Wikipedia talk:New pages patrol/Reviewers

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia


Very low unreviewed pages backlog: 1399 articles, as of 12:00, 30 March 2023 (UTC), according to DatBot

Very high unreviewed redirects backlog: 12518 redirects, as of 12:00, 30 March 2023 (UTC), according to DatBot


Shortcut

NPP backlog[edit]

(how to use this chart)

Skip to top
Skip to bottom

Script to highlight unreferenced passages[edit]

Unreferenced passages are highlighted in red.

I wanted to let you know that I wrote a script to highlight unreferenced passages. It may be useful to New Page Reviewers to get a first quick impression of whether a new page or a draft lacks references and where the problems may be. The script does not understand articles and is only meant to assist editors, not to replace their judgment. More information can be found at User:Phlsph7/HighlightUnreferencedPassages, including instructions on how to install and use it as well as information on its limitations. Questions and feedback on problems or new ideas are welcome. See also here and here for similar discussions. Phlsph7 (talk) 14:29, 7 February 2023 (UTC)Reply[reply]

installed, very cool...--Ozzie10aaaa (talk) 19:13, 7 February 2023 (UTC)Reply[reply]
Feel free to add this to WP:NPPSCRIPTS –Novem Linguae (talk) 19:32, 7 February 2023 (UTC)Reply[reply]
Thanks for the feedback and the info, I've added the script there. Phlsph7 (talk) 08:48, 8 February 2023 (UTC)Reply[reply]
Thanks for the script! VickKiang (talk) 09:43, 10 February 2023 (UTC)Reply[reply]
Thanks for (yet another) useful script Josey Wales Parley 21:00, 14 February 2023 (UTC)Reply[reply]

Resilience in dealing with frustration[edit]

I am wondering if there are any NPPs here willing to share their most effective ways of responding to/dealing with the frustration and personal attacks of editors in a thoughtful and non-escalating way. While it is often possible to effectively communicate with editors, it can get difficult and challenging when these exchanges get personal. Admins, of course, deal with these problems all the time, though I think that NPPs are often subject to a different kind of saber rattling where their authority is questioned even more aggressively. Any thoughts or experiences are welcome. Thank you! Ppt91 (talk) 18:03, 9 February 2023 (UTC)Reply[reply]

There are many more prolific reviewers than me (@Onel5969: of note!), but what works in my case is walking away for a while and doing something else when the fur begins to fly. I guess it is easy to say something like that, but it may be harder to actually do. I firmly believe that a successful Wikipedia editor (and NPP member) will have the hide of a rhinoceros and the heart of a dove. My $0.02, for what it may be worth. Geoff | Who, me? 18:15, 9 February 2023 (UTC)Reply[reply]
Focusing on easy reviews is one strategy to avoid conflict. Easy reviews include very obvious notability passes such as species articles, and politicians that pass WP:NPOL. Another strategy is, when questioned, to be willing to get a third opinion or engage in a consensus process (such as AFD). This helps to make sure that your interpretation of policy is correct, and it can be helpful for the other person to hear it from multiple people. Shoot me an email with details of your conflict if you want some additional tips. I couldn't find it on your talk page. Hope this helps. –Novem Linguae (talk) 18:44, 9 February 2023 (UTC)Reply[reply]
There's also a list of easy reviews at Wikipedia:New pages patrol/Reports/Easy reviews. Thanks. VickKiang (talk) 09:43, 10 February 2023 (UTC)Reply[reply]
@Glane23@Novem Linguae@VickKiang A belated note with many thanks. I appreciate it! Ppt91 (talk) 01:16, 18 February 2023 (UTC)Reply[reply]
In addition to all the advice above, I would also suggest avoiding fast-paced engagement. The person at the other side might also need time to cool down. So it's ok that your first response is relatively quick, but space out subsequent follow ups. Consider delaying your second reply by one day or even more. MarioGom (talk) 15:29, 18 February 2023 (UTC)Reply[reply]
@MarioGom Very good point. Thanks. It's easy to get entangled in a back-and-forth and inadvertently escalate despite one's best efforts to do the exact opposite. Ppt91 (talk) 22:38, 18 February 2023 (UTC)Reply[reply]
A few points: be sure you understand the nature of your decline. Whether the subject fails an SNG or just GNG, be sure you can explain what it is and why. If you're circumspect in your review you should have no problem. Then, just explain this issue to the author/ submitter. Your decline or rejection isn't personal and no one should have hurt feelings. That said, please have compassion for the editor who, whether paid or not, is somehow invested in the draft. They know little about Wikipedia and will often express anger, accuse you of bias, and make threats. If you know this is routine you might not let it bother you. Finally, don't be invested in being right or defending your decline. If the author questions you about it, explain your rationale and then leave the matter. If the author re-submits let someone else have a crack it. Anytime editor behavior crosses a line, report to COIN or ANI as needed. Stop reviewing when you need a break from vitriol; come back when you want to. This is just business; please treat it so. Chris Troutman (talk) 03:28, 9 March 2023 (UTC)Reply[reply]

Review-of-reviews needed after NPR removed for copyvio[edit]

I recently removed the NPR status of SuperSwift for repeated coypright violations. His misunderstanding (AïngGF) of the copyright policy means that none of their past reviews can be considered complete. His own articles will be handled by CCI, but for the ones that he reviewed, we should make sure that, at a minimum, each is clean according to Earwig's. Seems easier to do it this way than to reënqueue everything.

Reviews All checked

I've started us off with 5, of which 2 needed rewording. A 40% hit rate, even if it's for relatively minor copyvio, suggests it's worthwhile to check the whole list. -- Tamzin[cetacean needed] (she|they|xe) 22:19, 11 February 2023 (UTC)Reply[reply]

I came across this list per a request for help at Wikipedia:Discord. Akeem Sirleaf has a lot of copyvio from the North Carolina A&T source and present in the first revision. I'll work through it as it's a relatively short article. MrLinkinPark333 (talk) 22:34, 11 February 2023 (UTC)Reply[reply]
I see that @Sennecaster: went through Akeem Sirleaf already. MrLinkinPark333 (talk) 00:08, 12 February 2023 (UTC)Reply[reply]

Thank you so much, everyone! Three stray bits:

  1. @Glane23: I think something went wrong here? If you meant to change something at Hong Kong Hockey Association, it did not save.
Got it. Very slight reword for a 26% hit - don't know what happened, but saved this time. Geoff | Who, me? 13:33, 12 February 2023 (UTC)Reply[reply]
  1. @Paradise Chronicle: I don't see copyvio there, but, in terms of good article-writing, the quote usage at Killing of Andrew Brown Jr. § Police statement is inappropriate and should be rephrased.
  2. I pared back David Lanre Messan quite a bit, for a mix of copyvio, sourcing, and promo reasons. What's left makes me think it should probably go to AfD, but I haven't a businessperson AfD in some time and I'd welcome a second opinion.

-- Tamzin[cetacean needed] (she|they|xe) 05:49, 12 February 2023 (UTC)Reply[reply]

Add Capability to go through Drafts?[edit]

Currently, the new pages list filter allows you to go through mainspace or userspace. I’d like to add drafts to that, as I’m now doing some work in draftspace trying to head off problem editors early in the process (guidance and/or more significant action) before it gets to mainspace. This is trying to work upstream, which is generally a good thing, process-wise. Would others support this change? Seems like it should be simple, since the menu is already in place to switch namespaces.

Note: I tend not to worry too much about significance or wording in draftspace: mainly looking for serious issues like copyvio, ChatGPT use, paid editing, etc. — rsjaffe 🗣️ 17:11, 12 February 2023 (UTC)Reply[reply]

Whilst copyvio is still an issue in draftspace (as per WP:NPPDRAFT), isn't patrolling draftspace a bit out of scope for NPP? I wouldn't object to an extra option on the menu though. -Kj cheetham (talk) 18:06, 12 February 2023 (UTC)Reply[reply]
rsjaffe, Special:NewPagesFeed has a toggle to Article for Creation with its own filtering options which includes unsubmitted. Or are you looking for a broader category of drafts?Slywriter (talk) 18:10, 12 February 2023 (UTC)Reply[reply]
I mean adding draft here:
New page curation screenshot
— rsjaffe 🗣️ 19:34, 12 February 2023 (UTC)Reply[reply]
Screenshot of Options
- do you not have this toggle? It seems like these are the options you are looking for. Or I am thoroughly misunderstanding. Slywriter (talk) 20:42, 12 February 2023 (UTC)Reply[reply]
No, I don't have that. Is that from AfC reviewing? — rsjaffe 🗣️ 21:27, 12 February 2023 (UTC)Reply[reply]
Assume so, I have both NPP and AfC rights. Thought NPP could see it without AfC right but guess I am wrong there. Slywriter (talk) 03:40, 13 February 2023 (UTC)Reply[reply]
No, everyone can see it. @Rsjaffe: Just click the "Articles for Creation" radio button, as shown in the top right of Slywriter's screenshot. -MPGuy2824 (talk) 03:47, 13 February 2023 (UTC)Reply[reply]
Now I see it, but the side tool doesn't show up like it does for mainspace and userspace, like it does with the dropdown. That makes scanning drafts much more difficult. — rsjaffe 🗣️ 21:17, 14 February 2023 (UTC)Reply[reply]

Unreviewing new pages[edit]

Hello, could anyone point me to an official guideline or stance of this project on unreviewing new pages?

Yesterday I marked a page as reviewed [1], but it seems that Onel5969 (talk · contribs) has decided to mark the page unreviewed. I find it remarkable that someone can do this, especially when the original patrolling editor was an administrator. It seems that the NPP process is being weaponized to push a certain view of notability and to second-guess or even !supervote the decisions of established editors. Rschen7754 19:09, 16 February 2023 (UTC)Reply[reply]

