's quality standards.

To clarify the stipend amounts: the $397 USD figure is for the initial trial period. For regular editors, this increases up to $452 USD, and for those taking on mentoring/management roles, it ranges between $525 and $634 USD. While these amounts are modest, we ensure they are above the minimum wage in the countries where the editors reside.

Many of our recipients edit part-time or use the stipend to support their existing volunteer work. Our goal is to enable people to contribute who otherwise wouldn't have the financial flexibility to do so. We are very sensitive to the "precariousness" mentioned—which is why we don't treat this as a traditional job with "overloading responsibilities," but as a grant to support independent encyclopedic work. There are also no set working hours—people are free to work whenever they want to, and leave at any time without any notice.

I am happy to share more details on our internal workflows if it helps. However, our internal documentation includes sensitive payment details, so I need to be careful about where it is shared. I'm open to providing any additional details that the community feels that would help bridge the trust gap. 7804j (talk) 12:40, 20 January 2026 (UTC)

@7804j While these amounts are modest, we ensure they are above the minimum wage in the countries where the editors reside.

I have changed my mind. This is disgustingly exploitative. You're a Swiss-based non-profit exploiting residents of the global south to add AI slop to the project and expecting volunteers to clean this up. Take a long, critical, look at yourself and your operations.

At this juncture I want OKA editors to be prohibited from directly editing articles entirely, at the minimum. qcne (talk) 16:18, 20 January 2026 (UTC)

+1 --Gurkubondinn (talk) 16:26, 20 January 2026 (UTC)

While I don't know in which canton of Switzerland you are based, the lowest minimum wage (in Ticino) is 19.5 CHF, or 24.7 USD. In comparison, the minimum hourly wage for your freelancers is 2.29 USD. Chaotic Enby (talk · contribs) 16:37, 20 January 2026 (UTC)

I find the characterization of our work as "AI slop" and the personal attacks regarding "exploitation" to be both disrespectful and inaccurate. I ask that we remain focused on the substance of Wikipedia policy.

Regarding the stipend levels: OKA is a non-profit providing grants, not a corporation hiring employees. We provide financial support to editors in regions where the cost of living is lower, specifically so they have the freedom to contribute to Wikipedia full-time—something that would otherwise be impossible for them. Comparing Swiss minimum wages to the local economies of our editors is not a helpful metric for a global project. Our goal is to empower contributors from the Global South and under-represented communities by removing the financial barriers to their participation.

Whether a stipend is "reasonable" is a subjective labor discussion that is outside the scope of this noticeboard. The community’s role here is to evaluate the edits, not our internal financial structure.

As for the content, labeling thousands of reviewed translations as "slop" ignores the thousands of hours of human work and the value these articles provide to readers.

I am happy to discuss quality controls, but I will not engage further with personal attacks or hyperbole about our mission. 7804j (talk) 17:05, 20 January 2026 (UTC)

99% chance AI Generated. We are highly confident this text was AI generated

It is disrespectful to continue to talk to us via a chatbot, despite already being asked not to. This is now disruptive. I would hat and hide your comment if I was not involved. qcne (talk) 17:35, 20 January 2026 (UTC)

I have been open about using tools to help structure my thoughts and ensure my responses are clear and professional. There is no policy that mandates a specific drafting style, and I find it counterproductive to derail a serious discussion about OKA's operations into a debate over my personal writing process. 7804j (talk) 18:00, 20 January 2026 (UTC)

Your own use of LLMs, as the founder of the nonprofit in question, to communicate with editors here on Wikipedia, which your nonprofit relies on, is absolutely relevant. --Gurkubondinn (talk) 18:03, 20 January 2026 (UTC)

Your website indicates that your translators bill you for their time worked; is that because having them act as independent contractors is less legally regulated than if you just granted them the money? I mean, you keep calling them grants anyways… ExtantRotations (talk) 18:06, 20 January 2026 (UTC)

It's a mix of reasons, but the primary one is process simplicity. We don't need to actively keep track of who's working and who's not working: we wait for the editors to send us the list of what they've worked on, we review it, and pay. Given that we're running this as only 3 volunteers, it's important to keep processes as lean as possible for the volunteers. But this also means when an editor decides to no longer work with OKA, or if they want to reduce their activity rate or not work for a few weeks, they don't even need to let us know, they can just stop submitting their "bills". 7804j (talk) 18:51, 20 January 2026 (UTC)

Apologies but you are the one who created a nonprofit that is financially dependent on Wikipedia. I think it is extremely prudent to ask things like how much the charity raises per year or how much you earn in comparison to your employees. I don’t know about Switzerland but in Canada this info is publicly available for nonprofits. ExtantRotations (talk) 17:40, 20 January 2026 (UTC)

According to this article from 2025 -- the organization linked to the news article so I assume this is ok to post -- half the budget (80,000 Swiss francs) is from the founder's personal income and half is from "donations and fundraising." 10,000 Swiss francs last year came from a Wikimedia grant to evaluate AI-assisted translations on the encyclopedia (something all of us are currently doing for free, by the way). I don't know whether that 10,000 is part of the "donations and fundraising" or separate. Gnomingstuff (talk) 18:15, 20 January 2026 (UTC)

Actually we have just published the results of this research project here. 7804j (talk) 19:10, 20 January 2026 (UTC)

From your website (bolding mine):

To maximize the impact of our limited funds, we primarily award grants to translators based in countries with low costs of living. These countries also tend to be underrepresented among Wikipedia editors, which also helps us decrease Wikipedia’s systematic bias. ExtantRotations (talk) 17:50, 20 January 2026 (UTC)

Can you point to an on-wiki discussion where you proposed the translation project to the English Wikipedia community before creating your non-profit association and taking on translators? ClaudineChionh (she/her · talk · email · global) 21:59, 20 January 2026 (UTC)

@7804j: I'd really appreciate a response to this question because as it stands, it looks as if you initiated this translation project without considering the wishes of the broader community or asking for feedback beforehand. ClaudineChionh (she/her · talk · email · global) 21:48, 21 January 2026 (UTC)

As Qcne wrote in the timeline, no, I did not inform the community before starting the project, but many discussions took place shortly after, and at that point I took many steps to act on the community's feedback and be even more transparent and proactively communicating (e.g., formally associating OKA to Wikipedia:WikiProject Intertranswiki). Since then, the community has been overall supportive of our OKA, with the exception of isolated incidents (and this discussion so far, which I believe does not reflect the community's balanced opinion but rather the opinion of a self-selected subset of the community).

Why did I not inform the community before the project? Simply because at the time, I didn't realize it was expected, as I had very little experience with such types of projects, or even with interacting with the Wikipedia community in general (I was just a regular editor). Also, at the time I wasn't even sure I would be able to make it work, and the concept was still very experimental, thus not having very clear boundaries like it has today 7804j (talk) 07:05, 22 January 2026 (UTC)

Can you point to an on-wiki discussion where you proposed the translation [..] before [..]?
— User:ClaudineChionh

So the answer to her question is a "no" then?

P.S.: please use your own words if you choose to reply to me. --Gurkubondinn (talk) 13:15, 22 January 2026 (UTC)

Yes the answer is no, which I felt was pretty clear in my previous reply (which, by the way, was not LLM-generated). 7804j (talk) 13:47, 22 January 2026 (UTC)

Apologies, I didn't notice that you had given a clear answer. That's my bad, I wouldn't have written that reply otherwise. --Gurkubondinn (talk) 15:46, 22 January 2026 (UTC)

Thank you for your response yesterday (which was a long day at work for me). For what it's worth, I disagree with your interpretation that this discussion is not representative of the community's overall sentiments towards your initiative. ClaudineChionh (she/her · talk · email · global) 22:09, 22 January 2026 (UTC)

This discussion involves a dozen of editors that saw it through the admin noticeboard. Over the past few years, hundreds of people have been exposed to OKA's work, and I have personally received many messages of encouragement from a variety of editors and admins. Many editors have also thanked our editors directly for their contributions through barnstars, words of appreciation, etc. Several of our editors are active in Wikipedia projects, some participate in local Wikimedia chapter events, etc. So I don't know what's the overall community's view, but I know for sure that this thread doesn't represent it well 7804j (talk) 06:26, 23 January 2026 (UTC)

I second that. People don't come to AN to heap praise and give pats on the back, this is the place for grilling and hard knocks :P _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 06:28, 23 January 2026 (UTC)

At this juncture I want OKA editors to be prohibited from directly editing articles entirely, at the minimum. I agree in full. Stepwise Continuous Dysfunction (talk) 19:03, 21 January 2026 (UTC)

Firmly agree. There are numerous issues here - practical issues, labor issues, and community interaction issues - of which the most significant is that users are being told to add translations that they are not able to check for errors and copyedit. 7804j's responses here have made it clear they cleared this nonprofit without considering what the community would want, and have no intention of addressing the numerous major concerns. This effort needs to be shut down, and not allowed to simply skip over to harming a different project. I would prefer if someone else can formally propose the prohibition - I don't have any experience in doing so - but I will be it myself if need be. Pi.1415926535 (talk) 21:23, 21 January 2026 (UTC)

I could not agree more, the responses here by 7804j are concerning, almost all are AI generated and they seem to not respect the community's wishes at all. – LuniZunie_(talk) 23:21, 21 January 2026 (UTC)

I beg to differ. We don't know what community wishes are, except in the broad context of building an encyclopedia, and 7804j and his project have been certainly doing just that. As for the use of AI to streamline replies, there's a difference between having an AI reply, and having an AI rewrite/copyedit one's thoughts, and my understanding is that they've been doing the latter, not the former. While we are seeing a norm clash, with some folks considering the latter lazy, I think it's a sign of things to come, and overall an improvement, given low quality of online discourse; in either way, while I also support banning "using AI to talk", I have no problem with "using AI to help one refine one's response", and per WP:AGF I see no reason to believe this is not the case here. Lastly, I am also open to the consideration that AI may help some folks suffering from forms of disability (dyslexia, etc.) and I find criticism of folks that may benefit from such tool unwarranted and arguably, distasteful. Would you criticize a blind person for using a gadget to edit Wikipedia? In the end, it doesn't matter what tool one uses to write a response and communicate, as long as we are still talking to them (and not a bot). Ban the bots, let real people do what they want - it's not our place to force our thoughts on which tools they should be using. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 05:25, 22 January 2026 (UTC)

@Piotrus I absolutely agree with your point on the disabilities as I have dysgraphia and dyslexia, and I do use AI to help with spelling and reading (because I can't read for the life of me). However, what I am trying to say is, if an issue is brough up with you, and you use a bot to respond, that's a problem; you aren't responding or acknowledging the issue at all by doing that.

As per the community's wishes, that is true that we only know the broad context, but what I meant by that is that the responses here, and the subsequent responses by 7804j, are showing there is a pretty large mismatch in what is wanted. In a perfect world, we would tell 7804j that everything needs to be properly checked, but this isn't a perfect world and that just won't happen and I am not given confidence by the responses here to prove otherwise.

I of course, agree that minimal AI usage should be fine, I mean, it's the future whether we like it or not, and failing to adapt to the times will be this project's demise. But I just don't think this is the way to do it. – LuniZunie_(talk) 12:25, 22 January 2026 (UTC)

Well, I have read chunks of the discussion here, maybe I missed something. It's clear some OKA translators got lazy and used AI without double checking for common errors. I expect OKA (=7804j) to introduce safeguards to prevent this from happening, and discipline repeated offenders (because as everyone knows, if we have to do it here, it won't be pretty). I expect OKA will agree - it's not their first rodeo, and in the past, AFAIK, they have been ammenable to improve their standards and ensure the content they deliver is acceptable. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 12:51, 22 January 2026 (UTC)

I am a bit confused as to what gives the impression that OKA isn't willing to adapt. As Piotr mentioned, we are more than happy to make any changes to our processes that the community finds beneficial.

Also, as Piotr mentioned, I only used AI to organize my thoughts, I did not use AI to generate messages from scratch. It is therefore fully within the community's own guidelines. Also I have been extremely detailed and transparent in my responses.

LuniZuni, is there a concrete ask that you feel OKA should commit to but where you feel there is a fundamental mismatch between what you would want, and I would be willing to agree to? 7804j (talk) 13:39, 22 January 2026 (UTC)

LuniZuni is my doppelgänger :p

is there a concrete ask that you feel OKA should commit to but where you feel there is a fundamental mismatch between what you would want, and I would be willing to agree to?
I'm not asking about what I want, what I want is a website where AI is allowed in moderation, but not in content generation. What I am asking is that these things be communicated onwiki before they occur, not in discussions that take place after the fact, they should be done before. This is a major thing being done by your non-profit, that needs to involve the community. Right now, it seems like the community is involved, but only after the fact, which is not very helpful. – LuniZunie_(talk) 13:47, 22 January 2026 (UTC)

In all honesty, I (and others) complain elsewhere about how WMF does stuff without consulting community. Of course, we seem to be unable to change WMF behemoth; here, we have much more leverage, but I also think - again, based on past experience - we are much more likely to see OKA change what they do than WMF, sigh. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:51, 22 January 2026 (UTC)

While I of course agree, I don't think that just because WMF does it means the rest of us should also do it. – LuniZunie_(talk) 14:53, 22 January 2026 (UTC)

I am concerned about the accusation of exploitation made here; this is serious and I worry it was not fully considered in the heat of the moment. 7804j has said OKA work is not meant to be full-time, but fine, let's say people do it full-time anyways. 7804j says workers are paid more than minimum wage, but that doesn't mean much; even in Switzerland, most of the country has no minimum wage, so he could be paying one Rappen an hour and that would still be true. But he has given us numbers here that we can check against the statistics. Let's look at two countries in the global south with vast numbers of native English speakers: Nigeria and India. The World Bank says average annual income in Nigeria was 1663 USD in 2021 [11], which comes out to 139 USD per month. For India it gives 1907 USD per year, or $159 per month. The same list gives $115 monthly for Senegal, a (somewhat) Francophone country, and $427 monthly for Colombia, a Spanish-speaking country. (Many OKA articles are translated from Spanish or French. These are the first countries that came to mind for each language – if I were to cherry-pick the data, I could list, say, Chad and Bolivia instead.) For comparison, the minimum OKA stipend listed above was $397. Now, I don't know if the World Bank data truly represent a living wage in those countries, something that is very difficult to measure, and I don't know where OKA editors are (maybe they're in countries with far higher cost-of-living?). But this is a reasonable estimate, and it would seem to contradict the accusations of exploitation made above. Toadspike [Talk] 04:22, 24 January 2026 (UTC)

Thank you for looking into this, I was also rather shocked by the language used. No good deed, you know. Here, someone is paying out of their own pocket (or at least partially, and in either case not counting time spend managing this) to get volunteers (nobody is forcing anyone to do anything) to contribute to Wikipedia - and he is being accused of exploitation. I am at a loss of words, really, because I don't think anything I could say would be particularly nice. Sigh. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:33, 24 January 2026 (UTC)

@7804j: This sub-thread got somewhat derailed. I want to return to the issue of OKA's labour force. Set aside dollar/euro/franc amounts for a moment. To state my bias upfront, as an Asian immigrant to Australia I am very sensitive to "representation" being used as a fig leaf for exploiting workers in the global south, and "independent contractor" arrangements used to absolve your responsibility to your workers. Because that's what they are, your workers, and treating their half- or full-time employment as gig work is abhorrent to me. You have imposed a mass article creation program on the English Wikipedia without prior community discussion, which means the existence of this program and your workers' livelihoods is dependent on the community's goodwill. A noticeable number of your workers were given inadequate training or oversight which meant that they came to the community's attention as "problem" editors, though it's now becoming clearer to me that these problems are due to systemic issues within OKA itself. You are squandering any remaining goodwill with your evasive answers here. Like qcne I have become disgusted with this whole situation, but am also sensitive to the needs of the OKA workers who presumably have no form of protection from the consequences of their boss's behaviour. ClaudineChionh (she/her · talk · email · global) 04:35, 24 January 2026 (UTC)

And I look at it from exactly the other side. The boss is trying to help, both the workers and the community, and should not be held responsible for the fact that some of his workers are lazy and cut corners. The boss is volunteering and using their own money, not making a dime of profit. The workers are the ones who are making profit - and nobody is forcing them, they are free to choose other gigs. I can't understand how come we are criticizing a volunteer who invested immense amount of time, and their own financial resources, to help our community, instead of the paid editors, some of whom are obviously abusing the system. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:54, 24 January 2026 (UTC)

Who else's problem but the boss' can it be if the translators are not adequately trained and their work is not adequately reviewed? It's massively irresponsible to sic a cadre of unsupervised 'translators' on Wikipedia with the specific instruction to use general-purpose LLMs to translate massive swathes of text and not implement any kind of oversight or verification. OKA can't then wash their hands of that and say "well, some people are just lazy and don't review their work". Of course they are, that's why you have to plan for that in your operational model! For instance, they could have one or two more highly paid staff whose sole job is to internally review the translations made by the 'grunt' translators before they go live. Or they could at least implement some kind of peer-check system between the translators themselves. Either way, if you just leave them to their devices, obviously you're going to end up with one or two of them just shovelling slop into the encyclopedia with no second thought. Athanelar (talk) 10:58, 25 January 2026 (UTC)

Although I disagree with your assessment of OKA, which I think is quite responsible, I do think there is merit in the idea that they should hire some folks to do quality check, even if this would reduce the total output due to reducing the translator's total number a bit. It would, hopefully, prevent the issues we are seeing (as marginal as they are, IMHO). @7804j - this idea is worth considering seriously. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:18, 25 January 2026 (UTC)

I agree that having peer reviewers could make sense, and I've started to think about it in the last couple of days. However, it's not easy to put in place operationally and there are several different options to consider, so it will require some further thinking and fleshing out.

What I can do is draft a process and start a pilot internally with a few editors and focused on EN WP. If this works well, we can then scale it to the rest of OKA and other Wikipedias. This might take a bit of time, though I expect we can already get some results and feedback within a few weeks.

-----

As a side note, I really dislike the "boss" analogy. Would we require that every Wikipedia editor has a boss and an assigned reviewer? The whole point of OKA is to provide grants so that people can become regular editors. The more process we put on them, the more they lose the freedom that makes Wikipedia great in the first place.

OKA doesn't strictly require that grant recipients translate using LLMs, nor that they translate in fact. These are mere recommendations based on what we have found to be an easy way for people with dual language proficiency to have significant impact on Wikipedia. We do actually encourage editors to explore any other tools or types of Wikipedia contributions, but relatively few do, simply because they agree that LLM translations are a good way to have positive impact.

One of OKA's goal is actually to bring new editors to Wikipedia, which is why we don't advertise on Wikipedia. If someone promotes Wikipedia at their university or workplace, and as a result some mix of editors start editing, some being good and some being less so, is the editor who promoted Wikipedia in the first place going to be held accountable for having promoted it to the wrong people?

That being said, I acknowledge that due to the scale and organized nature of OKA, additional safeguards are useful, even if they restrict a little bit the freedom of OKA editors, hence my willingness to try the peer reviewer approach. 7804j (talk) 14:38, 25 January 2026 (UTC)

Your goal is understandable and admirable, but you of course have to be aware that paying people for their contributions is going to change the way they contribute. Similarly, of course someone who tells their friends "hey, editing Wikipedia is fun, you should try it" isn't liable for what those people do, but that's not what you're doing. You're specifically selecting, enrolling and supporting paid editors. You are responsible for the people on your payroll, there's no getting around that, because the mere fact that you are paying them is an explicit endorsement of their editing activities.

The "one or two bad apples" argument only works if you're making an active effort to catch the bad apples. It's very good that you take action when it's reported to you, but it's concerning that somebody is able to go on for so long without it ever being caught internally. You need to have some kind of internal review process, otherwise you can't be surprised when people blame you for the actions of people you're literally paying to edit. Athanelar (talk) 14:49, 25 January 2026 (UTC)

I don't agree with this perspective. If a university gives a full scholarship to a student (tuition and costs of living), it doesn't automatically become liable to all the activities and publications of that student, even if the university encourages its students to write and publish. E.g., if the student publishes a flawed and questionable paper on their own blog, the university may decide to cancel the scholarship, but shouldn't be held liable for having granted that scholarship in the first place. 7804j (talk) 16:22, 25 January 2026 (UTC)

Sure, but if one of the students accepted for said scholarship demonstrated that they lacked the qualifications one would expect for a recipient of that scholarship, it would raise questions as to how thoroughly the university is vetting its applicants. Athanelar (talk) 16:50, 25 January 2026 (UTC)

I disagree with this as well. I know many universities or systems where scholarships are funded without particular vetting of the applicant's competence, other than on the basis of a CV.

In either case, nobody would say that such universities are the boss of the scholarship recipients, even if such scholarships are tied to certain outcomes (e.g., successfully meeting course criteria). This is the model we're trying to replicate with OKA 7804j (talk) 19:28, 25 January 2026 (UTC)

@7804j, a brief comment on the "boss" wording: I didn't use it in relation to the editing workflow, but to draw out the economic and power dynamics in OKA's structure, regardless of whatever legal or taxation arrangement is used. @Athanelar has picked up on that aspect with paying people for their contributions is going to change the way they contribute. On reflection I think my critique of your gig work arrangement is a tangent from this specific discussion of quality control, but I am still extremely concerned (indeed, still disgusted) by this arrangement and may seek a better venue to continue that discussion. ClaudineChionh (she/her · talk · email · global) 00:11, 26 January 2026 (UTC)

Do keep in mind WP:BOOMERANG. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 01:22, 26 January 2026 (UTC)

There's also the Grok angle: that is the worst LLM one could possibly be using from an "oh this will be bad bad if it gets out" perspective, given that the three things the average person associates Grok with nowadays are Elon Musk, revenge porn, and Grokipedia.

I don't buy the "Grok knows wikitext" angle, all modern LLMs have a rudimentary-enough knowledge of wikitext to match the rudimentary level of the translations. Gnomingstuff (talk) 17:45, 20 January 2026 (UTC)

Wikipedia was mostly written by volunteers. But sometimes they need a bit of help helpful, indeed. qcne (talk) 17:52, 20 January 2026 (UTC)

We have actually just published the results of a study we have done on LLM quality for Wikipedia translation. Based on these results, we have proceeded to switch over from Grok to Claude and ChatGPT as the recommended tools, but there are certain use cases where Grok still performs better.

We had done a similar (though a lot less detailed) analysis in the past, and so far Grok was always outperforming the other models for this specific task. So we haven't picked Grok because we like Elon Musk or Grokipedia (in fact, I don't), but just based on data. 7804j (talk) 19:15, 20 January 2026 (UTC)

The mere willingness to test using the revenge-porn generator, the one that declared itself "MechaHitler", for this purpose is indicative of poor judgment. Stepwise Continuous Dysfunction (talk) 19:05, 21 January 2026 (UTC)

And CSAM generator. The level of doublethink needed to convince oneself of acting for the public good while wanting to be anywhere near that is breathtaking.  — Hex • talk 13:35, 22 January 2026 (UTC)

I find this response and the one above extremely concerning. Are you suggesting that because Grok is a very controversial tool that sometimes produces very problematic results when prompted to do so, that even the idea that I would want to test the effectiveness of Grok as a translation tool is a proof that we have poor judgement or poor intents?

Keep in mind that translation is a very mechanical task. Grok isn't going to suddenly add Hitler references or generate child porn when asked to translate a sentence about the History of France... I'm really not a fan of Elon Musk, but come on, there's really no need for extra sensationalism here... 7804j (talk) 13:43, 22 January 2026 (UTC)

Are you suggesting that because Grok is a very controversial tool that sometimes produces very problematic results when prompted to do so, that even the idea that I would want to test the effectiveness of Grok as a translation tool is a proof that we have poor judgement or poor intents? yes. qcne (talk) 13:49, 22 January 2026 (UTC)

If you knew a human who created CSAM, would you still work with them and trust them? – LuniZunie_(talk) 13:51, 22 January 2026 (UTC)

And now for argumentum ad hitlerum... _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:52, 22 January 2026 (UTC)

Unless I am misunderstanding something, I don't see how that applies. – LuniZunie_(talk) 14:55, 22 January 2026 (UTC)

Keep in mind that translation is a very mechanical task.

I believe that this shows a rather poor understanding of what goes into localisation and translation (and the unwarranted editorialising that LLMs perform whenever they're asked to merely translate something). --Gurkubondinn (talk) 08:17, 23 January 2026 (UTC)

I am not saying it is "mechanical" in absolute terms, but compared to other prompts you might give an LLM like "Tell me what you think of Nazis" 7804j (talk) 11:38, 23 January 2026 (UTC)

Ah, so you meant "mechanical task" in the context of LLM inputs/prompts? --Gurkubondinn (talk) 11:51, 23 January 2026 (UTC)

Yes 7804j (talk) 11:56, 23 January 2026 (UTC)

Aside. Iseult ^Δx talk to me 01:33, 23 January 2026 (UTC)

Also, fun fact. Wikipedia was created by one Jimbo Wales. From his bio: he also founded (earlier) "a web portal featuring... softcore pornography". How does it make you feel? :P _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:54, 22 January 2026 (UTC)

Well, I think CSAM and softcore pornography are a bit different... – LuniZunie_(talk) 14:57, 22 January 2026 (UTC)

Sure, just as Grok and CSAM are a bit different... _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 15:04, 22 January 2026 (UTC)

Never said Grok was CSAM, I said it generated it, my entire argument was how it generated it. If an AI doesn't have enough checks and has training data that allows it to generated CSAM, yeah, I don't trust it. – LuniZunie_(talk) 15:10, 22 January 2026 (UTC)

I would rather fear too much censorship per WP:NOTCENSORED. YMMV. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 15:15, 22 January 2026 (UTC)

I'm sorry, I am incredibly confused by this. Are you arguing you would rather the AI generate CSAM a few times than be censored enough to not generate porn? Or are you arguing something else? – LuniZunie_(talk) 15:17, 22 January 2026 (UTC)

Also, due to insufficient safeguards, I think most LLM models have been known to generate CSAM in the past. By that logic, all of them are not worth trusting. Now, when we consider CSAM has been created long before AIs, by the beings known as humans, why would we trust anything humans made? _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 15:17, 22 January 2026 (UTC)

I haven't seen anything that says other AIs generate CSAM, and if they do, then yeah I would not trust that AI.

