Talk:Community Wishlist Survey 2019/Admins and patrollers/Page Curation and New Pages Feed improvements

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search


@Insertcleverphrasehere: Note that proposals should be discrete tasks and not "needs improvements" or "has a lot of bugs" proposals. This proposal seems to be a catch-all which actually consists of many separate proposals. --AKlapper (WMF) (talk) 13:00, 30 October 2018 (UTC)

This proposal includes which specific things need fixing, and while it does make sense to limit the size of wishlist proposals, limiting them by the things requested even if they are smaller improvements all relating to one thing (page curation tools) that in total may be less effort than the bigger requests doesn't seem to make sense to me. There is Don't just say that "(x feature) is out of date", "needs to be improved" or "has a lot of bugs". That's not enough information to figure out what needs to be done. but this does give well enough information to figure out what needs to be done. I also note that I believe multiple people from the WMF have said that the community wishlist is the place to request improvements to the curation tools so there will be a lot of very annoyed people if this proposal isn't considered appropriate here. Galobtter (talk) 13:53, 30 October 2018 (UTC)
AKlapper (WMF), what Galobtter said. We have been repeatedly advised to request Page Curation improvements via the wishlist, and splitting these tasks into many smaller more discrete tasks doesn't make much sense to me as it would needlessly complicate the reason that we are here: Page Curation and New Pages Feed improvements. We have helpfully defined what those improvements should be in the Phabricator tasks, and have narrowed the list only to those deemed high priority to the New Page Review team. — Insertcleverphrasehere (or here) 16:23, 30 October 2018 (UTC)
Hi. The proposed improvements are as precise as they can get. The problem section isnt precise, but it does give the general background/context. The "Phabricator tickets" section pinpoints to the exact problems/proposals. These problems/proposals aren't new. A lot of them have been discussed with WMF personnels prior to this. Also, before submitting the proposal here, it has been vetted by NPP/R community, discussed thoroughly. We are not overdoing this, we are requesting fixes to the only stuff that is sourly needed :)
Regards, —usernamekiran(talk) 18:35, 30 October 2018 (UTC)

Hi Insertcleverphrasehere and all: I'm glad to see your proposal, and I appreciate all the work that you put into figuring out the highest priority tickets. As Andre said, this is a longer list of tickets than we usually want in Wishlist Survey proposals, but I've seen the master list of improvements and I know how much you cut to get here. :)

Assuming this is one of the top 10 wishes, the team will investigate and make estimates in January, and then talk to the NPP community and figure out the scope. With this many tickets in the list, there will probably be a few that are too technically difficult, compared to the amount of value you'd get.

For example, this ticket offers obvious value for patrollers, and looks technically straightforward -- "Add "previously deleted" as a possible issue (flagged in red) in the New Pages Feed/Page Curation Tool" (T189929). This would definitely be done.

However, at first glance, this ticket looks more difficult, possibly too hard for the value it's going to offer -- "Adding a "Potential COI" alert to the feed" (T207757). It's asking a machine to make a match between the username and the title of the article, which is an easy task for humans and possibly a much harder task for machines. If it turns out this would take months to train the machine, then it would probably be more worthwhile to work on something else instead. (Note: According to replies below, this task might be easier than it looked to me at first. But the point remains -- each task will need to be investigated and discussed, and there may be some things listed in the proposal that are too complicated technically to make it worthwhile.)

That's not a conversation that we need to have right now -- assuming this gets into the top 10, then the team will start looking at it in January, and we'll be able to show you all a breakdown of the relative difficulty of this set of tickets. At that point, we can have a conversation with you, and make decisions together about where we should focus. I think Marshall did something similar a few months ago, to figure out the scope of the changes he could make. That's how we approach any big project. So: feel free to canvass and cajole, and get people to vote for this proposal. When it's in the top 10, the team will definitely put significant work into the project, with the exact scope to be determined later. :) Let me know if you have any questions. -- DannyH (WMF) (talk) 21:23, 30 October 2018 (UTC)

