This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
Quality scoring and improved filter logic to be added to New Pages Feed tomorrow
Two enhancements are planned to be added to the New Pages Feed tomorrow as part of this project. This post summarizes them briefly, and I will post again with more information once they are in the feed and verified to be working correctly.
Quality scoring with ORES: previously announced here, this will add two new sets of filters to the feed, and new information for each page in the feed.
Improved filter logic: discussed in-depth here, this is a fix to a long-standing limitation to the filters. Previously, reviewers could only review redirects and pages nominated for deletion by lumping them in with all other pages. Going forward, reviewers will be able to make the feed show any combination of pages so they can produce exactly the list they need for their workflow. Specifically, the "Reviewed / Unreviewed" checkboxes will be separated from checkboxes for "Redirects / Nominated for deletion / All others". "All others" refers to pages that are not redirects or nominated for deletion. We will be migrating all reviewers' existing filter settings such that they reflect the same logic as they currently do.
If all goes as planned, these changes will appear in the feed over the course of tomorrow. However, if something goes wrong in the deployment, we will add the changes at the beginning of next week instead. Please let me know if you have any questions, and please speak up right away if anything looks wrong in the New Pages Feed tomorrow! -- MMiller (WMF) (talk) 19:43, 3 October 2018 (UTC)
Hi, I've just had a look at the changes to the new page feed. Theres a slight but not major issue. If I select a filter with no articles (such as Vandalism) it is not possible to see and click the "Set filters" to change the filters back as it extends out of the page. I have however found a temporary fix, which is to to zoom the site out till the page elements are smaller and the button can be seen again. ~Araratic|talk08:12, 5 October 2018 (UTC)
Thanks for trying things out, Araratic. Yes, that's always been an issue with the feed -- but I see now that it will occur more frequently, because with more ways to the filter the feed, it is easier to end up with zero pages in the list. We filed it here, and I'll talk with our team about whether there is a quick and easy way to fix this. -- MMiller (WMF) (talk) 17:53, 5 October 2018 (UTC)
MMiller (WMF) - I suspect I'm encountering the same issue: if I go into the history and revert to a previous version, then the resulting article does not display the sidebar anymore (so I can't mark as reviewed or add tags that way). I've been going back to the feed interface and selecting the article again, at which point it reappears. Bit of a speed bump! --Elmidae (talk · contribs) 15:21, 5 October 2018 (UTC)
@Elmidae: this sounds like an issue that we expect should be fixed by now. Are you saying that when you go into the "View history" tab of an article, revert, and then try to use the Page Curation toolbar, the toolbar is gone? Has that worked for you in the past? Could you give an example of article for which it happened? Thank you. -- MMiller (WMF) (talk) 17:53, 5 October 2018 (UTC)
@MMiller (WMF): yes, that was the issue at time of posting; however now it seems to work (as just tested when reverting John Paul's Rock to redirect) - bar was present upon reverting to version n-2. So I suppose I got the last of the pre-fix behaviour :) --Elmidae (talk · contribs) 19:17, 5 October 2018 (UTC)
@MMiller (WMF): I'm sorry to say the issue still persists (not sure why it worked for a spell there). Multiple instances in sequence, the latest with Konstantin Kastrioti: article displayed with sidebar; went into history, from there reverted to previous version; deployed; sidebar absent from article. Had to follow link from NPP browser again for it to reappear. --Elmidae (talk · contribs) 16:04, 8 October 2018 (UTC)
@Elmidae: I'm sorry you're still seeing that issue; very frustrating. We've heard a couple related issues, and we're tracking here, if you want to follow along. Could you log out and back in again, and let me know if you're still seeing it? I'll keep you updated as we look into this. -- MMiller (WMF) (talk) 00:07, 10 October 2018 (UTC)
@MMiller (WMF): Persists after logging out and back in. I'm surprised I appear to be the only one with that issue - as it's pretty in-your-face, I suspect no one else is encountering it. If it's somehow caused by my personal config then I guess it's unliley there will be an easy fix. --Elmidae (talk · contribs) 12:31, 12 October 2018 (UTC)
Deployment complete -- next steps
Hi all -- the deployment was successful yesterday, and in our testing we are seeing all pages in the New Pages Feed having the "Predicted class" and "Potential issues" classifications that we expect. The expanded filters related to "Nominated for deletion" and "Redirects" also seem to be working as planned. Please let us know if you see anything different.
