Wikipedia talk:Bots/Requests for approval

Change of Operation - User:NoSandboxesHere

So I am looking at switching User:NoSandboxesHere from AWB to a always-running PHP script. It does exactly the same thing as the AWB process (just does a regex search on anything in the Draft namespace with {{User sandbox}} or {{Userspace draft}}), however doesn't rely on Category:Non-userspace pages using User sandbox, it uses getTransclusions from RMCD bot's version of botclasses, filtering by the namespace.

Do I need a new/replacement BRFA for this or can I go ahead and just switch it out. Example diff in my userspace is just here ({{Fuuuuuu}} is just an invalid template I added to the script to make sure the API calls were working) - Rich^T|C|E-Mail 21:02, 10 July 2020 (UTC)

Go for it. Primefac (talk) 21:47, 10 July 2020 (UTC)

Semi-protected edit request on 31 July 2020

We are creating a bot to scrape information of Malls in United States from the Wiki pages. Mall Data Scraping Bot (talk) 21:44, 31 July 2020 (UTC)

 Not done: User:Mall Data Scraping Bot: If you don't want to make any edits, see Wikipedia:Database download. If you want to have a bot make edits, a request must be made using your own, non-bot, account. Mdaniels5757 (talk) 21:48, 31 July 2020 (UTC)

Approval for my pictures in an article

Hello, I have inserted two pictures, Dainikjagran.jpg and Hindustanarticle.jpg, in an article. Both are original and aren't copyrighted material. That's an extract from a newspaper, which is legal to use. But now both are nominated for speedy deletion. Please approve both the images back. Don't delete them. Enigmaticpravin (talk) 09:50, 10 August 2020 (UTC)

This is the talk page for discussing improvements to the Bots/Requests for approval page, you may want to follow up with whomever you were discussing that with on their own talk page. — xaosflux ^Talk 11:43, 10 August 2020 (UTC)

QEDKbot

QEDKbot (talk · contribs · deleted contribs · logs · filter log · block user · block log) was approved at Wikipedia:Bots/Requests for approval/QEDKbot on 20 May 2020, by @Primefac. The approval seems surprising, since there were several well-founded objections, and no community consensus for its functions. The BRFA was advertised at WP:Administrators' noticeboard/Archive317#Wikipedia:Bots/Requests_for_approval/QEDKbot, but not at any page which concentrates editors who work on categories, such as WT:CATP.

The bot's function is to populate Category:Empty categories with no backlinks. That category was nominated for deletion at Wikipedia:Categories for discussion/Log/2020 August 9#Category:Empty_categories_with_no_backlinks, and the discussion is still open. I spotted the discussion yesterday, and while the bot owner QEDK has been prompt and courteous in their replies, I haven't been able to glean from QEDK any clear explanation of what purpose is served by filling Category:Empty categories with no backlinks with category redirects. Those categories are supposed to be empty and to have no backlinks, so this amounts to categorising them as "all OK" ... which seems pointless.

In July, this was raised at User talk:QEDK#Category_redirects (permalink)by @Mclay1. There were two replies from QEDK, who could offer no explanation for the utility of this category beyond It's for housekeeping[1], which tells us nothing. Exactly what "housekeeping" is facilitated by this category?

Anyway with the CFD open I was very surprised to see that the bot has still been running. The CFD was notified to QEDK at 18:27, 9 August 2020 [2] by UnitedStatesian ... yet 2 minutes later, at 18:20, QEDKbot made its first edit[3] for two hours ... and in total, it has so far made 207 edits since the CFD opened. (see contribs list)

When the category's future is being discussed, it seems perverse for a bot to continue to populate it. At best that's WP:RECKLESS; at worst it's a bit WP:FAIT.

So I have two requests:

1. Please can the bot's authorisation be revoked for now, without prejudice to any future new authorisation if there is community consensus that this bot serves some useful purpose?
2. Please can all deletions by QEDKbot be reviewed, and any category redirects restored. I see for example that on 27 Feb 2020, the bot deleted Category:Fianna Fail leadership elections, which I presume was a redirect to Category:Fianna Fáil leadership elections; if not, it should have been a redirect. --BrownHairedGirl (talk) • (contribs) 02:36, 21 August 2020 (UTC)

Thank you, @BrownHairedGirl:, I of course endorse both your requests, and would add a third:

3. That no reauthorization be given until the operator demonstrates that a page loggging all its CSD tagging is fully functional, to facilitate review of any deletions, of the sort requested in 2. above or otherwise, by non-administrators such as myself and User:BrownHairedGirl. UnitedStatesian (talk) 02:43, 21 August 2020 (UTC)

I agree with that third point, @UnitedStatesian. On top of the general problem of the bot being pointless, it lacks transparency. --BrownHairedGirl (talk) • (contribs) 02:51, 21 August 2020 (UTC)

After my brief conversation about the bot with QEDK, I still don't understand the point of tagging category redirects, for the same reasons BrownHairedGirl pointed out. To me, it seems like the creator of the bot misunderstood the function of category redirects. They do not serve the same purpose as regular redirects, and they should have no backlinks. The category redirect function should be removed regardless of whether the bot serves some other useful function. M.Clay1 (talk) 02:55, 21 August 2020 (UTC)

@BrownHairedGirl: A lot of misleading accusations here and there. So, let me address them in order:

• Re: no community consensus The bot was advertised at AN because it was an adminbot, it received quite a few comments on the BRFA w.r.t. its feasability - and a lot of the points were discussed and implemented.
• Re: why only category redirects? I have explained this multiple times that I need more time to actually make the bot work with other kinds of categories, and that category redirects were simply for demonstration. I have university and a part-time job, how easy is it to write programs in the time that's left?
• Re: no transparency How is this bot any different from other bots? ClueBot's all edits are found in its contribution, same with QEDKbot. MusikBot's protection actions are found in its protection logs. If you want to see what edits QEDKbot makes, see contributions, if you want to see deletions, see the Deletion logs. Even then, I still had written the code - I simply didn't get enough time to test it and it was buggy.
• Re: category redirects don't have backlinks Except they do. And the point of keeping them lying around is so that we can preserve the ones with some usage. In fact, a lot of them have a lot of backlinks (feel free to write a script and check this), the ones with no backlinks are basically pointless (in most cases, but in some cases, not).
• Re: bad deletions The bot only deletes categories in narrow circumstances, in the only example cited, the category was tagged by the owner and stated to be "created in error", so the point is moot.