And the argument
Now, when we consider CSAM has been created long before AIs, by the beings known as humans, why would we trust anything humans made?
is ridiculous, most humans don't consume, generate, or even think about CSAM, why would you discard everything made by most humans? You can't claim all humans are something just because a few are that. – LuniZunie_(talk) 15:19, 22 January 2026 (UTC)

I'm not fond of views in which humans are automatically, permanently irredeemable if they do something, so let's talk about your black-and-white worldview for a moment. About one in 15 underage males admits to having created and distributed naked sexual photos of themselves.[12] That's officially considered CSAM material in most of the world, even when it's fully consensual. Shall we kick out one in 15 males out of society for the rest of their lives, because they engaged in illegal sexting? We get about three-quarters of a million registered editors here each year. It's mostly males. A quick back-of-the-envelope calculation suggests that something on the order of 5,000(!) of the Wikipedia editors who made any edit during the last year has also sexted at least one naked photo to at least one romantic partner before the age of 18. So I ask you your own question: Now that you know that thousands of Wikipedia editors technically created and distributed CSAM material, would you still work with them? WhatamIdoing (talk) 00:58, 23 January 2026 (UTC)

? really weird comparison. qcne (talk) 15:00, 22 January 2026 (UTC)

With all due respect, I think that the point the above users are trying to make is not that Grok is likely to do any of these things while translating Wikipedia articles, but rather, that using a model that is particularly notorious for its poor oversight was a lapse in judgement. While use of Grok in this context is not necessarily an endorsement of its creator or its controversies, even in the absence of your explicit dis-endorsement above, it's not unreasonable that people aware of the many controversies of Grok would be concerned about its use in this context. It's obvious that using Grok for translation purposes is not likely to produce questionable or illegal image content, or that OKA editors would be foolish enough to put such content on Wikipedia, however, Grok's biases are well documented, and its tendency to gas up or disparage particular concepts and individuals based on the politically-motivated instructions provided by its creators is relevant in the context of translating encyclopedic text. If paired with poor oversight, it's reasonable to be concerned that text that is unencyclopedic or outright biased may be more likely to slip through when using Grok as opposed to other models with less toxic reputations. It's good to know that you've switched over to primarily relying on Claude and ChatGPT (despite having their own problems, I think most users would consider them far less likely to produce biased outputs), but I hope you can understand why so many users are concerned and consider the use of Grok to have been a poor decision, even though I'm sure it was done with good intentions. Ethmostigmus 🌿 (talk | contribs) 02:51, 23 January 2026 (UTC)

I understand, and I have definitely taken these concerns into account. However my own extensive experience of using Grok was that it's only showing these biases or issues when either (a) prompted to do so, (b) on extremely rare occasions (that I never personally observed other than news report) where prompted about specific topics that are very close to Elon Musk. As you mentioned, for translations that would never happen due to the mechanical nature of the task.

So I am not surprised that some people in this thread would think that it's a sign of poor judgement, but the reality is that it isn't. It was carefully considered, and at the time, Grok indeed had a much better performance than other models, so the pros significantly outweighed the cons back then. Fortunately we can now switch to models that don't have this reputation since the data proves they now also perform better 7804j (talk) 06:36, 23 January 2026 (UTC)

Someone should hat this nonsense part of the thread. ~2026-47637-7 (talk) 16:41, 22 January 2026 (UTC)

To clarify, my argument was one of optics. After Grokipedia, after the massive controversy just recently, the already bad headline "An organization is paying people to flood Wikipedia with AI content" becomes the even worse "An organization is paying people to flood Wikipedia with Grok content." It's just an unforced PR error. Gnomingstuff (talk) 02:01, 23 January 2026 (UTC)

And that would be just fake news and should be ignored. AI translations that are vastly accurate (reported problems account for only a small minority of generated content) is not a "flood of AI content". _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:07, 23 January 2026 (UTC)

OK, but what about unreported problems? No one's gone and systematically checked all of them; the above dataset seems to mostly look for spelling error-type issues, whereas the issues AI Cleanup has found are far deeper. Most of everything the project has found, really, was undetected for months or years. Gnomingstuff (talk) 08:46, 23 January 2026 (UTC)

If they are unreported, we cannot even be sure they are real. Maybe there are no problems to find. All that we have is a proverbial storm in a teacup - one OKA translator was lazy and did not check the AI output. "Big deal". IF there were more translators like that - I haven't seen the diffs or usernames. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 11:35, 23 January 2026 (UTC)

To be clear, given we had over 50 translators that each published dozens or hundreds of articles, I am quite sure we will find more examples if we really look for them. The question for me isn't "do problems happen", but "how often do these problems happen, do we believe we can mitigate them, and if not are they so large that they offset all the great outcomes where no problems were found" 7804j (talk) 11:42, 23 January 2026 (UTC)

Indeed. To answer the first and third questions, we need data I am not sure we can easily get (and without the data, it's all pretty much subjective ILIKEIT/IDONTLIKEIT, sigh). To answer the second - well, that's something you can do (and perhaps have done already here?). _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 11:51, 23 January 2026 (UTC)

Yes, the second point is something that we've tried to do as much as we could, but where we're fully open to any suggestions for doing things differently 7804j (talk) 11:57, 23 January 2026 (UTC)

From what I've seen in the past, OKA delivers good quality content, although as with any large project, it will occasionally fail and produce a dud. I was not aware OKA has adapted LLM, but it makes sense - these are popular and efficient tools (and, of course, imprefect). The only realistic suggestion I've is that all OKA content output should be checked with tools like ChatZero, and that the translators are required to ensure that LLMs didn't add hallucinations, or lazily skipped over content. Other than that - good job, kudos to volunteers engaged in it, and move on. (And, sigh, it amazes me how much complains and scorn is heaped upon volunteers who go an extra mile and try to something ambitious). --_{Piotr Konieczny aka Prokonsul Piotrus| reply here} 05:11, 22 January 2026 (UTC)

Arbitrary break

@7804j, I'm not an admin but I've got a few questions:

1. I'm wondering how general editors would know to approach you if there were any concerns over an editor? The user-page disclaimers go to your website, but the Contact page of your website seems to be down and there's no indication of who we should contact on Wikipedia. Are you expecting general editors to report them using our normal escalation process then admins have to contact you in turn?
2. Do you perform regular quality checks (e.g. checking Talk pages for warnings) to ensure that quality is maintained, considering they're being paid? From what I've read above, it looks like they're only checked during probation then left to continue on their own (with self-reporting) but I may well have missed something.
3. What happens if an editor has passed probation but their work falls below the expected standard? Has this happened yet?
4. I'm also wondering how you accommodate for different projects having different guidelines/policies (e.g Wikipedia:Notability), considering you're essentially transplanting articles from one Wikipedia into a completely different one. I see there's a preset list of articles, but you also advise editors that they're free to create their own.

Apologies if some of these have been answered, but this is a long thread! Blue Sonnet (talk) 04:53, 23 January 2026 (UTC)

1. The OKA.wiki contact page is a redirect to mailto:info@oka.wiki so it seems to work for me. Regarding how to contact us: every OKA editor has a link in their talk page to our meta Wiki profile. There, people can find the username of senior editors and of myself, the website of OKA, and other information. People have used various channels to contact us, including tagging me, writing in my talk page, writing in one of the OKA's talk pages (which are monitored), or emailing us, etc. We don't have a particular preference or expectation on how people reach us -- we expect that most issues get resolved directly by communicating with each editor, and that if people want to escalate an issue, they get in touch with me or with OKA through one of these channels. If you feel this could help, we could for example try to make it more prominent on the Meta page what's the process to escalate issues?
2. Yes -- we have a script that lists all the maintenance tags of all pages ever created by an OKA editor (based on the OKA template added in the talk page). The senior editors regularly go through the list to ensure these issues are resolved, so even if a page gets tagged after 1-2 year, we still go back to it to fix the issues.
3. Yes this has happened, including for several of the issues listed in this thread. This was discussed in some of the above comments, but we then try to look into what happened, identify if these were process issues or individual issues, and if it's something that can be "resolved" or a fundamental skill or attitude problem with the editor. Typically, if this is the first time, it leads to a formal warning, but has led on several occasions to a suspension of the grant to the editor when the issue occurred again. In such cases, we then task the senior editors to review all the work of that editor to see if some "clean up" is required to fix any potential issues introduced
4. Generally we tell people to follow the English Wikipedia as minimum standard, even when translating to other Wikipedias. But we do have some language-specific guidelines as well to complement these. Editors are indeed free to create their own articles, but in my experience this almost never happened, because people prefer to translate, which feels safer or more familiar for most as it's their profession.

7804j (talk) 20:04, 23 January 2026 (UTC)

@7804j Sorry to ask yet another question, but is there a list somewhere of the targeted donations, and/or of which articles fall under them? I couldn't find it, although I believe it probably must be available somewhere. It could be reassuring to make it prominently visible so admins don't worry about issues of paid editing disclosure. Chaotic Enby (talk · contribs) 23:46, 23 January 2026 (UTC)

Currently not, because so far we haven't received any targeted donations. This statement on OKA's website is more about keeping that option open for the future. If we ever do, yes, I would list them publicly somewhere, with the details of the rules/requirements and affected articles (note that the affected articles could be vaguely defined, e.g., "Any article that is related to Switzerland" if the donation was from the Swiss government). 7804j (talk) 07:37, 24 January 2026 (UTC)

Thanks a lot, that is indeed very reassuring to know! Chaotic Enby (talk · contribs) 07:47, 24 January 2026 (UTC)

Thanks for the clarification - I'm on mobile so perhaps something went wrong there? It would be nice to have a link to the meta page on the disclaimer, I'd find that really helpful for those who aren't familiar with the project & want to find out more on-wiki. Blue Sonnet (talk) 07:44, 24 January 2026 (UTC)

I think that we've gotten a little sidetracked above with discourse about specific LLMs and LLMCOMM. My concerns, as expressed earlier, remain unaddressed. The last three discussions originally linked by qcne are troubling. The fact that obvious indicators of synthesis (at best) or hallucination are in translations is unacceptable and has always been so on this project. Less obvious/more insidious errors require more-or-less fully bilingual editors to identify and redress. That more errors haven't been found in the array of huge edits ([13]) made as a pattern from OKA-affiliated editors is almost certainly a question of volunteer scale and capability. AINB is already dealing with a lot. I do not want this community to be in a position where ten unpaid volunteers will have to verify the output of thirty paid editors while hamstrung. Iseult ^Δx talk to me 08:27, 23 January 2026 (UTC)

I am not convinced. Until someone finds more errors, it's pretty much fearmongering/ABF, particularly given the tremendous positive contributions made by OKA (hundreds of translated articles, if not thousands, with only a fraction ever being a problem). We should focus on real, not imagined, problems. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 11:40, 23 January 2026 (UTC)

Thanks you. You've made your position abundantly clear in many comments above: AI translations good, no sanctions needed, close this thread. What this amounts to is throwing the burden of verifying the translation on other people and dismissing their concerns. You and 7804 have still not addressed the concerns and diffs originally raised in this thread. You have not provided proof that the translations and expansions are actually good.

I spent an hour trying to verify four sources here a few weeks ago. It turns out that AI has a really high rate of making things up. I assumed good faith, just as I am here, and got burned. Why don't you go verify everything in this edit, for example? Or, if some error crops up in an OKA editor's contributions, will the blame be attributed solely to that editor and not systemically? It turns out that verification isn't easy!

Making things up is manifestly forbidden! Why is it that I have never seen an AI proponent involved with AINB or patrolling NPP for AI spam? Iseult ^Δx talk to me 18:16, 23 January 2026 (UTC)

Why is it that I have never seen an AI proponent involved with AINB or patrolling NPP for AI spam? That would involve work, and the point of AI is to offload work and not care about the results. Cremastra (talk · contribs) 18:52, 23 January 2026 (UTC)

Noting that, after previous issues were brought up, I decided to check the (at the time) latest OKA draft by that same editor, and found that the very first source I could verify was hallucinated. Of course, every find can be dismissed as "only a fraction" given the amount of text OKA publishes, but when this is the case for every translation that gets closely reviewed, statistically, this doesn't bode well. Chaotic Enby (talk · contribs) 21:00, 23 January 2026 (UTC)

It would help to have a full list of articles to audit rather than finding them piecemeal. Gnomingstuff (talk) 21:25, 23 January 2026 (UTC)

A proposal

Alright. I don't think we're going to get anywhere particularly helpful out of the forgoing discussion. I would like to remind everyone that we have two core ideals to inform ourselves of what to do here, namely WP:V and "anyone can edit". That is, we care very deeply about whether things are verifiable, but we also care very deeply about how this is an encyclopedia written by amateurs working together; that is, mistakes are part of the process, and we can expect new editors in particular to make a lot of them. I agree that OKA probably has something like "the WikiEdu problem", namely that people mostly only notice when it's going wrong, not when things are working, which happens far more often. And it also seems to me that OKA editors are being treated as a monolith, so we end up with "OKA screwed up again" not "editor x screwed up again". Neither of these are fair. But the problems are real. So let's set some rules for OKA translators:

1. everything an OKA translator adds to en-wiki must be verified by the translator; that is, if they haven't checked the information against the provided source, it should not be added.
2. OKA translators who have received, within six months, four (correctly applied) warnings about content that fails verification will be blocked without further warning if another example is found. (clarification added asilvering (talk) 04:43, 24 January 2026 (UTC))
3. content added by an OKA translator who is subsequently blocked for failing verification may be presumptively deleted as WP:V failures unless an editor in good standing is willing to take responsibility for it.
4. effective henceforth; that is, no blocks/pdels for content written before this proposal passes (if it does).

The "within six months" clause is there to prevent a scenario whereby a new OKA editor messes up three times and has the sword of Damocles hanging over their head for the rest of their wiki-career. The intent is not to prevent action being taken against an editor who is messing up at high speed and high volume - we'd just have to do that the usual way, at ANI. -- asilvering (talk) 20:53, 23 January 2026 (UTC)

I would support this, but the direct encouragement of LLM usage - Grok especially - still really worries me, and I do feel like @7804j inherently doesn't see the moral issue with using LLMs on Wikipedia. The outcome of Wikipedia:Village_pump_(policy)#Seeking_clarification_on_WP:NEWLLM_regarding_human-reviewed_translations will have a direct bearing on a OKA proposal going forward, since the LLM usage is integral to the OKA workflow. qcne (talk) 20:59, 23 January 2026 (UTC)

Yes, and that would require a different kind of discussion, as you note. the moral issue with using LLMs on Wikipedia is not something we are going to reach consensus about in time to usefully deal with the issues raised here about unverifiable content being added to Wikipedia, possibly at scale, by a team of editors. And we can solve that one the way we've always solved it - with blocks - and authorize presumptive deletions to make that less of a burden. -- asilvering (talk) 21:10, 23 January 2026 (UTC)

I think this is a really useful proposal. I do think rule 1 might be slightly too harsh; this effectively blocks translation of any content referenced to offline sources. It seems to me that there are two qualities I want from OKA translator content: (a) translated content is cited to the same source as is used in the source article or a new source verified by the editor, (b) if cited to an offline source, the cited source "seems reasonable" (e.g. sufficient bibliographic metadata to find the source in principle, title seems related, etc.). But I think your proposal has the significant advantages of concision and clarity. Cheers, Suriname0 (talk) 21:23, 23 January 2026 (UTC)

@Suriname0, it is my position that, since there are significant concerns about hallucinations and AI use with the OKA translators as a group, we've lost the ability to assume accuracy when it comes to offline sources. I wouldn't say it effectively blocks translation - someone with access to the offline source via a library, WP:LIB, WP:RX, or other methods could still translate those parts. And these days that does include quite a lot of sources that have previously been "offline only". -- asilvering (talk) 21:43, 23 January 2026 (UTC)

Yes, true! I was counting WP:LIB + shadow libraries as functionally online these days, but especially with resource requests it's probably quite reasonable. It would probably be straightforward to generate a resource request for offline sources before translation begins on an article: "I want to translate the French article fr:So-and-So, which refers to Some Encyclopédie, pages 54 & 65, and I'd like to verify the content before I include it on English Wikipedia, thanks!". Suriname0 (talk) 22:11, 23 January 2026 (UTC)

I disagree that it's harsh. Every editor is responsible for complying with WP:V and should verify all content they add to the encyclopedia. voorts (talk/contributions) 23:26, 23 January 2026 (UTC)

Yes, I agree except rule 1 needs tweaking. The point is to control for AI hallucinations, not to add extra burden on translators, one that even WP:AFC does not. Let me quote from WP:AFCSTANDARDS: "Avoid declining an article because it correctly uses general references to support some or all of the material. The content and sourcing policies require inline citations for only four specific types of material, most commonly direct quotations and contentious material about living persons." It does not make sense to require translators in general, or even OKA in particular, to be required to meet higher standards than what we expect from an average new editor. In other words, we accept, through AfC, articles that are undeerreferenced (which I am not very happy about, but that's how the rules are written). Now, AfC rules don't seem to be updated to AI era, and we are indeed dealing with a different issue here - lazy translators not checking for AI hallucinations. So yes, OKA translators (and translators in general) need to check for them, and if they fail to do so, blocks are in order (just like for any editor who adds AI hallucinations to Wikiepdia). But no, requirement to check all sources, which effectively means checking even offline publications, is way above and beyond what is required from anyone outside FA and some GA level content. It would be an extremely unfair burden, and one that could be gamed easily by using AfC system, meeting their minimum standards, and just stripping article from most references - hardly an outcome we want. PS. Yes, I think we should tighten AFCSTANDARDS, but that's a different discussion. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:02, 24 January 2026 (UTC)

But no, requirement to check all sources, which effectively means checking even offline publications, is way above and beyond what is required from anyone outside FA and some GA level content. Would a compromise such as that proposed by Suriname0 above be acceptable? My worry is, for example, if an AI model adds spurious citations to offline sources (which it couldn't have accessed). Requiring to check sources for citations that you add (that weren't there in the original) should presumably be a baseline. Chaotic Enby (talk · contribs) 04:30, 24 January 2026 (UTC)

This is all a good best practice, but it is not something we can reasonably require, and again, it would be more than we require from AfCs (or from a regular translator). What we need is a requirement to check for AI hallucinations - that I fully support. But any requirement that a translator should verify content from another Wikipedia is beyond what we have ever required, and it would make translating much more burdensome than writing an article from scratch. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:36, 24 January 2026 (UTC)

Checking sources that you add yourself (whether directly or through an AI model) that weren't in the original text is absolutely something we can reasonably require. When you publish an edit, you take responsibility for the content you add. Chaotic Enby (talk · contribs) 04:42, 24 January 2026 (UTC)

@Chaotic Enby I agree with that, as long as "sources you add yourself" does not involve copying within Wikipedia (including cross project). We do have WP:AGF and when few weeks ago I for example split the List of lost literary works from the main lost literary work I can assure you I did not check a single source, nor do I feel I should've had; nor did I check the refs in the List of Fading Suns books I likewise split a few days ago. Such splits, just like translations, are, IMHO, mechanical actions, not proper content creation that requires the creator to take responsibility for what's in the text body (as long as they do their work correctly - and being lazy and adding AI hallucinations is certainly not something we should condone, that's below acceptable levels of quality, as it makes new content worse, whereas proper translation or split should produce content that's identical to what already has been accepted by the community, and thus falls within AGF). _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:27, 24 January 2026 (UTC)

@Piotrus, we're talking about editors who have repeatedly come to the attention of patrollers for creating articles that contain hallucinations. It is time to add "extra burden" to these translators. We are not talking about AFC broadly or translation broadly, but this specific subset of editors. -- asilvering (talk) 04:39, 24 January 2026 (UTC)

@Asilvering Fair. What we need is a list of problematic editors, an understanding of the percentage they make of OKA editors, how that percentage compares to equivalent projects in good and bad standing (i.e. what percentage is a red line?), and then we can figure out what extra burden to add to them (and to be clear, I fully support a requirement for any and all translators - not just OKA ones - to check for and remove any AI hallucination, and to block repeated offenders). _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:58, 24 January 2026 (UTC)

so we end up with "OKA screwed up again" not "editor x screwed up again".

As it should be. This is a structural and process problem, not a problem of a few bad actors. It's hard to blame Editor X for using AI when they were specifically told to use AI. Gnomingstuff (talk) 21:30, 23 January 2026 (UTC)

This seems promising. Is there, though, a specific reason why you chose a four-strike rule as opposed to three strikes? Also, it would be useful to have a table or something along those lines of OKA editors and the articles that they make so that tracking/auditing is made easier, as per GnomingStuff. Then there’s also the LLM translation RfC, which covers OKA. If the RfC passes, would this proposal be made redundant or would the measures here be taken as precedent? Iseult ^Δx talk to me 21:35, 23 January 2026 (UTC)

This sheet and Category:Articles translated by an OKA editor should be helpful for that purpose. Chaotic Enby (talk · contribs) 21:37, 23 January 2026 (UTC)

@Iseult, I went with four strikes since our normal set of escalating warning templates have four levels. -- asilvering (talk) 21:38, 23 January 2026 (UTC)

While I strongly support this as a baseline from which we can start, and I believe the question of "moral issues" of LLMs is ultimately a red herring, there are a few more things I would like to see addressed. For example, the original remedies suggested transparency in the OKA process: the targeted donations program is an aspect of this, and our paid-editing policy would require these to be disclosed. Additionally, large-scale editing campaigns (such as lead format changes in many articles at once) should require explicit on-wiki consensus rather than OKA-specific consensus. Chaotic Enby (talk · contribs) 21:36, 23 January 2026 (UTC)

+1 Ethmostigmus 🌿 (talk | contribs) 01:19, 24 January 2026 (UTC)

Struck the part about targeted donations which has been clarified. I'm already much more reassured about the situation. Chaotic Enby (talk · contribs) 07:48, 24 January 2026 (UTC)

Support. This is a narrowly tailored solution to the problem. Technically #2 should be a 5-strike rule since strike 4 is the final warning. voorts (talk/contributions) 23:31, 23 January 2026 (UTC)

Thanks @Voorts, point taken. I'll clarify that. -- asilvering (talk) 04:41, 24 January 2026 (UTC)

Weak support as bare-minimum. This is a WP:CIR issue: every editor, regardless of what motivates them to edit and what tools they use, is responsible for verifying the information they put on Wikipedia. If an editor is utilising a tool that has a higher likelihood of making mistakes (hallucinating references or substantially changing the meaning of translated text), they should be extra vigilant in verifying both references and the integrity of the text. This is the absolute minimum level of competency we should expect of all editors, volunteer or paid.
Realistically, this is not an issue that can be fully addressed with onwiki sanctions alone - what's really needed is improvements to internal oversight and procedure at OKA, which is obviously out of our hands here, but I hope this thread has given 7804j food for thought about where improvements can be made. Currently, issues identified with OKA editors' work internally and onwiki seem to exist in separate streams: issues identified internally are, presumably, dealt with internally, and issues identified onwiki may be addressed onwiki or internally. I think we need better transparency here, so both OKA editors and the wider community can be on the same page when issues arise to allow both groups to better identify ongoing issues. Ethmostigmus 🌿 (talk | contribs) 01:16, 24 January 2026 (UTC)

I think we also need further clarification, transparency, and community consensus on the paid editing aspect of OKA. Chaotic Enby's suggestions above regarding appropriate disclosure of targeted editing and onwiki consensus for large-scale changes are a good start, but I have a few other things I think are worth discussing further.

• OKA's translation taskforce page states Since OKA translators are paid editors, according to WP:FCOI, all their contributions to English Wikipedia are considered to have a conflict of interest. As a result, they are not allowed to create new articles here directly, but must use the articles for creation (AfC) process - this is good, but the targeted donations add an extra wrinkle - I think a distinction should be made between edits directly related to and not directly related to these donations when declaring COI. As a starting point, if not being done already (I don't think it is, but apologies if I am incorrect!), I think specific disclosures ("I was paid to edit this article on [topic A] as part of my paid editing work funded by a donation from [donor associated with topic A], see [insert link to OKA targeted donations page]") would be much better than the standard COI disclosure. A list of edits made at the direction of donors would also be helpful, if it doesn't exist already.
• OKA's Meta page states that Individuals who donate over CHF 5,000 per year are eligible to join OKA as formal members with voting right in the annual assembly: @User:7804j, can you give us a bit of background on what this entails? What influence do donors who become formal members have, and what kind of matters would they vote on at these assemblies? Is there a public list of these members and their affiliations (ie. primarily editor/translator vs primarily donor)? This system appears quite opaque to those of us outside OKA, which naturally raises suspicion given that this involves paid editing, so I think transparency is in all of our best interests.

Ethmostigmus 🌿 (talk | contribs) 02:44, 24 January 2026 (UTC)

We so far never received any targeted donation, which is why we haven't disclosed any yet. This statement was more to keep the door open for any such situation in the future. If we ever do, we would make the list of such donations, their scope and requirements, public and clear. However, I would be against requiring it in the edit summary, primarily because it makes it extremely complex operationally for us to manage. Also in practice, I expect (and will require) that such targeted donations be sufficiently broad to avoid undue influence. For example, a donation could be for "any article related to Switzerland", so I wouldn't want people to think that because a translator worked on an article for a Swiss company, that this is because that particular company paid for it, when in fact it was just because they were asked to work on articles about Switzerland.

Regarding the donation of 5,000 CHF and above: this is linked to our legal status as an association, which requires a mechanism for adhesion. We would vet such members, so not everyone who donates 5000 will automatically be able to join as a member. We can disclose these members if the community is interested, but so far they only consist of myself and people I am close with who donated such amounts (none of them are active on Wikipedia). 7804j (talk) 07:50, 24 January 2026 (UTC)

Moral support. This is a sensible proposal, grounded in existing English Wikipedia policy. I'm not certain it is necessary, nor am I certain it will fix what it aims to fix, but it's better than the pitchfork mob approach above.

