Anomie is still around, mostly to maintain AnomieBOT. But after the WMF proved that office politics are more important to them than seemingly anything else, and otherwise generally seem more concerned with their own image than substance, Anomie is not engaging in technical work on MediaWiki.
I saw that youundid the change that I made toWP:ADMIN. I went back and re-read the changed wording that I introduced, and yeah... that actually made no sense at all. I appreciate you for doing that, and I wanted to leave you a message and thank you. :-)~Oshwah~(talk)(contribs)09:17, 12 September 2025 (UTC)[reply]
I have gone through the AnomieBOT codebase and I am really impressed by how well you have managed everything right from the start. The way you have documented even the smallest details and organised the code folders is excellent. I am sure I will spend more time exploring these pages and will definitely learn a lot from them. Thank you for all your efforts in powering so many important Wikipedia processes through AnomieBOT. :) –DreamRimmer■16:25, 15 September 2025 (UTC)[reply]
Thanks! A few thoughts that might be helpful asyou look at the code:
AnomieBOT predates Toolforge, and it used to run on a repurposed desktop machine in my house. Some of the architecture reflects this and doesn't fit all that well with Toolforge, e.g.
Having to have things running as one-off jobs withbot-wrapper.sh restarting things instead of running as continuous jobs (because ofT361405).
Possibly also thatbot-instance.pl does its own loop running multiple tasks in one job instead of having a job per task. Although possibly I'd need to request a bunch of increased quotas and run into other problems to be running 44 separate jobs. 😀
That I already have a push-to-deploy setup that theyonly now have something in beta for, and I'm unlikely to ever switch to their thing once they get it out of beta.
One thing I really wish I had done is include automated testing. Right now, all testing is manual. Which at least I have a pretty robust setup for, although not all of it is in the uploaded code: I have one extra script that downloads theAnomieBOT_Store table entries for a task from Toolforge to a local database, then runs the task viatest.pl and sends it SIGSTOP when 50 edits show up in the directory (and then empties the directory for the next 50).
I really appreciate you taking the time to explain all this, it makes everything much clearer. My own setup has been quite messy, with jobs and files scattered under random names, and I often struggle later to figure out what is what. Seeing how you organised AnomieBOT has given me a lot of ideas, and I plan to work on an automated setup of my own, making sure to structure things properly this time. I think I will be able to save a lot of time by adapting some of your code into Python rather than figuring out everything from scratch. Looking at the edit history, it's impressive how robust and automatic your system was right from the start. Thank you again, this really helps a lot. :) –DreamRimmer■15:08, 16 September 2025 (UTC)[reply]
Hi Anomie,it seems that your bot didn't createTemplate:POTD protected/2025-09-22 and so on; has this part of its job stopped running and could you restart it please? While I am here, do you know why we even have this setup with two versions of the POTD template? With my limited amount of imagination, I can't think of a good reason why POTD is different from OTD, which does not have a separate version, so I would appreciate being enlightened (I find the existence of two versions puzzling and sometimes annoying when responding toWP:ERRORS reports). The only thing I can imagine is that the setup is from before we had cascade protection and admin bots. —Kusma (talk)08:42, 21 September 2025 (UTC)[reply]
Thank you for the explanation, I hadn't seen that the unprotected version had been created so late. I might start that discussion, but I'll wait until the next time when it annoys me :) —Kusma (talk)13:05, 21 September 2025 (UTC)[reply]
Hey, I know that you don't really work on MediaWiki these days, but I just wanted to let you know that we're working on a version of your OAuth Hello World app that uses OAuth 2. We'll probably set up a Git repo somewhere to host both versions, but apart from that, it'll stick to the single-file no-dependencies approach. Thanks for writing that tool, and let me know if there's anything particular you'd like us to do. (For reference:T384442)Matma Rextalk22:13, 24 September 2025 (UTC)[reply]
