Wikipedia:Village pump (idea lab)/Archive 59

• Village pump
  • Policy
  • Technical
  • Proposals (persistent)
  • Idea lab
  • WMF
  • Miscellaneous

Village pump (idea lab) archive  

This page contains discussions that have been archived from Village pump (idea lab).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

---

< Older discussions · Archives: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59

Ready for the mainspace

I'm here to solicit opinions about what it means for an article to be "ready for the mainspace". This phrase has turned up in hundreds of AFDs during recent years. Here's the story:

You are looking at an article. You have determined that the subject is notable, and that none of the Wikipedia:Criteria for speedy deletion apply to the article. Another editor says to you: "I don't think that article is ready for the mainspace".

What would you guess that the editor means? Is that consistent with our rules, such as the WP:NEXIST guideline or the WP:IMPERFECT policy? WhatamIdoing (talk) 19:32, 28 May 2024 (UTC)

Personally, and this is just my own opinion here, I find this "ready for the mainspace" thing a little ambiguous. As you said, as long as WP:GNG is met, an article that is properly sourced (or at least whose topic does) deserves to be in the mainspace. Not all articles are perfect, and by having an article in the mainspace, more people will see it and improve it, which is exactly the purpose of Wikipedia. It's a work in progress! Cocobb8 (💬 talk • ✏️ contribs) 19:45, 28 May 2024 (UTC)

I think it's a little ambiguous, too, which is why I'm asking. ;-) WhatamIdoing (talk) 20:14, 28 May 2024 (UTC)

However, WP:DRAFTIFY clearly states that the aim of moving an article to draft is to allow time and space for the draft's improvement until it is ready for mainspace, so maybe a change to that guideline could be required to make it clearer? Cocobb8 (💬 talk • ✏️ contribs) 20:17, 28 May 2024 (UTC)

If we can figure out what it means, that might be helpful. WhatamIdoing (talk) 20:32, 28 May 2024 (UTC)

I would generally interpret it as "WP:N has not been shown." Gråbergs Gråa Sång (talk) 19:53, 28 May 2024 (UTC)

Then what about you have determined that the subject is notable per @WhatamIdoing's original comment? Cocobb8 (💬 talk • ✏️ contribs) 20:02, 28 May 2024 (UTC)

If I couldn't see for myself why the other editor would say that, I'd ask. For myself, I could see saying "not ready for mainspace" for something so poorly or inappropriately written that it does a disservice to the topic and the reader (although I'd probably say specifically what my concern was). Schazjmd (talk) 20:12, 28 May 2024 (UTC)

Wikipedia:Drafts#During new page review says that it's enough that the topic is plausibly notable to draftify. An unsourced article with a claim of significance (or notability) could fit this description, not being eligible for WP:A7 but still not meeting the referencing standards for mainspace. Chaotic Enby (talk · contribs) 20:34, 28 May 2024 (UTC)

In my opinion, if the article draft meets all of the following it's ready for mainspace:

• Is not being discussed at XfD
• Would not meet a speedy deletion criterion in article space
• Has no identified copyright, BLP, etc issues
• Has sufficient sources to demonstrate notability
• Has been at least minimally proof-read (perfection is not required, basic readability is).
• Has no in-line editing notes ("need to reword this", "add more info here", etc) (excluding templates and hidden comments).
• Has no obviously broken templates (if you don't know how to fix it, ask for help before moving). Thryduulf (talk) 21:22, 28 May 2024 (UTC) ("article" changed to "draft" for clarity Thryduulf (talk) 23:20, 28 May 2024 (UTC))

I'm not sure why you're asking in this venue. The only way to know is to ask the editor making the statement what they meant. Even if it could be done, I don't think it will be helpful to try to establish a common interpretation. Editors should be specific about their concerns. isaacl (talk) 22:15, 28 May 2024 (UTC)

It's difficult to ask hundreds of editors. Also, if everyone has their own ideas, then the phrase becomes useless. We might as well just say WP:IDONTLIKEIT in that case. WhatamIdoing (talk) 22:29, 28 May 2024 (UTC)

The phrase is useless on its own, as it's not specific. It sounds more like you want to revisit the criteria for deleting an article, to examine what should be considered showstopping shortcomings. Commenters in deletion discussions should be encouraged to list those shortcomings. They can optionally add that as a result, the article isn't ready for the mainspace. isaacl (talk) 22:48, 28 May 2024 (UTC)

I do not want to revisit the criteria for deleting an article. Also, if you take a look, this phrase frequently is given as a reason for not deleting the articles (but instead moving them to Draft: or User: space).

Consider Wikipedia:Deletion policy#Deletion of drafts: "If an article isn't ready for the main namespace, it can be moved to the draft namespace". Commenters in deletion discussions can listed specific shortcomings, but the deletion policy itself can't. Is this a matter of pure consensus, in which case it's nearly indistinguishable from IDONTLIKEIT (which sounds worse than it probably would be in practice)? Does it mean, e.g., what @Thryduulf said about "Has sufficient sources to demonstrate notability", in which case WP:NEXIST is no longer valid? Would a visibly broken template count as the sort of IMPERFECT thing that the deletion policy won't countenance? WhatamIdoing (talk) 23:09, 28 May 2024 (UTC)

To be clear, my criteria are for moving a page from draft space to article space, not for moving a page in the other direction (where such issues as broken templates should simply be fixed). Thryduulf (talk) 23:18, 28 May 2024 (UTC)

Articles don't avoid deletion to be moved to draftspace simply because they're not ready for mainspace by someone's measure, but because someone thinks there's promise to demonstrate that the topic meets English Wikipedia's standard for having an article. There's no point in trying to retroactively figure out what others have meant by a non-specific phrase they used in the past. Moving forward, users should be asked to provide specific details, assuming that it's not already clear from context what shortcomings are being considered.

Regarding the quote from the deletion policy, I agree that ideally it wouldn't use a vague phrase. I appreciate, though, that the sentence is trying to be a placeholder to cover any scenario where the participants in a deletion discussion agreed that the best course of action was to move the article to the draft namespace. It's essentially tautological. isaacl (talk) 00:05, 29 May 2024 (UTC)

If it means "by consensus at AFD", then it should say that. We could change the deletion policy to say that.

In re no point in trying to retroactively figure out what others have meant by a non-specific phrase they used in the past, I don't agree. This phrase seems to mean something to people. You are the only editor who thinks that understanding what we want to communicate (in about a thousand AFDs, in the deletion policy, twice in Wikipedia:Drafts, in more than forty thousand pages all told). When a bit of wiki-jargon has been used tens of thousands(!) of times, I don't think that figuring out what we mean, and whether we all mean the same thing, is pointless. If it doesn't interest you, then that's fine, but please don't tell other editors that what they've been saying is meaningless.

Also, I suspect that in a substantial fraction of cases, "not ready for mainspace by someone's personal standards" is exactly what is meant. WhatamIdoing (talk) 02:10, 29 May 2024 (UTC)

To clarify, I mean from my view, there's no point in trying to guess at the meaning in a village pump thread. If we're serious about trying to figure it out, we should be systematic: take a sampling and ask the editors in question if they're still around. We can also analyze the discussion threads to see if there is enough context to understand. isaacl (talk) 05:05, 29 May 2024 (UTC)

This phrase is used in WP:DELPOL and WP:DRAFTIFY. The village pump is the normal place to discuss confusion that affects multiple policy/guideline/help/etc. pages.

But I'm no longer hopeful that we can have that discussion. If you look at this thread, five editors thought they had something useful to contribute. Then you started posting that you thought it was not helpful to figure out what editors mean, that it's useless, that there's no point – and nobody else has shared their thoughts since. I think you have effectively discouraged editors from sharing their their views. WhatamIdoing (talk) 18:54, 31 May 2024 (UTC)

The main thing that it means to me is that most claims in the article are sourced, and that they're sourced to enough separate reliable sources to establish notability by just reading the references. Many topics are notable in the sense that sources exist out there somewhere, but implicit in the notability guideline is that the reason we're looking to establish there exist such-and-such many reliable sources about a topic is to use those sources to write the article. Any article that does not actually do this is half-baked. Loki (talk) 04:30, 1 June 2024 (UTC)

@LokiTheLiar, how many existing articles do you think meet the standard of "most claims in the article are sourced"? WhatamIdoing (talk) 21:56, 1 June 2024 (UTC)

I'm aware that there's lots of bad articles out there, if that's what you're asking. I'd still say that the majority of articles meet that standard, and that the overwhelming majority of traffic to Wikipedia is to articles that meet that standard.

Like, compare naked butler, which doesn't meet the standard I've set here, to complaint tablet to Ea-nasir, which does. They're both small articles on obscure subjects but the complaint tablet one is totally fine. Loki (talk) 22:12, 1 June 2024 (UTC)

I believe that the complaint tablet has about five times as many sentences as the median article and about ten times as many sources. So if that's the standard, we'd probably be deleting about 90% of current articles. WhatamIdoing (talk) 22:19, 1 June 2024 (UTC)

To put it another way: The median article is a stub. You have given a C-class article as an example that should be considered a "small article". A quick look at Wikipedia:Version 1.0 Editorial Team#Statistics suggests that my off-the-cuff 90% estimate is correct. Only about 10% of articles (excluding lists, dab pages, etc.) rate as C-class or higher. WhatamIdoing (talk) 22:25, 1 June 2024 (UTC)

Stub class articles don't necessarily violate this standard. So for instance, I just found a list of stubs and clicked randomly and found Ty Barnett, which clearly meets my standard. Or have Fred Baxter or William Beavers, literally the next two articles I clicked on. All stubs of obscure people, all definitely meet the standard I laid out. Loki (talk) 04:31, 2 June 2024 (UTC)

• ORES says the first is Start-class. I think editors might have different opinions about whether it's a long stub vs a short Start, but at 200 words/10 sentences long, it is at minimum on the long side for a stub.
• The second is a four-sentence, four-source stub, which might put it around the median article for length, but I think it is above average for sourcing.
• The third is also Start-class. It has 2750 bytes of readable prose and 450 words. This is about twice the length of the maximum described in Wikipedia:Stub#How big is too big? The stub tag was removed from the article during an expansion in 2006. I have corrected the WP:1.0 rating on its talk page.

Looking at Fred Baxter (the second one), would you feel the same way if it had only three sentences and three sources? WhatamIdoing (talk) 06:16, 2 June 2024 (UTC)

Yes. I don't care about length at all. Loki (talk) 13:44, 2 June 2024 (UTC)

Are you interested in the number of sources, or the percentage of sentences with inline citations? WhatamIdoing (talk) 18:05, 2 June 2024 (UTC)

Number of sources only has to be enough to meet the notability guideline. Otherwise it's fraction of claims that need to be sourced that aren't. Loki (talk) 23:39, 2 June 2024 (UTC)

WP:NEXIST says that the number of citations required to meet the notability guideline is zero. (Per that long-standing guideline, the sources have to exist in the real world, but they don't have to be cited in the article.) There are no claims in User:WhatamIdoing/Christmas candy that need to be sourced (nothing about BLPs, nothing WP:LIKELY, etc.). Is that "ready for the mainspace" in your opinion? WhatamIdoing (talk) 23:51, 2 June 2024 (UTC)

In my opinion, that article isn't 'ready for mainspace' because it is unreferenced. Cremastra (talk) 00:51, 3 June 2024 (UTC)

I realize that the notability guideline itself says that the sources just have to exist somewhere, and not be actually present in the article. However, it's pretty clear that the reason the notability guideline says the sources have to exist somewhere is so they can be used to write the article.

My big problem with the example article you linked is that it's not clear that "Christmas candy" is a notable subject separate from specific types of Christmas candy. I also think some of the list of examples is more WP:LIKELY to be challenged than you think. I think that for instance someone who did not know what a szaloncukor was is very likely to start out doubtful that it is Christmas candy. Loki (talk) 02:33, 3 June 2024 (UTC)

Do you really think I needed to consult sources to write that "Christmas candy is candy associated with the Christmas holiday season. Candy canes are one type of Christmas candy"? WhatamIdoing (talk) 05:25, 3 June 2024 (UTC)

No, but someone who doesn't celebrate Christmas and has lived in a Hindu/Buddhist/Muslim-majority country all their life might need to. WP:V still stands, whether you like or not. Cremastra (talk) 12:05, 3 June 2024 (UTC)

WP:V says that it must be possible to find sources (e.g., at a library). It does not say that sources must be cited in the article, except four types of material, none of which are in this article. WP:V is not violated by having those two sentences uncited. WhatamIdoing (talk) 19:31, 3 June 2024 (UTC)

I would guess the editor means:

• The article is completely unreferenced, and/or many of the claims are factually dubious
• The article is written in English, but is barely coherent. It can be understood, so isn't gibberish, but is an embarrassment and not very helpful.
• The article is blatantly and overtly promotional

Cremastra (talk) 20:30, 1 June 2024 (UTC)

I could also interpret "not ready for mainspace" to include glaring MOS or technical issues, like:

• templates outputting nothing but error messages
• external links peppering article prose
• infobox with default values for parameters
• entirely empty sections
• no subheadings whatsoever, just a giant chunk of text
• unintentional blockquotes from starting a paragraph with whitespace
• other Wikipedia pages incorrectly formatted as references instead of internal links
• etc

Folly Mox (talk) 14:22, 2 June 2024 (UTC)

First, to emphasize the obvious, "ready for mainspace" is a vague subjective term. Probably the only more objective term that could fall under that is "allowed to exist in mainspace" and the most universal standard for that is "likely to survive a reasonably well run AFD". And for an article (NOT article content) NPP and AFC passage ostensibly follow that. Which in turn (presuming no eggregious speedy or wp:not violations) the main criteria ends up being passing wp:notability. Many people (e.g. at AFC, during mentoring, and in this thread) set a higher standard for "ready for mainspace" which is that the content of the article and the article does not have any significant problems or shortcomings. Yes, this is a double standard, and can make AFC a somewhat rough and arbitrary path. But we need to recognize that it is only human by the person reviewing it. If somebody took an article to you that was allowed to exist in mainspace (usually a wp:notability decision on the topic) but which was in really bad or undeveloped shape, would you be willing to bless putting it into mainspace? Most people would want it to meet a higher quality standard before they would personally say "ready for mainspace". North8000 (talk) 19:21, 31 May 2024 (UTC)

• If I'm not mistaken the "not ready for mainspace" phrase originated in WP:DRAFTIFY and has since leaked into deletion discussions. As everyone here seems to agree, it is very poorly defined phrase and, far from the low bars proposed above, I've seen new page and AfC reviewers invoke it for things like a draft not being long enough or using plain text references instead of {{cite}} templates. Rather than trying to define it, I think we should purge it from guidelines and templates in favour of listing specific problems in an understandable way. – Joe (talk) 12:35, 4 June 2024 (UTC)

  Rather than trying to define it, I think we should purge it from guidelines and templates in favour of listing specific problems in an understandable way. I agree with this. U ideally we would not move something out of mainspace or disallow moving it into mainspace unless there are problems that are all of specifically identified, actionable, adversely detrimental* and not trivially fixable (anything that is trivially fixable should just be fixed). *"adversely detrimental" means things like failed verification or no evidence of notability, not merely lacking inline sources, cite templates or being "too short". Thryduulf (talk) 12:44, 4 June 2024 (UTC)

  I suppose we could try to re-define it as "does not qualify for deletion" (either CSD or AFD), but (a) it'd take a couple years for the usage to shift and (b) there is a strong demand from a minority of the community to have ways to get rid of "ugly" (i.e., short) articles. WhatamIdoing (talk) 16:42, 4 June 2024 (UTC)

  Following up on what @Joe Roe said about DRAFTIFY, I find this in that page:

  2a. The page is obviously unready for mainspace, for example:

  2a-i. is not a reasonable WP:STUB (e.g. has very little verifiable information, or is interchangeable with a short dictionary entry, but the definition is not good);

  2a-ii. or it would have very little chance of survival at AfD;

  2a-iii. or it meets any speedy deletion criterion.

