User talk:Dw31415

Advice requested: clarifying implementation questions from a closed RfC

I am seeking advice about process regarding the closed Wikipedia:Requests for comment/Archive.is RFC 5 (the NOMOREARCHIVETODAY RfC).

There’s pushback on most of the proposals I’ve made to implement the consensus. the practical implications of that consensus. Specifically, there appears to be disagreement over whether the consensus to “remove” archive.today links contemplated deterministic implementation based on the presence of archive.today links themselves, or whether individual editorial review of each citation was expected before any removal could occur.

I am not seeking to reopen the RfC or relitigate its outcome. Rather, I am looking for guidance on the appropriate venue and process for obtaining clarification when reasonable editors disagree about what a closed RfC authorized in practice.

Would the preferred approach be to:

• ask for advice on the original RfC talk page;
• seek input from the closer and/or participants in the original discussion;
• continue the implementation discussion in the venue where the question arose; or
• initiate a new RfC specifically addressing the implementation question?

AI Agent search

While generally discouraged, editors may use LLMs to find potentially relevant sources, provided they write the article text themselves. See Wikipedia:LLMRESP#Finding sources. Dw31415 (talk) 02:25, 24 July 2026 (UTC)

Your bot

Any idea why you chose "Cutlass" out of all usernames for your bot? Cutlass^Ciera 02:55, 24 July 2026 (UTC)

Yikes. I was playing with names about cutting out blacklisted links. It didn’t occur to me to check to see if there was an active user named Cutlass. I’m sorry. Didn’t even occur to me. How do you feel about it? Dw31415 (talk) 03:03, 24 July 2026 (UTC)

It's a bit confusing. I received a talk page message from someone asking if I operated it a few days ago. No big deal, but a rename could be useful if you want. Cutlass^Ciera 03:32, 24 July 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:Magazine (firearms) on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 22:31, 26 July 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:Slavic migrations to the Balkans on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 13:30, 5 August 2026 (UTC)

Handling External Links

During the discussion of CutlassBot's approval request, two editors asked that the bot skip Archive.today links in the External links section. Their key objection was that most such links should be removed under WP:ELDEAD, while some official-site links might need to be retained or replaced under WP:ELOFFICIAL. Other editors argued that the broader consensus to remove Archive.today links should also apply to this section. Following that discussion, I changed the bot to defer processing links in that section. While completing 47,000 pages, the bot found that 27,000 other links appeared in External links sections.

Proposal

The bot should begin processing Archive.today links in External links sections using {{Deprecated archive}}. The key principle articulated by the WMF still holds For readers to remain relaxed and trusting while using Wikipedia, they should be able to reasonably expect that links on Wikipedia to potentially dangerous websites are rare, and that those that do exist are dealt with quickly once spotted. wmf citation

Example at Afroasiatic languages § External links

* Afro-Asiatic at the Linguist List MultiTree Project: Genealogical trees attributed to Delafosse 1914, Greenberg 1950–1955, Greenberg 1963, Fleming 1976, Hodge 1976, Orel & Stolbova 1995, Diakonoff 1996–1998, Ehret 1995–2000, Hayward 2000, Militarev 2005, Blench 2006, and Fleming 2006

Proposed change

* Afro-Asiatic^{[link removed]} at the Linguist List MultiTree Project: Genealogical trees attributed to Delafosse 1914, Greenberg 1950–1955, Greenberg 1963, Fleming 1976, Hodge 1976, Orel & Stolbova 1995, Diakonoff 1996–1998, Ehret 1995–2000, Hayward 2000, Militarev 2005, Blench 2006, and Fleming 2006

Do you support allowing CutlassBot to process links in External links sections? ping for editors involved in approval @Dw31415: Dw31415 (talk) 13:43, 15 August 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:Mirza Fakhrul Islam Alamgir on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 15:30, 21 August 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:Antisemitism on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 07:31, 26 August 2026 (UTC)

Template RFC

This comment seems to capture the majority: thoroughly checked use is acceptable in practice, even if the guideline appears to prohibit it. Leaving exceptions unwritten makes community standards harder to learn and undermines trust that guidelines mean what they say. If thoroughly checked use is acceptable, the guideline should say so. Otherwise, it has Luddite vibes rather than emphasizing our objections to fabricated content and mechanical prose. Dw31415 (talk) 10:05, 1 September 2026 (UTC)

Plant

The facility's five generating units have a combined capacity of 13.7 megawatts and average annual net generation of 74,531.5 megawatt-hours. Dw31415 (talk) 17:44, 8 September 2026 (UTC)

You deserve a barnstar!