I will also note similar misbehavior here and here where other NPP reviewers later re-marked the pages as reviewed after Onel5969 had unreviewed them. --Rschen7754 19:15, 16 February 2023 (UTC)Reply[reply]
For what it's worth, I marked those as reviewed when they were redirects and they were later changed back to an article. That's part of why there will sometimes be multiple instances of people marking a page as reviewed. Hey man im josh (talk) 13:02, 17 February 2023 (UTC)Reply[reply]
The whole point of marking a page as unreviewed is triggering another review. So it's normal that whenever some pages are unreviewed, they are marked as reviewed again. MarioGom (talk) 17:02, 18 February 2023 (UTC)Reply[reply]
I'm amazed the above post is from an admin. Perhaps maybe review WP:NPA and consider whether your word choice here is quite poor and an insult against a hard-working volunteer. Slywriter (talk) 19:25, 16 February 2023 (UTC)Reply[reply]
Please indicate to me what portion is a personal attack. --Rschen7754 19:26, 16 February 2023 (UTC)Reply[reply]
weapononizing, supervote, second-guess. All in context of one editors reviews. Seems unnecessarily loaded language if you were looking to have a policy discussion. Slywriter (talk) 19:30, 16 February 2023 (UTC)Reply[reply]
I wasn't going to dredge up the previous discussions, but I suppose I will have to: [2]. --Rschen7754 19:32, 16 February 2023 (UTC)Reply[reply]
Ah, so a re-hash of an old disagreement. I don't see a community resolution in that thread, just a multitude of different opinions and suggestions that an RfC is needed. And it would have been better to lead with that since it provides context. Slywriter (talk) 19:39, 16 February 2023 (UTC)Reply[reply]
My overall point is that just as I can't use my admin tools to further a content dispute, NPP tools should not be used to further one either. Of course Onel5969 is free to take it to AFD, though given Wikipedia:Articles for deletion/M1 (Durban) that might not be a great call. --Rschen7754 19:40, 16 February 2023 (UTC)Reply[reply]
Wikipedia:New_pages_patrol#Unreviewing is the only "official" guideline I can think of for unreviewing. -Kj cheetham (talk) 19:43, 16 February 2023 (UTC)Reply[reply]
No editor, even an administrator, is above anyone else, which is why I find a comment like "especially when the original patrolling editor was an administrator" shocking. There is no issue with objecting to a review and sending back into the pile for another opinion. Curbon7 (talk) 19:50, 16 February 2023 (UTC)Reply[reply]
Reading your own guideline, it seems that the purpose of this was only to fix errors from a wayward patroller, not to further a content dispute. --Rschen7754 19:53, 16 February 2023 (UTC)Reply[reply]
I think one piece of context that is probably missing from all of this is the culture "older" NPP (myself included) have around peer review. I know I've had @Onel5969 unreview things I've done and that's something I'd just expect as par for the course. But I completely understand why it felt really different to Rschen. I think the piece that was missing here was a comment from Onel following the unreview about what the goal with the unreview was. I'm guessing the goal was a sort of WP:3PO but it would have been helpful to articulate that. Best, Barkeep49 (talk) 21:03, 16 February 2023 (UTC)Reply[reply]

This looks like a disagreement on whether or not it should be a separate article. Sending it back for re-review by a different set of eyes might been seen as a mild way to approach it rather than a trip to AFD to possibly decide on making it a redirect. There probably isn't something that definitively says whether or not such is an OK use of unreviewing. North8000 (talk) 20:36, 16 February 2023 (UTC)Reply[reply]

Anyone is welcome to unreview any page that I’ve reviewed. This is a collaborative project and my opinion doesn’t count for more than anyone else’s. If I’ve made a poor judgement about notability I’d hope someone else would put it right. I recall that Onel5969 has said more or less the same thing - if anyone disagrees with his reviews they should go ahead and undo them. Mccapra (talk) 21:09, 16 February 2023 (UTC)Reply[reply]
agree w/ Mccapra--Ozzie10aaaa (talk) 21:40, 16 February 2023 (UTC)Reply[reply]
+1 signed, Rosguill talk 21:56, 16 February 2023 (UTC)Reply[reply]
I feel the same way. I'd like to know what I missed so I can do better. Hey man im josh (talk) 13:04, 17 February 2023 (UTC)Reply[reply]
Same here. MarioGom (talk) 17:04, 18 February 2023 (UTC)Reply[reply]
Also I'll note that I have draftified some articles that were previously marked as reviewed by Onel, and I don't recall any instance where I got pushback for that. MarioGom (talk) 17:05, 18 February 2023 (UTC)Reply[reply]
  • Comment - I would not have unreviewed the page if it met either GNG or the SNG for roads. But it is not a state road, so notability is not presumed. Therefore it needs to pass GNG, and with the current sourcing of only google maps (which may or may not be considered reliable), and a second, unreliable source, there is not a single in-depth source about the road. This does not even come close to meeting GNG. The question should not be why did I unreview it, but why would any reviewer mark it reviewed in its present state. I have no agenda, other than that articles be adequately sourced so that they pass GNG (or an appropriate SNG) and VERIFY. The disagreement I had previously with many of the folks over at the road project was regarding VERIFY, that is not the case this time. I also find it troubling that Rschen thinks that being an administrator gives them some type of super!vote, since they have once again marked the article reviewed. Mccapra is absolutely correct, none of us is perfect, and any of my reviews is okay to be undone, as I've said to several other reviewers over the years. However, that being said, Barkeep49 is correct, and I probably should have left a message on Rschen's talkpage, but I thought my tagging it with a notability tag would have sufficed. Since they have removed the notability tag, I would appreciate another uninvolved reviewer to take a look at the page and see if they feel, with the current sourcing, it meets GNG. Onel5969 TT me 21:51, 16 February 2023 (UTC)Reply[reply]
    • WP:NEXIST is part of the notability guidelines. Given the outcome of Wikipedia:Articles for deletion/M1 (Durban) and the fact that this is a similar road in an urban area, we can expect that there will be sources and that GNG is met. This is a difference of opinion and the discussion really shouldn't be at this page at all, except that it involved use of the NPP tools. --Rschen7754 01:12, 17 February 2023 (UTC)Reply[reply]

(edit conflict)

It seems to me that per WP:INVOLVED, specifically about which they have strong feelings, Rschen7754 should recuse themself from administrative actions in articles concerning roads. I understand that marking pages as reviewed is bundled with admin rights, therefore an administrative action. Agree with Slywriter about WP:NPA and would also include WP:ADMINCOND.
With regard to M41 (Durban) and M25 (Durban). A second reviewer did indeed mark them as reviewed, but significantly added the {{Sources exist}} template in both cases, i.e. the article, as is, doesn't demonstrate notability.
It seems to me that unreviewing an article so that another reviewer can give a second opinion is perfectly reasonable. Marking an article as reviewed a second time after it has been unreviewed could well be seen as a form of editwarring.
As somebody uninvolved in any of the articles mentioned above, I'm sorry to say I see very biased interpretations of the facts and an 'I'm an admin' attitude. --John B123 (talk) 22:33, 16 February 2023 (UTC)Reply[reply]
I could have added the {{sources exist}} template, but would it have made a difference or would the article have still been marked unreviewed?
I could say that I am a fellow reviewer, but that according to MediaWiki is technically not the case. I also find it troubling how my review actions are being considered INVOLVED just because I am an administrator who happens to edit in the same subject area, yet any non-admin editor can perform the same actions and even edit war on the same article [3][4] and it can not be considered INVOLVED. In fact, the first sentence says In general, editors should not act as administrators in disputes in which they have been involved. and I don't see any use of NPP or rollback etc. as acting as an administrator.
I have seen too many cases where GNG and the SNG for roads were not followed by NPP [5] and editors were chased away and content was lost. (And that's not intended to be a negative reflection on the purpose of NPP. I'm aware of how much junk comes into small wikis and I know that it's a lot worse on this wiki.) I am however, concerned that certain perspectives and interpretations of policy are being pushed through NPP that are much higher than what the actual policies say - such as omitting NEXIST above and tagging or redirecting state highways for notability against GEOROAD. If you don't believe state highways are notable or that NEXIST is ridiculous, then fine - but don't use the NPP tools or project to further your interpretations of policy.
As far as the third review action: there apparently is no written expectation. If there was I would follow it.
And FWIW while I still find the "unreviewing" troubling, notably because it does take the page out of Google, and it seems a gross expansion of the scope of NPP which I thought was targeted towards newer editors - I do appreciate @Barkeep49:'s perspective about it not being seen as a Big Thing under ordinary circumstances. If I used the (actual) admin tools in the way the review tool was used on that article, I would not be an admin anymore, end of discussion. So indeed, there is a bit of culture disparity. --Rschen7754 02:29, 17 February 2023 (UTC)Reply[reply]
I legitimately don't see how the roads above meet the questions buried in Wikipedia:Notability (highways) under United States that give good guidance for all road articles -

Specifically, the article should answer the question, "why was this road built in the first place?", and "why are the taxpayers asked to keep spending money to keep the road maintained?" If the article does not answer the question of why does this road exist, that is grounds for deletion of the article.

These articles are prose describing a map. They shouldn't have been marked as reviewed, especially with no tags because they are not encyclopedic as they stand and make no claim to notability. That AfDs are coming to a different conclusion is troubling or shall we assume only US road articles are expected to meet this standard? Slywriter (talk) 02:54, 17 February 2023 (UTC)Reply[reply]
That is an essay that has numerous problems including the text that you have pointed out. I agree that road articles should explain those things, but I do not agree that it is grounds for deletion if they don't. And pragmatically, that means that you're expecting a near-perfect article on the first draft. Regardless, this is borderline off-topic. --Rschen7754 03:05, 17 February 2023 (UTC)Reply[reply]
Problematic or not, we link to it directly at WP:GEOLAND roadways, which makes it a pretty strong essay. And it does connect to your concerns about NPP as an article in mainspace needs to establish notability at publication, otherwise it should be tagged, draftified, prodded or AfDed. A NPPer doing none of these hasn't followed the process(we even have File:NPP_flowchart.svg to show this). Slywriter (talk) 03:34, 17 February 2023 (UTC)Reply[reply]
WP:NEXIST is an official guideline, however, that says that is not the case. NPP "guidelines" don't supersede policy. --Rschen7754 03:39, 17 February 2023 (UTC)Reply[reply]
WP:NRV is also a guideline. Directly above NEXIST and comes to a different conclusion when applied to the above road articles.Slywriter (talk) 03:51, 17 February 2023 (UTC)Reply[reply]

While I don't share Rschen's sentiment behind starting this thread, it might be a good idea to have a process by which if Person B in unreviews an article Person A reviewed, Person B has to tell person A why they did it. I'm fine with holding each other accountable, but there has to be discussion that goes along with it. -Fredddie™ 23:30, 16 February 2023 (UTC)Reply[reply]

I'm all in favour of reviewers feeling welcome to unreview each other's pages. There is no one better placed to notice when you slip up than your fellow reviewers, and we should be thankful for someone going to the effort of double-checking our work. (Admin status means diddly-squat in this regard, so let's just pretend that hasn't been said.) In my experience, the fact that you get an automatic notice about "I have unreviewed page X" has always been sufficient to get a discussion started if there was need for one. Sensibly to be supplemented by a bespoke explanation if the reason seems non-obvious. --Elmidae (talk · contribs) 07:43, 17 February 2023 (UTC)Reply[reply]

So there are sort of two questions here. On the "unreview" I thinks folks here agree that it was OK to do so but should have had an explanation. The other is the status of the article. My own opinion is it should not be passed. Sources and content are just a description of the map. For several reasons I would not agree that the result on the other article is a reason to pass. Sincerely, North8000 (talk) 15:49, 17 February 2023 (UTC)Reply[reply]

I've not been terribly active for a while, but pitching in here FWIW. If anyone unreviewed a page I'd reviewed and a) I cared deeply about it and b) it didn't seem to be for an obvious reason and possibly as a last filter c) the unreviewer was reputationally unknown to me, I'd ask 'em why they unreviewed it. A lot less work - and less tiresome - than whacking a thread in NPP/R... Best Alexandermcnabb (talk) 04:21, 21 February 2023 (UTC)Reply[reply]

When should newcomers be prompted to add sources?[edit]

Hi y'all – the conversation above about the nifty script @Phlsph7 wrote, led me to think this might be a good place to start a conversation about the kinds of edits that warrant references being included within them.

