Wikipedia talk:Editing policy

Would something like WP:LLMPROD provide a reasonable course of action for dealing with mass-created articles?

As part of the ongoing discussion at Wikipedia talk:Notability (geographic features)#Proposals from Workshop the point was brought up that the WP:MASSCREATE guideline doesn't currently address what to do with articles that have been created as part of un-approved mass creation campaigns.

One of the main reasons community approval should be required for mass article creation is to allow the community to review the sources to be used. It has often been the case in the past that articles have been mass-created using dubious sources or by editors who didn't have a good understanding of the sources they were using (examples ). This leaves a huge burden for reviewers who have to make sense of a huge number of stubs which are often poorly sourced to the point that it is difficult to even figure out the exact nature of the subject and if it meets wikipedia's notability criteria.

Given the recent introduction of WP:LLMPROD and the fact that the problems created by LLMs are very similar (a large quantity of dubious information that the community can't verify in a reasonable timeframe), I would like to gage if there would be any support for implementing a similar mechanism to remove unapproved mass-created articles as part of WP:MASSCREATE. Giulio 21:09, 18 May 2026 (UTC)

I wouldn't support that.

I would support identifying such unapproved mass creation campaigns early, and getting them stopped right away. It is much easier to deal with 500 articles than 5,000. WhatamIdoing (talk) 04:43, 19 May 2026 (UTC)

Firstly, 500 articles in no joke, while it could take a creator 1 or 2 days to use a semi-automated tool to put up, it could take a dedicated editor weeks to sort through that many articles, if we can even spot what is happening and someone is willing to put in the tedious work to sort through them.

Secondly, this is essentially the status quo, which I don;t think is working to anyone's satisfaction, do you have any suggestions for how such mass-creation attempts could be spotted earlier? Giulio 17:40, 19 May 2026 (UTC)

I wonder if what we need is a technical solution that limits someone to creating X new pages in Y time? Thryduulf (talk) 17:48, 19 May 2026 (UTC)

I think that we would probably want to have some way to override it, but yeah, it would make sense to have that as a technical constraint which eg. an admin can then override temporarily for someone once they've received approval at some board where people go to get approval for mass-creations. --Aquillion (talk) 18:12, 19 May 2026 (UTC)

I've been wondering the same thing. Maybe a limit of 10 or 20 articles per day, and editors could request that it be lifted for specific mass creation tasks. We'd have to follow up to make sure the privilege isn't being abused, but at least there would be some accountability. –dlthewave ☎ 18:13, 19 May 2026 (UTC)

In terms of temporary lifting, a "masscreate" flag applied to an account in the same way the bot flag is currently would seem to be the obvious way of doing that. Technically: able to be set and unset by admins, by default for a temporary period. I'd say talk pages and subpages (in namespaces where they are enabled) of pages created shouldn't count towards the limit (e.g. if the limit is 10, you can create 10 articles and the talk pages of those 10 articles or 10 modules and an unlimited number of subpages of those modules, and/or maybe the limit is 10 pages per namespace per day?). Socially: only set following consensus for a mass creation task at an appropriate venue (a village pump or WikiProject being the ones that come first to mind) and able to be revoked by any admin in the case of abuse, task expiry or uncontroversial request.

Questions: Should admins be allowed to self-grant? Should it apply to project-space pages? Should it apply to drafts? Should page moves count as page creations for this purpose? If so, should page movers be exempt? Thryduulf (talk) 20:03, 19 May 2026 (UTC)

I would focus on creating or moving pages into article space. Everything else has too much collateral damage: We often create talk pages when adding project templates or user welcome messages, for example, and project space would limit the number of AfDs per day. I like the idea of including draft space, but it might be easier to just do a blanket restriction on moves into article space so that someone can't get around it by creating in sandboxes or project space. Does that sound technically feasible? –dlthewave ☎ 21:18, 19 May 2026 (UTC)

Also turning redirects into articles.