  This was introduced by SmokeyJoe as a result of his proposal at Wikipedia talk:New pages patrol/Archive 5#Clarification and guidance for draftification. (The original proposal was that "unready for mainspace" mean "It does not meet WP:STUB.")

  This suggests that the definition of "not ready for the mainspace" is:

  • a very short stub, containing either a bad dictionary definition or very little information in general;
  • the article is not ready because the subject is non-notable; or
  • the article qualifies for speedy deletion.

  Based on this, I suspect that the definition could be reduced to "contains less than about 20 words of encyclopedic content", because a look at Wiktionary suggests that the mode for dictionary definition length is a mere four words, and 20 words would give you one long sentence or several shorter ones. WhatamIdoing (talk) 01:45, 9 June 2024 (UTC)

  A stub should be defined at WP:STUB, not at WP:Drafts.

  A stub is a very short article that is accepted in mainspace, despite not meeting other inclusion guildelines. They seem to be inherently acceptable topics, like natural species, capable of expansion. SmokeyJoe (talk) 02:05, 9 June 2024 (UTC)

  Per WP:STUB, a stub is any short article. Generally, it is taken to be less than about 250 words/10 sentences. There are no minimum requirements in WP:STUB. Cancer is a disease – a mere four words with no sources and no other content – would be a valid stub per WP:STUB.

  WP:IDEALSTUB (perhaps that's what you had in mind?) recommends adding "enough information for other editors to expand upon it" and to avoid a {{db-nocontext}} deletion. Cancer is a disease is realistically enough to fulfill that recommendation.

  IDEALSTUB also recommends that you "try to expand upon this basic definition", so we could add something like Sometimes people die from it or It is mostly treated with surgery or drugs.

  Finally, IDEALSTUB recommends citing a source (though our policies only require this for BLPs, not for articles about diseases), so we could add a link to https://www.cdc.gov/cancer/ or some similar website.

  I don't know what you mean by "despite not meeting other inclusion guildelines". The inclusion guidelines are at Wikipedia:Notability and its friends, and none of them require any length or particular content in the articles. Cancer is a disease, unsourced, with nothing else, meets the inclusion guidelines. WhatamIdoing (talk) 04:24, 9 June 2024 (UTC)

  You’re referring to the sourcing requirement speaking to sources that exist, not sources currently listed. Ok, yes you are right. SmokeyJoe (talk) 09:56, 9 June 2024 (UTC)

  @SmokeyJoe, I wonder whether this list of "three" items could be shortened to two:

  • The subject is non-notable (in which case, you should usually send it to AFD instead of Draft:)
  • The article qualifies for speedy deletion (on any grounds, but particularly for {{db-nocontext}}).

  The example of "has very little verifiable information, or is interchangeable with a short dictionary entry, but the definition is not good)" is redundant with {{db-nocontext}}. But perhaps there is a different example of "not a reasonable WP:STUB" that should be retained? WhatamIdoing (talk) 15:30, 9 June 2024 (UTC)

  Three is a pleasing number.

  Lists of two encourage binary thinking. SmokeyJoe (talk) 00:17, 10 June 2024 (UTC)

  Sure, a great many things could be.

  I wonder whether it’s actually not a good thing to attempt to tidy up definitions of edge cases. Edge cases are messy, subjective, and cause emotional disputes. Mistakenly precise language can make this worse, setting up a conflict between rules oriented wikilawyers and new content creatives.

  Where are the actual problems that you are trying to solve? SmokeyJoe (talk) 00:08, 10 June 2024 (UTC)

  The problem I need to solve is: People use the same words to mean different things, which results in (preventable) confusion.

  The problem I want to solve is: People have significantly different ideas of what the minimum acceptable amount/type of content for an article is, which results in some preventable disputes (and some non-preventable disputes).

  For example, one editor looks at an article and says "Wow, ten sentences, nicely written, I understand what the subject is, and it's even kind of a cool subject. It's WP:NOTFINISHED, but readers will be happy if they run across it, especially if they only need basic information (which is usually the case)."

  Another editor looks at the same article and says "It's soooo embarrassing! WP:ITSUNREFERENCED so the whole thing might be made-up nonsense, and readers hate uncited articles. There's been WP:NOEFFORT to improve it. WP:WEDONTNEEDIT, and we do need to hide that WP:Garbage to protect our reputation. There's no chance of it getting deleted at AFD, but it's obviously not ready for the mainspace!"

  Some divergence is a desirable thing, but there's very little overlap between those two positions. If we're going to function well, we need to have most of us mostly agree on what the minimum requirements are for something being "ready for the mainspace".

  If "ready for the mainspace" is even a soft requirement, then we need to have a shared understanding of what that means, and it needs to be the same for both going into and getting back out of the Draft: space. WhatamIdoing (talk) 05:16, 10 June 2024 (UTC)

I agree with WAID that there is a real problem here. If you read the subpoints of WP:DRAFTIFY#During new page review, then it's clear that obviously unready for mainspace is intended to refer to a fairly narrow set of seriously problematic articles: something less than a stub, deletion almost certain, etc. But divorced from that context, "not ready for mainspace" admits a much wider range of understandings, as we've seen above. For example, the draftify script leaves the canned edit summary Not ready for mainspace, incubate in draftspace followed by a selection of prespecified reasons why the article is not ready, which include things like it needs more sources to establish notability and it has too many problems of language or grammar – a far cry from very little chance of survival at AfD. A similar message is given to the creator the explain what happened to the article. If you look at the logs, the vast majority of moves to draft use one of these canned reasons: people take their cues on what they should and shouldn't do from the UI in front of them, not the guideline. Taken out of the guideline and into scripts and other pages, the phrase "not ready for mainspace" itself has taken on a life of its own and is used to systematically circumvent the deletion policy on a daily basis. – Joe (talk) 08:12, 10 June 2024 (UTC)

Who are these draftifiers? Are they sufficiently qualified/experienced? My biggest concern about NPR approvals was insufficient expectation of experience at AfD, draftifiers are t performing AfD-like decisions, but unilaterally.

The wording of the script, was there any discussion or consensus behind it.

“Not ready for mainspace”. They are very simple word. I think it might be worth an essay, WP:Not ready for mainspace.

While trying not to embarrass individuals, is it possible to show me a list of bad draftifications? SmokeyJoe (talk) 12:37, 10 June 2024 (UTC)

I wouldn't know where to start. I've been reviewing draftifications through WP:PERM/NPP requests, CSD R2 nominations and from the logs for years now and I'd say I come across an egregious example just about every time I look. To be clear, by 'bad', I mean something that clearly exceeds the boundaries set by WP:DRAFTIFY and/or what I understand community expectations to be, not my own. I don't want to unfairly single anyone out, but you could check my contributions to the draft namespace for a representative sample. – Joe (talk) 13:44, 10 June 2024 (UTC)

Egregious?

Some brainstorming questions, not having looked at your contribution history yet:

Are reviewers systematically applying a higher standard to new articles than would be applied at AfD? I’ve seen that AfC reviewers do this. Could this be explained by an increased expectation of article standard, and AfD voting lagging this change? I know that some people complain about how hard it is to get article deleted at AfD.

Are bad draftifications being done by editors who are not NPRs? And are they doing bad things randomly?

Is there any sense that draftifications are being done to endorse a reviewers POV bias on what content should be in Wikipedia?

Is the problem with this page’s asserted boundaries, or with poor training of NPReviewers? Or with bias from the draftification script(s) due to them proving an easily option for difficult cases? I don’t think that anything in the fairly heavy NPR and AfC training pages instructions to read WP:Drafts. SmokeyJoe (talk) 13:59, 10 June 2024 (UTC)

Egregious meaning systematic enough that I feel I have to discuss it with the user and, if they don't stop, pull their NPR right. I think the rest of your questions are good ones and, although I give my anecdotal conclusion on them, I don't have any data to hand (and unfortunately I don't think anyone does, which is why this issue has been festering for years now). I do think the lack of clarity in the phrase "not ready for mainspace"—taken out of context, as discussed above—has contributed to the problem and that's why I think WAID's original question (what is ready for mainspace?) is a good one.

To suggest a concrete next step, there is a list of specific, consensus-backed things that make a page "not ready for mainspace" at WP:DRAFTYES. We could brainstorm what could be added to those, and/or consider making a separate list of things that don't disqualify a page from mainspace. – Joe (talk) 14:13, 10 June 2024 (UTC)

My impression:

• Are reviewers systematically applying a higher standard to new articles than would be applied at AfD? – Yes. And it's not just one "rogue" editor; it's anyone who doesn't want to be seen "endorsing" or "accepting" an WP:UGLY article. You have to be willing to expend a lot of social capital to follow the written rules. This is one reason I recently suggested a "three strikes and you're out" approach to AFC: On the third time AFC rejects a submission, a bot should do a procedural nomination at AFD. There is no more reliable method of figuring out whether it will be kept at AFD than to send it to AFD.
• Could this be explained by an increased expectation of article standard, and AfD voting lagging this change? – Yes, but I don't think that "increased" is the right word. AFD still operates on the written rules (e.g., sources must WP:NEXIST in the real world, but don't have to be cited in the article). NPP and AFC functionally reject this rule and want notability "demonstrated".
• Are bad draftifications being done by editors who are not NPRs? And are they doing bad things randomly? – Yes, overly aggressive draftifications sometimes are done by anyone who believes they are defending Wikipedia against ugly articles, but it's not really random. It is an effort to "raise Wikipedia's quality" by forcing other editors to choose between improving the article or having it hidden from readers.
• Is there any sense that draftifications are being done to endorse a reviewers POV bias on what content should be in Wikipedia? – I have not seen evidence of, e.g., editors draftifying articles related to geopolitical disputes. There have been times in which we see editors draftifying articles about, e.g., Bollywood actors or African politicians. This could be due to cultural differences (the normal, everyday ways of describing powerful people in some cultures looks like "pure promotional garbage!" in others) and is probably often due to WP:NEVERHEARDOFIT (with that bias applying both the subject and to the newspapers/standard sources in that country).
• Is the problem with this page’s asserted boundaries, or with poor training of NPReviewers? Or with bias from the draftification script(s) due to them proving an easily option for difficult cases? – I don't think that training is the problem, because part of Wikipedia's notion of "training" is to watch what others are doing and follow their lead. The problem that I want to deal with is the problem of nobody knowing/agreeing on what those words mean. If we agree that ugly articles should be accepted, then the script should reinforce that. If we agree that ugly articles should be hidden, then the script should reinforce that (and WP:UGLY should be updated to say that ugly articles can be hidden in draftspace).

Joe, I like your idea of having "a separate list of things that don't disqualify a page from mainspace". WhatamIdoing (talk) 17:44, 10 June 2024 (UTC)

Me too. SmokeyJoe (talk) 11:44, 11 June 2024 (UTC)

• User:WhatamIdoing, User:Joe Roe, how about “isn’t acceptable in mainspace”? Eg1. Eg2. SmokeyJoe (talk) 13:23, 12 June 2024 (UTC)

  If two people argue about whether something is ready, it does seem to be a horribly subjective argument over an undefined and unimportant threshold.

  If two people argue about whether something is acceptable, one can say “it is acceptable because I accept it” and the other can say “it is not acceptable because I am not accepting it”. It goes to AfD where the decision will be made, deleted or pseudodeleted, or kept in mainspace, proving one of the two to be right. SmokeyJoe (talk) 13:34, 12 June 2024 (UTC)

  If two uninvolved editors disagree in good faith about whether something is or isn't ready for mainspace, I think it should be declared ready and moved to mainspace but explicitly without prejudice to AfD (obviously nothing is immune from AfD, but it should be made explicit so the psychological bar to nomination is lower). Thryduulf (talk) 13:46, 12 June 2024 (UTC)

  If two uninvolved editors disagree in good faith about whether something is or isn't acceptable for mainspace, I think it should be declared acceptable and moved to mainspace but explicitly without prejudice to AfD (obviously nothing is immune from AfD, but it should be made explicit so the psychological bar to nomination is lower). SmokeyJoe (talk) 13:52, 12 June 2024 (UTC)

  I think any new wording is worth a try but I'm not sure that it is sufficiently different from "ready for mainspace" to make a difference. We don't generally encounter problems when editors disagree about drafts, because the obvious course of action then is to discuss it at AfD. The problem is that the vast majority of articles moved to draftspace are only seen by two people: the creator, and the reviewer who draftifies it. Reviewers shouldn't, but unfortunately often do (not least through the wording of the automated script), imply that their 'decision' on an article is uncontestable. Even if they don't, creators, especially inexperienced ones, are often ignorant of the fact that they don't have to go along with what the reviewer says. So unless a third party happens to come across the draft, we don't get disagreement, just a creator trying to meet whatever arbitrary standard a particular reviewer has decided is required for mainspace, or just concluding that their contribution has been rejected and giving up. This is incidentally the precise opposite of what WP:DRAFTIFY and most frequent draftifiers say they want to achieve: to "allow time and space for the draft's improvement". – Joe (talk) 14:01, 12 June 2024 (UTC)

  I think that “acceptable” is better defined, via tautology with hindsight, than “not ready”. On reflection, I think “not ready” is suggestive that it is ok to Draftify a topic that is undoubtedly suitable, cf meta:Conflicting Wikipedia philosophies and extreme Immediatism.