Specifically, I'm curious to learn: in what cases do you think someone adding new text to an existing article would NOT warrant them accompanying that new text with a reference? E.g. when someone is adding new content to the lead section of an article, per WP:LEADCITE. When someone is adding new content to plot sections, per WP:PLOTCITE. [i]

A screenshot of an early design of the initial reference check experience (mobile)

I ask the above on behalf of the Editing Team team who is beginning work on a new project that is intended to offer people who are new to editing Wikipedia actionable feedback when the edit(s) they are in the midst of making could defy policies.

The first "check" will prompt people to add a reference when they are attempting to publish an edit that involves them adding new content to an existing article. See proposed UX that's pictured.

In line with the above, we're trying to craft the initial set of "rules" that will determine when this "check" is triggered, thus the question I'm posing here.

Okay, please let me know if there is any additional context that I might be able to offer that would make it easier for you to engage with the question above. And of course, if you think there is a better place for me to ask this question, please let me know.

---

i. Thank you to @Sdkb for sharing these examples. PPelberg (WMF) (talk) 02:53, 17 February 2023 (UTC)Reply[reply]

Comments from Sdkb[edit]

Here are two other instances I didn't think to mention in my Phabricator comment: disambiguation pages and external link sections.
If you're looking to role out the feature gradually, you could also approach this from the other end, of what definitely should have a source added. For that, it's helpful to look at the categories at WP:BURDEN:
all quotations. It seems perhaps tricky but not impossible to computationally determine when a quote is being added.
all material whose verifiability has been challenged. This would be reverting someone who has removed your addition to restore that addition.
all material that is likely to be challenged. This might be tricky to algorithmically determine. Perhaps articles in categories that indicate contentious topic areas, or using a semantic analysis of some sort to predict.
all contentious matter about living and recently deceased persons. "Contentious" might be tricky to algorithmically determine, but living people are easily identified via Category:Living people.
Cheers, {{u|Sdkb}} talk 06:36, 17 February 2023 (UTC)Reply[reply]
Here are two other instances I didn't think to mention in my Phabricator comment: disambiguation pages and external link sections.
Oh, excellent. I've added these additional cases/exceptions to the ticket in Phabricator (T324730) where we're drafting the initial heuristic. Thank you, @Sdkb.
If you're looking to role out the feature gradually, you could also approach this from the other end, of what definitely should have a source added. For that, it's helpful to look at the categories at WP:BURDEN...
To be doubly clear we're on the same page, when you say "gradually" are you referring to an approach that would seek to minimize false positives to start? [i]
Assuming the above to be true, I wonder what – if any – risks you could seeing in starting with this kind of approach?
A couple of initial thoughts...
  • Maybe experienced volunteers would become frustrated if Edit Check was too permissive...it was being too lenient.
  • On the other hand, I wonder if experienced volunteers would grow frustrated if they saw Edit Check was being too stringent and causing unsuspecting newcomers to add references in places that people reviewing edits would consider disruptive/unhelpful.
For context, this question is very much top of mind for the team as we think through T329988: Decide how sensitive reference check should be to start.
---
i. Said another way: to start, limit the scope of edits that would cause the reference edit check to be activated to edits that we think the largest population of people would agree warrants references and then overtime, broaden the scope to include more nuanced kinds of edits PPelberg (WMF) (talk) 19:52, 21 February 2023 (UTC)Reply[reply]
are you referring to an approach that would seek to minimize false positives to start? Almost, but not quite. The way I'd frame it is that those scenarios are the ones in which citations are the most important, so there is the most upside. The downside risk of recommending too many citations is more forgivable in that tradeoff, but not necessarily less likely.
Overall, too many references is generally less of a problem than too few, since removing them is very easy. That said, one way I could see editors becoming annoyed is if there is a false positive recommendation that keeps misleading newcomers and causing different people to add an unneeded reference over and over. This isn't currently an issue with the tool activating only when you add new text (which will only happen once for a given piece of text), but it could be if edit checks in the future start applying to existing text in an article. At that point, I think you'll want to build in some way for experienced editors to flag a passage that the tool thinks needs a citation and say, "nope, actually this is fine, stop recommending a change". Analyzing the collection of flagged passages could provide useful insights about edge cases.
Also, I should add, one risk of starting with the WP:BURDEN categories would be that those categories are filled with a lot of heated controversy and sensitive topics, so if you're worried about rocking the boat with a beta feature, maybe those aren't the places to start.
Best, {{u|Sdkb}} talk 20:12, 21 February 2023 (UTC)Reply[reply]

Comments from Elmidae[edit]

  • Generally speaking, I would suggest sticking to a precautionary principle here and identifying only those instances where referencing is very likely to be unnecessary. I suspect that approaching this from the other end (i.e. trying to determine whether referencing might be required) may be next to impossible. I'm more pessimistic than Sdkb about the "tricky" bit when trying to determine when an addition may be "contentious" or otherwise in need of referencing. Assuming the cautious approach, it seems to me that edits that should not trigger the notice might include:
- lead and plot sections, as noted
- short description
- image captions (the image metadata is supposed to contain any refs)
- gnoming - punctuation & capitalization changes, wikilinking & piping, formatting, adding external links or citations (natch), and possibly a cut-off for altering/adding only 1-3(?) characters
- if it is feasible to hook this into a spelling or grammar checker, then edits that fix such issues could be identified and excluded
--Elmidae (talk · contribs) 08:19, 17 February 2023 (UTC)Reply[reply]
Generally speaking, I would suggest sticking to a precautionary principle here and identifying only those instances where referencing is very likely to be unnecessary.
@Elmidae would it be accurate for me to understand you using "precautionary" in this context to mean something like the following?
"I think it would be prudent for this 'reference check' to be more sensitive to start?"
Assuming the above is accurate, can you please say a bit more about what is leading you to see more value in implementing a more stringent filter to start?
One idea: maybe you're thinking that minimizing the risk of people adding new content without references is of greater value/concern than the risk of experienced volunteers growing frustrated/becoming disrupted as a result of people adding sources in places that don't warrant them.
Assuming the cautious approach, it seems to me that edits that should not trigger the notice might include...
In the meantime, I've added the edit types you named that should not cause the reference check to become activate to phab:T324730.
Oh! And thank you for sharing this context...I think this might be the first time we're talking? If so, my name is Peter. I work as the product manager for the Editing Team ^ _ ^ PPelberg (WMF) (talk) 21:04, 21 February 2023 (UTC)Reply[reply]
@PPelberg (WMF): sorry for the delay! Yes, by "precautionary" I meant that it seems advisable to me that this process start with the supposition that any given edit may need a source, then whittle that down by excluding those cases where none may be needed. This is based on the impression that the other way round (assume that an edit will not need a source, then try to find those that do) is much more tricky and fuzzy to parse. Plus, false positives (excessive suggestions to source) I would consider to be less of a downside than false negatives (missing instances where a suggestion should have been given). Admittedly this is coming from the perspective of an editor who will not have to see these suggestions, so take with a grain of salt :p --Elmidae (talk · contribs) 07:57, 23 February 2023 (UTC)Reply[reply]
I appreciate you following up with the thinking that informed the "precautionary principle" you named above, @Elmidae and I'm sorry it's taken me two weeks to return here!
To bring a bit of closure to this particular decision, we're going to start with an approach with Edit Check that minimizes false positives and is implemented in ways (T327959) that empower volunteers, on a per-project basis, to evolve the heuristic (T324730) to become more robust/complex (read: minimize false negatives) over time.
Note: you can see the thinking that informed the above and the trade offs we're accepting in moving forward with this approach in T329988#8654867. PPelberg (WMF) (talk) 01:40, 9 March 2023 (UTC)Reply[reply]
I see that Phlsph7 below weights this just the other way round. Is there some general idea within the team on what is considered the bigger issue - excessive suggestions or too few suggestions? --Elmidae (talk · contribs) 07:57, 23 February 2023 (UTC)Reply[reply]
@Elmidae: great question and you asking it is creating an opportunity for me to make explicit what was was implicit in what I shared above...
To start, we think excessive suggestions is the bigger issue.
This thinking is built on the following assumptions (copied from phab:T329988#8654867):
  • Following an approach that minimizes false positives will increase the likelihood that newcomers and Junior Contributors will intuitively see the prompts Edit Check is presenting them with as relevant/applicable to the change they're wanting to make. As a result, we assume these people will be:
    • More likely to trust/consider/engage with the guidance/prompt Edit Check is presenting them with
    • Not be further discouraged from returning to edit again because they found the experience to be straightforward
  • Newcomers and Junior Contributors are going to be less likely to detect false positives and therefore, more likely to become confused by them (read: Edit Check being triggered unnecessarily)
  • Experienced volunteers will be more motivated to improve upon/contribute to a simple system that works in expected ways most of the time rather than a complex/robust system that works some of the time.
    • See: Gall's Law: "A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over with a working simple system."
  • The Editing Team, in partnership with volunteers, will feel empowered to make Edit Check more robust over time (and therefore minimize false negatives) because doing so will feel relatively fast and lightweight. Where "fast and lightweight" in this context mean the Edit Check heuristic (T324730) will be:
    • Represented in natural language (as well as in code) to increase the range of volunteers who can audit and iterate upon the rules that comprise it
    • Configurable on a per-project basis so that volunteers can ensure Edit Check conforms with local conventions and policies (T327563) without needing consider how these decisions could impact other projects
    • Auditable on a per-project basis (T327959) so that volunteers can evaluate how Edit Check is performing "in production" (T324733) and adapt the heuristics in ways that increase the likelihood that it is causing people who are new to accompany the new content they are attempting to add with a reference (T325713).
PPelberg (WMF) (talk) 01:46, 9 March 2023 (UTC)Reply[reply]

Comments from Phlsph7[edit]

