This page is for New Page Reviewers to discuss the process with each other and to ask for and provide help to fellow reviewers. Discussion also takes place on our Discord server (invite link)For discussions on other matters, such as bugs, etc., please navigate through the tabs, or go to the discussion pages of the relevant policies.For discussion on topics purely relevant to coordination tasks, such as backlog drives, please post at Coordination TalkPage creators and other editors, please do not post requests here for articles to be reviewed. These requests will be declined.
Latest comment: 1 month ago9 comments4 people in discussion
Is it possible that the page curation tool/New pages feed could be modified to show potential LLM in new articles, similar to the existing copyvio tag? John B123 (talk) 19:50, 12 August 2026 (UTC)Reply
AI detectors are pretty hit or miss in my experience, but it might be helpful to have a tag that is based on definitive signs like utm-source metadata in links or the new text watermarks that are being rolled out. 🌸wasianpower🌸 (talk•contribs) 15:39, 13 August 2026 (UTC)Reply
Thanks for your advice. I can see the problem I think. On my laptop the banner isn't hiding notifications but am I correct in understanding that is what is happening for you?
One other question. Is it better in the General interface (bottom right (at least for me)) or in the main menu interface.
On my tablet it's extending across the whole screen and hiding the menu at the top of the page. Adding to either menu would be fine for me. Regards.--John B123 (talk) 21:47, 7 October 2026 (UTC)Reply
Thanks, I think I have fixed this.
I really appreciate your feedback as I hope that this could become a useful tool. It has been moved to the "General" area in tools.
Latest comment: 27 days ago16 comments8 people in discussion
After I ran into articles on contentious topics that were in bad shape and needed action, but were a few months old and hadn't been reviewed yet, I started thinking about how to find the articles lurking in the backlog that most need review. This is the first version, and I'd love for people to try it and tell me if it's helpful or interesting: https://projo.toolforge.org/review (note that it's still in development! it loads a bit slow and is still a bit janky).
It's a bit like Special:NewPagesFeed but primarily sorted based on risk instead of freshness. The artice list is updated once an hour to remove reviewed articles, and it pulls in fresh unreviewed articles once a day. You can also enable "chill mode" to sort the articles in reverse order (lowest score first) for easy reviews of unlikely-to-be-controversial articles. I'm working on more ranking factors, including prior deletion, inexperienced creator, lack of categories, and lack of talk page. Dreamyshade (talk) 03:00, 11 September 2026 (UTC)Reply
So far I'm loving it. Being able to tackle the most-viewed pages feels like I'm making more of a difference in the quality of the encyclopedia than slogging through the back end of the queue in chron order. I will try to share some more specific feedback but this is directionally terrific. Thank you @Dreamyshade! Dclemens1971 (talk) 01:36, 20 September 2026 (UTC)Reply
Thanks for this. I found it much more helpful than the standard NPP feed. Is there any way to save my custom weights for the significance of the different factors, either as part of the URL or as a cookie? lp0onfire()21:56, 19 September 2026 (UTC)Reply
Cool! Working on a way to persist the filters, but not yet a way to persist the custom weights. I'll see what I can figure out. I'm curious, what custom weights are you setting? Dreamyshade (talk) 22:07, 19 September 2026 (UTC)Reply
I've tried a few things; a large negative weight (say -80) for stale articles combined with the UPE filter found me a few easy draftifications, and I would generally decrease the CTOP bonus if I'm in more of a spam-hunting mood. I haven't been using it for long enough to have established a workflow, but I imagine it'll involve tweaking the weights and wanting them to stay that way. As a side note, I'd love a dark theme. lp0onfire()22:22, 19 September 2026 (UTC)Reply
I just want to say that this seems like a really wonderful and clever tool! I've often thought that the NPP queue contains a lot of things of very different urgency/impact levels, and this looks like a really wonderful way to give priority to the things that would have more negative impacts to be left waiting. Kudos!! ~ le 🌸 valyn (talk) 02:53, 20 September 2026 (UTC)Reply
Thank you! That’s exactly the goal. We also just added the option to filter by WikiProject, which I find interesting because it helps with finding articles in a variety of potential subjects of interest, without having to hand-craft database reports - for example:
Aside from this point I do think there are a few things missing/in need of updating. For instance there is no mention about what to do if we suspect the article is produced by LLMs Giulio15:36, 13 September 2026 (UTC)Reply
I tried to update the flowchart a while ago but unfortunately its original creator said they lost the original Draw.io file, so it would be tedious to update. Catalk to me!13:11, 14 September 2026 (UTC)Reply
