> Main discussion page

Very high unreviewed pages backlog: 9289 articles, as of 14:00, 28 May 2024 (UTC), according to DatBot

All NPPers are encouraged to watchlist the main discussion page to keep on top of issues directly related to reviewing and policy. The current high backlog is generating a lot of "coordination-type" discussion that has been spread about various personal talk pages. It would be better if all coordination discussion were in one place. Since these discussions are not happening on the main discussion page (where this TP has redirected since 2018), moving them here makes more sense than having them scattered on User TPs. @Kudpung, Novem Linguae, and Barkeep49: MB 03:57, 7 July 2022 (UTC)[reply]

Per the discussion on my TP on this matter, I have requested a monthly database report to identify those with zero reviews in the last 6 and 12 months. I put in this request without an immediate need for it since it looks like it could take some time for a volunteer to create it. Before implementing any automatic removal (per NPP guidlines), we should heed Xaosflux's advice and clarify that the criterion (12 months inactivity) means 12 months of NPP inactivity (i.e. no reviews) - not 12 months of en.wp activity (attested by its author Kudpung) and advertise this on the reviewer discussion page and in the next NPP newsletter. @Joe Roe:. MB 15:51, 7 July 2022 (UTC)[reply]

I'm in the minority on this one, so maybe I shouldn't say anything as it will probably just bug people. But I am wondering how de-certifying competent but inactive people helps reduce our queue. Pragmatically speaking, doesn't this just add an unnecessary extra step if we do succeed at inspiring these people to help out again (e.g. via the newsletter)? Food for thought. –Novem Linguae (talk) 19:28, 7 July 2022 (UTC)[reply]

@Novem Linguae: how do we know they are competent? Kudpung กุดผึ้ง (talk) 05:03, 8 July 2022 (UTC)[reply]

Make a list of every one in one thousand reviews and review these reviews. Does it matter if someone who rarely reviews is only semi-competent, compared to a prolific reviewer who has become confident with their idiosyncrasies? SmokeyJoe (talk) 06:23, 8 July 2022 (UTC)[reply]

I think the WP:PERM/NPP admins do a good job of doing thorough background checks. In the absence of evidence of a big problem, I think this is a safe assumption. –Novem Linguae (talk) 07:37, 8 July 2022 (UTC)[reply]

This has always been my point of view. There's some risk of compromised accounts that make Xaos' cleanup helpful but I have generally favored a looser rather than stricter enforcement. Best, Barkeep49 (talk) 19:29, 7 July 2022 (UTC)[reply]

@Barkeep49: I know you don't agree with me on this, but the data I've posted below is fairly convincing, don't you think? Kudpung กุดผึ้ง (talk) 05:03, 8 July 2022 (UTC)[reply]

There is no benefit either to having a long list of people with the right who never use it. Of course, some people do some reviews and move on to other things for a while, then come back. Some people may never be very active and just to one review occasionally. But the threshold is very low - one review in a year. If we send out a message after six months with no reviews, and they still don't manage to do one in the next six months, that is probably a good indicator that they just aren't interested in the project. If someone like this changes their mind, it's easy to request it be restored (not denying that is is an extra step). MB 20:12, 7 July 2022 (UTC)[reply]

Just to clarify: while I'm very much concerned for several reasons that the physical number in the NPP user group should be slimmed down as is done is for other special rights and AfD, I have not been advocating doing anything drastic anytime soon. Hence my suggestion to first obtain the the Quarry lists based on the script by Insertcleverphrasehere, and waiting for the outcome of the current backlog drive. I think it has become evident over the years that suggestions are best based on concrete stats and data, and also RfCs where they are considered absolutely necessary. I have offered to do a thorough, unbiased analysis of the lists. It will be time consuming but I hope it will enable the new coordinator group to move forward one way or another. I have already asked at ICH's talk page if he can produce a first list based on a certain time frame. Kudpung กุดผึ้ง (talk) 20:40, 7 July 2022 (UTC)[reply]

@Kudpung https://quarry.wmcloud.org/query/65845 — Insertcleverphrasehere^{(or here)_{(or here)(or here)}} 22:43, 7 July 2022 (UTC)[reply]