In this post, I want to give some more background on how the new ORES classifications can be used for new page review. As this community updates any documentation around the New Pages Feed and how to review pages, our team is happy to help with any explanations or screenshots. Let me know!
ORES is a system built by the WMF Scoring Platform team, led by Halfak (WMF), that automatically classifies edits and pages using machine learning. ORES models are in use at the Recent Changes and Watchlist feeds, where they estimate "content quality" and "user intent". We have added two different models to the New Pages Feed, which estimate the "predicted class" and whether an article has "potential issues". The models are built by looking at existing examples of articles that have been given a class, or shown to have issues, and then the algorithm learns what it looks like when future articles have those same characteristics. The classifications will update in real time as pages change. As as an editor works on their new page, its "predicted class" may rise "Stub" to "Start" in the New Pages Feed.
For instance, as I write this, there are 5 articles in the New Pages Feed marked as potentially spam, vandalism, or attack. Those might be pages to review soon, because they might have the most egregious issues associated with them. There are also 31 pages predicted to be "B-class" or better. Those also might be good pages to review soon because of their high quality.
It's important to note, as was referenced many times in the community discussion around planning these enhancements, that these predictions are only predictions. Because they are only suggestions from an algorithm, they are often wrong. Reviewers are meant to use them to find pages that are more likely to have those characteristics, in order to help make reviewing work more efficient. They can also be taken into account when doing a review. But at the end of the day, as several experienced reviewers emphasized in the community discussion, human judgment is still what should be deciding whether a page is of high quality or not.
As reviewers work with these models, cases will come up where the models seem to be wrong. It is really helpful to the Scoring Platform team to report those cases! They can use those to recalibrate and improve the models. Here is where and how to do that.
Please let us know if you have any questions, concerns, or bugs with the new ORES classifications. Our next (and final) enhancement to the New Pages Feed will be the addition of copyvio detection, planned for the week of October 15 or October 22. I will be back with more information on that. -- MMiller (WMF) (talk) 18:33, 5 October 2018 (UTC)
IP reviews?
Something strange I noticed over at Wikipedia:Database reports/Top new article reviewers: there is an IP (with no other contributions) which has inexplicably marked 20 redirects as reviewed. These also appear in the patrol log, which would usually mean they were marked as reviewed using Twinkle, but that doesn't make any sense as I'm pretty sure that you can't even use Twinkle without being logged in. I can't really find anything to explain it. Any ideas folks? — Insertcleverphrasehere(or here)17:03, 4 October 2018 (UTC)
(edit conflict) +1, this puzzled me, too. I even welcomed them. It’s an IPv6 IP if that is of any gravity to the matter at hand.
I would like to note, however, that it’s [probably] possible for one’s contributions to appear on the patrol log without using Twinkle, as, I think when I patrol pages when I am on mobile, my contributions appear on the patrol log. Again, I am not totally sure, nor am I java script wizard. Regards, SshibumXZ (talk·contribs). 17:21, 4 October 2018 (UTC)
Thanks, Ammarpad. Yes, Xaosflux reported this yesterday in T206130 -- people were able to review via the API without having the right role. We think this was open in that way for about 6 weeks, introduced by accident in a code change. The fix was deployed late yesterday, and so this shouldn't be possible anymore. -- MMiller (WMF) (talk) 17:49, 4 October 2018 (UTC)
@Natureium: See: [1]. They seem like totally fine redirects (with pretty random subjects) to the point that I was wondering if the bug had been because an NPR reviewer's reviews had accidentally been assigned to an IP somehow. — Insertcleverphrasehere(or here)18:18, 4 October 2018 (UTC)