@Giuliotf, @1keyhole: The CSD are some of the oldest and firm deletion criteria on Wikipedia; they are clear and designed to be unambiguous, and the Page Curation tool descriptions are adequate. Created nearly 8 years ago long before LLM became an issue, while the flowchart may not reflect some observations by new NPPers, many user find it useful. It replaced an earlier, clearer overview flowchart. Nevertheless, the flowchart is not intended as a substitute for the tutorial which it is assumed that aspirants for WP:NPR have read and digested, and LLM is covered here. Kudpung กุดผึ้ง (talk) 01:37, 15 September 2026 (UTC)Reply
Yeah sorry about that. It was on an old computer of mine where the hard drive bricked itself... I lot a lot of other stuff too. The links did work at one point, and as an SVG I have a few times opened it in Inkscape and made minor edits and added links. Changing the layout would likely require remaking it though. — Insertcleverphrasehere(or here)(or here)(or here)04:52, 23 September 2026 (UTC)Reply
I'll point out that, because the flowchart is an SVG, it would be pretty easy to make small edits to the chart or convert to Mermaid or another draw.io-friendly format with an LLM, if you'd like. If you do want this flowchart converted to a particular format, I'm happy to do so. Cheers, Suriname0 (talk) 21:59, 19 September 2026 (UTC)Reply
For anyone interested, WP:FIELD is a good reference for CSD that one is unfamiliar with. It is linked as a hatnote at § "Other speedy deletion criteria" but personally I tend to forget where wikilinks are on a page (my memory is sadly rather poor). Alpha3031 (t • c) 15:24, 19 September 2026 (UTC)Reply
In patrolling the back of the queue, I've found many instances of articles that have come through AFC and that shouldn't have so you'll probably find the same at the front of the queue. I think you can draftify, but I generally go straight to AfD with something that passed AFC, since an AFC reviewer thought it passed the bar and that makes its presence in mainspace a contested issue. Dclemens1971 (talk) 21:10, 19 September 2026 (UTC)Reply
Yeah, tbph, they let some awful junk through sometimes. I got pulled up for doing this by someone citing DRAFTNO, sounds like they are probably right in practice then. We should probably put it on the flowchart if draftifying isn't an option for AFC-positive articles. Boynamedsue (talk) 21:24, 19 September 2026 (UTC)Reply
I would usually check with the AfC reviewer first; if they agree that they made a mistake then there's no issue with re-draftifying, but if they disagree then it should probably go to AfD. At the very least it would be courteous to notify the reviewer, so they can contest the draftification if they think that's appropriate. lp0onfire()21:23, 19 September 2026 (UTC)Reply
No because you precluding folk from addresing simple human error as well as all other sort of things like bias, and ideology, nutters, which is a significant portion of mainspaced articles. scope_creepTalk
It's from ages ago. It was a single source article for a medieval British MP. MPs are automatically notable, so being a single source wasn't a problem, but the source was from 1875 so I didn't feel it should be considered reliable.Boynamedsue (talk) 14:07, 20 September 2026 (UTC)Reply
An 1875 source isn't problematic to establish, for instance, that someone was a Member of Parliament - and therefore is - as you point out - automatically notable. It's not, I know you were being hyperbolic, exactly medieval... IMHO. Best Alexandermcnabb (talk) 15:34, 20 September 2026 (UTC)Reply
No, the MP was medieval, from the 13th century, the source was from 1875. For my money, we need a reliable source to say that he was an MP, as this is a historical fact (or perhaps not). A footnote in a book published in 1975 is clearly enough for this, but I don't believe a footnote in a book published in 1875 is enough. My concern is WP:AGEMATTERS, as it is difficult to establish reliability through use by others in a source that old. Certainly none had been established by any user in this case.Boynamedsue (talk) 16:51, 20 September 2026 (UTC)Reply
I don't think there is (or ought to be) a hard and fast rule about this. I have occasionally draftified articles that have gone through AfC, in cases where there was a very clear-cut reason that the AfC reviewer had clearly just missed for why the article shouldn't be in mainspace. I probably wouldn't dratify an article that had gone through AfC if I thought that there was any doubt about it, or just because I disagreed with the AfC reviewer's assesment regarding notability (in which case I would generally nominate to AfD instead). Dionysodorus (talk) 13:58, 20 September 2026 (UTC)Reply
Latest comment: 1 month ago8 comments5 people in discussion
There's a conversation at ANI right now that made me aware that consensus is that talk pages consisting only of a {{WikiProject Disambiguation}} should not be created. Should the NPP instructions be updated to reflect this? Currently they say:
"WikiProjects - tag as normal, but also include WikiProject Disambiguation
CLASS = disambiguation
IMPORTANCE = N/A, not low"
I'm not sure you read my message correctly. There appears to be consensus around the application of that WikiProject's banner templates to talk pages that the NPP instructions do not align with, so I'm proposing to align the NPP instructions. ~ A412talk!14:06, 25 September 2026 (UTC)Reply
