Just wanted to know if autopatrolled right will do any good when I am going around creating new user talk pages by giving warnings during RTRC vandalism patrolling. I have created around 4000 user talk pages like that which probably someone else have mark as patrolled. I do so if I see that button during patrolling. Will the user right help me with that, or is it just limited to article mainspace? The Herald (Benison) (talk) 17:15, 20 December 2023 (UTC)[reply]

@The Herald: It will be of no benefit as the WP:NPR team does not patrol anything other than main space creation. Hey man im josh (talk) 17:24, 20 December 2023 (UTC)[reply]

Cool. Thanks :) The Herald (Benison) (talk) 17:26, 20 December 2023 (UTC)[reply]

Hello all. Just wanted to bring to our attention that lately there's been a lot of autopatrol perms being revoked. These 3 are from this month:

• Booberring - Only 841 edits. UPE sockfarm.
• Tumbuka Arch - 3762 edits. Bogus sources.
• PD Slessor - Only 647 edits. Created a bunch of NPOL articles, got autopatrol, then immediately switched to NCORP.

I think it is especially risky to give autopatrol to editors in the 0–1500 edit range, and I don't think the risk reward calculation is worth it in that scenario. Better for NPPs to do a bit more work. Anyway, just wanted to bring this to our attention. We should consider tightening up our criteria for granting autopatrol. It doesn't even necessarily need to be a formal tightening up with an RFC or a survey or editing the criteria. We can simply start being stricter about granting it, especially with new users who don't have enough of a track record to convince us that they are not UPEs.

Pinging admins that have been active in granting autopatrol lately, in no particular order: @Joe Roe, HJ Mitchell, Callanecc, Schwede66, and Chetsford:.

I hope you find this information useful. Happy holidays. –Novem Linguae (talk) 11:53, 22 December 2023 (UTC)[reply]

• I wholeheartedly agree with the suggestion. I would only add that I think it should be formalized and added as a strict benchmark to the guidelines at WP:APAT, which is currently written in such a way that there are essentially no objective criteria, everything is "suggested". The result of this ambiguity are ocassional cases in which I have not granted someone permissions and then have to process a bevy of complaints including, sometimes, from other admins as to why I'm not granting so-and-so such-and-such. (Indeed, in the case of PD Slessor, above, I initially granted them Autopatrolled for 90 days and was immediately asked to justify by a third party why I limited it to 90 days!) This leads to a system in which (at least in my experience) the Admin faces the potential of being punished with essay-writing assignments if they decline marginal cases (e.g. [1]), while no such punishment awaits them if they just accept the request. I'd suggest this probably has some deleterious, subliminal effect on decisionmaking. Chetsford (talk) 14:21, 22 December 2023 (UTC)[reply]

  I think criteria being somewhat flexible is a good thing, but I agree some reform is needed. For example, I think autopatrolled should be primarily not self-nominated, and it should be rarely granted on self-nominations by relatively new editors. Pinging Firefly and Tamzin since I think they have some ideas in this area. In any case, I think a design principle should be that we should not make the lives of admins working on this backlog harder. MarioGom (talk) 19:28, 22 December 2023 (UTC)[reply]

  I think the biggest problem with autopatrolled right now is that it drives scrutiny from 100% to ~0% (especially if the articles are on relatively low-profile subjects). I think we could change that with only a slight addition to backlogs. Something like

  • Every 20th article by an AP, or one article every 3 months, whichever comes first, is unpatrolled by bot and left for an NPR/sysop to review
  • AP is given for a fixed term, maybe 1 year the first time and 2 years subsequently. After it expires, the user is automatically renominated once they've had 3 articles patrolled manually. The reviewing admin spot-checks a few articles made while AP and grants or declines primarily on that basis; in most cases the editor's participation won't even be needed. (For existing APs, start with a mass removal of anyone who hasn't created an article in over a year, then come up with a staggered timeline for switching remaining permanent APs to temporary.)