Thank you Danny for your reply and explanation of this process should this make the top 10. Barkeep49 (talk) 06:35, 31 October 2018 (UTC)
@DannyH (WMF): Thanks for your reply. And yes we can have that discussion when and if this makes it in the top ten. However, I must note that T207757 ("Adding a "Potential COI" alert to the feed") was chosen because it should actually be fairly easy, as it would simply be flagging any new article where a user triggered Abuse Filter 148 or 149, which already make matches between usernames and titles of articles/content of edits: (Please see and — Insertcleverphrasehere (or here) 07:07, 31 October 2018 (UTC)
@Insertcleverphrasehere: Oh, that's good to know. :) That's the kind of conversation that the team will have with you all once the survey's over, investigating the different tasks and figuring out what's involved with each of them. DannyH (WMF) (talk) 17:26, 31 October 2018 (UTC)
I think it would be a good idea to strike the comment that a task may be too complicated before we've actually had a discussion about it, lest I affects the number of supporting votes it may not receive as a result of that pre-emptive assessment. I wouldn't vote for something if that a representative of the WMF says is too difficult. But I'd be upset if I later found out that it actually wasn't.--Vexations (talk) 01:52, 1 November 2018 (UTC)
Okay, I added a note to my comment above. But my point is that there may be individual tickets listed on this proposal that will turn out to take more time than we have to spend. I agree that the people who will be voting for this proposal should know that it's possible that the team won't finish all 17 of the listed tickets. I want to manage expectations at this point, so folks know what they're voting for. Getting this proposal into the top 10 will guarantee significant work on the Page Curation tools; the exact specifications of what's going to be done will be figured out in January, in collaboration with the NPP community. Does that make sense? -- DannyH (WMF) (talk) 15:48, 1 November 2018 (UTC)
Something to ask here, since other language Wikipedia is mentioned here. Is this Page Curation thing embedded in the Mediawiki software where a local community just need to decide that yes, they wanted it, it can be turned on (subject to localisation with some Phab tickets of course) or should it need a major site deployment? Thanks a lot.--Cohaf (talk) 18:22, 4 November 2018 (UTC)
@Cohaf: mw:Extension:PageTriage is currently only used on the English Wikipedia. I personally would love to see it deployed to other projects. There's still some work to do to make it wiki agnostic (phab:T50552), but if we can devote some time to it, I don't think we're super far off from making it happen. MusikAnimal (WMF) (talk) 17:11, 5 November 2018 (UTC)
@MusikAnimal (WMF):thanks for the reply. Hope to see it soon. One of the thing that frustrate reviewers is that there isn't a marked as unreviewed option (i.e. if you mark a page as reviewed and you will like someone else to look at it, it cannot be returned to the new page feed (as yellow background)). I hope if something along the lines can be implemented first, it may be better. The rest PROD (not relevant to my wikipedia), CSD/AFDs Twinkle is adequate. The option to go to the next page is also attractive (especially for redirects) as if we clicked and went back, the process may sometime be a little of an hassle. That's all I hope for. Thanks much.--Cohaf (talk) 17:17, 5 November 2018 (UTC)

How is this sort of CANVASSing acceptable or fair to the other proposals? Will the other proposals be afforded their own Signpost article or is this only available to former Editors-in-Chief of the publication? 00:48, 9 November 2018 (UTC)