That's assuming the only person who patrolled without having the perm was that one IP. Do we know if that's the case? Natureium (talk) 18:19, 4 October 2018 (UTC)
However, I do not get the curation toolbar when following a link from the browser, which makes it kind of pointless. Anyone else having that problem? --Elmidae (talk · contribs) 07:32, 12 October 2018 (UTC)
@Rentier, Vexations, L3X1, TonyBallioni, and Elmidae: It is most probably because of some change in mediawiki API. If you stumble upon freshly created page from anywhere except new pages feed, even then you dont get to see the curation toolbar. In old days, one could get toolbar for reviewed pages even after few days. But you can get it by entering ?showcurationtoolbar=1 in the address bar after the current address, and ?redirect=no&showcurationtoolbar=1 for redirects. I think we should contact someone from WMF/tech team to show the curation bar upto 15 days even after the page was patrolled.Apologies for the mass ping. —usernamekiran(talk)10:29, 12 October 2018 (UTC)
One would think that setting showcurationtoolbar=0 would hide the toolbar, but it doesn't. Is there another way to turn it off? I don't always want to see it. Vexations (talk) 11:03, 12 October 2018 (UTC)
If you are having trouble with the Page Curation Toolbar not coming up, try clicking the "Curate this article" in the "Tools" menu on the left, it finally fixed my bugged PC toolset which refused to come up anywhere on any articles regardless of the ?showcurationtoolbar=1 string or where I navigated from. — Insertcleverphrasehere(or here)22:00, 16 October 2018 (UTC)
recent changes
@MMiller (WMF): I've noticed that I no longer see the mark as patrolled link that used to appear on unreviewed pages when I did not have the curation toolbar open. I also can't close the curation toolbar anymore. Is this intentional? --Vexations (talk) 13:02, 9 October 2018 (UTC)
@Vexations: thanks for posting. That definitely is not intended behavior, and sounds annoying. We created this task to track this issue and related ones. We've been looking into it today, and will do some more tomorrow. For now, it's helpful if you tell us what pages you see this behavior on. And also whether it goes away from logging out and back in. If anyone else sees something similar, please speak up! Thank you, and I will keep you updated. - MMiller (WMF) (talk) 00:05, 10 October 2018 (UTC)
Thanks for looking into this, MMiller (WMF). For example, after logging out and logging in again, when I visit Wan Mohamad Nazarie Wan Mahmood, the page curation toolbar is not open, and no "mark as patrolled" link is displayed. when I click on "curate this article" it opens the toolbar, but that now no longer has a way to close it. --Vexations (talk) 00:13, 10 October 2018 (UTC)
@MMiller (WMF): My experience mirrors Vexations. If I open a NPP from the queue I get the toolbar (no doubt thanks to ?showcurationtoolbar=1) but if I navigate directly to any page in the queue I do not ever get the toolbar. This is different than what I originally reported which was just affecting some pages and where if I navigated directly most pages would still automatically show the toolbar. Best, Barkeep49 (talk) 00:17, 10 October 2018 (UTC)
I noticed the same problem several days ago. If I navigate to a new page, I do not see a toolbar, but I still see a link inviting me to mark the page as patrolled (similar to e.g. such links on user pages). If I click the link, it disappears, but the page does not make it to my review log.--Ymblanter (talk) 07:54, 10 October 2018 (UTC)
MMiller (WMF) - just finished promoting a few drafts in AFC (on an iPad nonetheless) and love the tools (copyvio detect, etc.) being in the drop down menu at my disposal. Also love that the newly promoted page already has the curation tool in place and it’s checked ✅ as reviewed, so we’re now getting twice the work done in half the time. Thank you for this major improvement. Atsme✍🏻📧13:42, 15 October 2018 (UTC)
RfDs in the New Pages Feed
Over the past week there have been two incidents where about 400 total redirects were mass nominated for deletion (unicode & Australian rail stations). Since these were redirects that were no longer redirects they appeared in the New Pages Feed. I askedMMiller (WMF) if it was possible to treat these like articles nominated for deletion for purposes of filtering the queue. He said probably if there was community support. I am asking here for community support.