  I also would support a ban or very strong discouragement of self-noms, maybe even restricting nomination to users who have patrolled articles by the editor in question. -- Tamzin^{[cetacean needed]} (they|xe|she) 19:56, 22 December 2023 (UTC)[reply]

  @Utopes, FYI. Regarding this: Perhaps we should have supported the admin doing a temporary rather than permanent grant in this situation, since the AP applicant had weak credentials and ended up getting their AP revoked. progress could be reassessed in 90 days to make sure things are golden. I think that's the idea behind a temporary grant: to reassess it at the end of the temporary period. There would be no reassessment done with a permanent grant (there is no efficient way to trigger reassessments for permanent grants). However having to re-apply after 90 days guarantees a re-assessment. –Novem Linguae (talk) 20:33, 22 December 2023 (UTC)[reply]

IMO the criteria sort of works in reverse. IMO anyone who creates a large amount of new articles in a 22 year old enclyclopedia needs closer scrutiny, yet that is a requirement to receive the tool which removes all scrutiny. Maybe I'm biased because I got turned down because of that. :-) Sincerely, North8000 (talk) 15:28, 22 December 2023 (UTC)[reply]

Short version: I'm in favor of stricter access and closer scrutiny. North8000 (talk) 20:58, 22 December 2023 (UTC)[reply]

Thanks for the ping. I suggest that Tamzin's suggestions are sensible. Schwede66 05:06, 23 December 2023 (UTC)[reply]

• Unlike other user rights, autopatrol doesn't provide any benefit to the user themselves (unless they wish to use it to avoid scrutiny). It exists for the benefit of NPP. So I think it would make sense to remove the ability to self-nominate and only consider third-party nominations from NPPers. This could still be gamed, but it would make it somewhat harder. I also think that as a general rule, the right shouldn't be granted to editors who exclusively create brief stubs on notable-by-default topics (NPOL, NGEO, NSPECIES etc) as these creations don't demonstrate a broad understanding of notability or content policy and require minimal effort on behalf of NPP. Spicy (talk) 15:45, 23 December 2023 (UTC)[reply]

  As long as we advertise to NPPers that they can (and should) nominate people who clearly don't need their attention, I'm very happy with making PERM/A nomination only. The purpose of the right is to support NPP, so we as admins should be supporting them as they ask us to. Making it nomination-only will reduce the propensity for hat collecting and make it less of a feather in the cap. As an aside, I agree with N8k about more scrutiny for editors creating large numbers of new articles in this day and age. HJ Mitchell | Penny for your thoughts? 21:08, 23 December 2023 (UTC)[reply]

  From my NPP standpoint, I think autopatrol has created more work that it saved. A couple of times I ended up spending much time at AFD debates (which ended up as deciding that they failed wp:notability) where the main "keep" argument was that there were many similar articles in walled gardens created by autopatrolled folks. North8000 (talk) 15:38, 24 December 2023 (UTC)[reply]

  You couldn't be more wrong about that. Currently about 25% of all new articles are autopatrolled, and NPP is still chronically backlogged. Without autopatrolled, the system would collapse. – Joe (talk) 13:54, 28 December 2023 (UTC)[reply]

  I think raw number of articles affected can be a bit misleading, because the high-quality articles that are supposed to be the basis for this permission are typically faster to review than the average new article. Factor in the de facto re-reviewing of each autopatrol candidate's articles when their application is being considered (as well as the emotional attrition to admins and good-faith candidates alike when a candidate's record is not quite strong enough) and it's not as clear of time-saver. _signed, Rosguill ^talk 21:56, 28 December 2023 (UTC)[reply]

  We're still talking about thousands of articles a month, it adds up. – Joe (talk) 09:51, 30 December 2023 (UTC)[reply]

  I agree. Autopatrolled is quantitatively a large component of NPP. I think we should continue granting autopatrolled to even more editors who create a lot of good articles. I hope that's compatible with some of the reform proposals under discussion. MarioGom (talk) 14:25, 1 January 2024 (UTC)[reply]

  +1 on the nomination-only thing. Jo-Jo Eumerus (talk) 07:36, 24 December 2023 (UTC)[reply]

  I think nomination-only could help a little, but it's easily gamed. Just hop on Discord, butter up a newbie NPPer, and get them to request it for you. This already happens now, even when self-requests are allowed. It's likely to reduce the number of borderline requests we have to process (i.e. "I just hit 25 articles, can I have this badge now?"), but not deter any actual bad actors. – Joe (talk) 13:47, 28 December 2023 (UTC)[reply]