That's all. --qedk (t 愛 c) 07:37, 21 August 2020 (UTC)

@QEDK:

1. feasibility ≠ desirability
   Yes, the bot does work. But that doesn't mean that the functionality is wanted or needed..
2. if categorising the redirects is simply for demonstration ... the you have demonstrated that the bot can do something with then. Demo over.
   Lack of time to create useful functionality is not grounds for running bot with no utility.
3. On transparency, UnitedStatesian has explained the problem several times
4. On the purpose of category redirects, you clearly have views on their utility ... but there is no sign that you have ever attempted to seek consensus for your view. You should not be running a bot in pursuit of a personal view for which you cannot demonstrate a clear consensus. Basically, this bot is preparing the ground for deletion of category redirects ... and there is no consensus for that.
   Also, note that your stated desire the delete category redirects contradicts your earlier statement about the redirects just being a demo.
   Your statement that the ones with no backlinks are basically pointless is alarming, because it shows a severe misunderstanding of how category redirects work and how they are used.
   • A category redirect which has no backlinks but is navigated by dozen of readers every day, is in your view basically pointless. That is perverse.
   • A category redirect which has no backlinks but is used by editors to categorise articles, is your view basically pointless. That is also perverse.
     Do you even know how category redirects help to categorise articles? If the editor is using WP:HOTCAT, then HOTCAT will automatically resolve the redirect before saving. And if the category is added by editing the wikicode, and HotCat isn't around to fix it, then User:RussBot will pick it up and fix it: see User:RussBot/category redirect log. Your intended removal of category redirects will break both HotCat and RussBot's functionality. There are thousands of category redirects which exist to resolve variations in language or typography, e.g, Category:Sinn Fein→Category:Sinn Féin or Category:German-Turkish relations→Category:Germany–Turkey relations ... but your only measure of their utility is backlinks which have not yet been fixed. That ignore readers and editors who use the redirects to navigate and to categorise pages.
     Similarly the ~14,500 redirects between the ENGVAR spelling variations of organisation/organization. The vast majority of them have no backlinks, but your seem oblivious to the consensus to mass-create them: see WT:CATP#Organi[SZ]ations_category_redirects (permalink).

You have clearly done a fine job on the programming. But I am annoyed to find that is really all about your desire to pave the way for mass deletion of category redirects, because you didn't state that goal upfront, let alone seek consensus for it. That is no basis for a bot, let alone one which is adding otherwise pointless categories to redirects. --BrownHairedGirl (talk) • (contribs) 08:32, 21 August 2020 (UTC)

What do you mean by mass deletion? The bot doesn't delete even one category redirect, it's up to another administrator. I don't think you understand, because I said they are G6-able, not that they should be G6-ed. And I very clearly stated, "most" of them are basically pointless as is demonstrated by the handful of pageviews that most of category redirects get, so your counterpoint is baseless because I am generalizing and you are cherry-picking. I have no intention on removal at all, in fact, I think you're really misconstruing my suggested actions as something I intend to do. And you go on and on about category redirects but don't really understand that what the bot does is fundamentally different from your claims. I didn't seek consensus for mass deletion of category redirects because the bot does not do that, is that not simple enough to get? I have no issue with category redirects being around but some of them serve no use and that usage can be tracked with backlinks to some extent, that's it. --qedk (t 愛 c) 08:47, 21 August 2020 (UTC)

Grrrr. No, I am not misconstruing your actions. Your bot is not pulling the trigger, but you are making the hitlist for a pattern of deletions which you desire but for which you have not sought consensus. Despite being repeatedly asked both here and at CFD, you have not identified any utility for this bot other than creating that hitlist.

And I understand perfectly well what the bot does: it adds category redirects to your hitlist.

However, you still clearly do not understand the purpose and uses of category redirects, because backlinks are the least significant of their four main purposes. It just happens to be a usage which the bot can measure, but a little bit of logic shows that it is a completely bonkers, naive measure: if there are dozens of links to the redirects, and one or two new ones are added every few weeks, then when somebody cleans them all up your bot will see no backlinks and add the redirect to your kill list. A week, when someone creates another link, the redirect will be gone. We don't remove article redirects just because they don't currently have a hardcoded incoming link, and there is no basis for removing category redirects on that basis. Desired deletions of article redirects are discussed at WP:RFD, and category redirects at WP:CFD .... but you have decided to bypass the consensus-building processes at CFD, and instead create a hitlist for misuse of WP:G6. If you want category redirects to be culled in this way, then the proper way to do it is to open an WP:RFC. Instead you made a bot request to implement your kill list without even notifying WP:CATP. --BrownHairedGirl (talk) • (contribs) 09:11, 21 August 2020 (UTC)

What "hitlist"? Can you demonstrate that this "hitlist" has had mass deletions? If you have no evidence for this why are you choosing this sensationalistic wording? I've explained the rationale multiple times (and it was done at the BRFA as well) so me not identifying any utility is your perception, not fact. You seem to quite interesting in translating a tracking category into a "kill" list but forget that administrators could choose to delete them anyway (and anyone could nominate them at CFD), the bot doesn't do anything other than "track" potentially useless categories. I have no interest in "culling" and I suggest you drop that narrative. No one is saying that they have to be removed, but that some of them can be - and that seems to be nuance you're missing. I've already stated that tagging beyond category redirects and logging are feasible, and will be done, I don't see what else I can do here. --qedk (t 愛 c) 09:37, 21 August 2020 (UTC)