  I think that draftification should include a mandatory link to WP:DRAFTOBJECT, both in the edit summary / move log entry, and in the message posted to the author. I agree with you concern about content creators not knowing all of the rules. SmokeyJoe (talk) 14:41, 12 June 2024 (UTC)

  I don't think that wording helps much. For one thing, it implies that there is a consensus that some articles are "unacceptable", but gives nobody any idea what is "acceptable" and what is "unacceptable". WhatamIdoing (talk) 14:40, 12 June 2024 (UTC)

  Acceptable means it survives AfD. Unacceptable means it doesn’t survive AfD. SmokeyJoe (talk) 14:42, 12 June 2024 (UTC)

  So in the deletion policy, you wrote "If an article isn't acceptable in mainspace, it can be moved to the draft namespace" but what you mean is "If a subject isn't notable, it can be moved to the draft namespace"? WhatamIdoing (talk) 14:50, 12 June 2024 (UTC)

  I’ve been trying to remember what I meant seven years ago with “ready for mainspace”. One thing that comes up was that reviewers were expecting citations in a BLP to be “inline”. I think I was attempting to not engage with dubious reasons reviewers were using. This was then quite a new backwater page. I am disturbed to discover my verbiage to have been copied into deletion policy and to have become common phraseology at AfD.

  WP:Drafts is not supposed to rewrite WP:N. Pages that pass WP:N are sometimes deleted. Pages that fail WP:N are sometimes kept.

  When a reviewer moves a page to draftspace, the reviewer should be justifying their action, not quoting generic statements. SmokeyJoe (talk) 21:33, 12 June 2024 (UTC)

  I think the wording should be more direct regarding expectations. Taking some text from Wikipedia:Deletion policy § Incubation, it could be something like "If a recently created article shows potential but needs additional development to establish that the subject meets Wikipedia's standards for having an article, it can be moved to the draft namespace." isaacl (talk) 15:59, 12 June 2024 (UTC)

  Is the rule you have in mind that the subject must not only be notable (e.g., NEXIST) but also demonstrate notability (e.g., cite multiple sources)? WhatamIdoing (talk) 16:26, 12 June 2024 (UTC)

  Whatever is necessary to convince a consensus of editors that the subject meets English Wikipedia's standards for having an article. This doesn't necessarily require citing multiple sources appropriate for demonstrating that the standards have been met, though that would be an easier route. isaacl (talk) 18:14, 12 June 2024 (UTC)

  Does the English Wikipedia have any standards other than WP:N for having an article? WhatamIdoing (talk) 19:08, 12 June 2024 (UTC)

  Having an article, no. Having this article, yes - e.g. WP:V, WP:COPYVIO, WP:NOTENGLISH, WP:BLP, WP:G10, etc. Thryduulf (talk) 19:36, 12 June 2024 (UTC)

  WP:V is not in the business of deciding whether we should have an article, in any mainspace. It's about which discrete bits of material require inline citations. Uncited material (WP:ITSUNREFERENCED) is not grounds for deletion under WP:V.

  If the article violates BLP, then it can't be moved to the Draft: space, either, because BLP applies to all namespaces. Ditto for COPYVIO and G10, which are reasons for immediate deletion and apply to all namespaces.

  NOTENGLISH has a two-week timer for deletion. It also says "Please keep in mind that drafts are out of scope for this page." WhatamIdoing (talk) 19:55, 12 June 2024 (UTC)

  You are missing the distinction between whether there should be an article about a topic, and whether any given individual content written about that topic is appropriate. For example, Australia is a notable topic about which we should have an article, but an article reading "Australia is land of criminals and man-eating spiders that is permanently on fire. Citation: My ex-girlfriend" should not be in mainspace. Thryduulf (talk) 20:11, 12 June 2024 (UTC)

  I'm pretty sure that the hypothetical Australia example would qualify for Wikipedia:Criteria for speedy deletion#G3. Pure vandalism and blatant hoaxes. WhatamIdoing (talk) 00:44, 13 June 2024 (UTC)

  Possibly (it's definitely not a hoax, vandalism is debatable. Unarguably it would fail WP:V) however these are all matters that have absolutely nothing to do with notability and are relevant to whether a page should or should not be in the mainspace. Thryduulf (talk) 01:08, 13 June 2024 (UTC)

  Prcicely right. There are a lot of topics that pass GNG or ann SNG, and thus deserve to have AN article. But, that does not mean any specific attempt at creating that article is acceptable. That attempt may have serious issues with other policies and guidelines, and need a complete rewrite. Draftspace is a temporary holding pen where that rewrite can take place. Blueboar (talk) 20:05, 12 June 2024 (UTC)

  Yes, as you are aware, there are other considerations than those explicitly listed at Wikipedia:Notability. That page does link in its introduction to one of the other key guidance pages to consider, Wikipedia:What Wikipedia is not. I didn't get into it as I find it hard to discuss the entirety of Wikipedia guidance related to having an article at once, and so I linked to the most commonly referenced guidance page in this area. (It wasn't a final proposal for a different wording, just a starting point.) I appreciate you like to use Socratic questioning, but it feels like you're trying to elicit a response that you can counter with your knowledge of current guidance, rather than bringing up additional guidance to consider.. isaacl (talk) 22:58, 12 June 2024 (UTC)

• Perhaps some of the confusion here stems from the fact that deletion is based purely on notability, while draftification can be based on other criteria. Yet, “not ready for Mainspace” gets invoked in edit summaries as an “explanation” for both actions. Blueboar (talk) 00:39, 13 June 2024 (UTC)

  Yes, plus there seems to be no agreement about what the "other criteria" are. WhatamIdoing (talk) 00:46, 13 June 2024 (UTC)

  User:WhatamIdoing, I’ve noticed that “unsourced” is a reason. SmokeyJoe (talk) 11:18, 15 June 2024 (UTC)

  The "unsourced" claims are probably in WP:PGCONFLICT with WP:NEXIST.

  I wonder whether the typical claim is actually "unsourced" (e.g., if it were a BLP, it'd qualify for WP:BLPPROD) or if the claim is closer to "does not contain a sufficient volume of sources that, in my opinion, clearly demonstrate notability". WhatamIdoing (talk) 14:04, 15 June 2024 (UTC)

  My opinion is that in new page review, if the page is unsourced, the reviewer should make some attempt to find sources, and if they fail to find sources, they should PROD the article (or BLPPROD) noting that they cannot find evidence of sources, and they should not Draftify, becuase unsourced content is dubious content that should be considered junk.

  At AfC, submitted drafts are routinely declined as unsourced. SmokeyJoe (talk) 00:18, 16 June 2024 (UTC)

  Do you really believe that "unsourced content is dubious content that should be considered junk"? If someone starts an article on French Renaissance gardens, and it says "French Renaissance gardens were the style of gardens in France during the Renaissance", do you actually think that's worth a {{dubious}} tag? Is it WP:JUNK?

  I can imagine it being irritating for those few people who want a Wikipedia:Four Award, but that article wouldn't violate a single policy. WhatamIdoing (talk) 16:57, 16 June 2024 (UTC)

  User:WhatamIdoing, I wrote with a few unstated assumptions. I’m assuming a new article, it is completely unsourced, no external links, a single author who is no longer active, and a new page patrollers has tried to find sources, at least by google search. The content is unverified and possibly unverifiable.

  French Renaissance gardens is the sort of article I’m imagining. Doubtless it exists, there were gardens in France during the renaissance, and it seems likely that they had a unique style. The information in the page may be true, but may just as likely be made up, embellished, oversimplified, etc. I consider this dangerous, through the process of citogenesis. Is the risk managed by draftififcation? SmokeyJoe (talk) 11:22, 18 June 2024 (UTC)

  If I found a new article or draft “French Renaissance gardens” containing unverifiable content, I would redirect it to Gardens of the French Renaissance. SmokeyJoe (talk) 11:40, 18 June 2024 (UTC)

  Template:Dubious would be not applicable. That template refers to “a specific statement or alleged fact that is sourced but that nevertheless seems dubious or unlikely. The unsourced article more likely contains BLUESKY plausible stuff.

  The WP:JUNK essay is about notability. I am talking about pages that are unverifiable. I don’t agree with that essay defining junk as stuff that fails Wikipedia-Notability. SmokeyJoe (talk) 11:28, 18 June 2024 (UTC)

  If it's actually unverifiable (which, as you can see from the books cited in Gardens of the French Renaissance, this is definitely not unverifiable), then the material would have to be removed. All material must be verifiable – that means that it must be possible for someone to check whether a reliable source says the same thing, with "possible" defined as including actions such as "getting help from a reference librarian at your own library" or "finding sources through Wikipedia:The Wikipedia Library" – though a sentence such as the one I gave does not require an inline citation according to WP:V. For WP:V purposes, it does not require a source even if it is the only sentence in the entire article.

  It is not my experience that uncited content is "just as likely be made up, embellished, oversimplified". About half of all sentences in the English Wikipedia are uncited; in my experience, it is not true that half of them (representing a quarter of our content) is made up, embellished, oversimplified, etc. I generally find that only a small proportion of our uncited content is wrong. My impression is that the proportion of wrong-and-uncited content is not as different from the proportion of wrong-and-cited content as one might wish.

  I have given you an example of a definitely verifiable (though presently uncited) sentence about a definitely notable subject. Do we agree that "unsourced content" is not necessarily "dubious content that should be considered junk"? WhatamIdoing (talk) 15:38, 18 June 2024 (UTC)

  We agree, "unsourced content" is not necessarily "dubious content that should be considered junk". SmokeyJoe (talk) 09:54, 19 June 2024 (UTC)

  Another reason if seen for non notability reasons to Draftify is “COI”. Including “suspected COI”. SmokeyJoe (talk) 00:37, 16 June 2024 (UTC)

  COI has no effect on non-notability. If it's non-notable, it should go to AFD, regardless of whether COI is suspected. WhatamIdoing (talk) 15:39, 18 June 2024 (UTC)

Contemplation of a Proposal: Mandate edit summary linking to WP:DRAFTOBJECT in every unilateral draftification

Proposal: Mandate edit summary linking to WP:DRAFTOBJECT in every unilateral draftification.

The more I think on this years old idea the more I think it should be done. In practical terms, it is a simple thing to write into draftification scripts. For manual draftifications, these draftifiers are probably not experience and the rule is even more important. For consensus based draftifications, via AfD or informal discussion, they should link the discussion.

I suspect the rule should also strongly encourage including WP:DRAFTOBJECT in the usertalk explanation (automatic by the scripts), but not mandated due to occasional complications such as the first page author being an IP or banned user. SmokeyJoe (talk) 00:33, 16 June 2024 (UTC)

I think we could realistically make this happen in the scripts, but not in manual edit summaries. WhatamIdoing (talk) 16:52, 16 June 2024 (UTC)

I agree, but but would it be a good idea, to make it happen in the scripts, and to encourage it in manual edit summaries? It seems to me to be an easy fix to some of the problems you’ve noted (eg newcomers being intimidated). Would it have downsides? It would not fix everything. Would you support this proposal? SmokeyJoe (talk) 11:37, 18 June 2024 (UTC)

This is the wrong place for proposals, and I would oppose it anyway, as mandating a link to an essay is a bad idea, as it gives the impression that DRAFTOBJECT is a policy without going through the policy validation process. E.g. the "you can't draftify again" part is being misused by some people to object to redraftification a priori, pretending that it isn't allowed. Often the same people who then object to an AfD because AfD is not cleanup, leaving not much room for other options to deal with very poor articles which, yes, aren't ready for the mainspace. Yes, the drafter could in theory do a complete cleanup of the article, providing coherent prose, sources, ... for a subject they know nothing about, where the sources are in a language they don't speak. Realistically speaking though, the best solution is to move the page to draft again and again until the creator or someone else with the time and knowledge to deal with it turns it into an acceptable article. Fram (talk) 10:03, 19 June 2024 (UTC)

Hi Fram.

Wrong place? Yes, I know, actually I meant it as contemplation for formally proposing this. I have learned to not propose something without at least one person agreeing with me. If supported, I would start a new page tagged {{Proposal}}. SmokeyJoe (talk) 11:26, 19 June 2024 (UTC)

DRAFTOBJECT *is*, already, pseudopolicy, and the proposal would be defacto ratification.

Redraftification, excepting for WP:COI, is not allowed. It is move warring. Two people disagreeing should not move war, but should discuss, and the perfect forum is AfD.

AfD is not cleanup? No, it is not. Neither is draftification. Draftification is not for cleanup.

What do you mean by a very poor article that is not “ready for mainspace”, to ask the central question of this thread?

If there is any disagreement, it should go to AfD. I firmly disagree with you if you if you think it is ok for one editor to have the authority over another to send their work to draftspace until it meets the first editors undefined standard. —SmokeyJoe (talk) 11:37, 19 June 2024 (UTC)

"Pseudopolicy" doesn't exist. Draftofject isn't only about "one editor to have the authority over another to send their work to draftspace until it meets the first editors undefined standard.", it also means that a second editor may not send a page back to draftspace. And the essay gives the right to "one editor to have the authority over other editors to send their own work to main space", no matter how poor. I see no problem with this nor a reason to burden AfD with it. I moved Draft:2025 Rugby Europe Championship to draft twice because it had no sources about the topic but about different topics, but if it could be sourced to good sources it would be a notable subject. An article with such poor sourcing is "not ready for the mainspace". Jesus Calls was draftified, recreated, speedy deleted, and then recreated as Jesus calls. I draftified that one, is that a redractification? And if so, is it for some reason problematic? I redraftified Air 1 (airline) (another editor did the original draftification), why not? Same for Mangkunegara III.

As for "not ready for mainspace", things like Draft:Sahajanya (unsourced microstub), Draft:Azad Samaj Party (Kanshi Ram), Uttar Pradesh and Draft:Azad Samaj Party (Kanshi Ram), Madhya Pradesh (barely above speedy deletion), Draft:2025 in Belgium (explanation for the "not ready" statement: "So far, this contains 1 sure event only, plus speculative claims about who will be PM, links to unrelated articles, and the holidays for 2024."), Draft:San Sanana (not ready as in "No evidence of notability at the moment, chart performance section is not for this song" but being an Indian song not easy for me or many others to check for actual notability), ... Fram (talk) 13:09, 19 June 2024 (UTC)

Taking Draft:2025 in Belgium as the example: Why did you decide that we shouldn't have that page in the mainspace?