  Have you ever seen or heard of this happening? I haven't. NPPs tend to nominate for autopatrol in big bursts of multiple noms, suggesting that they are objectively looking over the new pages feed or a Quarry query, rather than getting "buttered up" or doing a favor for someone. –Novem Linguae (talk) 21:48, 28 December 2023 (UTC)[reply]

  Yes, of course. Why else would I bring it up? – Joe (talk) 15:25, 29 December 2023 (UTC)[reply]

  I personally have never seen this occur, and I’ve been in the NPP discord for months. I am disappointed that you think so lowly of those who act as the last line of defense against bad articles. Do have any evidence that this has happened, as you say it has? - 🔥𝑰𝒍𝒍𝒖𝒔𝒊𝒐𝒏 𝑭𝒍𝒂𝒎𝒆 (𝒕𝒂𝒍𝒌)🔥 01:36, 29 December 2023 (UTC)[reply]

  "Months" is the clue here. – Joe (talk) 15:26, 29 December 2023 (UTC)[reply]

  Then can you point to an example where it’s happened, regardless of time? - 🔥𝑰𝒍𝒍𝒖𝒔𝒊𝒐𝒏 𝑭𝒍𝒂𝒎𝒆 (𝒕𝒂𝒍𝒌)🔥 16:10, 29 December 2023 (UTC)[reply]

  I'm not going to start unnecessary drama just to convince you I'm not making stuff up. Besides, linking to Discord conversations is verboten. – Joe (talk) 21:23, 29 December 2023 (UTC)[reply]

  I fail to see how providing evidence for a claim that only you seem to believe will cause drama. And the linked proposal is for connecting Discord users to on-wiki accounts, and I don’t see how that applies here. - 🔥𝑰𝒍𝒍𝒖𝒔𝒊𝒐𝒏 𝑭𝒍𝒂𝒎𝒆 (𝒕𝒂𝒍𝒌)🔥 21:33, 29 December 2023 (UTC)[reply]

• I agree entirely that PERM/A should be "nomination from NPP" only, and that we should be exceptionally cautious about granting AP to those who create stubs on obviously-notable topics. Paid editors use this route frequently to game AP and then go on to spam rubbish about corporations, "influencers", or governments who are lining their pockets. Personally I would start there, and see how it goes - if that doesn't resolve the issue then we can look at more complex proposals like Tamzin's. I do not oppose such a solution, but I think we should try the simpler ones first. firefly ( t · c ) 11:45, 26 December 2023 (UTC)[reply]

  Paid editors use this route frequently to game AP and then go on to spam rubbish about corporations, "influencers", or governments who are lining their pockets. @Firefly: Could you provide some examples of this? – Joe (talk) 13:48, 28 December 2023 (UTC)[reply]

  @Joe Roe we have one example above. "PD Slessor" created a bunch of articles on minor politicians and then their first creation after gaining AP was Amini (startup). There are quite a few more in the history of the Ugbedeg SPI - Luciapop is one. They created a bunch of articles on African jurists and lawyers (their article creation history), gained AP and then created things like Parsiq (undelete link) and Victory Obasi (see AfD - some of these articles were deleted under G5, but the rough consensus at the AfD was that they were deletable on the merits). There's Soheelmoon from the same SPI who did something similar and then moved on to entrepreneurs. There are more - I will try to dig them out.

  "frequently" may have been an overstatement (apologies), but it doesn't need to be a terribly frequent occurrence for it to cause a mess. firefly ( t · c ) 15:25, 28 December 2023 (UTC)[reply]

  Yeah, I've no doubt it happens, but frequently enough to worry about? I'm not so sure. Especially if you consider these few examples as a proportion of the (probably hundreds?) of good faith stub creators with autopatrolled. – Joe (talk) 16:43, 28 December 2023 (UTC)[reply]

  I think it happens frequently enough for us to consider possible mitigations (some data). Other than UPE, there's a second group of concern, which are newbie users who get autopatrolled and end up creating a large amount of articles with serious problems, which go completely undetected for too much time. This is harmful for themselves, since they do not get the early feedback that would have helped them, and once it is noticed, it is a much more frustrating experience for everyone. MarioGom (talk) 14:21, 1 January 2024 (UTC)[reply]