Necessary? In my experience, 7804j has been very receptive to specific, actionable feedback, including rapidly firing an OKA editor who was falsifying sources (link to discussion). In that case, it took only one warning for action from OKA itself, far fewer than the four warnings proposed here. It is alarming that it seems nobody went to 7804j's talk page to raise concerns about the LLM rewrites of existing articles before bringing the issue to this noticeboard, which would have been a far more efficient way of resolving this.

Sufficient? The challenge with detecting issues in OKA drafts is that they tend to be very long and largely unproblematic. In the case I linked above, @MCE89 detected serious issues, but those same issues had been missed by several other reviewers – it's not in the purview of an AfC/NPP reviewer to meticulously check source–text integrity, esp. for OKA drafts which are usually fine (source: my own AfC reviews of OKA content, which often include spot-checks, and also CE's analysis below), when there is so much actual crap that needs their attention. This proposal theoretically makes it easier to block OKA editors, but will we actually do that? Will we actually check any more than we currently do? I doubt it. Toadspike [Talk] 02:58, 24 January 2026 (UTC)

Weakest possible support per my concerns here, I do not want OKA workers punished as a consequence of poor management. ClaudineChionh (she/her · talk · email · global) 04:40, 24 January 2026 (UTC)

Mostly supportive. As others mentioned, this proposal imposes stricter criteria on OKA editors than on regular editors, but I can nonetheless support it as the requirements and process listed is very sensible, and already consistent with our own guidelines.

The only part I disagree with, however, is to require that the content of every offline source is checked. As others have pointed out, it is often very difficult or impossible to check the content of a book, especially when these aren't digitized or extremely expensive. I agree that translators should be required to check the content of any new offline source they add to ensure it isn't hallucinated, and of any online source, but in the context of a translation of offline references, I think it should be sufficient to ensure that the reference was correctly placed and translated and thus not hallucinated. As others mentioned, this is the same standard that we would hold non-OKA translators to.

Without this exception, I expect that we would need to throw out so many articles or paragraphs that translation would become effectively impossible 7804j (talk) 07:29, 24 January 2026 (UTC)

Moral oppose - I'll just say it: I don't think we should be encouraging people, let alone paying people, to add AI content to Wikipedia at all. I'm not sure why that's being framed as an extreme stance; I suspect if we asked readers whether they think people should be paid to add AI content to Wikipedia, the median response would be closer to "no, what the fuck?". Scapegoating a handful of editors for doing exactly what they were told to do is closer to a "pitchfork mob" stance, imo.

This isn't analogous to WikiEdu because WikiEdu doesn't tell students to use AI -- they discourage it in their training guide, and they've been receptive to feedback on the specifics. Obviously some students are going to do it anyway, but that's entirely on them, it's not part of their instructions. Gnomingstuff (talk) 09:26, 24 January 2026 (UTC)

I suspect if we asked readers whether they think people should be paid to add AI content to Wikipedia, the median response would be closer to "no, what the fuck?" Are you referring to the same readers (i.e., consumers of information online) who are slowly shifting to using LLMs in lieu of search engines and coming to trust LLMs more and more? This opposition to LLM use within the community does not map onto the real world. voorts (talk/contributions) 21:42, 24 January 2026 (UTC)

If it didn't map onto the real world, then Wikimedia wouldn't be running an enormous fundraising campaign about how Wikipedia is "human-created". Generally speaking, people don't do ad campaigns about unpopular opinions. Gnomingstuff (talk) 23:29, 24 January 2026 (UTC)

Yeah its like they play both sides[14] and it doesn't matter what advertising they do. Voorts is right here. ~2025-38536-45 (talk) 17:06, 27 January 2026 (UTC)

I don't think we should be encouraging people, let alone paying people, to add AI content to Wikipedia at all. This. This. A thousand times this. Stepwise Continuous Dysfunction (talk) 19:56, 25 January 2026 (UTC)

Support I will support essentially any plausibly actionable accountability mechanism for machine-assisted translations. This will add accountability from day 1. It will also allow for enforcement without having to file a complex LLM/machine translation case at ANI, which is sufficiently unpleasant that it deters reporting. NicheSports (talk) 15:28, 24 January 2026 (UTC)

Support - If they can't verify a source due to not having the original book/document, perhaps they could tag it so someone else can check?

Since AI is prone to embellishments, I'm rather wary of just leaving those alone without any human verification.

Otherwise, my gut feeling is that it's reasonable to expect that an editor who is being paid (even a little) and has access to additional training & support when compared to the standard volunteer editor, should be held to a higher standard - especially when relying heavily on a tool that's broadly untested (this project was/is part of the test), constantly evolving and is well known to produce errors when it's not used correctly.

We may loosen the reins as time goes on, but I feel that the project needs extra guardrails until we have enough data and long-term experience of the project to make that determination.

BTW I was a bit surprised to see that the three Senior editors/Managers have around 5k, 3k and 2k edits respectfully, which isn't egregiously low but isn't as high as I was expecting to see for that position. That's not to say they're not competent as some of them clearly are, I was just expecting more experience (perhaps incorrectly). Blue Sonnet (talk) 15:54, 24 January 2026 (UTC)

To clarify regarding the senior editors: it's not based solely on the quantity of edits, but to a larger degree on factors such as the quality of their contributions, their ability to coach others and to improve our existing processes. Typically, once they receive that role, their edit counts tend to decrease as they spend more time on coaching. Also, note that OKA edits include more bytes on average than the typical edit, since they are often focused on very large articles. 7804j (talk) 16:08, 24 January 2026 (UTC)

Noted, thank you! It wasn't a major issue so I made sure to put it in small font - it's good to know how that works. Blue Sonnet (talk) 16:17, 24 January 2026 (UTC)

Are your workers compensated based on time, or based on how many words/bytes they produce? Apologies if you have already answered that elsewhere in this discussion. --Gurkubondinn (talk) 23:25, 24 January 2026 (UTC)

@Blue-Sonnet, edit count for someone doing high-quality translations is going to be very low. Of all my edits, the ones that took the most time and effort - hours for each - were my translations. For someone exclusively focused on content work 5k edits is quite high indeed and represents a lot of work. -- asilvering (talk) 21:51, 24 January 2026 (UTC)

Yeah, that makes a lot of sense - quality is definitely preferable to quantity. Blue Sonnet (talk) 23:59, 24 January 2026 (UTC)

• Oppose OKA should be barred from Wikipedia by any and all means possible until they forsake the use and even implicit encouragement of Grok, the flagship product of a deepfake porn site. Stepwise Continuous Dysfunction (talk) 19:48, 25 January 2026 (UTC)

  They did say above that they've stopped using Grok and switched to other LLM's. Blue Sonnet (talk) 01:32, 26 January 2026 (UTC)

  Not according to their own documentation, which endorses it in multiple places. Stepwise Continuous Dysfunction (talk) 03:49, 26 January 2026 (UTC)

  It's not endorsed; in fact, the second of the three mentions clearly says that it's inferior to Claude and ChatGPT. The third mention might warrant rewording/clarification that Grok is not recommended. @7804j That said, there are plenty of other tools, and it's hard to keep recommendations of best tools up-to-date - I don't know if we even have any on wiki? (And saying 'don't use LLM' is increasingly pointless, although this all reminds me of the early days of Wikipedia, when teachers in schools were telling students don't use Wikipedia - how things have changed since, eh?).

  Anyway, having read that page I think the general prompt could benefit from clear instructions about not adding anything that's not in the source text, and not omitting anything (being lazy) - the most common errors we see. Oh, and in the alread-present warning "Ensure there are no AI hallucinations! In rare cases, LLMs may completely invent content that's not in the original text. If you introduce AI hallucinations, your Wikipedia risks being suspended", isn't there a word missing after "your Wikipedia"? Account? _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:13, 26 January 2026 (UTC)

  Thanks for the suggestions! I have added them.

  (by the way, feel free to directly edit our instructions if you feel like it -- as long as there are no controversial changes, I don't mind having others making edits) 7804j (talk) 20:48, 26 January 2026 (UTC)

  Anything other than a full-throated rejection is an endorsement. That's how instruction manuals work: point people at a tool, and they will use it. Stepwise Continuous Dysfunction (talk) 21:43, 26 January 2026 (UTC)

  In either case, there is no reason to disallow use of Grok as a translation tool; it seems to preform reasonably where from what anyone can see - unless you are familiar with any reliable data that says otherwise? I doubt it is going to insert Musk POV or such into most tasks (and if its hallucinations are Musk-flavord, that doesn't make them any worse or better other AI hallucinations, which we want zero of here, regardless of which LLM model produces them). _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 06:43, 27 January 2026 (UTC)

  I think the objection is more specifically endorsing Grok. But if they're not doing that anymore, that's a step in the right direction. (As far as "introducing Musk POV," there have been some, uh, not great and hasty changes to Grok's system prompt, so that's actually a decent possibility.) Gnomingstuff (talk) 15:40, 27 January 2026 (UTC)

  The very first mention of Grok in the linked documentation page is, Even for technical articles, LLMs such as Claude, ChatGPT and Grok correctly translate >90% of text and handles most formatting. That's not just permission to use Grok; it's actively saying that Grok is a good thing. The third of the three is a plain instruction: Go to https://claude.ai/ or chatgpt (or https://grok.com/). In between the two is Grok used our preferred tool, and some editors prefer it because it tends to respond better to instructions and has a higher daily cap; however, it's writing style is poorer, so it's not recommended. A mild disinclination to recommend because of writing style and not the fact of its being a propaganda machine from a revenge-porn company is... insufficient. Wikipedia is willing to deprecate sources; organizations that seek to contribute to Wikipedia should be willing to deprecate tools. Stepwise Continuous Dysfunction (talk) 23:20, 27 January 2026 (UTC)

  And we still need more than IDONTLIKEMUSK to declare Grok bad for translation. In fact, I trust OKA (which provides a lot of data) more than POVed views here. Which LLMs to use should be based on data about their efficiency/error-rate or such, not on irrelevant anecdotes about how a given AI is used to create porn or whatever. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:40, 28 January 2026 (UTC)

  I don’t see why your first sentence is true: many people obviously would be happy with such a straightforward declaration. Maybe you mean that you personally object to it? But that’s obvious, you’ve been saying so at great length. ~2025-31850-11 (talk) 11:12, 28 January 2026 (UTC)

  I'm not arguing it's bad for translation, I'm arguing that it's bad for optics. Gnomingstuff (talk) 00:50, 31 January 2026 (UTC)

  I am afraid I don't understand either what you or the IP mean. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 02:30, 31 January 2026 (UTC)

Spot checks

I've decided to spot-check the first few articles in Category:Articles translated by an OKA editor translated from French (as that is my native language and I would be able to verify them more easily).

• 16 May 1877 crisis: In #Initial hesitations between republic and monarchy, some offline sources were swapped without a given explanation, such as Duclert 2021, p. 122-132. being replaced with Houte 2014, p. 27. (another source that was being used somewhere else). Since both sources are offline and would be hard to access, the change being made without explanation is surprising, and I'm not especially confident about text-source integrity here. Besides this one small issue, the rest of the article is a very good translation, adapting the French Wikipedia article to the English Wikipedia's writing style while staying accurate to the original (which was a featured article).
• 20,000 Colombian peso note: Very accurate, couldn't find any concerning divergences between the two versions (original in French was a good article).
• 1824 Saint-Louis coal mine disaster: A few word swaps here and there, but again, great translation all things considered.
• 1879 French Senate election: Now this is a very different case. While the entire article is marked as translated, and was added in a single edit, only half of it actually comes from the French original. The first two sections are fully new, with several new sources being cited, the main two being Providential Men: History of a French Fascination and Dictionary of Ministers (1789–1989). The latter is cited in pages 45–46 for the one/two-round electoral system. The relevant pages talk about completely unrelated personalities (Étienne Clavière, Charles Cochon de Lapparent, Jean-Baptiste Collin de Sussy and Emmanuel Crétet), all dead long before the relevant senate election, or even the Third Republic itself! I couldn't find an online version of the former, but it appears to be a series of biographies of "great men", which doesn't seem especially consistent with it being used to source the organization of a senate election.
• 1914 French mobilization: Despite the source being again a featured article, there are unsourced paragraphs, which were ported (and expanded!) in the translated version. Unsourced sentences like A small cadre of volunteer career officers and NCOs would be supplemented by millions of reservists, who had received training through their mandatory two-year-service at age 20 and periodic refresher trainings. appear from virtually nowhere (and is contradicted a few paragraphs below, where it is stated that the duration of military service varied as different laws were passed). Besides that, a few minor issues (the territorial army and its reserve being confused, état de siège being half-translated as state de siège, broken piped links like "including Algeria"), but nothing too dramatic.

Conclusion? The average OKA translation is certainly good, but errors are still present, especially when the source article has unsourced paragraphs or empty sections, with text-source integrity being one of the main worries. Chaotic Enby (talk · contribs) 22:47, 23 January 2026 (UTC)

@Chaotic Enby It's good to have some data. What you missed is identifying the specific translators responsible for the work. The one problematic article you found, 1879 one, is work of @Leeanah, and per WP:DUCK, I am concluding this particular translator (whose substandard work is what started this entire issue AFAIK) has been using LLM without checking for AI hallucinations. I would recommend pulling all her content from mainspace or at least tagging them with LLM warnings, and I would also ban that editor from any new submissions until they fix their past messes. Now, all the other articles were penned by different translators - which is a good sign, as it indicates most OKA translators are proofreading their work more carefully. So, really, all the evidence we have points to a single "bad egg" (although, of course, this is still just a sample, and it's possible more problems could be found - laziness is hardly a rare vice). _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 04:11, 24 January 2026 (UTC)

This is just not accurate - did you click through and read any of the discussions qcne linked in his original post? The linked discussions involve multiple editors, as well as 7804j themself being forthright about the fact that a small number of OKA editors were not doing the right thing in certain cases. There are some problems with a small portion of OKA editors, and denying that problems exist does not fix them. Ethmostigmus 🌿 (talk | contribs) 04:33, 24 January 2026 (UTC)

Someone should make a table, and clearly indicate what percentage of OKA editors have been found to be fall significantly below our standards, what action was taken by OKA and the community, and how that percentage compares to other projects such s Wiki Edu and so on. Otherwise we are talking apples, oranges, ILIKEITs and IDONLIKEITs. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:18, 24 January 2026 (UTC)

I agree that it seems like two editors here (Leeanah and Luizadnts) seem to have been the cause of quite a few of the issues raised throughout the thread. They had already received a formal warning, and it seems the articles here predate that warning. I've raised these specific points to them, and told them to no longer create any new articles until they have reviewed in depth all of the articles they have previously created. I have also asked them to keep a detailed internal tracker of their progress. It has been communicated to them that their ability to continue receiving OKA grants will be contingent on them being able to correctly review and fix all of their previously published content. 7804j (talk) 15:25, 24 January 2026 (UTC)

Checked the first short article in the W's, Walecznych Street (note I do not speak Polish). Noticeably AI. Promotional in tone, but besides the AI compulsion to turn "has" into "features," the original seems to have been somewhat promotional too. Some content has been removed, but mostly unsourced and/or trivial stuff, so that's fine. However, there are some meaning changes:

Google Translate mentions 11 Walecznych Street having a "glass staircase" whereas the translated article says "glazed staircase," which are two different things ("glazed" implies ceramic, wood, etc). I don't know which is more accurate -- "glass" does seem suspect, admittedly.

The translation changes "designed by Helena and Szymon Syrkus" to built by them, which seems likely to be factually wrong, especially since the same word is repeatedly translated as "designed" elsewhere.

The translation claims that 12 Walecznych Street is "similar to the Kiltynowicz villa," but the Polish article seems to only claim that the clinker slabs are, not the whole house.

The translation changes "window openings" for 37 Walecznych Street (per Google Translate again) to "rounded windows" out of nowhere. These don't look rounded to me (assuming this is the right house -- another angle seems to show a "37" marker on the wood house and it's the only wood-looking one, so I'm guessing it is -- and there aren't different windows out back or something)

Gnomingstuff (talk) 10:10, 24 January 2026 (UTC)

@Gnomingstuff Since I am rather fluent in Polish: 1) glass > glazed, it's a staircase with, hmmmm, that actually gave me a pause. Big windows? Walls made of mostly glass? It's not made from glass, obviously. Literary translation would be "glassy", which obviously is not correct, it's late at night and I am blanking on the right terminology. Anyway, glazed is an error. 2) Yes, designed. 3) The use of these slabs on the ground floor, to be exact (as in, it's the use that's called similar, not the slabs - nor the whole house). 4) 36A not 37 - which is a good reminder that human beings like you and me are perfectly capable of making errors too. Anyway, the original text was indeed wrong ("It features rounded windows, a curved corner, and a garden-facing terrace"). The correct translation would be : "Its characteristic features include windows*, a curved corner, and a garden-facing terrace". *-> Polish text says, literary, "window openings", which is unclear in Polish and seems to me (again, late night, and I am not an architecture expert or fan) to be just a pointlessly long ways of saying window. Anyway, yes, that translation had some minor problems, but whether there are LLM errors or human errors, I am not sure, and in either case, they are minor errors, not something I'd reprimand someone over. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:13, 24 January 2026 (UTC)

@Nihil novi How would you translate a "przeszklona klatka schodowa"? _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 14:16, 24 January 2026 (UTC)

I'm at a disadvantage, as i don't know what "OKA" is, and I don't have access to the context in which "przeszklona klatka schodowa" appears. Among perhaps other possibilities, it might be a "glass-enclosed staircase" or "glassed-in staircase". Is there a link to the Polish text? Nihil novi (talk) 01:41, 25 January 2026 (UTC)

Another possibility: a "staircase with a skylight at the top". Nihil novi (talk) 02:01, 25 January 2026 (UTC)

The article is Walecznych Street, specifically we are talking about a building at Walecznych 11 which features, according to text, a "przeszklona klatka schodowa". Unfortunately I have yet to locate a picture :( _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 07:05, 25 January 2026 (UTC)

The house is visible on Google Street View (link blocked by Wikipedia). While it's not obvious what the stairs themselves are like, the centre of the building, as seen from the street, is all window from ground floor lintel to top floor ceiling. Assuming the stairs are inside those windows, I'd call it a "glazed stairwell". Daveosaurus (talk) 07:18, 31 January 2026 (UTC)

Hah, then I stand corrected - seems glazed is correct English. Live and learn, and also, it goes to show that what some calls an error is not always one. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 11:18, 31 January 2026 (UTC)

FWIW I'm not sure I agree with daveosaurus. 'Glazed' typically implies a layer of some kind of lacquer applied onto another surface. I would not usually use it for something 'encased in glass' or 'surrounded by windows' or the like, nor would I use it for something made of glass; that would be 'glassy' or simply 'glass' Athanelar (talk) 11:57, 31 January 2026 (UTC)

See Glazing (window), which perhaps should be linked from "glazed". Glaze itself is a DAB with no primary topic. Donald Albury 15:36, 31 January 2026 (UTC)

@Nihil novi For the full context, see the top of this very thread, then the spot check of an article translated from Polish. ClaudineChionh (she/her · talk · email · global) 04:06, 25 January 2026 (UTC)

Thank you. Nihil novi (talk) 08:02, 25 January 2026 (UTC)

@Oliwiasocz Could you take a look at the above, and make changes where needed? As highlighted, it seems like several of the issues are from the source article, so I'd recommend fixing both the source and the translation 7804j (talk) 15:16, 24 January 2026 (UTC)

Why did this thread devolve (in my opinion) to just pointing out a few editors mistakes and cleaning them up? Or is it just yhis subsection? ~2026-51814-4 (talk) 15:51, 24 January 2026 (UTC)

This is a spot check, to see what the proportion of errors is on a random sample of articles. Of course, we can't physically check every single translation, so checking a sample of 5 or so allows us to see how frequent/rare errors are. Chaotic Enby (talk · contribs) 17:59, 24 January 2026 (UTC)

thank you. that spot check is a redlink though maybe the article you are thinking of is under a different name. ~2026-54906-2 (talk) 16:17, 25 January 2026 (UTC)

Yep, surprised we don't have an encyclopedia article on it. Wikipedia:Spot checking sources is one of our project pages on how we apply that practice. Chaotic Enby (talk · contribs) 21:17, 25 January 2026 (UTC)

Hi, concerning the mentioned issues:

1. I used "glazed" because in wiktionary it was described as "(architecture, construction) Having glass in the windows". Maybe using "stairwell with windows" would be better here? The word in the Polish article seems to be correct as well, since the Polish version of wiktionary also describes it as "one that has a large amount of glass surfaces".
2. Yes, it was supposed to be "...built between 1935 and 1937 and designed by...".
3. The original translation may have sounded confusing, so I think it could be reworded to "...slightly recessed ground floor clad in clinker tiles (similarly to the Kiltynowicz villa at 26 Katowicka Street)..."?
4. Yes, I have made a mistake with using "rounded windows". This phrase could be translated as "window openings" in a more technical way, but leaving just "windows" is completely fine. I also changed the phrase in the Polish article.

Oliwiasocz (talk) 00:28, 25 January 2026 (UTC)

Thanks for the clarifications on the Polish wording (and on the address as well, my mistake there) -- my gut sense was also that the original article was maybe leaning a bit much into architectural trivia.

Not sure, though, how pointing out factual inaccuracies in an article so they can be fixed and the article can be improved, though, is "devolving" -- it's literally the whole point of a wiki. Really not sure why you thought I was reprimanding anyone; if I wanted to reprimand someone I would just do that, there's a reason I didn't single out the editor. Gnomingstuff (talk) 23:44, 24 January 2026 (UTC)

Improving articles is all good. And I'll note that Oliwiasocz, who has translated many articles form pl wiki over the years, is, IMHO, a good example of positive value OKA has brought to the project. She has translated, among others, several of my articles, some of which were GA on pl wiki, and, after minor c/e, were recognized as GA here too. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 07:08, 25 January 2026 (UTC)

Issues with a DYK hook

Today, we had a DYK about a recently murdered woman. What did the people at DYK decide would be a good way to give her article attention on the main page?

• "... that Michele Singer Reiner shot Donald Trump for his first book?"

The added insensitivity of combining "shot" with someone who survived an alleged murder attempt by shooting was just a bonus. Discussion at WP:ERRORS, obviously, where the defence of this hook ranged from (paraphrased) "it's in the last position which is for quirky hooks!" (as if readers know that) to "it wasn't intentionally misleading" (but still this one was picked anyway to grab attention and to make it quirky).

Looking at Template:Did you know nominations/Michele Singer Reiner, it looks like everyone preferred a different hook but User:Surtsicna, who then continued to approve the hook against the clear consensus against his preferred hook.

Luckily, we have checks and balances at DYK, with two additional people (at least one an admin) looking at a hook before further promoting it to the main page. At least, that's the theory, in practice these two extra checks are way too often just rubberstamping, lke apparently here with User:Dclemens1971 promoting the hook, and admin User:Crisco 1492 simply copying it to the queues for the main page.

Wiser heads prevailed at WP:ERRORS, and in the end (after a lot of opposition by some DYK regulars) this got corrected, but why is it so often that such insensitive (and even in other cases racist or sexist, like "... that the first Slovenian restaurant to earn three Michelin stars was helmed by a woman?" from October) hooks make it to the main page in the first place? So please, admins, if some more of you could hang out around DYK and look out for such issues, things might perhaps improve. Fram (talk) 10:52, 20 January 2026 (UTC)

Hi there, I've also made a discussion inspired by this incident at Wikipedia_talk:Did_you_know#Specific_guidelines/rules_to_nominating_an_article_someone_else_has_worked_on?, regarding nominating articles that other people have created/expanded; I created the article and the nomination was made 2 hours later while I was still expanding the page by someone else. I had planned to make a nomination, but when the article was ready, but was pretty blindsided when somebody else did it. I had let it go since the nominator @Launchballer: is probably the most active person at DYK, but with this situation I feel like my proposed changes (see the section linked) should be implemented. jolielover♥talk 11:00, 20 January 2026 (UTC)

• Not going to get bogged down, but I am going to point out that I am taking "Simply copying it to queues for the main page" and the aspersion of incompetence as ABF. "Simply copying" is also demonstrably incorrect: I specifically indicated at WT:DYK that I expected that some may take issue with the term "shot" but that it was my judgement (perhaps biased by my own activities as a photographer) that the word was a common and established synonym for "photographed". Consensus was later against that judgement, and the hook was changed. That doesn't mean I "simply cop[ied] it to queues". It means that consensus was not with my judgement, which is a situation every editor finds themselves in at some point.

While the timeline was a bit tight in this instance, due to the DYK queues being empty at the time, there were still several hours for people to voice their concerns at WT:DYK after the potential issue was noted. Furthermore, WP:DYKQ is open to everyone, and the prep areas are not protected. Anyone who is concerned with areas of DYK can participate.  — Chris Woodrich (talk) 12:25, 20 January 2026 (UTC)

Consensus at the DYK nom was already against this hook though. If there are two synonyms, one which can be used without any issue and one which you know will cause raised eyebrows because of the very poor taste of the double meaning, then why did you go with the problematic one? Fram (talk) 13:13, 20 January 2026 (UTC)

You insist that we purposefully pushed something that we knew to be "problematic" and "very poor taste". You have labeled me personally as problematic for approving the hook. @Crisco 1492 is right in interpreting this as assuming that we are here to harm the project. There is no double meaning in "shot Trump for the cover of his book". Disagreement can be expressed without painting the rest of us as vandals. Surtsicna (talk) 13:38, 20 January 2026 (UTC)

If there was no double meaning (which was apparent to everyone else in the DYK nom discussion which you approved anyway), then why was it put in (and defended as) the "quirky" spot? It the meaning was so straightforward and couldn't be misinterpreted, then there was no reason to put it there. Your very first comment at the DYK nom defended the choice of "shoot because "it makes the hook that much more interesting" and in a later comment "we want to attract attention", and still later "attention-grabbing". Why would this specific word choice "grab attention" if not for the double meaning? Fram (talk) 14:00, 20 January 2026 (UTC)

Meaning 1: The photographer took a picture of the current president for his book.