For those of us with the Rater script, it is highly annoying for it to pop up constantly on dab pages without any talk/rating. It is basically begging for the dab workgroup to be added. ☾Loriendrew☽☏(ring-ring)21:15, 25 September 2026 (UTC)Reply
I did read it correctly, but the NPP instructions should probably not give reviewers the impression it's something they have to do. The only 'instruction' I can find is this one which doesn't address the project banners directly. Kudpung กุดผึ้ง (talk) 01:26, 26 September 2026 (UTC)Reply
Thanks, for that. Now I understand. I was misled by the WikiProjects - tag as normal not knowing that the project banner was meant. I don't know how that originally got into the NPP tutorial. It is definitely not in the NPP remit to add any project banners anywhere. As I said above: NPP instructions should probably not give reviewers the impression it's something they have to do, they have enough to do. Kudpung กุดผึ้ง (talk) 07:47, 26 September 2026 (UTC)Reply
Should there also be a reminder to patrollers that "Foo (disambiguation)" should not exist unless there is already a "Foo" article or redirect? (Chipping in here as a NPP editor who only works on redirects, but is also something of a dab geek.) PamD07:16, 26 September 2026 (UTC)Reply
Latest comment: 1 month ago2 comments2 people in discussion
It may have gone unnoticed but for some while this bot has been greatly unter-totaling the patrols for each reviewer. This has now been fixed but please check against your actual logs and if it persists or happens again, contact the bot maintainer directly. Kudpung กุดผึ้ง (talk) 09:07, 25 September 2026 (UTC)Reply
I've serviced this list for the last 5 years anyway. Its a thankless task. There is a problem how folk tag them in the modern era as there is far more getting tagged now than there was in the past in a like for like basis and its often for the most spurious reasons. We really need to start educating editors on what consistutes notability and how to evaluate it. Perhaps a black book on how discern notability, or perhaps create a special task force of a couple of dozen folk who can do say, a 1 day Afd, pile in, evaluate it quickly and pass/fail all in one day, with perhaps a drop out mechanism where it goes to a 7 day if its really complex. The task force would have been specially selected with a new wikiproject, perhaps made up of different specialisms. Folk that have created lots of articles in a particular subject. scope_creepTalk18:22, 30 September 2026 (UTC)Reply
That would be a nice option; I am finding it causes pages to load more slowly in my browser. Might be good to have on when I'm patrolling but not for general editing/browsing. Dclemens1971 (talk) 21:29, 30 September 2026 (UTC)Reply
@Dafootballguy: The problem with the script is that, every time you load a page, the script loads as well. The script is resource-intensive and takes a while to load, and this has negative impacts on reading articles and editing in other areas. Because it is only really needed when you're checking for LLM edits/verification, when you don't need it, it would be good to have the option to turn it off. What asilvering is suggesting is that it would be off by default and would be run only when prompted. Chorchapu (talk|edits) 02:07, 1 October 2026 (UTC)Reply
Looks useful! Can you have it ignore special:NewPagesFeed? It shows up in the <whatever the left-toolbox area is called> once for each article in the visible feed. If you scroll down you will have hundreds of instances. ☾Loriendrew☽☏(ring-ring)14:32, 2 October 2026 (UTC)Reply
Well, @Fermiboson:, when I tried unchecking the "Run automatically on page load" box in the tool Settings, I get "Save failed — kept for this browser only". Is that the expected result - no persistence beyond the current browser session, in essence? In any case, this tool is a real help when reviewing for NPP, but I do agree with others that it does slow down regular editing of articles and having a "turn off" switch is very helpful. Terrific work and extremely useful when patrolling. Kudos! Geoff|Who, me?13:08, 2 October 2026 (UTC)Reply
It is not the expected result. According to my bot, this happens when something prevents the editing of sourcecheck-options.js, but I don't know enough JS to determine the truth of that statement except empirically - check if it exists for you? If not you can manually edit your options page with the json format which Ozzie has linked above. Fermiboson (talk) 14:01, 2 October 2026 (UTC)Reply
@Fermiboson: Thanks - your response gave me the clue I needed. The issue was my VPN, which I have engaged most of the time, but turn off to edit due to an IP address block on the VPN's IP range. Using the VPN prevents all editing, including that of source-checkoptions.js. Turned off VPN, reloaded a sample article, unchecked the box and save completed without the limit to "this browser only", and the tool remains toggled off when I turn the VPN back on reload an article. Geoff|Who, me?14:32, 2 October 2026 (UTC)Reply
Latest comment: 1 month ago3 comments2 people in discussion
Hi there,
I have recently been granted temporary NPP rights and I would appreciate if anyone was to check that I am doing things correctly and if I am doing any thing wrong so that I can fix it before I reapply.