Such a prompting system sounds like a great idea since many inexperienced users are not familiar with the requirements for references on Wikipedia. The technical problem seems to be similar to the problem faced by the script to highlight unreferenced passages: the newcomer should be prompted if the text they intend to add would be highlighted. The most basic way to implement such a prompting system would be: check whether the added text contains a reference; prompt if it doesn't and don't prompt if it does. This approach would result in many false positives (as when a summary of a referenced section is added to the lead) and false negatives (as when two paragraphs are added at the same time but references are only provided for one).
As was already mentioned, this could be refined depending on where the paragraphs are added. My script currently excludes the lead and sections with the following names: 'Further reading', 'See also', 'External links', 'References', 'Bibliography', 'Notes', 'Selected publications', 'Selected works', 'Plot'. But I'm sure more could be added. If the text is added as part of an existing paragraph, one could check whether this paragraph has references and "citation needed"-tags and whether the text is added before or after the references. Regular paragraphs and probably lists should be checked but image captions and individual table cells in most cases not. There are many different templates that could be taken into account. For example, Info-boxes usually do not provide references for every single fact while for quotation templates, references are essential. Another relevant factor could be the size of the added text: if someone adds 5 words to a paragraph, chances are that this is just meant to clarify an existing statement and to introduce a new one. A further question would be how to deal with explanatory footnotes in contrast to actual references. Even a refined system is bound to lead to false positives so prompt messages should be carefully formulated to take this into account. Phlsph7 (talk) 08:46, 17 February 2023 (UTC)Reply[reply]
@Phlsph7: One section name you can add to the exclusions is "Synopsis," which is preferred over "Plot" for nonfiction works. {{u|Sdkb}} talk 18:07, 17 February 2023 (UTC)Reply[reply]
It's probably not worth pinging newcomers about citing, but I always thought you were supposed to cite the content of nonfiction works. I've always made sure I cited a page range in the nonfiction book, a book review, or some other source. (t · c) buidhe 18:20, 17 February 2023 (UTC)Reply[reply]
@Sdkb: Thanks for the info. But I think Buidhe is correct that WP:PLOTCITE and WP:PLOTSOURCE only apply to works of fiction. Phlsph7 (talk) 08:36, 18 February 2023 (UTC)Reply[reply]
I think it somewhat depends on the work and one's personal editorial views (e.g. this essay doesn't necessarily have full community buy-in). When I was writing Boys_State_(film)#Synopsis, for example, I didn't feel compelled to source the synopsis section — the documentary is telling a story, and the only way to source it would have been to cite reviews that just used the film itself, which would have been rather redundant. Some synopsis sections may need references, but I don't think we can be so certain that all or almost all need them that we should have tools like edit check or your script flag them. Cheers, {{u|Sdkb}} talk 08:45, 18 February 2023 (UTC)Reply[reply]
If a passage is highlighted, it only means that there could be a problem, not that there definitely is a problem. In this regard, the script is not meant to replace human judgment but only to assist users in finding passages that may need references.
First off, @Phlsph7, I feel inspired by how you've described the HighlightUnreferencedPassages tool. Specifically, I appreciate how by framing the tool as you have, I (someone who might consider using it) start to think about it as a way of helping me direct/allocate my attention rather than trying to relieve me of exercising judgement altogether and just setting me up not to have my expectations met.
Now, I'm going to respond to the points you raised above in separate comments in an effort to make exploring each a bit easier... PPelberg (WMF) (talk) 00:04, 22 February 2023 (UTC)Reply[reply]

False positives and false negatives[edit]

The most basic way to implement such a prompting system would be: check whether the added text contains a reference; prompt if it doesn't and don't prompt if it does. This approach would result in many false positives (as when a summary of a referenced section is added to the lead) and false negatives (as when two paragraphs are added at the same time but references are only provided for one).
Well put and I agree with what you described above. A resulting question: what risks can you see with a tool like Edit check having too many false positives?
I ask the above with two thoughts in mind:
1. I assume you've spent some time thinking about where on the imagined spectrum of "too many false positives" and "too many false negatives" you think HighlightUnreferencedPassages existing would make it most useful
2. As I mentioned above [i], the Editing Team is actively thinking about how sensitive to make the first edit check to start. PPelberg (WMF) (talk) 00:05, 22 February 2023 (UTC)Reply[reply]
I think there is an important difference between my script and your project: scripts are used mainly by experienced users while your targets are inexperienced users. I assume that, for experienced users, having more false positives is not so much of a problem since they are able to discriminate between true and false positives on the spot. But false negatives are more of a problem since more time for an in-depth reading would be required to spot them. In this regard, the red color is just a tool to direct their attention. But new users may naturally rely much more on what an automatic message prompt says. False negatives will be handled as before: by other editors who check the edits later. False positives, on the other hand, may confuse the new user and make it less likely for them to continue editing since they may not know what to do and may not want to spend time to figure it out. So I speculate that, in your case, a more conservative approach would be better, i.e. having fewer false positives with the risk of having more false negatives. This may also help you avoid or reduce criticism by experienced users who do not like this change. But, of course, getting the right ratio between false positives and false negatives is a balancing act. Depending on the feedback and how well it works, you could slowly expand to cover more cases.
Two factors are how reliable the detection is and how severe the violation of WP:V is. For example, adding a full paragraph without sources should be easy to detect and is a severe violation. But adding an unsourced entry to an otherwise sourced list is more difficult to assess (do the sources of the subsequent entries cover it or not?) and is probably less severe. My script would highlight it but you may decide not to if the list has sources at the end or the beginning.
As a side note: you may also find the discussion of my script at Wikipedia_talk:Good_article_nominations/Archive_27#Script_to_find_unreferenced_passages helpful to judge some reactions to false positives. Initially, the script did not exclude the lead section, which was criticized. Phlsph7 (talk) 09:15, 22 February 2023 (UTC)Reply[reply]
Mmm, well put and I agree with everything you stated above@Phlsph7.
In fact, I started to quote and respond to specific statements you made only to realize halfway through that I was agreeing with and restating all of what you already said.
We (the Editing Team) met today to decide how sensitive the reference check ought to be to start with and ended up converging on what you describe here:
...a more conservative approach would be better, i.e. having fewer false positives with the risk of having more false negatives. This may also help you avoid or reduce criticism by experienced users who do not like this change. But, of course, getting the right ratio between false positives and false negatives is a balancing act.
Doing the above assumes that, as you said, we'd iterate upon this heuristic over time to iterate towards the "balance" you described. We hope this "iteration" could happen in partnership with volunteers at specific projects and we'll be thinking through the logistics of this a bit later in phab:T327959.
Wikipedia_talk:Good_article_nominations/Archive_27#Script_to_find_unreferenced_passages helpful to judge some reactions to false positives. Initially, the script did not exclude the lead section, which was criticized.
This is the first time I'm seeing this conversation; thank you for drawing my attention to it.
A few questions/ideas this conversation has me thinking about:
  1. What might we do to: A) minimize the likelihood that, as you said, people use Edit Check mindlessly and B) minimize the harm that can be caused by the subset of people who, I assume, will invariably use the tool in this way?
  2. How might we communicate the limitations of Edit Check in prominent enough ways that the experienced volunteers who are evaluating it are likely to see and consider?
  3. Might it be worthwhile for the Editing Team to consider omitting edits to lead sections from the initial reference check heuristic, as it looks like you decided to do?
Note: I do not expect you to engage with the above! I'm mostly speaking them aloud for my own memory/accountability. Of course, if you or anyone else feels inspired to join in on my quest to answer them, I'd value that ^ _ ^ PPelberg (WMF) (talk) 01:26, 23 February 2023 (UTC)Reply[reply]
Thanks for the detailed feedback! As for questions 1A and 1B: One thing you can do is to communicate the limitations clearly in the description of the project. Another thing would be to formulate the prompt messages carefully to get the editors to think about the issue instead of telling them what they did wrong. So instead of saying "this is wrong" you could say "there may be a problem". This has also certain disadvantages since simple messages are often easier to understand than carefully worded messages.
I'm not sure about question 2. One rather obvious point would be to make its purpose and limitations prominent in the project description, assuming that the editors assessing it read that description first. I guess its purpose, at least initially, is not to ensure that all edits are free from violations of WP:V but to catch the most severe violations and to make new editors aware of them.
As for question 3: What to do about the lead is a difficult decision since the lead is the most important part of the article. In many cases, WP:LEADCITE does not apply since the new content added is not referenced in the body of the article. But checking this automatically would be very difficult to impossible. If avoiding false positives is a high priority then you probably should exclude it. Or, maybe better, you decide after you have your first prototype to see how serious it is. A further way for you to handle the issue would be to make the user aware of the problem by giving a different prompt message. Something like "please either provide a source or ensure that these claims are supported elsewhere in the article (see WP:LEADCITE)". Another way would be to exclude it for new editors but still list it in the filter log so that experienced editors know what to look for. The idea would be that you have 2 filters: a weak one for prompts given to new editors and a stricter one for the filter log. This way, you minimize false positives for new editors adding content and minimize false negatives for experienced editors going through these logs.
Strictly speaking, my script does not exclude the lead in all cases: the lead is still highlighted for drafts and stub articles. The reason is that, in these cases, it's less likely that WP:LEADCITE applies. Phlsph7 (talk) 06:03, 24 February 2023 (UTC)Reply[reply]
Thanks for the detailed feedback!
You bet, @Phlsph7 ^ _ ^ And thank you for being patient with me over the past couple of weeks.
Now to the points you raised...
One thing you can do is to communicate the limitations clearly in the description of the project.
Great call. I've added a reminder to WP Talk:Edit check to hold @Whatamidoing (WMF) and I accountable to doing the above.
Another thing would be to formulate the prompt messages carefully to get the editors to think about the issue instead of telling them what they did wrong. So instead of saying "this is wrong" you could say "there may be a problem". This has also certain disadvantages since simple messages are often easier to understand than carefully worded messages.
Mmm, yes: design the prompt messages to actually prompt people to consider/think about whether the change they're attempting to make warrants a reference rather than simply telling them what to do. The latter does little to volunteers learn/internalize the ways of editing Wikipedia.
I guess its purpose, at least initially, is not to ensure that all edits are free from violations of WP:V but to catch the most severe violations and to make new editors aware of them.
Would you be okay with us using this language, or some modified version of it, on WP:Edit check or mw:Edit check? I find what you wrote to clearly and succinctly capture how the Editing Team is currently conceiving of this project...
Or, maybe better, you decide after you have your first prototype to see how serious it is. A further way for you to handle the issue would be to make the user aware of the problem by giving a different prompt message. Something like "please either provide a source or ensure that these claims are supported elsewhere in the article (see WP:LEADCITE)".
Oh, interesting. I've filed T331583 to hold the Editing Team accountable to revisiting the prospect of doing what you described above.
The idea would be that you have 2 filters: a weak one for prompts given to new editors and a stricter one for the filter log.
Two filters...can you please say more about this? Specifically, when you say ,...minimize false negatives for experienced editors going through these logs. are you meaning something like the below?
"Experienced editors will be more likely to develop trust and find value in Edit Check if they can see the edits which are likely to need references, but Edit Check prompts were not shown. PPelberg (WMF) (talk) 02:23, 9 March 2023 (UTC)Reply[reply]
Please feel free to use any of my formulations here for the project.
To me, it seems that prompting new editors to add sources and having a log of possible violations of WP:V are, in principle, 2 different projects: you could have one, or the other, or both. The filter log could have the same purpose as the prompt system (catch the most severe WP:V violations) or a different one (make it easier to find all kinds of WP:V violations). And you have 2 quite different target groups: inexperienced editors for the prompts and experienced editors for the log. The experienced editors can deal with more complicated cases and they are probably less confused by false positives (if it is clearly communicated to them that the log is far from perfect). And, of course, the more violations the log captures, the more useful it is (assuming that the false positive rate is not too high). Maybe tags could be used in the log to indicate whether the inexperienced editor was warned or not.
But having two different filters also has disadvantages. For example, it makes the project more complicated and, if the log has the same name as the prompt system, it may be confusing why it also logs cases where no prompt was given. Phlsph7 (talk) 09:19, 10 March 2023 (UTC)Reply[reply]

Excluded sections[edit]