Meaning 2: This individual attacked the current president with a gun for reasons to do with the cover of his first book. ExtantRotations (talk) 14:31, 20 January 2026 (UTC)

@Surtsicna Language exists in context. A recently murdered person combined with a contentious president who survived an assassination attempt is extremely relevant context to the DYK. It is surprising and disappointing that you would not understand that immediately-obvious context. qcne (talk) 16:59, 20 January 2026 (UTC)

The exact wording appears to have been "Michele Singer Reiner shot Donald Trump for his first book". No mention of the cover, which would have made this less bad. Indeed, it seems several of the defenses for this here and on the Errors page specifically say "cover",[15][16][17] unsurprisingly, because that is indeed useful additional context and more natural wording. Lack of context makes it easy to misinterpret, as it seems many did. LordCollaboration (talk) 17:41, 20 January 2026 (UTC)

• Yes, it's unfortunate that this stayed on the Main Page for nine hours after the obvious problems with it had been raised. This isn't the first time this has happened and I am perpetually bemused how many experienced Wikipedia editors don't get why we shouldn't be posting juvenile "quirky" hooks about living or recently deceased people on the front page of a website that gets millions of hits a day. Black Kite (talk) 12:35, 20 January 2026 (UTC)

  While there's no accounting for taste, japes about shooting a sitting President (regardless of one's persuasion – political or otherwise) are akin to joking about hijacking (or bombing) a plane while going through airport security. Likely to attract attention, for sure, but risky business all the same. If one is inclined towards this sort of humour, proceed with great caution (and assume responsibility for any unintended outcomes). -- Cl3phact0 (talk) 12:58, 20 January 2026 (UTC)

  These kinds of jokes wouldn't fly in a reputable encyclopedia, and I'm not surprised to see that this one landed poorly. Chaotic Enby (talk · contribs) 13:36, 20 January 2026 (UTC)

  yeah. whether or not it would have actually misled people, there's something to be said for the fact that even joking about a very-possibly-murdered BDP "shooting" someone who hated her and made (much worse) jokes about her death just isn't appropriate. theleekycauldron (talk • she/her) 13:59, 20 January 2026 (UTC)

  Out of curiosity, is there a simple way to look at the hourly page views for this hook during the time is was displayed vs. the modified hook(s) that replaced the original? -- Cl3phact0 (talk) 13:32, 20 January 2026 (UTC)

  there isn't, unfortunately. pageviews are released by day only. if we wanted, we could start running experiments on hooks to get a finer idea, but we'll never know for this one. theleekycauldron (talk • she/her) 14:02, 20 January 2026 (UTC)

  Noted. As there is a "vph" column here, I thought perhaps there might be a simple way to extract this information. -- Cl3phact0 (talk) 16:04, 20 January 2026 (UTC)

A few thoughts:

1. I reject Fram's assumption of bad faith. (See just rubberstamping, which an inaccurate aspersion; anyone who thinks I rubberstamp preps for sets should look at my body of work at DYK.) I conducted all required checks of prep builders and I did so thoroughly, and Fram should retract any statement otherwise. The most relevant provision in WP:DYKPBI is to check that no outstanding issues raised in the nomination remain relevant. @Surtsicna had conducted a full QPQ review and had approved. No one else in the discussion had questioned Surtsicna's review, as often happens, and no one had raised an objection at WT:DYK, as often happens, and three weeks had elapsed with no additional comments in the nom or at DYKT. It was a validly approved review with no apparent remaining objection. (Moreover, unanimity is not required in a discussion for a hook to be promoted.) You may dislike the hook, but don't say the promoter didn't do his or her job.

2. The hook was in prep for several days and @Crisco 1492 commented about it when he moved it to queue and again no one objected. (I do wish the page creator had raised a concern in the discussion or at DYKT; had I known they strongly objected to the hook I would not have promoted that hook, and I've apologized for the distress I caused by promoting a hook they objected to after the fact.)

3. There is no policy violation being alleged here. In fact, I considered carefully if questions of taste should have meant this hook not be promoted, but DYKPBI does not ask promoters to make such a judgment. Indeed, the governing policy is WP:NOTCENSORED. I have promoted a good many hooks whose framing or subject matter I find distasteful. I do this because my personal preferences should not govern what appears on the homepage; rather, policies and guidelines and community consensus should. No one participating at ERRORS actually alleged an error; the concerns were instead matters of taste, which ERRORS says should not be raised there. As a result, we should look at the pull as an WP:IAR outcome.

4. If Fram's desire is to get more admins to participate at DYK, this is going about it all wrong. Setting up a scenario where good-faith editors follow instructions and procedures and are then hauled to AN for very poor taste seems unlikely to build enthusiasm. We already have a chronic shortage of prep builders and queuers at DYK; indeed, noticing the extensive backlog of approved hooks is why I started pitching in to help with DYK last summer. This kind of Monday-morning quarterbacking of a process that already has ample opportunities for community input and consensus-building before anything reaches the main page seems calculated to put good-faith volunteers off of DYK entirely.

The page creator belatedly objected to the hook, so it seems like the right outcome was reached here in the end, even if there wasn't necessarily a consensus for the pull at ERRORS when it was made. Since there's no arguing about matters of taste, I'm not going to do so further here. If the community believes that DYK participants should add considerations of taste to the required checks, by all means, but that will make an already somewhat subjective process even more so. Dclemens1971 (talk) 14:30, 20 January 2026 (UTC)

I think the problem is that, regardless of whether the hook was misleading or not, if you think it might be on the boundaries of tastelessness (especially with living/recently deceased people) it's almost certainly better to go with the less interesting hook that isn't going to get dragged to WP:ERRORS 20 minutes after it goes live on the Main Page... Black Kite (talk) 14:52, 20 January 2026 (UTC)

"Rubberstamping" is not a bad faith accusation, it's the good faith accusation. The bad faith accusation would have been that you looked at the DYK nom discussion and the hook, and decided that yes, this was a good thing to post on the main page for an article about a recently murdered person.

The whole "no one at DYK objected" (well, apart from the objections at the actual DYK nom) is the reason I ask for more people (and admins) at DYK, as the difference between the opinions of people who frequent WT:DYK and those who don't should be apparent from this discussion (and too many similar previous ones). "There is no policy violation being alleged here. " apart from WP:BLP? "Indeed, the governing policy is WP:NOTCENSORED. " When you have the choice between two otherwise equally correct words, one neutral or unproblematic and one bound to be causing problems (as could be seen from the DYK nom discussion, and was the actual reason this word was chosen over the other), then hiding behind "butnotcensored" is a very weak cop-out to justify running with this. "I have promoted a good many hooks whose framing or subject matter I find distasteful. " Perhaps, but in this case you called it "an excellent DYK hook."

"t seems like the right outcome was reached here in the end, even if there wasn't necessarily a consensus for the pull at ERRORS when it was made." It was first changed at 07:15, 20 January 2026. At that time, objections had been raised by Krisgabwoosh, theleekycauldron, LuniZunie, HurricaneZeta, ~2026-41738-4, Randy Kryn, Rhododendrites, and jolielover (plus some people already present at the DYK nom). Support for the hook mostly came from the people who had proposed or approved it, you even tried to get the discussion closed just some 15 minutes after it had started. Fram (talk) 15:05, 20 January 2026 (UTC)

I am ~2026-41738-4, forgot to log in HurricaneZeta^C 16:27, 20 January 2026 (UTC)

I think there are ways to read this that are not misleading even if I did object to the hook as it was written.Rjjⁱⁱⁱ (talk) 14:31, 20 January 2026 (UTC)

Looking at Template:Did you know nominations/Michele Singer Reiner, it looks like everyone preferred a different hook but User:Surtsicna, who then continued to approve the hook against the clear consensus against his preferred hook. Fram clearly has a different definition of "everyone preferred" to me. Five people took part in the discussion; aside from Surtsicna, Launchballer (who proposed the hook originally) never indicated preferring any other hook and thus presumably preferred the hook which they had themselves suggested. Bagumba, who suggested the alternative "took a shot of" in response to an objection to the original wording didn't clearly indicate a preference either way. Only two of the five participants in the discussion, by my reading, clearly preferred to avoid "shot Donald Trump".

I agree with Fram that the hook was bad. I think in general DYK is too willing to run misleading clickbaity hooks. Where there's dispute about whether a hook is appropriate, one of the people involved in the argument shouldn't close the DYK discussion in their own favour. But two out of five people (seven if you count the people who then promoted and queued the hook) objecting to it is hardly "clear consensus", let alone "everyone". Caeciliusinhorto-public (talk) 14:58, 20 January 2026 (UTC)

Both Bagumba and Launchballer had no objections to reword it, only surtscina insisted on the original hook and then picked that one. If you have two hooks, where two people prefer hook 2, two don't really care either way, and one insists that it should be hook 1 because that will grab more attention (never mind that this will happen for negative reasons), then it's bad form for the outlier to review the nom and hook in his preferred version. Wikipedia works by consensus, and whatever way you look at it, hook 1 was not the hook with consensus at the time of the review. Never mind that it was 100% predictable that the issues raised during the DYK discussion would reappear when it ran on the main page. Some of the people involved may not have realised this, which I hope they will learn from. Some did this deliberately, which should disqualify them from DYK. Fram (talk) 15:12, 20 January 2026 (UTC)

I agree with the substance of your complaint, Fram, but the way you've raised it is unacceptable. Mistakes and errors in judgement can be made in good faith and the important thing is to find the root cause of the problem and prevent it recurring, not drive off the people who made the mistake. "More or better admins" in the section header (which I've changed) is an implicit personal attack on the admin involved and all admins who work at DYK and "rubberstamping" and "simply copying" are unfounded accusations of misconduct or incompetence. The sarcastic tone of your original post is only likely to antagonise the people involved. Constructive criticism helps people to grow and learn but your approach is only likely to make them defensive. HJ Mitchell | Penny for your thoughts? 16:47, 20 January 2026 (UTC)

As can be seen from later comments, "rubberstamping" was the best option to explain this. In reality, they looked at it, realised the probable reactions it would cause, and agreed to run it because of this. E.g. Dclemens already was very defensive at the Errors page long before I had posted anything, and nothing much has changed since. The root cause of the problem is too many people at DYK caring more about clicks than about getting it right (in different meanings of that word, from getting things factually right to being neutral, BLP-compliant, and open to criticism as with that DYK discussion). Some of the people involved outright stated that they wanted the possibility of a misunderstanding in the hook because that would be "attention-grabbing", which was apparently more important than, I don't know, being an encyclopedia which tries to avoid misunderstandings, and which tries to treat living and recently deceased people with some respect (just like in other older DYK issues some DYK regulars cared more about attention-grabbing than about basic respect for the subjects of the hooks, which just happened to be woman or people from non-ENglish speaking countries).

Constructive criticism was raised in the DYK nomination discussion. This is where it brought us. Fram (talk) 16:54, 20 January 2026 (UTC)

You'll catch more flies with honey than with vinegar. Constructive criticism resulted in the hook being amended. Your personalised and sarcastic post here has led to at least one participant temporarily withdrawing for their own mental health. I have my issues with the DYK process, which is why I don't participate there these days, but your approach to this thread is achieving precisely the opposite of your stated aim of encouraging more admins to participate there. HJ Mitchell | Penny for your thoughts? 17:37, 20 January 2026 (UTC)

+1. Ed ^{[talk] [OMT]} 17:57, 20 January 2026 (UTC)

+1 DoubleGrazing (talk) 18:00, 20 January 2026 (UTC)

Look, I agree with your criticisms of the hook. But your repeated misrepresentations of the discussion are not helping your case here. You originally said that Bagumba and Launchballer preferred the revised hook, which simply isn't true. Now you are claiming that Launchballer had no preference either way, which may be true but absolutely is not expressed by them in the discussion, and you could just as legitimately – in fact I would argue significantly more legitimately – read the discussion as them preferring the hook that they proposed. Caeciliusinhorto (talk) 19:04, 20 January 2026 (UTC)

I had written: Could reword to "took a shot of Trump", unless the community is ok with the quirkier wording and possible misinterpretation. —Bagumba (talk) 08:33, 21 January 2026 (UTC)

Maybe this part of the discussion is over (I know Wikipedians hate prolonged discussions) but I, a native English speaker who'd never heard of Michele Singer Reiner before this thread, definitely read it as her having shot Donald Trump with a gun because of his first book. In fact, I thought Fram's objection was that Donald Trump was made the focus of a woman's death. I had to read the replies to find out what the hook actually meant. ~2026-43450-9 (talk) 16:30, 20 January 2026 (UTC)

Everyone who would look at it would read it that way, which is likely what someone wanted. Lulfas (talk) 20:03, 20 January 2026 (UTC)

As Crisco 1492 noted, they did flag this for discussion, so they weren't merely rubberstamping. DYK can stand to improve its sensitivity toward negative hooks and quirkiness at the expense of living people, in this case getting "shot" (perhaps with the same vigilance DYK has with "first" hooks). Unfortunate that some participants continued to double down on "shot" being merely "technically correct". Still, any improvement in guidelines re: double entendres at the expense of BLPs is for the DYK community to sort out, not AN. The hook did eventually get addressed at WP:ERRORS, so the system isn't totally broken. —Bagumba (talk) 08:26, 21 January 2026 (UTC)

Suggestion: create a mechanism called "potentially controversial hook". Perhaps it's a template, or perhaps it's just a parameter of the nomination template. Either way, permit anyone to add it to a nomination. It would have two functions: 1) delay promotion by, say, 48 hours, and 2) solicits additional eyeballs either through an automatically updated running list of potentially controversial hooks at WT:DYK or some other mechanism. My impression is we do not need "better" admins, but that the scope and rapidity of the DYK system makes it easy for something to fall through the cracks. Having a forced delay and a way to flag things for others seems like a good idea. — Rhododendrites ^talk \\ 15:55, 20 January 2026 (UTC)

I don't think that would help. The problem is fundamental to DYK: in order to make the hooks "interesting", they have to be stripped of encyclopaedic context. When you add that to the speed and volume that are required to keep DYK going, which is what gives us so many poor-quality articles despite all the rules which get increasingly byzantine every time we have a conversation like this, something like this will inevitably happen again. That's not to say anyone participating at DYK is not doing so in good faith, though. HJ Mitchell | Penny for your thoughts? 16:14, 20 January 2026 (UTC)

It is true that DYK is fast-paced (dizzyingly at times), whereas Wikipedia as a whole tends towards slow and incremental. That said, it seems a stretch (and somewhat of a logical fallacy) to equate the quantity of poor quality articles added to Wikipedia with the supposed voracity of the DYK project. -- Cl3phact0 (talk) 16:25, 20 January 2026 (UTC)

I was equating the speed with poor-quality articles making it onto DYK, not with poor quality articles on Wikipedia in general. DYK is very much an insider process. HJ Mitchell | Penny for your thoughts? 16:28, 20 January 2026 (UTC)

Noted. Even so, in defence of DYK, WP:DYKNOT specifically states that it is not expected that articles appearing on DYK would be considered among the best on Wikipedia. Room for improvement? Sure. Candidate for censure or termination? Nah. (Nb. This defence of DYK is coming from a low-level participant who has had my fair share of frustrations with the insider processes you mention.) All in all, in spite of whatever criticisms one may have, DYK seems like a net positive (in the WP:1Q sense). -- Cl3phact0 (talk) 16:47, 20 January 2026 (UTC)

I don't believe it is. It consumes a lot of resources—editors' time willingly given to the nominations but also all the effort that goes into keeping the process running, the admins needed for promotion to the main page, and then the resources consumed when we have discussions like this—but for what benefit? We've had discussions like this innumerable times over the years and all that changes is that the rules have become more complicated and the expectations on reviewers have increased, making it even more of an insider process (because what new editor is going to understand all those rules!). Hence my reply to Rhododendrites' (good faith) suggestion, that the problems run too deep to be fixed by just adding more rules and my proposed solution of abolition. HJ Mitchell | Penny for your thoughts? 17:56, 20 January 2026 (UTC)

Again, I've certainly had my share of frustrations with the project, but I don't see eliminating DYK as a solution or the right outcome here. It's not beyond repair, although I would hope the "regulars" might be more open to suggestions for improvement after this most recent episode. -- Cl3phact0 (talk) 08:29, 21 January 2026 (UTC)

I was actually introduced to DYK as a new editor back in 2018. Someone noticed my first article and nominated it on my behalf. I was able to make sense of the general idea easily enough, although I was surprised that DYK functioned as some sort of effort to recognize newly created/expanded articles instead of simply collating interesting existing information. Clovermoss🍀 (talk) 23:47, 21 January 2026 (UTC)

@HJ Mitchell, here's a courtesy link to another recent example of DYK as a "net positive" (in my view, of course). Cheers, Cl3phact0 (talk) 09:38, 1 February 2026 (UTC)

I only do DYK noms when I have something "interesting", and I've never had this issue. I don't always agree with Fram, but I don't think they've really stepped out of line here. It's important that things like what happened here do not happen. I like a lot of the people who messed up here, but it was a mistake, and I don't think we should fault Fram for pointing it out. I tend to be somewhat direct myself, but I don't nessecarily see that as a bad thing. It makes someone's message more clear (like what I've been saying in this comment, for example). I do think Fram went a bit too far with the original section heading, but I don't think the general focus should really be on Fram here. There was a pretty big mistake made, it wasn't the first time something of this nature has happened, and that is magnitudes of order more important when what we do does not exist in a vacuum. Creating the perception that someone shot the president of the United States is a very noticeable thing that you do not want happening on one of the most visited websites in the world. This administration is already unfavourable towards Wikipedia and this is the country in which the WMF is based is not a good look. But generally speaking, I think we have a moral obligation to get things right when they appear on the main page. If something is too "quirky" to not be misleading when stripped of context, it should be rephrased or not included on the main page. Clovermoss🍀 (talk) 23:45, 21 January 2026 (UTC)

• (first editor who opposed the hook) As I am not a regular at DYK, I was surprised that Surtsicna did the review, as he clearly favored the original hook, and commented in favor of it. I would've wanted another editor to come do the review who did not have preconceived opinions on the hook. I did not want to comment on this at the nom, as I had already said too much and wanted to WP:DROPTHESTICK. After the hook got promoted and then posted, I found the thread on ERRORS was pinged to the thread by Bagumba. Natg 19 (talk) 17:47, 20 January 2026 (UTC)

  Situations like this are why I don't touch anything on the main page with a 10-foot pole. If the "regulars" want more people helping out, they don't act like it. —Rutebega (talk) 20:41, 20 January 2026 (UTC)

There are several levels of problem here:

• Surtsicna was in a WP:1AM situation at Template:Did you know nominations/Michele Singer Reiner and should have recognized it. The nom, Launchballer, proposed an ALT1 that added the word "photographer" and, had it been the hook that was approved, would have avoided this entire situation at WP:ERRORS. Bagumba suggested "took a shot of Trump," which also would have avoided WP:ERRORS. DrewieStewie agreed with "took a shot of Trump" and also suggested "took a photograph," a third suggestion that would have avoided WP:ERRORS. Natg 19 noted, presciently, that Sensitivity is absolutely warranted, supported ALT1, and said, again presciently, we should avoid the 'x shot Donald Trump' phrasing. Natg and DrewieStewie repeatedly pointed out that A simple reading of ALT0 sounds like an assassination attempt, or an actual shooting with a gun. and The problem is that “shot” means many different things. The objections could not have been clearer. Only Surtsicna advocated for ALT0 (repeatedly!), and their statements like So what if there have been assassination attempts? The hook is not saying or in any way implying that she attempted an assassination., "Shot" is as indisputable and appropriate, and ALT0 is concise, clear, and attention-grabbing, and I prefer it over ALT1. beggar belief. There is no doubt in my mind that Sursicna knew that "shot" would be interpreted to mean "with a gun," because multiple editors pointed that out, and that the reason they preferred the version without "photographer" (ALT0) rather than with "photographer" (ALT1) is because, in their own words, it was attention-grabbing. Also, since when do DYK reviewers get into WP:1AM arguments with others and still promote approve their preferred hook? I thought they were supposed to follow consensus, not exercise a veto over consensus? Because consensus was obviously against them. This was a major lapse in judgment and I hope it is not repeated.
• Dclemens1971 and Crisco 1492's statements are even more concerning that Surtsicna's. In Crisco's case, Admins are supposed to be the firewall in this process, the adults in the room, the editors chosen to be admins in part due to their good judgment. But Dclemens's statement at WP:ERRORS It is both technically correct and attention-getting, which makes it an excellent DYK hook. and here that says, basically, "taste" is irrelevant, and Crisco's acknowledgement that Possible DYK Errors posts about "shot", make me seriously question their judgment . They should have read that nomination, seen the 4-1, and not promoted the hook. They should have realized it was in bad taste and misleading, and that "technically correct and attention-getting" are not the applicable criteria. Another major lapse in judgment that I hope is not repeated.
• I sincerely hope both Surtsicna and Dclemens and Crisco learn from this.
• Jolielover raises a good issue about nominations made by an editor who is not the article creator, without the consent of the article creator. It ought to be a matter of courtesy to tell an article creator you're planning on nom'ing their article and make sure it's OK with them. One of the benefits of having the article creator involved in the nomination, even if they're not the nominator, is that the article creator will likely be more familiar with the article topic than any other editor, and thus may be able to flag any potential problems. I hope the discussion Jolie started at Wikipedia talk:Did you know#Specific guidelines/rules to nominating an article someone else has worked on? will lead to improvements in this area.
• In my view, all of the above are ancillary to the real problem, which has been noted already on this page by multiple editors:

  the timeline was a bit tight in this instance, due to the DYK queues being empty at the time

  We already have a chronic shortage of prep builders and queuers at DYK

  the scope and rapidity of the DYK system makes it easy for something to fall through the cracks

  When you add that to the speed and volume that are required to keep DYK going, which is what gives us so many poor-quality articles despite all the rules which get increasingly byzantine every time we have a conversation like this, something like this will inevitably happen again.

  It is true that DYK is fast-paced (dizzyingly at times)

  And if we needed further evidence, see these threads:

  Wikipedia talk:Did you know/Archive 210 § No queues

  Wikipedia talk:Did you know/Archive 210 § Queues

  Wikipedia talk:Did you know/Archive 210 § Suggestion for backlog

  Wikipedia talk:Did you know/Archive 210 § No queues, admin required (resolved)

  Wikipedia talk:Did you know/Archive 210 § All queues are empty

  Wikipedia talk:Did you know/Archive 210 § Modifying the required checks for admins

  Wikipedia talk:Did you know/Archive 211 § DYK is almost overdue

  Wikipedia talk:Did you know/Archive 211 § Empty queues

  Wikipedia talk:Did you know/Archive 211 § Empty queues 2

  Wikipedia talk:Did you know § Queues are empty

The real problem, IMO, is too much demand and not enough workers. Whether you go with Harry's Honey™ or Fram's Vinegar™, it is not realistic to expect that we are going to convince volunteers to volunteer to do something they are not interested in doing (this is true not just for DYK, but for sock hunting, copyvio cleanup, and all the other little tasks that few are interested in). The problem is: 9 slots is too many slots. While we have a backlog on the front end (235 approved noms, with many nominators having 100+, 500+, or even 1000+ prior noms), on the back end we just don't have enough people to keep those queues filled. And haste makes waste. In the rush to keep the pipeline moving, people aren't being careful enough. This is a very old story, it's been plaguing DYK forever. The answer is to stop trying to do 9 hooks/day (or sometimes 18/day), and just accept that we don't have enough volunteers to carefully check 9 hooks/day or more, and therefore we should reduce the number of hooks per day. Drop it to 4 or 5 a day and quality will improve. Put the nom's with 100+ prior DYKs at the back of the line, and new nominators will still be able to get their noms posted in a reasonable time even with a 4/5-hook set. Win-win, if only we start being realistic about our volume capabilities. Levivich (talk) 20:05, 20 January 2026 (UTC)

To clarify @Levivich, I am not an admin. (All DYK participants are invited to promote hooks, which was my role here.) I have certainly seen the feedback here and understand where the weight of the community's consensus is on hooks like the one in question, and obviously would not promote such a hook again. I also strongly recommend that a check for tastefulness be added to the required checks for reviewers, promoters and queuers, since so many commentators here believe it is a necessary consideration. (For my part, I doubt I will be promoting any more hooks. Doing promotions and queues at DYK is a chore that I was helping with because I perceived help was needed there, but the failures of AGF at many points in this conversation have made a mostly thankless task even less rewarding.) Dclemens1971 (talk) 20:15, 20 January 2026 (UTC)

Sorry Dclemens, my bad, I conflated you and Crisco. I've edited my post accordingly. Levivich (talk) 20:24, 20 January 2026 (UTC)

Agree completely with the last paragraph, as a long-term prepper and queuer who now rarely bothers to wade through gargantuan sets. ~~ AirshipJungleman29 (talk) 20:19, 20 January 2026 (UTC)

Just a note, Crisco 1492 (Chris Woodrich) is taking a Wikibreak due to stress caused by this issue. Natg 19 (talk) 20:36, 20 January 2026 (UTC)

Levivich's suggestion is a bit less dramatic than my call for abolition but I largely agree with it. If it were accompanied by an increase in the quality requirements for nominations (at minimum, double the amount of required prose) it might work. But I think such a proposal would get considerable pushback from people with many DYKs. HJ Mitchell | Penny for your thoughts? 21:15, 20 January 2026 (UTC)

Almost as if my "politicking in projectspace" does contribute to the goal of the community. Levivich (talk) 21:29, 20 January 2026 (UTC)

At the end of the day, the issue is this: telling people that you have spotted a flaw in their article, or that they had a bad idea, is not fun. I am just hard headed enough that I can drag somebody up at notice boards(as one editor here can attest, you should know who you are but you caused far more damage to a living person over 10 years than anybody at DYK did today and did nothing to clean it up) or Errors if needed, but its uncomfortable. That's not DYK exclusive, (Looking at you, Wikicup) and nor is it something only socially enforced at DYK (looking at you, recent memories of what happened when an editor tried to bring an admin up here for DYK PAG issues)- maybe tweaking the number of hooks changes that, maybe it doesn't, but the fundamental problem of "don't be a yes man" remains, and the day anybody solves that is the day that person gets their Nobel.

