Wikipedia

Wikipedia talk:New pages patrol/Backlog drives/September 2025#Concern about Agent VII's reviews

Error?

[edit]

Hi there, just wondering if this is an error or misinterpretation from my part. Per the leaderboard (1) User:JTtheOG has 313 points, however, their logs show their newest article patrolled was in July? Doesn't this only count patrols in September? jolielover♥talk 18:27, 1 September 2025 (UTC)[reply]

Hello. This is due to my stubborn reliance on Special:NewPages instead of Special:NewPagesFeed. JTtheOG (talk) 19:27, 1 September 2025 (UTC)[reply]
I filed phab:T403429 to get these "Mark this page as patrolled" patrols into the Page Curation log. This has been confusing for many years (WP:PatrolVsReview), and I think I see a clear technical solution. Now someone just needs to code it up :) –Novem Linguae (talk) 20:02, 1 September 2025 (UTC)[reply]

What to do if someone isn't exactly reviewing pages to a high standard?

[edit]

Alright, I know there's no definitive rulebook as to what a NPP should do - but from my experience, I take the bare minimum should be creating a talk page and adding relevant WikiProjects, adding categories, basic copy-editing, short spotchecking, and adding short descriptions. I do all of these when I'm reviewing pages, anyway. This is essentially what Wikipedia:New_pages_patrol#Serious_content_problems calls for. I don't want to start drama or anything, but I feel like some users are just marking pages as patrolled despite doing very little/none of these in order to bump their numbers up for the drive. I feel like NPPs should be able to do these things, as they're often the small details people miss when making pages - reviewing the articles despite these issues make it harder to find. jolielover♥talk 13:47, 6 September 2025 (UTC)[reply]

JL, please take a look at the simplified NPP flowchart. Adding Wikiprojects, categories and stub sorting are marked as optional. Adding short descriptions isn't mentioned there but I would consider that as optional too. Spotchecking refs would be automatically done while checking for notability (a required step). Do feel free to continue doing the optional things though, as it may take a long time for a gnome to notice these things outside the NPP process. -MPGuy2824 (talk) 15:00, 6 September 2025 (UTC)[reply]
I agree with MPGuy2824, and will add that copy editing is also optional. NPP is mainly checking for CSDs and notability and article title. Anything extra you do is great, but is not required. Evaluating WP:GNG can take a lot of brainpower to do properly so there is still plenty of work to do per article.Hope that helps. Thank you for reviewing! –Novem Linguae (talk) 17:44, 6 September 2025 (UTC)[reply]

Discussion at Wikipedia talk:New pages patrol/Reviewers § Great job backlog drive participants

[edit]

 You are invited to join the discussion at Wikipedia talk:New pages patrol/Reviewers § Great job backlog drive participants. –Novem Linguae (talk) 19:42, 22 September 2025 (UTC)[reply]

Concern about Agent VII's reviews

[edit]

I'm not trying to start drama and I've gone back and forth on starting this topic because I do believe the editor in question is acting in good faith, but I believe a discussion is needed about User:Agent VII's reviews from this drive. After finding copyvio issues at Bholaganj, an article they reviewed, which I outlined on their talk page, I took a look at Agent VII's review log and found several more problems. I'm now worried about the quality of other reviews. Some examples of problematic articles I found which were passed by Agent's review:

  • Lovachora was also clearly promotional, and also contained copyvio issues (though less extensive) from the BN Wiki. It has since been speedy deleted per G11.
  • Okongo Community Library does not meet WP:GNG or WP:NBUILDING, the only sources available in the article or that I could find online are primary sources about its opening.
  • Fernand Hazan Éditeur does not cite any secondary source coverage or WP:SIGCOV, and a search is not turning up anything other than trivial references to books they published. It was also a non-attributed translation of the French Wiki article.

This wasn't a thorough search, all three articles of these articles in addition to Bholaganj were within Agent VII's last 20 reviews at the time that I checked, and none of these were difficult issues to catch. @JTtheOG has also documented a number of additional problems during re-reviews, and other issues have been brought up on their talk page. These are issues that a review should have caught, and I'm concerned the rate at which issues with reviews are being found.