Support If they are nominated for deletion, then they should be filtered by the 'nominated for deletion' tick box, regardless of whether they are redirects or articles. If you want them to appear in your feed, you can simply leave the 'nominated for deletion' box ticked at Special:NewPagesFeed. — Insertcleverphrasehere(or here)03:45, 16 October 2018 (UTC)
Oppose as this defeats the purpose of why we originally decided to have redirects turned into articles in the New Pages Feed: it's one of the easiest ways to sneak paid spam in. As pro-WMF as I am, I'm skeptical of any solution to this problem that wouldn't make the more important issue of catching redirect hijacks more difficult. Additionally, the easiest solution here is to tell anyone who nominates 400 redirects for deletion to stop being disruptive, because that is what nominating that large a number of redirects for RfD is. It floods our capacity both in NPR and at RfD for things that 99/100 don't really matter and are only commented on by a very small subsection of the community. TonyBallioni (talk) 04:34, 16 October 2018 (UTC)
@TonyBallioni: I am not suggesting any redirects or former redirects leave the feed. Simply that "redirects" with the RfD template be marked with the trash icon and be able to be sorted as nominated for deletion with the now working filtering options. Best, Barkeep49 (talk) 04:42, 16 October 2018 (UTC)
Meh. I don't really see any benefit to that, but I thought you meant filtering of out the feed rather than the filter list. Not sure how much that goes with the WMF's goal to make Page Curation wiki-agnostic, which is a reason to oppose it, but I guess that wouldn't really harm anything. I guess you can make this "oppose as a waste of limited human resources when there are probably more pressing concerns" since the number of redirects at RfD is normally substantially less than 400. TonyBallioni (talk) 04:48, 16 October 2018 (UTC)
Also asked at the BRFA, this had some sample runs on betawiki, but I'm not seeing an interface to use it there either? — xaosfluxTalk19:35, 15 October 2018 (UTC)
Barkeep49 and Natureium -- yes, this is the trial period for the bot written by the Growth Team and ערן that will serve copyvio detection results from CopyPatrol into the New Pages Feed as part of this project. I'm sorry that this has been momentarily confusing -- I had not yet been through the release process for a bot, and I didn't know this trial period would take place. But yes, you will be able to test out copyvio detection at https://en.wikipedia.org/wiki/Special:NewPagesFeed?copyvio=1 within a couple days. We're expecting it to be the same functionality that the community already tested in Test Wiki. Once the bot starts populating the feed, I'll let reviewers know that there is something to play with. -- MMiller (WMF) (talk) 22:43, 15 October 2018 (UTC)
I checked the first one on the log - John Bertolino (photographer) - 7.4% chance, and all were the name of a school. I’m wondering if there should be a higher minimum percentage before it’s logged as a potential copyvio. Atsme✍🏻📧12:29, 16 October 2018 (UTC)Adding... checked the 2nd Dua Allahumma kun li-waliyyik - it came up 37.5% vio unlikely - based on quoted passage. I always check for copyvios anyway, and the main advantage I see is having the tool in the drop down menu for convenience. Is there more that I’m not seeing? 12:52, 16 October 2018 (UTC)
Two of the three I checked were COPYVIOS and have been deleted as such, Herbert E. Meyer which copied a speaker's bio for him and Baniaganti S. N. Academy School & College which was also a G11. The third was Dua Allahumma kun li-waliyyik which correctly triggered the alert but had a long block of text (a prayer) which is going to be OK to use. I would say those three results were very promising. Best, Barkeep49 (talk) 13:30, 16 October 2018 (UTC)
@Atsme: If you redo your lookup of John Bertolino (photographer) with the "Use Turnitin" option turned on, you'll see that Turnitin actually caught much more duplicated content than Google did. (You can see the copied content here and here.) Most of that content is quotation and bibliography (and thus OK), but it seems to have functioned correctly. There is a minimum percentage before an edit is flagged, but I don't remember what the current threshold is set to. If it seems too sensitive consistently, let us know. Ryan Kaldari (WMF) (talk) 16:32, 16 October 2018 (UTC)
Hi all -- I just wanted to post here that copyvio detection is ready for reviewers to test in the New Pages Feed. Thanks to those of you who have already been trying it out after Xaosflux's official announcement that the bot was deployed for its trial period. Here's how you can test the new features:
Go to this URL, instead of the usual New Pages Feed URL.
Open the "Set filters" menu and select "Copyvio" under "Potential issues".
This will filter to the pages that have been flagged by CopyPatrol (via the Turnitin service) as having revisions with potential violations.
You can click the "Copyvio" link in each entry to inspect the potential violating text in the CopyPatrol interface.
Sometimes, the reviewers working in CopyPatrol will already have deleted the violating text, deleted the page, or marked the page as "No action needed".