I think this is worth pursuing. I think that it would be best to start with a bot that counts per-day and per-week article creations and drops a note at a suitable page (could be this one, at least initially). WhatamIdoing (talk) 21:49, 19 May 2026 (UTC)

I agree. This is an idea worth pursuing. - Enos733 (talk) 15:07, 22 May 2026 (UTC)

Questions: Should admins be allowed to self-grant? - definitely not in usual cases, although I'm not sure we'd need a technical prohibition against that. The purpose of the flag would be to say "there is a community consensus for it"; the ability of admins to grant it would just be so we have enough people who can quickly respond to a clear consensus by granting it. Granting themselves the flag would be a violation of WP:INVOLVED. (Possibly in the case where someone else made an uncontroversial NAC on a mass-create request by an admin, that admin could then grant themselves the flag, since that's a technical action and, in theory, not a judgment call, but it'd still be a violation of INVOLVED as written and I feel that having an actual admin have to implement the results of an RFC that are hard to reverse for WP:FAIT reasons is important for adding an extra layer of security and, therefore, probably shouldn't be done by an INVOLVED admin even if it's not usually something that involves their judgement.) --Aquillion (talk) 15:53, 20 May 2026 (UTC)

I think that monitoring (in the first instance) and perhaps constraining (in the future for those editors without MASSCREATE approval) prolific creation of various sorts is a really sensible move. For what it's worth, here are a couple of illustrative Quarry queries that I whipped up with the help of Microsoft Copilot: the query logic looks sound to me, and I sampled a few of the outputs to check that that the queries are not returning absolute rubbish.

• Prolific creators yesterday, where prolific is "more than 2"
• Prolific un-redirectors yesterday where prolific is "more than 2"

Obviously the bar can be set higher. I do note that when redirects are tagged for discussion, that counts as an un-redirect, so I might try and remove those edits if I can figure out how best to do that. I'm also going to try and make one for "Prolific movers into mainspace yesterday". Cheers, SunloungerFrog (talk) 05:51, 20 May 2026 (UTC)

Some of the prolific creators are creating WP:DAB pages, which we might or might not care about. If you want to exclude those, then there's some code in https://quarry.wmcloud.org/query/103977 that might be helpful.

Obviously, the bar should be higher than just two pages. But it should maybe lower than MASSCREATE's "25 to 50" upper limit, and/or we should run a weekly check as well.

The footnote says that more than just the mainspace should be included: "articles, most visible categories, files hosted on Wikipedia, mainspace editnotices, and portals". I'm not sure that we need to include all of these, but we should probably include whichever ones are easy to count. WhatamIdoing (talk) 06:07, 20 May 2026 (UTC)

Thanks for the pointer to the code for identifying DAB pages; I'll take a look.

Obviously, the bar should be higher than just two pages. But it should maybe lower than MASSCREATE's "25 to 50" upper limit, and/or we should run a weekly check as well. Yes of course. What I had imagined was that the putative bot might run two main routines:

• first, logic that identified editors who met a (lower than MASSCREATE, but TBD) threshold of prolificness, to get them on a "prolific list". Such editors would stay on the prolific list for e.g. 7 days; if they met the threshold again, the counter would reset.
• second, for every editor on the "prolific list", logic to retrieve more detailed stats, probably in the form of graphs linked to Special:Contributions web queries, relating to prolific contributions.

The aim being to have a short-ish boiled down page of (tens of, not hundreds of) prolific editors and their recent history of contributions that editors/administrators can eyeball and use as a starting point for more detailed investigations about whether the boundaries of MASSCREATE are being exercised. Cheers, SunloungerFrog (talk) 07:16, 20 May 2026 (UTC)

How often do you expect this to run? WhatamIdoing (talk) 17:51, 20 May 2026 (UTC)

Once a day, I think. Cheers, SunloungerFrog (talk) 18:03, 20 May 2026 (UTC)

That's what I was thinking, too.

If we can get all the things (creations + un-redirecting + etc) in the same list, then I think that flagging anyone doing, e.g., 15 in a single day or 50 in a week might be good. That would keep the list short enough to focus attention, but not miss many.