Wow! That was quick. Thank you so much, ICH. Kudpung กุดผึ้ง (talk) 23:00, 7 July 2022 (UTC)[reply]

@Kudpung Well, I tried to do it last night but some of the coding had changed since I last ran it, so had to ask for help over at WP:QUARRY. — Insertcleverphrasehere^{(or here)_{(or here)(or here)}} 23:11, 7 July 2022 (UTC)[reply]

@Novem Linguae I guess it is a security risk. Additionally, if people knew that they had to wear their hats occasionally after collecting them or they would be taken away, maybe some of the hat collecting members of the NPP group might do more patrolling? — Insertcleverphrasehere^{(or here)_{(or here)(or here)}} 22:46, 7 July 2022 (UTC)[reply]

I think possibly a lot that has to do with it is that many users apply for the right in good faith without having read or realised the implications of WP:NPP where it is nevertheless spelled out in bold type in in the lede. They haven't done their homework, the bar for obtaining the right is too low, and not all admins at PERM do sufficient due diligence or accord the right on a 30-day probation. Hence there are dozens of reviewers who have bitten off more than they can chew. Kudpung กุดผึ้ง (talk) 23:00, 7 July 2022 (UTC)[reply]

@Kudpung I don't think the bar for inclusion is too low; I think a lot of the learning necessarily has to be 'on the job'. But I really think that the 30 day probation is a good way to weed out hat-collectors who never intend to use the tools, and potentially to annoty anyone with nefarious intentions (UPE black hats). I used to be against the idea of paring down the list, but I've come around more to the idea that some minimal amount of activity should be required to keep the tools, if only to encourage those who are inactive to do a few reviews. I do think that re-establishment (pending activity on a 30 day probationary period) should be assumed essentially automatic for those who want it reinstated. — Insertcleverphrasehere^{(or here)_{(or here)(or here)}} 23:07, 7 July 2022 (UTC)[reply]

Paring down the reviewer list is necessary. It's not an exercise to name and shame anyone, and most user rights are 'easy come; easy go', and easy come back again. Learning 'on the job' is of course important, that's why we have the NPP School, and more motivation is required to help new reviewers find their feet. I've caught some pretty sticky persistent UPE and blocked and banned them, and only yesterday while patrolling, I caught and CSD flagged an article that I immediately recognised as a hoax (if I had admin tools I could have deleted it on the spot because it was potentially also a negative BLP). But that kind of instinct only comes with experience. I am sure that it would not have been detected by a new(ish) reviewer.

The problems for AI solutions which some NPPers are now asking for, is that the WMF (or more exactly the boss of Phab) is still claiming only yesterday, lack of resources for addressing even simple but urgent software bugs. I contend that as the WMF is awash with money, there is no excuse for not hiring more software engineers. There is going to have to be a major shake up of NPP and I'll stick around to help see it through. I've been talking to some leaders at Growth and Community Tech where I still have a voice (I think). I have a good feeling that now the issues of coordination appear to have been addressed by MB and this page restarted, some progress is going to be made. Paring down the reviewer list will provide an important overview of the performance of NPP.Kudpung กุดผึ้ง (talk) 23:46, 7 July 2022 (UTC)[reply]

Kudpung กุดผึ้ง (talk) 23:46, 7 July 2022 (UTC)[reply]

Activity report

Of 740 reviewers in the period of 12 months up to 30 June this year,

• 320 had made 0 reviews
• a further 45 had made 1 edit.
• 153 had made 2 to 20 reviews.
• 112 had made 100 or more reviews.
• 71 had made over 200 reviews
• 25 had made over 1,000 reviews.
• 12 had made an average of 5.55 reviews per day or more.
• A significant number of the 136 < 10 review patrollers have not patrolled in the last 6 months.

Still to analyse: How many of the 1 to 9 review users are new users.

Anyone is welcome to do a better or more accurate analysis. I won't be offended. A further report will be made to cover the period of the July backlog drive to compare and measure its impact. Kudpung กุดผึ้ง (talk) 00:11, 8 July 2022 (UTC)[reply]

.