For this task, it gets the content fromTemplate:Ffd log day. Note that if that template gets changed, the bot will try to replace the existing header on all recent-ish discussions with the new version. Make sure it continues to end with a header like=== October 1 === or the bot will get confused and probably break things.Anomie⚔21:19, 1 October 2025 (UTC)[reply]
I hope you know what you're doing in declaring the metadata class "unnecessary". I'd be wary that people's tools or scripts might use it.A quick search turns up a lot of possible references to it, although how many are for something else I couldn't say.Anomie⚔20:04, 2 October 2025 (UTC)[reply]
It wasremoved from{{Afd top}} to "fix in mobile view", so this could achieve the same thing. A quick look at the link you put shows that most are just to put "display: none", which could simply be done by directly targeting this class. I see no reason to target this element with that class specifically, and it appears to just have been absent mindedly added when someone created that template. —Matrixping mewhen u reply (t? -c)16:06, 3 October 2025 (UTC)[reply]
To the admin with the impeccable logic: I was so impressed by your philosophical argument. Well reasoned!Your edit summary was a masterpiece of irrefutable logic. For such philosophical brilliance on the wiki, I hereby award you the Socratic Barnstar! 😊TheEagle107 (talk)23:57, 5 October 2025 (UTC)[reply]
I'm not doubting you after seeing the blanking either, that indicates enough to me. It was more so to be aware that like others assessing a dispute, I'm only considering what I read in that discussion, as someone who doesn't know that editor either (otherwise I might have a very different outlook). It's also only my perspective and I think I've been judgemental enough for one day when it was never asked for in the first place to be honest.Also trying to avoid replying where I said I wouldn't reply. Regards,CNC (talk)00:38, 10 October 2025 (UTC)[reply]
To answeryour question: I think that our handling of (true) even splits is weak, with a higher than average risk of a supervote or of throwing up our hands and saying that nothing can be done. But if you encounter that, I'd suggest two approaches: First, ask the QUO people which of the other colors is "least bad" in their opinion. It could be that a 3:3:3:3 four-way split becomes a 0:3:6:3 three-way split with a dominant option. Second, ask everyone to list a second choice. It's unlikely that the result would then be a perfectly even split for first choice and also a perfectly even split for second choice.
Why only ask the status-quo people for another opinion? Again, why should they have a higher threshold to get their preference?As for asking people for second choices, that gets towards either doing approval voting or ranked voting as I suggested elsewhere in that discussion.Anomie⚔00:53, 13 October 2025 (UTC)[reply]
It's also 9:3 against every other option. Again, what makes the status quo special that you only apply that logic to it?Anomie⚔01:09, 13 October 2025 (UTC)[reply]
But also 9:3 against red = consensus for not changing to red, etc. Again, why is the status quo special here? It's just another color option like all the others. You keep dodging that question, just asserting that it is.So say you get your way and they change to red. Then 2 months later someone asks ifconsensus has changed, it's still 3:3:3:3, but now since red is the status quo we need to change away from that? And repeat every 2 months thereafter? Seems great for neophiles, not so great for people who like stability.Anomie⚔12:22, 13 October 2025 (UTC)[reply]
The status quo is special because a four-way tie for four colors is actually communication about two things: A vote for red says I would preferred, but also, it implicitly says I would preferchange. In this example, there are three votes for each of four individual colors, and there arealso nine votes for change vs three votes for no change.
(Contrast this with a two-way even split: six votes for status quo vs six votes for red. That would also be an equal split for change vs no change – and, having no consensus for anything [=neither for a color nor for change in general], we would usually default to no change.)
