Wikipedia:Village pump (proposals)

From Wikipedia, the free encyclopedia
 Policy Technical Proposals Idea lab WMF Miscellaneous 

The proposals section of the village pump is used to offer specific changes for discussion. Before submitting:

Discussions are automatically archived after remaining inactive for nine days.

Enable RC patrolling (aka patrolling for edits)[edit]

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.

This is something that has been bothering me for years now. Many other wikis have patrolling for edits in recentchanges and watchlist and it drastically improves efficiency of vandalism patrollers. Edits done by autopatrolled users automatically gets marked as patrolled, reverted and rolled back edits also automatically get marked as patrolled too (as there is no need to be reviewed again) but also when someone reviews an edit and it doesn't need to be reverted, they can mark the edit as patrolled which avoids double, triple or more reviews by others.

You can enable the filter to only look at unpatrolled edits (or combine that filter with other filters, such as ores damaging ones). The software for it already exists and this feature is enabled in Wikidata, Commons, French Wikipedia, Italian Wikipedia, Dutch Wikipedia, Chinese Wikipedia, Portuguese Wikipedia and Vietnamese Wikipedia (Wikis with above >1M articles) and many more wikis.

One concern was that it might start a massive backlog of edits needed to be reviewed (like pending changes) but the patrolled status gets purged after 90 days so no large backlog will be made (and it's better than status quo where there is no status stored). It was enabled in here in January 2005 but some people didn't like the "!" it adds next to unpatrolled edits. That was almost nineteen years ago when we didn't have ores or highlighting or filtering in RC/Watchlist. We can change that exclamation mark if people want it to or completely hide it (I can make the software changes necessary if there is consensus for anything different). That's not really an issue now.

What do you think? Ladsgroupoverleg 17:39, 10 October 2023 (UTC)Reply[reply]

  • Support as proposer. Ladsgroupoverleg 17:39, 10 October 2023 (UTC)Reply[reply]
    Heh. Disclosure: We had talked about this in person and I was invited to this discussion via e-mail. To my understanding, the one huge difference to Pending Changes is that all edits are immediately visible to the public. There is no queue of edits that need to be reviewed before they're displayed to readers. However, the patrol permission required to mark edits as patrolled seems to be the same one we grant at WP:PERM/NPR (cf. Special:ListGroupRights#patroller). We can either start granting that flag liberally (not going to happen) or practically forget about the idea then... ~ ToBeFree (talk) 18:25, 10 October 2023 (UTC)Reply[reply]
    @Ladsgroup: From your link: Patrolled edits is a feature which allows specific users to mark new pages and items in Special:RecentChanges as "patrolled" We have WP:RCP and WP:NPP which seems the same, or very similar. RudolfRed (talk) 18:28, 10 October 2023 (UTC)Reply[reply]
    @RudolfRed They are the same code and software but many wikis allow the same for edits (not just new page creations). Ladsgroupoverleg 18:39, 10 October 2023 (UTC)Reply[reply]
  • Question: First, is this compatible with mw:Extension:PageTriage that we already have? Second, I think at the least it uses the same groups/etc - which means we'd likely end up greatly expanding those that bypass parts of new page review. — xaosflux Talk 18:30, 10 October 2023 (UTC)Reply[reply]
    @Xaosflux Why would it interfere with PageTriage? This is about edits not page creations.
    They use the same rights as the NPP which can become a problem as @ToBeFree mentioned but the fix is not that hard in the software, we can split the right and define a new group. That part is fixable. Ladsgroupoverleg 18:42, 10 October 2023 (UTC)Reply[reply]
    @Ladsgroup I don't recall the details, but something about how the very enwiki customized pagetriage hooks the patrol system already? — xaosflux Talk 18:44, 10 October 2023 (UTC)Reply[reply]
    It uses patrolling for new pages but it can't interfere with the edit patrolling in any way, the only complication is that the rights are the same which can be fixed easily. Ladsgroupoverleg 18:48, 10 October 2023 (UTC)Reply[reply]
    The entire MediaWiki extension would be changed to make it compatible with enwiki's new page patrolling? ~ ToBeFree (talk) 18:48, 10 October 2023 (UTC)Reply[reply]
    nah, the code for it is not too hard to change and if we want to deploy NPP to more wikis that might have edit patrolling, we have to change it anyway. Ladsgroupoverleg 18:50, 10 October 2023 (UTC)Reply[reply]
  • Seems like trouble - every non-user revision would be un-patrolled, which I expect would instantly flood queue in to a backlog that will never get cleared. — xaosflux Talk 18:46, 10 October 2023 (UTC)Reply[reply]
    The queue gets automatically purged after 30 days, as I said above. Ladsgroupoverleg 18:47, 10 October 2023 (UTC)Reply[reply]
    And the point of it is not to clear any queues either. You will just have a a new filter to avoid re-reviewing edits and wasting your time. Ladsgroupoverleg 18:55, 10 October 2023 (UTC)Reply[reply]
  • Oppose. If the right is new-page patroller, a critical feed which is nearly always backlogged, we don't want to divide the attention of the small group of people who are trusted to work in this area into something significantly less pressing. Espresso Addict (talk) 21:53, 10 October 2023 (UTC)Reply[reply]
    Actually the more I think about this, the less I like it. Frank vandalism is usually easy to spot and revert, so wastes little time. The really problematic edits are PoV pushing of one form or another, and it would be trivial for a PoV pusher/paid editor to get the permission and then reduce scrutiny on problematic edits. Espresso Addict (talk) 00:19, 11 October 2023 (UTC)Reply[reply]
  • Comment I review most edits on my watchlist by hovering over (diff) (or "(3 changes)", etc.) with Popups. This is quick and easy. I'd be much less productive, and less likely to bother, if I had to click (diff) against every edit, wait for the page to load, click "mark as patrolled" and find my place in the watchlist again. Certes (talk) 22:11, 10 October 2023 (UTC)Reply[reply]
    The support for marking edits patrolled can easily be added to popups. Ladsgroupoverleg 11:05, 11 October 2023 (UTC)Reply[reply]
    Or... you could just not mark things as patrolled, @Certes.
    I don't think that everyone is understanding this proposal. Think of the question this way:
    • We could have a system that allows admins and other editors we trust to optionally signal to other editors – the ones who are voluntarily choosing to use this feature – to signal to the other editors in the opted-in group that one editor personally thinks a given edit no longer needs extra eyes on it (e.g., because it was reverted already, or because the patroller knows that it's good).
    • If you choose to use this optional system, then you can filter Special:RecentChanges to show you only those edits that other editors either haven't checked or are uncertain about. You would not need to see (and re-check) edits that other editors have already marked as acceptable. You would know which ones have no been checked, instead of assuming that the entire list of recent changes needs your personal attention.
    • Nobody will be required to use this system. In fact, it would probably be good to have a mix of editors: some being more efficient (so that everything gets checked, instead of some edits being checked many times and others being overlooked), some looking at their watchlists, and others looking at everything they can (to double-check).
    • Note that not re-re-re-re-checking known-good edits is what anti-vandal tools like Wikipedia:Huggle do automatically, so this is not a new idea.
    I think the question really is: Do you want to ban other editors from voluntarily and easily sharing information about which edits they've already checked, keeping firmly in mind that if you personally don't want to use their system, then it doesn't need to affect you at all? WhatamIdoing (talk) 01:07, 16 October 2023 (UTC)Reply[reply]
  • (edit conflict) If this happens, we definitely need separate "new page autopatrolled" vs. "recent change autopatrolled", as well as "new page patroller" vs. "recent change patroller" user-rights. I'm not convinced this is a good idea anyway, but I'm not an active RC patroller so I'll leave the merits to others.
    Although creating that split means that global rollbackers would lose new-page autopatrolled, instead of me having to manually unpatrol every article they create. * Pppery * it has begun... 22:13, 10 October 2023 (UTC)Reply[reply]
  • Comment. This reminds me of a conversation a few months ago at Help talk:Citation Style 1/Archive 89#Proposal: 'Verified' parameter, and is a subgenre of the mark something as "no action needed" problem. Personally I feel like the proposal here would create an immense amount of paperwork and possibly false sense of security, but I do think a general optional tickbox for "looks good to me" could be beneficial. Generally, we want to double check each other's work but not octuple check it. It's a grey area for sure.
    Folly Mox (talk) 22:29, 10 October 2023 (UTC)Reply[reply]
    While I understand your point, I want to mention that this has been working for over a decade in many large/medium wikis including mine without a a lot of paperwork and has been quite useful. A wiki can turn it into a paperwork nightmare or something lean if they chose to. Ladsgroupoverleg 11:22, 11 October 2023 (UTC)Reply[reply]
  • I'm generally supportive. If it can be made to work technically, i.e. there's no unresolvable conflicts with other extensions, and we can get a sensible set of permissions, I think this would help prevent a lot of duplicated effort. -- zzuuzz (talk) 22:42, 10 October 2023 (UTC)Reply[reply]
  • Support individual changes for rollbackers and administrators. Old vandalistic and other disruptive edits often go unnoticed for years. Oppose new pages inside of mainspace for anyone who is not a new page reviewer or autoreviewer, as those require a solid understanding of the deletion policy beyond the criteria for speedy deletion. Awesome Aasim 17:16, 11 October 2023 (UTC)Reply[reply]
    giving the right to rollbackers is a good idea to avoid creating yet another group of users and extra paperwork. There is a major overlap between rollbackers and patrollers too (if not the exact same). Ladsgroupoverleg 17:55, 11 October 2023 (UTC)Reply[reply]
.Oppose If you believe something might be wrong you should check it yourself, not rely on other editors. Either this user right is limit to a small group, making it ineffective, or a larger group, opening it up to all kinds of abuse. -- LCU ActivelyDisinterested transmissions °co-ords° 19:36, 11 October 2023 (UTC)Reply[reply]
I think you're wrong. You can choose to ignore this system and still check every change (do *you* check every change now?). You'll give the patrol right to your trusted users, with the possibility of having the right removed if they're no longer trusted; patrol actions are logged, and you can easily tell who marked each change as patrolled. Can you say that *every* change has been seen by someone (trusted) now? I'm pretty sure a user with, say, 2000 clean edits is a good patroller candidate, someone I'd trust. But I'd still sometimes take a look, even if the change was marked as patrolled. Nothing to lose here, trust me! Ponor (talk) 03:25, 15 October 2023 (UTC)Reply[reply]
@ActivelyDisinterested, you say If you believe something might be wrong you should check it yourself, not rely on other editors, but there's nothing in this proposal that would prevent you from doing exactly that. So why the opposition? WhatamIdoing (talk) 01:09, 16 October 2023 (UTC)Reply[reply]
I didn't say "I should check", maybe I should have said "should be checked". This system and many other recent discussion on similar things would mean that editors are relying on other editors for verification. That shouldn't happen, it's fundamentally wrong.
This has nothing to do with currently checking all edits, the question is trusting current edits. I don't blindly trust current edits, no editor should, but this would directly implement a system of doing exactly that. -- LCU ActivelyDisinterested transmissions °co-ords° 12:11, 16 October 2023 (UTC)Reply[reply]
Our current system works like this:
  • Alice looks at Special:RecentChanges and looks at the diffs for edits 1, 2, 3, 4, and 5. She wasn't sure about edit 2, but she didn't think it was bad enough to revert. Maybe someone else will look at it.
  • Bob looks at Special:RecentChanges and looks at the diffs for edits 1 and 4.
  • Chris looks at Special:RecentChanges and looks at the diffs for edits 1, 2, and 4.
  • David looks at Special:RecentChanges and looks at the diffs for edits 1 and 4.
Result: Nobody looks the diff for edit 6 or 7. Four people have checked edits 1 and 4. Two people have checked edit 2. One person has checked edits 3 and 5. If you wrote it out in order, editors looked at diffs 1111223445XX.
Imagine that we instead said:
  • Alice looks at Special:RecentChanges and looks at the diffs for edits 1, 2, 3, 4, and 5. She marks edits 1, 3, 4 and 5 as okay, but isn't sure about edit 2.
  • Bob looks at Special:RecentChanges and looks at the diffs for edits 1 and 4, just like he always did.
  • Chris looks at Special:RecentChanges, turns on the filter so they can see what others haven't already accepted, and looks at the diffs for edits 2, 6, and 7. Chris marks edit 7 as okay, but isn't sure about edits 2 or 6.
  • David looks at Special:RecentChanges, turns on the filter so he can see what others haven't already accepted, and looks at the diffs for edit 2 and 6. David – who wouldn't have looked at either of these diffs under the previous system – is able to resolve the question about edit 2 but leaves edit 6 in the queue for another editor.
In this scenario:
  1. Any editor can still look at whatever they want, like Bob did.
  2. Any editor can voluntarily focus on the un-patrolled edits, like Chris and David did.
  3. The result is that more people look at the uncertain diffs, and somewhat fewer editors look at the obviously good edits. If you wrote it out in order, editors looked at diffs 1122234456667.
What I'm not hearing from you is any reason to believe that requiring all editors to randomly checking RecentChanges is a better system (as measured, e.g., in the percentage of diffs that get looked at by an experienced editor) than having the old system plus allowing some editors (i.e., those who want to) to check specifically the ones that other editors haven't indicated is okay. Adding this system takes nothing away from you. Why do you want to take away the opportunity to use a different system from the editors who want to use it? WhatamIdoing (talk) 17:42, 16 October 2023 (UTC)Reply[reply]
I was going to say almost the same. We like to think that every change gets 'seen' by someone, but in reality I'm observing a lot of propaganda, wp:refspam, false statements, original research (...) being added and never reverted. This system will raise some red flags for those edits, especially as they get closer to the 30 day limit. Ponor (talk) 18:28, 16 October 2023 (UTC)Reply[reply]
Nothing you added changes my point. It isn't, and has never been, about what editors can check and again it has absolutely nothing to do with what I can check.
As you have just said the purpose is for editors to not have to check obviously good edits. But that is just a backwards way of saying editors won't check edits based on the reliablity of other editors, and is directly in opposition to how things should be done.
Nothing you have said, nor anything anyone else has said in support of the proposal, changes that this change would have a large negative effect on the project. -- LCU ActivelyDisinterested transmissions °co-ords° 18:29, 16 October 2023 (UTC)Reply[reply]
Or as an exaple.
1) Editor A marks a edit as approved
2) Editor B sees that editor A has marked as approved and doesn't check it.
3). FAIL -- LCU ActivelyDisinterested transmissions °co-ords° 18:31, 16 October 2023 (UTC)Reply[reply]
That could only happen if everyone is using it (I rate this as being exceedingly unlikely), and it could only make things worse if we additionally assume that Editor B does nothing (e.g., reviewing other edits and therefore finding other problems; I rate this as unlikely).
But again: Why should you get to effectively ban other editors from using this tool? Nobody's going to force you to use it, but why shouldn't other editors be permitted to do so, if they wanted to? WhatamIdoing (talk) 18:47, 16 October 2023 (UTC)Reply[reply]
Other editors shouldn't use it because using it would have a negative impact on the project, for the reasons I have outlined (a few times). This is again, for the second time, nothing to do with my using it or not. Please stop trying to twist my words in that way.
The situation in my last comments happens every time an editor doesn't check an edit because another editor they trust "approved" it. -- LCU ActivelyDisinterested transmissions °co-ords° 19:10, 16 October 2023 (UTC)Reply[reply]
The situation in your last comment happens only if no editors check an edit because another editor they trust "approved" it. It does not happen when some editors move on to other, unscreened edits.
This is already being done by anti-vandalism tools such Huggle, and the situation you hypothesize is already not happening. Why do you think that having one more tool doing this will create your situation, when your situation has not materialized through the tools that are already doing this? WhatamIdoing (talk) 21:03, 16 October 2023 (UTC)Reply[reply]
Some or most makes absolutely no difference, as I have already stated about whether the right is given to a large group or a small one. Also the suggest 2000+ edits for an editor in good standing is not a small group, it would just be the new watermark to reach for POV pushers.
This is not done by current tools, or at least to my knowledge, at the moment edits are highlighted as potentially bad by an algorithm. This would mark edits as "good" in the opinion of an editor, the inverse of the current situation.
I've shown using your own example that there is a problem. -- LCU ActivelyDisinterested transmissions °co-ords° 21:36, 16 October 2023 (UTC)Reply[reply]
This is done by existing tools. Some of them, including Wikipedia:Huggle, set up a feed with the idea that it's better to check all edits once than to check some of them multiple times and some of them never.
This green button is used in Huggle to mark an edit as a "Good Edit".
If the edit is cleared by one Huggle user, then it's not shown to other Huggle users. See mw:Manual:Huggle/Quick start#Controls, especially the green icon associated with the words "Clicking on the green pencil with the checkmark will flag the edit as a good edit."
You don't have the user right necessary to use Huggle, so I'm sure you've never personally seen it, but it's still happening whether you've seen it or not. WhatamIdoing (talk) 01:38, 17 October 2023 (UTC)Reply[reply]
I don't care for such things, due to the many issues such things cause. If this is already happening it's not something we should encourage, as per my previous statements. Making a bad situation worse, doesn't make it better. -- LCU ActivelyDisinterested transmissions °co-ords° 12:32, 17 October 2023 (UTC)Reply[reply]
  • it's been happening for years, and
  • you believe it causes many issues, but
  • you can't point to a single example of any issue it's ever caused.
Okay. WhatamIdoing (talk) 17:52, 18 October 2023 (UTC)Reply[reply]
Or as an example. (1) Editor A marks a edit as approved (2) Editor B sees that editor A has marked as approved and doesn't check it. (3). FAIL
The issue with this reasoning is that patrol actions are logged, and you can easily tell who marked each change as patrolled, quoting Ponor. In other words, there's accountability. Make a few mistakes, fine (page watchers probably couldn't have caught it either), you'll get some feedback and learn. Make consistent mistakes and your RC patrol right is revoked. DFlhb (talk) 23:24, 16 October 2023 (UTC)Reply[reply]
So great we now have to have watchers for the watchers. How about instead don't have such a system, and all editors are watchers of all edits. I made a similar point in my original statement. -- LCU ActivelyDisinterested transmissions °co-ords° 12:34, 17 October 2023 (UTC)Reply[reply]
You don't need to watch the watchers. No one in other wikis do that, there are many ways to surface incorrect patrol including highlights, people who don't use the edit patrolling, etc. Ladsgroupoverleg 14:08, 17 October 2023 (UTC)Reply[reply]
That other wikis don't do something I feel they should do is the epitome of OTHERSTUFFEXISTS. -- LCU ActivelyDisinterested transmissions °co-ords° 18:28, 17 October 2023 (UTC)Reply[reply]
In you list of checks, 1122234456667. The problems is that 3 and 5 could be the edits with issues, but because everyone takes Alive as a reliable source they are not checked again. This change creates an environment where editors are trained to think in ways that the opposite of how they should think about the situation. -- LCU ActivelyDisinterested transmissions °co-ords° 18:41, 16 October 2023 (UTC)Reply[reply]
Sorry, I should have specified from the beginning, from my viewpoint as the omniscient narrator that edits 2 and 6 were the suspicious edits. WhatamIdoing (talk) 18:49, 16 October 2023 (UTC)Reply[reply]
So yes, your suppositions are correct if you control all the variables. But life is not like that. In a real world situation were you are not the "omniscient narrator" edits 3 and 5 could be the edits with problems, and your system would mean they are not checked when they should be. -- LCU ActivelyDisinterested transmissions °co-ords° 19:05, 16 October 2023 (UTC)Reply[reply]
Edits 3 and 5 weren't checked in the existing system, either. WhatamIdoing (talk) 01:23, 17 October 2023 (UTC)Reply[reply]
Maybe, maybe not. But this system ensures they are not, because it's a system that trains editors to not check such edits. -- LCU ActivelyDisinterested transmissions °co-ords° 12:35, 17 October 2023 (UTC)Reply[reply]
At most, it would "train" the ~9,000 of admins and editors who have rollbacker rights to divide the workload rationally. The rest of us – more than 100,000 editors each month – wouldn't be able to use it. WhatamIdoing (talk) 14:53, 17 October 2023 (UTC)Reply[reply]
It would train them to do something I have repeatedly said I feel is a negative the project. That you don't have the same opinion is obvious, that you don't seem able to take onboard that I disargree is verging of IDLT territory. -- LCU ActivelyDisinterested transmissions °co-ords° 18:27, 17 October 2023 (UTC)Reply[reply]
I have no idea why my opposition in particular has gained so many replies, but I've had enough of it. I won't be replying to any further comments. -- LCU ActivelyDisinterested transmissions °co-ords° 18:30, 17 October 2023 (UTC)Reply[reply]
  • Oppose per EspressoAddict and ActivelyDisinterested. Quis custodiet ipsos custodes? Edward-Woodrowtalk 20:41, 11 October 2023 (UTC)Reply[reply]
  • Support. I'm not sure how other wikis use it, but I don't see a disadvantage to keeping track of which pages have been checked at least once, whether this is applied to all pages, a specific subset or as needed. If people want to, they can continue to check the feed of all edits, it's not like turning this on would disable to normal recent changes feed. Patrols don't have to always be good, as long as they're correct more than 50% of the time, that's potentially useful information. I don't see the merit of not collecting information for fear of people using it badly in this specific case. Maybe it would be better if instead of defaulting to everything needing to be checked, everything gets automatically patrolled but RCPers can unpatrol things they find suspicous, I know there's a "suspicous edit" button in Huggle that I don't use very much (when I actually do RCP, which isn't much these days), but I'd say the best way to see how well it'll work is to enable it as a trial. If nothing else, having a log of patrol actions would help collect data on RCP work (or some subset of it), even if the data isn't directly used on-wiki at all. Alpha3031 (tc) 14:56, 14 October 2023 (UTC)Reply[reply]
  • Support. I don't understand how you work here without this bookkeeping system, but I do see how and why some false statements I discovered over the years found their way into enwiki and stayed for a long time: I myself sometimes think someone else will take care of the things, but do they? By enabling this extension, you have nothing to lose. I've patrolled thousands of edits on another wiki, even made it easier to patrol from recent changes, page history, and user contributions pages by using two scripts for "inline" patrolling (much easier than that floating window, Certes!) from this proposal (check animated pics to see how patrolling works). As for who should patrol: all advanced rights users, any additional "trusted" users... as I said, this is just another bookkeeping system, let those who want to use it - use it. Nothing will change for those who don't. Ponor (talk) 02:59, 15 October 2023 (UTC)Reply[reply]
  • Support because it's an optional system that can be used by those that want to use it and ignored by those who don't want to be bothered with it. Ladsgroup, I suggest talking to the maintainers of the Wikipedia:Cleaning up vandalism/Tools. There's no reason to require two clicks. People like Remagoxer (on the RedWarn/UltraViolet team) would probably like to know about this, assuming it gets implemented. WhatamIdoing (talk) 17:51, 16 October 2023 (UTC)Reply[reply]
  • Support. This seems great; it enables optional 'coordination' when scrutinizing the watchlist. Even though it's not perfect, I think it's valuable to reframe: fighting against bad edits is best done through defense in depth. ORES, page watchers, people who check recentchanges, this new RCpatrol, anything that helps, by any amount, is good to have in the toolbox. Fully optional, too - DFlhb (talk) 00:02, 17 October 2023 (UTC)Reply[reply]
  • Support as a great idea – easy way to improve our accuracy. As it stands an immense amount of unsourced and incorrect additions get through the system, and by spreading out who checks what more evenly the percentage of bad edits getting through will dramatically lessen. Even requiring two editors to check every good edit would IMO improve efficiency. J947edits 03:43, 17 October 2023 (UTC)Reply[reply]
  • Oppose as not a great idea ~ per ActivelyDisinterested and EspressoAddict. Obvious vandalism is easy to spot and is then reverted; less obvious stuff benefits from as many eyes as we give it. I fail to see the benefit. Happy days, ~ LindsayHello 14:58, 18 October 2023 (UTC)Reply[reply]
    @LindsayH, but what makes you think it won't get enough eyes? The only thing that's added is a little red exclamation mark in recent changes *for users with patrol rights*, which only they can remove, and they still get to see all edits on their watch list. There's no change whatsoever for nonpatrollers. Ponor (talk) 07:09, 19 October 2023 (UTC)Reply[reply]
  • Support as opt-in. It should be a user right you can apply for, or maybe an opt-in that any rollbacker/maybe pending changes reviewer can choose to enable. Many patrollers won't go through the effort. One of (admittedly, a bunch of) the reasons I don't do much recent change patrolling is that most edits have already been checked by the time I see them and there are plenty more that I'd either mark good or leave purposefully for someone else, but there's no difference between the last two categories and sometimes nobody gets to the ones being left purposefully. Maybe a trial run would be better, to see if it seems to be beneficial or not, but given it's opt-in and many editors seem to have issues with it/won't have the right to use it, edits should get double-checked anyways even if they've been approved. However, given the concerns raised, I think the status of an edit being patrolled or not shouldn't be easily visible on Special:RecentChanges/etc. unless you specifically opt into it being shown (assuming any user would be able to see the status, just not change it), so that newer users get the current interface and aren't fed into the pool of people who rely on patrol status, as that seems like the most likely way for it to backfire (everyone starts opted-in and then we over-rely on patrols being accurate). Skarmory (talk • contribs) 21:14, 23 October 2023 (UTC)Reply[reply]