Latest comment: 1 month ago2 comments2 people in discussion
As shown by the graph at the top of this page, our efforts in September to reduce the number of articles waiting for review has only had a marginal effect. Indeed, we still have over 25,000 articles requiring attention. Have we made any progress on introducing new ways and means to deal with the problem or are we prepared to carry on as before, as long as the number of unreviewed articles does not exceed 30,000?--Ipigott (talk) 09:09, 4 October 2026 (UTC)Reply
Latest comment: 29 days ago7 comments5 people in discussion
About 11 years ago, after I resigned non-controversially from being an administrator, I was granted new page patrol privileges. Since then, I've been flamed quite a few times for being too easy on the newbies. I haven't really used the privilege for a few months, except for clicking on pages at AfD. I just had a request for review from Aficionado538 in the past few hours, but I don't want to work on it until I have community feedback. Should I continue to have these privileges, or should you take them away from me? I'm not going "to die on this hill." Ping me if you can answer that question, or need anything. Bearian (talk) 14:46, 7 October 2026 (UTC)Reply
I'd suggest keeping them. I've seen you around NPP discussions quite a few times, and I think you could still help out with reviews whenever you get the chance. Even if it's only a review or two now and then, I think that would still be helpful. Max (talk) 15:45, 7 October 2026 (UTC)Reply
Please stay. While I have not always agreed with you, I do respect your opinions. Very frequently you have picked up on older articles with issues, making good calls there.
I had you down as one of the GOATs:) Administration-type work can feel thankless at times on Wikipedia, but your skill level is high and you are definately needed imho. Aszx5000 (talk) 16:22, 13 October 2026 (UTC)Reply
Tool for finding editors qualified for autopatrol or NPP
I also made a report of editors who may be qualified to serve as new page reviewers, so that people can check them out and potentially invite them to request the permission: AutoScout NPP. It includes editors who have more than 2,500 edits, registered more than one year ago, made at least 100 edits in the past 30 days, have at least one DYK, GA, or FA (as a simple threshold for competence and interest in article review processes), and have not been invited already according to this list. The ranking method counts AfC and AfD participation. The tool tries to check for recent warnings about edit warring as a proxy for potential conduct issues.
This tool is largely vibecoded and is basically a proof of concept rather than a proper robust thing, but I would be interested to hear thoughts about how to make this more useful. For example, if you check some of the top auto-ranked editors in either list and consider them ineligible, what disqualifies them? Do the base criteria likely exclude a bunch of qualified editors or include too many unqualified editors? What potential metrics are missing that might help pop the most qualified editors to the top? (For example: counting DYK reviews, GA reviews, PROD nominations, and/or speedy nominations?) Dreamyshade (talk) 03:12, 10 October 2026 (UTC)Reply
Latest comment: 20 days ago6 comments3 people in discussion
The following discussion is closedand will soon be archived.
I am reviewing a couple of articles and had discovered that the PHP bug caused the toolforge to be closed down. This made the regular way of checking for copyvio impossible. Any suggestions for a (temporary) replacement? Thankful in advance, Викидим (talk) 17:09, 22 October 2026 (UTC)Reply
@Викидим I am an ex-new page reviewer but, I think I can help here. Copy pieces of text in quotation marks then paste into google, any exact matches would be like the copyvoi would flag.