This testing period will continue into next week, at which point we'll decide whether we're ready to make the feature available at the usual URL. If you have feedback, reactions, or questions, please post here or on the project's talk page. And to read more about the specifics of this implementation and the rules behind how it works, check out this project update. -- MMiller (WMF) (talk) 23:15, 17 October 2018 (UTC)
redirecting talkpage(s) here
There is clear consensus to redirect/merge the pages discussed below, except for Wikipedia talk:Page Curation/Suggested improvements. As it is mostly about the technical side of a toolset, and not exactly about the immediate process of page reviewing; it should remain attached with its parent page for better convenience. I will perform the merger over the next 24-48 hours. Regards, —usernamekiran(talk)00:26, 20 October 2018 (UTC)
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Hello. In February 2018, I suggested to redirect/soft redirect Wikipedia talk:Page Curation to here. There is no point in having discussions scattered over multiple vrnues which arent watched by many. In the original discussion, there were no opposes, and two supports. I think we should discuss it again here, including any other talkpages that might require to be redirected here. —usernamekiran(talk)08:35, 13 October 2018 (UTC)
The "suggestions" section links to Wikipedia:Page Curation/Suggested improvements. Talkpage of that has been edited 20 times in total, including a "wrong venue" discussion, and closing of it.
"Coordination" section is for co-ordination of the project, mostly used for preparing the newsletters. Other co-ordination related stuff usually takes place here. The talkpage of co-ordination isnt a redirect, but mostly inactive.
"Reviewers" section Standard wikipedia user rights page. The talkpage of it, is this page. —usernamekiran(talk)13:30, 14 October 2018 (UTC)
Proposal
I hereby formally propose to redirect following three pages to "Wikipedia talk:New pages patrol/Reviewers", and to move their archives to the archives of this page.
I've suggested this at least 3 times, so I would be in favor, but you may want to find and read the past discussions on this. Natureium (talk) 13:06, 13 October 2018 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Update
I finished the redirecting/mergers of the talkpages here. The archives of the redirected talkpages have been moved into the subpages of Wikipedia talk:New pages patrol/Reviewers/Archives/Page Curation/, and Wikipedia talk:New pages patrol/Reviewers/Archives/Coordination/ accordingly. I have also added an archive box (below the current one) of the moved talkpages. Regards, —usernamekiran(talk)23:25, 22 October 2018 (UTC)
As the Page curation tools currently won't load up, I used Twinkle to patrol Rachel Green (biologist). This adds it to my patrol log, but not my page curation log. It is marked as 'reviewed' at Special:NewPagesFeed, but doesn't show under 'all public logs' for the page (if patrolled via the page curation tools, this does show up in 'all public logs for the page). Why does 'all public logs' contain the log entry from the page curation log, but not the one from the patrol log? Is this a bug? — Insertcleverphrasehere(or here)13:57, 16 October 2018 (UTC)
Charles Dickinson (writer) is a year old redirect that got vandalised and reverted in such a way that it was tagged as a New Redirect and showed up here. Just wondering if this is a bug/ can this be prevented... We obviously don't need extra work. MB01:43, 23 October 2018 (UTC)
@MB: This has been identified as an issue for quite some time (2015!) and it is in the task list for the community watchlist above as a bug in need of being fixed.
T92621: Implement addition of un-redirected pages to Special:NewPages and Special:NewPagesFeed (articles converted to redirects are sent to the feed, they should be sent back out again automatically if that edit is reverted). — Insertcleverphrasehere(or here)06:21, 23 October 2018 (UTC)
@Atsme: Looking at the page history it looks like the page was moved in April of 2016.
If you are looking for the creator of the page that was moved, you'll need to look at the history of The Novo by Microsoft, which indicates it was created as a redirect on January 13, 2009. The first substantive content was added on January 11, 2015 by Smilerslove.
If you're wondering who and when the current article at Club Nokia was created, the first non-redirect content after the page move was added by KaukoHaapavesi on October 22, 2018.