As was already mentioned, this could be refined depending on where the paragraphs are added. My script currently excludes the lead and sections with the following names: 'Further reading', 'See also', 'External links', 'References', 'Bibliography', 'Notes', 'Selected publications', 'Selected works', 'Plot'.
Would it be accurate for me to understand the rationale for how you arrived at the section exclusions you named above as the following? I ask this seeking the policies that support these exclusions so that I include them in phab:T324730.
  • Further reading: "By contrast, Further reading is primarily intended for publications that were not used by editors to build the current article content, but which editors still recommend." | Source
  • See also: "The "See also" section should not include red links, links to disambiguation pages (unless used in a disambiguation page for further disambiguation) or external links (including links to pages within Wikimedia sister projects). As a general rule, the "See also" section should not repeat links that appear in the article's body." Source: Source
  • References: "This section, or series of sections, may contain any or all of the following..." | Source
  • Bibliography: "A bulleted list, usually ordered chronologically, of the works created by the subject of the article." | Source
  • Notes: See "References" above.
  • Selected Publications: I'm not sure about this one...?
  • Selected Works: I'm not sure about this one...?
PPelberg (WMF) (talk) 00:05, 22 February 2023 (UTC)Reply[reply]
My main reason for these exclusions was rather practical: I didn't exclude them in the beginning and the script produced many false positives. These sections contain mainly lists of items that currently do not need references. For example, Wikipedia does not require a reference that a link added to a "See also" section is relevant. Or new books added to a "Selected Publications" section do not require a reference to support that this book is actually about this topic of the article. There could be cases where contents added to these sections require references but they are quite rare. If your solution was sophisticated enough to spot such cases then maybe you wouldn't need to exclude these sections. But this may not be worth setting as a goal since there are few cases.
I'm not sure about the specific policies but your suggestions make sense. For "Selected Publications" and "Selected Works", you could use the same as for "Bibliography" unless you find something better. There could also be more synonyms for the section names listed above. Phlsph7 (talk) 09:39, 22 February 2023 (UTC)Reply[reply]

Others[edit]

If the text is added as part of an existing paragraph, one could check whether this paragraph has references and "citation needed"-tags and whether the text is added before or after the references.
Oh, checking for the presence/absence of citations or [citation needed] templates within the paragraph the person is adding text to...great idea! I've added this to phab:T324730.
The basic principle of how my script handles this is roughly the following. I'll use the example of a regular HTML-paragraph but the principle is the same for lists and various templates. If you want to find out whether a sentence (or a word or any type of string) somewhere in a paragraph is unreferenced, you keep going to the right of that sentence until one of three things happens: (1) you encounter a reference, (2) you encounter a citation needed tag, or (3) you have reached the end of the paragraph. If (1) happens then the sentence is referenced. If (2) or (3) happen then the sentence is unreferenced. This algorithm works in most cases to get the intended result. Phlsph7 (talk) 12:39, 22 February 2023 (UTC)Reply[reply]
This algorithm assumes that the sentence itself does not contain references somewhere in the middle. If it does then the algorithm can be used for the parts of the sentence that do not contain references. Phlsph7 (talk) 14:39, 23 February 2023 (UTC)Reply[reply]
Regular paragraphs and probably lists should be checked but image captions and individual table cells in most cases not. There are many different templates that could be taken into account. For example, Info-boxes usually do not provide references for every single fact while for quotation templates, references are essential.
Great call and agreed; I've added the cases you shared above to the ticket we're drafting the initial heuristic that will determine if/when newcomers are prompted to decide whether the edit they're making warrants a reference or not (phab:T324730).
If you wanted to double down on this, you could go through the most-used templates and decide in each case separately. Phlsph7 (talk) 09:42, 22 February 2023 (UTC)Reply[reply]
Another relevant factor could be the size of the added text: if someone adds 5 words to a paragraph, chances are that this is just meant to clarify an existing statement and to introduce a new one.
Agreed. We're actively investigating whether we'll be able to specify not only words or bytes added, but whether we can use the number of sentences that someone is adding as a filter as well. See more in phab:T324363. PPelberg (WMF) (talk) 00:05, 22 February 2023 (UTC)Reply[reply]
Even a refined system is bound to lead to false positives so prompt messages should be carefully formulated to take this into account.
I think the point you're making here is astute. In fact, it leads me to wonder: How might you expect to be able to audit the efficacy of the system Edit check is introducing? E.g. might you expect to see a set of configuration pages like are currently offered for various abuse filters, like Special:AbuseLog? PPelberg (WMF) (talk) 00:06, 22 February 2023 (UTC)Reply[reply]
Unfortunately, I'm not very experienced in this field. As a rather general comment, I think experienced editors may need to go through individual edits and decide for each one whether the filter should have been triggered. Their judgment can then be compared with whether the filter was actually triggered. This way, you get some feedback on your ratio of false positives and false negatives in comparison to the number of edits. Something like the automatic abuse log you mentioned would be useful to get feedback on false positives. Phlsph7 (talk) 09:46, 22 February 2023 (UTC)Reply[reply]

Other comments[edit]

SDKB summarized the cases where a cite is required which I'd guess is about 5% of additions. I really don't see this as a good or workable idea. Sincerely, North8000 (talk) 18:50, 17 February 2023 (UTC)Reply[reply]

I'm not sure I would go that far. The larger the text added, the more likely they are adding content that needs support and risks being problematic. Though one of the smallest byte changes could be the most troublesome, removing or adding 'not', which obviously changes the entire meaning of a sentence. Slywriter (talk) 19:01, 17 February 2023 (UTC)Reply[reply]
I would think any addition that adds or changes a number in the text or infobox should generate a prompt to cite source. ~ ONUnicorn(Talk|Contribs)problem solving 19:06, 17 February 2023 (UTC)Reply[reply]
  • The lead is the most significant part of the article because that's what the reader sees first and often they don't go any further. But there's no mechanical linkage between the lead and the body to ensure that they are in sync. So, exempting leads from citation requirements is a big loophole in the verification scheme. If you have automated nagging about updates to the body then this will tend to place even more focus on the lead.
Andrew🐉(talk) 19:16, 18 February 2023 (UTC)Reply[reply]
  • I'll begin with a question, and follow-up with a suggestion. Will this tool be able to verify the growing number of foreign language books that are being used in en.WP? This is a real problem considering the obscurity of some of those sources, our inability to determine their reliability, and difficulty in obtaining them. To my suggestion: I agree 100% with Andrew in that the lead is the most significant part of the article. If we cannot verify material in the lead, we may well be on a goose chase. Some of our articles are more like a collection of detailed theses rather than a summary of topics; i.e., tl;dr territory. If we had a tool that could scan the lead in a way to verify the added material with a cited source in the body (or perhaps an internet scan with an evaluation of the verifiable source) we would have a better feel for where to go with the tool.
Second question: will this tool begin scanning at the beginning of article creation at AfC and/or namespace? If so, then is it possible to use keywords that trigger a prompt for the editor while they are editing? For example, on Twitter, certain keywords prompt a pop-up that advises the author if the term is offensive, etc., whereas our pop-up could advise the author relative to WP's needs. We could do something similar at AfC relative to terminology, copyvio and the need for citations and verifiability, and possibly do the same when a new article is being created in namespace – IOW nip the problems in the bud while helping authors at the same time. The tool could be limited to article creation by new editors with say, under 2500 edits, or no limit if it involves creation of or adding material in contentious articles. Keywords could be set to trigger a prompt if the editor is not autopatrolled, or it could go a higher in user rights such as NPP reviewer+. The algorithm will add an inline tag specifying what caused the prompt. Keywords could be used on superlatives, insults, copyvio, or anything else the algorithm finds challengeable. Typically, if the lead is properly verifiable with citations in the body, there is likely little to no issue. Atsme 💬 📧 12:24, 23 February 2023 (UTC)Reply[reply]

Meeting[edit]

The mw:Editing team is hosting a meeting about their plans for a new mw:Edit check feature in the visual editor this Friday and hopes to talk about this subject more. If you'd like to have a real-time conversation with User:PPelberg (WMF), this would be a good opportunity.

If you are interested, please see mw:Editing team/Community Conversations#3 March 2023 and plan to join the meeting (17:00 UCT/9:00 a.m. California). Whatamidoing (WMF) (talk) 20:16, 1 March 2023 (UTC)Reply[reply]

"Minimize" button in Page Curation toolbar[edit]

Tracked in Phabricator
Task T327722

Does anyone use the "minimize" button/feature of the Page Curation toolbar? Any objections to replacing this with a "close" button? You'd hit an X to close it, then you can re-open it with left menu -> tools -> Open Page Curation. Current process is almost identical, but you have to hit minimize before you can see the X. Motivation is that this was requested by someone, and it would also simplify the workflow and the code. –Novem Linguae (talk) 06:46, 18 February 2023 (UTC)Reply[reply]

I didn't do so before and I don't object to the request to replace it with an X.Paradise Chronicle (talk) 08:31, 18 February 2023 (UTC)Reply[reply]
I've used this every now and then, and I have no objections to this, especially if this will help simplify things behind-the-scenes. echidnaLives - talk - edits 10:37, 18 February 2023 (UTC)Reply[reply]
Nihil obstat Mccapra (talk) 13:46, 23 February 2023 (UTC)Reply[reply]

Uzbekistan articles[edit]

The new pages feed is seeing a surge of stubs about Uzbekistan in the last 24 hours. Most look like plausibly notable topics, but most have only one source. Many that look like they have two sources are just the same source in pdf and web page versions. Most also rely on primary sources. There seems to be a very active group of new editors which is great but there’s going to be a lot of frustration and disappointment if they keep churning out stubs like this. Any thoughts on how to approach this? Mccapra (talk) 10:07, 25 February 2023 (UTC)Reply[reply]