Discussion (Enable RC patrolling (aka patrolling for edits))[edit]

Could someone comment on the other large wikis that the proposer is referencing? Do they have a large editor base that is similar to en-wiki? The smaller wikis of my acquaintance tend to have sufficiently few highly active editors that triaging patrolling by the "have I heard of this editor" method is a rational strategy.

Also, could an admin active in permissions comment on how much scrutiny is given to rollbackers? My guess is not enough to grant this kind of right, but I don't work in this area. Espresso Addict (talk) 00:45, 12 October 2023 (UTC)Reply[reply]

They are mentioned in the proposal Ladsgroupoverleg 10:22, 12 October 2023 (UTC)Reply[reply]

Two notes:

  • The patrol logs can be used to improve ML models of vandalism highlights and revertbots. I already used the patrol logs to build the revert bot in my home wiki. I can see this could be used to improve general AI models.
  • I feel there is a bit of misunderstanding on how this works, possibly a trial period to show its pros and cons and then we can decide? Ladsgroupoverleg 20:05, 15 October 2023 (UTC)Reply[reply]
@Ladsgroup, have you talked to your colleagues in Tech, just to make sure that the servers won't fall over if this is enabled here? WhatamIdoing (talk) 00:54, 16 October 2023 (UTC)Reply[reply]
@WhatamIdoing yeah, specially since this is enabled in wikidata (which has a massive flood of edits) which caused issues in 2017 and I overhauled how the data is stored around that time and now it's quite healthy. Ladsgroupoverleg 12:11, 16 October 2023 (UTC)Reply[reply]

What is the proposed workflow for a bad edit? If User:Vandal serves up spam and User:Helpful reverts the addition, is the first edit marked as patrolled because it's been assessed, or left unmarked because it wasn't approved, or marked in some other way as bad but dealt with? What if the first edit isn't marked "reverted"? (Perhaps the second edit also fixes a typo, or it keeps part of the first edit because it mixed useful information with a spammy citation.) Certes (talk) 11:14, 18 October 2023 (UTC)Reply[reply]

@Certes If it's a rollback, it automatically gets marked as patrolled by the software itself. I think that is also the case by manual revert but not 100% sure. But conceptually a reverted edit is a patrolled one since there is no further action is needed. Ladsgroupoverleg 14:05, 18 October 2023 (UTC)Reply[reply]

Should this be mentioned in CENT or if there is a wikiproject for patrolling (I couldn't find one), notify its members? Ladsgroupoverleg 14:22, 18 October 2023 (UTC)Reply[reply]

The most relevant one would probably be WP:CVU. Alpha3031 (tc) 22:32, 18 October 2023 (UTC)Reply[reply]
done Ladsgroupoverleg 14:44, 20 October 2023 (UTC)Reply[reply]

the bot tried to archive this, I rather make sure it gets closed first. Ladsgroupoverleg 11:25, 2 November 2023 (UTC)Reply[reply]

Another note to avoid the bot archiving it Ladsgroupoverleg 11:28, 13 November 2023 (UTC)Reply[reply]
You could use {{do not archive until}}. WhatamIdoing (talk) 17:13, 13 November 2023 (UTC)Reply[reply]
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.