If we have to have separate lists for each, then maybe 5 per day of anything is the right place to start. WhatamIdoing (talk) 02:41, 21 May 2026 (UTC)

OK, User:SunloungerFrog/sandbox/Prolific creators is my latest stab at this, essentially the output of quarry:query/105582. At the moment it is only article creations in mainspace (not un-redirects or moves into mainspace), and I will be manually updating it once a day, as I don't have a bot. It shows:

• the preceding seven days of article creations
• in date/time order of time of creation per editor
• for all editors who, in the preceding seven days
• have created five or more articles on at least one of those days.

I think it's 49 editors for the current version of the page.

On un-redirects, I am currently struggling to filter out submissions to RfD because to do that I need to be able to query the article text, and that isn't possible with a straight quarry query because the database replicas don't contain the article text. So I have my thinking hat on.

On moves into mainspace, I haven't got a good working query yet.

Grateful for feedback and suggestions, though make them simple as I am not the world's most natural coder. I will try and improve the table a bit. Cheers, SunloungerFrog (talk) 19:42, 23 May 2026 (UTC)

Un-redirects get a Special:Tags from the AbuseFilter, which I believe would be visible to in Quarry.

Would you like to take this to Wikipedia:Request a query? (Ping me if you do!) WhatamIdoing (talk) 20:19, 23 May 2026 (UTC)

I think the issue is that nominations for RfD also get that tag, e.g. so it's not easy (possible?) to distinguish between converting a redirect to an article or disambiguation page and nominating a redirect at RfD. Thryduulf (talk) 20:25, 23 May 2026 (UTC)

If edit summaries are visible then excluding those that include "Redirects for discussion" and "RfD" (in any capitalisation) will catch most (17 of the 21 redirects listed on today's page had at least 1 of those phrases in the edit summary of the edit that removed the redirect (most commonly as a link to the discussion, e.g. "Listed for discussion at Wikipedia:Redirects for discussion/Log/2026 May 23#Oxjam". The four that didn't had no edit summary at all. Thryduulf (talk) 20:38, 23 May 2026 (UTC)

Exactly. quarry:query/105446 essentially uses that tag. But you need the page text to filter out the chaff of RfD nominations. I think I could do it by filtering on edit summary, so am going to look at that. It's a bit manky, but maybe the best we can get. Cheers, SunloungerFrog (talk) 20:39, 23 May 2026 (UTC)

I think this is a useful starting point.

Locking the ability to publish more than 25 articles a day behind a new permission that admins would need to grant if the proposed mass-creations get a community approval would be an effective way to contain the problem, and such approval would have to be dependent on the sources being provided in advance for the community to review and ensure they are understood correctly (maybe we could also have a limit on articles per week and articles per month that is lower than a sustained 25 articles per day).

I don't think this should cover things like disambiguation and redirects as problematic mass creations seem to be a lot rarer and its quite easy for a page you publish to require multiple redirects, particularly if the subject has a name that changed a lot over time (e.g. Kharabet Ihrit), or might be the subject of many weird transliterations (e.g. Mobye Dam). The issue here is I could see it being technically challenging to apply the restriction only to article creation without blocking the creation of redirects. Also I don't think its necessary to limit non mainspace pages as I am not aware of similar issues happening there.

I also think a tool that would flag users who produce more than say 10 articles in a day as a priority for e.g. NPP to review would help as I think a user producing 25 articles per day for an extended period of time is still a massive problem if all the articles they produce have systematic issues like a misinterpretation about what a source is saying about a series of subjects. Giulio 21:31, 20 May 2026 (UTC)

I'm no programmer, but AIUI redirects are a technically distinct type of page to articles so I expect excluding them would be technically possible (I make no comment about how easy it would be). There would be at least edge cases that would need defining though:

• Pages changed to redirects after initial creation. This could be due to e.g. syntax errors, page moves, being manually converted, etc.
• Pages created as redirects but later unredirected. In good faith this could be to convert it to a disambiguation page, a breakout article, etc. or, in bad faith, an attempt to bypass article creation limits.
• Pages that change state multiple times (edit warring is the first reason that comes to mind but I'm sure there are good faith reasons too).