QEDK, please stop this evasion, and please own your own words. You have made it very clear that A) the ones with no backlinks are basically pointless (in most cases, and that B) they can be deleted by G6. So it is entirely clear that the whole purpose of this bot is making a hitlist to cull redirects. You have now made your goal crystal clear (which you didn't do at the grossly under-notified BRFA), and your attempt to cast that as my "narrative" or "perception" is a viciously nasty response: it's gaslighting. Please conduct yourself much better. I have seen this pattern before: -manipulation of consensus-formation (by woefully inadequate notification and lack of upfront clarity about goals) and then a belated admissson of the real goal followed by an attempt at gaslighting the objectors, It is very nasty stuff; please stop it. --BrownHairedGirl (talk) • (contribs) 10:02, 21 August 2020 (UTC)

Quit the virtue-signalling. The very person you're against, speaking out against you is a standard expectation, it's not gaslighting - and again, quit the sensationalism, this is a collaborative environment, not your primetime American talk show. You keep saying goals being made "crystal clear" and what not but have no evidence for any such mass deletion, what are you on about? The BRFA had 14 unique editors which is at at the higher end for a BRFA, and you call it manipulation of consensus, so I don't know what your words are supposed to mean. You also seem to suggest a pattern, so maybe you can point out the pattern for the visitors to this page? And I know you're quite into labelling detractors as gaslighters and what not, but using that word in such a minor context just takes away from the actual roots of the word, based in emotional abuse and gender inequality - which is not what any respectable editor in this community should do. --qedk (t 愛 c) 11:09, 21 August 2020 (UTC)

A shameful response. This is indeed a collaborative environment, which is why I am horrified at your manipulative conduct.

Yes, the BRFA had 14 editors ... but it didn't have any of the editors who do most of the consensus-building work on deleting categories and developing guidelines... because you chose to ignore the instructions at WP:BRFA point I, 2nd bullet to

If your task could be controversial (e.g. most bots making non-maintenance edits to articles and most bots posting messages on user talk pages), seek consensus for the task in the appropriate forums. Common places to start include WP:Village pump (proposals) and the talk pages of the relevant policies, guidelines, templates, and/or WikiProjects.

You didn't notify the talk pages of the relevant policies, guidelines, templates, and/or WikiProjects, and you didn't ask anywhere for any help in finding such places. --BrownHairedGirl (talk) • (contribs) 16:25, 21 August 2020 (UTC)

@QEDK: Can you please explain what you think constitutes a useful category redirect? In my view, categories are only used to categorise pages; whether or not there are any other links to those categories is meaningless. With category redirects, there shouldn't ever be any pages in them, and unless there's been a discussion about them, there probably won't be any links either. As I understand it, they're used to prevent duplicate categories and aid searches. Very few category redirects are just bad misspellings in the way many article redirects are. They're not automatically created from page moves. They're made deliberately. M.Clay1 (talk) 13:43, 21 August 2020 (UTC)

It's certainly possible that they are useful and it's also quite plausible that they are not, and having a backlink doesn't automatically consitute usefulness automatically but rather serves to prove that an unlikely redirect that has no backlinks is quite possibly not worth keeping around. Not all of them are created deliberately though and even the ones that are, might not be worth keeping around. A category is only as useful as the members it contains, we have C1 for categories that don't, but it's quite possible that some categories could be exempted from that which shouldn't be, that's as simple as creating pointless redirects from А to A (the former is a Cyrillic A) to something like Category:All deaths that have occurred in 2020 which is a long-winded name for a category that already exists and these would be exempted from C1 because they are redirects, but if they have no backlinks and this is checked by a bot, deletion of such categories is much easier to handle - especially because a lot of the categories we presently have are permanently empty and it's impossible to determine if all of them should be kept so anything that makes clean up easier has a rationale to be around, and none of this is mass deletion because an administrator will still have to review the category of their own volition at CfD/CSD before deleting it. This tracking category serves only to be an indicator of (non-)usefulness of an empty category, category redirects are but a small aspect of it. --qedk (t 愛 c) 14:01, 21 August 2020 (UTC)

I still don't see how it makes it easier. If we're saying that having no backlinks makes a category redirect more likely to be useless (which I disagree with), on top of all the false positives and false negatives of that logic, there will be so many redirects tagged that no one is going to go through them all to check. Most of them will have no consensus to delete. It's much better to just nominate individual redirects as you find them. As you say, no category redirects are valid under C1 because they're supposed to be empty (and we have bots to automatically move any pages placed in them), so under what criteria would a "pointless" category redirect be nominated for deletion? I can't see how they could be deleted in any way other than a regular deletion discussion, and I can't see how the bot is making that easier. Do you have an example of a category redirect with a backlink that isn't from a discussion about it? I can see how it might be useful for empty categories that aren't redirects, though surely it would be much better to just track that and not have the tracking category polluted by all the redirects (which will greatly outnumber any others). But how many empty categories that aren't redirects are there? Don't we already have bots for patrolling that? There are a few tracking categories that are sometimes empty, but they are useful and are highly unlikely to have no backlinks. M.Clay1 (talk) 14:21, 21 August 2020 (UTC)

@Mclay1: It was in the works and I pushed it today, so it's no longer exclusive to category redirects. And even though we have around 92k category redirects, only around 50k seem to have no backlinks. So, clearly there is a disparity and the "no category redirects have backlinks" claim is false if so many don't. --qedk (t 愛 c) 15:44, 21 August 2020 (UTC)

It's quite extraordinary that after all the discussion on this page and at CFD, QEDK still persists in the false notion that a category redirect with no backlinks is somehow a problem which needs to be categorised. --BrownHairedGirl (talk) • (contribs) 16:16, 21 August 2020 (UTC)

It's also quite extraordinary that when there are two current discussions in which all participants oppose the bot's work, QEDK's response is not to pause the bot and seek consensus ... but instead to expand the scope of the bot's work. That is not compatible with the instructions at WP:BOTISSUE that Bot operators are expected to be responsive to the community's concerns and suggestions. The collabrorative way to respond to objections is to pause and discuss ... but QEDK has chosen to do the exact opposite, by engaging a higher gear.