Post-close (RC patrolling RFC)[edit]

The lack of participation and consensus is really too bad. This seems like a proposal best tested on a trial basis, so the claims of supporters and opponents can be experienced instead of conjectured. Would this be worth proposing again, explicitly as a trial? DFlhb (talk) 00:41, 30 November 2023 (UTC)Reply[reply]

I don't think many people will join in if there is only discussion, because experience shows that these patrol proposals will come to nothing anyway.
The technical implementation of the patrol system is also not particularly convincing. It would be wrong to claim that everyone on e.g. Commons would be happy with the existing patrol feature enabled. Killarnee (talk) 19:42, 2 December 2023 (UTC)Reply[reply]
I think this RfC could've gone better if the technical details were clarified (e.g. how this will affect new page patrol) and if it was advertised on CENT. Galobtter (talk) 20:16, 2 December 2023 (UTC)Reply[reply]

Scoring for Wikipedia type Articles Generated by LLM[edit]

Our research team of students are building a LLM-based system which can generate a full-length Wikipedia page for a given topic without the need for supplemental information (e.g., human written outlines, curated references, etc.). Besides automatic evaluation, we would like to have frequent wikipedia editors collaborate with scoring the articles and providing feedback. Due to GDPR rules we will be limiting participation to those in the USA. Our goal is only for educational research, and we are not intending to try to publish these LLM generated articles on Wikipedia. Our LLM will ideally generate Wikipedia style articles with citations, and different sub-points. We will be scoring the essay based on 1. Well Written, 2. Verifiable with no original research, 3. Broad in its coverage, and 4. Qualitative comments (The first three metrics for a Good Article + Qualitative comments). We would take a subset of our articles produced and score them by actual Wikipedia editors as a way to verify our scoring is within reason.

Please fill out this form if interested and we will send you a consent form in compliance with IRB standards. Link[1]

Below is our ethical statement.

In this work, we study the automatic Wikipedia generation problem as a way to push the frontier of automatic expository writing and automatic knowledge curation. All the studies and the evaluation in this work are designed to prevent the dissemination of misinformation by not publishing generated content online and implementing strict accuracy checks. We avoid any disruption to Wikipedia or related communities, as our system does not interact with live pages. Also, although we try to generate grounded articles, we believe there is no privacy issue related to this work as we only use information publicly available on the Internet.

The primary risk of our work is that the Wikipedia articles written by our system are grounded on information on the Internet which may contain some biased or discriminative contents. Currently, our system relies on the search engine to retrieve high-quality information but does not include any post-processing module. We believe improving the retrieval module to have good coverage of different viewpoints and adding a content sifting module to the current system will be a critical next step to achieve better neutrality and balance in the generated articles. In our experiment, we manually go through all the topics in the test set to ensure the topics themselves are not biased or discriminative.

Another limitation we see from an ethical point of view is that we only consider writing English Wikipedia articles in this work. Extending the current system to a multilingual setup is a meaningful direction for future work as there are more interesting topics that do not have their Wikipedia pages in non-English languages. Terribilis11 (talk) 01:19, 10 November 2023 (UTC)Reply[reply]

Who is we? What will you be doing with participants’ personal data that would not be allowed by GDPR? Barnards.tar.gz (talk) 08:02, 10 November 2023 (UTC)Reply[reply]
Based off of the google link above, "we" appears to be some team at Stanford. Some Stanford people have done work on LLMs and Wikipedia before (see: Wikichat). Should OP not respond, it may be worth reaching out to the WikiChat folks to figure out what the team is behind this/what the IRB approval looks like for this. — Red-tailed hawk (nest) 03:52, 12 November 2023 (UTC)Reply[reply]
To be frank we aren't doing anything with participants' personal data. However, we just aren't familiar with the standards of GDPR and are choosing to avoid the issue as a whole. If there are many Wikipedia collaborates interested in participating in the EU we can evaluate the GDPR more closely to determine if we are in compliance. Terribilis11 (talk) 00:37, 14 November 2023 (UTC)Reply[reply]
@Barnards.tar.gz @Red-tailed hawk Given feedback I have received, the large percentage of European editors, I believe we will be confirming compliance with GDPR shortly. Terribilis11 (talk) 00:48, 14 November 2023 (UTC)Reply[reply]
Hello we are students at Stanford working on a research project. Terribilis11 (talk) 00:37, 14 November 2023 (UTC)Reply[reply]
If there is no current English Wikipedia policy that prevents such exploitation of this project, there should be. There is no consensus for use of LLM on this project for any purpose that changes content here. Your statement is a very poor introduction to Wikipedia. I oppose any such endeavor. — Neonorange (talk to Phil) (he, they) 18:09, 10 November 2023 (UTC) —Reply[reply]
@Neonorange, could you expand on your objection a bit? The proposal states that none of the generated pseudo-articles will be published and "does not interact with live pages". Schazjmd (talk) 18:20, 10 November 2023 (UTC)Reply[reply]
I oppose any use of LLM that might make it easier for any LLM generated material to appear in this project. This is a consensus project where all content is generated by volunteers. And where all error and vandalism are remedied by volunteers. I consider your project as injurious to Wikipedia and exploitative of our volunteers. Ethical and empathetic AI is still twenty years off. — Neonorange (talk to Phil) (he, they) 18:31, 10 November 2023 (UTC) —Reply[reply]
They are free to create this large language model because there is a CC-BY-SA 4.0 license for content here. You may think of it however you like but they are allowed to do that.
Whether we use it depends on how well they do it. For now ChatGPT sucks to write Wikipedia, particularly if you are less experienced, but this may change later. Training LLMs on Wikipedia only is bound to fail because we've got so much crap here that sieving through it would probably take about as much time as fixing it.
I wonder what is the GDPR data they are talking of? Are you only going to deploy this tool for non-EU editors? Szmenderowiecki (talk) 20:25, 10 November 2023 (UTC)Reply[reply]
@Neonorange, what's your overall view? "Nobody should do research on LLMs, because that 'might make it easier for any LLM generated material to appear in this project'"?
If you're concerned about them posting it, they already said we are not intending to try to publish these LLM generated articles on Wikipedia. I'm assuming you missed that, because otherwise your comments sound like you telling other people what they're allowed to do on their own computers and with their own time. WhatamIdoing (talk) 03:24, 12 November 2023 (UTC)Reply[reply]
To reiterate we will not be posting this articles on wikipedia or any other website as a news soruce. Rather we are simply pursuing a model that produces models of similar quality to Wikipedia. We will be using our own scoring mechanism, but feel that to best determine the accuracy of our scoring we would like ot compare it with scoring by actual wikipedia editors. Terribilis11 (talk) 00:35, 14 November 2023 (UTC)Reply[reply]
All of the LLMs are already trained on our content [2]. The main issues here are covered by SnowFire below: review of this AI output is unlikely to demonstrate that it doesn't hallucinate because that requires WP:FAC-level examination by subject-matter experts, and WP needs its volunteers to remain focused on doing this at FAC and WP:GAN, not helping some off-site AI project.  — SMcCandlish ¢ 😼  19:29, 20 November 2023 (UTC)Reply[reply]
Hi @SMcCandlish We have found the feedback here to be very useful. As such we are going to be creating a UI that will have the article with in place citations. In addition clicking on a citation should bring up the website pointing to the exact location in article our model pulled from for the generation. We believe this will make it both efficient and more practical to determine what is hallucinated and what is supported.
Of course I hear your opinion on what volunteers should focus on, but we are of course going to honor their autonomy in this matter. Terribilis11 (talk) 08:30, 21 November 2023 (UTC)Reply[reply]
You should probably read WP:LLM before going any further. -- LCU ActivelyDisinterested transmissions °co-ords° 20:48, 10 November 2023 (UTC)Reply[reply]
Thank you for the suggestion. Terribilis11 (talk) 00:49, 14 November 2023 (UTC)Reply[reply]
To clarify, that page is an WP:ESSAY and does not reflect community consensus - in fact, a proposal to elevate it to policy status was explicitly rejected by the community just recently. That said, it is probably prudent to stick with your plan of not posting these LLM-generated articles on Wikipedia (at least not in the mainspace, i.e. as actual articles). Regards, HaeB (talk) 03:14, 14 November 2023 (UTC)Reply[reply]
  • The fundamental problem here is that it is usually easier and quicker (and certainly more enjoyable) to write an article from scratch than to carefully review one created by anyone whose competence is questioned, such as an AI. Espresso Addict (talk) 00:45, 11 November 2023 (UTC)Reply[reply]
    Yup. @Terribilis11, it can take hours to review a single article under the Wikipedia:Good article criteria. For some articles, it can take 30 minutes just to read it – and that's without doing things like figuring out whether the sources exist, are reliable, and WP:Directly support the content they've been associated with. You should probably be budgeting for one or two hours per article. WhatamIdoing (talk) 03:48, 12 November 2023 (UTC)Reply[reply]
    Okay, this is valuable feedback! I will take this back to the team. Terribilis11 (talk) 00:42, 14 November 2023 (UTC)Reply[reply]
    This is my main concern, too. I'm happy to review an article written from the three best sources, but if one of the metrics is 'broad in its coverage' does that mean the project will attempt to exhaust the sources and could end up with twenty or eighty or more, some of which may include multiple pages in books? Valereee (talk) 14:05, 12 November 2023 (UTC)Reply[reply]
    Thank you very much for this input! This is great feedback. We will take this into account. In general most citations will be web-based. Regardless, I will take this input back to our team. Terribilis11 (talk) 00:41, 14 November 2023 (UTC)Reply[reply]
@Terribilis11: As a general practice, it is recommended to start a documentation page for your research project at meta:Research:Projects. No need to get super detailed or fill out every field in the "add your project" form - you can probably mostly reuse the text from your message(s) above. That page can then also serve as a reference point later, instead of the post and discussion here which will get archived soon. Good luck with the project, and once you publish something about it (even if in preprint form), feel free to ping me and my @wikiresearch collaborators, we would be happy to help publicize it. (E.g. here is our coverage of the "WikiChat" effort mentioned above, which made it to the front page of Hacker News.) Regards, HaeB (talk) 03:24, 14 November 2023 (UTC)Reply[reply]
  • Comment. score them by actual Wikipedia editors - I would caution accepting just any volunteers. You can get some baseline feedback on 1, 3, and 4, but 2 ("Verifiable with no original research") requires either subject matter experts or somebody with the time who's willing to dive into all the citations and verify them. The worst outcome - and by far the scariest one for LLMs - is output that looks right but is in fact wrong, and "casual" volunteers giving a 5-minute skim will miss these cases. And I can assure you that editor time for that sort of a deep-dive is quite limited - it's hard enough getting people to do a citation-by-citation level verification at places like WP:FAC, and there at least the work will have "meant" something. IMO, any volunteer willing to do such deep, untrusting, total verification for a machine-generated article should probably spend that kind of valuable time over at WP:GAN instead. Basically, if you really want to verify #2, I suspect that actually paying vetted researchers for their time may be a better option. This is especially true if it turns out that your LLM really is an improvement - saying a LLM doesn't hallucinate falsehoods is cheap, but verifying it is very hard. SnowFire (talk) 23:12, 17 November 2023 (UTC)Reply[reply]
    @SnowFire Thanks for the feedback. You make some great points on the dangers of LLMs. We are part of Stanford's Open Virtual Assistant Lab [3] which is focused primarily on the grounding of LLMs. i.e. eliminating hallucination. We feel for the scale of our current project, Wikipedia editors are the best option available, although you are right that finding topic specific experts would be ideal, this would unfortunately be an intractable challenge for a model that should be functional in a wide variety of topics. If you have more thoughts, it could be helpful to checkout our research page, we recently created due to suggestions in this thread. [4] Terribilis11 (talk) 09:36, 21 November 2023 (UTC)Reply[reply]
    @Terribilis11: Thanks for the reply. To be clear, and I hope this doesn't come across as nitpicky, but my post wasn't about "the dangers of LLMs", which I'm sure we're both familiar with. Rather, it was a post about how difficult and expensive it is to detect hallucinations, especially subtle ones. Imagine a disease that sometimes results in obvious symptoms, but sometimes is subtle and hard to detect without a trained doctor who knows what to look for, or a nurse with a lot of time on their hands. Now imagine a paper is released saying that a special new treatment reduces the disease 50%, but with evidence being random volunteers who looked over patients for an unknown amount of time. This won't be convincing, except that it perhaps reduced the overt cases of the disease by 50%. It's even more dangerous if you use the feedback to help guide the model adjustments - you might be being mislead by adjustments that just make hallucinations more subtle.
    If there isn't budget to hire Real Experts to check, I would humbly suggest to have your model generate text on topics that people on the project itself know very well, and that way you can perhaps self-verify in the "middle" of your experiment, and save the budget for a few independent verifications at the end. SnowFire (talk) 20:54, 21 November 2023 (UTC)Reply[reply]
    You're absolutely right the problem of hallucination is quite difficult. I imagine that's why there are already dedicated companies for the same issue when checking politician's statements. We are taking a very structured approach and are giving it our full attention. Thank you for the suggestion. Terribilis11 (talk) 03:46, 24 November 2023 (UTC)Reply[reply]
  • What SnowFire said. All of it.  — SMcCandlish ¢ 😼  19:29, 20 November 2023 (UTC)Reply[reply]
    My bad-science alarm has gone off. This project needs to decide whether the intent is to create a LLM that writes accurately and reliably in the style of Wikipedia, or whether it's to create a LLM that writes articles that will pass AfC. These are totally different aims. Wikipedia defines itself as an unreliable source because we do not believe our editors get it right all the time. Therefore if you want to design a LLM that is accurate you cannot use Wikipedian volunteers, as they are, by definition, no more accurate than Wikipedia itself. If you want to create a LLM that passes AfC then using Wikipedians is a more reasonable choice, but you need to make sure your volunteers are an accurate match to those at AfC. Vast numbers of articles written by Wikipedia volunteers are rejected at AfC, so you might not want those volunteers to be assessing the products of your model! Do you have a strategy to select volunteers to match AfC? On accuracy, you should also be thinking that Wikipedia's accuracy (or not) stems not only from a careful review by an AfC person, but from the fact that articles are typically read by huge numbers of people, including subject experts, from across the globe. There is no guarantee that a panel of even 50 selected volunteer Wikipedians will give the same accuracy as this panel of an unknown number of casual readers, many with prior, in-depth knowledge. I'm not saying that LLM's can't write accurate articles, or write AfC-passing articles, but I'm saying there are lots of reasons why a model trained by Wikipedian volunteers might end up doing neither. Elemimele (talk) 17:32, 28 November 2023 (UTC)Reply[reply]
    It sounds like they are trying to build an LLM that could pass GA. WhatamIdoing (talk) 05:46, 3 December 2023 (UTC)Reply[reply]
    @Elemimele We want to design a LLM that is as accurate as Wikipedia, because as you pointed out accuracy is a very complicated metric. It is also not with out disagreement on what is and isn't accurate, we need merely look at American politics which frequently argue over "fact". As such, our research hopes to make clear the limitations of our results. They are verified by our own metrics as well as Wikipedians. Our model isn't being trained by Wikipedian volunteers, rather the results of our model's scoring will be compared to the Wikipedian volunteers scoring. This will give a frame of reference to the results of our model. We aren't going to be making unwarranted claims of complete accuracy as that would certainly be bad-science. Our results will be in terms of our own coring and the Wikipedians, to hopefully give an idea of what level of verifiability and hallucination our model is demonstrating. Obviously if we were able to tap into the hundreds of experts that edit Wikipedia articles to give us hope of high accuracy, we would love to do that. This is unfortunately not feasible in this case. Terribilis11 (talk) 23:07, 3 December 2023 (UTC)Reply[reply]