And, fundamentally, this could have been prevented by enforcing the rules we do have: @Dclemens1971, you know that checking for tastefullness in hooks covered by the BLP PAGs is needed; we literally had a conversation about it one month ago. [18]. I pointed you to WP:DYKBLP by name. While its understandable that you may have missed my ping, multiple other editors discussed the importance of not coming across as tactless. GreenLipstickLesbian💌🧸 11:11, 21 January 2026 (UTC)

I've learned from this discussion that the action I took within the chain of approval was well outside of community expectations for DYK, and I retract the defense of the hook I made at ERRORS. For my misjudgment, I apologize to all of the community members who have participated here and at ERRORS. As mentioned above, I am stepping back from promotions and queuing at DYK so you shouldn't expect to see a problem of this kind from me in the future. Dclemens1971 (talk) 19:39, 21 January 2026 (UTC)

@HJ Mitchell I'm not sure how doubling the minimum amount of required prose would help. I've promoted preps to queue and it's a LOT of time consuming work. In order to do it properly you should read the entirety of each article in the set, check the references to ensure they say what they are claimed to say, etc. Doubling the amount of required prose would double the amount of time needed to check each article. It might reduce the number of articles being nominated, but it wouldn't improve the time commitment that makes admins shy away from the work. One of the reasons I wanted to become an admin was wanting to help with that particular task, but I just don't have the time, which I don't have. ~ ONUnicorn^{(Talk|Contribs)}problem solving 16:05, 22 January 2026 (UTC)

noting that it wouldn't have prevented this case, as Michele Singer Reiner is nearly five times the current minimum length requirement. theleekycauldron (talk • she/her) 16:07, 22 January 2026 (UTC)

Indeed, this seems more a matter poor judgement and questionable taste rather than a question of any specific policy. The issue here is editorial discretion not the quality or length of the article itself. -- Cl3phact0 (talk) 16:29, 22 January 2026 (UTC)

I went for that because it's easily measurable, but it would reduce the number of nominations. An alternative would be to consider only GAs, as a review as through as it's supposed to be for DYK is almost a mini GA review. I still favour scrapping it altogether and replacing it with something else though. I don't think it serves the encyclopaedia. HJ Mitchell | Penny for your thoughts? 16:48, 22 January 2026 (UTC)

Blowing up DYK completely for one inappropriate hook seems overkill. -- Cl3phact0 (talk) 16:59, 22 January 2026 (UTC)

@Cl3phact0 I struggle to see how that could be a good-faith paraphrasing of my comments. A cursory reading of this thread would show that issues with DYK go back years. We were having conversations like this when I became an admin 16 years ago and they weren't new then. HJ Mitchell | Penny for your thoughts? 13:04, 23 January 2026 (UTC)

I obviously don't have your experience or longstanding involvement. I apologise if you took my comment as anything other than a good-faith contribution. Not my intention. If you look at my recent history with DYK (and the Kafkaesque labyrinth that on-boarding has been), you might also wonder why on earth I would defend the project. Alas. -- Cl3phact0 (talk) 15:52, 23 January 2026 (UTC)

I don't think it serves the encyclopaedia I view DYK and WP:ITN more as drives to create and improve content. Whether the actual hooks are blurbs on the Main Page are encyclopedic is secondary. —Bagumba (talk) 20:45, 22 January 2026 (UTC)

I can't speak for anyone else, and with my seven DYK noms I'm still a spring chicken, but DYK was a huge motivation when making my first article and continues to be. It got me to make and improve articles because it is, for lack of a better word, awesome that my work is displayed on the front page of one of the world's largest websites. 1brianm7 (talk) 21:12, 22 January 2026 (UTC)

I agree, it was huge for 16 year old me to write an article that anyone could read on Wikipedia, let alone the feeling of it being on the main page. Clovermoss🍀 (talk) 21:19, 22 January 2026 (UTC)

It was huge for my middle-aged self as well. It's what originally got me drunk with power hooked on editing. Levivich (talk) 22:05, 22 January 2026 (UTC)

You say that you "favour scrapping it [...] and replacing it with something else". Leaving aside how we would explain to the millions of readers who start their day on our main page to check out our features (as I do) that it was all a big mistake and we're sorry, but DYK doesn't actually meet Wikipedia's editorial standards or "[serve] the encyclopaedia" – out of curiosity, what would you replace it with? -- Cl3phact0 (talk) 07:46, 23 January 2026 (UTC)

Also, since when do DYK reviewers get into WP:1AM arguments with others and still promote their preferred hook? I thought they were supposed to follow consensus, not exercise a veto over consensus? Because consensus was obviously against them. We do not. I reviewed the hook. I did not (and could not) promote it, and I did not (and could not) veto anything. The reviewer is there to ascertain whether the hook and the article, in their opinion, meet DYK criteria, not whether there is a consensus that they do. Surtsicna (talk) 21:11, 20 January 2026 (UTC)

Sorry, you approved the hook, not promoted it, in DYK parlance; I edited my post to fix that. But the point stands. Your comment was ALT0 is concise, clear, and attention-grabbing, and I prefer it over ALT1. I guess there's a little ambiguity there as to whether you had approved just ALT0, or both ALT0 and ALT1. But your statement about preference... I don't see anywhere in WP:DYKRI that says the reviewer should express a preference as to which hook they like better. Did both hooks meet the criteria, or only one? It's been years since the last time I was directly involved in a DYK nom, but as I remember it, when reviewers became parties to the discussion about hooks and got into expressing preference or suggesting hooks themselves, they would do the (ask for a fresh reviewer) rather than (approve what they personally felt was the best hook). I believe this is the point made by Natg above. Levivich (talk) 21:25, 20 January 2026 (UTC)

The reviewer may indeed express a preference and reviewers do that regularly, but their preference is not binding on the promoter. A reviewer cannot approve a hook they themselves proposed, which is when another reviewer is requested using . Surtsicna (talk) 22:08, 20 January 2026 (UTC)

The implied suggestion that an approving editor should not express a preference as to which hook proposed by others should be used is, in a word, preposterous; that's a standard thing to do and always has been, and there's absolutely nothing wrong with it. - The Bushranger _{One ping only} 02:01, 21 January 2026 (UTC)

yep. the promoter is free to respect the reviewer's preference or not; they can promote any validly approved hook (i.e. one that there is consensus for). theleekycauldron (talk • she/her) 02:13, 21 January 2026 (UTC)

Wait, that parens seems to contradict the rest, I don't think I'm understanding. A reviewer can approve a hook that has consensus, or whichever hook they think is best? And a promoter can promote any approved hook, or the one that has consensus? (I get that all the approved and promoted hooks must meet the criteria.) Levivich (talk) 03:42, 21 January 2026 (UTC)

Consensus is usually assessed only as up-or-down for each hook individually, not as between hooks. If consensus prefers ALT0 to ALT1 (which I've never really seen happen, it's not something that ever gets assessed), the promoter can still pick ALT1; if consensus is that ALT1 is unsuitable for DYK, the reviewer cannot approve it and the promoter cannot promote it, as they need to check whether the hook is validly approved. (This is also why promoters are generally not supposed to promote a hook when the review is very clearly deficient, such as a comment that is just a check mark.) Reviewers can indicate preference between the hooks they've approved, but they can't reject a hook just because they like another one better (although they can pick which ones they want to review in the first place). theleekycauldron (talk • she/her) 03:54, 21 January 2026 (UTC)

I don't think DYK is irreparable. It does a lot of good for the project, both on the reader and community sides of the equation. But I agree that it has been trying to do too much. We run a lot less than we used to – 12-hour sets are a lot less frequent these days – and I would say that errors have gotten less frequent as the years go by, but we should be pushing further in this direction. I don't agree that we should be deprioritizing experienced nominators – they are not perfect, but they're more likely to be reliable sources of hooks that you generally know are going to be interesting and factual. Newer nominators are a grab bag, although they're vital to the equation too. I think DYK should move in a direction kind of like FAC; if there's not a sufficient amount of momentum gathered in a certain amount of time, that hook is probably not a better choice than one of the 200+ other hooks waiting in the pile. If we run fewer hooks, and extend the reach of the kind of no-fault fail FAC has (WP:DYKTIMEOUT is currently the best we have and it's doing ... meh), I think that'd be a big step in the right direction. theleekycauldron (talk • she/her) 04:07, 21 January 2026 (UTC)

@Theleekycauldron: I'm assuming you have a lot of thoughts on how such things could be prevented? Regardless of the framing Fram used, I do think the general idea of more people involved is one of the easiest ways to prevent something like this from happening. I like Levivich's idea of a priority queue for the more experienced nominators as a newbie that got to have my day in the sun early on. I haven't done much from the reviewer side of things on DYK, but I've always wanted to. I've reviewed one hook and I think I've pointed out some possible issues when I've noticed them in the past, but I've never really been actively involved in the process beyond the occasional nomination myself. It's hard to know when things are dire enough that my help would be useful instead of a hindrance. This may be a bit tangential to ANI itself at this point, but if there's a conversation being held elsewhere about these things, feel free to tell me about it. Or have a generalized discussion on my talk page. Clovermoss🍀 (talk) 00:01, 22 January 2026 (UTC)

ITN operates at a much lower volume - what people are proposing as the fix here - and there are still issues not to mention many observe how unpleasant ITN discussions can get. I think that if the community feels the only acceptable rate of ERRORs - and it's worth noting that this was not an error in the more typical sense that information was wrong either in the hook or the article, it was an error in judgement - is zero for DYK on the front page than we either need to radically rethink DYK (operating at an even slower rate than what Leeky proposes here) or abolish it altogether. Or we come to some kind of consensus about what an acceptable error rate for DYK is and measure ourselves against that consensus and make tweaks if/when our error rate exceeds our consensus for what is tolerable. As someone who basically abandoned doing DYK after a factual error was introduced into one of hooks without my knowledge, I would still prefer this latter approach. No encyclopedia is without errors and we have many other incredibly high profile articles which receive traffic over much longer periods of time which do not meet the standards we set for DYK. I acknowledge the difference that we're not promoting those articles we do in the way we promote DYK (and thankfully we don't have algorthymic promotion) so it's not quite the same, but I think a goal of 0 but an acknowledgement that we're willing to accept a bit of WP:TIMELYFASHION would be helpful. Though crucially even in that world we'd still likely have had this thread because wider input was needed in order to establish that there was an error (in judgement) in the first place. Best, Barkeep49 (talk) 20:12, 22 January 2026 (UTC)

I'm a big proponent of measuring and reducing the error rate. It doesn't have to go down to 0, which is unrealistic. Two years ago, I measured the error rate over two months and found it was about 95% correct / 5% errors (meaning 5% of hooks had an error that had to be corrected at WP:ERRORS). Some might say that's very good, or think there isn't much difference between a 95% accuracy rate (5% errors) and a 99% accuracy rate (1% errors). But in my mind, the difference is significant. At 9 hooks/day, 63/week, about 270/month, a 5% error rate is about 3 errors per week, or 13 per month. That's an error on the main page almost every other day, on average. That's too high (and even higher if we run 18/day double sets). The error rate should be, in my opinion, "a couple times a month or less," or about 1%. Moving from "almost every other day" (5%) to "couple times a month" (1%) is a worthy goal to pursue. Levivich (talk) 20:29, 22 January 2026 (UTC)

I personally think I'd be at 1.5% which would result in about 1 error a week, but that's all skipping a step. My bigger hope is to come to agreement that this is how we should filter these problems. One thing I would hope it also would do is to depersonalize some of this stuff. That is when the error rate creeps above our consensus rate we can have a productive process oriented discussion for the most part, but in cases where we would need to have a personal discussion (because an individual is causing some real % of errors) it would hopefully be closer to Harry's "build volunteers" approach. Best, Barkeep49 (talk) 20:39, 22 January 2026 (UTC)

I think that beyond statistics about errors, the type of errors matters too. Controversial hooks about living people tends to be are the type that leads to the conversations where people want to abolish the DYK process, so we need some better way to flag potential issues among the immense pool of hooks. There's a lot of work and a lot of burned out regulars, from my understanding. I don't think anyone is being actively malicious. Clovermoss🍀 (talk) 20:44, 22 January 2026 (UTC)

... we need some better way to flag potential issues among the immense pool of hooks: But it was flagged at the nomination—the approver ignored it. Then it passed through the set builder and promoter, and nobody flagged it in the queues either. It was flagged again at ERRORS, and hours passed before the hook was modified. There were many points in the existing process that the hook could have been changed earlier by the community. —Bagumba (talk) 22:33, 22 January 2026 (UTC)

Yes, my point is we need some better way of doing this because this isn't the first time our existing processes have failed. There needs to be a better way of getting people's attention, like a filter or something for hooks about living people, at the very least. That way someone can more easily check in on that without having to search for it amongst the other mountains of approved hooks. An alternative solution is to just not have hooks about living people at all, but I'd prefer not to throw the baby out with the bathwater. I'd imagine most of these hooks are really uncontroversial yet interesting facts about people's lives. Clovermoss🍀 (talk) 22:51, 22 January 2026 (UTC)

(somehow my reply was deleted by a bot) I am not an administrator but I wanted to share my recent experience with DYK. In my second DYK nomination, Template:Did_you_know_nominations/Andreas_Papandreou (died in 1996), I provided eight different hooks, reviewed without any objections. On the day, the selected hook presented on the main page, it was replaced with alternative not once but twice [19]. Each hook is factually accurate as they were supported by sources (and more could have been provided), but they were pulled out (as I understand) to calm down political sensitivities. So were these errors? I do not mind which one of those eight hooks provided would have been promoted, but I do find frustrating (to say politely) when there is backtracking for the wrong reasons.

IMO, it is a slippery slope to censorship once we impose restrictions of what constitute a hook and which articles are suitable for DYK. A good hook is like a good joke, whether one laughs or finds offensive is tricky and requires mastery but imposing restrictions is like prohibiting all jokes. My 2 cents in case this was helpful, otherwise ignore me. A.Cython(talk) 00:09, 23 January 2026 (UTC)

I don't think we should be making jokes about people on the main page, especially jokes of a political nature, and especially jokes about living people (even if in this specific example, the person in question was dead). Seems like a better option than avoiding hooks about people entirely, or getting rid of the whole DYK process. Censorship can be a slippery slope, but I don't think tasteful framing has that much risk. On the other hand, jokes have a well-known tendency to land poorly when you have a large enough audience. Clovermoss🍀 (talk) 19:12, 23 January 2026 (UTC)

I never said the hooks to be jokes. I said or at least what I wanted to say was that hooks no matter how tasteful or thoughtful will have the same problem as in making a joke. Some will like the hook and hopefully most of the readers, but there will be always some in the audience who will not like it. You will never going to please everyone. If we try to make DYK policy changes to please this minority then WP will wither, wane, and die. By pleasing everyone is a recipe for boring hooks. You say "tasteful framing", and i will ask you according to whom? Tasteful is subjective. Maybe I am wrong but why would I or anyone else read about something that is not interesting enough... People are interesting and we should not be afraid to tell stories and their failures. It is our failures that makes us human. A.Cython(talk) 20:30, 23 January 2026 (UTC)

I appreciate the clarification, some people do try to place jokes in their DYK hooks, especially on April Fool's Day. See [20]. There's lots of interesting facts people can use about people's lives without treading anywhere near that territory. The guideline at WP:DYKBLP exists for a reason and I don't think we need to think there's no point in setting limits because subjectivity exists. If anything is remotely controversial, it should be delayed, not rewritten right before landing on the main page. There should be a clear consensus and a tendency to err on the side of caution. Being tasteful isn't some impossible task. Clovermoss🍀 (talk) 20:37, 23 January 2026 (UTC)

I think part of the problem here is that people might be reluctant to mess around with hooks once they have been included in a prep set? It throws the order off. Maybe there could be some place aside from WT:DYK where hooks can be temporarily held if someone is concerned until there's a consensus for inclusion. I'd be willing to watch a page like that specifically even if I'm not constantly checking in on how DYK is doing. Clovermoss🍀 (talk) 20:43, 23 January 2026 (UTC)

It wasn't quite a factual error (although it gutted the point of the hook), but I stopped submitting articles to DYK 12 years ago after the approved hook for an article was changed without my knowledge (or any discussion that I was aware of) and the article jumped up in the queue by a couple of days so that by the time I was aware of the change the article had been up for nine hours. That may seem a petty thing to abandon DYK over, but I've found I don't need getting articles into DYK as a motivation for creating (and expanding existing) articles. Donald Albury 22:21, 22 January 2026 (UTC)

• I have no idea where to put my thoughts at this point in the discussion, so I'll stick them down here. I'm another who used to be very active at DYK (including in my case briefly adminning there) but walked away from it years ago. My history with DYK is old enough that I remember resenting Fram as a scold there. Crisco 1492 came along as a DYK admin later and I know him as a good one. When I adminned at DYK, I saw it as important to check the hooks and the articles for issues, and I sometimes found some; in this case both the prep builder, Dclemens1971, and the admin who promoted the set to the queue, Crisco 1492, apparently saw issues but didn't follow their better instincts and change to an alternate hook, which is all it would have taken. And I was also active at ERRORS for a while (because I saw a call on this noticeboard for more admins to watchlist it), and for a BLP issue such as this to be raised and not actioned until the hook has been up for 9 hours is hard to excuse.
  Levivich talks a lot of sense here. I once would have pushed back against "firehose" arguments, but cutting the number of DYK hooks appearing daily would be a lot better way to get enough scrutiny on problematic articles and hooks than would eliminating DYK. It's become clear that there just isn't enough editor-power to go around, and I can't think of a replacement for the quid-pro-quo reviewing requirement, which is always going to result in some inadequate nomination passes.
  The question is how to limit the number fairly. "Interestingness" can't be legislated (a solution that's been proposed again and again), although it is supposed to be a factor in nomination discussions. I don't think giving preferential treatment to experienced nominators is a good idea (or fair): anyone can mess up (in this case the nominator, who proposed and strongly preferred the problematic hook, was highly experienced), and with repetition of any task, there's a tendency to drift from standards; and one of the purposes of DYK is to showcase the work of new editors, which is the original reason for having third-party nominations as well as creator nominations. Like Clovermoss, I first became aware of DYK when one of my articles was nominated. I was against the incorporation of newly-minted GAs into DYK, where they fit oddly at best with the newly created and newly expanded articles. The simplest way to limit the number of DYKs would be to reduce the number of nominations by reversing that decision (which was controversial at the time, with a messy RfC); or even to reduce the number of nominations more sharply by restricting eligibility to new articles, which would have the advantage that the articles to review would tend to be relatively short. Another angle would be to deprecate third-party nominations unless the article is by a relatively new editor; let's say, one of someone's first 5 articles created. At the least, it was impolite to jump in and nominate someone else's article 2 hours after creation, without prior consultation, as Launchballer did here with Jolielover. I was glad to see there is now an "I've nominated your article for DYK" template; I don't believe there was one in my day; but still! And there's a perennial issue with tussles between article creators and others at DYK nominations over the hook; instances have been raised in this discussion. Plus editors who for whatever reason don't want their articles to appear in DYK (maybe they resist the notion of reducing the topic to an interesting factoid). Reducing the number of third-party nominations would help with those conflicts too. (I know DYK nominations are worth points in the WikiCup, but massive DYKing for WikiCup points has triggered rule changes in the past. Contestants will find other ways to score points.) Yngvadottir (talk) 01:02, 23 January 2026 (UTC)

  Regarding "enough scrutiny on problematic articles": but this one did get scrutinized when nominated, when promoted, and just 20 minutes after hitting the main page. I may write up a timeline to reflect on, but I feel wary about alienating folks I both respect and see taking a step back. There is a flaw in the way DYK currently operates: if someone has a bad idea, sticks by it, and nobody has a better idea, that bad idea can sometimes make it to the main page if they just wait it out. Natg 19 spotted this issue with the hook's language over a month ago. That should have been enough but wasn't. Rjjⁱⁱⁱ (talk) 08:07, 23 January 2026 (UTC)

  Sometimes that's it, but I don't think that was it this time. Multiple people raised the objection at the nomination, and proposed alternate hooks / tweaks to the original hook (better ideas); yet someone who disagreed with the objection accepted it with that hook. Crisco 1492 had doubts, so flagged it as maybe needing discussion at WT:DYK; that's where it should have been caught. I see a combination of not wanting to make waves and it getting lost in the other discussions at WT:DYK, including the others that Crisco 1492 flagged in the same edit. And since there will always be mistakes, please, more admins watching ERRORS. Yngvadottir (talk) 08:32, 23 January 2026 (UTC)

01:02, 23 January 2026 (UTC)

• Like Yngvadottir I stepped away from DYK some time ago. Not all the reasons are germane here, but one of them is, which is DYK's broad reluctance to reject hooks. Whether because interest is subjective, or hurting the feelings of established editors is difficult, or some other reason, we have been unwilling to say "no" to as many hooks as we should. Not every article is suitable for the main page and not every article is suitable for DYK. Those facts aren't generally reflections on the people who wrote them. I don't doubt that cutting throughput at DYK will help a little with the error rate, but dropping the number of hooks per set simply makes the backlog balloon and creates pressure to move to two sets a day, unless we also cut the input. There's a number of ways to do this. I disagree with Yngvadottir that interest cannot be legislated, at least in theory - we can certainly afford to tighten our guidelines. We could further limit eligibility, but as we move toward an encyclopedia with more bad articles relative toward nonexistent articles I would prefer we didn't do that. Another idea which may help our present predicament is to forbid altogether hooks related to living people and/or commercial ventures, which in my experience most often have concerns around appropriateness. But ultimately we need a cultural shift in the willingness to say "this isn't the best fit for DYK" and moving on. Vanamonde93 (talk) 18:15, 23 January 2026 (UTC)

I think that’s reasonable. Dronebogus (talk) 11:37, 29 January 2026 (UTC)

Proposal to change topic ban from "artificial intelligence" to "Stable Diffusion"

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

Since 2023, I am currently serving a topic ban covering artificial intelligence. I am not contesting any wrongdoing on my behalf which led to this topic ban, and fully accept the non-constructive nature of my actions during that time period, however I would like to propose that my topic ban be narrowed down to the topic of Stable Diffusion specifically. My rationale is as follows:

1. As of 2026, it has become increasingly difficult to avoid the topic of artificial intelligence when editing tech-related articles. Want to write about Google? They have made AI advancements in recent years, and it's all over recent tech news. How about Microsoft? They've made similar inroads, and have invested billions of dollars into datacentres and R&D in this area. Nvidia? The same applies. The US Air Force? You bet. Kanye West? His latest album is made using AI. The broader topic of AI has become more commonplace than it was back in 2023, and thus it is becoming more difficult to avoid AI when writing about topics that might touch upon it.
2. Since the topic ban, I have observed a strict 1RR, have made zero personal attacks, and have participated in zero edit wars. I've made a genuine attempt to approach editing with more constructive behaviour compared to my earlier style in the past.
3. Based on third-party tools, 124 people have my talkpage in their watchlist. If I started to misbehave tomorrow, I'm sure all of you will find out very quickly. Logically speaking, there's no benefit to cutting off my nose to spite my face, so the likelihood of this topic ban adjustment resulting in any increase in disruptive editing is extremely low. That is, of course, if you believe that I am a rational actor on this project.
4. I am not requesting for a removal of the topic ban, only that it be narrowed down to the topic which I was originally disruptive on, given the recent change in world circumstances. In order to fully regain the trust of the wider community, I expect that more time and more demonstrable action on my behalf is required. Thus, I do not have any plans to resume editing the Stable Diffusion article, or topics which mention it.