There are currently 53,145 pages in Category:Empty categories with no backlinks, of which https://petscan.wmflabs.org/?psid=17146842 finds that 53,018 are category redirects. So 99.76% of the the contents Category:Empty categories with no backlinks are redirects, which shouldn't have backlinks. This is a useless set: it would be massively more useful to simply make a list of the 127 pages which are not redirects. --BrownHairedGirl (talk) • (contribs) 16:48, 21 August 2020 (UTC)

I was demonstrating that the functions previously requested by Liz and US were dealt with. I was planning to disable it once I could demonstrate that, so I will be disabling it shortly. As I've stated before, it isn't a problem but it's being tracked because it could be an indication of a problem. The point of contention differs between all of you on the bot and category. That's not consensus but I understand the sentiment anyway. --qedk (t 愛 c) 17:56, 21 August 2020 (UTC)

• If I understand correctly, the bot has been revised to exclude redirects, so that we will get the list of 127 pages. It may be worthwhile to wait and have a look what these 127 pages actually are. Marcocapelle (talk) 21:01, 21 August 2020 (UTC)
  • I hope that the changes will lead to creating a list, rather than the category. --BrownHairedGirl (talk) • (contribs) 01:39, 22 August 2020 (UTC)

QEDKbot malfunctioning, bot owner not fixing

I just spotted that at 14:27, 22 August 2020‎, the bot added[4] to Category:Empty categories with no backlinks, even though it is a disambiguation category.

https://petscan.wmflabs.org/?psid=17167124 shows that there are 441 categories which are in both Category:Empty categories with no backlinks and Category:Disambiguation categories -- presumably all added by the bot.

Note that at Wikipedia:Bots/Requests_for_approval/QEDKbot, WEDK pledged that: "If the category is tagged with {{Db-c1}}, {{Possibly empty category}}, {{Disambiguation category}}, {{Cfd full}} (Cf* to be accurate) or its redirecting templates, it will skip the page."

It isn't doing that: see also User talk:QEDK#please_fix_your_bot

I am pleased to see that QEDK says they have shut off this wretched disruptive bot ... but it seems very clear that it is not being competently run. Even if QEDK can't made the code work, they should at least be checking the bot's actions so that they spot these errors yourself ... and then fix them. Instead, nearly 36 hours later, these 441 errors are un-noticed and unfixed.

At 12:23 on 21 August, QEDK said[5] that the bot had been revised to stop placing category redirects in Category:Empty categories with no backlinks, but too has not happened: https://petscan.wmflabs.org/?psid=17167197 shows that 53349 category redirects are still in Category:Empty categories with no backlinks, which is over half of the 97,802 categories in Category:Wikipedia soft redirected categories.

This beaches at least two counts of WP:BOTACC:

1. "Bot accounts should not be used for contributions that do not fall within the scope of the bot's designated tasks."
2. "In particular, the bot operator is responsible for the repair of any damage caused by a bot which operates incorrectly"

Please BAG, can you just revoke this bot's authorisation? There is no consensus to support its declared actions, the bot owner is failing to:

1. make the bot work within its remit
2. check the bot's edits
3. fix the bot's prolific errors

Pinging some WP:BAG members. @TheSandDoctor, Headbomb, Xaosflux, and Primefac: please stop this bot. If it creates any more mischief, I will take this to ANI to ask for the bot to be blocked. --BrownHairedGirl (talk) • (contribs) 00:49, 24 August 2020 (UTC)

The bot has not edited for almost two days. There is no need to have the bot blocked. I haven't had a chance to look into this too deeply, but I will say that just from a quick skim there seems to be an inordinate amount of vitriol being thrown around. I know it's a subject certain editors are passionate about, but it's also part of the reason why I've wanted to see how things shook out before diving in. Primefac (talk) 00:58, 24 August 2020 (UTC)

After review here, I see no grounds to revoke or even amend the scope of QEDKbot. QEDK shut the bot down while he investigates an issue with its behaviour. This is responsible bot owner behaviour. You don't get to ask for a block, jumping ahead in the steps outlined at WP:BOTISSUE, simply because a bot operator has a different schedule than yours. Headbomb {t · c · p · b} 01:12, 24 August 2020 (UTC)

I concur with the above points by Headbomb and Primefac. Please review WP:BOTISSUE prior to making such requests. At this point in time, I am going to say that that request is  Denied under current circumstances. Bringing this to ANI requesting a block is certainly within your choice, but I would recommend against it at this time. --TheSandDoctor ^Talk 04:20, 24 August 2020 (UTC)

@TheSandDoctor, Headbomb, and Primefac: the problem here is that QEDK promised to fix the bot to remove the ~50,000 category redirects from Category:Empty categories with no backlinks. Instead of doing that, QEDK set the bot off on a spree of adding further pages to a category which is up or deletion at CFD, including the cat dab pages ... and did not remove the category redirects. QEDK's statement below that they have identified the issue and fixed the code does not state what the revised code is intended to do. Will it untag the category redirects? Will it untag the cat dabs? QEDK gives no clarity.

There is also not commitment from QEDK to monitor the bot's operation: the flaws so far have been detected by others, not by the bot owner.

This wretched bot was approved without proper community consensus, and it is clearly running in experimental mode across the whole set of categories. Even if BAG is still happy to overlook the lack of community consensus for this bot's actions, it should not have gotten past trial stage until those issues had been resolved. And on top of that, the bot owner fails to communicate with clarity. --BrownHairedGirl (talk) • (contribs) 08:06, 24 August 2020 (UTC)

Just a few thoughts, and I genuinely am not trying to dump it all on you but I still haven't had an opportunity to really dig into the substance of the initial concern and all that has been posted here; I'm just replying to your latest statement. First, I seem to recall that the bot does not need to untag the categories in the "no backlinks" category, because when the CFD closes as "delete" a different bot will handle the orphaning. Second, you keep saying essentially "no one knew this bot existed, and it was approved lightning fast without any trial." It did go to trial (twice!) with 140 edits made to categories, and the process took four months from time of first trial to time of acceptance. As far as the "gotten past trial" comment re: the BRFA, every single time someone said "what about..." or "this is broken" the issues were dealt with; that's why we do trials (and second trials) in the first place.