The latest trend seems to be articles using the exact same source and wording like Sirli Mosque, Namangan and Jonobod Mosque. Believe it or not, these are allegedly written by different users! I get that these technically meet our notability guidelines but would these not be better all merged into a list article? I mean I've looked at over 100 of these and none of them contain any unique info at all. They are stand-alone articles for the sake of having them. Spiderone(Talk to Spider) 12:59, 25 February 2023 (UTC)Reply[reply]
It might be beneficial to talk to them on their talk pages or ping them here. Maybe they'd be willing to slow down while this is being discussed. –Novem Linguae (talk) 17:36, 25 February 2023 (UTC)Reply[reply]
AfC has been inundated as well with these. Curbon7 (talk) 18:11, 25 February 2023 (UTC)Reply[reply]
This is an interesting style of editing and we should do something about it. A discussion with similar concerns was started by TheLongTone. One of the Uzbek editors seemed to be blocked for a month somewhen in January, but I couldn't find anything on it in the block log. Maybe they feel they have to hurry before being blocked again? I don't know.Paradise Chronicle (talk) 20:10, 25 February 2023 (UTC)Reply[reply]
This was not a block, they were just (unsonsciously) editing from a blocked proxy. Nothing to do with any behavioral issues. Ymblanter (talk) 12:02, 26 February 2023 (UTC)Reply[reply]
Thanks for the explanation. I already assumed and hoped it was a good faith issue. Paradise Chronicle (talk) 15:41, 26 February 2023 (UTC)Reply[reply]
I declined about 20 of these yesterday at AfC -- all cookie-cutter articles with only a few words changed between them. They have since been moved to mainspace by Artemev Nikolay. I'm not really sure what can be done about these, potentially a mass AfD? — Ingenuity (talk • contribs) 20:12, 25 February 2023 (UTC)Reply[reply]
Ah, it appears to be an editathon going from the 20th to 26th. There's a list of the articles created in it here: meta:IbratWiki/Content_Created#English_Wikipedia. Unfortunately, from a sampling of these articles, very few, if any, are actually ready for mainspace. — Ingenuity (talk • contribs) 20:18, 25 February 2023 (UTC)Reply[reply]
All in all, they are pretty poor. Standard early editathon ware? CSD on some, Afd on others. scope_creepTalk 23:16, 25 February 2023 (UTC)Reply[reply]
Do all the mosques not meet WP:GEOFEAT due to being protected by the state and do all of the towns not meet WP:GEOLAND? Spiderone(Talk to Spider) 00:02, 26 February 2023 (UTC)Reply[reply]
They are all templated. WP doesn't do templated or generic or cookie-cutters articles. The pushback on AI reinforces that point. scope_creepTalk 00:33, 26 February 2023 (UTC)Reply[reply]
Fair enough. In that case, let's get some AfDs going. Spiderone(Talk to Spider) 00:47, 26 February 2023 (UTC)Reply[reply]
I got into contact with two of the organizers. One was Umarxon III, who with 226 edits appears to be the most experienced editor. I'd appreciate if those Ibrat camp–wikipedia cooperation could be coordinated a bit better. I am not sure if this happens a lot. I believe to reach out to so many editors is quite an accomplishment. Only the result of so many almost identical articles is a bit disappointing.Paradise Chronicle (talk) 15:37, 26 February 2023 (UTC)Reply[reply]
  • Comment: Good day, everyone! I'm a longtime Wikimedian and an admin on uzwiki and have been closely involved in the WikiStipendiya edit-a-thon. I wasn't able to participate in person in the IbratWiki camp, though. Basically, what we have here are dozens of new users who are trying to contribute to enwiki in good faith. They're all very young (mostly high school students), and none of the coordinators of the camp has much experience editing enwiki. I wasn't able to consult them much, and it seems some of the coordinators had this unfortunate idea of teaching the camp participants how to create cookie-cutter stubs. Some of the articles they've created are decent and can be salvaged, though. (A few examples include Jasliq Aerodrome, Mehrinoz Abbosova and Turob Toʻla; still, most of the entries they've created lack in-text citations and are written in poor English). With the weeklong camp now officially over, we don't have to worry about the influx of new articles anymore. Given that they were all physically located in one place, SPIs are unwarranted. Let's cut them some slack and see what can be salvaged and what needs to go. And do let me know if I can be of any help - I can contact all the coordinators and users not only through wiki projects, but also through Telegram. (Our Telegram group is huge and very active). Nataev talk 15:41, 26 February 2023 (UTC)Reply[reply]
    @Paradise Chronicle @Nataev If that was a edit-a-thon so
    It was an unfortunate coordination, but they kept track of their articles here, so we can clean up good. Paradise Chronicle (talk) 15:56, 26 February 2023 (UTC)Reply[reply]
    I've already instructed the camp coordinators to refrain from moving drafts to mainspace. A list of all the entries they've created is available on Dashboard. I've also suggested that the participants add the {{User IbratWiki}} userbox to their user pages. Nataev talk 16:05, 26 February 2023 (UTC)Reply[reply]

IbratWiki Edit-a-Thon - Week 2[edit]

Good day! I've just found out that the IbratWiki edit-a-thon is going to run for one more week. Apparently this week they're going to be working with a different cohort of students. I'm not sure this is the best way to go about it, but this whole event has been primarily organized by Ibrat, which is a community of language learners (funded by the Youth Affairs of Agency of Uzbekistan). The IbratWiki coordinators are just volunteers trying to help out. While they have years of editing experience in between them, I don't think any of them are fluent English speakers. So, let's cut them some slack and help them out.

Given the concerns raised above about the poor coordination of the first week of the edit-a-thon, what do you think should be done differently this week? For a start, I've told them not to create cookie-cutter stubs. I've also suggested that the participants primarily focus on improving the articles and drafts created by the previous cohort, as most of them are in a rather sorry state. Meanwhile, I'll go ahead and revise the Meta page to reflect the fact that there are two cohorts of students. Nataev talk 15:00, 27 February 2023 (UTC)Reply[reply]

Hi @Nataev. Thanks for checking in with us. I think bullets 1 and 3 in LordVoldemort728's comment above would prevent almost all problems that were encountered this time. To recap, #1 is publish all articles in draftspace (with the prefix Draft: in the title) instead of mainspace (and don't forget to add the {{subst:Submit}} template to the top), and #3 is mentioning the editathon on their user pages so we know they aren't sockpuppets, perhaps with the {{User IbratWiki}} userbox. Hope this helps. –Novem Linguae (talk) 17:15, 27 February 2023 (UTC)Reply[reply]
Got it. I've passed on the word. Thank you! Nataev talk 18:59, 27 February 2023 (UTC)Reply[reply]

e-pao[edit]

Hi. Anyone know anything about the reliability of this website? Here's a sample of their content. I can't find any discussions about this source, and can't seem to find anything about their editorial policy/practices. There are some editors who are suddenly relying very heavily on them. Onel5969 TT me 11:26, 26 February 2023 (UTC)Reply[reply]

The site doesn't give you confidence it is well used and prominent. It has no SSL certifificate, copyright is out of date, browsers requirements message are still present and the comments on the contribution page indicate it is the third site in the country for news and it examines non-standard news content in that country. It does seem to be updated quite heavily, but couldn't find any editorial or team pages, or privacy policy or editing standards page or anything like that. The fact you can submit any story you want and its seems to be a kind of aggregator put it well down the hierarchy of reliableness (if you want) with AP News at the top. I would take it with a slice of toast as being non-reliable unless it can be shown externally as being reliable, bv somebody in-country. If it was of that type, it would have a much much nicer interfaces as newspapers are generally paywalled now. Hope that helps!! scope_creepTalk 11:40, 26 February 2023 (UTC)Reply[reply]
Thanks. I think I'll bring it up at WP:RS. Onel5969 TT me 12:27, 26 February 2023 (UTC)Reply[reply]
WP:RSN is a good place to discuss sources and get a consensus on source quality. –Novem Linguae (talk) 15:59, 26 February 2023 (UTC)Reply[reply]

WMF Moderator Tools interview research published[edit]

Hi - I just wanted to drop a note here to let you know that we've just published some research findings from interviews with new page patrollers. We'd love to know what you think! Samwalton9 (WMF) (talk) 14:26, 27 February 2023 (UTC)Reply[reply]

Tagging for missing photographs with NPP toolbar?[edit]

Is this possible with the NPP toolbar? If there is another solution, it would be great to know. Paradise Chronicle (talk) 03:17, 28 February 2023 (UTC)Reply[reply]

If there is a template (that is added to the article page) for this, then we can add it to the toolbar's tagging section. Till now, I've been adding image-requested=yes to some of the wikiproject templates (via the Rater script). -MPGuy2824 (talk) 03:56, 28 February 2023 (UTC)Reply[reply]
I found this template. Image requested. You just need to add a set of brackets {{}} to it. Paradise Chronicle (talk) 04:26, 28 February 2023 (UTC)Reply[reply]
Maybe presented better like this {{Template:Image requested}}.Paradise Chronicle (talk) 04:30, 28 February 2023 (UTC)Reply[reply]
Tracked in Phabricator
Task T330728
PageTriage/Page Curation toolbar doesn't currently do any talk page tagging, only article tagging. WP:RATER is currently the best user script for talk page tagging, although I think it only does WikiProject tags. I created a PageTriage ticket (phab:T330728) to discuss this idea further. –Novem Linguae (talk) 04:55, 28 February 2023 (UTC)Reply[reply]
Thank you MPGuy2824, I just learned how this is done over the rater script. Didn't know it was possible like this. And to Novem Linguae, thank you for opening a discussion for the PageTriage. Paradise Chronicle (talk) 20:35, 2 March 2023 (UTC)Reply[reply]
Sorry, but the Image Requested one is better, as then it also appears on the talk page banner. Else it is just mentioned in a hidden category. Just an update. Paradise Chronicle (talk) 21:18, 20 March 2023 (UTC)Reply[reply]

Autopatrolled for stub creators?[edit]

Autopatrolled for stub creators has lead to quite some discussions on several venues. Has this been discussed here? I don't see it beneficial for wikipedia if stubs are created for years (by the same editor). Stubs are likely deficient in some way and in my opinion should get a review so they can get tagged for deficiencies.Paradise Chronicle (talk) 03:24, 28 February 2023 (UTC) :I have added the (by the same editor) after Novem Linguae responded.Paradise Chronicle (talk) 05:24, 28 February 2023 (UTC)Reply[reply]

Microstubs are a thorny issue that has risen to the level of arbcom, e.g. Wikipedia:Arbitration/Requests/Case/Conduct in deletion-related editing and Wikipedia:Arbitration Committee/Requests for comment/Article creation at scale. I'm not sure which way the wind is currently blowing on this issue, but hopefully those links are useful. –Novem Linguae (talk) 05:02, 28 February 2023 (UTC)Reply[reply]
Stubs are likely deficient in some way - where does that idea come from? A well-constructed stub is only deficient in that it does not cover the topic as extensively as could be done, which is the base state of the vast majority of articles. As noted before, someone who can reliably create stubs that have no other issues is just the kind of person we do want to have the autopatrolled right. --Elmidae (talk · contribs) 09:07, 28 February 2023 (UTC)Reply[reply]
As long as notability is met. My problem is that some mass-stub creators were not verifying notability prior to creation. — rsjaffe 🗣️ 20:35, 28 February 2023 (UTC)Reply[reply]
I do not mean that stubs are deficient per notability, but for information. If stubs are created lets take the example of the one of the species as mentioned by Joe. What about if an editor creates thousands of stubs on species in danger but in the very vast majority doesn't include the info that they are in danger, nor since when they were considered to be in danger or why their scientific name is like this (often for a habitat or some scientist) nor since when they were considered a species in danger? In the articles I reviewed (out of NPP, as stub creator has autopatrolled), all this info was accessible in the sources used. When I review, I often add wls to other articles so they are no orphan no more. To such species articles I'd likely do that as well and encourage the editor to include such info etc. before they request and be granted Autopatrolled. Paradise Chronicle (talk) 10:41, 1 March 2023 (UTC)Reply[reply]
Speaking as an administrator who has been known to grant autopatrol, I don't think someone whose main creation work is creating stubs qualifies for the permission according to the criteria as stubs are often not "clean" articles. This is qualitatively different than short articles which sometimes get mistaken for stubs. All that said, I would just note that New Page Patrollers have an interest in autopatrol but I find it a little out of scope for NPP and so I would hope meaningful discussion would be held elsewhere. Best, Barkeep49 (talk) 23:01, 28 February 2023 (UTC)Reply[reply]
There's a difference between a 200-word stub with 3 decent references and a one-sentence stub with one source. The former is fine, but there are several reasons why autopatrolled shouldn't be granted to people who mainly create the latter, namely: these kinds of articles do not take a long time to review and don't pose an excessive burden to the NPP queue; in the past UPEs have used rapid stub creation to game the system and obtain the right illicitly; and writing short cookie-cutter stubs on "notable-by-default" topics such as geographical locations and members of legislative assemblies does not necessarily demonstrate a broad understanding of notability and other content policies. There was a case a few months ago where an editor was granted autopatrolled after creating a number of very brief stubs on Oscar winners (which were fine). However, they did not have a strong grasp of English and when they created longer articles, they were full of factual errors and so poorly written as to be unreadable. Spicy (talk) 23:40, 28 February 2023 (UTC)Reply[reply]
See a previous discussion of this. As one of the most active admins at WP:PERM/A, I don't share the view that there is something wrong with most or all stubs, and I'd be interested if anybody has any empirical evidence for that claim. An editor who creates hundreds of verifiable stubs on notable topics (species for example) clearly has a strong case for autopatrolled because they have a significant impact on the size of the backlog and individually reviewing each of them is a waste of patroller time. Indeed they have a much stronger case than an editor who has created 25 short-to-medium-length articles over the course of a couple of years and wants to round out their collection of hats, which is what most requests for autopatrolled are these days. Both types of request are susceptible to gaming, which is a separate issue.
A problem can arise when someone granted autopatrolled for stub creation switches to writing substantial articles, and it turns out they're not competent at that. But the same problem can happen when someone granted it for writing long articles on individual topics starts an ill-conceived mass creation. Or when someone granted it for writing good articles about apples starts writing terrible articles about oranges. The problem isn't stub creation, it's that autopatrolled is permanently assigned and almost never reviewed. – Joe (talk) 06:09, 1 March 2023 (UTC)Reply[reply]