Reduce WP:ADVOCACY in Module:Find sources/templates/Find sources[edit]

Module:Find sources/templates/Find sources currently contains WP:ADVOCACY for Google/Alphabet Inc, one of the world's most powerful and monopolistic, highly criticised corporations that pays 10s of billions of dollars to try to maintain a monopoly with its criticised search engine. This advocacy is contrary to our mission. Moreover, this advocacy violates the spirit of the Universal Code of Conduct in the sense that it encourages Wikipedians to disclose their personal data, and in some cases puts us at risk of harm. What editorial changes should we make in the module to reduce the level of advocacy, without favouring one particular attempted monopoly and without harming Wikipedians' privacy rights, while helping readers find reliable, independent sources using notable search engines? Boud (talk) 16:44, 10 November 2023 (UTC)Reply[reply]

  • Proposed discussion structure: Please add Support or Oppose and arguments for or against the following specific proposals in the sections below, or add additional proposals, so that someone uninvolved can summarise the consensus after a reasonable period of debate.
  1. tech step 1: Add missing engines to the list of available search engines at Module:Find_sources/links, for engines that catalogue high numbers of pages, according to Comparison of web search engines, or are notable scholarly search engines:
    1. Add Qwant
    2. Add Mojeek
    3. Add Semantic Scholar
  2. tech step 2: Add some metasearch engines that respect privacy to the list of available search engines at Module:Find_sources/links
    1. Add
    2. Add one or 3-4 Searx instances such as
  3. policy step 1: modify Module:Find sources/templates/Find sources that defines the list of search engines at Template:Find general sources
    1. Replace Find sources: by Use a [[List of search engines|search engine]] ([[Comparison of web search engines|comparison]]) to find sources: (which renders as: Use a search engine (comparison) to find sources:)
    2. remove the general link to Google and replace it by Startpage, Qwant and/or Mojeek and/or [add a proposal]
    3. keep Google Books unless someone can propose a similar equivalent
    4. replace Google News by a link to the Startpage or news link (needs tech solution for direct URL) and/or [add a proposal]
    5. replace Google Scholar by Internet Archive Scholar and/or Semantic Scholar and/or [add a proposal]
    6. (Other modifications)
  4. (Other)

Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

tech step 1: Add missing engines[edit]

Add missing engines that catalogue high numbers of pages, according to Comparison of web search engines, or are notable scholarly search engines. The following should be added to the list of available search engines at Module:Find_sources/links (support/oppose NAME(s) or NUMBER(s) + evidence + reasons):

Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

  • Support Mojeek + Semantic Scholar (proposer) - Both are Wikipedia-notable, reasonably well-established over several years, and no known privacy concerns are listed in their Wikipdia articles. Qwant is more controversial, and despite claiming to protect privacy, there are RSed concerns about privacy. On the other hand, this point is only about adding these search engines to the list available for templates to use, so maybe it should be added anyway. I may update on Qwant after seeing other evidence for/against. Boud (talk) 16:53, 10 November 2023 (UTC)Reply[reply]

tech step 2: Add some metasearch engines that respect privacy[edit]

The following should be added to the list of available search engines at Module:Find_sources/links (support/oppose NAME(s) or NUMBER(s) + evidence + reasons):

Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

  • Support Startpage + a Searx instance. (proposer) Per the sources in, Startpage has been around for many years, it claims to protect privacy, and there don't appear to be any sources contesting the privacy claims. Startpage does do contextual advertising, so it's not ad-free, but it does not do targeted advertising, so that makes it a lot closer to the spirit of UCOC than Google. Searx: the Searx software package is mature, stable software (available in Debian in the old-old-stable, old-stable, stable and unstable distributions). The practical tech question is whether (i) to list one or a few specific Searx instances (such as, or (ii) a list of instances, or (iii) whether Wikimedia community techies should install our own Searx instance. I would tend to go for (i) or (ii), at least in the short term, since (iii) would require waiting for the techie work/server resources to be done. Boud (talk) 17:04, 10 November 2023 (UTC)Reply[reply]

policy step: Modifications to Module:Find sources/templates/Find sources[edit]

Modify Module:Find sources/templates/Find sources that defines the list of search engines that is used in Template:Find general sources in the following ways (support/oppose + evidence + reasons). Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

3.1: Replace 'Find sources'[edit]

Replace Find sources: by Use a [[List of search engines|search engine]] ([[Comparison of web search engines|comparison]]) to find sources: (which renders as: Use a search engine (comparison) to find sources:) Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

  • Support (proposer). This would avoid advocacy for any particular search engine and would allow 3.2 to remove all specific search engines. Boud (talk) 17:07, 10 November 2023 (UTC)Reply[reply]

3.2: Replace the generic Google link[edit]

Remove the general link to Google and replace it by Startpage, Qwant and/or Mojeek and/or [add a search engine]. Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

  • Support removal of Google, with no replacement (proposer). Along with point 3.1, which would link to our NPOV, non-advocating list of search engines, removing Google, with no replacement, would satisfy our desire to help the reader find sources while also avoiding advocacy. Boud (talk) 17:10, 10 November 2023 (UTC)Reply[reply]

3.3: Keep Google Books or propose alternative[edit]

Keep Google Books unless someone can propose a similar equivalent. Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

3.4: Replace Google News[edit]

Replace Google News by a link to the Startpage or news link and/or [add a news search engine]. Startpage and Searx may need extra tech work since they don't display direct URLs to their news sections. Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

  • Support Startpage and/or Searx (proposer) The influence of any single organisation in the selection of news stories is a particularly sensitive question in terms of biasing Wikipedia content. Startpage and Searx are both meta-search engines, so they will still be affected by Google's selection biases, but they each have their own mixes of sources. Since they protect users' privacy, either (or both) would be better than Google for news. Boud (talk) 17:14, 10 November 2023 (UTC)Reply[reply]

3.5: Replace Google Scholar[edit]

Replace Google Scholar by Internet Archive Scholar and/or Semantic Scholar and/or [add a scholarly search engine]. Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

  • Comment (proposer) I personally use the Science tab in a Searx instance, but I've proposed several links to Searx above, and having too many links to a Searx instance would be as risky in terms of biasing our source selection as having too many links to Google. Boud (talk) 17:45, 10 November 2023 (UTC)Reply[reply]

3.6: Other modifications[edit]

Other changes to the current content of Module:Find sources/templates/Find sources and its template form at Template:Find general sources could be proposed. However, for NYT vs AP, please go to the orthogonal RfC at Wikipedia:Village pump (proposals)#RfC on Module:Find sources - replace New York Times with Associated Press. Boud (talk) 16:45, 10 November 2023 (UTC)Reply[reply]

4 Other related changes[edit]

As an open-ended component of this RfC, feel free to add other (preferably concrete) proposals or 'see also' links.

  • This is an invalid RfC. From Wikipedia:Requests for comment: Keep the RfC statement (and heading) neutrally worded and short. The opening statement isn't even remotely neutral, and the remainder of the proposal is hopelessly complex. I suggest that it be closed, and instead a simple, brief and neutral RfC be started in its place. AndyTheGrump (talk) 18:58, 10 November 2023 (UTC)Reply[reply]
    Not bad for an RFCbefore though. Could ditch the tag for now and reopen later. Selfstudier (talk) 19:05, 10 November 2023 (UTC)Reply[reply]
    What is non-neutral in the RfC statement? The Wikipedia mission is clear: we are opposed to advocacy. This is not an RfC about whether advocating for a monopoly is advocacy or not. Advertising and advocacy are almost universally considered to be in opposition to Wikipedia's mission. We don't have to have an RfC about whether the sky is blue. The remainder of the proposal gives concrete subquestions on which to work, in order to avoid abstract blabla. There is a minimum of structure in order to have a chance of emergence of consensus on the current text. Wikipedia modules and templates cannot avoid a minimum of structure. That is not hopeless complexity, it is structure. The choices are dictated by the current content of the module. Without structure, we wouldn't have Wikipedia. Boud (talk) 19:32, 10 November 2023 (UTC)Reply[reply]
    Sure, but look at Wikipedia:Village pump (proposals)#Discussion (Find sources: NYT vs AP) above, dead simple but still going. Selfstudier (talk) 19:39, 10 November 2023 (UTC)Reply[reply]
    Boud, I'm not going to argue with you - if you aren't going to close this malformed dogs-breakfast of an RfC yourself, I will find an admin to do it instead. AndyTheGrump (talk) 19:42, 10 November 2023 (UTC)Reply[reply]
    This doesn't look anything like an RFC. And it's actually advocacy of an anti-GOGGLE position. (Although the ADVOCACY essay likely isn't relevant outside of mainspace.) Just remove the RFC tag, although too complex even for a discussion. O3000, Ret. (talk) 19:59, 10 November 2023 (UTC)Reply[reply]

I have removed the RfC tags -- please read through WP:RFC thoroughly (especially its sections WP:RFCBEFORE and WP:RFCBRIEF) before attempting anything like this again. --JBL (talk) 20:02, 10 November 2023 (UTC)Reply[reply]

Shit, you've beaten me to it.
I removed "RfC" from title so that nobody thinks there is an RfC.
Boud, I second the advice directly above, but I also think you can discuss this but not as an RfC but more as a preparation to it. Szmenderowiecki (talk) 20:08, 10 November 2023 (UTC)Reply[reply]

@AndyTheGrump, JayBeeEll, Objective3000, Selfstudier, Szmenderowiecki, and Hobit: The best way that those of you who have suggested that the RfC could be worded in a different way while still being concrete, actionable would be to edit the draft or discuss changes to be made to it. Thanks in advance. :) Boud (talk) 22:34, 11 November 2023 (UTC)Reply[reply]