Again, I'm literally just replying to your last comment, and I totally understand your frustration, but it's starting to sound like a reasonable issue is being blown out of proportion (though I do understand the juxtaposition of saying "reasonable" matched with 50k potentially problematic edits, more that it's a single issue). I'll do my best to take a proper look at this later today and give my full thoughts further down the page. Primefac (talk) 10:32, 24 August 2020 (UTC)

Thanks, @Primefac. The issue here is that more 50% of all en.wp's category redirects have been tagged with a category whose stated purpose is that they have a flaw which may merit speedy deletion per WP:G6 ... even though the alleged flaw is in fact an indication of a healthy category redirect. That is a real biggie: a potentially massive amount of damage to the category tree, which would have devastating effect on the ability of editors to accurately categorise articles. And since G6 requires no prior notification, the bot's tagging of the categories may be the last warning before deletion.

That's why I am so concerned about this ... and the reason I am angry about it is that the bot owner is repeatedly deceptive, evasive and gaslighting (if needed, I will provide multiple diffs to support each of those assertions). Quite the most unpleasant bot owner I have encountered since the dark days of BetaCommand, and I have no doubt at all that their appalling handling of objections is alone enough to leaves them totally unfit to run a bot, even w/o considering the other factors.

The underlying issue is that there is no consensus for the bot's purpose, because no category-related page was notified, so most editors who work on categories did not know about it (I didn't spot it until the category was at CFD). QEDK has explained that they were unaware of any relevant pages ... which, if true, would mean that they were unaware of WT:CFD, WT:CATP, WT:CSD, WT:OC. I find that highly implausible (an admin who doesn't even know that one of the XFDs exists?), but in the unlikely event that it is true, it was QEDK's responsibility to ask "where should I leave notices to catch the attention of editors specifically interested in categories". QEDK did not do that, and instead claims hilariously that he did due diligence. Sorry, but dunno + didn't ask ≠ due diligence.

With respect, I think you made a rare misjudgement (I truly do mean rare) in authorising the bot without any notifications on category-specific pages, let alone any prior discussions there. Compare and contrast with the extensive discussion at WT:CATP on pre-proposal of BHGbot 6 and pre-proposal of BHGbot 7. Editors there do play close attention to bot proposals, if asked ... but in this case there were not asked. And one of the things that really annoys me about this is that almost 100% of the work of WP:BHGbot 7 is now in the Category:Empty categories with no backlinks, which QEDK regards as mostly fit for deletion. In other words, massive consultation led to BHGbot 7's output; but a stealthy proposal may wipe it all.

As the CFD, yes, you're right, if the CFD is closed as delete, then all it pages will be removed from the category.

However, the CFD is still open after 15 days, and even tho the !voters are 100% delete, a potential closer faces an unusual dilemma: how to weigh the apparent contrast between the CfD !votes and the fact that the BRFA was approved? How does this all fit with WP:LOCALCON? Any possible close there is problematic, because deleting the cat effectively kills the bot ... and not deleting the cat ignores the fact that at CFD, nobody backs the bot-owner.

AND QEDK has outright refused to remove the category redirects from the Category:Empty categories with no backlinks:[6] I will not be uncategorising category redirects because that does not fall within the remit of the bot's stated functions (and would thus be considered misuse of the bot)

That's why I think this is back in BAG's court. Revoking the bot authorisation allows the CFD to be closed. And a fresh BRFA at some future date, properly notified after pre-discussion, can consider whether there is consensus for any of the bots functions ... as well as whether this bot-owner is competent to run a bot.

But if BAG fails to revoke, then we will have a clash of two processes which will have to be escalated (to be clear, I will escalate if this isn't resolved, but I am unlikely to be alone). It seems to me to be vastly preferable to go back to the start and do the proper consensus-building rather than to plough ahead in an escalating drama. Sorry this has ended up so long. --BrownHairedGirl (talk) • (contribs) 13:22, 24 August 2020 (UTC)

(Redacted) --qedk (t 愛 c) 21:33, 24 August 2020 (UTC)

(Redacted) --BrownHairedGirl (talk) • (contribs) 00:24, 25 August 2020 (UTC)

Enough. I've removed both of your recent comments because neither one is at all helpful to this situation. You've both said more than enough about this situation, and you've long since stopped saying anything useful. Let me look over the damn situation and knock off the sniping. Primefac (talk) 00:27, 25 August 2020 (UTC)

@Primefac, this has turned sour because of failings in BAG's oversight on two key issues: consensus for the bot's operations, and the failure the bot owner to uphold basic standards of effective communication, let alone the enhanced requirements of WP:BOTCOMM. I appreciate that you and the other BAG members are volunteers with limited time, but this does need resolution. --BrownHairedGirl (talk) • (contribs) 00:53, 25 August 2020 (UTC)

• Per my reply to Trappist on my TP, I've identified the issue and fixed the code. --qedk (t 愛 c) 06:57, 24 August 2020 (UTC)

BAG analysis

I'm not going to bother going into a lot of detail about my findings, because quite honestly there has been about 40k worth of text already dedicated to it. I'll do my best to summarize. The bot, as described in the BRFA, performs the following operations:

1. Goes over all existing category pages with zero members
2. Tags the page if:
   1. It is a category redirect with no backlinks (via the oft-mentioned Category:Empty categories with no backlinks)
   2. It is not tagged with some sort of "possible empty", dab, or CSD template (via WP:C1 nomination)
3. Deletes a C1-tagged page if it does not meet 2.1 or 2.2's criteria and has been empty for 7 days

From everything I've read, the only contentious part of this bot run is 2.1, which in hindsight is reasonable given the scale of the editing and the ensuing apparent disruption to those heavily involved in category maintenance. Unless I'm missing part of a discussion, though, tagging or deleting as C1 has not been a concern of this bot. And yes, I know that there were some issues with mis-tagged C1 pages, but those bugs have apparently been dealt with and are not concerns with the task itself.

In other words, I find no reason to completely revoke permission for this bot, only the revoking of subtask 2.1. Unless one of my fellow BAG members feels otherwise, or there is demonstrated consensus against the other subtasks of this bot, I will consider this matter closed. QEDK, in the interest of good faith, please do not restart this bot task until one of us (BAG) gives the go-ahead, if only to allow for any further discussion that might arise from other parties. Primefac (talk) 00:51, 25 August 2020 (UTC)

Thank you, Primefac. This is valuable progress.