	Link Solutions Barnstar
	Thank you for your patient perseverance with the WP:ATODAY problem. You have done more than anyone else to protect readers from this dangerous website. WhatamIdoing (talk) 01:44, 14 September 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:List of disability-related terms with negative connotations on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 05:30, 15 September 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:First Fitna on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 18:30, 18 September 2026 (UTC)

Citations RfC

Whether the Wikipedia:Writing articles with large language models guideline prohibits using AI agents to generate or modify wikitext markup and templates remains ambiguous. A recent RfC found clear consensus against adding an explicit exception for citation formatting, but did not resolve whether all such use is prohibited or should attract sanctions. Editors who use an LLM for these technical tasks should exercise particular care to verify every generated parameter and ensure that no LLM-generated prose is introduced into the article. Dw31415 (talk) 23:13, 18 September 2026 (UTC)

AWB CheckPage addition

Hi Dw31415, I'm leaving this message to inform you that I've added you to the AWB CheckPage per your request. Please (re)-familiarise yourself with our rules around (semi-)automated editing, and don't hesitate to reach out if you've got any questions. I'll also ping @Zackmann08 who might want to offer some support as well, based on the thread you linked. Keep up your good work and happy editing. Best, Staraction (talk · contribs) 07:24, 30 September 2026 (UTC)

Happy to answer any questions you may have! Zackmann (^{Talk to me}/_{What I been doing}) 23:09, 30 September 2026 (UTC)

Date and incident in the first sentence

My edit to add the incident to the first sentence was reverted. I think the first sentence should identify the date and the incident that makes the flight notable rather than starting with the scheduled route. MOS:FIRST says that for a topics notable for only one reason, that reason should usually appear in the first sentence. This is also consistent with many aviation incident articles.

I propose

On 30 September 2026, Flydubai Flight 1073 broadcasted the squawk code for hijacking, diverted from its flight path, and made an emergency landing at Tabuk Airport in Tabuk, Saudi Arabia. The flight was a Boeing 737 MAX 8 flying a regularly scheduled international passenger route from Dubai International Airport in the United Arab Emirates to Ben Gurion Airport near Tel Aviv, Israel.

Of the first ten dedicated crash articles listed at 2025 in aviation#Events, eight give the date in the first sentence:

• 2025 Light Air Services Beechcraft 1900 crash
• 2025 Potomac River mid-air collision
• Bering Air Flight 445
• Delta Connection Flight 4819
• Aerolínea Lanhsa Flight 018
• 2025 Hudson River helicopter crash
• 2025 San Diego Cessna Citation II crash
• 2025 Dhaka Chengdu J-7 crash

The two exceptions are Med Jets Flight 056 and Air India Flight 171.

Putting the date and incident first is the case for the majority of articles. For this article, the emergency and diversion are what establish notability.

Do you support or oppose this change? Dw31415 (talk) 01:31, 1 October 2026 (UTC)

Superpower

In 1991, Buzan argued that the emergence of multiple independent centers of power and the weakening of spheres of influence made the term ‘‘superpower’’ less appropriate than ‘‘great power’’.Buzan, Barry (1991). "New Patterns of Global Security in the Twenty-First Century". International Affairs. 67 (3): 431–451. JSTOR 2621945. Dw31415 (talk) 12:31, 5 October 2026 (UTC)

Your thread has been archived

Hello Dw31415! The thread you created at the Teahouse, WP:SEMI for Flydubai flight, has been archived because there was no discussion for a few days. You can still read the archived discussion. If you have follow-up questions, please . See also the help page about the archival process. The archival was done by lowercase sigmabot III, and this notification was delivered by KiranBOT, both automated accounts. You can opt out of future notifications by placing {{bots|deny=KiranBOT}} on top of the current page (your user talk page). —KiranBOT (talk) 03:04, 9 October 2026 (UTC)

Thank you

Thank you, for your inspiration with the draft article. After more research, I found the most common spelling of the name. Created article. Article now located at = Yaniv Hayun.