I'm also guilty of this I once nominated an editor for autopatrol because they were making a bunch of articles annnnd they ended up being a sock. I don't know if there's a way to assess content creation as part of granting autopatrol or something like that. I figured I would point out my own mistake since it happened to me. Dr vulpes (Talk) 05:13, 27 December 2023 (UTC)[reply]

• I agree that the criteria for autopatrolled should be stricter (and have been advocating this for years), but I don't agree that individual admins responding to PERM requests can "simply start being stricter about granting it". The current minimum criterion is 25+ "clean" articles and though the guidelines allow some discretion, when someone who meets clearly meets that standard (like all of the three users mentioned above), we can't simply say "sorry, no, I'm getting bad vibes". It's not how admin tools are supposed to be used (to enforce consensus, not supplant it), not fair, not consistent (what if the next admin doesn't get bad vibes?). Besides, we haven't got magical bullshit detectors – spotting UPEs is hard, and anyone who tells you to otherwise hasn't spent enough time doing it. So if tightening needs to be done—and again, I think it does too—we need to roll up our sleeves and figure out some new criteria that work better. – Joe (talk) 13:41, 28 December 2023 (UTC)[reply]

Time-limited grants

I think this is the most immediately practical of the suggestions so far above, so spinning out its own section: should we make autopatrolled a time-limited grant by default? That functionality wasn't available when the current criteria were written, but since it was introduced it's been really useful with other perms (e.g. WP:NPR, where initial 'trial periods' are now the norm), and I think it makes a lot of sense for autopatrolled specifically. But rather than coming up with a fixed period for everyone, I'd suggest a more flexible tiered system:

• For self-requests, the requester should specify how long they need autopatrolled for (e.g. "I plan to create a lot of articles on frogs over the next three months"). The reviewing admin could then weigh the specific risks of granting versus the specific length of time requested and we could do away with the general minimum article criterion. There'd be no automatic renewal, the user would have to re-request and show a specific need again.
• Nominations from NPPers would be for a set time period (6 months or a year maybe). When it's close to expiry, and if the editor is still actively creating articles, a bot would automatically post a new request so the user's creations can be re-reviewed and the grant possibly renewed.
• Exceptionally, highly prolific users could have an indefinite grant. This could happen automatically after a certain number of limited grants, but I think it really should be limited to the few editors who really do pump out vast numbers of articles over a sustained time period and do not need to be reviewed.

I think the main advantage of tiering things like this is that it increases the degree of scrutiny on autopatrolled editors without majorly inflating the workload at WP:PERM/A (which, since only a few admins regularly work it, I'd be grateful if people kept in mind here). Hat collectors probably won't bother to keep making requests after the first expires, people who go inactive in article creation will naturally drop out of the system, and people who are genuinely high-volume, problem-free article creators will naturally gravitate towards indefinite grants. – Joe (talk) 14:14, 28 December 2023 (UTC)[reply]

I would happily support this. firefly ( t · c ) 15:25, 28 December 2023 (UTC)[reply]

I like this proposal. I would still prefer my suggestion of letting temp grants expire and then having a bot renominate after a certain number of manual patrols, but ultimately either order works well enough. I'm not convinced that there's a need to allow indefinite grants, though. For someone like Epicgenius who makes a large number of high-quality articles, sure, it wouldn't do much harm to grant indef, but it also doesn't cost us much to re-up every year. What I'd be more inclined to support is something like "6 months on first grant, 1 year on second, 1-5 years on all subsequent grants at admin's discretion". The Epicgenius of today obviously doesn't need patrolling, and the Epicgenius of 2028 likely doesn't either. The Epicgenius of 2040 who's returning from a 10-year wikibreak? Indefinite is a long time. And even if we add inactivity removal criteria (which would be my next proposal if something like this one passes) there's a risk of someone being active but gradually drifting away from community norms. -- Tamzin^{[cetacean needed]} (they|xe|she) 18:24, 28 December 2023 (UTC)[reply]

I don't really see the need for the manual patrols because If an admin is going to (re)grant autopatrolled, they're going to have to review some articles anyway, and hopefully more than three. So why the double the work and make the rules more complicated?