Yes, the contentious issue has been the tagging of category redirects with no backlinks (via the oft-mentioned Category:Empty categories with no backlinks). Revoking that task is the key first step to resolving the dispute.

However, revoking 2.1 does not of itself sort everything. That would still leave us with a bot which populates Category:Empty categories with no backlinks, which there is a currently consensus to delete. We just await the CFD's closure).

If the category is not deleted, then the ~50,000 category redirects need to be removed from it. QEDK has refused to do that ... so who will do it?

If the bot continue its WP:C1 task, it would better to create a report rather than edit the category. Many categories fluctuate around emptiness, and the bot's approach of categorising will lead to a significant amount of editing as pages are added to and then removed from Category:Empty categories with no backlinks. Other similar task create reports, and the objection to doing that in this case was based on the sheer volume of category redirects. That no longer applies.

When the catredirs and catdabs are excluded, we are left with about 1898 pages potentially eligible for C1 (see https://petscan.wmflabs.org/?psid=17176395) ... but a significant proportion of those are WikiProject assessment categories, many of which which also fluctuate around emptiness and should be tagged with Possibly empty category. Those should also not be C1ed, and even excluding Category:Articles by quality and importance+subcats and Category:Wikipedia vital articles+subcats still leaves a set of ~600 pages of which about half are WikiProject assessment categories (see https://petscan.wmflabs.org/?psid=17176443). I took those Petscan rosults offline to filter out the pages whose titles include "-Class" or "-importance, and was left with 503 pages. But that set of 503 still contains ~#100 subcats of Category:Articles containing non-English-language text, which should also not deleted when empty, plus dozens of need-infobox categories which should also be tagged as possibly empty ... and there are probably more sets of false positives.

In other words, even without task 2.1, the bot needs masses of ongoing tuning to resolve the wide range of false positives which it is generating. Its owner needs to work with the community to deploy Possibly empty category more widely, and to engage openly with the wide range of issues that raises. That is important in building a useful list, but if the bot is doing any automatic CSD tagging or deletion, then it is absolutely essential that such false positives be entirely eliminated.

Such fine-tuning of the bot's scope requires a bot owner with high communication skills who actively seeks out feedback, welcomes criticism, responds clearly and promptly to concerns, restricts the bot to non-contentious actions, and above all works hard to build consensus. Unfortunately in this case, we very clearly do NOT have such a bot operator.

If this bot continues in any form, we remain with the unresolved problem of QEDK's appalling communication style, and especially their contempt for WP:BOTCOMM. This harks back to the Betacommand fiasco, who was a very competent programmer but entirely lacked communication and consensus-building skills, leading to mega-dramas. BAG at the time paid insufficient attention to the non-technical issues, which thankfully appeared to be resolved by the demise of those bots and by a sort of reboot of BAG. I welcome the move to resolve the biggest technical issue with this bot, but I am concerned that there has been far too little scrutiny of the rest of its tasks, and that BAG is not addressing the bot owner issues which have been starkly revealed over the last week. This bot's scope involves masses of nuance and complexity, but its owner does not even communicate effectively on simple issues, and is actively hostile to nuance and complexity.

And I am sorry, this long ... because there are so many unresolved issues --BrownHairedGirl (talk) • (contribs) 03:09, 25 August 2020 (UTC)

I guess I'm a little confused at your reply. I'll try to break it down

• ...revoking 2.1 does not of itself sort everything. That would still leave us with a bot which populates... no, if 2.1 is revoked then the bot will not be populating that category, since that is task 2.1.
• If the category is not deleted... it currently has unanimous consensus to delete
• If the bot continue its WP:C1 task... This one requires a little more than a hand-wavey reply, but what I am gathering from your concern is that a whole lot of categories that should have the required "exclusion templates" but don't will be nominated, in other words the false positives. This is not a problem with the bot, or the operator, or even the task itself, but rather with the actions (or inactions, in a way) of the editors who set up the categories falling into this false-positive group. It is not qedk's responsibility to make sure every category that is empty and nominated should be nominated (at which point it would be a 100% manual task and defeat the purpose of a bot), and I suspect that's one of the primary reasons why db-c1 has a one-week delay time; to give other editors enough of an opportunity to see why the category they created has suddenly been nominated. If anything, this bot will help in that process by pointing out potentially valuable categories that have been mis-managed (through intention or accident). Additionally, if any of the false positives do slip through and get deleted, REFUND is a tragically easy process to navigate.

I think the only major modifications to the task might be a limitation on how many pages to tag as C1 in a single run/day/week/etc, as we of course don't want to flood Category:Empty categories awaiting deletion (or the "ready to delete" cat) with so many cats they cannot be checked in a reasonable time frame. Would that satisfy concerns about over-tagging false positives? Primefac (talk) 15:16, 25 August 2020 (UTC)

@Primefac, my understanding is that what you propose is to stop the bot adding category redirects to Category:Empty categories with no backlinks. Surely that would still leave it still doing what you label as task 2.2, i.e. adding non-redirected categories to Category:Empty categories with no backlinks?

That's why I see a clash between the bot task and CFD. As you note, the currently has unanimous consensus to delete. But AIUI, the bot Task 2.2 will still be adding non-redirected cats to Category:Empty categories with no backlinks, so the we will end up with the bot authorised to populate a non-existent category. That would be a breach of WP:REDNOT, and would flood Special:WantedCategories ... which is why I suggested that instead the bot create a list.

The false positives issue can be handled in two ways:

1. by the bot skipping certain types of categories,
   or
2. by a big effort to tag all such categs as {{Possibly empty category}}

There are merits demerits in both approaches, so some discussion is needed. The best outcome will probably be some sort of hybrid of both approaches, and I don't think it's helpful for you describe this grey area as mismanaged.

The bot is clearly already excluding some sets of categories (e.g. it seems to be skipping cats which transclude {{Category class}}), so there is no great issue of principle in adding more exclusions. It's just a mater of forming a consensus on what should be excluded, and ensuring that bot's documentation is clear about what is excluded.