There is no question about the notability. It would easily pass AFD, and AFC's mandate is to accept pages that will pass AFD. If Thief-River-Faller submits it to AFC, they ought to accept it immediately (assuming they follow their own rules, about which there has been some doubt).

It would also be more likely to get corrected if it were in the mainspace. So why hide it in Draft: space? WhatamIdoing (talk) 17:05, 19 June 2024 (UTC)

Whoever would accept that page as is would need their reviewer rights removed. "AFC's mandate is to accept pages that will pass AFD." Among many other things. Accepting pages with almost exclusively blatantly incorrect information just because the topic is notable is making Wikipedia worse, not better. The page at the moment has one correct entry, "7 – 17 August: Belgium at the 2025 World Games"; everything else is either speculation or just factually wrong for the topic. I would urge you to reread Wikipedia:WikiProject Articles for creation/Reviewing instructions, which contains a lot more than "notable = accept". Fram (talk) 21:28, 19 June 2024 (UTC)

Sure, I've read that page. I even helped write it. In particular, see Wikipedia:WikiProject Articles for creation/Reviewing instructions#Core purpose, which says (second sentence): "Articles that will probably survive a listing at Wikipedia:Articles for deletion should be accepted."

See also the second sentence of the next section: "Article submissions that are likely to survive an AfD nomination should be accepted and moved to mainspace" (bold in the original). WhatamIdoing (talk) 21:57, 19 June 2024 (UTC)

All the while ignoring the detailed checklist and workflow, which give further, more detailed instructions than the (by definition) simplified summary. I see no good arguments why this page should be in the mainspace as it is now, and putting it in the mainspace while knowing about the issues (which is what you claim a reviewer should do now, if asked by the creator) is basically vandalism, deliberately and knowingly putting incorrect information in the mainspace. Fram (talk) 07:43, 20 June 2024 (UTC)

I believe it would be best to blank the obviously wrong information. WhatamIdoing (talk) 20:24, 20 June 2024 (UTC)

Taking Draft:San Sanana as an example, why did you choose to hide it in Draft: space instead of blanking the apparently incorrect information (KjjjKjjj, that song isn't "Falling Behind", like it says in Draft:San Sanana#Charts, right?) and tagging it with {{notability}}? WhatamIdoing (talk) 18:14, 19 June 2024 (UTC)

So, if it is notable, I should keep it in mainspace. If it is not notable, I should nominate it for deletion. And if notability is unclear I should tag it with "notability", even if there are (like here) clearly other problems as well. Is there any scenario where you believe draftifications is an acceptable course? Fram (talk) 21:30, 19 June 2024 (UTC)

Yes, there are a couple of scenarios for which I would accept and even recommend draftification. The first and most obvious is when the editor(s) working on it want to take that route. The second obvious case IMO is when the subject is not currently notable but is reasonably likely to become notable within the next couple of months. For example, we know that certain events, such as the US State of the Union speech or the United States census, will continue to happen on a predictable schedule, but future events frequently fail Wikipedia:Notability (events) until shortly before they happen. If an article is created a bit too early, when we don't have enough sources/attention from the world at large, but when we also believe such sources will be forthcoming in short order (e.g., a press conference has been scheduled for an announcement), then I think draftification is better than either deletion or keeping it. WhatamIdoing (talk) 22:15, 19 June 2024 (UTC)

I don't think that extremely minimal approach to draftification is shared by many, and would leave many very poor new articles in the mainspace. Something like Draft:Science Centre, Patan has now been draftified twice, which is a good thing. Would the subject survive an AfD? No idea, and as discussed elsewhere, it isn't the job of reviewers to do a WP:BEFORE check. Fram (talk) 07:43, 20 June 2024 (UTC)

I think your extremely maximal approach to draftification, openly ignoring the only written guidance on the subject because it's "just an essay", is shared by even fewer. – Joe (talk) 11:16, 20 June 2024 (UTC)

See below, we actually have policy about this, which is what I follow. Fram (talk) 11:43, 20 June 2024 (UTC)

the proposal would be defacto ratification if in effect the idea is to make DRAFTOBJECT policy then that should be the proposal, rather than discussing edit summaries. -- LCU ActivelyDisinterested «@» °∆t° 19:13, 19 June 2024 (UTC)

Something doesn't have to be a policy (or even a guideline) to be linked in a tool-generated edit summary. WhatamIdoing (talk) 21:51, 19 June 2024 (UTC)

...which should be obvious given that the current edit summary used by scripts is a paraphrase of the very same essay (Not ready for mainspace, incubate in draftspace). Apparently it's okay to use a non-policy to justify moving tens or hundreds of thousands of articles out of mainspace, but not to remind the creators that they're entitled to move it back? – Joe (talk) 11:21, 20 June 2024 (UTC)

My point had nothing to do with adding anything to the summary of automated edits. It was that if editors wanted to make suggestions for new policy they should do so. The comment I was replying to was suggesting that policy should be made via discussion on another topic. -- LCU ActivelyDisinterested «@» °∆t° 14:35, 20 June 2024 (UTC)

I don't use that edit summary either, and if you both want a policy about draftification; WP:ATD-I: Recently created articles that have potential, but do not yet meet Wikipedia's quality standards, may be moved to the draft namespace ("draftified") for improvement. Fram (talk) 11:43, 20 June 2024 (UTC)

Where are those quality standards defined? Thryduulf (talk) 11:45, 20 June 2024 (UTC)

Nowhere probably, just like many things around here. Having reliable sources and intelligible prose, being factually correct, and actually being about the topic as suggested by the title, is what I (and in my experience most others who do new page checking) apply. I don't think any of these can be considered really controversial. Fram (talk) 12:01, 20 June 2024 (UTC)

This discussion demonstrates that every one of those is controversial:

• The only requirement for sources is that they exist - if they don't exist the article should be prodded or sent to AfD, if they do exist add them to the article, if you don't know then look.
• If there is no intelligible prose then the article should be deleted (speedy deletion criteria G1, G2, A1, A2 and/or A3 almost certainly apply)
• If the article is factually incorrect then it should be corrected or nominated at prod or AfD (unless it's a blatant hoax, in which case it should be speedily deleted under criterion G3).
• If it isn't about the topic as suggested by the title, then either rename the article or nominate it for prod/AfD. Thryduulf (talk) 12:14, 20 June 2024 (UTC)

Nobody would disagree that those are desirable qualities in an article, but whether they constitute a required standard for mainspace is indeed controversial. As just one data point, the existence of {{unreliable sources}}, {{incomprehensible}}, {{disputed}}, and {{off topic}} would suggest that all of the problems you list have been tolerated in mainspace in the past. The lack of a definition of "Wikipedia's quality standards"—AKA being "ready for mainspace"—is the problem that motivated this discussion and, as the discussion shows, it leaves room for a wide range of understandings. – Joe (talk) 12:17, 20 June 2024 (UTC)

Feel free to ping me when a policy proposal is put up for a vote. Until then, I don't think anything useful will come from continuing this discussion with you three. The requirements put up here, basically requesting the reviewers needing to do all the work the creator should have done and can do much easier, are mainly based on misreadings of policy ("The only requirement for sources is that they exist", well, no: "Any material lacking an inline citation to a reliable source that directly supports[b] the material may be removed and should not be restored without an inline citation to a reliable source. " and "The burden to demonstrate verifiability lies with the editor who adds or restores material" (bold in original). A completely unsourced article may thus be blanked (which would make it eligible for speedy deletion, not really preferable), or by draftifying it "removed" from the mainspace; and the burden to add sources lies with whoever wants it back in the mainspace, the one who "restores material". Putting that burden on new page reviewers is not acceptable (of course they may do so if they wish, but it should never be a requirement). Similarly, "If the article is factually incorrect then it should be corrected or nominated at prod or AfD (unless it's a blatant hoax, in which case it should be speedily deleted under criterion G3)" yes, it should be corrected by the creator or whoever wants the material in the mainspace. Otherwise it will be deleted. It shouldn't be brought to Prod or AfD as these are not for cleanup. "If it isn't about the topic as suggested by the title, then either rename the article or nominate it for prod/AfD." No, if you actually do new page patrol then you will encounter many cases where someone has created an article for topic X by copying their own previous creation about topic Y, and forgot to change all or most of the text. Topic X is notable, the creator is probably knowledgeable and interested in correcting this, but until then we have a completely wrong article in the mainspace (not incomplete, poorly sourced, just wrong). Speedy deleting this as a hoax is very WP:BITEy and draftifying the much more friendly, gentle solution, the middle ground between keeping the mainspace factual and the editor encouraged to continue working on it.

The approach taken by you three seems to be "we need a policy or you can not do this" (even though we have a policy encouraging draftification in such cases), and "you are not allowed to do things which go against this essay here". Oh, and "all the work should be done by the reviewers, not the creators" or (judging from their contributions) people who never patrol new pages, edit draft space, or nominate pages for deletion (like Whatamidoing and Thryduulf) I'll continue to ignore this until you get a policy that actually supports your positions, or until you get a consensus at ANI or so that I should change my approach. Fram (talk) 12:49, 20 June 2024 (UTC)

Most of that is not opposing DRAFTOBJECT, which is about empowering the newcomer to get their week at AfD if they want it. Opposing part of DRAFTOBJECT are a few example of something draftified twice, where WP:ATD-I rolled with DRAFTOBJECT would mean that Fram is supposed to send the bad article to AfD with a nomination to Draftify. Maybe “do not Draftify twice” is a soft rule, maybe newcomers mainspacing a draftified article do not actually mean that they want to debate it at AfD. SmokeyJoe (talk) 13:15, 20 June 2024 (UTC)

I find your (Fram's) response interesting, because underneath it, it feels like there is a question about whether Wikipedia is a collegial, collaborative project. Several of the examples you give sound like a Wikipedia-as-a-game model: Any sentence could be required to have an inline citation, so if "you" don't have "enough" (or any) inline citations, then "my" move is to capture your article. If you make the right moves, you can get your article promoted back to the mainspace, but I see my role as fundamentally adversarial: I will prevent you from sharing information until you do so in a way that I believe is appropriate, and I will not help you fix any problems you encounter. WhatamIdoing (talk) 21:06, 20 June 2024 (UTC)

And here I thought the singular purpose of Wikipedia is to be a factual, verifiable encyclopedia that anyone can edit. But all this time it has actually been a social platform where the real goal is to get more precious users by zealously protecting their right to publish whatever they want to the first page of Google. Editor retention above all else. JoelleJay (talk) 09:11, 21 June 2024 (UTC)

How the fuck are we going to maintain a "factual, verifiable encyclopedia" if we don't retain editors? – Joe (talk) 09:27, 21 June 2024 (UTC)

Editor retention can still be achieved without militantly assuming every article creation is inherently encyclopedic as a standalone regardless of sourcing and content. JoelleJay (talk) 10:15, 21 June 2024 (UTC)

Assuming bad faith, hiding their work in place that nobody will ever find it, and refusing to put any effort into even checking whether what they wrote is correct, let alone making trivial improvements to things new editors cannot reasonably be expected to master is not the way to retain editors. Thryduulf (talk) 10:31, 21 June 2024 (UTC)

Perhaps not, but frankly speaking, if new editors are not either of the mindset to proactively learn those "things new editors cannot reasonably be expected to master" beforehand or of the mindset to look for help afterwards on figuring out the issues with their creation(s), how to fix them, and then try again? Chances are, you won't retain them long-term with your suggested course of action either.

Instead, you'll likely as not lose them shortly after that newness has worn off just enough that people will start expecting them to have picked up on the basics of "articles need to be factual, non-promotional, in readable English-language prose, with sources", the major policies and guidelines, and some slightly-beyond-basic skills like how to create a reference without scattering CS1 errors all over the page, edit a table or infobox without breaking it, and so on. At that point, people will stop fixing their issues for them and expect them to do it themselves—with skills and knowledge they cannot reasonably be expected to have mastered if other people have silently fixed all issues for them so far.

The solution is not "let them figure it out entirely by themselves" nor "fix it for them without even making them try". It is guidance on where they went wrong and how to do it right. And yes, depending on the severity of the issues and how long fixing it is likely to take (and how likely an editor is to even give it a try), sometimes that guidance is better done outside mainspace. AddWittyNameHere 12:39, 21 June 2024 (UTC)

But what happens is not "move to draftspace, teach them what they need to know, including where to get help, assist them to improve their article and welcome them as a productive editor" but "move to draftsapce where someone can delete in six months". Thryduulf (talk) 13:01, 21 June 2024 (UTC)

Yes, hence my won't retain them [...] either (emphasis added) and [t]he solution is not "let them figure it out entirely by themselves" nor "fix it for them without even making them try" (emphasis), i.e. if the aim is predominantly "retaining editors", then absolutely, the current method does indeed not work well for that purpose. I just do not believe the course of action suggested by you would work any better, for the reasons outlined above (and would come with an additional hidden cost: still no long-term retaining of productive editors, but an increase in workload as a result of these editors leaving a little later)

That said, I don't think the primary intended objective of draftification-as-concept is or has ever been "retaining new editors", it is "guarding mainspace from incorrect, dubiously notable and/or unverifiable, but potentially improvable, new articles while retaining the contents somewhere so that (at least in theory, as we all know that this only rarely happens in practice) someone could work on improving it without the hassle of having to get it undeleted first". It is a slightly less BITEy alternative to deletion, but also only slightly so.

That in practice it ends up being less of an alternative to and more of a delayed form of deletion is exactly where the guidance I mentioned comes in: such guidance is lacking, it should not be, and if it were not, it would work towards both objectives (editor retention/mainspace not getting flooded by New Editor's Clueless First Article) and make it a significantly less BITEy alternative to deletion that produces some actually-mainspaceable articles instead of an almost-as-BITEy-delayed-deletion that produces a heap of stale drafts to be cleared like clockwork. AddWittyNameHere 14:09, 21 June 2024 (UTC)

And where was it decided that mainspace needed to be 'guarded' from such things? Not to state the obvious, but this is a wiki; we don't need a special place where people can work on things. Our editing policy even explicitly states that poor articles, if they can be improved, are welcome. – Joe (talk) 14:19, 21 June 2024 (UTC)

And where was it decided that mainspace needed to be 'guarded' from such things? Among various places, in the WP:ATD-I section of the WP:Deletion policy, which states Recently created articles that have potential, but do not yet meet Wikipedia's quality standards, may be moved to the draft namespace ("draftified") for improvement, with the aim of eventually moving them back to the main namespace, optionally via the articles for creation (AfC) process.; the WP:DRAFTIFY of the WP:Drafts explanatory essay, which states The aim of moving an article to draft is to allow time and space for the draft's improvement until it is acceptable for mainspace; during the RfC which proposed the creation of a Draft namespace and which explicitly described one of its potential uses as a successor of the now-historical WP:Article Incubator; during the various discussions and decisions which led to the existence of said now-historical Article Incubator. AddWittyNameHere 20:54, 21 June 2024 (UTC)

I've heard (but not read myself) that if you want to retain new editors who add content, then one of the best things you can do is add an inline tag like {{fact}}, which they will often fix the next day.