Without trying to tackle the question of stubs, IMO one of the the criteria for autopatrol is a bit in reverse. IMO someone who an experienced editor who has not created a large quantity of articles is the safest to grant the right to. But one of their criteria is having created a large amount of articles. North8000 (talk) 13:57, 1 March 2023 (UTC)Reply[reply]

That misses the only real reason for granting the right though - lessening reviewer load. Someone who only creates the occasional article does not fill the NPP queue. Exempting someone who adds substantially to the queue, however, saves review work. After all, the right is not meant to be a badge of honour but a tool to mitigate reviewing effort. --Elmidae (talk · contribs) 14:11, 1 March 2023 (UTC)Reply[reply]
Yes, I knew/know. That relates to the "need" aspect. My point was that regarding "safe to give?" that criteria works in reverse. North8000 (talk) 03:00, 3 March 2023 (UTC)Reply[reply]
I would also support a change into that that quality is weighed more than quantity when admins have to decide on editors who apply for autopatrolled.Paradise Chronicle (talk) 01:19, 20 March 2023 (UTC)Reply[reply]
IMHO someone creating of large amounts of new articles in a Wikipedia that is somewhat mature is a reason for extra scrutiny rather than a reason/requirement for bypassing scrutiny. North8000 (talk) 21:42, 20 March 2023 (UTC)Reply[reply]

Newspapers.com - Now available without application[edit]

For those who are unaware, Newspapers.com is a newspaper database which contains more than 840 million pages from over 22,000-plus newspapers. Newspapers.com is normally a subscription service but it was also previously available to editors who applied for access through the Wikipedia Library. Some changes recently went through and, provided you have access to the Wikipedia Library, the database should be available by default without an application or subscription.

Eligibility: Any editor can use the library if they meet a few basic requirements:

  • You have an account that is a minimum of 6 months old
  • You have made a minimum of 500 edits to Wikimedia projects
  • You have made at least 10 edits to Wikimedia projects in the last month
  • You are not currently blocked from editing a Wikimedia project

It should automatically appear in your library here, but the direct collection link can be found here.

I HIGHLY encourage editors to check out the site and what it has to offer. Hey man im josh (talk) 15:44, 10 March 2023 (UTC)Reply[reply]

Glenn Fleishman[edit]

The article Glenn Fleishman was deleted at AfD. Today it was undeleted by DavidLevinson with the edit summary undeleted article - Glenn Fleishman is not unnotable. Additions in progress, more later. Undeleting an article legitimately deleted at AfD because of an admin's personal opinion seems an abuse of privilege. Surely the correct way to have an article on Fleishman if you thought he was notable would be to create a new article in draft and submit to AfC. John B123 (talk) 21:11, 12 March 2023 (UTC)Reply[reply]

An admin doing a self-refund probably doesn't violate policy. Seems that their only change so far though is this small addition. That probably isn't enough to avoid G4. Perhaps they'd be willing to move this to draftspace until it is ready for mainspace? –Novem Linguae (talk) 22:22, 12 March 2023 (UTC)Reply[reply]
A few things bother me about this. Firstly an admin undeleting an article that was deleted at AfD based on their personal opinion makes AfD a waste of time. Why should people bother participating at AfD if the consensus could potentially be overturned if an admin doesn't agree with it? Secondly, per WP:TOOLMISUSE, admins should not use their tools in a non-neutral way. Clearly having an opinion that there should be an article on Fleishman and undeleting the article is not acting neutrally. Lastly, restoring to mainspace with the article marked as reviewed (thanks to Pppery for marking it unreviewed) comes across as attempting to minimising third party scrutiny. --John B123 (talk) 23:45, 12 March 2023 (UTC)Reply[reply]
While I certainly don't like admin cowboy undeletions (which David seems to have a history of), I think you're reading too much into restoring to mainspace with the article marked as reviewed (thanks to Pppery for marking it unreviewed) comes across as attempting to minimising third party scrutiny - most likely he was not aware of the mew page patrol process at all, or the intricacies of the way undeleting a page worked. I add lots of slipped-through-the-cracks pages back to the queue, and see nothing malicious about it. * Pppery * it has begun... 23:54, 12 March 2023 (UTC)Reply[reply]
Ummm...a refund into mainspace is absolutely an abuse of admin power. Self refund makes it worse especially as no admin undeletes directly into mainspace. It goes to draft or user. Slywriter (talk) 00:08, 13 March 2023 (UTC)Reply[reply]
The article should have been restored to draft space in my opinion. It was deleted at AfD for a reason. Hey man im josh (talk) 00:40, 14 March 2023 (UTC)Reply[reply]
Pop it back into AfD? Best Alexandermcnabb (talk) 04:31, 14 March 2023 (UTC)Reply[reply]
I think draftifying it is the least dramatic option. The other option is CSD G4. I don't think the article has been changed enough to qualify for a 2nd AFD. –Novem Linguae (talk) 08:50, 14 March 2023 (UTC)Reply[reply]
I did bold. Best Alexandermcnabb (talk) 09:22, 14 March 2023 (UTC)Reply[reply]
Thanks. --John B123 (talk) 21:14, 14 March 2023 (UTC)Reply[reply]

Repeated draftification of the same article[edit]

@Onel5969, M.Ashraf333, and Silikonz: Recently, I stumbled upon User talk:DifficTones while looking at that user's SPI case. I noticed that the three of you on separate occasions moved pretty much the exact same article from mainspace to draftspace. I'm thinking it may have been more productive to nominate the article to WP:AFD at some point. Getting the article deleted via AfD has the side-benefit of allowing us to use WP:CSD#G4 to speedily delete any future recreation of the article—rather than requiring us to whack-a-mole the drafts as they are repeatedly recreated. While not a policy, I would also note that WP:DRAFTOBJECT states that A page may only be moved unilaterally to the draftspace a single time. If anyone objects, it is no longer an uncontroversial move, and the page needs to be handled through other processes, such as deletion, stubbing, tagging, etc.

With all of that being said, I'm mindful that a recent RfA candidate made an argument that "double-draftification" is appropriate in some circumstances and that WP:DRAFTOBJECT was not written based on any well-attended discussion (see Q13 at Wikipedia:Requests for adminship/MB). I'd be curious to hear your thoughts. Mz7 (talk) 23:03, 12 March 2023 (UTC)Reply[reply]

Hi. I think there are two separate issues at play here. First, I see there are several messages on the user talkpage, but when I look at the draft history, on the first three, I'm only seeing a single draftification, the one on 2/25. Not sure what happened, but the editors who draftified it had no clue that it had ever been draftified before. The fourth time is similar, although the reviewer knew there was an existing draft about something with the same title. So, if they didn't check the existing draft, they also had no idea that the article had been draftified. It could simply have been created in draftspace, and then recreated in mainspace. Which happens all the time.
I didn't see the RfA discussion about this (I had already cast my vote prior to that question being posed), but I would agree that sometimes DRAFTOBJECT is not written in stone. But I would only do so in cases of UPE or COI editing. The only other instance I can see is if an article is wholly unsourced, gets draftified, and then is returned without improvement. In those instances, I feel WP:VERIFY and WP:BURDEN trump DRAFTOBJECT. I am not shy about sending stuff to AfD, and when I've draftified something and it gets returned without improvement, I usually send it that way, although I am a believer in Wikipedia:Using deletion as cleanup, as sometimes that is the only course of action available to get an article to be improved. Onel5969 TT me 00:21, 13 March 2023 (UTC)Reply[reply]
Regardless, Silikonz's draftication (see page history of Draft:Draft:Megha Bhardwaj 2) was done only two minutes after page creation, which seems sub-ideal. Now, creating pages repeatedly in mainspace hat are not ready are inappropriate, but I also had a look at their move logs and found 1, 2, 3 in the past week merely less than twenty minutes after page creation (one of the linked ones is an unsourced BLP, but IMO a good way to deal with BLPs would be to promptly remove the offending material, nominate it for WP:BLPPROD or deletion within an hour, etc). M.Ashraf333 also drafted this less than an hour following page creation. Now everyone makes mistakes and I also really appreciate the thankless work that everyone does in reducing the backlog, the NPP guidance is never a firm rule, WP:IAR is always a thing, and reviewers have good reason to suppose that this user in creation might be a bad-faith user, or has a WP:UPE and WP:COI. Regardless, given that none of these are explained during the draftications (if UPE/COI or other reasonable presumptions of the user not acting in good-faith is the reason to not follow minimum draftification rules, then these should at least be pointed out in some venue...), I think that experienced users with advanced permissions not following NPP guidelines is a bit sub-optimal. Now these obviously aren't the biggest deal, but I'd appreciate everyone's thoughts on this. Sorry this is too long, thanks. VickKiang (talk) 06:34, 13 March 2023 (UTC)Reply[reply]
I'll leave it to folks that have more thoroughly analyzed it to comment on this case. But I do favor the extension to 1 hour. And also that we should not categorically prohibit a 2nd draftification. But IMO if we know that an article should exist, then it should be in mainspace even if it is in bad shape. But the big "limbo" zone is when that has not been determined and in that case I think that the burden for putting in wp:notability references to establish it should be on the zillions of editors, not thrown onto an overloaded NPP'er to have to determine that they don't exist. And due to other flaws in the system, draftification is the best way to implement that. North8000 (talk) 14:21, 13 March 2023 (UTC)Reply[reply]
I think that "limbo" zone is pretty central to my motivation for starting this thread. Nominating an article to AfD does require a bit more effort on the part of the reviewer than draftification because of WP:BEFORE: at AfD, the onus is on the nominator to argue for non-notability, whereas an article can be sent to draft space immediately if the page creator hasn't added sufficient sources to demonstrate notability. The main conflict I see right now is that the written rules state that draftification is sort of like PROD: you can do it once, but once it gets contested, then you can't do it again. Is that guidance accurate? Are there cases where we should try to "enforce" a draftification after a page creator objects by recreating the draftified article in mainspace? Mz7 (talk) 22:11, 13 March 2023 (UTC)Reply[reply]
Probably the best solution is to develop a mindset that step one in developing an article is finding GNG references, and that is a main part of developing a article. Without that you've done nothing worth keeping. Just like I can't say "here's a windshield wiper. I want it to be given the status as a partially built automobile".
Removing or modifying any categorical rule against multiple draftifications would also be good
Finally, getting rid of or modifying wp:before would be a good idea. North8000 (talk) 18:26, 14 March 2023 (UTC)Reply[reply]
I think that the proscription against re-draftifying does need to be done away with, and I also think that WP:BEFORE should be done away with, but only during NPP patrolling. Onel5969 TT me 01:08, 21 March 2023 (UTC)Reply[reply]
Worth an RfC on both? I'd agree on both counts... Best Alexandermcnabb (talk) 06:14, 21 March 2023 (UTC)Reply[reply]