I would urge that Task 3 (actual deletions by the bot) be suspended until the other issues are sorted out. That way the community can assess one of the bot's functions at a time. --BrownHairedGirl (talk) • (contribs) 16:01, 25 August 2020 (UTC)

Task 2.2 is tagging empty categories (that aren't excluded because of {{pec}} etc) with {{db-c1}}. The disputed category at CFD has nothing to do with it. Primefac (talk) 16:03, 25 August 2020 (UTC)

Regarding the rest of your post, yes, skipping specific types of categories is another option (and is already being done for WikiProject assessment cats), but that somewhat favours ignoring a problem because we can't be bothered dealing with it. Primefac (talk) 16:11, 25 August 2020 (UTC)

@Primefac, as I noted above, Category:Empty categories with no backlinks does include pages which are not part of task 2.1: see https://petscan.wmflabs.org/?psid=17176395, which currently shows 1,898 of them.

Is this a bot malfunction? Or is there some discrepancy between your understanding of the bot's task and QEDK's understanding? --BrownHairedGirl (talk) • (contribs) 16:50, 25 August 2020 (UTC)

They appear to be listed because of a bug, which seems to have been dealt with here. Primefac (talk) 17:16, 25 August 2020 (UTC)

I don't think it's a bug, @Primefac. I think it's a feature: see [7] where QEDK (with typically unhelpful terseness) describes as a "fix" the adding of a non-redirected category to Category:Empty categories with no backlinks. (That one was a possibly-empty category, so it should have been excluded on those grounds, but the relevance here is that it is a non-redirect).

So far as I can see, the issues raised by Trappist at User talk:QEDK#please_fix_your_bot relate to the failure to exclude categories with various tags such as possibly empty. They do not relate to the inclusion of non-redirects.

The simplest way to clarify this simply to ask QEDK a direct question: will removed task 2.1 make to bot sop populating Category:Empty categories with no backlinks?

If the answer to that is "yes", then problem solved. --BrownHairedGirl (talk) • (contribs) 18:27, 25 August 2020 (UTC)

@BrownHairedGirl: If you keep this constant incivility and WP:ABF mentality, I will take you to WP:ANI. Consider yourself warned for the second time here. Headbomb {t · c · p · b} 18:31, 25 August 2020 (UTC)

@Headbomb, I presume that relates to my comment about typically unhelpful terseness. The reply to which I linked was unhelpfully terse, as are many of QEDK's other replies. It is not uncivil or ABF to note that communication problem. --BrownHairedGirl (talk) • (contribs) 18:36, 25 August 2020 (UTC)

It applies to that (the response was very clear, and there's no need to spend 342 paragraphs to say something that takes 1 line to say). But it also applies to your entire interaction with QEDK and on his talk page, like assuming bad faith when QEDK said something was a bug, and you insist it's "a feature". Headbomb {t · c · p · b} 18:40, 25 August 2020 (UTC)

@Headbomb, I disagree: the response was not clear, for the reasons I have set out.

The point about bug/feature is not ABF: it is about clarifying whether the bots' adding of non-redirected categories to Category:Empty categories with no backlinks was an error or what QEDK intended. Please look at [8] where QEDK describes as a "fix" the adding of a non-redirected category to Category:Empty categories with no backlinks. That was in response to Marcocapelle's point about the Category:Empty categories with no backlinks being a subset of Category:Wikipedia soft redirected categories, so I read it as meaning that the addition of non-redirects fixes the overlap. If QEDK meant something else, they should have explained exactly what they were fixing, rather than linking to a diff and leaving the reader to infer what was meant. We would not have needed so many paragraphs to unravel this mess if QEDK communicated clearly. --BrownHairedGirl (talk) • (contribs) 18:55, 25 August 2020 (UTC)

That was what I meant by restricting 2.1, yes; since the category should not exist/have members, the addition of pages to that category should not be happening. Primefac (talk) 18:30, 25 August 2020 (UTC)

Thanks, @Primefac. If that category is not to be populated, then that issue is sorted. Your 2.1 as worded was narrower than that, so I am glad we have cleared up the misunderstanding.

To wrap this up, please could you leave a note at WP:Categories for discussion/Log/2020 August 9#Category:Empty_categories_with_no_backlinks confirming that the bot is no longer approved to populate this category? That will assure the closing admin there that there is no clash between the CFD consensus and the bot's mandate. --BrownHairedGirl (talk) • (contribs) 18:40, 25 August 2020 (UTC)

I have to ask but what is it going to take you to realize that Wikipedia has no place for your incivility? I've ignored it time and time again but I don't think you're going to stop or improve, and I really don't know if you think your behaviour is a prime example of good WP:BOTCOMM or it's meant to be an ironic comment everytime you bring it up. That's all I'll say re: typically unhelpful terseness. I'm not sorry if your "communication" entails suffering a barrage of personal attacks, ad hominem and whatnot, I'm fine without those communications, thank you.

Now to provide some numbers to words, the bot nominates around 1-3 categories per run (that amounts to 3-9 per week), the reason that number is low is because 1) it's organic and probably arises out of someone decatting and 2) the criterion is slimmer than C1 (which itself is very restrictive). From the very first trial itself, I've been working to ensure that false positives are minimal for deletions and nominations; ofcourse it's plausible that some do fall through the cracks, but a quick look over the last 500 shows that one deletion was "bad" (it was nominated by another user and wasn't rescued), I have not received any concerns regarding this on my talk page till date so I can only presume that it works alright. In any case, @Primefac: I have no intention of running my bot until BAG gives a clear remit to do so, I totally understand that until it's sorted, you as a BAG member should put the functioning on hold (as it currently is) but I think it's clear that running a bot for this community no longer has any benefit. Coming to the question of tagging {{Pec}} categories, it was definitely not a feature (as I clarified with my reasoning to Trappist), the feature was to tag categories other than category redirects, in which process, I introduced the bug and it was fixed once I was made aware of it. Lastly, regarding "reports": I very much doubt that listing the category in a page format is suitable because MediaWiki has pagesize limits and would require the bot to make multiple pages, which would be quite difficult to navigate through, totally not worth the hassle imo, it would be simultaneously harder on end-users and me. --qedk (t 愛 c) 19:11, 25 August 2020 (UTC)

@QEDK: A lecture from you on incivility is mighty ironic.