I think we can discuss it here once we started it and, once we are ready, we can start an RfC in a subsection. It would be better for you because a lot more people will visit this page than your draft. But I will post in your draft anyway because I'm not sure everyone agrees.
And btw, good you've listened. Szmenderowiecki (talk) 22:45, 11 November 2023 (UTC)Reply[reply]
(ec)On the contrary. Given that the draft entirely failed to even approximate to a proper RfC proposal, the best way would be to start discussion again from scratch, working on the premise that there seems to be something of a consensus at Wikipedia talk:Articles for deletion that the present wording needs revision, and that this is best accomplished without engaging in soapboxing about monopolies etc. It seems likely that we can get a general agreement that we should avoid implying that Google is the only option amongst search engines, and that we need to make clear that other options are available in most cases. Given the complexities involved when considering the appropriateness of specific engines in specific circumstances, attempting to find wording for RfC at this stage would seem premature. Start a discussion on the general subject, in a thread that doesn't scream 'advocacy' in the title but instead presents the topic neutrally, and see what people have to say. Few have commented so far, and I suspect that more would do so if they weren't being told in advance how to think... AndyTheGrump (talk) 23:10, 11 November 2023 (UTC)Reply[reply]
A good place to start would be noting that WP:ADVOCACY defines its subject as "promot[ing] a person's or organization's beliefs or agendas at the expense of Wikipedia's goals and core content policies, including verifiability and neutral point of view". I do not see how the recommendation of Google in modules or at AfD is harming Wikipedia's verifiability or NPOV. Could you explain, Boud? ~~ AirshipJungleman29 (talk) 00:48, 12 November 2023 (UTC)Reply[reply]
Don't mean to interrupt a response from Boud. Just thought I would mention that ADVOCACY is an essay, not a PAG. Also, I'm not certain that it applies out of mainspace -- which should perhaps be cleared up in the essay. In any case, good question. O3000, Ret. (talk) 01:12, 12 November 2023 (UTC)Reply[reply]
I personally feel that there is absolutely no reason it should apply out of article/article talk spaces, but perhaps Boud feels differently. ~~ AirshipJungleman29 (talk) 01:19, 12 November 2023 (UTC)Reply[reply]
Google/Alphabet is an organisation that has beliefs and agendas. Its legal aim is to maximise revenue from advertisements. Our verifiability and NPOV depends on where we choose our sources. By promoting Google over other search engines, we are promoting POVs selected by Google as opposed to POVs that other search engines select. The fact that Google/Alphabet paid 10s of billions of dollars to Apple and Samsung in order to convince them to make it difficult for smartphone users to use a search engine other than Google only makes sense if Apple and Samsung consider other search engines to be at least equally good at providing sources that would make their customers happy. Neither Google/Alphabet nor Apple nor Samsung are academic institutions that run transparently, with collegial decision-making university-style or public git repository style about how their algorithms run or how their preferred search engines run, so they are inconsistent with the usual criteria for the search for knowledge. Criticism of Google goes into quite a fair amount of detail of why Google does not neutrally select its sources. Given Google's current power, we are unlikely to completely avoid advocating for Google. But we can reduce our advocacy for either Google or whichever search engines Apple and Samsung presumably would prefer if they weren't paid 10s of billions of dollars. (Even Mozilla is also is paid by Google/Alphabet to make Google its default search engine.)
As for article space vs non-article space, {{Find general sources}} is used on about 868,000 pages; while these are not directly in article space, they affect the content of articles by advising people about where to seek sources. Boud (talk) 01:34, 12 November 2023 (UTC)Reply[reply]
Kinda hate to disillusion you, but Wikipedia:Nobody reads the directions, so sometimes an editor will toss out a few WP:UPPERCASE claims without knowing what they're linking to. WhatamIdoing (talk) 03:10, 12 November 2023 (UTC)Reply[reply]
It's true that the essay WP:ADVOCACY suggests that it's mainly about WP:MAINSPACE advocacy, without saying specifically that it's restricted to article content alone. I've edited the draft to clarify that the advocacy is non-mainspace advocacy. It would be hypocritical to advocate for a particular organisation, especially in a way that biases our selection of sources - which affects our core values of verifiability and NPOV - while claiming that we should not advocate for particular organisations in our articles. Google has an agenda that affects our content. Currently we advocate for Google. Boud (talk) 12:44, 12 November 2023 (UTC)Reply[reply]
Here's a prediction. You carry on with your 'advocacy' soapboxing. People look at your arguments, recognise them as the great-wrong-righting grandstanding they are, and decide not to get involved in pointless diversions when a simple discussion over wording and links without all the waffle would solve things quicker. Nothing gets decided. The issue remains... AndyTheGrump (talk) 13:28, 12 November 2023 (UTC)Reply[reply]
When some people are trying to say that the sky is not blue, it's important not to let a reader of the conversation think that there is consensus that the sky is not blue, but is instead the colour of higher frequency optical light that is inevitably scattered down during the daytime on a cloudless day. But I agree that we need to cut through the distractions. Boud (talk) 18:22, 12 November 2023 (UTC)Reply[reply]
Starting at AfD and continuing here, you have made the following statements:
  • (using Google) puts us at risk of harm
  • one of the world's most powerful and monopolistic, highly criticised corporations that pays 10s of billions of dollars to try to maintain a monopoly with its criticised search engine.
  • it encourages Wikipedians to disclose their personal data
  • would-be totalitarian (though fortunately only authoritarian) organisation
  • Trying to impose that as a unique choice - a totalitarian choice - on the world
  • Text presents a totalitarian point of view - "Thou shalt worship no other search engine than Google!".
  • Google's would-be totalitarian nature
  • "Authoritarian" and "would-be totalitarian" are accurate adjectives in this case.
You have used the words totalitarian and authoritarian seven times each. It sounds like it is you who have an agenda – an anti-Google agenda. Look, if it makes sense to add to the Find Sources template, why not? But, it should be based upon our evaluations of their usefulness, not on some personal dislikes about the source's corporate activities (which I also despise). As I said at AfD when this began, throwing contentious political terms in likely makes editors' eyes glaze over. Which is to say that I totally agree with what AndyTheGrump said above. O3000, Ret. (talk) 14:19, 12 November 2023 (UTC)Reply[reply]
We cannot know the usefulness of Google sources, because the agenda is done by secret algorithms and secret committee and board meetings at Google. Meta-reviews, which in some sense are what we do for articles, depend on the quality of the selection of sources. Cochrane meta-reviews are obliged to detail exactly how the sources were selected. In Wikipedia, we do not (and cannot) require editors to say how they searched for sources, but we can reduce our advocacy for particular organisations that have opaque ways of selecting sources, or at least diversify our advocacy. Repeating arguments that appear to have been missed or misunderstood is me being patient, not "having an agenda". As stated on the talk page of the draft, I quite deliberately did the best to avoid my own biases by including search engines that are objectively useful and independent of Google, but that I personally do not find useful. I also strongly recommend that you look at the Wikipedia article political science: the adjectives that I have used are not my invention ("would-be" is not a standard term, but the equivalent type of nuance is common), and they are used in numerous peer-reviewed publications. They are objective terms that can be supported or refuted by evidence. The links to Wikipedia pages are to NPOV-ed articles that have survived the usual Wikipedia public peer-review procedures; the titles and section titles are not mine. Boud (talk) 18:22, 12 November 2023 (UTC)Reply[reply]
On the contrary: We can know the usefulness of Google Search, because editors can tell us whether they find it useful. Useful is not a synonym for transparent. WhatamIdoing (talk) 17:40, 13 November 2023 (UTC)Reply[reply]
What Wikipedians perceive individually as useful overlaps with, but is not the same as, what is useful for Wikipedia. The search engines we use have search engine bias and some of them have customised results and filter bubbles. Knowing these biases and filter bubbles is difficult; in this sense, knowing the usefulness of Google searches is not impossible, but is a difficult field of research. Boud (talk) 01:59, 15 November 2023 (UTC)Reply[reply]
(after edit conflict) We are not talking about the content of the encyclopedia, but a tool that may help editors to find sources. I personally find Google Books and Scholar very good for this purpose, with News less good and a general web search (anyone's general web search) hardly at all. It should stick to what is actually used by most people. In the English-speaking world Google seems to be the most used search engine, whether one likes it or not, just as Windows is the most used operating system. There is no advocacy involved in saying this. Phil Bridger (talk) 14:29, 12 November 2023 (UTC)Reply[reply]
Your personal perception of usefulness does not change the fact that Google has an agenda and that we advocate for Google; and it does not make Google's selection unbiased. (As for operating systems, as far as I know, Google uses Debian GNU/Linux, Facebook uses Fedora GNU/Linux, and most servers use one or another GNU/Linux distribution; smartphones mostly use Android with a Linux kernel or Darwin/XNU (which is UNIX certified); and Wikipedia runs on Debian. MS only dominates the PC market.) Saying that Google is popular is not advocacy: I agree. Saying on nearly a million pages that people should primarily use Google is advocacy. Boud (talk) 18:22, 12 November 2023 (UTC)Reply[reply]
In going for brevity I omitted the all-important qualification that Windows is the most used operating system on home PCs in the English-speaking world. I apologise. The rest of my post stands as written. Phil Bridger (talk) 18:36, 12 November 2023 (UTC)Reply[reply]

New proposal for the initial statement:

In the Find sources module and in the {{Find general sources}} template, should we balance the non-mainspace advocacy recommendations for any particular search engine(s) or meta-search engine(s), with the aim of diversifying the selection biases in finding sources and for encouraging Wikipedians to protect their privacy? If yes, then what changes should be made?

Boud (talk) 18:40, 12 November 2023 (UTC)Reply[reply]

  • v2: version 2: s/non-mainspace advocacy/recommendations/ Boud (talk) 19:31, 12 November 2023 (UTC)Reply[reply]
Done. There is no advocacy in the Find Sources template. O3000, Ret. (talk) 19:01, 12 November 2023 (UTC)Reply[reply]
Considering that most of the people in this discussion have disagreed with the salient points of that, how about Should we consider that the use of Google in the Find sources module and in the {{Find general sources}} template violates WP:NOTADVOCACY or WP:NPOV? I mean, that's if you want to form a consensus. ~~ AirshipJungleman29 (talk) 19:02, 12 November 2023 (UTC)Reply[reply]
I don't see the point - that version would be unlikely to lead to concrete, actionable changes and the conclusion would be more or less foregone. WP:ADVOCACY tends to focus on the mainspace content rather than the common-sense meaning of how we actually select the sources used for mainspace content, so debating that would get into WP:WIKILAWYERING about how to interpret the essay. As for Google selection violating NPOV, we don't have good data about the details of its selection, so we can only make the reasonsable expectation that the selection biases NPOV, which is not quite as strong as violating it. In any case, it would be unclear what the result of a hypothetical "yes" consensus would be: I fail to see any chance of a consensus to completely remove Google from the module + template.
Regarding consensus, the consensus to be sought is among the people in the future who may participate, not just those most active right now. Several people have already said that they are interested in diversifying the current list.
How about if we replace non-mainspace advocacy by recommendations, since people are uncomfortable stating that advocacy is advocacy? Boud (talk) 19:28, 12 November 2023 (UTC) (I made this v2.) Boud (talk) 19:31, 12 November 2023 (UTC)Reply[reply]
Do try and click on links, instead of assuming you know where each one goes (see WhatamIdoing's relevant comment above). I would also suggest being less condescending, as that may put off "the people in the future who may participate", along with, well, myself. Best wishes with your proposal, ~~ AirshipJungleman29 (talk) 20:01, 12 November 2023 (UTC)Reply[reply]
I don't like the wording that runs "with the aim of diversifying the selection biases in finding sources and for encouraging Wikipedians to protect their privacy". In particular, this meant to prioritize these over everything else, but it does not say what any of those other things are. It's rather like saying "You support motherhood, apple pie, and the flag, right"? and then finding out that the guy's emptied your bank account, burned your house to the ground, and stolen your dog, because, hey, you said you support motherhood, apple pie, and the flag, but not money, housing, and pets.
I think in this case, the request is: Could we please link to search engines with transparent algorithms and decent privacy practices instead of search engines that are more likely to find relevant results?
I say this as a person who primarily uses the privacy-oriented DuckDuckGo, and who also fairly often gets inadequate or incorrect search results from it, and switches to Google to find the thing that I need. DuckDuckGo is usually sufficient, but not always. WhatamIdoing (talk) 17:57, 13 November 2023 (UTC)Reply[reply]
DuckDuckGo as a default with Google as a backup is what I used to do, not so long ago. Now I generally use Startpage + Searx (roughly equally) with DuckDuckGo as a rare alternative. In any case, my own preferences are not the issue here. Back to the question of the RfC statement/question.
I don't understand why diversifying ... encouraging would be interpreted as instead of. "Diversify" says nothing about what sort of compromise to choose: it does not exclude any search engine completely; "encourage" does not necessarily mean excluding any particular privacy-violating search engine either. To continue with your analogy, once someone says "Yes" to "motherhood, apple pie, and the flag", that same someone gets to answer the If yes, then what changes should be made?" question, to give his/her proposal of how to balance motherhood, apple pie, the flag, money, housing and pets. Boud (talk) 20:04, 13 November 2023 (UTC)Reply[reply]
Because, realistically speaking, if people were to express support for "diversifying the selection biases in finding sources and for encouraging Wikipedians to protect their privacy", I expect you to turn up the next day insisting that we have an RFC-backed mandate to remove links to Google, because in your opinion retaining a link to Google would not be consistent with support for "encouraging Wikipedians to protect their privacy". WhatamIdoing (talk) 00:29, 17 November 2023 (UTC)Reply[reply]
I see your point, but I guess we disagree on interpreting encouraging ... to .... For me, this is not as strong as insisting that .... Anyway, the encouraging ... to ... wording is no longer in the draft. Boud (talk) 19:07, 21 November 2023 (UTC)Reply[reply]

Any objections to the current draft RfC at User:Boud/sandbox/draft RfC Reduce advocacy in Find sources Module? Boud (talk) 20:55, 15 November 2023 (UTC)Reply[reply]

I think the question is still non-neutral. How about: "In the Find sources module and in the Find sources: Google (books · news · scholar · free images · WP refs· FENS · JSTOR · NYT · AP · TWL template, should we diversify our recommendations for specific search engine(s) or meta-search engine(s)?" Firefangledfeathers (talk / contribs) 21:00, 15 November 2023 (UTC)Reply[reply]
My opinion is to simply go with:
Hey there, here are three options of the template:
Option 1. Leave as is
Option 2. A position we can hash out here for now in broad strokes, and which we haven't yet
Option 3. Boud's original proposal
We should cut on the demagoguery that RfCs like these will inevitably attract, so one RfC only, limited options, in the worst case we may add some articles one by one.
For me, singling out NYT is ridiculous, but AP only isn't the way to go. We could instead link to our articles called Newspaper_of_record#Selected_existing_examples and News_agency#List_of_major_news_agencies, with a link to WP:RSP and advice on WP:RX just in case. Szmenderowiecki (talk) 21:48, 15 November 2023 (UTC)Reply[reply]
@Firefangledfeathers: I've updated the first question to match your version but kept in the second question, and haven't changed the Additional comments below first statement and timestamp. I kept the second question to make clear that anyone answering "yes" is expected to give concrete suggestions of changes. The additional comments are also aimed to help focus the "yes" answers on making concrete suggestions, with arguments/evidence. Boud (talk) 22:29, 15 November 2023 (UTC) (separate responses) Boud (talk) 22:41, 15 November 2023 (UTC)Reply[reply]
@Szmenderowiecki: There is currently no Option 3, because the requirements of neutrality required me to not make a specific overall proposed change. I don't see much point having an RfC that has Option 3 Boud's one-against-all option in opposition to Option 2 a community rough consensus option :). If you can facilitate a discussion to develop an Option 2, then Option 3 as Other alternatives would seem better to me, although it would be clearly harder to summarise by the closer if Option 3 was the option best supported by arguments. Boud (talk) 22:41, 15 November 2023 (UTC) (clarify Boud (talk) 12:34, 16 November 2023 (UTC))Reply[reply]
WP:RFCNEUTRAL does not require you to hide your goal or to make only vague proposals. WhatamIdoing (talk) 00:31, 17 November 2023 (UTC)Reply[reply]

I don't see any work done on Szmenderowiecki's alternative with a specific "Option 2", but we already had changes to the RfC draft by Firefangledfeathers, which were implemented on the draft, and we've had further simplifications by whatamIdoing, giving the current draft RfC of

{{rfc|policy|tech}} The {{Find general sources}} template produces this list of search options: "Find sources: Google (books · news · scholar · free images · WP refs· FENS · JSTOR · NYT · AP · TWL". Should we change the items in the list? If yes, then what changes should be made? ~~~~

with Additional comments below first statement and timestamp.