So if you want stuff WP:Glossary#cited, hiding the whole page in Draft: space probably isn't the right way to go about it. WhatamIdoing (talk) 21:10, 21 June 2024 (UTC)

Yes, exactly this. Frankly I don't expect the editors who aren't bothered to do even the bare minimum in making their first article PAG-compliant (citing sources in any format) to have any interest in sticking around long-term. If they don't understand they need citations in 2024, despite all the hand-holding alerts and training modules available for new editors nowadays and the ubiquitousness of "citation needed" in English online discourse, then best case scenario is they're a child or geriatric person who doesn't know any better. More likely they're careless, incompetent, a vandal or amateur self-promoter, and/or don't speak English at all, and would be both highly unlikely to continue editing anyway and not the type of editor we'd want to retain regardless. A person who actually cares about contributing would put in some effort and not be discouraged by the mildest difficulty, and a person we'd want to keep around would be familiar enough with "citations" and "what is an encyclopedia" that we wouldn't need to explain very basic concepts that have been universal in secondary education for 30+ years. JoelleJay (talk) 23:45, 21 June 2024 (UTC)

@AddWittyNameHere, I'm struck by your comment about skills and knowledge they cannot reasonably be expected to have mastered if other people have silently fixed all issues for them so far. I have said for years that one of the reasons that I stuck around in the early days was precisely because an editor silently fixed wikitext errors for me. I mastered wikitext despite this (welcome) help. I probably would have quit if everything had been reverted or someone had yelled at me for making mistakes. WhatamIdoing (talk) 03:57, 22 June 2024 (UTC)

(@WhatamIdoing: Apologies up-front for the somewhat lengthy response, but this is about as far as I managed to condense it after four rounds of removing extra verbiage.)

Yes, I think that's a common thing for a decent portion of the editors we actually retain: autodidacts and adjacent, to whom seeing someone else do it (or reading the documentation) is enough to pick up on what to do. But that's not the way everyone learns best, or finds comfortable, or has the time and energy to spare for to engage in what is a voluntary internet hobby. "cannot reasonably be expected" was meant to be read as "not a reasonable ask of all or most new editors", not as "impossible for any and every new editor".

"Revert everything/yell at" are certainly not better, but I am not advocating for that, and I am a little puzzled that every response I have gotten in this conversation so far seems to assume I must be in agreement with the status quo simply because I see issues with a specific alternative presented. Rather, I am saying "hey, instead of assuming these are the only options, let's look at what other options might exist. How about, say, Z: neither silent fixing nor silent draftifying, but actual personalized guidance."

(The hows of that are a separate matter. On account of this message already being lengthy, all I'll say about it here is that imo, it's probably best done through a different process than (but if possible, working closely with) NPP/AfC, both because of pre-existing chronic AfC/NPP backlogs and because of different personal inclinations and skill sets between "check large volume of articles for compliance with core content policies" and "guide individual newbies through creating a single core content policy compliant article, tailoring approach to said newbie".) AddWittyNameHere 09:24, 22 June 2024 (UTC)

As a matter of efficiency (i.e., getting both decent-ish content and not scaring off the newbie; NB that this is different from the fastest way to hide imperfect content), I think that silent fixing will prove better when the newbie is in the first few edits. Most newbies don't manage to make three edits, or to edit on two different days. Purely as a practical matter, then, I wouldn't routinely attempt any personalized guidance at that stage.

The Wikipedia:Mentorship tools might be effective, but I understand that enwiki is the only Wikipedia where too few people actually want to help newcomers for that to work out. WhatamIdoing (talk) 22:49, 27 June 2024 (UTC)

I also think that it's supposed to be a verifiable, factual encyclopedia. However, that's not a full and complete description. Consider:

• A verifiable, factual encyclopedia produced by people working together collaboratively (e.g., if you created an article by copying/pasting a previous one, and you forget to remove something from the old article, I could blank that off-topic content for you), versus
• A verifiable, factual encyclopedia produced by people working adversarially (e.g., instead of fixing an obvious problem, I'll hide the whole thing in the Draft: namespace).

WhatamIdoing (talk) 21:06, 21 June 2024 (UTC)

What is this "pseudopolicy" and why is it exempt from actual policies, like Policy and guideline pages are seldom established without precedent[3] and require strong community support. Policies and guidelines may be established through new proposals, promotion of essays or guidelines, and reorganization of existing policies and guidelines through splitting and merging. [...] Proposals for new guidelines and policies require discussion and a high level of consensus from the entire community for promotion to guideline or policy status. Adding the {{policy}} template to a page without the required consensus does not mean the page is policy, even if the page summarizes or copies policy. and Wikipedia has a standard of participation and consensus for changes to policies and guidelines. and The burden to demonstrate verifiability lies with the editor who adds or restores material and Because a lack of content is better than misleading or false content, unsourced content may be challenged and removed.? JoelleJay (talk) 08:39, 21 June 2024 (UTC)

I’m not sure what the point in the green stuff is, but pseudo policy is something that is in practice as policy but without being documented. The most obvious pseudo policy is the guideline WP:N, which is not policy, but is enforced as policy through WP:DEL#REASON#8 (used to be #6). SmokeyJoe (talk) 13:56, 21 June 2024 (UTC)

The thing that bothers me about quoting "The burden to demonstrate verifiability lies with the editor who adds or restores material" in this context is that it seems to mean "Editors are supposed to magically know that I wanted an inline citation for that, even before I saw the page, so they should have provided one in advance of me actually WP:CHALLENGING the content, but since they didn't read my mind, the content should be hidden until they (a) find where I've hidden it and (b) fix it up well beyond the level of adding a source, but so that an AFC reviewer will feel comfortable publicly endorsing it".

That's not really what the policy says, but it appears to be what's meant in the specific context of draftification. WhatamIdoing (talk) 21:03, 21 June 2024 (UTC)

Fertiliser

"Not ready for mainspace" is pure "I don't like it"-ism

if an article is:

• in some form of English
• has at least one citation
• meets notability guidelines
• is not an attack page
• is not vandalism
• is written in good faith

it *is* ready for mainspace. it does not need to be draftified.

It can be tagged to the skies. It can be ignored in NPP for months or years. It does not need to be put into a box labeled "go away you suck". New articles and new users are awkward teenagers that need a little patience and encouragement and many of them *will* grow into beautiful competent adults. Sending them away to reform school in the mountains is just avoiding dealing with our own discomfort with our own flaws and imperfections. NPP should ideally be a "gates wide open come on in" group of greeters who are there to welcome new articles and new users to the party. "Hi here's a cocktail. It's crazy up in here. Here's the syllabus and another cocktail and a cookie and also a kitten. Don't mind them, that's a WikiProject, they're kindly fanatics."

Shitty articles are good actually. Shit is fertilizer. Shit is rich in nutrients and promotes growth. Scrubbing the world of shit reduces cholera transmission but also increases the prevalence of autoimmune disorders. There's got to be a balance.

Anyway, IMHO, "not ready for mainspace" is mean and vague and more harmful than helpful. Even the worst article that meets the standard above should be greeted with a compliment sandwich: "Thank you so much for contributing to Wikipedia! Your passion for this topic is so evident. I wanted to let you know that that according to our current guidelines, this article may be [list top 3 problems here]. Let me know if you need any help resolving these issues. We really appreciate you contributing to the sum of all human knowledge. There's so many topic domains that still need attention and we so appreciate your participation in growing the project."

Anyway, please enjoy the snacks and thanks for coming to the party. jengod (talk) 16:46, 21 June 2024 (UTC)

WP:FERTILIZER should be blue! – Joe (talk) 17:47, 21 June 2024 (UTC)

done. Let's groooooooowwwwww! jengod (talk) 18:29, 21 June 2024 (UTC)

Meh… I’m now considering writing a contrary essay, and calling it WP:Prune the weeds. Weeds need to be pruned in order for a healthy garden to grow. Blueboar (talk) 18:41, 21 June 2024 (UTC)

I would very much like all Wikipolicies to be presented as a series of (increasingly unhinged?) rambunctious garden metaphors, and would like to be pinged at the creation of each entry in the series please and thank you. :) jengod (talk) 18:57, 21 June 2024 (UTC)

Keukenof, said to be the most beautiful garden in the world, requires an awful lot of maintenance. The best Wikipedia articles, in terms of maintaining volunteer sustainability, are spontaneously maintained and updated by passing readers. SmokeyJoe (talk) 01:30, 22 June 2024 (UTC)

This is operating under the assumption that a primary purpose of Wikipedia is cultivating a social community that effusively courts all potential new users, even when it comes at the expense of encyclopedia quality and requires established editors to take time away from editing to mentor newbies. That's fine for the people who want to do that, but a huge proportion of editors are mainspace-only and so would not know about or care to participate in any kind of newcomer-welcoming behavior regardless of whether "building a community" was an actual WP goal, and would interact with newcomers the same way they always would with any other editor they encounter. Meanwhile I would guess a large percentage of those who do get involved in Wikipedia-space are only there for reasons directly related to improving the specific pages or topics they're working in rather than a desire to be part of a broader "Wikipedia movement" or to socialize or do outreach. These are the editors who have their individual motivations for expanding articlespace, enforcing content rules, molding the PAGs, etc. towards what they think the encyclopedia should be; why should they be forced into additional social roles, beyond what is needed for civil discourse between colleagues, in furtherance of the WMF's or other people's agendas for "Wikipedia the Institution", that are not actually backed up by empirical evidence showing they'd have the intended effect, and for which it isn't even clear how the intended result would improve the encyclopedia? JoelleJay (talk) 01:52, 22 June 2024 (UTC)

For User:WhatamIdoing especially, I was coincidentally listening to a radio program interviewing an academic horticulturist, who was asked for the definition of a weed: “A weed is a plant that you don’t like”. SmokeyJoe (talk) 07:08, 22 June 2024 (UTC)

@JoelleJay, I think you have identified two key points of tension.

The first is that some people are doing things that don't meet their stated goals. For example:

• They say that their goal is to comply with policies, but they take actions directly against the Wikipedia:Editing policy, which says Wikipedia is better off when it has more information instead of less.
• They say that they want more citations added to articles, but they take actions (e.g., moving pages into Draft: space) that reduce the likelihood of that happening, and don't take actions (e.g., adding {{fact}} tags to newly created articles) that would increase the likelihood of that happening.

The second is that some people are preventing other editors from doing the work that they'd like to do. This, I think, is why we have discussions such as this one. For example, if Alice moves the page into Draft: space without a redirect, that reduces the chance that Bob has for mentoring the new editor, because when articles disappear, new editors are less likely to edit again. If Bob's going to be able to do his work, Alice needs to be a little slower at hiding the WP:IMPERFECT and WP:UGLY articles. WhatamIdoing (talk) 21:17, 27 June 2024 (UTC)

I don't know how you're getting any of that from what I said, at all. 1. Per our editing policy, Because a lack of content is better than misleading or false content, unsourced content may be challenged and removed, so just no to the first bullet point. 2. This is a nonsensical dichotomy that isn't worth addressing. 3. Literally every edit is "preventing other editors from doing the work that they'd like to do", just because YOU have decided it's "worse for the encyclopedia" if "Bob" doesn't get a chance to mentor someone doesn't mean this is objective fact, and it certainly doesn't mean it's objectively better that "Alice" doesn't get to remove unsourced and potentially false or misleading junk from the encyclopedia, or that it's "better" if we prevent her from "doing what she wants to do" by forcing her to do more than "consider" options other than removing the material. The author of the content certainly wasn't prevented from "doing what they wanted to do" despite failing to comply with core content policies. If Bob is so worried about draftification hurting a new editor's feelings, then it's up to him to be faster at NPP, or to watch Alice's contribs, so he can insert himself as a mentor for this hapless noob; it's certainly not Alice's responsibility to prioritize Bob's goal of "retaining editors who didn't put even minimal effort into reading our rules" over her own immediate goal of "improving encyclopedia quality". JoelleJay (talk) 00:44, 28 June 2024 (UTC)

1. Per the start of the paragraph from which you quote, Wikipedia summarizes accepted knowledge. As a rule, the more accepted knowledge it contains, the better. NB that "uncited" is not the same thing as "misleading or false content".
2. What is a nonsensical dichotomy?
   • I never said that it's better for Bob to get what he wants. I only say that the conflict between Alice's actions and Bob's desire result in these conversations. Alice might be right. Bob might be right. Probably both of them are right sometimes and wrong sometimes, and probably they even agree that Alice really ought to draftify some (worse) pages and Bob really ought to mentor newcomers on some (better) pages. What we won't get is Alice and Bob conflicting with nobody ever starting these l-o-n-g and repetitive discussions about whether the line is drawn in the right place. This symptom (the discussions) indicates that not everyone in our community has the same idea about what's best. This symptom does not mean that either Alice or Bob is wrong.
   • Note that I never said anything about the quality of the article Alice is draftifying; there is no reason to believe that it contains any "false or misleading junk". Most unsourced pages don't, and many long, heavily cited pages do. Citations are not a magic incantation that protect articles from containing false and misleading junk.
   • Your proposal that Bob "be faster at NPP" is non-functional. Nothing Bob can do – whether seeing the article first, or editing it, or contacting the original author – can prevent Alice from draftifying the article whenever she thinks that's the best action.
   • WP:Nobody reads the directions. If your baseline for promising new editors who are "worthy" of mentorship is that they read all, or even some, of the directions before WP:Being bold, then Wikipedia will not outlive you.