I did not suggest that tagging {{Pec}} categories was a feature. The issue I raised was that the bot had place a non-redirected category in Category:Empty categories with no backlinks, which you described as a "fix". That's why I assumed that you intended this to be a feature. If that action was in fact an error, then your decision to describe it[9] as a "fix is bizarre. Note that the diff you posted was Special:Diff/974163791, which is a tagging of PEC category ... and I presume that you didn't intend that. --BrownHairedGirl (talk) • (contribs) 19:53, 25 August 2020 (UTC)

• @Primefac: due to the length of this discussion I haven't gotten to reading all of this, but is there anything you need me on for this?—CYBERPOWER (Chat) 19:07, 25 August 2020 (UTC)

  If you think my post at the top of this section seems reasonable (i.e. I didn't miss anything obvious), that's the main thing. Primefac (talk) 22:09, 25 August 2020 (UTC)

  Primefac, I've never known for you to misrepresent a discussion, so it seems very reasonable to me. —CYBERPOWER (Around) 00:25, 27 August 2020 (UTC)

  Primefac, FWIW, I believe the post at the top is reasonable, and revoking 2.1 for the time being is the way to go. SQL^{Query me!} 15:25, 27 August 2020 (UTC)

  Excellent, thank you both. Primefac (talk) 15:38, 27 August 2020 (UTC)

Moving forward

qedk, I believe we're at the point where this can start being rolled out again. As a summary of my summary above, the bot has approval to:

• tag pages with {{db-c1}} if they are not a) category redirects, or b) pages with "exclusion templates" such as {{pec}} and {{catdab}}. Ideally, it would be beneficial to also skip WikiProject Assessment categories (e.g. "Stub-class XYZ pages") if that is easily done.
• delete pages tagged as {{db-c1}} that meet the criteria listed in the BRFA

I am also going to temporarily include a rate limit on the C1 taggings - no more than 100 per week, if only to avoid flooding the category and allow for checks to be more easily made to ensure any pages that should have {{pec}} etc actually get them. We'll let it run like that for a few weeks and reevaluate. If this sounds acceptable, let me know, but I'm still happy to discuss any concerns. Primefac (talk) 15:38, 27 August 2020 (UTC)

@Primefac: That sounds good to me. Regarding your first point, the bot already skips those kind (WikiProject-related) for nomination and deletion but it's more of a bruteforce strategy since a lot of those categories have no tags at all, so I'm using a combination of tag-based, title-based and content-based checks. I should also drop a note that revoking 2.1 also effectively takes away half of the 2nd BRFA's task of patrolling the category - so there needs to be roadmap on what's next. Finally, 100 as an upper limit is alright because I run the bot for a few times per week and I don't think it will get close to the upper limit at all (and I don't want to deal with the fallout of making 100 nominations by running the bot continuously). On a final note, there's also a log at User:QEDKbot/Catlog if anyone wants to see the deletions the bot makes (if it makes wrong ones, let me know) - although the titles are a bit strange even though the return types are supposed to be "string" (see here). That's about it from me. --qedk (t 愛 c) 08:39, 28 August 2020 (UTC)

Yeah, I think that subtask needs the sort of discussion BHG was referring to in the original concerns about this task - asking the categorization projects how they feel it would be best to maintain and track these sorts of pages to get more buy-in. It might not happen right away, but I'm sure that some form of 2.1 will be re-implemented eventually. Primefac (talk) 12:06, 28 August 2020 (UTC)

@Primefac:. After all the discussion above, it would been helpful to pinged about this. That omission is surprising.

Because of the inadequate notification of the initial BRFA, these tasks have not been adequately scrutinised. Please do not re-start the bot until consensus is properly established for whatever task it is to perform.

A few specific points

1. Up above, I set out a range of category types which should not be deleted, in my post[10] of 03:09, 25 August> this includes:
   • Category:Articles by quality and importance+subcats
   • Category:Wikipedia vital articles+subcats
   • subcats of Category:Articles containing non-English-language text
   • There is a misunderstanding in Primefac's comment above about WikiProject Assessment categories (e.g. "Stub-class XYZ pages"). Stub categories (e.g. Category:Japan stubs) are populated by templates applied to article pages. They are wholly separate to WikiProject Assessment cats (e.g. Category:B-Class Japan-related articles), which are populated by talk pages, through WikiProject banner templates.
     I don't expect that editors not working intensively with categories should be fully aware of all these nuances ... but that's why more eyes and wider discussion is needed.
2. There may be other types ... but right now there is no clear documentation of what is skipped and what isn't. For example, it seems that the bot has been skipping some WikiProject assessment cats, but this is not documented
3. Until there is clear consensus for what should be skipped, any runs of the bot should make lists of category pages, rather than tagging anything for deletion. Please can we have an assurance from BAG that C1 tagging will not start until there is clear consensus for that task at a discussion which has been properly notified n category-related pages? (including WT:CFD, WT:CSD, WT:CATP)
4. It seems to be to be highly unlikely that any variant of task 2.1 will be supported. The clear consensus at WP:Categories for discussion/Log/2020 August 9#Category:Empty_categories_with_no_backlinks is fundamentally misconceived. Please can Primefac or other BAG members give an assurance that this will NOT be restarted without a new BRFA which has been clearly notified to a range of category-related pages (including WT:CFD, WT:CSD, WT:CATP).
5. The last round of logging at User:QEDKbot/Catlog was broken: see e.g. [11]. If and when there is consensus to restart the bot, please can the bot NOT be restarted until until testing has verified that logging actually works, and that its edit summaries link to the log?
6. Please can we have an assurance that if the bot is to be run by QEDK, they will uphold WP:BOTCOMM by improving their communications?
   It's highly disruptive to have a bot where the owner replies with cryptic comments like[12] (what was the "fix" in the chosen example? tagging a PEC category, or tagging a non-redirected category?), or where the bot owner deletes[13] from their talk page a notification[14] of a problem which has not been previously raised?

I do welcome the end of task 2.1. But, please please ... it's now time to build consensus. --BrownHairedGirl (talk) • (contribs) 17:19, 28 August 2020 (UTC)