We seem to have rough (not full) consensus. Any objections to proposing this as an RfC here on WP:VP/PR? Boud (talk) 21:09, 21 November 2023 (UTC)Reply[reply]

As I said, I hate open-ended questions, but it's a much better RfC question than you asked originally, so if others think it's OK to start such a question, it's fine. It's that my understanding of VPR/VPP is that you have something concrete to propose and then you vote on change/no change. If your proposal is just "let's do something but I don't know exactly what", then it belongs to the idea lab.
Regardless of all of the above, it's a good question and it's great you have broached this topic. I will definitely have something to say, as I hope will many others. I definitely am not holding you from starting this RfC, but you have been warned.
Also, wait a sec and I will post what I see to be the template I think I would definitely use. Szmenderowiecki (talk) 22:05, 21 November 2023 (UTC)Reply[reply]
Find your source
The Wikipedia Library offers wide access to scholarly articles and academic books for eligible users.Explain who is eligible
Consider also using:
Fellow Wikipedians can help you find what you were searching for if you still cannot.
Without the comments, it looks like this:
Find your source
The Wikipedia Library offers wide access to scholarly articles and academic books for eligible users.
Consider also using:
Fellow Wikipedians can help you find what you were searching for if you still cannot. Szmenderowiecki (talk) 23:08, 21 November 2023 (UTC)Reply[reply]
Thanks for letting this proceed and for the warning. A flaw in your specific proposal is that the template is currently used as a very compact, single-line (depending on browser width) guide to sources; I think that people would prefer any replacement to also be very compact. Boud (talk) 12:48, 25 November 2023 (UTC)Reply[reply]

Wikidata connected sections[edit]

I think we should have wikidata connected sections. For such situations the section header will have a language dropdown indicator. This could go to articles or other sections in foreign language wikipedias and potentially solve the bonnie and clyde problem. Depending on how it's implemented it could help with translation between wikipedias too by letting people track which subtopics are covered in articles in which languages, someone could use an expand-language template and specify a section from that language's page.

I imagine such wikidata linked sections as being kind of like anchors, with links being more persistent. Renaming the sections could automatically update section link redirects.

Ideally we would somehow change links on wikipedia to more properly respect links to wikidata connected sections. interlanguagelinks could be corrected by User:Cewbot to section links, and we could have some kind of process for spinning off a section into its own article or merging an article into a section of another one. Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 20:33, 12 November 2023 (UTC)Reply[reply]

I want to bring this to next year's proposals too Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 23:22, 12 November 2023 (UTC)Reply[reply]
  • Not gonna happen. How to integrate WD into WP has been discussed many times before, and (except in very limited ways) the idea has been continually rejected (I would say it is becoming a perennial proposal). The primary problem is that Wikidata continues to have issues with reliability. It’s a great resource for locating potential sources, but we still need vet those sources manually. Blueboar (talk) 23:52, 12 November 2023 (UTC)Reply[reply]
    @Blueboar The main reason I want such an implementation is just so that foreign language wikipedias can be connected more than they currently are. I think that is different than some other proposals. Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 01:44, 13 November 2023 (UTC)Reply[reply]
Courtesy links: d:Wikidata:Sitelinks to redirects (the "Bonnie and Clyde" problem); meta:Community Wishlist Survey 2019/Wikidata/Solution to the ‟Bonnie and Clyde” problem. Folly Mox (talk) 03:50, 13 November 2023 (UTC)Reply[reply]
@Folly Mox thank you. My proposal is more specific (miht be too specific though idk) but do you agree with the other person that this proposal is dead in the water from the beginning? Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 18:20, 13 November 2023 (UTC)Reply[reply]
I don't agree that the proposal is necessarily a non-starter. It seems like it's just increasing the kind of connectedness Wikidata already has with the rest of the ecosystem, and doesn't have a direct impact on content, so the reliability concerns come into play to a lesser degree than many other Wikidata integration proposals.
I do think that this thread as formulated should probably have gone to the Idea Lab first, but I understand this was the venue recommended to you.
As to the merits of the proposal, I don't understand Wikidata well enough to assess whether or not this is likely to be able to solve the "Bpnnie and Clyde" problem as you suggest— I linked those pages above because I had never heard of the problem and didn't know what it meant.
The chief difficulty I see in implementation, at first glance, is how article sections are so much more mutable than article topics. Any major reorg, and some minor reorgs, have the potential to result in broken anchors. It might also be difficult to assign with any degree of certainty that topic of a section. I'm sure multiple topics (or Wikidata items; as I said, I'm unfamiliar with the system) could be assigned, but for anything other than navigation between different language projects – like extraction of structured data – I could see it being fraught, due to loose associations compounding to produce misleading results. I'm no expert though, so I could be wrong about any and all of this. Folly Mox (talk) 18:56, 13 November 2023 (UTC)Reply[reply]
@Folly Mox my idea is that at least at first it would solely be for linking articles together cross-language. I think this is less prone to failure than general article sections Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 02:50, 15 November 2023 (UTC)Reply[reply]
Ok, well that sounds pretty safe, at least. What advantages does your proposal have over an interwiki link to another language project with more information on the topic? Folly Mox (talk) 03:32, 15 November 2023 (UTC)Reply[reply]
@Folly Mox sorry for the late response. The big advantage I see with it is that different wikis have different standards for what warrants a page, or have particular contributors that are interested in certain things. Notable examples I've seen include that Gerrman Wikipedia is a lot better at historical timekeeping topics, Russian wikipedia seems to have a lot more on anthropology of religion, and military strategy. So when an interested reader sees a small section on one article they might find it worth looking at the full article on the thing in another language.
My main area of contribution on Wikipedia is Shinto and I particularly notice Japanese wikipedia has about 3-4 times the articles, but English articles are held to a much higher standard of notability. In this article Kamiyonanayo for example we could have a section dedicated to each of this group of gods that links to the article in Japanese. We could do the same for this article Sect Shinto with each section being linked to an article in Japanese.
At least in the past I've attempted to push for articles to match the structure of Japanese wikipedia, even in areas where them being in one article makes more sense. I imagine wikidata linked structures would allow for more flexibility in the use of wikidata links like this. Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 08:59, 22 November 2023 (UTC)Reply[reply]
Nothing will be cited to wikidata, this will all serve the purpose of making interlanguage stuff easier Immanuelle ❤️💚💙 (talk to the cutest Wikipedian) 23:43, 29 November 2023 (UTC)Reply[reply]

Date format for year articles[edit]

I checked the year articles for 2021 onwards and they look a tad different from the previous revisions. Recently, the date formatting was changed from MDY (November 16, 2023) to DMY (16 November 2023). According to the edit summary, it stated that "MDY had just been chosen without consensus and for a global article like this [year article], and DMY is more global". Example is given here. So what do you guys think? Should we shift to DMY dates for the year articles? Or keep it at MDY? RMXY (talkcontribs) 11:46, 16 November 2023 (UTC)Reply[reply]

  • I'm a little confused (FYI, you haven't added an RfC tag {{rfc}}, and I'm not sure this merits an RfC so I'm not adding one yet), but I assume you're referring (broadly) to MOS:DATE. I am 99% sure that both MDY and DMY are acceptable, and I'd like to keep it that way. There's no need to specify which format we should use for particular group of articles. I mean, DMY is clearly better, but... Edward-Woodrow (talk) 12:51, 16 November 2023 (UTC)Reply[reply]
Both are acceptable, mass changes should not be done, editors should be able to come to a consensus on which to use for the page on its talk page. -- LCU ActivelyDisinterested «@» °∆t° 17:53, 16 November 2023 (UTC)Reply[reply]

Change all articles about generic years to the DMY format At the Wikipedia talk:WikiProject Years, we've been discussing converting all generic articles about calendar years to the DMY format ("1 October 1998" as opposed to "October 1, 1998"). This is not a proposal for a guideline/policy change, but is a proposal to set a standard for a very narrow set of articles, namely articles about generic calendar years, i.e. 1987. I will personally convert all articles from 1899 to 2029. I would like to create consensus for this standard, is necessary by a RfC here. Reason: There was never created consensus about the use of MDY. It was just chosen. DMY is, to put it shortly, a much more global date system used in the majority of English-speaking contries. I therefore find that DMY is considerably more suitable for use on articles about generic years.--Marginataen (talk) 20:43, 16 November 2023 (UTC) When writing about international subjects, the DMY format is the standard format. I find it baffling that year articles use the MDY format. This is properly because many editors are north american but this is not a proper reason. I've looked in the archives and this is not something that's ever been properly discussed. I would like to suggest chanding articles about dates, years, decades and centuries to the DMY format. This should not apply for some articles about years in specific countries, e.g., 2023 in the United States. DMY = 29 October 2023. MDY = October 29, 2023.--Marginataen (talk) 17:03, 29 October 2023 (UTC)Reply[reply]

@Thebiguglyalien @Deb @Alsoriano97 @Carter00000 @Wjfox2005 @PaulRKil @JohnAdams1800 @4me689 @Donner60 @DementiaGaming @Austria Football 02 @XTheBedrockX @Yeoutie

I'll just elaborate a bit. I've look through the archives. There were a few mentions e.g. "Date formatting in decade article" but those never went anywhere. There has been no discussision first before MDY was just implemented. The precedence for using DMY for anything that isn't specifically North American is extremely strong. DMY is used by most English-speaking countries and most of the world. From a lingustic point of view, MDY is simply illigionally as it goes 2-1-3 instead of 1-2-3 (from smallest to biggest). No consensus about using MDY was ever established. Marginataen (talk) 15:56, 31 October 2023 (UTC) Date format by country. Map showing which countries primarily use which endian date format. Colors for base formats CMY, mixing using CMYK color model. Reply[reply]

  DMY – Day/Month/Year – Little-endian
  MDY – Month/Day/Year – Middle-endian
  YMD – Year/Month/Day – Big-endian (used internationally as ISO 8601)

--Marginataen (talk) 20:51, 16 November 2023 (UTC)Reply[reply]