Thoughts, concerns, feedback, or suggestions? --benlisquare_T•C•E 15:57, 22 January 2026 (UTC)

Given the reasons the topic ban was originally imposed, I think this is a reasonable loosening. I would suggest something down the lines of "AI-generated art", instead of specifically Stable Diffusion, though. SarekOfVulcan (talk) 16:08, 22 January 2026 (UTC)

I would be fine with "AI-generated art" or "AI image generation" as the topic ban. --benlisquare_T•C•E 16:15, 22 January 2026 (UTC)

Given how few people will actually know what 'stable diffusion' is, it would seem wise to make the scope of the topic ban more understandable. 'Ai image generation' seems clear enough. AndyTheGrump (talk) 16:30, 22 January 2026 (UTC)

Yeah, that works for me too. Anyone else want to chime in, or should we call it good? SarekOfVulcan (talk) 18:03, 22 January 2026 (UTC)

@SarekOfVulcan, we can't "call it good" inside of 24 hours, as this is a ban appeal. -- asilvering (talk) 18:07, 22 January 2026 (UTC)

I agree that this topic ban has become broader than it was when it was set, and furthermore the locus of disruption was AI-generated images in particular. I support changing the tban to "AI-generated images". -- asilvering (talk) 18:01, 22 January 2026 (UTC)

Support AI-generated images. — rsjaffe 🗣️ 18:36, 22 January 2026 (UTC)

Support restricting the topic ban, given the much broader scope that AI took over these last few years and the lack of disruption. Chaotic Enby (talk · contribs) 18:41, 22 January 2026 (UTC)

Comment Personally I wouldn't have downgraded their indef to a topic ban in the first place (Original blocking thread), but OTOH I presume they're intelligent enough to understand that any repeat of the nonsense that got them an indef in the first place will get them another one. Black Kite (talk) 18:44, 22 January 2026 (UTC)

• Oppose - did I read a different original blocking thread? It's the one where the appellant was blocked for repeatedly insisting that there was no problem with them prompting a genAI to create dozens of images of sexualized young women with comically large ... attributes ... and inserting them into articles as encyclopedic visualizations? This comment by Levivich and Benlisquare's bizarre defense of it demonstrate an editor who should be nowhere near the topic of AI. Ivanvector (^Talk/_Edits) 19:44, 22 January 2026 (UTC)

  Those comments are entirely about AI images. Frankly, the issue here in my mind is much less the AI bit and much more the... everything else. But the tban he got was from AI. -- asilvering (talk) 19:53, 22 January 2026 (UTC)

  I don't know if I would have unblocked, but given that unblock, the proposed topic ban serves the community better than the original would, in my opinion. SarekOfVulcan (talk) 14:26, 23 January 2026 (UTC)

• i can't imagine what more we'd ask for from an editor in an unblock request and a subsequent amendment request – there's a clear understanding of what was wrong with their conduct and every attempt made to improve on it. 3+ years is a long time, especially for a sanction stemming from immaturity. Unless there's more recent evidence of misconduct I'm missing, I support a narrowing to "AI-generated images", would probably support narrowing it to "adding AI-generated images to articles", and could pretty easily be convinced to support the topic being vacated entirely. theleekycauldron (talk • she/her) 20:04, 22 January 2026 (UTC)
• Oppose. I'm with Ivanvector. We already have enough people writing about AI. There are almost infinite topics that Benlisquare can write about. NinjaRobotPirate (talk) 23:00, 22 January 2026 (UTC)

  The difficulty with writing about those infinite topics is that the moment AI brushes against those topics, there is now a need to carefully tiptoe around it, or avoid that topic completely altogether. This was manageable back in 2024, but it's not as easy now. The US economy is seemingly convinced that it needs to shovel hundreds of billion dollars towards investing into AI until it bursts, and because of this, writing about Dynamic random-access memory, Greenhouse gas emissions, Spotify, Media coverage of the Gaza war, Sustainable Development Goals, Andrew Cuomo, Call of Duty: Black Ops 7, Adobe Creative Cloud or Samsung Galaxy S25 becomes an issue. While the easy and obvious answer would be to just "be careful" about it, the overlooked aspect is that doing so would still risk a misinterpretation powderkeg waiting to happen, and the last thing I desire over the next 10 years is another trip to WP:ANI. --benlisquare_T•C•E 01:02, 23 January 2026 (UTC)

• Oppose I am the administrator who indefinitely blocked Benlisquare. Their behavior was both atrocious and bizarre in the extreme. The editor can write about Google as long as they do not write about Google's AI work. The same is true of Microsoft, Nvidia, the US Air Force and Kanye West. The editor can write about historical tech topics that predate AI as a "big deal". The fact that this editor has just now attempted to alter their topic ban to something as narrow as Stable Diffusion is a warning sign. There are countless AI tools in addition to that one that could get this editor in trouble. As for misinterpretation powderkeg?? Give me a break. It's simple. Do not edit about AI and there will be no powderkegs for Benlisquare to detonate. Cullen328 (talk) 05:01, 24 January 2026 (UTC)

  Your suggestion "do not edit about AI" excludes me from editing Age of Empires, a game released in 1997: Age of Empires allows players to choose to play either along specialized, story-backed conditions or as individual battles against the AI (and other players). Choosing to battle against the AI – rather than following the storyline – allows the AI to adapt to players' strategies and even remember which games it won and lost. The AI eventually overcomes players' strategies and easily destroys their villages after several games.. AI does not only cover LLMs and GANs invented after 2020, it covers anything that a computer is able to make a decision over. A computer program made in 1970 that emulates a game of chess uses AI. Your suggestion is as good as "do not edit about anything that you are an enthusiast about, or a professionally-qualified SME for, you may only edit about crocodile species in Queensland". Personally, I'd rather get rid of the uncertainty from living in the grey zone, and have what I am and am not allowed to write about better defined - this benefits you as much as it benefits me. --benlisquare_T•C•E 06:49, 24 January 2026 (UTC)

  I do not care in the slightest that you are unhappy that you cannot edit Age of Empires. Had you not engaged in grotesque and shocking misconduct, we would not be having this conversation and you would be editing merrily wherever you want. Your wikilawyering here just makes me feel more strongly that the current topic ban should stay. You are not limited to crocodiles, or even alligators or caimans. Or tortoises or asteroids or butterflies or the Trojan War. There are literally millions of articles that have nothing whatsoever to do with AI and countless opportunities to improve articles about companies like Google, Microsoft and Nvidia without breaching your topic ban. Cullen328 (talk) 07:15, 24 January 2026 (UTC)

  There is no "uncertainty ". There is no "Grey zone". Do not edit about AI. Cullen328 (talk) 07:39, 24 January 2026 (UTC)

  I'm not wikilawyering, I'm explaining my point-of-view, which I am entitled to do. Were our roles reversed, in the spirit of collegiality I would gladly grant you the right to explain your point-of-view, even if I disagree with it. I'm also civil in my communication with you, and I'd like to formally request reciprocation from this point onwards. We have not interacted in 3 years, so I do not understand the choice of words. Last time we talked, you had a different president, and I just got a new king.

  As for the point about crocodiles, I would like to make clear that I have no competency to write about such topics, as my understanding in such areas is minimal, and my expertise lies elsewhere. This is the ultimate crux of why I'm requesting for this tban change; I am not conspiring to secretly engage in disruption, I am seeking, with the community's permission of course, more wriggle room to write about the topics I have competency in. --benlisquare_T•C•E 08:56, 24 January 2026 (UTC)

  @Cullen328: I think that the original behavior was tasteless at best, but we're trying to build an encyclopedia, no? It's entirely possible that Benli could be qualified to write about Age of Empires, and if so, it's good for the wiki as a whole. Stikkyy (talk) 22:10, 30 January 2026 (UTC)

  Stikkyy, to be perfectly clear, this editor has not yet earned back my trust. I am aware that other editors disagree. Cullen328 (talk) 03:35, 31 January 2026 (UTC)

• I'm with Asilvering on this – the original behavior was horrendous, but it's really not clear how the wider topic ban prevents repeat offenses any better than the narrow one would. Neither prevents further Commons uploads in the same vein and neither would prevent additions of such images to Wikipedia – there was a separate unblock condition banning that. As such, I guess I support narrowing the tban to only cover AI image generation. Toadspike [Talk] 05:17, 24 January 2026 (UTC)
• I was initially inclined to be supportive of the narrowed scope (to AI image generation broadly; not Stable Diffusion specifically). Benlisquare's repeated arguments above give me pause, however. They say to Cullen above that the current topic ban prevents them from editing the article Age of Empires because it mentions that the computer controls units using AI. I don't believe this is true: so long as they stay away from edits about the Age of Empires AI, by my reading they are within the bounds of their TBAN. If Benli believes that the scope of their TBAN is that extensive, however, then they should not be making edits like [21] and [22], both of which are to articles which mention video game AI. By my reading of the current restrictions, these are perfectly acceptable edits – but this suggests to me that either Benli is making these edits while believing that they are in breach of their topic ban, or at least without any particular care about whether they are complying with their restrictions, or that Benli is in this appeal disingenuously overstating the breadth of their topic ban. In none of those cases does narrowing the restriction seem wise. Caeciliusinhorto (talk) 21:51, 24 January 2026 (UTC)

  Of the two diffs you linked, I believe that they were within the "grey zone" I mentioned earlier above. In other words, I hope they're okay, and if I were hypothetically enforcing the tban against someone else, I would personally consider them okay, but I can also imagine a sysop who might disagree with such an assessment, and that is what I fear. In many jurisdictions, jaywalking is technically illegal, but often tolerated by police who do not care, but if someone really wanted to get rid of someone and lengthen their sentence, they'd lay on jaywalking and 15 other charges - thus, I would describe jaywalking as within a "grey zone". I'd like some peace of mind that even as I make every effort to follow what is expected of me conduct-wise, I don't get stung by something that I do within the "grey zone". --benlisquare_T•C•E 06:16, 25 January 2026 (UTC)

  Unfortunately, AI has overbroadened as the old uses for the word AI are now appropriated onto LLMs, GPTs and ML models. So for instance pathfinding and LLMs are now in the single category but they are very different. ~2026-53299-5 (talk) 12:22, 25 January 2026 (UTC)

• Support narrowing to AI image generation. The current TBAN seems overbroad to me. voorts (talk/contributions) 21:58, 24 January 2026 (UTC)
• Support narrowing. The world has changed since 2023 and a topic ban from "AI" is vastly broader and more nebulous as a topic area now than it was then despite Benlisquare doing nothing in the interim to merit an expanded topic ban. The original disruption was contained to the topic of AI image generation and they haven't (as far as I've seen) come close to that topic or otherwise repeated that behaviour in the interim so it would not be unreasonable to appeal the whole ban but they aren't doing that. They're just asking for the topic ban to be narrowed to better reflect the circumstances that it resulted from, and for it to better defined - it is better for everybody if those who are subject to topic bans and those enforcing topic bans have the same understanding of what is and is not covered. Thryduulf (talk) 13:23, 25 January 2026 (UTC)

Support narrowing to AI image generation on the basis that a TBAN of such breadth is no longer necessary to prevent disruption to the encyclopedia. The editor in question was blocked as a result of (a) their insertion of AI-generated sexualized imagery into articles and (b) their incivility and disruptive editing with respect to the former. As a condition of their unblock, they were subjected to three conditions including a TBAN from artificial intelligence, which they have observed for approximately two and half years. Given that the actions which resulted in their block were confined to the realm of AI-generated images rather than artificial intelligence more broadly, the need for a widely-scoped restriction has in my view lapsed with respect to (a) given their behavior post-unblock. Likewise, seeing as the incivility in question also occurred within the realm of AI-generated images vs. artificial intelligence more broadly, that (as a condition of their unblock) further battleground behavior is grounds for an immediate block, and that the editor seems to be avoiding such behavior, I see no reason to reach a different conclusion than with (a). — chrs || talk 08:19, 26 January 2026 (UTC)

• Support - AI is used in a vast amount of modern technology, in some form or another (gaming, phones, watches, design, fridges, scientific research, maths). The original issue related to generative AI specifically, so a ban from AI image generation (or even "generative AI" as a whole) seems fair. Just "artificial intelligence" is much too broad IMO. Blue Sonnet (talk) 04:55, 27 January 2026 (UTC)

• Support - narrowing to AI image generation. AI is becoming nearly inescapable, and it appears that the user in question has acknowledged the issues behind the tban and in fact, in requesting simply a modification instead of a whole removal shows, in my opinion, learning from their mistakes. -- Aunva6^{talk - contribs} 21:18, 28 January 2026 (UTC)
• Oppose You don’t need to edit this area. Getting to edit at all is a privilege you should be grateful for. Anyone else more trusted and competent can. You can just avoid editing about most topics from the 2020s onward and computer topics in general. That still leaves a huge part of the encyclopedia. Your Tban was barely three years ago, and you were lucky to avoid a cban considering how egregious your actions were. If a topic ban was correct three years ago it is correct now. Dronebogus (talk) 09:07, 29 January 2026 (UTC)

  If a topic ban was correct three years ago it is correct now. this is fundamentally incorrect. A three-year-old topic ban might still be necessary but in many cases it will not be. What matters is whether the topic ban is needed now', not whether it was needed three years ago. I sincerely hope the vibes of this being sour grapes from your appeal not going the way you hoped are wrong (if I'm not wrong then that's more reason not to grant your appeal). Thryduulf (talk) 12:20, 29 January 2026 (UTC)

  To quote you in that discussion, That Dronebogus doesn't support the restriction now does not invalidate that they supported it at the time, so I'm seeing no procedural grounds on which to overturn. My extrapolation from that to this discussion is as follows: that the widespread adoption of artificial intelligence technology was not foreseen at the time of the ban does not invalidate the topic ban now because AI wasn’t exclusively used by image generation then either. Dronebogus (talk) 01:12, 31 January 2026 (UTC)

  Indeed it does not invalidate the topic ban, I have not argued otherwise. Nobody is arguing to overturn or narrow the scope on procedural grounds. What we are doing is evaluating, on substantive grounds and without prejudice, whether there are grounds to amend the topic ban to fit the circumstances of both the topic area and benlisquare's behaviour as they exist in 2026. Most editors here believe that there have been changes in one or both those aspects sufficient to justify a change. In contrast most editors in your appeal saw no evidence of changes in either your behaviour or the relevant topic area that warranted removing the topic ban.

  If you genuinely cannot tell the difference then I'm going to have to consider whether there is a WP:COMPETENCE issue here. Thryduulf (talk) 01:30, 31 January 2026 (UTC)

  I will also cite Cullen328’s reasoning: 1) his behavior was atrocious and the community should never trust him in this area again 2) he shouldn’t have done it in the first place if he didn’t want to be sanctioned 3) he is only slightly inconvenienced by a topic ban that is extremely easy to follow. Dronebogus (talk) 01:52, 31 January 2026 (UTC)

  1. His behaviour related to AI images was bad, the community is not being asked to trust him in that area now. Nobody has presented any evidence of poor behaviour in relation to anything broader than AI image generation.
  2. Irrelevant as he is not complaining about being sanctioned, nor is he asking for the sanction to be overturned.
  3. He is more inconvenienced by the topic ban today than he was in 2023, and the boundaries are much harder to identify now than they were in 2023. No evidence has been presented on any behaviour since 2023 that would justify banning them from a broader topic area. A topic ban works much better for both the sanctioned editor and the rest of the community when everybody can look at the same page and come to the same conclusion about whether it is or is not covered by the topic ban.

  Can you see how these three reasons are fundamentally different to all your previous comments? Thryduulf (talk) 02:04, 31 January 2026 (UTC)

• Support narrowing to AI-generated images (both the images themselves and the process of generating them). I'm convinced by the arguments that the existing ban is unnecessarily broad. If the same problems recur, we should probably be looking at harsher sanctions than a topic ban anyway. —Compassionate727 ^(T·C) 17:12, 29 January 2026 (UTC)
• Oppose Two major problems: (a) far too many red flag related WP:SAFEGUARDING issues and (b) the nub of this TBAN appeal appears to be that it should be narrowed because of its inconvenience. No. --Goldsztajn (talk) 20:40, 30 January 2026 (UTC)

  Isn't "inconvenience" the main reason why almost every editor who appeals a TBAN wants it removed? voorts (talk/contributions) 22:12, 30 January 2026 (UTC)

  As an editor who's spent almost 20 years here and not gone near literally 1,000s of topics, I remain unconvinced by appeals to convenience. Regards, Goldsztajn (talk) 23:13, 30 January 2026 (UTC)

  (a) is a very serious allegation, I don't think it's appropriate to be throwing it around on a whim without ample and firm reason. Such claims are potentially life-altering and reputation damaging, and it's not like a "whoops, my bad" afterwards can solve the damage caused. I abhorred the allegation back in 2022, and I will still abhor the same allegation today; my use of colloquial speech, the same colloquial speech used in modern-day rap music to say the least, is not advocacy or attraction towards juvenile harm, the idea of such harm disgusts me, and the allegation that I would partake in such harm also angers me. It is a nonsensical stretch completely devoid of objective fact, and back in 2022 it was a scorched-earth smear used by a minority of Wikipedia editors to poison the conversation so that there was no possibility that I could ever return to the community. Please retract your statement, because otherwise it appears that you believe the smears, and intend to spread them further, despite the evidence to the contrary. --benlisquare_T•C•E 22:27, 30 January 2026 (UTC)

  The link above to the descriptions of images generated by pattern-recognition algorithms - with repeated variations around themes of nude, young, female - *are* grounds for caution. An editor who has been here 20 years should be aware of this and understand why the issue requires such sensitivity. Posting lengthy responses to every singe oppose does not help the case either. Goldsztajn (talk) 23:54, 30 January 2026 (UTC)

  If WP:SAFEGUARDING governs what we are allowed to think about 19-year olds and 20-year olds, then I highly doubt that would be following the spirit of the policy. That policy is designed to protect children, in both the spirit of the policy, and the letter of the policy. I don't dispute these images were uploaded in poor taste and that obscene sexual content has no productive place on a non-sexual technical topic, but you should be bringing up WP:GRATUITOUS in this case. How else am I supposed to interpret your use of WP:SAFEGUARDING, other than an attack on my morals and character? --benlisquare_T•C•E 00:03, 31 January 2026 (UTC)

  I agree that sexualizing fictional adult young women isn’t grounds for serious concern. Unless you’re uploading realistic looking material, explicit lolicon-type stuff, or lots of softcore lolicon type stuff, anything fictional is not grounds for serious concern. Dronebogus (talk) 00:17, 31 January 2026 (UTC)

• Support a narrowing to the use of AI in images. I'm unconvinced that a broader scope than this would be preventative rather than punitive. Stikkyy (talk) 22:10, 30 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Blocking TAs

This has come up only a few times for me, but I wonder: When blocking a vandalistic TA, is it better to just block the TA or to block the underlying IP address? I think I've tended to block the IP with account creation disabled so that the IP couldn't just immediately create a new TA, but I'm not really sure whether there's a difference. Would blocking the TA with account creation disabled be an equivalent action? Deor (talk) 19:53, 25 January 2026 (UTC)

Would blocking the TA with account creation disabled be an equivalent action? Yes. voorts (talk/contributions) 20:02, 25 January 2026 (UTC)

Thanks, Voorts. Deor (talk) 20:58, 25 January 2026 (UTC)

As IP information isn't retained indefinitely, wouldn't it be equivalent to a 90 days block (at the software level) that is actually an indef block (at the "social"/block evasion level)? Chaotic Enby (talk · contribs) 21:19, 25 January 2026 (UTC)

One problem with not blocking the TA is that, when the IP block ends, you can't block a new TA for block evasion, as there's no block to evade. Indeffing the offending TA sets up a return TA for block evasion if it is clearly the same person. — rsjaffe 🗣️ 21:26, 25 January 2026 (UTC)

Yes, but that is exactly the issue. A new TA can't appeal the original block (since you need to be on the same account to do so) and will be summarily blocked for block evasion without even realizing it, as they might think their block just expired. Giving them what is functionally a 90 day block while actually making it an indef is basically setting them up for a block evasion block the next time they edit, even if in good-faith. Chaotic Enby (talk · contribs) 21:47, 25 January 2026 (UTC)

If they come out of the gate with different behavior from the prior TA, there's no indication to block. But if they repeat the prior behavior, I think a block evasion block is appropriate. I'm mainly thinking of disruptive people who edit logged-out. Some of the behavior takes a while to block on its own--but if you know they're a repeat of the prior person on that IP address, a block can occur quicker. — rsjaffe 🗣️ 23:52, 25 January 2026 (UTC)

Thanks! Chaotic Enby (talk · contribs) 23:55, 25 January 2026 (UTC)

What happens if you have blocked one TA on some IP address/cidr, and someone else tries to create a TA (edit) from the same IP addr/cidr? Are they blocked as well, or is it only tied to the session cookie for the blocked TA? --Gurkubondinn (talk) 22:55, 25 January 2026 (UTC)

Cookies are set on the user's system, so if a different person tries to create an account on the same IP, they will be able to. That is unless blocking an account sets an autoblock, the instructions are not clear. Ivanvector (^Talk/_Edits) 17:58, 26 January 2026 (UTC)

Yeah that was my question, if it would only block the session (identified by the cookie sent by the client), or if it also blocks or flags the underlying IP(s). Of course there are also ways to correlate a blocked session with a new one, was just curious what is usually done when a TA/session gets blocked. --Gurkubondinn (talk) 18:39, 26 January 2026 (UTC)

During the autoblock, TAs cannot be created on that IP. The second table at Wikipedia:Temporary accounts#Impact for administrators makes that clear. Most blocks include autoblock. — rsjaffe 🗣️ 21:32, 26 January 2026 (UTC)

Just to clarify: @Deor asked whether blocking the TA with account creation disabled is equivalent to blocking the IP with account creation disabled. @Voorts said yes. But I don't think that's correct. @Ivanvector wrote: "Cookies are set on the user's system, so if a different person tries to create an account on the same IP, they will be able to." That means if we block the TA with account creation disabled (but not the IP), the blocked person can simply clear cookies and create a new TA. Right? — Chrisahn (talk) 20:06, 26 January 2026 (UTC)

No. Blocking account creation means you're blocking account creation on that IP address. That's what autoblock does. voorts (talk/contributions) 20:14, 26 January 2026 (UTC)

Thanks. Again to clarify: This means that if a different person tries to create an account on the same IP, they will not be able to. Right? If yes, then @Ivanvector's answer wasn't quite correct. — Chrisahn (talk) 20:41, 26 January 2026 (UTC)

Oh, I just noticed that Ivanvector also mentioned autoblock. So it comes down to the question: Does blocking a TA always set an autoblock? — Chrisahn (talk) 20:45, 26 January 2026 (UTC)

No. it's a setting admins can turn on or off. voorts (talk/contributions) 20:46, 26 January 2026 (UTC)

That's the part I wasn't sure about. Autoblocks are set if you block an account, on the server side, so the user dumping their cookies doesn't matter. But temporary accounts aren't accounts, so I don't know if blocking one triggers an autoblock. voorts says yes; my experience suggests otherwise, but I really don't know for sure. Ivanvector (^Talk/_Edits) 20:50, 26 January 2026 (UTC)

Case in point: Yesterday, ~2026-52716-3 was blocked for 31 hours, but came back as ~2026-55974-9 27 hours later, on exactly the same IP. I guess there was no autoblock? Chrisahn (talk) 21:22, 26 January 2026 (UTC)

Autoblock runs for 24 hours after the block and 24 hours after the latest log in attempt. So an autoblock could have been in place but then expired. — rsjaffe 🗣️ 21:26, 26 January 2026 (UTC)

Thanks. Here's my current understanding: Blocking a TA for any length of time only blocks that IP for 24 hours. After 24 hours, the person can clear cookies and create a new TA. Which means that if we really want to block someone for longer than 24 hours, we need to block the IP. (Of course, IPs can also be changed, but that's usually much more work than clearing cookies.) — Chrisahn (talk) 21:43, 26 January 2026 (UTC)

Depends upon what we believe the blocked person will do. A decent number don't know about cookies and will remain blocked. Those who do do eventually need an IP block. See my comment below about scanning the IP history for clues as to future behavior. Most (or even all) of us start with the TA block unless we have a suspicion that cookie dumping will occur. And, indeed, that's what the first bullet point in Wikipedia:Temporary accounts#Impact for administrators suggests. — rsjaffe 🗣️ 21:53, 26 January 2026 (UTC)

If the TA hops IP addresses (I don't count moving around an IPv6 /64 as "hopping"), I just block the TA. If not, I might block the IP as well (almost always block the TA)--more likely to include the IP if there's a history beyond the current TA of disruption. If I fail to block the TA, it tends to confuse people who think the problem is unaddressed. — rsjaffe 🗣️ 21:25, 25 January 2026 (UTC)

My experience as a checkuser: if you reveal the IP address of the TA, you can see other TAs that have edited from the same IP, and you can expand that view to a CIDR up to /16 (/32 for IPv6). If they're using an IPv6 /64 you should just block it. If they're IPv4 and there are other TAs that have edited recently and are behaviourally similar, you're probably looking at a user rapidly logging out of and creating new temporary accounts to evade scrutiny, and in that case you should block their IP. Otherwise, blocking the TA is fine. You shouldn't block a temporary account and its IP without a good reason, because the block log is public, but sometimes the user gives you no other choice, and that is not the same sort of "you will lose your rights immediately" action as a checkuser revealing an account's IP address. Ivanvector (^Talk/_Edits) 17:58, 26 January 2026 (UTC)

Regarding possible disclosure through logs, WP:TAIVDISCLOSE says: "However, users with TAIV access are allowed to take administrative actions that, by their timing, imply a connection between an account and an IP (e.g. indefblocking ~2026-77890-57 and then moments later blocking 192.0.2.1 for a week [...])" Sounds like it's no big deal. (But of course, it's best avoided.) Thanks to @EvergreenFir for posting the reference in Wikipedia talk:Administrator intervention against vandalism#How to report ranges now? — Chrisahn (talk) 12:07, 27 January 2026 (UTC)

Yes, that's a better summary. A TAIV user should try to avoid it, but sometimes you can't. That contrasts with the much more rigid advice given to checkusers that we must avoid disclosure of an account's IP, but temporary accounts are not accounts in the same sense. Ivanvector (^Talk/_Edits) 16:24, 28 January 2026 (UTC)

For cases involving sockpuppetry with temporary accounts, it is policy-compliant to block the temporary accounts along with the associated IP addresses or ranges with a block reason such as "Block evasion" or "Long-term abuse" (typically for the IPs), which is the standard text used when specific IPs or usernames cannot be disclosed in the block log for privacy reasons. I would not hesitate to do this in response to sockpuppetry or long-term abuse. — Newslinger talk 10:45, 31 January 2026 (UTC)

Lifting Topic Ban

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

Saair (talk · contribs · deleted contribs · logs · filter log · block user · block log)

I was reading Topography of Pakistan, and a section mentioned area of Thar desert as 468 square kilometers, actually it is over 200,000 square kilometers. I was topic banned in 2021 from Pakistan, India and Afghanistan geography. It is about five years and my contributions confirm I didn't edit the particular articles. After five years, I am really not going to repeat my five years old self, and get myself bashed anymore. I assure positive contributions to Wikipedia from now on. I request the ban be lifted so that I'm not banned from editing the particular geography articles. Saair (talk) 15:16, 26 January 2026 (UTC)

I don't have an opinion on the topic ban, but I have corrected the error you mention in Topography of Pakistan. I've used the value and source given in Thar Desert#Geography although I don't have access to that source. Thryduulf (talk) 17:52, 26 January 2026 (UTC)

• Oppose. Although lots of time has passed, Saair has made almost no edits. Nothing to articles in 2026, nothing at all in 2025, only 7 edits in 2024, 2 in 2023. I'm not saying the topic ban can never be lifted, but I'd like to see a meaningful amount of editing outside of that topic area, given the years of problems before. --Yamla (talk) 18:10, 26 January 2026 (UTC)