Thanks again, Princessa Unicorn (talk) 14:42, 13 October 2026 (UTC)

A barnstar for you!

	The Teamwork Barnstar
	Thanks for your inspiration and research help, with the new article page, Yaniv Hayun. Most appreciated! Princessa Unicorn (talk) 17:04, 13 October 2026 (UTC)

New cross-language page link: Yaniv Hayun

A new page ইয়ানিভ হায়ুন on the Assamese Wikipedia has been linked to the page Yaniv Hayun, which you created, via Wikidata (Q141638044).

You will not be notified about future links to pages you created. To opt in to these notifications, follow the opt-in instructions.

NervousBot (talk) 14:53, 14 October 2026 (UTC)

LLM at VP

Draft talk:John Michael Faile (Athlete)#c-Gurkubondinn-20260911091900-MWFwiki-20260911080900 Dw31415 (talk) 13:01, 16 October 2026 (UTC)

AINB idea

I have a possible idea (I am not a bot programmer or script writer and have no idea if this is feasible).

Lately we have had some monstrously big cases. We have resorted to mass LLMPROD via AWB and edit removal via mass rollback. However, this does not play well with the article tracker system. I end up spending a lot of time following the LLMPROD team around and marking the articles they have PRODed on the trackers, and also following myself around (as the biggest user of mass rollback) and marking the articles I have rolled back. For cases like the recent 3500 LLMPROD one this is not going to get done. (In that one we are planning to wait five days, remake the tracker, do another round, remake the tracker....)

It's important to know which articles have been handled, particularly for LLMPRODs where the tracker allows us to easily spot articles that fail to vanish in the expected time frame, and may have been inappropriately deprodded.

Is there any way to bridge between the semi-automated prodding or rollbacking of articles, and the tracker markings? Can something ask "Does this article have an LLMPROD tag? Does it have an AI-generated tag? Has it been deleted? Are the edits in this list of edits all reverted?" and update the tracker accordingly? M kuhner (talk) 23:38, 16 October 2026 (UTC)

I'll do a little research and get back. Do you have a handy example I can review? Dw31415 (talk) 01:33, 17 October 2026 (UTC)

For LLMPRODs you could look at the monster, Wikipedia:AI noticeboard/Carigval.97 with 3500 LLMPRODs now done. There are four trackers attached but no one is updating them, it's infeasible, we will hope most of the LLMPRODs go through and then remake the trackers. (Trackers die if you go over about 1200 articles in a single one.) I am also planning to ask consensus for mass rollback on this case but have not done so yet.

For a current mass rollback case, look at Wikipedia:AI noticeboard/Pristino. There are 739 uncompleted articles in the tracker. I have used mass rollback and rolled back an unknown number of those. Now I am going through one by one, seeing if I've rollbacked it and trying to resolve if not. It would be a lot easier if the maybe 200 I rolled back were marked automatically. (Yeah, I could have sat there with the mass rollback script open in one window, the tracker open in the other, and marked each one in the script and again in the tracker. But this really reduces the value of the automated rollback.)

Some of these problems could be fixed by doing all LLMPRODs and mass rollback before making the tracker. But that imposes a 5 day delay on regular cleanup which is not desirable (since the tracker has to be made after the LLMPRODs resolve to avoid having to hand-mark them) and is also not a natural workflow for us: sometimes I don't really grasp that a case is a monster until I try to make the tracker.

AINB-helper currently has a button that allows all deleted articles to be auto-marked as resolved, which is huge in cases with mass LLMPRODs. Three more like that (but much harder, I suspect): Mark (in the tracker) all articles with an LLMPROD tag currently on them. Mark all articles where all edits from this editor have been reverted. (Alternatively, all articles where the "how much text remains from this author?" comes back low or zero--but I suspect that will run too slowly to work.) And mark all articles which bear an AI-generated tag.