WhatamIdoing (talk) 01:47, 28 June 2024 (UTC)

Diagram

I made a diagram that visually translates my understanding of what the phrase "not ready for mainspace" means. What do you think about it? (The boundary between the "not ready for mainspace" and the implied "not unready for mainspace" is the line between the grey and greenish block in the background.) —Alalch E. 16:14, 28 June 2024 (UTC)

That looks reasonable to me, with the caveat that it is difficult to accurately detect all LLM content. WhatamIdoing (talk) 02:03, 29 June 2024 (UTC)

"Raw LLM" is not a reason to do anything and doesn't belong in this discussion or on this diagram. If it has no problems then it belongs in the article space, if it is a copyvio then it needs to be deleted, if it has other problems (bad grammar, no reliable sources, etc) then it needs to be treated identically to a human-written page with those problems. Thryduulf (talk) 22:00, 1 July 2024 (UTC)

I don't think that 'Raw LLM' means the ordinary dictionary definition. I'm assuming it means 'Whopping great mess of nonsense, usually including made-up citations to non-existent sources'. WhatamIdoing (talk) 23:14, 1 July 2024 (UTC)

A page that is a "Whopping great mess of nonsense, usually including made-up citations to non-existent sources" should be fixed (if possible) or deleted (if not). Whether it's in that state because of a human author, an LLM-author or a combination of both is completely irrelevant. The problem is that it's a mess, not who authored it. Thryduulf (talk) 23:35, 1 July 2024 (UTC)

Yes, I agree, but I understand that many reviewers believe that it's expedient to shift the mess out of the way while you figure out whether it can be (or will be) fixed. WhatamIdoing (talk) 00:05, 2 July 2024 (UTC)

It's not currently possible to use an LLM to generate text that doesn't have problems from Wikipedia's point of view. The available models are incapable of generating citations to real, relevant reliable sources, incapable of fact checking, and all their output is of legally ambiguous copyright status. Therefore, if someone uses LLM output without taking steps to address those problems (hence "raw"), it is automatically not mainspace-worthy. If the technology ever improves to the point where it could plausibly generate something we could use directly, then I agree we should evaluate it as if it came from a human, but since that's currently science fiction I think the practice of moving unedited LLM output to draftspace is sensible. – Joe (talk) 10:01, 2 July 2024 (UTC)

Again, that it was written by LLM is irrelevant. If something has problems those problems should be dealt with by whatever means is most appropriate to the specific problems that it has. e.g. if it has no real, relevant reliable sources then the problem is that it has no real, relevant citations not that it was written by an LLM. Thryduulf (talk) 11:58, 2 July 2024 (UTC)

+1 Donald Albury 13:22, 2 July 2024 (UTC)

If something has a 100% chance of causing articles with mainspace-disqualifying problems, I think it's utterly immaterial whether we say we're removing it from mainspace because of the problems or because of their cause. – Joe (talk) 16:25, 2 July 2024 (UTC)

I've not used LLMs myself, but someone showed me one recently that provides a list of real citations. My guess is that it's a type of Retrieval-augmented generation (first you retrieve real sources; then you apply LLM to their contents). The idea seems to be both to minimize hallucinations and also be able to check for them afterwards.

If the result is not an 'accidental' copyvio, and it is providing real sources that actually verify the contents, is that still a 100% chance of mainspace-disqualifying problems? WhatamIdoing (talk) 16:42, 2 July 2024 (UTC)

@Joe I couldn't disagree more. We should only be removing things from mainspace for articulated reasons that are directly relevant to why it is unsuitable to be in mainspace. Who wrote it is only relevant if it is being speedily deleted under criterion G5. Even if 100% of things user:Example has written are A7 failures, we remove their newest contribution for being A7 failures not because it was written by User:Example. The distinction is important because otherwise we would be deleting their next contribution even if it is about a very clearly notable topic. Thryduulf (talk) 16:48, 2 July 2024 (UTC)

This is a tangent, but I asked for some information about newbies' first edits, and it appears that if the first edit is to create a page, a third of those will get deleted within the first week. Longer term, the result depends on the namespace: User: space pages mostly get retained past that point, but ultimately 95% of Draft: space pages [if created as the newly registered account's first edit] get deleted. There's lots more over there, for anyone who is interested. WhatamIdoing (talk) 17:48, 2 July 2024 (UTC)

I think it's pretty relevant whether text was authored by LLM vs human, especially for niche complex topics. At least with human-written text if something is confusing or problematic there's the possibility of asking the author to clarify what they mean, or how they drew a particular conclusion from a given source, or why they weighted one source more than another. With an LLM there is no possibility of the "discuss" part of BRD, because neither the LLM operator nor the LLM itself can answer "why did you summarize this in this way". We can't backtrace their rationale from the sources they use because there is no rationale that would be meaningful to humans. What happens when an LLM is contributing a bunch of content in like Hodge theory, where very few people have expertise? How comfortable are we letting that content stand when we know it's been generated to look very plausible, but could easily contain significant errors only experts would detect? Comfortable enough to allow it on a few pages? What about an arbitrary number of pages? JoelleJay (talk) 01:29, 3 July 2024 (UTC)

Absolutely none of that is specific to LLMs. Any content can have any, some or none of those problems regardless of author. Address the actual problems with the content not unverifiable personal suppositions about how or why the content might or might not have been created. Thryduulf (talk) 02:59, 3 July 2024 (UTC)

In answer to: How comfortable are we letting that content stand when we know it's been generated to look very plausible, but could easily contain significant errors only experts would detect?

Apparently we are comfortable enough with this that we still let anyone edit any sort of technical, niche, or complex material, even if they have no expertise. See Wikipedia:List of hoaxes on Wikipedia for a small sampling of the times when that's caused problems. Those cases were generated by bad-faith humans, but the only difference I can see between a bad-faith human and a good-faith human with an LLM is the potential volume for the latter. (The good-faith human with an LLM even has a chance of getting things right.) WhatamIdoing (talk) 03:59, 3 July 2024 (UTC)

@Thryduulf: It's true that we could judge content purely on its merits and get a similar result but, in practice, editors that are active in checking content for problems find it labour-saving to consider contextual information like who or what created the article. I understand that you object to this in principle—not just with LLMs but also with COI edits—but I don't understand where this principle comes from or why it's important to preserve. I'm genuinely interested to learn. – Joe (talk) 06:45, 3 July 2024 (UTC)

Contextual information is useful for helping to evaluate the content, but it is the content that we are evaluating not the contextual information. Ultimately the only thing that actually matters is the words on the page. If the words on the page are bad then they should not be in Wikipedia (i.e. they should be fixed or removed), if they are good they should remain. This is true regardless of who wrote them, how they were written and why they were written. Thryduulf (talk) 09:01, 3 July 2024 (UTC)

This principle appears in places such as Wikipedia:Comment on content, not on the contributor. It's also the principle behind the POV and COI article templates: the time to remove {{COI}} is when the article is fixed, not when the accused editor has been either declared innocent or ejected from the community. WhatamIdoing (talk) 17:20, 3 July 2024 (UTC)

The measure of a ban is that even if the editor were to make good or good-faith edits, permitting them to edit in those areas is perceived to pose enough risk of disruption, issues, or harm, to the page or to the project, that they may not edit at all, even if the edits seem good.

Anyone is free to revert any edits made in violation of a ban or block, without giving any further reason ... the presumption in ambiguous cases should be to revert.

Given we have consensus that LLM-generated content is unreliable, and multiple editors have been blocked/banned for repeatedly adding "raw LLM" text, why shouldn't material that is identifiably "raw LLM" be considered the product of a banned editor and reverted without discussion even if it isn't obviously bad? The LLM is functionally just as banned as the editors who use it, therefore any content it generates can be presumed likely to feature the same issues that led to it being deemed unreliable. JoelleJay (talk) 04:06, 4 July 2024 (UTC)

The LLM is functionally just as banned as the editors who use it Where is the consensus for this? The only discussion I recall came to the opposite consensus - i.e. it's not LLMs that are bad it's bad output from LLMs.

any content it generates can be presumed likely to feature the same issues that led to it being deemed unreliable. again, citation needed. If the content it generates has issues deal with those issues, if it doesn't then there is no problem. Thryduulf (talk) 08:15, 4 July 2024 (UTC)

What part of Large language models (LLMs), such as ChatGPT, are unreliable. is unclear? As the raw output of LLMs is consistently inaccurate/fake enough for there to be consensus that LLMs (/their output, whatever) are unreliable, and that any news media created via LLMs is also unreliable, that raw output should not be added to mainspace. Editors whose natural product is routinely that inaccurate, hallucinatory, and/or nonsensical would be blocked, and editors who repeatedly introduce raw LLM content would also be (and have been) blocked.

Therefore, in situations where it is identifiable (e.g. retaining "I am a large language model" or "regenerate response" in the most obvious cases), or where its origin is otherwise strongly suspected (e.g when a portion of an edit contains hallucinated sources, then the whole edit should be treated as LLM-generated), raw LLM output can be removed immediately the same way we permit deletion of edits by banned users, even if the (rest of the) content isn't blatantly problematic. JoelleJay (talk) 20:46, 4 July 2024 (UTC)

It's perfectly clear but also perfectly irrelevant. We care about the problems with the output, we don't care about the input or the method because they don't matter. If the sources are fake we delete it because it has fake sources - it's irrelevant whether they are the product of human, machine, both or neither. Thryduulf (talk) 21:00, 4 July 2024 (UTC)

We should reject bad content because it's bad content, regardless of the way that bad content came into being.

The problem with "reject raw LLM" is that it's mistaking the cause for the effect.

If the LLM creates a sentence like "The capital of France is Paris" – a statement that is given in policy as an example of a statement that is so obviously factually correct that it will never require an inline citation – and an editor drops it in a relevant article with an edit summary that says 'copied and pasted with no changes from my LLM', then that's "raw LLM output" but neither wrong nor against the spirit or letter of any present or past rule (including copyvio, as that sentence is not copyrightable). We should accept that particular instance of "raw LLM output" because there is nothing wrong with that specific edit.

But if a human creates a sentence like "The capital of France is London", then we should reject that, because it's factually incorrect.

The simple and accurate rule is "reject bad content and accept good content", not "reject LLM content, which is good in a certain percentage of cases, and accept human content, which is bad in a certain percentage of cases". WhatamIdoing (talk) 22:22, 4 July 2024 (UTC)

We endorse deleting any content added by socks, especially from socks that we know make subtle shit up or do copyvio, so yes, we do care about the origin of an edit even if it is not immediately apparent that it is bad. JoelleJay (talk) 04:06, 13 July 2024 (UTC)

If the text is so good that we cannot tell it is a LLM output, and all content is verified, we may as sell keep it. But I had a case recently where the content looked like LLM, and references were all fake, and content was unverifiable, I deleted it as a hoax, but it had me fooled as it is a real thing. Maybe that was a bit harsh. (Kamafugite) Graeme Bartlett (talk) 09:50, 4 July 2024 (UTC)

If the references are all fake and the content unverifiable then it shouldn't be in mainspace (whether AfD, speedy deletion or draftification is best will depend on specifics). Whether an LLM was involved in its creation in any way is indeed completely irrelevant. Thryduulf (talk) 09:57, 4 July 2024 (UTC)

Some kind of reminder at AN/I

This is not a joke proposal.

I'd like to propose some kind of reminder/checker that automatically runs at AN/I, in the style of [1]. Actually reading it back to the user is good, but a given user's computer probably doesn't have the sound turned on all the time, so a pop-up dialogue box, with a five-second delay before the user can click "continue" would probably work too. Something like this:

Here's what you just posted to AN/I:

“ And what the hell is that supposed to mean? You're being very unconsructive here, in fact, you're being a bit of a jerk. Go away, because you are SO not helping. — Swanky Signature (talk :: contribs) ”

Yes, post this comment.Wait, go back.

Thoughts? Cremastra (talk) 20:05, 16 June 2024 (UTC)

Having a delay for every single reply can really make things really annoying. You also have all the different userscripts for replying (CD, Factotum, etc.) that to make that warning appear to every one of them would be really hard. Aaron Liu (talk) 22:41, 16 June 2024 (UTC)

Forced preview was a feature(?) in the early days of the visual editor. However, it previewed the entire page, which would be a terrible experience at ANI.

There's another Wikipedia (Korean, maybe? I can't remember) that does something like this for all posts to their village pump. I don't think it shows your comment. It's more like a message that says "This is the village pump, which is not for random chatting". I don't know for certain how they set it up, but it might be implemented with the 'warn' setting in Special:AbuseFilter. That method would probably work on (almost) all tools. WhatamIdoing (talk) 18:24, 19 June 2024 (UTC)

I avoid ANI as much as possible but I like this idea, and I'm very impressed by this design. That's some legendary Wikitext work. Toadspike [Talk] 07:49, 25 June 2024 (UTC)

We could just have it restricted to popping up for comments that contained the string shit, piss, fuck, cunt, twat, hell, damn etc. jp×g🗯️ 00:27, 7 July 2024 (UTC)

I like this idea. Pecopteris (talk) 00:31, 7 July 2024 (UTC)

And that could be easily done with an edit filter (I think). Cremastra (talk) 01:07, 7 July 2024 (UTC)

Why those? Scunthorpe problems aside, profanity is only very mildly correlated with the actual problems at ANI. —Kusma (talk) 11:59, 7 July 2024 (UTC)

Well: a script to detect if someone's saying "asshole" takes fifteen seconds to write, whereas a script to detect if someone's being an asshole takes fifteen years of dedicated research. jp×g🗯️ 20:01, 7 July 2024 (UTC)

I think we should have a script that detects if people are using a predetermined list of insults and offensive words on non-articles and asks if you actually want to post this, flags it as a potential NPA violation, and maybe even prevents posting for non-autoconfirmed users. Dronebogus (talk) 12:20, 13 July 2024 (UTC)

New shortcut

Tracked in Phabricator
Task T369993

A recent "feature" on article pages (PC, Chrome) is an indelible box near the top left of article space, giving access to the list of headings. This is an unhelpful irritant to someone who simply wants to read an article, as it obscures the first words of several lines. Assuming it's a useful feature, can it be made minimizable or semi-transparent and becomes opaque when moused-over ? Doug butler (talk) 15:02, 13 July 2024 (UTC)

I'm not seeing this in Firefox or Chromium on Linux, logged in or logged out, using monboook, vector or vector2022 skins. If it is something that's from Wikipedia rather than your browser this is a question that's better suited to the technical village pump or phabricator than this board. Thryduulf (talk) 15:20, 13 July 2024 (UTC)

Screenshot with table of contents expanded

@Doug butler, are you seeing something like this, at the top of World? This screenshot shows the table of contents covering up part of the first paragraph. If you click the button right next to the =World= title, it will collapse. WhatamIdoing (talk) 19:59, 13 July 2024 (UTC)

Thanks, exactly that box, functions the same, but on article space, covering the first three or four letters on the first two or three lines of text (Vector (2022) everything vanilla standard). When an article is opened, the three dotted lines symbol is there alongside the title of the article, then appears (as my problematic box) as soon as I scroll down sufficiently that the title line is goes off-screen. Doug butler (talk) 23:40, 13 July 2024 (UTC)

@Doug butler, please try this link:

https://en.wikipedia.org/wiki/Katharine_Viner?safemode=1

The safemode=1 bit will screw up the infobox formatting, but don't worry about that. Just see if that fixes the problem or if the problem is unchanged. (I expect it to be unchanged, but the answer will tell us who to contact about getting it fixed.) WhatamIdoing (talk) 00:15, 14 July 2024 (UTC)

Thanks for the prompt reply. Behavior of that box exactly as before. Doug butler (talk) 00:24, 14 July 2024 (UTC)

Then I think we want @SGrabarczuk (WMF) and @Jdlrobson from the Web team. You said you're using Chrome on a PC. What version for Chrome and for Windows?