If we were designing autopatrolled from scratch, then I'd be with you on the no indefinite grants thing. But as it is, we have nearly 5000 users who got the right indefinitely based on knocking out 25 stubs or just the whim of an admin friend way back when. I think we can and should try to gradually transition as many of those as possible temporary grants, but realistically doing so en masse will be more trouble than it's worth, so it doesn't make sense to me to be so strict. – Joe (talk) 19:12, 28 December 2023 (UTC)[reply]

I have no problems with admins having indef grants, especially as they are likely to create all sorts of non-article pages that also are subject to patrolling - so many people forget this is page patrol, not only article patrol. — xaosflux ^Talk 19:20, 28 December 2023 (UTC)[reply]

We could probably start by removing autopatrol for users who created 0 articles in 5 years or something like that. I think that should be a low drama threshold. After some time with the new rules in place, transitioning all indef grants to a 1 or 2 years grant will probably be quite natural. MarioGom (talk) 14:44, 1 January 2024 (UTC)[reply]

Is this solving the right problem though? The problem isn't that we grant autopatrol for too long, the problem is that we accidentally grant autopatrol to UPEs. Also, if we start making all grants from now on temporary, then that 25% number that Joe mentioned above will start declining, which would be bad for the backlog.

I'd be in favor of implementing a high minimum edit count. That would increase the amount of work UPEs have to do to prepare an account for autopatrol, and would also filter out newer users that may not be familiar enough with notability. Looking big picture, editors with lower edit counts are much higher risk for granting autopatrol. Two of the three autopatrols revoked this month were for users with less than 1000 edits. However I also know that Wikipedians hate edit count red lines that are above 500, so I doubt this idea will gain traction. –Novem Linguae (talk) 22:06, 28 December 2023 (UTC)[reply]

Well, the goal is to increase the overall level of scrutiny, which should also mean that mistakes like that happen less often. I've no objection to also adding an edit requirement and/or raising the article requirement, but like Tamzin I think the primary problem is that autopatrolled is an all-or-nothing proposition. Admins handling it are essentially asked to predict the future -- will this user who's created X good articles continue to do so, or suddenly start creating paid-for spam? -- and that's a fundamentally hard thing to do, no matter how high the minimum criteria are – Joe (talk) 15:39, 29 December 2023 (UTC)[reply]

If autopatrolled on self-noms is limited to short timeframes, the effort-reward ratio of gaming AP would be far worse for UPE. Hopefully this leads to some of the recurrent offenders giving up on this particular tactic. Those who still succeed (which will eventually happen) will be subject to earlier scrutiny. This is still compatible with other reforms though. MarioGom (talk) 14:39, 1 January 2024 (UTC)[reply]

Wondering, would these proposals have any effect on those who already have autopatrolled rights? BeanieFan11 (talk) 21:52, 30 December 2023 (UTC)[reply]

I was browsing this Page and noticed that no Administrator was tending to the requests, I saw a request that has been pending since the 1st Of March, It is not a protocol to do this, but This is a game-changing tool, It is pretty important, even if this may not be regular, Someone should check up on the Request page at least twice a month

Cheers, Arotparaarms (talk) 11:52, 16 March 2024 (UTC)[reply]

I check on it every week or so when I remember, though lately due to forces outside of my control I am not always notified when a new request is made. Primefac (talk) 14:05, 18 March 2024 (UTC)[reply]

Ok, At least You try though :D Arotparaarms (talk) 16:46, 18 March 2024 (UTC)[reply]

Howdy folks. I just created User:Novem Linguae/Scripts/UserTalkErasedSectionsDetector.js, which seems like it might be useful to WP:PERM admins. It will alert you with a yellow banner at the top of a user talk page if there is a lot of self-deletion going on on the user talk page. This could be a red flag for a user deleting warnings without archiving them. Feel free to leave feedback on the user script's talk page. Happy editing. –Novem Linguae (talk) 12:00, 21 March 2024 (UTC)[reply]

Ooooh, I just added it.

Warning: This user has removed content from this page (probably without archiving it) in 49 of the last 500 revisions. Click here to see diffs.