Additionally, in a discussion on their talk page Agent promised to slow down, but a cursory look at their log shows they're still doing 4-5 reviews per minute. Just from today, there are more issues that someone else will have to clean up — Dum Dum Road as an example does not contain any secondary source coverage and does not appear to pass WP:NROAD, but they passed it and 3 other articles in the course of a single minute. They've done more than 4000 reviews this month and someone else can't manually re-check them all. Courtesy pings to people involved in previous discussions on Agent's talk page: @Asilvering, @Toadspike, @Vanderwaalforces, @Klbrain. 🌸⁠wasianpower⁠🌸 (talk • contribs) 19:42, 27 September 2025 (UTC)[reply]

This person regularly performs 100-200 reviews an hour, and regularly performs eight reviews per minute. It is obviously not possible to perform any of the duties for NPP, much less all of them, at this pace. JTtheOG (talk) 19:46, 27 September 2025 (UTC)[reply]
Fair enough. I am stoping and withdrawing from this backlog drive. I will review normally as was doing before the start of the backlog drive now onwards as per merit and notability guidelines. Maybe after a day or so as a cooling break. Thank you. Agent 007 (talk) 19:55, 27 September 2025 (UTC)[reply]
I am un-reviewing my reviews all done TODAY during the backlog drive. Thank you. Agent 007 (talk) 20:13, 27 September 2025 (UTC)[reply]
@Novem Linguae, DreamRimmer, and Utopes: Request you, kindly remove my name from the WP:SEPT25 leaderboard. I have already removed my name from Wikipedia:New pages patrol/Backlog drives/September 2025/Participants. Thank you. Agent 007 (talk) 20:32, 27 September 2025 (UTC)[reply]
I have not checked the content of your reviews, but the speed at which you completed them is concerning. You reviewed 136 pages with a gap of less than 5 seconds each. In addition, 792 pages had a gap of less than 10 seconds, 1484 had a gap of less than 20 seconds, 1930 had a gap of less than 30 seconds, 2277 had a gap of less than 40 seconds, 2553 had a gap of less than 50 seconds, and 2781 had a gap of less than 1 minute. Considering you only received permanent rights last month, it is unusual to review such a large number of pages so quickly, with most taking less than 30 seconds. I have documented all your reviews this month along with the time taken at User:DreamRimmer/RandomPage, and there is a separate list at User talk:DreamRimmer/RandomPage highlighting pages where you took less than 20 seconds. Given that thousands of reviews are involved, I am pinging @Rosguill, who granted the NPP rights, to make them aware of this. – DreamRimmer ■ 09:10, 28 September 2025 (UTC)[reply]
@DreamRimmer First of all thank you for the list. For the backlog drive only I had two/multiple windows for Special:NewPagesFeed one from oldest sort and another newest sort. I was reviewing in parallel based on WP:RS, WP:GNG, content from well established editors with 1500+ to 10,000+ contributions of track record at first. Then I went through normal ones just recently starting Friday this week. Also, many of the WP:COPYVIO were revdeled like Bholaganj and were less than 5-10% similarity match which excludes quotations per WP:COPYQUOTE on checking Earwig's tool. Many of the websites from where copied were in public domain, that I cnfirmed on visiting the sites. For AI I took help from GPTZero as when required and I have tagged the pages when ever applicable, messaged the creator and moved to draft non-notables when required. For transparency see User:Agent VII/CSD log and User:Agent VII/XfD log logs. Since you listed all of the past month reviews, @Rosguill and @DreamRimmer if permitted I will go through all of them one by one and rectify any errors (tag them without over-tagging) and if required send to WP:AFD strictly based on Wikipedia notability guidelines. I am willing to create a list similar on the lines of Wikipedia:New pages patrol/Backlog drives/September 2025/Re-reviews for myself so that anyone can verify after I re-review them.
Waiting for your reply.
I never expected this much harsh and bitting will be faced if a new person working hard day and night on Wikipedia. Just for information I am also an IEEE and Spring journal(s) reviewer in real life and I know what it goes though Wikipedia review is different and not comparable.
@Rosguill since you belived in my understanding, rest assure and I will not let you and others down. Just pardon for this time and let me fix as proposed above if you think so. I am also not going to take part in future backlog drives as I think it will be against the NPP community.
Thank you all and sorry that I let you all down. Agent 007 (talk) 11:27, 28 September 2025 (UTC)[reply]
@DreamRimmer @Rosguill I have started maintaining the log of self-re-review at User:Agent_VII/September_2025_re-review_log. Thank you. Agent 007 (talk) 12:28, 28 September 2025 (UTC)[reply]
I appreciate your comments here and your efforts to self-correct per others' concerns. I think that if you're able to take on the criticism expressed here and apply a slower pace going forward it will address the concerns identified thus far. signed, Rosguill talk 17:34, 28 September 2025 (UTC)[reply]
Hi Agent, a few thoughts. First, I'm very sorry you're in this position, I know having all this attention on your reviews cannot possibly be fun and that you were only acting in good faith. I have no doubt you were working hard to improve Wikipedia, and I appreciate your willingness to work with other editors on this situation. Second, you've made oblique reference to this a few times and maybe I'm misunderstanding, but I think this might speak to part of the issue here. You shouldn't really be reviewing any differently just because a backlog drive is ongoing. Yes, we are working to get numbers down, and many of us are reviewing in higher volume than usual, but this doesn't mean we should be skimping on parts of reviews so we can go faster. If you just treat them as normal reviews, I think that will help. Additionally, as we discussed on your talk page, an editor having a high edit count does not mean an article they wrote is exempt from thorough review. The fact that you were focusing on those articles does not invalidate any issues with those reviews, please try to keep that in mind.