Does this behavior change if you change the size or zoom of the window/screen? WhatamIdoing (talk) 00:54, 14 July 2024 (UTC)

Windows 10 Pro 22H2 installed 25 July 2020 on Lenovo 4205 laptop i5

Chrome is up to date:

Version 126.0.6478.127 (Official Build) (64-bit) Doug butler (talk) 01:12, 14 July 2024 (UTC)

To clarify, my problem box remains in the same position on the screen as I scroll down through the article, so it masks just the leftmost three or four letters on the top two or three lines, assuming "Small" text selected. The box remains the same screen size with "Standard" or "Large" text selected. Size of the box increases and decreases proportionately with text size as Ctrl+ or Ctrl- invoked. Doug butler (talk) 01:01, 14 July 2024 (UTC)

This is included within Vector 2022 when the screen size is too small for a sticky header. To make the button semi-transparent unless hovered, add the following CSS to your personal CSS for Vector 2022:

#vector-page-titlebar-toc {
  opacity: 50%;
}
#vector-page-titlebar-toc:hover {
  opacity: inherit;
}

I'll see if I can create a patchset for this. Aaron Liu (talk) 02:49, 14 July 2024 (UTC)

Actually, add the following as well if you want it to be opaque when extended:

#vector-page-titlebar-toc:active {
  opacity: inherit;
}

Aaron Liu (talk) 03:15, 14 July 2024 (UTC)

I've opened the task, submitted the patch, and attached a {{tracked}} before the opening... ask. Aaron Liu (talk) 03:27, 14 July 2024 (UTC)

I might leave CSS alone and report back ? Should I subscribe to the task ? As you can tell, I'm not at all HTML-savvy. Doug butler (talk) 04:25, 14 July 2024 (UTC)

You don't have to subscribe to the task, but you're welcome to.

If you don't feel any need to have this solved urgently, then you could do nothing, and perhaps it will get fixed for everyone on an upcoming WP:THURSDAY. If you want it fixed sooner than that, then updating your account files will fix it just for your account. WhatamIdoing (talk) 05:47, 14 July 2024 (UTC)

It was an inconvenience, is all, and another week is not a problem. Thanks all for taking it seriously — most impressive service. Doug butler (talk) 10:30, 14 July 2024 (UTC)

Redefining ECP

Once again concerns about ECP being too easy to work towards have been raised at AN. My concern with this is that we are WP:BITING good faith editors who see the requirements and, not knowing that we don't want editors working towards them, decide to do so through productive and good faith edits.

Rather than repeatedly taking such editors to AN or ANI, or otherwise sanctioning them, the best solution is to change our requirements so that even if an editor chooses to work towards them we can have faith than they have obtained a minimum level of understanding and competency. As a simple and conservative change towards this, I suggest that to obtain ECP an editor must have:

• 500 edits outside of draft and user space (noting that articles written in draft or user space but moved to article space will be considered article space edits)
• 200 of which are more than 250 bytes

Implementing this can done with a simple adminbot, which would check whether users have met the requirement and grant them ECP if they have. Users who have already been granted ECP will be grandfathered in.

@Iskandar323, Amayorov, EggRoll97, Swatjester, Starship.paint, Sean.hoyland, Selfstudier, and XDanielx: Notify editors who participated in the recent AN discussion. BilledMammal (talk) 20:56, 9 July 2024 (UTC)

I know there have been concerns about me gaining ECP in less than a week, but I must say that it certainly wasn't easy. It took me around five days of non-stop editing and learning, practically without taking a break except for sleep.

But anyway, I think yours is a great suggestion! Is there some concept like a cumulative edit to a page? Because otherwise a user could simply insert-delete a wall of text multiple times to meet the requirements. Amayorov (talk) 21:02, 9 July 2024 (UTC)

I'd say a user inserting and deleting a wall of text would be blocked as WP:NOTHERE before they get to ECP. Also, I genuinely don't think non-stop editing for days without taking breaks except for sleep is healthy. Not in terms of your edits, but in terms of actual health. Chaotic Enby (talk · contribs) 21:05, 9 July 2024 (UTC)

I agree with this. I'm not concerned about editors who obtain ECP through bad faith edits, as they are easy to identify and there is no WP:BITE issue in sanctioning them for it. BilledMammal (talk) 21:10, 9 July 2024 (UTC)

I agree that a qualitative strengthening of the requirements would be beneficial. I mentioned the spirit of the 500/30 rule before, and in my mind, the spirit is indeed to build up both editing experience and community and guideline familiarity. This actually helps editors who are interested in getting into contentious topic areas in the long run, as it ensures that they begin editing in such areas on a more solid footing and are less likely to inadvertently run afoul of the brighter red lines. The measure ignoring user spaces would improve the qualitative bar a little, though I wonder if it is enough. I might also exclude talk, since chatting in talk, especially with the more recent reply function, often really isn't anything akin to editing. Also, I have certainly seen disruptive users rapidly racking up edits through very vapid or actively disruptive talk page contributions. I totally agree with the 200x250 part –assuming it is feasible to implement. However, I might also add a further dimension, and that is time. The 30 days rule only requires that an account exist for 30 days, but I think the original spirit of this was that an editor edits for 30 days. Again, if practically implementable, it would be good to tighten this to 30 days in which actual edits were made – a metric that, again, would provide some assurances that a user has spent a decent amount of time actually becoming familiar with everything. Iskandar323 (talk) 21:13, 9 July 2024 (UTC)

If we're using an adminbot to apply the rights, almost anything is feasible.

Personally, I would prefer to see some talk space participation, as collaboration is an important part of editing. Adjusting how the 30 day rule works could be useful, although I'm not sure how you propose this is done - perhaps the clock starts after the 100th edit, rather than when the account is registered?

One thing to keep in mind is that previous discussions suggest that the more complicated the requirements are the less likely they are to be supported by the community, which is why I've kept them simple this time. However, we could always run an RfC with multiple options? The status quo, a conservative option, and an ambitious option? BilledMammal (talk) 21:22, 9 July 2024 (UTC)

Even the clock starting from the first mainspace edit would be an improvement – so 30 days since an account nominally goes active in mainspace. This would weed out accounts that just boot up to sniff around, play around in the sandbox, comment in talk and gentrify their own user pages, among others. Iskandar323 (talk) 21:54, 9 July 2024 (UTC)

Is there perhaps another way to frame this? I'm not opposed to the general idea of using a minimum edit size as a proxy for edit quality, and why that works neatly from an automation standpoint, but it's not without it's problems either. But it seems to me that the real problem we're trying to solve for is edit count abuse that leads to post-confirmation problems on ECR articles, specifically. Communicating expectations of editors is good, but we can take a step back and look at the reason why that expectation arose -- it's not really about the 500 edit requirement and what they did to meet it, it's about demonstrating that one knows how to play by the rules and can edit constructively with others (particularly AFTER their 500th edit). Now, let's say we had a hypothetical user who achieves ExC status through 500 meaningless, insubstantial edits -- but then goes on to have a productive career subsequently and either never edits on ECR articles or does so only in a constructive and harmless manner. Is that user a problem? No. More importantly, is the fact that the user gamed the system to get their ExC status a problem, if they're not actually displaying any problematic behavior? IMO, no. So, separate from the above discussion about edit size and counting active editing days instead of account age, what I'd like to see, is a way that specific problem users (and ONLY them, not every single extended confirmed editor) can be reported and have their pre-confirmation edits put up for review if they're behaving badly post-confirmation; with the result of having their extended confirmed status be revoked if necessary. I'm assuming that would take the form of a new noticeboard, but possibly could already be covered by an existing one? I hope I'm explaining it clearly enough. ⇒SWATJester ^{Shoot Blues, Tell VileRat!} 23:28, 9 July 2024 (UTC)

If the goal is to expect constructive contribution from people before they can get ECP, then no automated way of assigning it will be sufficient, as it will still not show constructive ability to collaborate. Any set of "metrics" one can define, including edit count, days active, edits in certain namespaces (such as Article or Talk), size of edits, etc. will either be gameable just as easily as it is now, or will result in it being unnecessarily difficult to obtain to prevent that gaming. Only if there is truly an issue with the current way (assign it automatically after 30d/500 edits), then the only real other option is for it to be assigned manually, such as by request on a permissions noticeboard (or a new board set up for "experienced confirmed editor" confirmation or similar). I do not believe that is in the spirit of the ECP, however, which was intentionally not supposed to be a "trusted editor" but just a "second level autoconfirmed". -bɜ:ʳkənhɪmez (User/say hi!) 03:15, 10 July 2024 (UTC)

I tend to agree with Berchanhimez's rather pessimistic view. ECP apparently can't do what it was hoped it could do, with the exception, in my view, of reducing the temperature on talk pages by excluding fire-starters. More comments later... Sean.hoyland (talk) 05:30, 10 July 2024 (UTC)

Other comments.

• I'm unconvinced that we are WP:BITING good faith editors. I think they trickle into contentious topic areas organically and don't even notice ECP because they will already have crossed the barrier editing the millions of other articles in good faith. In other words, they don't have an objective that is inconsistent with the Universal Code of Conduct. It would be great if there were a whole bunch of good faith editors out there who wanted to join the PIA topic area for example, but I'm not seeing it, before or after ECP was implemented. In reality, I think it is more like a cloud of people with no intention of complying with the prohibition against 'Systematically manipulating content to favour specific interpretations of facts or points of view'. ECP doesn't appear to help them gain the ability to leave their biases at the door before they enter contentious topic areas.
• re: "is the fact that the user gamed the system to get their ExC status a problem, if they're not actually displaying any problematic behavior? IMO, no." I think the answer to this question depends in part on whether Wikipedia requires editors to be honest, to comply with all policies or only some policies. Editors with a "criminal record" that removed their right to be here are more likely to game the system, more likely to make an effort to appear to be a constructive editor, at least for awhile, until their problematic behavior manifests again and the block->sock cycle restarts. Many of the numerous blocked editors are very experienced editors, they are pretty good editors if you ignore their dishonesty, advocacy, racism, ultranationalism or whatever peculiarities they have that get them into trouble. One of the objectives of ECP was to increase the cost of entry for editors who employ deception via sockpuppetry. There is no evidence that it does that, not as far as I can tell anyway. This is a very significant issue because it means there are two classes of editors, and only one of the classes is affected by sanctions, the class that does not employ deception via sockpuppetry. So, for me, "if they're not actually displaying any problematic behavior" is not a very useful test in itself if honesty matters, and it doesn't really tell me anything useful about the value of ECP. Certainly, in PIA, we are dealing with many experienced ban evading editors with a good understanding of the content rules who unfortunately choose to behave like sociopaths. But maybe honesty doesn't matter because it is not enforceable in practice. Wikipedia could make that decision, that we no longer care whether you are using sockpuppetry to evade a sanction because you are generating decent content some of the time. Sean.hoyland (talk) 06:41, 10 July 2024 (UTC)

  I think the answer to this question depends in part on whether Wikipedia requires editors to be honest, to comply with all policies or only some policies.

  My problem with this is I don't think they are breaking any policies by "gaming" ECP. We tell them that they need 500 edits and 30 days tenure to edit; we don't tell them that working towards that through productive edits will be seen "gaming" the system.

  There is no evidence that it does that

  This is part of what adding the minimum byte size will do. It's harder to make large productive edits than it is to make small ones; it will make it harder for editors game ECP, or it will make them more likely to do it through unproductive edits, which will make them easier to catch. BilledMammal (talk) 04:50, 11 July 2024 (UTC)

  Byte size is not the best thing either though. Many things, such as templates (with parameters that are not used but valid parameters for the template), references (which can be 250 bytes on their own in some cases), and a plethora of other "trivial" contributions can be over 250 bytes. And things under 250 bytes can be productive edits too. I would be against a byte size rule for automatic granting - at that point, it needs to just be moved to be granted by administrators after evaluating the edits. Because any size of edit you come up with will be just as easy to "game" as it is now, and would prevent legitimate editors who don't meet it from getting it. That's why I think the only options are leave it as is, or to move it to no longer being granted automatically at all - if "gaming" truly is a concern.