We explicitly encourage people to canvass for their proposal, which also helps all of the other proposals. People can cast support votes for all of the proposals that they like, so somebody who's directed to the survey will also have the opportunity to read and vote on other proposals too. Please feel free to spread the word about your proposal as well. -- DannyH (WMF) (talk) 01:14, 9 November 2018 (UTC)
It's a terrible system which biases proposers who know the system better and/or are more popular. My only comment for the IP is that we are being forced into it, and due to the system have no choice but to canvass. The idea that the top ten proposals should be chosen via vote count is a bit ridiculous as it is gameable (and gaming the system even encouraged). There are many unpopular but easy proposals that would be of high benefit that are overlooked each year, which is sad. — Insertcleverphrasehere (or here) 13:03, 9 November 2018 (UTC)
  • DannyH (WMF) can we please move this lengthy 'discussion' section to the talk page before voting begins? It has essentially resulted in "lets vote on it then discuss it later", and the last thing I want is a massive discussion section below the proposal to discourage people from voting. — Insertcleverphrasehere (or here) 07:41, 13 November 2018 (UTC)
Insertcleverphrasehere: Sure, I've moved it to the talk page. -- DannyH (WMF) (talk) 23:19, 13 November 2018 (UTC)
  • So far so good, we're at over 110 supports which is nearly double what the next most supported proposals have. ( disclaimer, I didn't get to the last 7 categories). enL3X1 ¡‹delayed reaction›¡ 03:08, 19 November 2018 (UTC)

Moved section from my support[edit]

I knew this will be passed nevertheless, and ICPH and the team (Inc kudpung as I read the Phab tickets) had done a hell load of good work which should be recognized so I don't see the point of opposing. However, I would like to register my strong dislike of Kudpung comments above. Yes, English Wikipedia is important but not until the fact that all other projects seems to be in a totally dependent state. I simply doesn't buy it. --Cohaf (talk) 19:24, 18 November 2018 (UTC)