While I was being grumpy and researching other tools (nothing free comes even close, for the record), somebody had fixed the checker. Praise to the PHP wranglers! --Викидим (talk) 20:00, 22 October 2026 (UTC)Reply
Latest comment: 20 days ago7 comments4 people in discussion
If I understand correctly, New Page Patrollers see something that indicates that an article is unreviewed when they visit an article. Could a patroller upload a screenshot of this, so that it can be included in the RfC for Q7? I think it will be helpful to show non-NPP patrollers what EC editors will see if that proposal passes. BilledMammal (talk) 03:21, 23 October 2026 (UTC)Reply
That wouldn't be relevant. What NPRs see is the page curation toolbar. This is presumably not what non-NPRs would see when looking at an unreviewed article. Whatever that indicator would look like hasn't been specified by the question. In solidarity, asilvering (talk) 03:25, 23 October 2026 (UTC)Reply
Good point (obviously I installed it at some point, and since then have been thinking it's part of the suite of core Page Curation features). There are almost certainly lots of other really useful js scripts I have installed over the years that I simply take for granted, some of which, NPPers should be obliged to install. Kudpung กุดผึ้ง (talk) 04:48, 23 October 2026 (UTC)Reply
When AINB finds that an editor with autopatrolled is clearly abusing LLMs, we request autopatrolled to be revoked. We are wondering whether or not this should be extended to the NPP right. There's discomfort with having an LLM user patrolling, as LLM use is one of the main things they should be looking for. I'd appreciate hearing views on this as it affects you directly. M kuhner (talk) 14:52, 23 October 2026 (UTC)Reply
I have revoked the permission. LLM use (and denial of it) breaks the community's trust, and NPP is fundamentally a perm that requires trust. If we expect patrollers to hold other editors to Wikipedia's LLM standards, we should expect patrollers to adhere to the same standards. If TheGreatEditor024 is later unblocked, they can reapply for NPR at a future date. Dclemens1971 (talk) 15:06, 23 October 2026 (UTC)Reply
There are two other common cases: the editor doesn't respond at all, or responds by admitting LLM use and pledging to stop. What is your view on those? M kuhner (talk) 15:22, 23 October 2026 (UTC)Reply
An admission + pledge to stop still likely warrants a revocation as a significant error of judgment, but would be more discretionary. Documented use followed by denial and then a block should be accompanied with revocation of high-trust permissions like NPR and AP. (And TheGreatEditor024 had other problems too, like this cursed DYK.) Dclemens1971 (talk) 15:43, 23 October 2026 (UTC)Reply
Also, there needs to be a review of all the drafts this NPR has approved - I took a quick look at their talk page archives and there are dozens of requests to "positively review" various articles that I personally don't think are notable.
They also used AI to review drafts and to respond to requests to review drafts. Many of the drafts were created by editors who are blocked for UPE/promo. aesurias (she/her) (talk) 19:29, 24 October 2026 (UTC)Reply
This talk page seems to serve the same purpose as an NPP noticeboard. If the suggestion is to create a second page, I'd recommend against it because it would fragment and de-centralize this page. –Novem Linguae (talk) 14:22, 26 October 2026 (UTC)Reply
Yes. When I finished NPP school I took my patrolling duties seriously. It the first time where I felt the community trusted me to do something important. Dr vulpes(Talk)23:45, 26 October 2026 (UTC)Reply
Do you have reasons to suspect AI use beyond the number of past contributions? Are there AI signs that give you that feeling?
If I'm suspicious, I look for AI signs in the article history, I look for a pattern of use, I look at logs, and I look for source-text integrity and do additional verification. If I'm still suspicious after that, I will gently ask the user on their talk page. If there's something that seems to confirms AI use, I will warn them. If there's a pattern, I go to WP:AINB. I don't see particular red flags for AI with this article. —BrechtBro(talk)18:08, 1 November 2026 (UTC)Reply
It's also helpful to run the text through one of the AI detectors available, such as ZeroGPT's free tool. A caveat is that such detectors are but one tool and should not be solely relied on. The verification checks outlined by BrechtBro should also be employed. Geoff|Who, me?18:57, 1 November 2026 (UTC)Reply
I find the GPTZero Wikipedia tool has some utility for now, and it is easy to use, but like Geoff says it should not be solely relied on. I run most articles I review through it and if I get a 1.0 for AI, I treat that as a redflag for a more time-consuming closer inspection similiar to a concerning CopyVio result. If an article looks OK but I'm suspicious or has that 1.0, often looking at the user's other contributions will show a pattern, sometimes with more definitive evidence, and then the issue can be dealt with as a whole at AINB. —BrechtBro(talk)18:46, 3 November 2026 (UTC)Reply
It involves a series of proposals for changes to the NPP process. The intention of these proposals is to reduce the NPP mainspace article backlog, which has been steadily growing and recently exceeded 30,000 unreviewed articles.
There is a very large articles backlog
You are invited to join the discussion at Wikipedia talk:Requests for permissions § 25 new articles that merely pass all tests...what about a smaller number of *much higher-quality article creations*?. Extraordinary Writ (talk) 18:26, 17 October 2026 (UTC)