  On further reflection, I'm going with Neutral here. --Yamla (talk) 23:27, 26 January 2026 (UTC)

• In your post earlier today about your IP being blocked, you mentioned that you discovered your IP was blocked when you tried to correct an error in Topography of Pakistan, an article covered by your topic ban. The block that prevented that edit is anonymous-only, meaning you were not logged into your account when you tried to make the edit. I also notice that your account hasn't made very many edits since you were topic banned, and before today your account didn't edit at all since October 2024. Have you been making other corrections like this while not logged into your account? Ivanvector (^Talk/_Edits) 18:13, 26 January 2026 (UTC)
  • I've not run Wikipedia logged out for years. I have been logged in for years. I think, the IP address surrounding my area is blocked since a long time. The last time I tried to make an edit was in August last, the IP address was blocked even then. Checkusers might confirm the IP block is there for a long time. This morning I was stunned to know the block was still there, so I requested for an exemption that Yamla gifted me with. I'd request the ban be lifted, as I'll be responsible for all my acts. Actually, I'm most interested in Geopolitics of South Asia and Middle East, and mostly visit these articles while using Wikipedia. I don't know why I was banned from editing India and Afghanistan geography related articles. I don't remember ever having edited geography articles about these two countries. And perhaps, someone would find little disruptions in the edits I made to geography articles related to Pakistan. I want where is the community discussion which would have established I was specifically disrupting geography related articles of these countries and a ban on me was necessary to save the particular articles from vandalism emerging from me. I want there is a fair evaluation of my respective edits, and if I'm really found disrupting the articles, let me face the consequences. Saair (talk) 18:46, 26 January 2026 (UTC)
    • To be honest, I'm not seeing the logic behind that topic ban either. You seem to have been originally blocked for sockpuppetry and disrupting an AFD, but I don't understand at all how that turned into an offer to conditionally unblock you with a ban from this topic area. You have a handful of edits to Chak Shafi and Arif Wala Tehsil and other Pakistani geography articles from way back in 2017, months before you were blocked. @SpacemanSpiff and 331dot: the conditional unblock seems to have been your idea, do you have any more context? Ivanvector (^Talk/_Edits) 19:03, 26 January 2026 (UTC)
      • I've made a handful of edits to Pakistan geography related articles, but those edits were in no way controversial. The topic ban was never imposed on me by any community discussion. If sock is in question, I was blocked for that in 2017, and I ceased to hold the sock accounts. The block regarding sock was lifted in 2021, but the topic ban without any rationale is following me since then. Saair (talk) 19:42, 26 January 2026 (UTC)

      I have reviewed the discussion but it was five years ago and nothing about this matter stands out to me. I assume that the TBAN was somehow related to the block but I can't recall how. 331dot (talk) 20:44, 26 January 2026 (UTC)

      • ⛏⛑ Digging time, though the best I have is that the numerous articles they started that got deleted, the disruption at Teahouse (including blanking the Teahouse once), at the one AfD, the sockpuppeting incidents from around May and June 2017, the multiple sockpuppeting incidents up to November 2017, and the focus on articles around Pakistan up to the end of 2017 got rolled up into a belief that Saair was being disruptive in the area. (Regarding the deleted articles, I don't know of an easy way to link to them or see all of them, other than the talk page notices that they received. If my count is correct, there were three A7 deletions (with one duplicate), six G7 deletions (with one going to Articles for deletion first and with another being a draft), one Redirect for discussion that was an article Saair created prior to the redirect with 104 revisions in the end, two Article for deletions (excluding the G7 deletion), multiple user pages that were deleted for reasons including non-free content, and their sandbox was deleted with part of the reason being non-free content.)
      • From June to August, Saair was very active at the Teahouse with over three dozen discussions that they started, with one of the longer discussions being about blocking certain pages from appearing for them: Has Wikipedia any method by which some of its articles can be blocked from navigation for a specific user or a user can block certain pages from his vision? This topic also continued into other discussions that they started. Eventually, Saair was asked by an administrator at the time to edit the article space more. (Technically not the first time, but this was the final warning about it.)
      • While Saair became more productive in August and even created Wheel construction, he ended up in a dispute with the creation of Pornography in Saudi Arabia which got sent to Articles for deletion. While that was not an issue, Saair decided to end the discussion by taking the article and sending it to drafts. As can be guessed, that did not go over well, leading to a block by SpacemanSpiff. (Note: I am skimming a bit here as there were other things that happened that were not addressed above, such as a sockpuppet incident prior to this in the same month and this edit to someone else's unblock request.)
      • From there you have more sockpuppeting until November 2017, the request to be unblock in 2020 which led to the TBAN suggestion, and the successful unblock in 2021 after agreeing to a TBAN.
      • Since then, Saair tried to get the TBAN removed in April 2022, but was denied by a unanimous decision not to do so at the time: Consensus is clear; [Saair] needs more editing experience to demonstrate they understand Wikipedia policies; promotional editing is a major problem thus far. Ban will remain in place for the time being. (To skim a bit here, the main issue was apparently an attempt to make an article about himself that he tried to defend at AfD.) --Super Goku V (talk) 11:36, 27 January 2026 (UTC)

There's no doubt I had been much disruptive around 2017, for that I remained blocked until 2021, and now it's 9 years since 2017. None of those disruptions have ever been repeated by me since then. But I've never disrupted geography related articles, even around 2017; and I hope those who dig in this regard would agree with me in this regard, and that the TBAN issue was never discussed at any extensive form; and it is not required as it is not backed by any type of disruptive behavior in the specific field. So, I'd once again repeat my request to lift the TBAN. Saair (talk) 13:27, 27 January 2026 (UTC)

I get that. I was just trying to figure out why the TBAN idea ended up being brought forward and I gave the best answer I could determine.

That aside, I also was going to add a part about this request to overturn the TBAN. From my perspective, the TBAN doesn't make sense. At the same time, you haven't edited that much since being unblocked. So, I would suggest some sort of probation based on edits to unique articles if that is something that is possible. (A time based one wouldn't make much sense for a probation period given 15 edits to articles in 3.75 years or so.) --Super Goku V (talk) 13:44, 27 January 2026 (UTC)

Hey, Saair. I just want to quickly address this: I don't know why I was banned from editing India and Afghanistan geography related articles. For background, we used to have something called Discretionary Sanctions. DS was basically special sanctions imposed on editors by the Arbitration Committee in specific topics due to significant disruption. In 2022, DS was replaced with the similar Contentious Topics, with your TBAN being rolled over to the CT system. (There are some differences, but I would say that most are technical in nature.) With the change to CT, it was made clearer that there are specific topics that over the years have had increased disruption compared to the rest of Wikipedia.

Now to your TBAN. Back in 2012, ArbCom enacted DS (...) "for all pages related to India, Pakistan, and Afghanistan, broadly construed." While you were not editing articles tied to India and Afghanistan, you were editing Pakistan articles. So, when you accepted the TBAN from Pakistan articles related to geography, you were accepting a ban from all three. --Super Goku V (talk) 12:08, 27 January 2026 (UTC)

• Support fully lifting ban with no conditions - thanks to all who did more digging and provided more context above. What I see here is a user who got into "enthusiastic new user" trouble nine years ago, who took the only unblock offer they were given after three years of trying, by two admins who stated plainly at the time that they were busy and just trying to clear a backlog. I get the sentiment that they should build more experience, but that doesn't account for the fact that the topic ban made no sense in the first place, and isn't preventing any disruption. More importantly it isn't preventing the sort of disruption they were engaging in nearly a decade ago, yet they have not continued to cause problems in the meantime, and that to me is a strong indicator that the topic ban is not useful and should be lifted. Ivanvector (^Talk/_Edits) 14:03, 27 January 2026 (UTC)
• Support per Ivan. —Compassionate727 ^(T·C) 00:57, 28 January 2026 (UTC)

Support per Ivan. Dronebogus (talk) 15:09, 28 January 2026 (UTC)

• Support: I know I suggested a probation above, but I am more convinced by Ivan's points than my own suggestion. --Super Goku V (talk) 09:22, 29 January 2026 (UTC)
• Support Ivanvector sums it up best: More importantly it isn't preventing the sort of disruption they were engaging in nearly a decade ago, yet they have not continued to cause problems in the meantime, and that to me is a strong indicator that the topic ban is not useful and should be lifted. Regards, --Goldsztajn (talk) 20:00, 30 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Arbitration motion regarding NorthernWinds

The Arbitration Committee has resolved by motion that:

NorthernWinds (talk · contribs)'s topic ban from the Arab-Israeli conflict is rescinded.

For the Arbitration Committee, EggRoll97 ^(talk) 06:45, 27 January 2026 (UTC)

Discuss this at: Wikipedia talk:Arbitration Committee/Noticeboard § Arbitration motion regarding NorthernWinds

WP:Categories for discussion needs more closers

It now has a backlog of 250 discussions, dating back to November. I've been closing a lot of discussions myself, but am approaching burnout without making a significant dent in the backlog and would really appreciate others helping out more. Instructions for admins at Wikipedia:Categories for discussion/Administrator instructions, see also User:Qwerfjkl/How to close CfD discussions. * Pppery * _{it has begun...} 07:44, 27 January 2026 (UTC)

It seems that a decent portion of these are by a few editors that tie heavily to sports categories. Of the 43 discussions still open from 2025, 28 of them are sports sub-categories with 23 of those discussions by just one user. --Super Goku V (talk) 12:44, 27 January 2026 (UTC)

Yep. I was worried about this. I've asked @Johnpacklambert several times to bundle and consolidate the nominations he writes. He's not responded/acknowledged my requests on this: User talk:Johnpacklambert/Archives/2026/January#Please consolidate your merge nominations. Other users have tried to help: User talk:Johnpacklambert/Archives/2026/January#Friendly CFD Cleanup, like @RevelationDirect And other users (@Super Dromaeosaurus) have raised concerns about the consequences of his nominations User talk:Johnpacklambert#Categories with a less than satisfactory response. [23] ^SMason_Garrison 17:20, 27 January 2026 (UTC)

Johnpacklambert is under a suspended topic ban from CfD, which means any admin could reimpose it as an AE action. After having read through those talk page comments I'm seriously tempted to do so (which is unusual for me since I almost never involve myself as an admin in conduct matters). * Pppery * _{it has begun...} 18:16, 27 January 2026 (UTC)

Support reinstatement this is just a different flavor of the same disruption that led to the bans. Clearly nothing has changed. Star Mississippi 19:37, 27 January 2026 (UTC)

Back in December, I encouraged Johnpacklambert to hold off on further college sports nominations because they didn't seem headed anywhere but "no consensus". I haven't checked everywhere, but I'm pretty sure he took my advice, which eliminated the largest group of nominations but the backlog remains. RevelationDirect (talk) 22:22, 27 January 2026 (UTC)

If memory serves, I believe the last sports related discussion was December 28th. I believe most of the January batch of discussions are about people born or living in countries. --Super Goku V (talk) 23:16, 27 January 2026 (UTC)

Pppery, thanks again for all your hard work. I'll see if I can clear out some non-controversial ones with non-admin closures. RevelationDirect (talk) 22:25, 27 January 2026 (UTC)

• When I was asked to wait on nominating more categories related to sports until the existing ones were closed I stopped nominating more categories related for sports. I have tried to seek consensus. I have tried to respond to advice. I have not nominated one sports Category since he made that request. I am teying to be collaborative with others. I am trying to discuss the issues involved in building categories. I am trying to work through the issues of making it so categories. The issue with sports nominations was brought up. I responded by not making any others. I have tried to be cooperative. I have not made even one more sports nomination since then.John Pack Lambert (talk) 22:41, 27 January 2026 (UTC)
• I did listen to requests about making larger nominations. I made a nomination with 108 categories related to Category:Generals by nationality. I also made a nomination related to all the categories under Category:Slaves by nationality. I am sorry if my actions have caused disruption. I have been trying to follow the general guidelines of categorization. I have been trying to seek consensus. I will try to do this more in the future.John Pack Lambert (talk) 22:56, 27 January 2026 (UTC)

  I know that you're trying. I'm glad to hear that your changing your behavior, even if you're not articulating that on your talk page. But, like, it's a lot of comments and nominations, and just sheer volume of text. Remember that other people have to read the full conversation in order to process the nomination. Perhaps the 2nd choice remedy of limiting to 3 per day could be implemented instead? Wikipedia_talk:Arbitration/Requests/Case/Conduct_in_deletion-related_editing#Motion:_Johnpacklambert_topic_ban_amended_(limited_categories_exception) ^SMason_Garrison 22:34, 28 January 2026 (UTC)

  I am sorry. I am trying to make sure I clearly explain my position. I will seek to be more concise in presenting my position. I will also seek to where possible bundle nominations in larger groups. I appreciate all the work you and others do to improve Wikipedia and hope that we can find ways to do so more in the future. I will try to be more concise in my statement in the future.John Pack Lambert (talk) 11:46, 29 January 2026 (UTC)

  I appriciate your efforts. I really do. I wonder if there's a way to help you calibrate the volume. What if you limited yourself to 3 nominations per day and had a max word count? I think that would help a lot by encouraging you to think thru the arguments and prioritize the important ones. ^SMason_Garrison 13:59, 29 January 2026 (UTC)

  I will consider this.John Pack Lambert (talk) 14:07, 29 January 2026 (UTC)

  I'm really struggling with what to do about User talk:Johnpacklambert#Category:Generals by nationality because the challenge is also here the sheer volume of changes you made related to a bundled category nomination that wasn't going your way. @FromCzech can probably summarize the nuances better.^SMason_Garrison 21:28, 30 January 2026 (UTC)

  I have apologized for rhat. I am very sorry. I have tried to correct my mistakes. I was trying to adapt to the advice of others and apply it. I was trying to heed the advice ilof some editors that both the country served for and nationality are defining. I am sorry that I jumped the gunin trying to implement it.John Pack Lambert (talk) 03:17, 31 January 2026 (UTC)

  I know that you've apologized. But like, this is illustrative of the challenge that I don't think you really appreciate. Now there are 37 subcats of Category:Generals by country to clean up with nearly 1000 people (according to petscan https://petscan.wmcloud.org/?psid=42588159), and that's already after you emptied 13 of those categories. Practically, the only thing I did today was try to add all the missing parent categories of FOOian generals in the vain hope of restoring the category tree to the status quo. ^SMason_Garrison 04:55, 31 January 2026 (UTC)

  I am trying to correct this error. I am very sorry about this.John Pack Lambert (talk) 05:25, 31 January 2026 (UTC)

  I have tried to restore all categories and articles back. Most of what is left is in the French and Spanish categories that may take a little longer.John Pack Lambert (talk) 05:27, 31 January 2026 (UTC)

  I know that you've apologized. It's not just this specific case, it's that you make an overwhelming number of changes, and a sizable number of them are inconsistent with the consensus about nationality versus country. I feel like I ask you to do the same handful of maintenience things over and over again that are consistent with how most people think about nationality.. it's one of the reasons I don't feel like I have the bandwidth to help with closing CFDs. ^SMason_Garrison 05:40, 31 January 2026 (UTC)

• Question Is there any potential appetite for unbundling access to WP:CFD/W as a stand-alone permission that non-admins could request? - RevelationDirect (talk) 01:06, 28 January 2026 (UTC)

  The previous few times that has been proposed it hasn't gone anywhere (For example Wikipedia_talk:Categories_for_discussion/Working/Archive_2#Protection_of_WP:CFD/W,_take_3). Despite having written an exasperated edit summary in the protection log for the page I'm not actually completely against it.

  But I don't think it would solve the issue here - the backlog at WT:CFDW has not been that huge lately, although again that may only be because I've been handling it. * Pppery * _{it has begun...} 17:23, 28 January 2026 (UTC)

  I do think that giving more people permission would be helpful. We can have a vetting process like NPP does. (I've still not got the hang of closings myself, so finding ways to make it easier for folks to learn the process might be helpful regardless).^SMason_Garrison 13:55, 29 January 2026 (UTC)

  As I have said in the past, I think giving access to Template editors could be a way forward for this. We trust them with pretty disruptive templates. Plus if one starts posting "delete" results, I think that'd be noticed rather quickly. Please ping me if someone starts a discussion on this. - jc37 03:31, 31 January 2026 (UTC)

  Please no. The template editor right is already heavily overloaded. Competence in closing deletion discussions has very little to do with competence in editing high-risk templates (or DYK queues, which I also think was done wrong). If we want to unbundle access to CFDW then do it right and create a separate "CFDW editor" user group. * Pppery * _{it has begun...} 16:49, 31 January 2026 (UTC)

• It is a difficult trade-off. When bundled, participants in the discussion may complain that it should be unbundled, and vice versa. Marcocapelle (talk) 07:21, 28 January 2026 (UTC)

  Marcocapelle is absolutely correct here. Plus, there is the problem, that "bundled" group noms can end up a train wreck, due to many moving parts. And even the most well-researched nominator can be surprised by the direction a discussion may go. - jc37 03:31, 31 January 2026 (UTC)

• Just to broaden this out to the general issue of CfD closures, is there any sort of tool that reduces straightforward closes to a few clicks, as there is for closing requested moves? BD2412 T 22:40, 30 January 2026 (UTC)

  WP:XFDC. voorts (talk/contributions) 22:44, 30 January 2026 (UTC)

  The problem is that XFDC doesn't know how to add content to WP:CFDW. User:Qwerfjkl/scripts/CFDlister is the only tool that comes close, but that right now only helps non-admins (or admins who want to cosplay as non-admins). No tool knows how to add content to CFDW itself.

  Aside from all that, CFD tends to use a much less rigid format than other deletion discussion venues so there are many nominations that no tool can parse. * Pppery * _{it has begun...} 01:17, 31 January 2026 (UTC)

  BD2412 was asking for tools in a broader context than CfD. And XFDC can at least make closing the CfD discussion easier. voorts (talk/contributions) 01:30, 31 January 2026 (UTC)

Unequal DRV process - despite repeated requests

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

I am choosing to finally escalate this matter after waiting patiently for nearly 7 days since the DRV process started at following page: https://en.wikipedia.org/wiki/Wikipedia:Deletion_review/Log/2026_January_19 - wherein I have repeatedly requested the DRV coordinator as well as the AfD closer more than once - in the DRV forum as well on their talkpage that the DRV in question is currently being conducted - without any "choice" available to any reviewer to read the article as it stood on 16-Jan-2026 when the AfD was closed - and/or compare the subsequent "significant new content" listed as a key point in DRV request. But my repeated request has since been politely stonewalled with the message that - "Temporary undeletion is sometimes requested by participants in a DRV who haven't seen the deleted page, and wish to consider its contents before forming an opinion. You, Jn.mdel, have seen the page, and none of the other participants in that DRV seems interested in its contents.".

I am really amused and unhappy at the same time - because with due respect to everyone involved, if some reviewer is "not interested" in reading the content when the DRV request is actually based on content vis-a-vis last version of article - then why should such a reviewer be counted or their feedback be even considered. In my personal opinion, the reviewers do not realise the value of their own votes or its impact in using it, without necessary application of mind on the content before voting - especially when this DRV's basis is about content - but the more unfortunate part is that we are currently appearing to be ok with this "no choice" as a process.

Ok fine, we cannot change or mould the perspective or choose who all is reviewing the DRV in the manner they choose to do - but atleast the basic requirements of a logical and fair process should still be fulfilled so that the DRV process sounds equal for all sides - meaning for a content based DRV request, atleast the deleted version of the article as it stood on 16-Jan-2026 should de-facto be "available" for review if any reviewer wants to use that "choice" to peruse.

If they still do not read or use that "choice" then maybe it is bad luck for the article and for my efforts too - but if the article version itself is not even available - and the reviewers are too lethargic to ask for it or consider it too tedious to read and are still submitting a general opinion without meaningful reasoning, then later the DRV closer should be able to consider what weight to be given to such feedback.

My only humble submission here is that the DRV process should atleast appear fair and be as fool-proof as possible - because during the DRV, every participant may act as per their own wishes and perspectives - but then again at the closing stages of the DRV, the closer again has the responsibility and duty to uphold the due sanctity of the reviews - meaning the process does always remain in the background - but that it is still there, that much atleast should always be assured.

Otherwise why should anyone spend time to try and contribute to this article since Sep. 2024 - and even after AfD nomination on 2-Jan-2026, still continue to dig for more and more information and or relationships/relevance with other topics - so much so that I could even add a completely new sub-section of "Geometric Interpretations" on 15-Jan-2026 - but then nobody got to review it - and even now when the DRV process has been requested, still no one has the "choice" or option to review it. In which case what is the point of the proposed for deletion masthead stating - "... Users may edit the article during the discussion, including to improve the article to address concerns raised in the discussion...".

I mean why should I even have to repeatedly request for such a read-only access during DRV process - I do not understand how as a process - one can expect the reviewers to be able to give a fair opinion on something which they have not even read - specifically when the DRV is based on "significant new content" - and thus, read-only access to deleted article for such a DRV review should have been a mandatory auto-available I think.

Thus, only if the article version of 16-Jan-2026 is made available - can I still go back and think of requesting the already existing reviewers as well as any new forthcoming reviewers to atleast read the article once and then maybe revisit their already formed opinions if they feel like - or maybe move towards a consensus to rename the article or any other path to rehabilitate the valid efforts already put in - rather than just delete some sincere efforts. Jn.mdel (talk) 16:22, 27 January 2026 (UTC)

I read it, and I agree with the people who found it to be original research. I think it's time for you to accept that the process has decided against you. Believe me, it happens to all of us. SarekOfVulcan (talk) 16:39, 27 January 2026 (UTC)

@Jn.mdel: you didn't like the result of the AfD, which was unanimous apart from your !vote, and could only be closed the way it was. So you took the matter to DRV, wholly without merit. Now you don't like the way the DRV is going, so you bring this here. (What's next, ArbCom?) This would be an excellent time to drop the stick, IMO. -- DoubleGrazing (talk) 16:40, 27 January 2026 (UTC)

You are welcome to your opinion @Jn.mdel but it is clear that the community does not agree and we operate under consensus. Please see WP:1AM. Note that there is no DRV coordinator. We are all equal participants including you, the AfD closer and anyone else who participates. Star Mississippi 17:29, 27 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

By the time I have come back, both this and the DRV is closed, so I only wish to thank all of you for your mature advise that I should defer to the community consensus - though I still retain my opinion that the DRV process as was conducted was not equal as it made reviewers give opinions on something which they could not even read !!! - and that too when the DRV request was content-based. Also wish to clarify humbly that by responding during AfD and then later opting for DRV also, I do not think it was anything wrong in trying to stand for the sincere efforts put in - but I was not here on the Administrator's Board for content re-discussions - but instead to highlight the process deficiency in DRV - especially for content-based DRV requests - so that atleast someone else's future DRV procedures may be better streamlined - just as a suggestion. Jn.mdel (talk) 05:39, 28 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Assistance Needed

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

BlueboyLINY made cleanup changes to several radio station pages. I attempted to discuss this with him only for him to revert it without discussion there and say I needed to make the discussion on the station's talk page. He is now mass reverting my changes as "Non-constructive edit[s]" or "ain't broke". This change just dumbfounds me. The article, images included went through GA scrutiny, and haven't moved in 10 years, but he calls them "sandwiched". These include other changes, like updated sources, updated history, source addition, etc. Again, with no discussion. His reverts reintroduce incorrect information, outdated information, unsourced information into these articles with no reason.

Now, I can't get into an edit war and since I can't bring my concerns to my unblocking admin, Deepfriedokra, as he has laid down his mop and is semi-retired, so I bring this to the community at large for guidance. - Neutralhomer • Talk • 22:10, 27 January 2026 (UTC)

I have notified him of this thread. - Neutralhomer • Talk • 22:14, 27 January 2026 (UTC)

What is the content consensus? GoodDay (talk) 22:15, 27 January 2026 (UTC)

@GoodDay: Well, in the case of this edit, I don't think consensus is needed. It's moving some sources around, correcting incorrect information (admittedly, one of those is mine), and removing affiliations that are no longer valid. None of it is particularly controversial. This actually added back incorrect information to the page. So, I guess the consensus is correct information is good? - Neutralhomer • Talk • 22:19, 27 January 2026 (UTC) 22:19, 27 January 2026 (UTC)

I should note that he has continued to make changes to articles I have edited today, either reverting them or making changes that make no sense. All of this instead of discussing them. - Neutralhomer • Talk • 22:54, 27 January 2026 (UTC)

@BlueboyLINY: please respond. GoodDay (talk) 22:59, 27 January 2026 (UTC)

The majority of my edits are done using WP:SCRIPTS in an effort to comply with WP:ACCDD. That being said, I'm not sure why we're even having this discussion as those edits are considered minor, and all @Neutralhomer need do is click on them for an explanation and not be "dumbfounded". Yes, I do occasionally edit articles to prune out WP:SEAOFBLUE, MOS:OVERLINKING and MOS:NOTBROKEN, or to remove redundant wording, or fix grammar. @Neutralhomer states: "His reverts reintroduce incorrect information, outdated information, unsourced information into these articles with no reason." I would appreciate knowing exactly what needs to be corrected, that's why I posted on @Neutralhomer's talk page to discuss on the articles talk pages. Discussion there would be open to all editors to weigh in. ~ BlueboyLINY (talk) 00:00, 28 January 2026 (UTC)

@BlueboyLINY: Nice of you to join us. I'd like an explanation here as to why the wholesale reverts? Explaining what was wrong with these edits? You made a grammatical error on the WFSP-FM page and never corrected it. I did and you reverted it. You reverted the addition and correction of information like WFSP Radio, LLC instead of WKMM Radio, LLC owning WKMM. You reverted clean up of the sources within the page. Explain these changes. These are changes you made and we are discussing your behavior. - Neutralhomer • Talk • 00:07, 28 January 2026 (UTC)

@BlueboyLINY: Thank you for your re-revert to the WKMM page. Now, on the WFSP-FM page, you reverted a correction to the owner of WFSP-FM, sources for that correction, adding WKMM to the sister stations, sources for that, and again readded the Use mdy dates template to the article (there aren't any DMY dates within the page). - Neutralhomer • Talk • 00:28, 28 January 2026 (UTC)

@BlueboyLINY: Thank you for your re-revert to the WFSP-FM page. Now, on the WXRK-LP page, the callsign meaning was added as "A hat tip to the former WXRK-FM in New York" with source added, which is already within the page. The bottom part just moves somethings around, moving the September 2024 items behind the January 2026 item, which has an updated date from November 2025. This is something that is currently ongoing. I hadn't had the chance to check for more updated information (EDIT: Nothing new as of close of business). The formatting of the infobox is the same as the WKMM and WFSP-FM pages.