    • Cohaf, Keen to chat with you about this more and have you involved in later discussions. Activating the New Page feed on other language wikis is I think going to be a high value-for-effort task, and I'm keen to push for it if there is interest from other language wikis. There is a phab task for this already (phab:T50552). As I noted above, I think that Kudpung was referring to the severe amount of promotion and spam that is disproportionately burdened with. Other wikis are just as important, as can't provide a chinese language article for a native chinese speaker who doesn't speak English, as one example. Those wiki should also have the tools they need and if the New Page feed and ores scoring as well as the Toolbar shell (which can be modified by on-wiki .js pages) can help you guys too, then it should be made available. I've been told by Danny that they will approach us afterwards to discuss what can practically be done (see talk page of this page), at that point I'll be keen to push for (phab:T50552) to be addressed at the same time. From the most recent comment on that phab task, it seems the growth team wants this to happen eventually as well. — Insertcleverphrasehere (or here) 00:25, 21 November 2018 (UTC)
      • @Insertcleverphrasehere:Pardon me for moving here to make the front page less cluttered. A few issues, first, I definitely hope for more conversations with you to bring new page feed to zhwiki. Second, ORES is quite useless for issues regarding Chinese language articles, I ran earwig on numerous articles which I found copyvios manually and all turned out 0%. So that can be omitted if it makes tethering easier. For growth team, I don't think it's relevant to my wiki as we are the 8th largest but I do have patroller flags at zhwikiversity as well as here at meta, so these may be relevant. Thanks for your efforts and I hope to speak to you soon. One sidenote, I don't think you have to speak for Kudpung, if they can make such statements they should be the one answering it, you basically own your statements, isn't it. No need to help someone else who make such irresponsible statements to shoulder the flak. Yes, also I do get English language spam articles also at zhwiki, without an A7 and a tighter G11, i often had to raise an AFD. Spam / undisclosed PE are indeed ubiquitous. Best Regards,--Cohaf (talk) 00:43, 21 November 2018 (UTC)
        • @Cohaf and Insertcleverphrasehere: Now I'm interesting that, if you both really think that this is good for a lot of zhwiki users, why not just post your wishlists at zh:Wikipedia:互助客栈/消息, zh:Wikipedia:互助客栈/技术, or their guestbook? Just discuss on this talk page, to the best of my experiences, can be a reason that why you can't hear the actual answer of what local community members are considering. And afaik, there's still a little lack of zh native users (I mean, are they matching 25% of joined-voting users?) on survey. --Liuxinyu970226 (talk) 12:37, 6 December 2018 (UTC)
          • @Liuxinyu970226:.I had in fact spoken to several patrollers as well as admins to discuss about this toolbar informally. The key thing now is to make it wiki agnostic. No use to post on the village pump and get a consensus to use this toolbar where in the end, it isn't deployable. AFAIK, I am the only Chinese Wikipedia New Page Patroller who voted in support of this proposal. I didn't canvass for this as it is almost certain this will pass.--Cohaf (talk) 12:42, 6 December 2018 (UTC)
            • @Cohaf: Ehh, I also voted in favor for this, but only because the "maintainers" group is now changed (from Collaboration team which had some bad histories e.g. disallow uninstalling Flow on Commons, to Growth team which I more and more trust). --Liuxinyu970226 (talk) 12:48, 6 December 2018 (UTC)
              • @Liuxinyu970226:I thought this is handled by Community Tech not Growth? In addition, you are not a Chinese Wikipedia Patroller based on SUL but a Chinese Wikivoyage Patroller. So yes, I am still the only.--Cohaf (talk) 12:54, 6 December 2018 (UTC)
          • (Edit conflict.) I agree with Cohaf's assessment entirely. Liuxinyu970226 To be perfectly honest, I don't really know of the workings of Zh wiki or of the existence of those pages before you mentioned them. I'm happy for members of other wikis to come forward and express their desire to have the tools on their wiki, but honestly I'm a bit busy with at the moment to be trying to connect in with patrol teams on all the big wikis. I'll push for the tools to be fixed so that they can be used on other wikis, but enabling them will require some work from the native users of each of those wikis, and talking to the other wikis now is a bit premature when the tools are not in a state to be deployed anyway. — Insertcleverphrasehere (or here) 12:45, 6 December 2018 (UTC)
            • @Insertcleverphrasehere:.Just to add, those pages are wrong. The correct page to discuss this isn't the 3 that are mentioned. I will be willing to start a thread in the correct place once it's wiki agnostic. Thanks and appreciate for your help and do keep in touch.--Cohaf (talk) 12:50, 6 December 2018 (UTC)
            • @Insertcleverphrasehere: Be fair to say, I indeed was opposing the deploymenting of this on zhwiki, mainly because of the Language variant compatibility (and you should even know that the members of past group Collaboration team don't fixed issues on any time, by any trial patches, but just said a lot of "we will fix it in the future" and nothing else). --Liuxinyu970226 (talk) 12:53, 6 December 2018 (UTC)
              • This discussion shouldn't be here. It should be done locally. I am just wanting the mark this page unpatrolled as well as next page function which have nothing to do with language variants. FWIW, those doesn't require any words. Tagging could be done via TW.--Cohaf (talk) 12:57, 6 December 2018 (UTC)


@Insertcleverphrasehere: Please explain what exactly you mean by "wiki-agnostic". Thanks in advance. 4nn1l2 (talk) 21:13, 24 November 2018 (UTC)

4nn1l2 I think I can answer that. English Wikipedia does some things differently than other Wikipedias. That makes it difficult to use that same tools for similar, but different tasks. For example, nominating an article for deletion is implemented as a workflow in the page curation tools, but not all Wikipedias have AfD, so the curation tools can't work for them in their current form. Ideally, the tools would be written in such a way that all Wikipedias can use them, regardless of how they choose to implement their policies. Such a tool is wiki-agnostic, because it doesn't "know" which wiki it runs on. Making the tool wiki-agnostic would also benefit patrollers on English Wikipedia, because it would give users more control over the configuration, which is separate, in stead of hard-coded in the software. Vexations (talk) 21:44, 24 November 2018 (UTC)
What Vex said. Also, it is defined in much more detail in the Phab task description. — Insertcleverphrasehere (or here) 23:11, 24 November 2018 (UTC)
BTW I changed "wiki-agnostic" to "" for clarity throughout. I had originally used 'wiki-agnostic' due to the title of the phb task but it isn't clear enough apparently. — Insertcleverphrasehere (or here) 11:25, 25 November 2018 (UTC)