Having determined that there is a preference for change does not permit us to choose one of the non-status quo colors. (How could you pick one over the others, without either supervoting or flipping a coin?) Therefore they can't, on the basis of this 3:3:3:3 discussion, change to red. You'd have to have additional discussion to decide which of the non-status-quo colors to change to. For red to actually be present in your scenario two months later, we'd have to form a consensus that red really was preferred, if status quo couldn't be chosen. In that case – imagine a result of x:3:6:3 – a subsequent discussion with a 3:3:3:3 result would represent an actual change from the previous consensus, and yes, red could be moved away from, and yes, this could repeat endlessly. But this seems unlikely to happen in practice.WhatamIdoing (talk)15:53, 13 October 2025 (UTC)[reply]
You're making an unwarranted assumption that everyone who votes for not-the-status-quo really prefersany change over no-change. Other than for those voters (like you!) who actually express a preference for "anything else would be better", we can't know that from the sort of ad hoc !voting we normally do. In our 3:3:3:3 example, it may well be that all the red partisans prefer the orange-y status quo to blue or green (because it's closer to red), and some of the green partisans prefer it to blue or red too (because it's closer to what they like about green than blue is), even if all the blue partisans absolutely hate it (because it combines the worst of red and green). A bartender's close is only appropriate when discussion indicates that "change" really is preferred or required (e.g. we can't keep the page at "Bob Smith" because this particular Bob isn't the primary topic, or in a color vote we explicitly do approval or ranked ballots), even if people can't agree on what to change it to.Anomie⚔19:14, 13 October 2025 (UTC)[reply]
I'm making the entirely warranted assumption that everyone who votes for not-QUO has a first preference ofa change over no change. In the case of color choices, I'd assume that nobody would approve of all color options. I could confidently vote for "anything" because I knew I could rely on the community to not choose (e.g.,) a black-text-on-black-background color combination.WhatamIdoing (talk)19:50, 13 October 2025 (UTC)[reply]
Supporting "a" change as a first choice is not the same as having no-change as the last choice. As I already pointed out. Just because you personally might be happy with any change that you imagine people would support over no-change doesn't mean everyone who didn't express that preference feels the same.Anomie⚔20:02, 13 October 2025 (UTC)[reply]
Yes, which is why I said above that you can't just jump from "consensus to change tosomething" to "consensus to change tothis color specifically". In the absence of additional information (e.g., a comment that says "I'd prefer blue, but if I can't have blue, then I prefer the old color"), I think that we would have a consensus against the old color, but not (yet) a consensus that can be implemented. Reaching a consensus-for-new, rather than a consensus-against-old, would IMO require additional conversation.WhatamIdoing (talk)20:58, 13 October 2025 (UTC)[reply]
Again with the unwarranted assumption that people whose first preference is for a change implicitly support any change over no-change. If they don't state either way, we shouldn't assume either way. And since we can't assume either way, we also shouldn't have different thresholds for no-change versus any other option.Anomie⚔23:27, 13 October 2025 (UTC)[reply]
Supporting one potential "a" change doesn't mean supporting any possible "a" change over non-change. I already explained above how this could play out in the example.Anomie⚔23:48, 13 October 2025 (UTC)[reply]
Yes, you're right: One could support "a" change without supporting "every single possible change". But by supporting "a" change, one has also inherently and unavoidablyopposed "no change".
Because we don't know what people's second choices are, it's necessary to have further discussion.
And for purely practical this-is-how-ordinary-humans-work reasons, we have to narrow the responses. If we say "Tie vote – everyone vote again!", the response will be the same. So we can't do that. We can say "Tie vote on the details, clear majority on doing something different – everyone vote again, but this time, 'no change' isn't a valid option".
What some might call an artificial narrowing of options is an ordinary approach that is used in real-world elections all the time. With aPrimary election, some voters' first-choice candidate might not win. That means they may not like any of the candidates on the ballot. Thinking "Oh, dear, I'd have voted for Paul Politician back then, if I knew that we were going to get these two crooks otherwise" doesn't change the names on the ballots. Sometimes you have to narrow down the field.
In the case of editors with a four-way tie on the exact color, then we can identify one difference, which is that there are a lot fewer "no change" vs "supportssome change" votes.WhatamIdoing (talk)00:21, 14 October 2025 (UTC)[reply]
So you're admitting you're supervoting away some options to try to force a different result. You should look into how doing that can have disenfranchising effects, which is why I'd suggest a ballot method that takes into account secondary preferences instead as that's less likely to be as problematic.Anomie⚔12:43, 14 October 2025 (UTC)[reply]
No, narrowing down a long list based on what most people vote for is not a supervote. A supervote is "equally divided, and I pick red". A supervote is not "equally divided for colors, unequally divided for whether to change the color: That's a consensus to change to something, so now please have a second discussion about which color to change to".