A few other minor points: GPTZero is not a great indicator of AI text. I've personally found it useful to bolster my confidence that text is AI or not if I see other indicators of AI text, but in general we should be using WP:Signs of AI writing for tagging, or, especially relevant for NPP work, the WP:G15 criteria. This was also discussed on your talk page. I also feel it's a bit unfair to pull out WP:BITE here, people came to your talk page first and gave reasonable feedback. You've also been an editor for 2 years and have 10k edits, I'm sorry but you're hardly a new here. 🌸⁠wasianpower⁠🌸 (talk • contribs) 17:12, 28 September 2025 (UTC)[reply]
I've gone ahead and written Wikipedia:New pages patrol#Hasty reviewing to help document/train newer reviewers that reviewing this quickly is against community norms.
I think we need to put back into the NPP queue (mark as unreviewed) any hasty reviews done by this reviewer, since these reviews were not up to our normal standard. @Agent VII, you mentioned that you unreviewed some articles already. It sounds like you might have unreviewed one day worth of articles, but that you did hasty reviewing for approximately 27 days. Am I understanding that correctly? If so, then we may need to write a bot to unreview an additional 26 days of articles. –Novem Linguae (talk) 17:06, 29 September 2025 (UTC)[reply]
They have unreviewed 273 pages in the last 72 hours. I have listed them at Special:Permalink/1314087859. What do you think, how many pages should be unreviewed for this month? Should we only count those with a gap of less than one minute, or use some other criteria? – DreamRimmer ■ 17:34, 29 September 2025 (UTC)[reply]
An auto-unreview of any reviews with a gap of less than 1 minute sounds reasonable. However I'd also like to invite input from @Wasianpower and @Agent VII. –Novem Linguae (talk) 18:23, 29 September 2025 (UTC)[reply]
I agree, sounds reasonable to me. 🌸⁠wasianpower⁠🌸 (talk • contribs) 18:23, 29 September 2025 (UTC)[reply]
I agree what is being proposed. Anyhow, I am going through User:Agent VII/September 2025 re-review log and self re-checking again (with comments for each). Agent 007 (talk) 14:25, 2 October 2025 (UTC)[reply]
Thanks @Novem Linguae, I went ahead and created WP:NPPFAST as a shortcut for this if that looks okay. 🌸⁠wasianpower⁠🌸 (talk • contribs) 18:23, 29 September 2025 (UTC)[reply]
@Wasianpower, WP:TOOFAST was available for use so I've boldly added that too! This can be redirected elsewhere in future should the need ever arise! 11WB (talk) 21:24, 29 September 2025 (UTC)[reply]
While I do see that some of their reviews have issues, most of them are still fine and don't justify rolling everything back. If we unreviewed all of those pages, the queue would shoot back over 10,000 again and undo the solid backlog drive that just happened, which really doesn't look good as an NPP point of view. A better way is for patrollers to voluntarily recheck pages as needed. That way the bad reviews can be fixed without flooding the queue or erasing the progress made. If there are bigger concerns with the reviewer's overall work, that can be taken up separately instead of mass unreviewing. - The9Man Talk 10:25, 1 October 2025 (UTC)[reply]
They have reviewed over 4400 pages, and unreviewing those with a gap of less than one minute would mean over 2700 pages, or around 2500 if we exclude disambiguation pages. This is a huge number, and we cannot just ignore it or leave it for later, just because it will increase the backlog again. I personally think we should not compromise the quality. I do not think people will spend the time rechecking each page and marking which ones were re-reviewed, because it would take too long. I am posting a message at WT:NPP to get more comments. – DreamRimmer ■ 11:30, 1 October 2025 (UTC)[reply]
If some or all of the articles reviewed by Agent VII are marked as unpatrolled, would it be possible to create a separate bot-updated list of these articles like the ones at User:SDZeroBot/NPP sorting? Given that Agent VII has probably judged rightly in some cases, a few experienced editors aware of the problem could probably re-patrol many of the entries on such a list fairly quickly, eliminating many less problematic cases more efficiently than if they are just being handled in an ad hoc fashion by anyone who happens to come across them on the main feed. (I'd be very happy to go through such a list and try to quickly eliminate any unproblematic biographies, for instance.)
Or alternatively, if we decide not to mass-unpatrol the articles, could a bot create a non-self-updating list of the articles, and then editors could go through manually ticking off the ones that are ok, and manually marking others as unpatrolled where necessary? Dionysodorus (talk) 12:09, 1 October 2025 (UTC)[reply]
I haven't looked through Agent's reviews so I can't pretend to know the entire extent of the issue, but from my understanding they are a good faith editor who ended up reviewing too fast and getting in over their head, naturally making mistakes. Regardless of their intentions, this is an issue, and I think unreviewing all of the quickly reviewed pages (less than a minute per review?) and keeping that at a list, somewhere for experienced reviews to peruse, updated by a bot, will be best solution. As you said, we shouldn't keep possibly incorrect reviews for the sake of making the backlog look smaller. ULPS (talk • contribs) 13:24, 1 October 2025 (UTC)[reply]
Yes, it's possible to generate a list that updates automatically, whether we decide to unpatrol or not. Since most of the commenters seem okay with unpatrolling 2500 to 2700 pages, are you interested in handling the bot part, @Novem? Let me know if you need any help with the bot or the list. – DreamRimmer ■ 16:26, 1 October 2025 (UTC)[reply]
I'm super busy with IRL stuff this month. Would be better if someone else could handle. –Novem Linguae (talk) 16:47, 1 October 2025 (UTC)[reply]
Alright, I will take care of this tomorrow. It’s bedtime here. – DreamRimmer ■ 17:22, 1 October 2025 (UTC)[reply]