  Personally, I don't see gaming as a concern - because ECP is not to say whether an editor is productive or not - just that they've at least spent more time here than the 4 days/10 edits for autoconfirmed. Even if they gamed it, so what? If they are disruptive, we have processes to deal with that already (such as Arb Enforcement for Contentious Topics, AN(/I) for other behavioral issues, etc.) and I have seen no evidence that those areas are failing to deal with problematic editors. Ultimately, I feel the people arguing to make it harder to get misunderstand its purpose - which is not to segregate editors into productive/non-productive, but is to merely make sure someone who is still brand new can't just hop in to certain pages/topic areas. It's in the name - extended confirmed. A confirmed/autoconfirmed editor is not someone who is guaranteed to be productive - they're just someone who has not made an account an hour ago and has at least some experience with the edit window so they're less likely to "accidentally" delete text or mess something up because they don't understand the edit window. Likewise, EC is just that, but with a bit longer and a bit more edits under their belt. -bɜ:ʳkənhɪmez (User/say hi!) 05:32, 11 July 2024 (UTC)

  Hear hear. Curbon7 (talk) 06:15, 11 July 2024 (UTC)

@BilledMammal:, I agree that people are not breaking any policies by "gaming" ECP. For interest, for your minimum byte size considerations, some plots of the first 600 (or less) edits for a few editors who were blocked as socks. The signals are quite diverse. Sean.hoyland (talk) 13:47, 12 July 2024 (UTC)

I think they [good faith editors] trickle into contentious topic areas organically and don't even notice ECP because they will already have crossed the barrier editing the millions of other articles in good faith. Many (maybe even a majority, though I can't say for sure) high edit-count editors who work in ECP areas tend to limit their edits almost exclusively to these areas, sometimes on a daily basis for years on end. We certainly don't question the extent of these editors' good faith on the basis of where they choose to spend their editing efforts. I don't see why we can't extend the same courtesy to new editors. For example, in your first 500 edits, you made changes to articles which are today ECP. As far as I know, nobody questioned your good faith on the basis of the areas where you chose to start editing. We should continue that with new editors. spintheer (talk) 06:00, 12 July 2024 (UTC)

@Spintheer:, I understand this way of thinking, but "I don't see" is not a good state to be in when you are under attack and thinking about countermeasures. Here is an example where you would have been wrong over a thousand times just for one person employing deception via sockpuppetry to edit in the PIA topic area. Here's another, and another, and another, and it just goes on and on, dishonest people exploiting weaknesses in the ways we guard content. To slightly correct the record for your entertainment, in my case, although it long predated the Arbcom ECP remedy, lots of people questioned my good faithiness, made all sorts of dumb and confusing accusations together with a sprinkling of the occasional threats of violence. But it makes no difference to me whether people question my good faith. Good for them, everyone should question everything all the time, including everything they believe to be true about the world and themself. Assuming anything, including good faith, tells me nothing useful about an editor. The relevant question for me is whether a person is complying with the rules, all of the rules, including rules against advocacy and the use of deception. There need to be countermeasures. There is agreement on that at least. There need to be countermeasures because the strength of the attractive force that content exerts on editors is apparently, in a large number of cases, inversely proportional to their capacity to comply with section 3.3 of the Universal Code of Conduct. I don't know what those countermeasures should be. ECP is not an effective countermeasure. AGF is not an effective countermeasure. Quite the opposite. It is something that is exploited. Rules need to deal with the unpleasant reality rather than a faith based optimistic model of Wikipedia. Sean.hoyland (talk) 08:38, 12 July 2024 (UTC)

I would like to see new measures tested, alternatives or additions to ECP, things that try to address root causes. Unfortunately, I lack the imagination to think of them.

e.g.

* A test that ensures that an editor understands that they must comply with section 3.3 of the Universal Code of Conduct all the time i.e. they understand and accept that they can't convolve their biases with content, regardless of whether it is in good faith or bad faith.

* A place to report non-compliance with section 3.3 that handles reports as routinely as reports of non-compliance with things like revert rules, civility rules etc.

* Routine frictionless use of checkuser in contentious topic areas i.e. effectively make fishing a countermeasure (I'm sure there are arguments against this that I would probably find unconvincing).

But tweaking ECP wouldn't hurt. Sean.hoyland (talk) 07:24, 10 July 2024 (UTC)

@Sean.hoyland, UCOC 3.3 is the section about "Deliberately introducing biased, false, inaccurate or inappropriate content" and "Systematically manipulating content to favour specific interpretations of facts or points of view" (among other things). Are you mostly concerned about POV pushing?

On the assumption that you're thinking about that (and not, e.g., about 3.3's prohibition of hate speech), then I think that it's important to remember that one person's POV pushing is another's genuine attempt to help Wikipedia.

If the article seems too "red" to you, then you try to make it more balanced. And the next person thinks your version is too "blue" and tries to make it more balanced ...which now looks too "red" to you again. This is 100% good-faith editing. It's not all editing in compliance with the community's own POV about what "color" the subject Really™ Is, but it is good-faith editing, and nobody's exploiting anything here. We're all just trying to do our best.

If we keep at it, with just a little intellectual humility (like many Wikipedians, I am confident that I am always right; also, I have been wrong in the past, and it might happen again) and a little willingness to learn (huh, I never thought about it from that angle before), then we usually end up with a better article than if we kick out all the editors who disagree with us or challenge the article's neutrality. Sometimes that's hard. Sometimes we're in the wrong space emotionally to see beyond our own POV. Our needs can overwhelm our ability to understand someone else's view.

But those needs can be managed. We can have a group writing an emotional subject like Mother that includes editors with every type of parents as well as every type of family member themselves. We end up with a better article than if we leave the article entirely to the starry-eyed would-be mother, or to the child abuse survivor, or to the father. What we can't do, though, is do that work without going through the effort of working with other humans. https://www.newyorker.com/news/letter-from-silicon-valley/the-lonely-work-of-moderating-hacker-news talks about an approach to moderating group chats that I think is very similar to our aspirations. The net result is that it's a long game that requires a lot of investment in individual education. Every single editor has to be brought through the process: "Oh, I see. You think this, because you are looking it at from the POV of poverty. Okay, I'm looking at this subject from the POV of personal fulfillment. Can we find a way to fit both of these views into the article?"

I think we keep hoping for a model of "everyone can edit, but it won't be a big, time-sucking, energy-draining endeavor". This is not realistic. Democracy does not have a monopoly on being "noisy and messy and complicated". Wikipedia is all of those things, too. WhatamIdoing (talk) 02:12, 13 July 2024 (UTC)

@WhatamIdoing:, well, don't get me started on modelling the dynamics of the system over time or I'll be here all day.

• I'm mostly concerned about the consequences of there being 2 classes of editors - a class that must follow the rules because sanctions have consequences for them and a class that is unaffected by rules because they reincarnate in sock form. The existence of 2 classes has a significant impact on the dynamics of the system at all times scales, at a particular article at a particular time for example and over long time scales.
• I'm also concerned about POV pushing, not because I have opinions that differ from other people or because of something like a community POV. A reason POV pushing is significant for me is captured by the notion of genetic drift, because just like allele frequency, the incremental corrective processes that move the content dot towards policy compliance, are critically dependent on population size. I don't believe that the population size is large enough to do that in the PIA topic area for example in many cases. And this is another area where the existence of a class of disposable accounts can have a influence on the dynamics. 
• There is the idealized spherical cow type model of the dynamics of Wikipedia. I understand this is the way many people think about things like neutrality and completeness in Wikipedia, as an emergent property of the system. It's an optimistic model and perhaps it is a good first approximation in the majority of cases over long time scales. But what I would say is, if someone thinks they have a decent model of the dynamics in Wikipedia, it probably doesn't work in the PIA topic area. The dynamics are more complicated and diverse, noisy and messy as you say.
• Simple optimistic models have all sorts of assumptions that are not always valid e.g. 

• The notion of a shared objective function, that the system has a single attractor, that individuals acting in good faith are all trying to nudge things towards it. This is observably not the case, and we know that because we can observe efforts to push content up the hill, away from what sources say.
• That we are dealing with individuals where there is a one-to-one mapping between an account and a person. Neither of these things are true in all cases. There is account sharing and coordination between account holders to push the dot up the hill to a preferred location inconsistent with policy. We know this is the case. 
• That the dynamics moves the content dot towards policy compliance over time because of the interactions of many competing agents. Maybe, but what we also observe is instabilities rather than gradual movement towards an equilibrium, again, often because of the existence of 2 classes of agent. These instabilities manifest as things like rapid transitions of content between 2 states via edit warring, over the presence or absence of a word for example.

• As a general point, it's probably true to say that nobody can control the dynamics. They're too complicated and we're not that smart. Attempts to steer things in a direction will probably fail. What can be done though is to try to control what we have the power to control, the behavior of individual agents, the editors, to make sure that they follow the rules, all of the rules, in the hope that what emerges is okay. I look to other complex systems made up of many interacting agents that are meant to have a shared objective function but in fact have complicated competing interests, social insects in colonies where the queen mated with many different males for example. There is misalignment between the interests of subsets of workers. How do they deal with this conundrum? They invest a great deal of energy in worker policing to make sure everyone is following the same set of rules. Sean.hoyland (talk) 07:06, 13 July 2024 (UTC)

Sean, I agree with basically everything you say, except that I prefer to use a Spherical frictionless chicken in my models. ;-)

1. We already have two classes of editors: Those of us with social capital, and everyone else. Disposable accounts are a thing, but they are primarily a thing for spam, rather than geopolitical POV pushing, especially when non-ECP editors get kicked out of discussions
2. One of my concerns about the ECP approach is that it reduces the size of the pool that could edit PIA articles. IMO we need some limits, but we might actually benefit from more editors there, rather than fewer. See Template:Registered editors by edit count – click to Table 2 if you want to see the numbers calculated against those who have made 1+ edits, rather than those who managed to figure out Special:CreateAccount. BTW, while 14 million accounts have made 1+ edit ever, only about 800K made 1+ edit last year, and about half of those never came back after the first day. User:WhatamIdoing/I am going to die compares this against what it would take to replace an editor like me.
3. I think you are right about not everyone prioritizing the same goals in exactly the same ways. In other areas, I have seen editors make decisions based on a desire to influence sources. For example, years ago, some editors were upset with a source about Chemotherapy. The source wasn't wrong (i.e., chemo is vital for some forms of cancer and has limited value for others), but they were upset about the overall value (specifically, chemo is vital for uncommon forms of cancer and has limited value for common forms, which means that the "median" patient probably gets a smaller benefit than they believe they're getting). They were doubly upset that since the Wikipedia article started citing this paper, the paper had been cited several more times in the formal literature, and they suspected that these authors were finding the source in the Wikipedia article. So we removed it, because editors wanted to influence future sources. OTOH, I don't think this was a case of trying to push content up the hill, away from what sources say. I think it was an attempt to push the content towards "alignment with the sources I happen to be familiar with" (e.g., an oncology textbook that emphasizes the importance of chemotherapy as one part of a comprehensive treatment plan, but never tried to calculate the isolated benefit of just the one part).
   I assume that the same thing happens in many areas: The politician wants the article to say "known for freedom" instead of "known for conspiracy theories"; the company wants the article to say "fourth largest producer of widgets" instead of "responsible for the massive pollution scandal"; and presumably everyone on all sides of PIA want their side to be held up as the One True™ Righteous Cause. But I still think that the article will be better if people from all those sides talk it through, than if I get to decide on my own what the article(s) will say. (There's an open question about whether we will be better off, as these conversations are doubtless wearing and sometimes painful for the community members involved, but I am convinced that WP:YESPOV is correct about what's best for the article.)
4. Strict rule enforcement is also wearing and sometimes painful, and I'm not sure that it's always humane. When someone is terribly upset, it might be better to gently and privately encourage them to take a break than to sanction them for experiencing reasonable emotions. Right now, we likely have people editing PIA articles who are mourning family members, and many more who are a little more distantly but still significantly affected. When people are being told that their lives don't matter because they have the 'wrong' ethnicity, religion, or nationality, we can't expect them to be totally impersonal about it. The 2024 United States presidential election is being written by editors who are afraid about what will happen if the wrong guy gets re-elected, based on sources written by authors who are afraid about the same thing, and we expect this to just get worse. We have some young trans editors who are very upset about the Cass Review, which looks like a harbinger of a worldwide shift in the way gender care has been handled, and they're watching this unfold and doubtless thinking that the very thing they valued most when they were trans teens is being denied to the next generation, just because it has only been proven to be very, very, very wanted by the trans teens, which the establishment doesn't care about, and not proven to have certain practical long-term outcomes that the establishment does care about. We could kick all of these affected people out of these articles, but (a) that leaves us with a smaller population working on them, (b) it leaves us with fewer voices for certain POVs, (c) we'll probably end up with worse articles as a result, and (d) it doesn't seem very kind – or effective, when you think about the socking problem – to tell people to go away from the thing they most care about.

WhatamIdoing (talk) 08:02, 13 July 2024 (UTC)

Is this intended only for AI area? ECP is broader than that, although AI is Arbcom specified.Selfstudier (talk) 13:39, 10 July 2024 (UTC)

It would have to be all areas - I don't think there is a reasonable way to separate it. BilledMammal (talk) 20:02, 10 July 2024 (UTC)

I'm not sure this would make much of a difference. The "half in the mainspace" rule will likely make no significant difference, as editing the mainspace is what newer editors do. The "big edits" rule can be gamed; just hang out at RecentChanges for a few days to find blanking vandals, or run a script to add citation templates to articles. Two or three copies of {{cite web}} will usually produce the desired number of bytes.

ECP originally wasn't based on any sort of evidence that 500 was better than any other number, and this is also not based on any sort of evidence. I suggest that you collect some numbers. For example, we know that during 2023, about 800,000 accounts made 1+ edits (new and old accounts together). We also know that very few of those 800,000 editors have achieved ECP status. Can we figure out how many editors achieved ECP in 2023, and then how many would have done so, under the proposed rules? The first should be pretty straightforward; just figure out how many people in this list created their accounts during 2023. It's probably a bit more than half a percent of those who edited at all, so I'd expect an answer around 4K or so. I'm not sure what it would take to check the new requirements, but you only have to check them for the people who achieved ECP last year, because the ~795,000 editors who either already had it, or who didn't manage to make 500 edits yet, wouldn't be affected by these changes anyway.

If very few would be affected, then what's the point? It'd just be rearranging the deck chairs. But if a lot of editors would be affected, we will have to ask ourselves whether we truly want our small percentage of eligible editors to be even smaller. Our dispute resolution processes are built on the principle that when we have problems like POV pushing, we generally need more eyes on the situation, rather than fewer. WhatamIdoing (talk) 06:19, 14 July 2024 (UTC)

I ran some queries. There were 1968288 accounts created in 2023 (UTC), excluding any blocked with the "hideuser" option. 487317 have made at least one edit since creation (note that edit may have been in 2024). 1776 have edited at least 500 times since creation, 1773 more than 500 times. 1691 are currently extendedconfirmed; this link currently shows those 1691, plus the first from 2024 and the last from 2022 (note it displays the creation dates in your configured timezone rather than UTC). There are 106 accounts with at least 500 edits that aren't in the group, presumably most because they haven't made a 501st edit since reaching the 30 days. The 21 accounts in the group without 500 edits were manually granted extendedconfirmed. Anomie⚔ 13:40, 14 July 2024 (UTC)

Thanks. Is there any way to test those 'new' (2023) extended-confirmed accounts against the proposed rules, without having to do it manually? It's possible to do 2K if we have to, but that could easily take a week of full-time work. WhatamIdoing (talk) 23:49, 14 July 2024 (UTC)

Sure, someone could write code implementing whichever checks are proposed, and then run that list of users through the check. The above discussion is a bit too TL;DR for me personally to want to try to extract what the proposal(s) is/are though. But as a quick check, it seems 401 of the now-1693 extendedconfirmed accounts registered in 2023 have made fewer than 500 mainspace edits, and 235 have made fewer than 500 edits when excluding only the User and Draft namespaces (294 when also excluding User talk and Draft talk). Here's a query listing them: quarry:query/84853. Anomie⚔ 12:07, 15 July 2024 (UTC)

Thanks v much for putting some numbers to this. Selfstudier (talk) 12:16, 15 July 2024 (UTC)