You can claim it's not all you like, and justify it as much as you want, but in the end you're still eliminating an option you don't like based on flawed reasoning. 🤷Anomie⚔18:21, 14 October 2025 (UTC)[reply]
I notice that you removed my second post on grounds that it is a "duplicate AI post". To clarify, another administrator asked me to rewrite my post manually. I did not use AI at all to write the new post; it was written from scratch manually. If there is anything you need clarification on, please let me know, but for your reference, see discussion at:User talk:Muboshgu#Re: Removal of proposal from Village Pump Idea Lab.86.33.69.28 (talk)22:38, 14 October 2025 (UTC)[reply]
There's no way you wrote it from scratch manually and came up with the same ridiculous structure of headers and such. It seems more like you lightly revised it, if you didn't just ask the LLM to shorten it or "make it less like an LLM" for you. Besides which, you should check the archives for other similar proposals, which have also been rejected.Anomie⚔22:59, 14 October 2025 (UTC)[reply]
The collapsed portion of my post was an example, and that was still AI-generated. It was to illustrate a case study of the tool. The proposal itself was manually written. Additionally, if you can, kindly reference or link specific examples of proposals similar to mine that were rejected?
With all due respect, I ask for permission to repost my proposal with your approval. And no, I didn't ask the AI to make it "less like an AI"; the text was written manually. I am not sure what kind of proof you are looking for. Regardless of whether or not you are opposed to my proposal, I would like to present it to the community and see what they say. If you have any concerns regarding how I am going about this, let me know.86.33.69.28 (talk)23:09, 14 October 2025 (UTC)[reply]
You're not going to get my approval, and I'm not going to do your research for you. But I can't stop you from trying to post it again if you insist.Anomie⚔23:32, 14 October 2025 (UTC)[reply]
Thanks for your reply. I understand there is a good chance my proposal will be rejected. However, I just want to make sure I am compliant with Wikipedia policy regarding posting AI-generated content. I want to clarify that the new post was manually written. Do you advise me to fully remove any AI-generated text from the post? For reference, here is a permalink to the post in question:[1]
The only AI-generated content in this post is collapsed under "Example article bias analysis (for starters)". If you want me to remove this so that there are no misunderstandings, let me know. I can also provide links to our discussion in the edit summary to avoid misunderstandings.86.33.69.28 (talk)23:58, 14 October 2025 (UTC)[reply]
Category:Use Malaysian English from November 2025 was deleted as unused on November 13, because that project isn't actually doing monthly tracking at all anymore (its template has been deleted outright by TFD), but then AnomieBot recreated it on November 14 because it's in the bot's list of expected monthly categories despite not actually being in use. So could you edit the bot to remove "Use Malaysian English" from the creation list? Thanks.Bearcat (talk)14:58, 16 November 2025 (UTC)[reply]
Also, please let me know once that's been done, so that I can redelete the category. I've left a "wait to delete because of the recreation situation" note on the category in the meantime, however. Thanks.Bearcat (talk)15:02, 16 November 2025 (UTC)[reply]
Hello! Voting in the2025 Arbitration Committee elections is now open until 23:59 (UTC) on Monday, 1 December 2025. Alleligible users are allowed to vote. Users with alternate accounts may only vote once.
TheArbitration Committee is the panel of editors responsible for conducting theWikipedia arbitration process. It has the authority to impose binding solutions to disputes between editors, primarily for serious conduct disputes the community has been unable to resolve. This includes the authority to imposesite bans,topic bans, editing restrictions, and other measures needed to maintain our editing environment. Thearbitration policy describes the Committee's roles and responsibilities in greater detail.