Help needed

[edit]

Hi! Today, when I review but not marked as reviewed an article called Mountain View Voice. I had send a NPP note to Yngvadottir but Yngv said to me that, "I dropped a NPP to the wrong editor." Since, Yngvadottir created a page as a redirect only and Funginailuser expanded an article. As a NPPer, to whom do I need to send a NPP note? Fade258 (talk) 15:35, 29 September 2025 (UTC)[reply]

Yngvadottir is the page creator, and the software sends these types of messages to the page creator. While it'd be better if the software could be smart enough to detect that a multi-year redirect had recently been switched to an article, and then identify who did it and send the feedback to them, this is a hard problem to solve from a programming perspective since there'd be a bunch of edge cases.
For now, I would just make a mental note that the software sometimes sends messages to the wrong person, and ignore this with no action. Hope this helps. –Novem Linguae (talk) 16:36, 29 September 2025 (UTC)[reply]
Thanks for feedback and I understand it. On that case, Is it better to leave a custom message to article expansion user? Hope you understand. Fade258 (talk) 16:56, 29 September 2025 (UTC)[reply]
Sure, if you have the time, feel free to copy the message that was sent to Yngvadottir's talk page and paste it on Funginailuser's talk page. –Novem Linguae (talk) 17:19, 29 September 2025 (UTC)[reply]
Sure, I will be doing that from onwards. Fade258 (talk) 17:23, 29 September 2025 (UTC)[reply]

Please submit deletion logs

[edit]

Please submit your deletion logs that were made using Twinkle and not the page curation tool. We will review the logs and award points as appropriate. You have 72 hours to submit them. cc @Utopes – DreamRimmer ■ 15:56, 1 October 2025 (UTC)[reply]