Thryduulf (talk) 21:55, 20 May 2026 (UTC)

If you look at the links given:

• Prolific creators yesterday, where prolific is "more than 2"
• Prolific un-redirectors yesterday where prolific is "more than 2"

it's easy to include or exclude un-redirectings. It's more difficult to differentiate between turning a redirect into a proper article vs sending it off to Wikipedia:Redirects for discussion. WhatamIdoing (talk) 02:45, 21 May 2026 (UTC)

In terms of differentiating between nominating a redirect for discussion vs converting it into an article, a bot should be able to distinguish article pages that have a link to Wikipedia:Redirects for discussion from those that don't, but whether software can do this I don't know. The only reasons I can think of for an article (not it's talk page) to have a link to RfD are:

• Disambiguation pages or articles being drafted in response to an RfD nomination, but these are still fundamentally RfD nominations (and redirects converted to articles tend to get quickly procedurally closed at RfD, some subsequently get taken to AfD).
• An article mentioning RfD from an encyclopaedic perspective, but the number of these is going to be vanishingly small and they should be using external link syntax anyway.

Thryduulf (talk) 09:21, 21 May 2026 (UTC)

I think that such approval would have to be dependent on whatever the community wants, when the community is making that decision. WhatamIdoing (talk) 02:43, 21 May 2026 (UTC)

I oppose the proposed creation of a technical limit on the rate of article creation. (1) Such a limit would indiscriminately obstruct the creation of high quality articles about notable topics. (2) Once such a limit is introduced, it can be lowered; and there would probably be endless campaigning and bludgeoning to lower the limit that would never stop unless and until the limit was lowered to "zero articles per forever". (3) In general, I think that, both as a matter of principle and for practical reasons, we cannot place a limit on the number of articles someone can create, unless we also simultaneously place a limit on the number of AfD nominations and delete !votes, PRODs, CSDs, draftifications, and talk page comments advocating for deletion (or against creation), etc, someone can make. We would also need to place a word limit on the permissible length of such !votes and talk page comments. James500 (talk) 01:35, 23 May 2026 (UTC)

If an editor is creating 25 high quality articles about notable topics per day, they can simply request that the limit be lifted for their account so they can complete that task. –dlthewave ☎ 03:43, 23 May 2026 (UTC)

The limit already exists as a policy WP:MASSCREATE, this proposal is merely intendent to prevent people from simply ignoring the policy as happens now and leaving a mess for others to clean up if the articles don't meet wikipedia's other policies and guidelines. Editors who have a large number of high quality articles to submit can either request community approval to have the limit lifted for them (as in the current policy), or if they have a one off batch of articles can prepare all the articles in draft or user space and move them to main space in batches over a few days. Giulio 08:33, 23 May 2026 (UTC)

(1) The proposed limit does not exist as policy. The policy only applies to "automated or semi-automated" creation, and says that "no specific definition of large-scale was decided". (2) No evidence has been presented that mass creation is happening now, or that it has happened recently. James500 (talk) 11:15, 23 May 2026 (UTC)

Will you accept a thousand new articles by one editor in the last six weeks as evidence of mass creation happening? (The editor has permission to create these articles at this speed; I believe his record was something around 100 articles in a single day.) WhatamIdoing (talk) 18:56, 23 May 2026 (UTC)

The relationship between WP:PRESERVE and WP:ATD-E has an RfC

The relationship between WP:PRESERVE and WP:ATD-E has an RfC for possible consensus. A discussion is taking place. If you would like to participate in the discussion, you are invited to add your comments at WP:Village pump (policy)#RfC: The relationship between WP:PRESERVE and WP:ATD-E. Thank you. Flatscan (talk) 04:27, 17 July 2026 (UTC)

Proposed addition

From the "Wikipedia is a work in progress" essay, one of the pictures states that "The missing puzzle pieces in the Wikipedia logo reflects the fact that Wikipedia will never be truly complete; it is a constant work in progress." Therefore, we should add that sentence into the "Wikipedia is a work in progress: perfection is not required" section. Redzone2 (talk) 21:40, 29 July 2026 (UTC)

Good to know, but it seems a bit like wp:KUDZU to me. - Butwhatdoiknow (talk) 22:11, 29 July 2026 (UTC)