  • Comment - I don’t care one way or the other, but I will just mention the 2000 pound gorilla in all this: The MDY format is standard in the US, and an overwhelming number of our readers are from that country. While Americans are aware that other places use DMY , they do find it a bit jarring when they see it (the exception being 4th of July). Perhaps more importantly, an overwhelming number of our editors are from the US. No matter what rules/conventions we may adopt, these editors will continue to use MDY when adding new events. It is what comes naturally to them. We will thus have to perform continuous maintenance on the articles to conform them to whatever set standard we may choose.
So… I would just apply ENGVAR… allow whoever starts the article to set the variant, and don’t change it. Yes, that means some articles will use MDY while others use DMY. I don’t have a problem with that.
Or… we could adopt YMD, and upset everyone equally (😉). Blueboar (talk) 21:16, 16 November 2023 (UTC)Reply[reply]
While Americans are aware that other places use DMY , they do find it a bit jarring when they see it applies equally in reverse to people who use DMY dates (maybe also those who natively use YMD dates experience the same for both other systems, but I'd have to research that). This is therefore not an argument in favour of MDY that holds any weight. Thryduulf (talk) 21:31, 16 November 2023 (UTC)Reply[reply]
an overwhelming number of our editors are from the US. No matter what rules/conventions we may adopt, these editors will continue to use MDY when adding new events. It is what comes naturally to them This is the exact same the other side around. I constantly see articles with a tag to use MDY using DMY during the text. This is not an argument. Marginataen (talk) 21:44, 16 November 2023 (UTC)Reply[reply]
@Blueboar, according to, less than half of our page views come from the US. The breakdown is 42% from the US, 11% from Great Britain, 11% from India, 5% from Canada, 3% from Australia, 1% from Ireland.
But I agree with you that this seems to be an end-run around the usual rule that whoever does the work for that individual page gets to set the standard for that initial page. WhatamIdoing (talk) 00:20, 17 November 2023 (UTC)Reply[reply]
The concept that DMY is better than MDY has been rejected in many discussions, many of which can be found in the archives of Wikipedia talk:Manual of Style or Wikipedia talk:Manual of Style/Dates and numbers. Since that argument has no merit, if the format is to be harmonized among year articles, I believe we should use the format that is in the majority among those articles. I had Excel choose 25 random numbers between 1 and 2022. All 25 turned out to use the MDY format. All but one had a {{Use mdy}} template with a date in 2011, so perhaps something systematic was done in that year. Jc3s5h (talk) 21:24, 16 November 2023 (UTC)Reply[reply]
Of course they are all MDY. That's what I'm trying to change. The fact that MDY was chosen does not mean it should be used for eternity. I'm not saying that MDY is a priori worse than DMY. I'm simply of the belief that DMY is more suitable for the use in articles about generic years that should have a global and not just American appeal. Look at the map. Marginataen (talk) 21:37, 16 November 2023 (UTC)Reply[reply]
I reject the map because I don't think customary formats for information can be compared between languages. The way dates are formatted in Chinese or Hungarian is not relevant for an English encyclopedia. Jc3s5h (talk) 17:41, 17 November 2023 (UTC)Reply[reply]
  • This sounds like an attempt to overturn WP:ENGVAR. There's nothing special about year articles vs. any other article on Wikipedia, and no reason to have a special standard there. The "map" being cited would be a reason for any article to have a specific date format, an idea specifically rejected by early Wikipedia very wisely. It doesn't matter if one year uses MDY and another year article uses DMY and another year article uses YMD. Just use whatever format is already established. Consistency isn't important nor expected, just as articles already differ in the variety of English used. SnowFire (talk) 22:30, 16 November 2023 (UTC)Reply[reply]
    What WP:ENGVAR actually says is When an English variety's consistent usage has been established in an article, maintain it in the absence of consensus to the contrary. (my emphasis). This discussion is an attempt to form such a consensus. Regardless of whether the discussion arrives at consensus for (or against) the proposal, the discussion is proper and not an attempt to overturn the guideline. Thryduulf (talk) 23:08, 16 November 2023 (UTC)Reply[reply]
    Yes, but I'm sure you understand what was meant here. The whole point of ENGVAR and RETAIN is to avoid having to do the same poll all across the site 9999x times. It protects both sides, as you can't have Americans all start whining and holding a vote to move the other way to MDY, either. When there's a consensus, it's always about something local like MOS:DATETIES - e.g. 2023 in the United Kingdom should clearly be in DMY format. Unfortunately, Marginataen phrased the contested edit as that DMY should be used because it's "more global," which is classic ENGVAR-warring stuff. That's basically saying "let's standardize on DMY except for articles on specifically American topics", because a lot more than just years are global topics. Which, well, fine, but that's challenging the truce of ENGVAR like I said in my initial comment. If there's a specific consensus to be had, it needs to be on something unique to years, which I can't find any at DATETIES. (But honestly, based on Marginataen's comments, it really just sounds like an argument to use DMY everywhere.) SnowFire (talk) 23:48, 16 November 2023 (UTC)Reply[reply]
    What MOS:DATEVAR, which specifies "consensus on the article's talk page", IMO doesn't need is a "Ha ha, we had a big RFC once six years ago and we voted to not let you use that date style in this group of articles. That's consensus!"
    I could argue that all sorts of subjects are "global" in nature (hmm, a global pandemic?) and therefore should use "my" date style, but that's exactly the kind of thing that DATEVAR says is a bad idea. WhatamIdoing (talk) 00:23, 17 November 2023 (UTC)Reply[reply]
    @SnowFire: this discussion, which explicitly excludes articles with ties to a specific country, is happening here to have one centralised discussion rather than one per article. You can oppose change if you wish, but do so on the merits of retaining MDY, the disbenefits of DMY and/or the disbenefits of changing, but having the discussion is appropriate, in the proper place and (given that someone acting in good faith feels the change is desirable) required by the guideline. That would be different if there was a recent consensus on the topic, but the evidence suggests that if there has been previous discussion it took place at least 12 years ago in a location that is sufficiently non-obvious that it has not been found. Thryduulf (talk) 00:55, 17 November 2023 (UTC)Reply[reply]
    Okay, I guess you didn't understand what I meant after all. Let me be more blunt. This is an invalid RFC. See WhatamIdoing's comment. The whole point of RETAIN is precisely to avoid polls of "what's your favorite English variety" which lead to anger, bad feeling, and wasted time. When it's talking about consensus, it's talking about "consensus on whether special factors apply or not", e.g. whether national ties are strong enough to override the usual RETAIN rule. It's not an invitation to start a poll on which date format is better, or to make people have to ridiculously defend "the merits of MDY" (it's arbitrary! there are no benefits either way!). It would be a valid RFC, albeit one I opposed, to suggest just overturning DATETIES and replacing it with "We use DMY on Wikipedia". That'd be fine. But there's nothing here so far that is really about years, and is instead really about sneakily overturning RETAIN on date style. Better to just make it explicitly about that. SnowFire (talk) 02:06, 17 November 2023 (UTC)Reply[reply]
    Fully concur with SnowFire on this point. --Coolcaesar (talk) 07:25, 20 November 2023 (UTC)Reply[reply]
As expressed on other talk pages, I am wholly in favour of the switch to DMY on all main year articles as per Marginataen, and will support any such RFC on the matter. TheScrubby (talk) 23:46, 16 November 2023 (UTC)Reply[reply]
According to MOS:VAR: "If you believe an alternative style would be more appropriate for a particular article, discuss this at the article's talk page or – if it raises an issue of more general application or with the MoS itself – at Wikipedia talk:Manual of Style". This is not a request for a particular article but a call for consistent use of DMY across all year articles. Within this very specific and narrow group of articles, I do find it legitimate and despirable to wish for a common format/standard. Before starting a RfC, a better move might therefore be (as recommended in the quote) to suggest a change to the Manual of Style? This might be as simple a change as adding what I've put in bold: "If you believe an alternative style would be more appropriate for a particular article or group of articles, discuss...". Marginataen (talk) 16:28, 17 November 2023 (UTC)Reply[reply]
Just to be clear here... if your concern is consistency... would you be okay with standardizing on MDY or YMD in the name of consistency? To be clear, I'm not advocating this. I don't think consistency is important at all, personally. But just a thought - would going the "wrong" way in the name of consistency be acceptable? SnowFire (talk) 16:53, 17 November 2023 (UTC)Reply[reply]
My wish is the DMY format to be used consistently across all year articles. Marginataen (talk) 17:06, 17 November 2023 (UTC)Reply[reply]

Pre-RfC: Date format for year articles[edit]

I get that an RfC is the likely next step here. Can we consider what the options should be, and what would be a good brief, neutral question? Are there options beside:

  1. All year articles should use DMY
  2. All year articles should use MDY
  3. All year articles should use YMD (added per WhatamIdoing)
  4. Neither style should be established, and normal MOS:DATEVAR rules should apply?

Can the question be as simple as "What date style should be used for year articles (e.g. 2023, 1491)? Firefangledfeathers (talk / contribs) 21:06, 16 November 2023 (UTC) adding YMD and numbering options 13:29, 17 November 2023 (UTC) striking YMD 15:00, 18 November 2023 (UTC)Reply[reply]

I would just like to empertize that this is only about generic year articles, not an article like 2023 in the United States :) Marginataen (talk) 21:40, 16 November 2023 (UTC)Reply[reply]
Ok. We could be clear about that with "What date style should be used for generic year articles (e.g. 2023, 1491)? This RfC would not apply to more specific year articles like 2023 in the United States." Look good? Firefangledfeathers (talk / contribs) 21:42, 16 November 2023 (UTC)Reply[reply]
For me, yes. Marginataen (talk) 21:51, 16 November 2023 (UTC)Reply[reply]
FFF, I think those are the three main options, though technically YMD (YYYY-MM-DD) is possible. WhatamIdoing (talk) 00:26, 17 November 2023 (UTC)Reply[reply]
Thanks WAID. I added it. I think it's unlikely to be the winner, but I don't think a four option RfC is too complex. If the RfC runs, I'm likely to add in a comment pointing out that YMD in a year article would make it seem very MDY-ish, since the year rarely needs to be mentioned. Firefangledfeathers (talk / contribs) 13:29, 17 November 2023 (UTC)Reply[reply]
I support the third option--stick to the normal MOS:DATEVAR rules. For prior years (2022 and before) I'm fine with switching to the DMY format, but for 2023 leave it in MDY format until the year is finished to permit a transition period, and for future years use the DMY format. This is a grandfather clause so to speak, where prior articles on generic years don't have to use DMY format, but are encouraged to, and future articles (2024 and beyond) will begin with DMY format.
The 2023 article can continue to use MDY format until the end of the year, and then switch to DMY format--it's easier to change it all at once in January 2024. JohnAdams1800 (talk) 02:37, 17 November 2023 (UTC)Reply[reply]
Thanks JA1800. We're workshopping the RfC question and options right now, so you may need to paste your !vote elsewhere later on. A few clarifying points:
  • since I added another option, the one you're referring to is now the fourth option, sorry about that
  • the process you outline would not exactly be Option 4, since we would be determining the date style "from above" and not via article-level consensus
  • the article 2023 is actually using DMY dates right now per some recent rough consensus at its talk page.
Firefangledfeathers (talk / contribs) 13:29, 17 November 2023 (UTC)Reply[reply]
  • I think the RfC question and options are in decent shape. I am not likely to start it while the interesting and important discussion above is ongoing, about whether or not this RfC is even valid. Firefangledfeathers (talk / contribs) 13:29, 17 November 2023 (UTC)Reply[reply]
  • Per above comments, I don't think this is valid RFC - at least probably not. It would break longstanding precedent to avoid unproductive ENGVAR polling. If the concern is consistency, I guess, but that also sets a scary precedent- imagine some other group of linked topics (other than years) where an American editor is happily editing and maintaining Article A1 in American style, and a British editor is happily maintaining and editing Article A2. A third editor comes along and tells them that these two articles, as part of a linked topic like years, actually need to have consistent date formatting. Great, now one editor is going to be unhappy at best, and maybe driven off from working on their article at worst after being told they can't format dates how they're used to. A pointless loss when readers really do not care at all about this and everyone can understand the "other" format just fine.
    • Also, who exactly has proposed standardizing on MDY? I don't think anyone has.
    • If the RFC went forward anyway, my suggestion would be phrasing the question to be strictly related to the importance of consistency rather than which date format is the best. The RFC should be "Is date consistency important in year articles? Option 1: Status quo, no, different articles have different ENGVAR styles and that's fine. Option 2: Date style should be consistent across year articles." Then if option 2 wins, somebody flips a coin while streaming video of it, and we go with whatever the coinflip says. Really. Avoids unproductive "which format is better" arguments, that way, because the truth is it's arbitrary and if consistency is really important, then we'll get it either way. SnowFire (talk) 16:53, 17 November 2023 (UTC)Reply[reply]
      I agree with SnowFire in saying that the underlying (primary) question is whether we maintain the current Status Quo regarding date format consistency or not. The current Status Quo is that we need consistency within an article, but not between different articles, and so we apply WP:ENVAR and WP:RETAIN … retaining whichever format was used first in a given article. It may be that consensus has changed on that question (I don’t mind asking)… but that is the core question. Blueboar (talk) 17:39, 17 November 2023 (UTC)Reply[reply]
      I agree with User:SnowFire and User:Blueboar should be the status quo. Its the same as language, the use of English for the relevant article (i.e. American English for American article, British English for British). Davidstewartharvey (talk) 16:41, 18 November 2023 (UTC)Reply[reply]
  • Dates in the YYYY-MM-DD must not be used for Julian calendar dates, or any date before 1583, for the reasons explained in the "Manual of Style/Dates and numbers". Such usage should be regarded as falsehoods and the editors adding them should be dealt with accordingly. Jc3s5h (talk) 17:51, 17 November 2023 (UTC)Reply[reply]
    @Firefangledfeathers, this means that YMD shouldn't be in the list after all. WhatamIdoing (talk) 03:37, 18 November 2023 (UTC)Reply[reply]
    Struck! Firefangledfeathers (talk / contribs) 15:00, 18 November 2023 (UTC)Reply[reply]
In terms of context, it might be appropriate to say that most articles happen to be using MDY at the moment, and some editors would like to change them to DMY.
I assume that notifications will be posted for the relevant MOS pages (WT:MOS and WT:MOSDATE should be enough?). WhatamIdoing (talk) 03:40, 18 November 2023 (UTC)Reply[reply]
Yeah, let's make a notification there Marginataen (talk) 20:49, 18 November 2023 (UTC)Reply[reply]
I've made a request at Wikipedia talk:Manual of Style#Style for group of articles Marginataen (talk) 17:02, 19 November 2023 (UTC)Reply[reply]
A detailed comment I made at WT:MOS about all this is also pertinent here [5].  — SMcCandlish ¢ 😼  21:20, 19 November 2023 (UTC)Reply[reply]
  • Question: The proposal says it is limited to “by year” articles… but what about “by month” sub-articles (examples: April 1966 or October 1975)? After a quick glance, it seems these are currently consistently formatted with “Month Day”. Is it envisioned that we would have to reformat all of these sub-articles as well? Blueboar (talk) 12:54, 20 November 2023 (UTC)Reply[reply]
    I would expect that to be a separate discussion. If the outcome of this discussion is that each article should be discussed individually and/or opposing any change at all then it seems unlikely that discussing month articles as a set will be worth anybody's time. Conversely if there is a strong consensus in favour of changing year articles to DMY then it will likely only require a short and simple discussion to determine if that also holds for month articles. Thryduulf (talk) 13:21, 20 November 2023 (UTC)Reply[reply]
  • We used to have an option to choose date formats, which had some problems. Unfortunately instead of fixing these problems (by comparing to the options available at the Chinese Wikipedia, a fix seems feasible), the option was removed. In any case, I can't think of a reason to go against the wisdom of WP:DATEVAR in cases like this where the article topic does not have any clear WP:TIES. It is possible to work with both formats, and nothing breaks if different articles use a different format; this is a feature of Wikipedia's approach to national varieties of English, not a bug. —Kusma (talk) 14:07, 20 November 2023 (UTC)Reply[reply]
    I think you may be referring to wikilinking dates to enable autoformatting of dates. This discussion is a good introduction to the topic. In my opinion, a major problem for that solution, and one of the reasons it was removed, is that it only worked if you were signed into an account and had chosen your default format in your preferences, and a date in an article had been wikilinked in a particular format. So, it did not help any reader/editor who was not signed into an account, or, if they were signed into an account. had not designated their preferred format in their preferences (namely, almost everyone using Wikipedia), and even then it didn't work if the date had not been correctly wikilinked in an article. The only way to make something like work for the general Wikipedia user would be to somehow pop up a window when the user first connected to Wikipedia, and ask them to chose which format they wanted dates to appear in. Even after making such a choice, it is likely that they would see some unformatted or incorrectly formatted dates that would not display in their chosen format. I think most of us will agree that whole process would be undesirable. Donald Albury 16:24, 20 November 2023 (UTC)Reply[reply]
    Given how easy it is to switch (without logging in) between different varieties of Chinese (that sometimes use different words) on the Chinese Wikipedia, I believe that all of these problems can be solved. —Kusma (talk) 18:26, 20 November 2023 (UTC)Reply[reply]
    That's not actually easy. It's almost a case of Any sufficiently advanced technology is indistinguishable from magic. WhatamIdoing (talk) 21:50, 21 November 2023 (UTC)Reply[reply]
    I'm not expecting magic, I am thinking of something like suitable markup in the wikitext that makes it possible to display one of a finite set of options, just like on the Chinese Wikipedia. —Kusma (talk) 21:59, 21 November 2023 (UTC)Reply[reply]
  • This pre-RfC discussion has gone stale, with no clear consensus on whether or not an RfC is appropriate for a question like this. Since participants of the RfC will be able to express within it that they view the whole exercise as pointless, or harmful, I think the best option here is to proceed. I'm giving it another 24h or so in case there's any final feedback, and unless there's something major I plan to start the RfC with the above options. Firefangledfeathers (talk / contribs) 17:48, 28 November 2023 (UTC)Reply[reply]
    Hello Firefangledfeathers. After reading SMcCandlish's comment where he writes "A detailed comment I made at WT:MOS about all this is also pertinent here [5]", I think there should be absolutely no doubt that the discussion and RfC themselves are valid. I anyone please say read SMsCandlish's comment and tell why you still disagree (disagreeing with the proposal itself is a different issue).
    With the validity of the discussion and the ligitimacy of the village pump as venue for it, I was thinking reposting the proposal in a week or two as this current discussion has become messy and not largely not about the proposal itself. Marginataen (talk) 17:27, 29 November 2023 (UTC)Reply[reply]
    I'm fine with waiting a week or two. Firefangledfeathers (talk / contribs) 01:37, 3 December 2023 (UTC)Reply[reply]

RfC about the criteria for existence of emoji redirects (2nd)[edit]

What should be done with emoji redirects, particularly with emoji redirects that are found to represent vague concepts that are not well reflected on Wikipedia?