I have raised again a request I made over 6 months ago on MMiller (WMF)'s talk page. Please see Suggested Improvements. It concerns additions and updates to the criteria on the filter dialogue panel. There appears to be Phab tickets but it's not clear if this request was included. Could someone please follow up on Phab? Kudpung กุดผึ้ง (talk) 03:32, 8 July 2022 (UTC)[reply]

The July Admin newsletter may not be sent for a few more days. I added a request that admins consider doing some reviewing based on some prior discussion. Kudpung, you didn't say anything. I'd like someone from the project to concur it sounds OK. Wikipedia:Administrators' newsletter. I'm not sure what editorial control there is - there is a note that says contributions are welcome from all users, so I just went ahead and added something. MB 04:23, 8 July 2022 (UTC)[reply]

MB I already saw that. It's perfect. I didn't see any need to say anything. Kudpung กุดผึ้ง (talk) 04:47, 8 July 2022 (UTC)[reply]

Looks good. – Joe (talk) 07:06, 8 July 2022 (UTC)[reply]

Efforts on Phabricator appear to have stalemated. Do we want to slap an RFC tag on the WT:NPPR thread and add it to T:CENT? Is it worth the time/effort/drama? It doesn't directly reduce the backlog, so I feel a bit meh about it, but I guess we should finish what we started. –Novem Linguae (talk) 18:18, 12 July 2022 (UTC)[reply]

Novem Linguae IMO even more important than reducing the backlog is preventing junk, spam, UPE, and other vanity pages from becoming indexed by Google. Kudpung กุดผึ้ง (talk) 03:02, 13 July 2022 (UTC)[reply]

We could do that, but an RFC could drag on for another month. My reading of all the discussion is that 365 has wide support and the devs should go ahead and do that. We could try one more time by asking a non-involved party (at Wikipedia:Closure requests) to assess if there is consensus for 365 which might be enough for the devs to move forward. Then we could give up on indefinite, hold an RFC for that, or just wait awhile and reconsider maybe in a few months. MB 20:26, 12 July 2022 (UTC)[reply]

The only problem with that is that when if we need to do a proper rfc, we would need to start the whole new discussion again with everyone repeating their comments. tbh neither is great and it should just be done, but I get that's not feasible. Lets try ANC, might as well not give everyone even more to do. Happy Editing--IAmChaos 20:59, 12 July 2022 (UTC)[reply]

One potential issue with ANRFC is that the discussion may be hatted, then rejected by the Phabricator volunteer again, then because the discussion is hatted we'd be pressured to start the discussion again from scratch.

It may be worth pinging the Phab volunteer and asking what they'd accept as sufficient consensus. –Novem Linguae (talk) 21:22, 12 July 2022 (UTC)[reply]

Proceed with caution. The conventional wisdom among most WMF staff is that the Wikipedia needs as many new articles as possible irrespective of quality or appropriateness. The official management section of Phab considers itself to be the WMF's upholder of this fallacy and will refuse or stall as many requests as possible from the communities which might be measures to reduce the flow of new article submissions. Volunteers developrs, who appear to be mainly concerned with using MediaWiki for non Foundation projects are also apt to quietly shunt community requests into an overgrown siding in the forest. Any well intentioned comments at Phab from the community will also be interpreted wherever possible as a 'no' vote for a bug repair or feature request, and that's why NOINDEX was thrown out again on the eve of its being rolled out. At the English Wikipedia, those who have longer memories are still smarting from the grossly uncivil behaviour of Buzilla (Phab's predecessor) governance obsessives over ACTRIAL which then took nearly 10 years to resolve. The bottom line is that strictly speaking, Phab should fully recognise the discussion at WT:NPR as an affirmation of the 2012 consensus which Okeys (WMF) promised to enact but as he often did, swept it under the carpet. Kudpung กุดผึ้ง (talk) 02:52, 13 July 2022 (UTC)[reply]

In this case, I believe this ticket was shut down by 2 respected admins from the enwiki community and a Phab volunteer who interpreted that opposition as insufficient consensus, rather than WMF staff. Anyway, what is to be done? I think good options are to ping the Phab volunteer and ask for criteria, or RFC this. I think we should just RFC it. –Novem Linguae (talk) 03:15, 13 July 2022 (UTC)[reply]