Then there's a dropdown list of all the difs, linked with the edit summaries. That's cool. Thanks Novem! Hey man im josh (talk) 12:45, 21 March 2024 (UTC)[reply]

 You are invited to join the discussion at Template talk:New Page Reviewer granted § Survey on what bullets/tips/Discord links to include. –Novem Linguae (talk) 19:04, 1 April 2024 (UTC)[reply]

In spite of being logged in, I get a message in a green box that reads Sorry, unregistered users cannot be granted permissions due to technical restrictions. Please create an account in order to request user account permissions. I'm IP-block exempt and wonder if that might be the issue (but I get the same message with my VPN temporarily off)?

I had the same issue during a previous drive, and I just worked around it by manually added myself in a page edit. If this is a general issue, however, it might be worth addressing at the source. Since NPP is a relatively thankless task, I think Wikipedia should make it as easy as possible for anyone plausibly qualified to have a trial go—especially during a backlog drive.

Admittedly clueless, however, about the underlying technical infrastructure —

Cheers, Patrick J. Welsh (talk) 21:36, 17 April 2024 (UTC)[reply]

Prob something wrong with the pagenotice, are you still able to edit the box below that, or is it read only? — xaosflux ^Talk 22:32, 17 April 2024 (UTC)[reply]

There should be a CSS that hides that from you, be sure you are not blocking scripts. — xaosflux ^Talk 22:34, 17 April 2024 (UTC)[reply]

I have not attempted to make any changes, but it appears that, as an autoconfirmed user, I could could make edits basically anywhere.

My usual browser is Firefox, and I have multiple ad/script–blocking add-ons installed. I get the same issue, however, on Safari, which I have not loaded up with the same kind of privacy protection. Patrick J. Welsh (talk) 22:47, 17 April 2024 (UTC)[reply]

Also, there is no "box below": this disqualifying message cuts directly to the list of individual requests for permissions.

I believe I can still add myself, but only by manually adapting the format of the requests above—which I think is not how this is supposed to work and appear. Patrick J. Welsh (talk) 22:57, 17 April 2024 (UTC)[reply]

A screenshot could be helpful for diagnosing here, if you're willing. –Novem Linguae (talk) 00:11, 18 April 2024 (UTC)[reply]

Just responded by email. Could not get Wikipedia to accept my screenshot. Patrick J. Welsh (talk) 00:43, 18 April 2024 (UTC)[reply]

Per your email response, the default screenshot filename was indeed the problem. Hope this helps! Patrick J. Welsh (talk) 00:53, 18 April 2024 (UTC)[reply]

That's helpful, thank you. Bear with me while I techno babble a little bit, which may help me or someone else solve this.

From that screenshot we can tell 1) you're for sure logged in, 2) it's not an edit notice that's the problem but rather the template Wikipedia:Requests for permissions/Subpage.

From a technical perspective, anything that is only to be viewed by someone with X perm is given an HTML class of something like anonymous-show or user-show, both used here. These are selectively hidden through some wizardry in MediaWiki:Common.css#L-193 and related .css files.

If your ad blockers or noscript browser extensions were interfering with this process, I'd expect there to be two green boxes there, like I get when I visit the page with ?safemode=1 on. It's very interesting that you're logged in but you do not see the "guidelines for granting" box.

Can you try blanking User:PatrickJWelsh/common.css and see if that does anything? Right now you have JavaScript in this CSS file so the file is broken and won't run.

Can you also try visiting the safemode=1 link above and tell me which 1 or 2 boxes you see? –Novem Linguae (talk) 01:29, 18 April 2024 (UTC)[reply]

Patrick J. Welsh (talk) 01:51, 18 April 2024 (UTC)[reply]

PatrickJWelsh, it would be good if you could add a category to files when you upload them to Wiki Commons. Schwede66 02:52, 18 April 2024 (UTC)[reply]

OK, 2 boxes with safe mode. That makes sense since safe mode turns off MediaWiki:common.css. And blanking your User:PatrickJWelsh/common.css didn't fix anything in normal mode?

Maybe try temporarily disabling your privacy extensions in your browser and see if that does anything? –Novem Linguae (talk) 06:28, 18 April 2024 (UTC)[reply]