On the WKEY (AM) page, I don't understand this. I agree it's sandwiched, but there are better ways of doing this and smashing them under the infobox isn't one of them. As stated, this passed GAN scrutiny and the sandwiching was allowed to pass. - Neutralhomer • Talk • 00:50, 28 January 2026 (UTC)

Ten years ago there was no group of editors with disabilities working on articles. Now we have WP:WPACCESS (a group I'm part of} which is helping bring articles up to WP:ACCDD standards. I understand that it's not the best look to be under the infobox, however it's better than be being a MOS:SANDWICH. Feel free to put the images into a gallery, if it's really bothering you. BlueboyLINY (talk) 01:15, 28 January 2026 (UTC)

@BlueboyLINY: Not sure what disabilities have to do with anything, also Hi, Autistic person here, but GA and FA articles are held to a higher standard. Having the images like that is not perferred and I've seen GAs and FAs taken to review for less. Now, a gallery, we can do, still gonna be in the same place and still gonna sandwich. - Neutralhomer • Talk • 01:21, 28 January 2026 (UTC)

@BlueboyLINY: Gallery made, which is what I planned on doing before I had to do this. Still would like to see an edit on the WXRK-LP page.

Now, Star Mississippi| is right, this was a content dispute, but it really didn't have to be. I hope we can work together in the future more constructively. Neutralhomer • Talk • 01:36, 28 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Disruptive editing at the Chiara Ferragni article

• Chiara Ferragni (edit | talk | history | links | watch | logs)

I would like to request administrator attention to this article due to continuing disruptive editing that may be considered as an attack page against a living person.

An anonymous editor is repeatedly reverting well-sourced and policy-OK content from independent, high-quality sources, including The Guardian and other major international media which cover the topic signficantly. These sources clearly meet WP:RS standards, however the anonymous user removes it as “puffery” or “advertising” without justification.

That is resulting in edit warring and clear WP:POV pushing

The mentioned article can be found here: https://en.wikipedia.org/wiki/Chiara_Ferragni Lorraine Crane (talk) 23:07, 27 January 2026 (UTC)

I don't think there's a need for administrator involvement at the moment, there's only a couple of edits back and forth each way so far; WP:BLPN or the article's talk page can be used to hash out the underlying content dispute. theleekycauldron (talk • she/her) 00:23, 28 January 2026 (UTC)

Appealing topic ban from XfD

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

I would like to appeal my topic ban from deletion-related activities. I have obeyed my topic ban since its implementation and while I have no enthusiasm for returning to my primary historical areas of activity I have recently encountered a lot of spam articles in Category:Technical fabrics that could uncontroversially be PRODed but am unable to. I also never experienced any complaints or issues with my conduct outside of MfD or AfD and would like to be able to edit in areas such as redirects/files/categories for discussion constructively again. Finally I don’t like having my pages/redirects/etc. come up for discussion without being able to participate in the discussion. If this request is still objectionable I’d at least like my topic ban narrowed to AfD and MfD, the only areas I actually received conduct complaints in, instead of having a blanket ban that also affects many areas I only ever edited constructively in. Dronebogus (talk) 15:07, 28 January 2026 (UTC)

• It's always helpful if you link to the ban discussion in your appeal; it's here, anyway. I haven't looked to see if there have been any previous appeals. Ivanvector (^Talk/_Edits) 16:32, 28 January 2026 (UTC)

  Oh there's a second one, here, which resulted in the current blanket ban from XfD. That discussion was closed by Lourdes, who was a sock of a banned user. On that basis alone I support reverting the ban conditions to those set in the first discussion: a ban from miscellany for deletion and from closing any XfD discussion.

  On the matter of the request, I think it would be reasonable to add a condition that you may participate in an MfD when you are a significant contributor to the target page, since from your talk page that seems to have been an area of difficulty. Ivanvector (^Talk/_Edits) 16:40, 28 January 2026 (UTC)

  @Ivanvector: Aside from the fact that Lourdes's close implementing the XfD-wide ban looks to be, on first brush, fairly uncontroversially correct, I would say that Dronebogus accepted the wider ban voluntarily by voting 'support' on it. theleekycauldron (talk • she/her) 16:54, 28 January 2026 (UTC)

  If only we had some kind of evidence that they do not actually want to be under this ban, like a request to lift it or something? An appeal of some sort? Ivanvector (^Talk/_Edits) 17:38, 28 January 2026 (UTC)

  We have evidence that they don't want to be under this ban now. We have an equal amount of evidence that they didn't object to being under it at the time. Given that an editor supporting the lifting of a topic ban imposed doesn't invalidate their support of it being imposed six months earlier, I see no reason why Dronebogus changing their mind on this after almost 2½ years is at all relevant? Thryduulf (talk) 19:04, 28 January 2026 (UTC)

  Well, sure, but the appeal should be assessed on its merits; a voluntarily accepted ban doesn't become procedurally invalid just because it was appealed. (and even if it did, you'd still have to get the close overturned for the ban to be procedurally invalid.) theleekycauldron (talk • she/her) 19:08, 28 January 2026 (UTC)

• Oppose. You claim "I also never experienced any complaints or issues with my conduct outside of MfD or AfD " which doesn't really match with this ANI discussion from October 2025. Fram (talk) 16:50, 28 January 2026 (UTC)

  I meant within XfD. I have never had complaints about my conduct in the sub-fields under that branch besides AfD and MfD. Dronebogus (talk) 17:52, 28 January 2026 (UTC)

• Support - If I understand correctly, one of the t-bans was closed by a sock puppet. -- GoodDay (talk) 16:54, 28 January 2026 (UTC)
• My opposition here should come as no surprise, given my stated view that Dronebogus should be community-banned. Nonetheless, as long as we're caught in the endless cycle of Dronebogus nearly getting themself indeffed, accepting some lesser sanction, and then continuing to disrupt things—which saw its most recent iteration a mere three months ago, as seen in that link—it is hard to think of a worse idea than unleashing Dronebogus back into the realm of XfD. We do not need to speculate what Dronebogus would be like at XfD again. They've given us ample evidence in their participation on Meta throughout their XfD ban. In the October AN/I thread, I pointed to m:User talk:Dronebogus § Your recent proposals, concerning Dronebogus' quixotic attempts to shutter various Wikipedias, or prevent the creation of new Wikipedias, based on nothing more than their personal feelings. (Full disclosure, in one instance this concerned a request I shepherded; there was certainly some reasonable opposition there, but Dronebogus' !vote is notable for directly contradicting the language proposal policy, the exact sort of behavior one would expect someone to avoid if trying to show that they won't disrupt XfD on enwiki.) When I pointed this out in October, Dronebogus replied that I haven’t proposed any more wiki closures on Meta; yet, since writing that, they have participated in a proposal to close the Navajo Wikipedia based on entirely hypothetical concerns raised by someone who doesn't speak the language, claiming without evidence that its article titles are made up (based on an apparent misunderstanding of how descriptive titles work in a language that they, too, do not speak), and resorting to conspiracy-theorizing and personal attacks against the editor who has diligently written 20,000+ articles in that minority language (If new learners are using words made up by one random guy on the Internet no other fluent speaker uses (and has anyone actually established this user even IS fluent?), it’s outright anti-educational).
  In short, Dronebogus is really bad at judging whether things should be kept or deleted on collaborative projects. We can also see this in their Commons block log in the same time period, although I'm less familiar with all the backstory there. If Dronebogus wants to reënter one of the most sensitive areas of community governance—determining which of their peers' content should be retained on this project—they should start by not throwing gasoline on fires in related areas on sister projects. Curiously, Dronebogus has argued, both here and on Meta, that their deletion-related disruption on one project should be inadmissible as evidence on the other. This is a misconception usually only found among halfway-trolling newbies, and I find it curious that someone of Dronebogus' tenure could think such a thing. But we are not obliged as a community to pretend that Dronebogus will be a productive presence in deletion discussions here when they have already presented ample evidence on sister wikis that they remain incapable of reasoned judgment in such discussions. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 18:45, 28 January 2026 (UTC)
• Oppose per Tamzin's evidence I am entirely unconvinced that Dronebogus will be a net positive at XfD. Separately, That Lourdes was a sockpuppet does not change the fact that they correctly read the consensus of the discussion they were uninvolved with, and any uninvolved admin in good standing would close it the same way. That Dronebogus doesn't support the restriction now does not invalidate that they supported it at the time, so I'm seeing no procedural grounds on which to overturn. Thryduulf (talk) 19:01, 28 January 2026 (UTC)

  I am going to be blunt: I accepted it under duress with the assumption it was either this or blocking. I did not enthusiastically propose it and say “no matter how much I beg, or plead, don’t let me back into XfD” (to paraphrase a great artist). There was a lot of reasonable opposition at the time, i.e. it was not a landslide (avalanche?) consensus. I do not agree that anyone would have made the same call, therefore the point about Lourdes’s administrative incompetence is entirely valid. Finally, Tamzin is cherry-picking incidents and finding fault on technicalities (ex. I have not proposed any more Wikipedias for closure; participating in a discussion where several people already voted to close the wiki isn’t that). Dronebogus (talk) 19:38, 28 January 2026 (UTC)

  Lourdes was banned for being a sockpuppet, not for any sort of incompetence. I realise I could have been clearer though, I have two independent opinions:

  • I think the topic ban as it stands is valid on both procedural and consensus of the 2023 discussion grounds.
  • I oppose lifting the topic ban.

  Thryduulf (talk) 20:06, 28 January 2026 (UTC)

  People, including admins, were complaining about Lourdes’s conduct at arbcom before the sockpuppet thing came out during that same session. If Lourdes hadn’t been a sock they still could’ve been sanctioned for administrative misconduct. Dronebogus (talk) 20:16, 28 January 2026 (UTC)

• Comment. The closure that resulted in the 2023 ban from MFD mentioned 'incivility, ragpicking, and a "general combative attitude"'. More generally, Dronebogus has, in the past, demonstrated a talent for loudly proclaiming how their 'opposition' are fools, biased, wrong, etc. And I say this as someone who often agrees with Dronebogus's points, so this isn't tone policing as an excuse for opposition, this is "this is embarrassing to our side and just going to cause the other side to dig in deeper". I don't see anything in this request mentioning the historical civility issues, which seems a major omission. Dronebogus, have you turned over a new leaf on this? If so, great, let's hear about it, and ideally commit to hardcore assuming good faith about others, named or unnamed. If not, then it's for the best not to go back in to one of the most fraught and bad-feels parts of Wikipedia. SnowFire (talk) 19:10, 28 January 2026 (UTC)

  I would like to turn over a new leaf, but I can’t do that when I’m topic banned. I was admittedly too much of a hothead at AfD/MfD and didn’t put in enough research when nominating articles. I probably will only nominate articles rarely if I get unbanned as it’s no longer a primary interest. My main focus is still on getting un-tbanned from XfD outside of AfD/MfD. I can take or leave the topic ban from those two, but I never did anything wrong in the other XfD departments to warrant such an all-encompassing topic ban. Dronebogus (talk) 19:18, 28 January 2026 (UTC)

  This was the wrong answer, I'm afraid. Dronebogus, you absolutely can turn over a new leaf, right now, before being unbanned, and that's what people want to hear. You need to say "Yes, I've changed my approach, and here are some examples in the recent past of me working productively, collaborating with others, de-escalating tension, and handling disagreements cordially." Can either be in normal editing on WP that isn't XFD related, or XFDs on other WMF projects. Give it a shot! If you don't have examples just yet, that's fine, spend 6-12 months gathering them by being awesome and friendly for that long. That's what you need to do first if you want to have a shot at getting the restriction rolled back. SnowFire (talk) 20:01, 28 January 2026 (UTC)

  I kind of meant in the most literal sense of I can’t turn over a new leaf at XfD when I’m not there. I understand the confusion but the meaning you construed is not my intended message at all. As for working productively, collaborating with others, de-escalating tension, and handling disagreements cordially… I’d say a majority of my edits across all projects are constructive, but are not noticed because they aren’t big and flashy. I don’t collaborate much because I’m naturally a loner and don’t seek out collaborative projects. I tried to de-escalate tension and handle disagreement cordially here but it was sadly unsuccessful because the other user didn’t want to cooperate, but at least it ended without “bloodshed” (i.e. long-term blocking). I also have worked files for deletion on Commons for the entire time I was there and rarely received any complaints (albeit acknowledging vastly different standards between Commons deletion and ), which disproves Tamzin’s assertion that I am too incompetent to be trusted in any deletion-related area of any project. Dronebogus (talk) 20:13, 28 January 2026 (UTC)

  I also politely collaborated with user:FantasticWikiUser here and here, which started with a misunderstanding and ended with the nomination of several low-quality, non-notable, likely promotional articles for deletion. (1 2 3 4 5) Dronebogus (talk) 20:29, 28 January 2026 (UTC)

• Oppose per Tamzin and Snowfire and, ironically, Dronebus's reply to Snowfire. -- Deepfriedokra (talk) 19:37, 28 January 2026 (UTC)

  What is wrong with my response? Dronebogus (talk) 19:39, 28 January 2026 (UTC)

  It speaks volumes. Best, -- Deepfriedokra (talk) 19:49, 28 January 2026 (UTC)

  That was really not helpful. I’d appreciate it if you actually explained what you think is wrong with my attitude that prevents me from coming back to XfD rather than acting like it’s stupidly obvious. Dronebogus (talk) 19:53, 28 January 2026 (UTC)

• Support — I'm confident their editing in this area will be monitored by plenty of editors.— Isaidnoway (talk) 21:34, 28 January 2026 (UTC)
• Oppose per Tamzin. I'll also agree with Tamzin that a site ban would be warranted here. Sugar Tax (talk) 23:35, 28 January 2026 (UTC)

• Dronebogus, I've spent a fair amount of time looking over the links and your edit history, as well as your comments here.

  You seem to keep implying that others are misunderstanding what you mean. So I'd like to provide you the opportunity to express yourself better. We don't need to hear anything about other editors, including whoever closed the past discussions. This is solely about your behaviour. My direct questions are:

  a.) Why do you feel that you need the current topic ban lifted. And when responding, please keep in mind that we volunteers here all act in service to our readers, either directly or indirectly.

  b.) What examples of exemplary behaviours can you indicate you have done since being banned - in particular, project-space and talk page collaborative discussions.

  c.) When looking at not-so-exemplary behaviour in collaborative discussions (including examples others have brought up), can you explain why the behaviour was inappropriate, and how it might be better handled in the future?

  Thank you for taking the time to consider these questions.

  I think these may be impotant to you in that the tone here seems to be starting to shift from merely retaining your current restrictions, to becoming a project-space discussions ban or even a full community ban. But I think you still have a chance to turn this around and possibly change minds. Which I doubt will happen, if you continue as you have been. - jc37 00:11, 29 January 2026 (UTC)

• Oppose. The request and replies to it give no indication that the issues that led to the initial tban won't recur; also, speaking frankly, the main part of the reasonings given for the appeal just aren't reasonably valid. I have recently encountered a lot of spam articles in Category:Technical fabrics that could uncontroversially be PRODed but am unable to Literally any other editor is capable of PRODding these articles if they need to be. There is no deadline, and Wikipedia doesn't need you to do it. I'm a little more sympathetic to I don’t like having my pages/redirects/etc. come up for discussion without being able to participate in the discussion, but on the gripping hand you do not own these articles just because you started them - they are not "your pages/redirects". - The Bushranger _{One ping only} 00:26, 29 January 2026 (UTC)

Absolutely not

Deflecting including People, including admins, were complaining about Lourdes’s conduct at arbcom before the sockpuppet thing came out during that same session. If Lourdes hadn’t been a sock they still could’ve been sanctioned for administrative misconduct. is part of the problem with Dronebogus' interactions in a collaborative space. Claiming duress now and manifestly showing they have not turned a new leaf. Nothing in this discussion or their conduct makes me think we won't be right back here because they do not show an understanding that their behavior is an issue. While I'm not to the point of supporting a c-ban, we're closer to one than we are lifting restrictions. Star Mississippi 00:56, 29 January 2026 (UTC)

• Consider this request withdrawn. I am clearly never going to get unbanned from this area. I didn’t expect this to succeed anyway. Dronebogus (talk) 08:50, 29 January 2026 (UTC)

  If you didn't expect it to succeed why did you make the request? TarnishedPath^talk 09:35, 29 January 2026 (UTC)

  One can hope for the best while expecting the worst Dronebogus (talk) 09:46, 29 January 2026 (UTC)

  Opportunities were provided (not just by me) for you to better explain yourself. What you chose do do with those opportunities was completely up to you. _ jc37 12:52, 29 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Request removal of TPA at Bigblueheron

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

I am requesting upgrade of Bigblueheron's indef to include removal of TPA at User talk:Bigblueheron based on this comment at 09:55, 29 January 2026, per WP:CIVIL while indeffed, per WP:NOTHERE (besides their contribs, see this block appeal), and WP:PA (while indeffed). I am not offended (despite being the target of this attack—not the only one on that page while indeffed) and would have posted here regardless who he had named in that post. The sooner they stop quacking, the better. Thanks, Mathglot (talk) 10:21, 29 January 2026 (UTC)

Pretty non-controversial case of removing their talk page access based on their personal attacks and NOTHERE editing.  Done —⁠k6ka 🍁 (Talk · Contributions) 11:38, 29 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Unblocking Request for Techoliver298

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

I would like to request administrator attention to unblock User: Techoliver298. I understand I violated BLP by mistake and will not repeat this offense. — Preceding unsigned comment added by Techoliver298 (talk • contribs) 00:05, 30 January 2026 (UTC)

• Your user talk page is the correct place to make the request. I see you've already done that: User talk:Techoliver298#January 2026 2. — rsjaffe 🗣️ 00:12, 30 January 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Changes to the CheckUser team, January 2026

Following a request to the Committee, the CheckUser permissions of Vanamonde93 (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) have been restored.

For the Arbitration Committee
Daniel (talk) 22:12, 30 January 2026 (UTC)

Discuss this at: Wikipedia talk:Arbitration Committee/Noticeboard § Changes to the CheckUser team, January 2026

Topic ban appeal

I would like to appeal the topic ban to which I am subject: [24]. I always try to stay out of trouble, and, as my block log shows, I have never received any sanction other than the topic ban. Furthermore, my account contribution list shows a long and continuing record of constructive contributions to Wikipedia. I feel that my topic ban can now be lifted.TH1980 (talk) 00:41, 1 February 2026 (UTC)

The scope of the topic ban is "any edit, or editing any page, related to the topic area of Japan.", and was imposed in July 2020 appealable after six months. It was formally imposed by Swarm. The entry at WP:EDRC#TH1980 notes it as the reinstatement of a ban originally placed in 2016:

• Original TBAN, July 2016
• Appeal declined, January 2017
• Appeal granted, May 2020
• Ban reinstated, July 2020

The closer of the May 2020 discussion and filer of the July 2020 discussion Girth Summit wrote at the latter First, the request for an unban was not, it appears, entirely accurate. TH1980 said in their appeal "I always try to stay out of trouble, and, as my block log shows, I have never received any sanction other than this topic ban, which was related to only a single article". That is not true: the findings of fact in this arbcom case included that TH1980 has in the past both edit warred and hounded another editor, and it ended up with an indefinite IBan with Hijiri88 - that is a sanction. The assertion that the concerns were about a single article is also not correct - the discussion that led to their topic ban actually concerned edit warring on two articles, Korean influence on Japanese culture and History of Japan. There is no claim in this request about a single article, but the rest looks rather similar.

I am explicitly not expressing an opinion either way on this appeal (although I may do in the future). It's worth noting that I was one of the arbitrators active on the Catflatp08 and Hijiri88 arbitration case referenced above, but I didn't even remember that TH1980 was involved with that case before reading Girth Summit's comment, let alone any details of how they were involved or what findings were made regarding them. Thryduulf (talk) 01:40, 1 February 2026 (UTC)

I'm not seeing any warnings & over 1500 edits since the reinstated TBAN which is good.

That said, immediately returning to problematic behaviour straight after having the ban lifted is a little concerning. That was a gap of four years (banned 2016 & lifted 2020), not so far off the current five year gap.

@TH1980 could you please explain a little more about why you want to return to editing this specific topic, and what happened in 2020? Blue Sonnet (talk) 03:35, 1 February 2026 (UTC)

It's a speciality field of mine, and I have much knowledge and expertise in the field. My clean block log shows that I could make constructive contributions in this area without stepping on any toes, because I have learned how to cooperate with other users.TH1980 (talk) 02:55, 3 February 2026 (UTC)

Epstein files and biographies of living persons

As you may have heard, a massive dump of files have been released from the Jeffrey Epstein investigation. I've already seen some BLP violations related to persons who are very peripheral to the issue, with only a few mentions in the files, in which information from the files was used to imply something that may or may not be true. I've had to remind people of WP:BLPPRIMARY and WP:UNDUE, and have had to temporarily protect one bio. Hopefully we'll get better secondary sources in a few days or weeks as people review all the data, but in the meantime, this is a heads-up for y'all. We've been talking about an edit filter to at least temporarily make it easier to see Epstein-related BLP issues--we'll see if that materializes. — rsjaffe 🗣️ 02:11, 1 February 2026 (UTC)

Changes to the functionaries team, February 2026

Pursuant to the procedure on CheckUser & Oversight inactivity, the CheckUser & Oversight permissions of Ks0stm (talk · contribs) are removed.

The Arbitration Committee also acknowledges the resignation of Mkdw (talk · contribs) from the CheckUser team.

The Committee thanks both Ks0stm and Mkdw for their service in these roles.

For the Arbitration Committee
Daniel (talk) 03:43, 1 February 2026 (UTC)

Discuss this at: Wikipedia talk:Arbitration Committee/Noticeboard § Changes to the functionaries team, February 2026

Wikipedia:In the News/Candidates#(Needs decision) North American winter storm needs urgent closure

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

Definitely has enough comments to reach consensus one way or the other; is due for archival in under 12 hours. Multiple admins have been pinged and a request at CR was put up, and all went unanswered. ~2026-69318-9 (talk) 12:24, 1 February 2026 (UTC)

Was just posted. ~2026-69318-9 (talk) 12:44, 1 February 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Topic ban

The following discussion is closed. Please do not modify it. No further edits should be made to this discussion.

---

I have had a topic ban for more than a year - categorisation and living persons. I was unblocked in 2023. I have been careful to avoid these areas, and mostly I have been working on Commons. Can my topic ban now be lifted? Rathfelder (talk) 10:20, 2 February 2026 (UTC)

I can see from your talk page that you have a topic ban from categorisation. Do you also have a topic ban from biographies of living persons? If yes, please can you add a link to the wording of your topic ban. This is important because you made edits on four occasions to an AFD discussion for a biography of a living person. -- Toddy1 (talk) 11:18, 2 February 2026 (UTC)

The topic ban is here: Wikipedia:Editing restrictions#Voluntary: Unblock conditions. Rathfelder is subject to a one-account restriction, and subjected to indefinite topic bans from categorisation and from XfD discussions. — rsjaffe 🗣️ 11:28, 2 February 2026 (UTC)

Oppose – It looks like your TBAN was actually from categorization and XfD rather than living persons. However, whether we consider one or the other, the edits provided by Toddy1 show that you have been violating this topic ban up to today, in fact, even after making this appeal. The categorization part of the topic ban was also breached back in September. Chaotic Enby (talk · contribs) 12:08, 2 February 2026 (UTC)

Er, oppose, for the reasons stated, and also because I just blocked for the tban vio. -- asilvering (talk) 19:27, 2 February 2026 (UTC)

The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

User:CONTRIẞUTIONS

Please take a look here. This is CONTRIẞUTIONS; commenting from temporary account b/c I can't log in. ~2026-74610-5 (talk) 00:23, 3 February 2026 (UTC)

We don't delete user talk pages that have a relevant history except in extraordinary circumstances. Given the contents of the talk page and the age of the last edit, I did blank the page, as I don't think that, even without confirming your identity, that blanking the page is problematic. — rsjaffe 🗣️ 01:03, 3 February 2026 (UTC)

Thank you for that. I actually meant the user page but I wanted the talk page blanked too. By any chance, can the edit REVISIONS be crossed out too? And could the user page be deleted? I meant to tag the user page but I can't edit user pages. Thanks again, from CONTRIẞUTIONS! (moved to different device) ~2026-73837-7 (talk) 03:49, 3 February 2026 (UTC)

I've blanked the user page and removed all the revisions from its history. Like this, or if the page is deleted, it will still be visible to any administrator. You can email Oversight at oversight-en-wpwikipedia.org to have it fully suppressed. Rjjⁱⁱⁱ (talk) 04:43, 3 February 2026 (UTC)

Arbitration motion regarding TenPoundHammer

The Arbitration Committee has resolved by motion that:

Remedy 9.1 of Conduct in deletion-related editing is amended as follows: TenPoundHammer (talk · contribs) is limited to one XfD nomination or PROD per 24-hour period. Disruptive nominations may be reported directly to WP:ARCA.

For the Arbitration Committee, EggRoll97 ^(talk) 05:53, 3 February 2026 (UTC)

Discuss this at: Wikipedia talk:Arbitration Committee/Noticeboard § Arbitration motion regarding TenPoundHammer