I daresay do the ping first. Worst thing that could happen is they say RfC and we do the RfC anyway. At least we wont waste time going to AN/C, or making an unneeded RfC, if it doesnt go that way. Happy Editing--IAmChaos 03:24, 13 July 2022 (UTC)[reply]

I've asked a number of (uninvolved) admins to consider reviewing & closing this discussion, but all have seemed reluctant — I am very keen to get a firm(er) answer to the question y'all have raised, so that this can be implemented. I'm not sure an entire RfC is required, and I'm confident an uninvolved close (i.e. a summarisation of the discussion) would un-stall the phabricator task and get the patch implemented — TNT (talk • she/her) 04:26, 13 July 2022 (UTC)[reply]

I agree with TNT and Kudpung. We should be leading this, not following. We just have to find a willing uninvolved Admin to summarize/close. MB 05:29, 13 July 2022 (UTC)[reply]

Unless things have changed since I deleted my watchlist on 4 March 2010, as far as I know there is no policy that states that discussions have to be closed by an admin. Any experienced and trusted editor of admin calibre can do it. Kudpung กุดผึ้ง (talk) 05:45, 13 July 2022 (UTC)[reply]

I went ahead and made a request at Wikipedia:Closure requests#Wikipedia talk:New pages patrol/Reviewers#Proposal: increase unreviewed new page search engine NOINDEX duration. However don't let this stop you from recruiting a closer who can potentially do it faster :) –Novem Linguae (talk) 05:46, 13 July 2022 (UTC)[reply]

I'm quite happy to let the new NPP coords look for someone. I'm somewhat out of touch these days since 4 March 2010 - and of course since Wikimania stopped happening. The sooner I can get back to my semi-retirement, the better 😉 Kudpung กุดผึ้ง (talk) 06:29, 13 July 2022 (UTC)[reply]

Novem Linguae, I'll just point out that that discussion is by way of an affirmation of an earlier RfC which closed with a consensus 64:11. The WMF promise to roll it out was never fulfilled. You may wish to add that to your closure request. OTOH, it states quite clearly at closure requests: Many discussions result in a reasonably clear consensus, so if the consensus is clear, any editor—even one involved in the discussion—may close the discussion. The default length of a formal request for comment is 30 days; if consensus becomes clear before that and discussion has slowed, then it may be closed earlier. The bolding is mine. Kudpung กุดผึ้ง (talk) 06:52, 13 July 2022 (UTC)[reply]

Feel free to reply to my ANRFC request and add the info about the old RFC if you'd like. I don't recommend an involved close of... really anything on Wikipedia. I'm surprised that such contrarian advice is given by an official page. –Novem Linguae (talk) 07:06, 13 July 2022 (UTC)[reply]

I don't think I will, Novem Linguae, again it's something I'll leave up to the coords. Suffice it to say however, as well as the 2012 RfC, NOINDEX was ironicaly part of the WMF's original development concept of Page Triage / Page Curation in 2012 and again in their 2818 overhaul. Those seem to me to be pretty strong arguments, but the management at Phab and one or two isolated Wikipedians seem to have other opinions. Kudpung กุดผึ้ง (talk) 08:28, 13 July 2022 (UTC)[reply]

I tried the equivalent of a self-close by stating at the ticket that there was clear consensus at least to extend to 365. As that had no affect on the Phab, I agree we should get a formal close by a non-involved closer, preferable an Admin or "trusted editor of admin calibre". MB 15:25, 13 July 2022 (UTC)[reply]

@MB and Novem Linguae: I may have misread his comment - it's full of Phab jargon - but it now looks as if another WMF dev is finding reasons to deny this request. You might not be subscribed to that thread. Kudpung กุดผึ้ง (talk) 02:00, 14 July 2022 (UTC)[reply]