It did answer my question, thank you. Unfortunately, through no fault of your own, it didn't clear-up my confusion. I'm just not grasping why 2 totally unrelated articles were connected at all much less redirected, or why the current header of Club Nokia states: For the venue formerly known as Club Nokia, see The Novo by Microsoft. Club Nokia was a loyalty program, originally an internet site created by Nokia, not a concert hall or similar venue. Going back even further, why was this edit made in 2015 or this edit in 2016? Was it vandalism or am I totally missing something? Atsme✍🏻📧01:15, 24 October 2018 (UTC)
That's because Club Nokia was The Novo. The website is an entirely different entity and the page was created over a redirect. Nothing untoward here. Polyamorph (talk) 06:30, 24 October 2018 (UTC)
Perhaps you can explain by whom, when and how the current Club Nokia was created because I am unable to locate the article’s actual edit history. Who authored the material? All I’ve seen to date points back to the move and redirects to The Novo and the name changed Club Nokia which are completely different topics. See my discussion above with ONUnicorn. KaukoHaapavesi is the editor who most recently removed the redirects, and it appears that, according to their TP, they happen to be a Nokia enthusiast (and have already denied any COI), so it’s quite possible they know how/when the current article was created. I can usually find an article’s edit history with a bit a maneuvering through redirects & moves, but for some reason, I’ve been unable to find the edit history for this one - it magically appeared as the result of a redirect. The same applies to the TP history & activity. It is important to know so we can find out if the article was a prior AfD, an AfC submission that was approved and by whom, or was simply created by an autopatrolled user, etc. Perhaps admin Fish and karate can help clear-up the confusion, and is able to trace it back to the actual origin of the current edits that created this Nokia promo. It just appears to me that there should be more to it than an article suddenly appearing as a completed article from a reverted redirect that pointed to an entirely different topic, be it a dab notice or not. Atsme✍🏻📧15:47, 24 October 2018 (UTC)
A user with only extended confirmed user rights created an article in main space over a move/redirect of another topic with the same name and you’re ok with that? The user does not have autopatrolled rights, did not use AfC and by doing so, created this mess. Something needs to be fixed. Winged Blades of Godric, Insertcleverphrasehere, any suggestions? What happened is not the customary procedure for creating stubs/articles, and I am a bit concerned by Polyamorph’s nonchalant reception as a reviewer, especially considering the underlying circumstances and the content of the stub itself, not to mention notability issues. The only reason it ended up in the NPP queue is because it was a redirect. I also want to be sure we are not changing anything in the way redirects show-up in the queue, especially in light of this particular situation. Atsme✍🏻📧16:47, 24 October 2018 (UTC)
@Atsme: That's exactly what happened. There was a redirect. KaukoHaapavesi pushed edit, took out the redirect, wrote the content that they wrote, and pushed save. That happens all the time. You seem to think it's unusual - it's not. That's why when a redirect becomes an article it shows up in the new pages feed again. There's nothing wrong with turning a redirect into an article, but the article should be reviewed by NPP. Any user can do what KaukoHaapavesi; even an IP. I'm not sure why you think KaukoHaapavesi created a mess; I don't see a mess. ~ ONUnicorn(Talk|Contribs)problem solving16:53, 24 October 2018 (UTC)
This does indeed happen all the time, and they are currently sorted into the feed by the date of the redirect creation, rather than the date of article creation, which is one of the issues in the List below that needs to be fixed. If this were fixed, a notice that the article was converted from redirect would also be useful. I'll add that to the request. — Insertcleverphrasehere(or here)17:20, 24 October 2018 (UTC)
Quoting AtsmeWhat happened is not the customary procedure for creating stubs/articles, and I am a bit concerned by Polyamorph’s nonchalant reception as a reviewer, especially considering the underlying circumstances and the content of the stub itself, not to mention notability issues. - yes it absolutely is customary procedure, and I've done nothing wrong. You on the other-hand shouldn't have un-reviewed the page, but not big deal. Please assume good faith on behalf of both the creator and fellow NPPs.Polyamorph (talk) 17:30, 24 October 2018 (UTC)
Thanks, ICPH. I’ve seen redirects removed and content added to the article that was redirected, but the topic was the same. I have not seen this happen when it involved 2 different topics of the same name. Common sense tells us to create/move the new article/stub with an identifier that distinguishes it from the other, such as Club Nokia (mobile) or (programme) or the like. If it’s not too much trouble, I would be very appreciative if those who are familiar with the repeated occurrence of creating stubs/articles of the same name/different topic over redirects to unrelated/totally different topics would provide a few diffs so I can at least see how these situations have been handled in the past. Oh, and Polyamorph, while we’re here trying to get new tools and procedures implemented/ironed out, how did this mistake occur, and do you have any suggestions how it can be avoided in the future? Atsme✍🏻📧17:50, 24 October 2018 (UTC)