Would it not save us all a lot of heartache if you could only move articles from Draft to Mainspace via the AfC process? I've seen a number of drafts where the user's gone to AfC, lost patience and banged the article into mainspace (and I've seen a number where the AfC history is lost in the process). Draft:Sachiyo_Ito is, I think, one such. Best Alexandermcnabb (talk) 06:13, 21 March 2023 (UTC)Reply[reply]

AFC (inadvertently) has a higher bar. I don't think that it should be required. Sincerely, — Preceding unsigned comment added by North8000 (talk • contribs)
It's a good idea in theory, but there are many quality content creators who gradually develop drafts and move them to mainspace afterwards. In this case, entirely prohibiting moves from drafts to articles except by AfC reviewers would create a negative effect (though they can also techically use userspace if they desire, but much fewer editors use userspace to develop). However, my other concern is that in addition to North8000's points (and I'm not sure about mandating AfC), wouldn't new users just do cut and paste moves to space? That will still be objecting drafting, and many reviewers would still opt for the practices of WP:DRAFTOBJECT and blanking a cut and paste mainspace move with a db-r2 tag. Or is cut and paste moves also prohibited by some way? IMO such moves are bad, but if these are additionally prohibited that would basically make drafting much more powerful in that article creators couldn't object it in any way unless they submit from AfC (if so, that seems strange because it would make drafting far different compared to BLARs and PRODs, which are also not subject to a community or admin review process but can be easily reverted; also, if it's drafted improperly by a non-NPPer who don't know what they are doing, who give the article creator no info on how to improve, there is no option except to submit to AfC, which has a high standard and will take months)? Thanks. VickKiang (talk) 21:06, 22 March 2023 (UTC)Reply[reply]

IMO Eliminating the ability to create a Wikipedia article except through AFC is a gigantic change that would never pass and IMO not a good idea. But I think that there are two other good ideas above. One is indicating that the wp:before is not intended for NPP situaitons (don't forget wp:before is just a suggestion and not enforcable) and also let NPP put articles back into draft. If we did a through job on drafting, providing rationales, addressing concerns and supporting the RFC IMO both could probably pass. North8000 (talk) 22:01, 22 March 2023 (UTC)Reply[reply]

As I said above, I agree both of Onel's cunning plans. And fine FINE my passing thought on AfC wasn't the brilliantest of ideas. Alexandermcnabb (talk) 07:03, 23 March 2023 (UTC)Reply[reply]

Resources[edit]

Okay, I'm not sure how many of you struggle with this, but one of the biggest obstacles I have when doing patrol is when the sources are in a foreign language. Oftentimes a machine translation will be very helpful in determining notability, but sometimes it's quite difficult. Might we create a page, and link it on our Resources tab, where we can list, by language, editors that speak a non-English language and would be willing to get pinged to look at certain articles? If so, I'll create the blank article, with sections by language and then others can add to it (sections) or list themselves under a section they'd be willing to focus on. To me the big ones are the languages which use an entirely different alphabet (Cyrillic, Arabic, Chinese, etc.). Thoughts? Onel5969 TT me 14:06, 14 March 2023 (UTC)Reply[reply]

What does AfC do for situations such as these? Hey man im josh (talk) 14:26, 14 March 2023 (UTC)Reply[reply]
Hey man im josh mostly I believe people either use machine translation or leave for another reviewer. We also have Wikipedia:WikiProject Articles for creation/List of reviewers by subject that includes reviewers other languages to help when machine translation is not good enough. Cheers KylieTastic (talk) 16:10, 14 March 2023 (UTC)Reply[reply]
Agree, foreign language sources are a particularly big problem when you patrol people's category pages, however I always use machine translation to avoid any possibility of error but we need to work on it. Creating a separate page for non-English speakers to review pages would be helpful and make our lives easier. M.Ashraf333 (talk) 14:28, 14 March 2023 (UTC)Reply[reply]
"use machine translation to avoid any possibility of error"? Really? I thought that in general we didn't have that much faith in machine translation yet. PamD 15:42, 14 March 2023 (UTC)Reply[reply]
Hhhm, although machine translation has flaws but it helps a lot in understanding the context of the content. M.Ashraf333 (talk) 15:51, 14 March 2023 (UTC)Reply[reply]
I agree with PamD that machine translations tend to contain errors. It is a tool to help a non-native speaker understand the jist of an article, but it has significant limitations. –Novem Linguae (talk) 18:23, 14 March 2023 (UTC)Reply[reply]
With Google translate the accuracy varies from language to language. The more common a language, the more people use it and suggest a better translation and so it improves. For example translations from Spanish are pretty good whereas from Catalan they are pretty poor. In my experience, machine translations from languages that use a non-latin alphabets are too poor to be relied on. --John B123 (talk) 21:39, 14 March 2023 (UTC)Reply[reply]
Just to clarify, I agree that machine translations are iffy, but I thought I was making it clear that I used them to get the gist of an article, not to verify specific facts about the subject. For example, if I'm looking at an article about Joe Russia, and all the sources are in Cyrillic, if the article has a parenthetical translation of "Joe Russia" in Cyrillic, I'll cut and paste that into the Cyrillic version of the article. If I find numerous matches, then I'll look at the machine translation and see the context of those matches. If there is no parenthetical translation, I'll simply do the machine translation and look for matches. However, this can be problematic, as the machine translation can leave out characters in the translation, so F5 searches don't always indicate how detailed the article is about the person. Onel5969 TT me 00:27, 15 March 2023 (UTC)Reply[reply]

I’d be happy to share my languages on such a resource page. Mccapra (talk) 07:41, 15 March 2023 (UTC)Reply[reply]

User:John Cummings[edit]

Hello everyone, there is a ANI about User:John Cummings's promotional content creation that may be of interest to NPP. Frankly, I'm shocked at the promotional nature of the recent content from this editor. Having worked with many new editors, I would argue that if any of the recent pages created by User:John Cummings came through NPP they would not fair well. AugusteBlanqui (talk) 08:50, 24 March 2023 (UTC)Reply[reply]

thank you for posting--Ozzie10aaaa (talk) 01:23, 25 March 2023 (UTC)Reply[reply]

Recent addition to WP:NPP[edit]

Does everyone agree with this recent addition to WP:NPP, or does it need more discussion? I don't look forward to folks throwing around the shortcut WP:NPPCON in potential ANIs against NPPers. This addition seems to have a POV that NPPers should never double draftify, never double BLAR, etc. Perhaps this is true, but I feel this needs discussion first. –Novem Linguae (talk) 17:18, 28 March 2023 (UTC)Reply[reply]

I think it needs a carveout for cases of COI/UPE, but otherwise seems reasonable. signed, Rosguill talk 17:23, 28 March 2023 (UTC)Reply[reply]
I've gone ahead and added language to that effect in a relevant footnote. signed, Rosguill talk 17:26, 28 March 2023 (UTC)Reply[reply]
Nicely worded and looks reasonable. scope_creepTalk 17:48, 28 March 2023 (UTC)Reply[reply]
I think, personally, it was a mistake for such extensive content to be added boldly rather than through consensus, and I would discourage it happening in the future, but I think it's a fine addition when reviewed in retrospect. It is my understanding that the community has made very clear, in unfortunately drama related venues, that they are against NPP doing double of anything, really (excluding the circumstances Rosguill mentioned). Regardless of how the community has let us know, I otherwise see no issue with it. The rest of the content of WP:NPPCON is consistent with what I've learned at WP:NPPSCHOOL or at various other guidance pages. It doesn't hurt to have a more thorough reminder at WP:NPPCON considering the work we do. —Sirdog (talk) 05:39, 29 March 2023 (UTC)Reply[reply]
I agree it was perhaps a bit too bold, but I don't disagree with the actual content of it. -Kj cheetham (talk) 16:31, 29 March 2023 (UTC)Reply[reply]
I think it's an excellent summation of the current consensus regarding NPP. ~ ONUnicorn(Talk|Contribs)problem solving 17:11, 29 March 2023 (UTC)Reply[reply]

It's wide-ranging with some good stuff in it and other stuff that has structural issues and can have bad intended consequences. IMO for addition to such a prominent and influential place it should receive a more granular review. To do that we could take it out and then put it back in piece by piece. Sincerely, North8000 (talk) 17:00, 29 March 2023 (UTC)Reply[reply]

That seems sensible. -Kj cheetham (talk) 17:04, 29 March 2023 (UTC)Reply[reply]

Be on the lookout for articles created through ChatGPT and other LLMs[edit]

New page reviewers should be aware of this ongoing ANI case related to the use of ChatGPT (or another LLM) to create articles with falsified citations. Make sure to follow through and actually check cited papers when reviewing articles, don't just assume that a plausible-sounding paper title is real. In particular, closely inspect articles whose creation or expansion follows a pattern of "full content first, citations later", as this can be a sign of LLM use. signed, Rosguill talk 18:27, 29 March 2023 (UTC)Reply[reply]

Wow. It's great that they provided prior examples, I highly suggest every NPPer read it thoroughly. And I also highly recommend we all bookmark GPTZero to be able to run articles through there. It'd be great if one of our programming gurus could make a quick link, similar to the one to earwig, that we could use on suspicious articles. Onel5969 TT me 20:15, 29 March 2023 (UTC)Reply[reply]
Good idea. I've added a link to GPTZero to User:Novem Linguae/Scripts/NPPLinks. It is now the third link down, named "AI/LLM check". Unfortunately that particular website doesn't make it easy to pre fill anything, so if you guys find a better website, please let me know. –Novem Linguae (talk) 20:28, 29 March 2023 (UTC)Reply[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:New_pages_patrol/Reviewers&oldid=1147263079"