I hope that makes sense. Thanks for your interest! M kuhner (talk) 02:07, 17 October 2026 (UTC)

Wow. Thanks for your work! I reviewed Pristino. That’s sooo much to be done! My first thought on a better process/tool would be some kind of work flow like:

1. Identify the user
2. Go to a Toolforge web app and put the user and a date range in a form.
3. The app loads a list of all article pages the user touched in the date range
4. The cleaner (you) requests a lock on 10 pages to review
5. The app confirms the text added by the slop-er are still on the page
6. The app assigns 10 pages to you
7. You update in the app as you remediate
8. The app updates its DB
9. The app updates the tracking table on the AINB

Is that at all on the right track? Which part sounds valuable? Dw31415 (talk) 02:39, 17 October 2026 (UTC)

I can see how #7 works for hand editing. But the current AINB-helper script is not bad at all for hand editing. The problem is getting any of this to play well with mass editing (AWB based mass LLMPROD or script based mass rollback).

If I have to rollback Carigval (it's under discussion right now) I might catch as many as 1K rollbacks in their last 10K edits. (Or not; the success of rollback varies a lot from case to case. Carigval has a lot of very recent edits, so it might.)

I know that Chaotic Enby is working on an LLM-cleanup browser; you might want to chat with them if you're heading in that direction. In the long run that's what we need. But in the short run, I note that the "mark all deleted articles as completed" button in AINB-helper has probably saved me hours already, and could save a lot more hours in a case as big as Carigval; are there more low-hanging fruit like that? M kuhner (talk) 03:34, 17 October 2026 (UTC)

Just as a historical note: Pristino is pushing the upper edge of past success in clearing a whole case. Everything bigger, we've done LLMPRODs and then given up. The non-creation edits in Carigval will probably not get done, beyond what I may do with rollback if we decide I can. Most cases, luckily, are in the 10-100 article range. If we are motivated, those can get done. (Kowal2701 does the LLMPRODs and I do the minor edits, after filtering out edits smaller than +/-35 and some tags like undo and redirect.)

AINB basically has two bottlenecks: determining the date range, and doing the cleanup. Cleanup seems easier to automate, though date range is a HUGE problem as it is very time-consuming. I cheer when I see someone who started recently and can easily be seen to have been using LLM from the start. When they've been here since before 2023 and you have to decide when they started using LLM, it's very tedious. I have little or no clue how to improve that step. M kuhner (talk) 03:41, 17 October 2026 (UTC)

Hopefully the paste-check will help us find slop earlier Dw31415 (talk) 03:43, 17 October 2026 (UTC)

After you do some clean up if you post here I can check your contributions to get a better sense of the work. Dw31415 (talk) 03:42, 17 October 2026 (UTC)

Your comment

I don't want to overreact if I'm misinterpreting you, but can you explain your cryptic comment at diff. Is this just a helpful "btw" or are you saying "I disagree with something you wrote and see this essay for why"? SnowFire (talk) 19:24, 17 October 2026 (UTC)

Oh, sorry. I was in a rush and just wanted you to be aware of it. Please expand that section of the essay if you have experience that would be good to share. Dw31415 (talk) 19:26, 17 October 2026 (UTC)

Thanks for the clarification. For future reference, a curt comment saying "See WP:SOMESHORTCUT" usually means "the above statement is wrong, please see this official policy for the correct guidance." I recognize that wasn't what you were trying to say per the above, but I'd suggest editing your comment to something like "Interesting, consider checking and adding to WP:SOMEESSAY" for the "friendly" version.

I'll take a look, thanks. SnowFire (talk) 19:42, 17 October 2026 (UTC)

Thanks! I appreciate the feedback. Dw31415 (talk) 19:46, 17 October 2026 (UTC)

Feedback request: All RFCs request for comment

Your feedback is requested at Talk:Slavic Native Faith on a "All RFCs" request for comment. Thank you for helping out!
You were randomly selected to receive this invitation from the list of Feedback Request Service subscribers. If you'd like not to receive these messages any more, you can opt out at any time by removing your name.

SodiumBot (botop|talk|contribs) 13:30, 18 October 2026 (UTC)