You did it properly when you moved it ONUnicorn so it is not the same as simply creating a new stub over the top of a redirect/move with the same exact name but a completely different topic (mobile promotion vs a concert venue). What you did left a proper edit history trail so there was no confusion. That is not what happened with Club Nokia and why I thought it best to bring it up here to hopefully keep it from happening again. Of course we initiate moves and redirects all the time, but we typically make distinctions when naming an article (which helps explain why we initiate some moves) - and we do it properly because when we don’t, it creates confusion. Thank you for going to the trouble of finding that diff - it is much appreciated. Atsme✍🏻📧20:50, 24 October 2018 (UTC)
I don't see how it is any different. There existed an article "Club Nokia" which was the name of a concert venue. This was moved when the name of the club changed to "The Novo by Microsoft", preserving the edit history at the new location and leaving a redirect. Then an editor created a stub at "Club Nokia" about a different entity. There is nothing wrong with this workflow and no editor is at fault. Polyamorph (talk) 21:08, 24 October 2018 (UTC)
I'll be very suprised if anyone will satisfy your request for diffs. Blow me down, kudos to ONUnicorn :) I'm not sure why my talk page history is of any interest to you, but that was a simple mistake the NPP involved re-reviewed the page immediately after, nothing to do with you but thanks for your interest in me. Polyamorph (talk) 18:13, 24 October 2018 (UTC)
(edit conflict)@Atsme. PRehse accidentally clicked unreview I would guess, then quickly re-reviewed it. It would have showed up in his watchlist because he was the original reviewer in April.[3] The redirect in question would have gone back into the feed when it was merged recently and converted to redirect(this is normal). Then the merge was reverted, and then the revert was reverted. This probably didn't have any effect on the situation, but would have brought the redirect to PRehse's attention. Polyamorph marked it as reviewed (probably found it in the feed) and then PRehse must have accidentally clicked the button while he was there inspecting the situation (probably just muscle memory), then quickly reverted his own action when he realised that he had unreviewed, rather than reviewed it. There is an item in the wishlist to be able to optionally disable this automated message, and it is also affected by the one about messages not being hard coded. The issue is that the 'unreviewed' message reads as if intentional, and the reviewer who sent it really has no idea that they have sent it at all unless they check their contribs as it is entirely automated and you aren't prompted. I'll note on the Phab task that a prompt before sending the 'unreview' message is probably a better solution. — Insertcleverphrasehere(or here)18:17, 24 October 2018 (UTC)
Makes sense, thanks ICPH. PRehse is a highly productive editor (NPP gnome of sorts), works quietly in the background getting things done. We cross paths from time to time - it’s a good feeling knowing he’s out there...or at least that’s my take on it. Atsme✍🏻📧20:50, 24 October 2018 (UTC)
Have a look at Vasant_Gajera. It was created by a blocked user, Idigitalsindia and deleted several times already. See [4]. How did it sneak past getting reviewed? Vexations (talk) 13:33, 28 October 2018 (UTC)
I suspect it's a reversion of a move that caused it. cymru.lass might remember. I'm trying to find out more about exactly what happened. Autopatrolled is extremely dangerous IMO and such unintentional patrols should be considered a bug. Vexations (talk) 13:54, 28 October 2018 (UTC)
@Vexations:--Located the bug (??) precisely.Whenever anyone with the auto-patrolled flag moves something from draftspace/otherspace to article-space, they get auto reviewed.I am an autopatrolled editor and moved Draft:Central south slavic to main-space, which immediately got auto-reviewed. The reviewing in this case was thus unintentionally executed by NRP's crossspace move:-)∯WBGconverse14:11, 28 October 2018 (UTC)
I kind of see that the logic that an autoreviewer (autopatrolled user) accepting an AfC submission is a sort of patrol. Maybe we can live with that. However, it seems that ANY move by an autoreviewer results in the article getting autopatrolled. That is undesirable, especially when the move is performed by an autoreviewer who is not also a patroller, because they cannot undo the removal from the unreviewed new pages feed. Vexations (talk) 14:41, 28 October 2018 (UTC)
I am pretty confident that it's only cross namespace moves which are autopatrolled. So Afc acceptances are the most likely scenario. Best, Barkeep49 (talk) 14:58, 28 October 2018 (UTC)
OK, fair. But I see that's an external tool. If you want see what actually happen to a page, I think the easiest way is myriad of logs right here on wiki, before going external. One click to the page's history could've saved you feeding all these conditions to quarry which at the end brought us here. –Ammarpad (talk) 18:05, 28 October 2018 (UTC)
Yep, definitely didn't mean to mark it patrolled. Probably did happen when moving it from Help: or Wikipedia: space, or wherever the hell the spammer put it lol. I was trying to clean it up and see if maybe some of the controversy could be brought into the article before deciding whether or not it should be deleted. cymru.lass (talk • contribs) 18:27, 28 October 2018 (UTC)
Do we need solutions, or do we need a toolkit so we can solve problems ourselves
When I was reading over the items in the Wishlist Proposal, it reminded me of an essentially defunct WMF project. Back when the WMF first conceived Flow, they also envisioned something called the Workflow project. Workflow is not specific to patrol, but it could be significant here. The original Workflow concept was badly tied to Flow, but I think the underlying idea has enormous potential if divorced from Flow. As a relatively simple example, consider the steps to nominate a page for AFD:
Add a template at the top of the article.