Majavah aka taavi wrote a patch that allows PageTriage NOINDEX to be set to indefinite, and the patch was accepted, and Krinkle thinks he found a major bug in it. We should go silent for a bit and let those two talk and fix the bug. –Novem Linguae (talk) 02:52, 14 July 2022 (UTC)[reply]

Too late, I've already opened my big mouth there, but it's due to my long and gruelling experience with the WMF and the many serious concerns voiced about that organisation's management from all quarters of the volunteer community. Perhaps all it needs now to assuage Krinkle's concerns is to address any non covered points about the two available patches. Our en.Wiki is not his home project and he does not appear to be aware of what is as stake. Kudpung กุดผึ้ง (talk) 04:19, 14 July 2022 (UTC)[reply]

Krinkle works for the WMF. He's a software developer contractor on the performance team. I think a likely scenario here is taavi sees the comments and fixes the bug. Pressure may not be needed at this exact moment. –Novem Linguae (talk) 05:17, 14 July 2022 (UTC)[reply]

Novem Linguae, I've just finished adding all the issues with open tickets on the suggestion page. We are going to have to go review everything and prioritize (from the NPP perspective). I see some of them were suggested by just one person, and then a ticket opened without much discussion. We need to talk about the best way to do this. Maybe ask everyone to pick their top 10 or something like that. But before we get to that stage, what's missing? Most everything there are enhancements. Aren't there more bugs that need fixes?

Also, can you archive the one item that is closed - this one. I don't know how to use that one-click archiver, I assume that is another script you have to install.

And can you take a look at this. This claims that redirects drop out of the queue after 30 days, and asks for the noindex time to be changed to 90 like articles. MB 05:53, 14 July 2022 (UTC)[reply]

Sure, I went ahead and archived the requested thread. The user script you're thinking of is User:Evad37/OneClickArchiver.js.

I don't think most folks could name 10 PageTriage bugs and feature requests off the top of their head. Asking them if they have one or two they'd prefer we work on should suffice. I'll go ahead and create a thread at WT:NPPR. –Novem Linguae (talk) 08:39, 14 July 2022 (UTC)[reply]

I meant ask them to go through the list and pick the their top 10. We can still do this in the future after we have a complete list. You queried phabricator directly recently, does thst show more that should be added to our list? MB 10:23, 14 July 2022 (UTC)[reply]

Very interesting about the 30 day redirect thing. I didn't know about this and this is kind of alarming. I found the code that is responsible for this at https://github.com/wikimedia/mediawiki-extensions-PageTriage/blob/58ef5381d2c1d5455a92090477bd6488544f3bf0/cron/updatePageTriageQueue.php#L80-L102. A patch for this would be easy. I may write one soon. @Rosguill, any objections to not autopatrolling redirects after 30 days? Would this overwhelm redirect patrollers like you, or would you prefer that every redirect get a set of eyes? (Right now they appear to get autopatrolled after 30 days) –Novem Linguae (talk) 08:39, 14 July 2022 (UTC)[reply]

Rosguill asked for this in 2019 according to what Barkeep49 said on the Suggestions page thread. The oldest redirect is currently from June 13, so it appears that they are indeed autopatrolled at 30 days. I have noticed the oldest redirect was a month old for a while now, but never realized this was why. We have probably not reviewed hundreds or thousands of 30 day old redirects by now. MB 10:47, 14 July 2022 (UTC)[reply]

I've been manually patrolling the back end of the redirect log since I raised this in 2019, so we've only been missing thousands of redirects prior to 2019. _signed, Rosguill ^talk 14:47, 14 July 2022 (UTC)[reply]

• Here's my list:

1. Extend No-index time
2. Opening AFDs
3. Put undeletes back in queue
4. Autopatrolled filter
5. New icon for patrolled with tags
6. Disable unreview message appears to have been done - or wasn’t it? Winged Blades of Godric very abruptly stopped editing from India on 10 July 2020. I hope he is ok.
7. Decline CSD/PROD msgs
8. Welcome new user (this is more useful and relevant than having a WikiLove feature)
9. Tool for moving to draftspace (there is an independent script, but it should be functional from the Curation toolbar)

Kudpung กุดผึ้ง (talk) 12:50, 14 July 2022 (UTC)[reply]