Create a new page for the discussion. This page will contain some routine AFD wikitext at the top, and add a user-supplied deletion rationale below.
Add that new page to the daily list of AFDs.
An editor could go to the Workflow builder and select options matching those three steps. For step 1 they specify what template to add, for step 2 they supply the standard AFD header text, in step 3 they say where the daily list is. I somewhat simplified things, but it comes pretty close to a point and pick process to define or modify a simple workflow. An admin could then approve that workflow to appear on a Twinkle-type list. (More complicated workflows might need a tech-comfortable editor to be sure that all possibilities are handled correctly.) Depending how powerful the workflow builder is, it could replace Twinkle, many of our work scripts, much of the patrol system, and we could build workflows for many other tasks.
The reason I mention the Workflow project is because I estimate about HALF of these Wishlist items could be fixed by us, on-wiki, if patrol were built around a Workflow type system. We could add or redefine (at least some of) the buttons on the Page Curation Toolbar ourselves, by adding or modifying the workflow for that button.
I also think it's worth noting one of the major motivations the WMF had for Workflows. The WMF can't realistically build and adapt tools like Page Curation to countless other wikis, with their varying needs and varying community processes. (For example Commons adds new deletion-nominations to the bottom of the page of any existing deletion discussion, and small wikis may simply toss assorted tasks onto one Village-Pump type page.) The Workflow system would allow each wiki to copy, modify, or build workflows themselves, as needed.
I don't know if the WMF is willing to revive the Workflow concept. I don't know if they'd be willing to reconceive it independent of Flow. I don't know how much a Workflow builder would be capable of. It would be a major project if they did build it. But I find it painful that the WMF sinks massive resources into questionable projects while allocating a token few percent of time to community requested tasks.
I'm not specifically proposing anything here. I'm tossing this concept on the table to see how much interest it generates. Alsee (talk) 12:31, 22 October 2018 (UTC)
@Alsee: As for 'Workflows', I don't know much about it, but sounds promising and could be nice as an independent submission to the Community Wishlist. With regards to the Page Curation Toolbar, It actually IS configurable for different language wikis, and at least some of the items above actually are fixable on-wiki by modifying the various .js pages that the toolbar pulls some of its tools from. For example, tags are contained within the .js pages, so we can add/remove tags ourselves, or at least, a tech-admin can. Winged Blades of Godric is actually looking into a number of the tasks that can be completely or partially solved ourselves (see his edits to my talk page, this page, and the suggestions page for more info). Some of the stuff is hardcoded though, like the author messages, which is something that ongoing WMF support would be useful for. — Insertcleverphrasehere(or here)12:50, 22 October 2018 (UTC)
An IFTTT or Zapier for Wikipedia would be a game changer. Even Flow wasn't a bad idea, but its implementation, messing up on-going discussions, was a heavy-handed clumsy disaster. Divorce Workflow from Flow (and its tainted history) and there's the possibility of some interesting & useful developments. Cabayi (talk) 10:47, 25 October 2018 (UTC)
I'd support reactivating the Workflows project -- I don't immediately see that it would have Flow as a pre-requisite. The question is whether this is such a substantial job to produce that we need to work for it independently of the wishlist. DGG ( talk ) 15:20, 29 October 2018 (UTC)
A script that may be of use to y'all
Well, I recently decided to write a very basic script which allows someone to post a note on the t/p of the creator of an article. What I would like to know if there is a need for such a tool ?(I have created a rough proof of concept here) — fr+18:25, 29 October 2018 (UTC)