Note that the discussion only pertains to emojis that do not have their own pages (as in 😂 which links to Face with Tears of Joy emoji), in which case the redirect consensus is clear in that it should direct to the article. microbiologyMarcus (petri dishgrowths) 17:43, 16 November 2023 (UTC)Reply[reply]

Please see Wikipedia:Requests for comment/Emoji redirects. I've split this off to its own page because it has more than 100 comments and the page overall is too large for some editors (e.g., on mobile devices or with slower internet connections) to participate easily. Please see over there. WhatamIdoing (talk) 02:40, 30 November 2023 (UTC)Reply[reply]

More barnstars for WikiLove[edit]

Hello, I am proposing the following barnstars be integrated towards WikiLove throughout Wikipedia:

The Most Improved Editor's Barnstar
The Most Improved Editor's Barnstar
{{subst:The Most Improved Editor's Barnstar|1=message ~~~~}}
The Most Improved Editor's BarnstarThe Most Improved Editor's Barnstar

The Most Improved Editor's Barnstar is awarded as part of editor retention efforts, to editors striving to learn and adopt best practices of the project, aiming to improve their contribution.

Introduced by Santasa99 on 16 January 2021.

The Anti-anon Barnstar
The Anti-anon Barnstar
{{subst:The Anti-anon Barnstar|1=message ~~~~}} The Anti-anon BarnstarThe Anti-anon Barnstar

The Anti-anon Barnstar is awarded to editors who have encouraged and/or assisted unregistered users to join the Wikipedia community.

Introduced by Jerium on 6 November 2023.

Jerium (talk) 19:50, 21 November 2023 (UTC)Reply[reply]

@Jerium, I like these. I wonder if that last one could have a new name, though. IPs aren't anonymous, and people who encourage them to register aren't "anti" them. WhatamIdoing (talk) 21:56, 21 November 2023 (UTC)Reply[reply]
Yeah, I like them. Like WhatamIdoing said, they are very nice additions, if you change the last one. ThatOneWolf (talk|contribs) 21:59, 21 November 2023 (UTC)Reply[reply]
WhatamIdoing I based the title, partial on the criteria, and design on Template:user anti-anon. I'm fine with changing the title, someone just come up with one. Jerium (talk) 22:05, 21 November 2023 (UTC)Reply[reply]
@Jerium: Maybe something like Recruiter Barnstar? Or, we could do something like IP account suggester. Maybe one of those? ThatOneWolf (talk|contribs) 22:16, 21 November 2023 (UTC)Reply[reply]
@ThatOneWolf, "recruiter" seems to have the right tone. Schazjmd (talk) 22:18, 21 November 2023 (UTC)Reply[reply]
IMO, I do like "Recruiter Barnstar" better than the other one. ThatOneWolf (talk|contribs) 22:19, 21 November 2023 (UTC)Reply[reply]
I also like the "Recruiter" idea. WhatamIdoing (talk) 18:46, 22 November 2023 (UTC)Reply[reply]
Schazjmd It makes no sense getting consensus from there, when I am WP:WPWPA. I am the current lead for making new barnstars and remastering older ones to meet WP:B2G (see barnstars made) once Antonu stopped editing. Jerium (talk) 19:55, 22 November 2023 (UTC)Reply[reply]
The names of both of these are problematic. The first implies a contest or review process by the community to select a single person as "most improved", but that's not what it is (and we wouldn't want such a process). Might be better as "Greatly Improved Editor" or something. The second is even more obviously wrongheaded, in setting itself up as opposition to anonymous editors. This should probably be more like "Registration Helper Barnstar". And the text in in it wonky as well. "Unregistered" (i.e. anon IP) users are still part of the Wikipedia community, just less integrated into it. Maybe somethign like "encouraged or assisted unregistered users to more fully join the Wikipedia community by creating accounts".  — SMcCandlish ¢ 😼  20:49, 23 November 2023 (UTC)Reply[reply]
The Anti-anon was changed to the “ Recruiter Barnstar” already, and I like “Greatly Improved Editor Barnstar “. Jerium (talk) 21:30, 23 November 2023 (UTC)Reply[reply]

Modify Module:Find sources/templates/Find general sources[edit]

More sources are proposed to be included in Module:Find sources/templates/Find general sources. AP has been added as a result of #RfC_on_Module:Find_sources_-_replace_New_York_Times_with_Associated_Press.

also, currently the only 2 given news sources nyt and ap are both american organisations. by adding the 4 below, there will be 3 american vs 2 british and 1 global, which will be less usa-centric as a whole.--RZuo (talk) 15:19, 24 November 2023 (UTC)Reply[reply]

Note. The format of this RfC was discussed here and here (see talk and draft), which should satisfy WP:RFCBEFORE. Szmenderowiecki (talk) 22:54, 25 November 2023 (UTC)Reply[reply]

i wrote proposals.
User:Szmenderowiecki put an rfc tag . whatever "should..." blah blah blah is not my concern. RZuo (talk) 08:54, 26 November 2023 (UTC)Reply[reply]

Note. The module is rendered by the template {{find sources}} and appears as follows. Boud (talk) 19:40, 26 November 2023 (UTC) Reply[reply]

"Find sources: Google (books · news · scholar · free images · WP refs· FENS · JSTOR · NYT · AP · TWL"

Notified: WP:CENT. {{u|Sdkb}}talk 23:41, 29 November 2023 (UTC)Reply[reply]

Add reuters[edit]

Add BBC[edit]

  • support because bbc news is probably the most reputable among the most visited news websites, and the most visited among reputable news websites. and it's free, no login and whatnot.--RZuo (talk) 15:19, 24 November 2023 (UTC)Reply[reply]
  • Oppose in favor of removing all individual news outlets. {{u|Sdkb}}talk 17:16, 24 November 2023 (UTC)Reply[reply]
  • Oppose in favour of a general custom search engine that searches for all reliable outlets, something WP:WRS was supposed to offer before being abandoned. I proposed a mock-up here, but I will listen to all your suggestions. Szmenderowiecki (talk) 20:24, 25 November 2023 (UTC)Reply[reply]
  • Oppose in favor of something more WP:WRS-like, as suggested just above. We don't have links to individual scientific journals; why should we have links to individual news outlets? XOR'easter (talk) 17:52, 28 November 2023 (UTC)Reply[reply]
  • Oppose: BBC is usually ok but it's even more prone to the political winds of one country; not suitable. Nemo 07:12, 29 November 2023 (UTC)Reply[reply]
  • Oppose per Sdkb, remove individual news outlets. Eddie891 Talk Work 15:31, 29 November 2023 (UTC)Reply[reply]

Add WSJ[edit]

Add The Times[edit]

  • support because The Times is better than nyt. for example, a company has created an archive of it for scholars to study it. do you see people doing that for nyt? as the most important newspaper of a country that once ruled many countries around the world, it reported a lot more on news around the world for a much longer period, compared to the usa-centric nyt.--RZuo (talk) 15:19, 24 November 2023 (UTC)Reply[reply]
  • Oppose in favor of removing all individual news outlets. {{u|Sdkb}}talk 17:16, 24 November 2023 (UTC)Reply[reply]
  • Oppose in favour of a general custom search engine that searches for all reliable outlets, something WP:WRS was supposed to offer before being abandoned. I proposed a mock-up here, but I will listen to all your suggestions. {{search for}} is great! Szmenderowiecki (talk) 20:24, 25 November 2023 (UTC)Reply[reply]
  • Oppose in favor of a general custom search engine, as suggested above. We go through the trouble of identifying "generally reliable" sources; we might as well benefit from all that work. XOR'easter (talk) 17:45, 28 November 2023 (UTC)Reply[reply]
  • Oppose, it seems even more paywalled. Nemo 07:12, 29 November 2023 (UTC)Reply[reply]
  • Oppose per Sdkb, remove individual news outlets. Eddie891 Talk Work 15:32, 29 November 2023 (UTC)Reply[reply]

Remove all individual news outlets[edit]

Yikes. Let's take a step back here. As background, the module we're talking about is what produces the links seen mainly at the bottom of the 700k instances of {{Talk header}}. The goal is to help make it easier to find sources on a topic. However, that needs to be balanced with the imperative to keep the links in Talk header as minimized as possible to combat the notorious banner bloat on talk pages. So when we're thinking about which links to include or exclude, the frame should not be "might a few people find this helpful?" but rather "is this essential?"

In light of that, let's consider how people use this template. When I'm searching for sources for a Wikipedia article, I'm not interested only in what The New York Times, or Reuters, or the WSJ, or any other individual news outlet has to say on it (since Wikipedia articles are supposed to summarize all the reliable sources on a topic, not just a single source). So the main link I want to find news coverage is the Google News search, which will turn up those outlets as well as any others. And lucky me, that link already exists in the list, along with other links to places that collect works from many different publications (like JSTOR). The NYT link long stood out as the sole link to an individual source, and frankly including it was a mistake from the beginning. The way to remedy it is to remove it, along with the recently added link to AP, not to add more and more links to try to achieve some sort of balance.

What we're seeing above is the start of a path we don't want to go down, where we establish a new "worthy of Find sources inclusion" tier of source reliability and spend countless hours debating which sources to include in it and end up listing every newspaper of record across the globe to avoid the (legitimate) fears of geographic bias. Let's turn back from that and establish a simple standard that avoids all that ugliness and comports with how people actually do search: Find sources is for collections of sources, not for individual news outlets.

  • Support as proposer. {{u|Sdkb}}talk 17:02, 24 November 2023 (UTC)Reply[reply]
  • Support: hell, same. jp×g🗯️ 20:39, 24 November 2023 (UTC)Reply[reply]
  • that's exactly my point when i raised my objection to nyt back on 13 July 2023 Wikipedia:Village_pump_(proposals)/Archive_203#Replace_nyt_with_reuters. there had never been any argument for why it's chosen over all other sources. yet it took 125 days(!) for enwp to come to a conclusion of adding AP and not removing nyt, and no one has over the 125 days come up with argument for why nyt was chosen over other organisations but only some condescending jokers kept lecturing me.

    so here they go, to balance out the 2 american sources, at least 2 non-american must be added. :) RZuo (talk) 07:12, 25 November 2023 (UTC)Reply[reply]

Find sources: Google (books · news · scholar · free images · WP refs· FENS · JSTOR · NYT · AP · TWL
  • This is why I created {{search for}} to be a comprehensive suite of searches that one can pull out the type of searches that one wants from. And if there's something useful to add to it, we can certainly consider it.

    But {{find sources}} isn't the same thing. In the event that the multiple searches of {{search for}} are needed, simply replace {{find sources}} with {{search for}}, which is definitely the workman's multitool here. Otherwise {{find sources}} really should be focussed upon a few broad search engines. It's a spork to the multitool.

    It's somewhat questionable that it gives such prominence to Google, which is another reason that {{search for}} exists, but the very reason that I designed {{search for}} as a series of collapsing boxes is that if you add everything in a single line it becomes enormous. I tried that with {{search for}}.

    Something that is a single line mid-dot-separated list needs to be minimal, and if you keep adding more and more useful specialist searches to it you'll end up reinventing {{search for}} but badly.

    Uncle G (talk) 10:44, 25 November 2023 (UTC)Reply[reply]

I can imagine Search for replacing Find sources. In fact, I think you could make the tool even better by integrating some tools like WP:WRS, by expanding the scholarly articles section, by integrating the subject-specific recommendations or by emphasising you may get help to find your source. But that one is very good and practically fulfills the same purpose. Szmenderowiecki (talk) 23:25, 25 November 2023 (UTC)Reply[reply]
Comment. I just spent at least three or four minutes at maximum zoom trying to figure out whether bits three and seven (from the left) in the "multitool" photo are star drive, and I can't tell and it's upsetting my professional sensibilities. Even if they are, with only two included you're bound to run into screws you can't undo, either T15 or T20. Folly Mox (talk) 09:41, 26 November 2023 (UTC)Reply[reply]
  • Support but replace the privacy-violating Google link by a link to either the Searx meta-search news link at, and/or (alternatively, it should be possible technically to set up a round-robin selection from the best-rated Searx instances listed at, or to (though I don't know of how to directly link to the News section there). Startpage also protects privacy, so would satisfy UCOC, but it does do some advertising, so that would count as advocacy conferring financial benefits, as in the case of Google, with a financial motive for search engine bias in its search results, affecting the neutrality of our finding sources. Boud (talk) 13:16, 25 November 2023 (UTC)Reply[reply]
  • Support and I see no issue in a longer but more informative template. I only see the use of replacing Google with other browsers if they offer a comparable quality of search or it's slightly inferior but otherwise usable and respects privacy better. Privacy isn't a goal in and of itself, so we must weigh tradeoffs. Szmenderowiecki (talk) 20:24, 25 November 2023 (UTC)Reply[reply]
    Yeah, and we already have (sorta defunct) WP:WRS, which was supposed to be a one-stop tool to search all news.
    My general raw idea, w/o coding and so on, is here.
    Keep the number of links at the minimum, maximise searching within one custom search engine.
    It's not really possible with peer-reviewed articles, but we have Google Scholar and equivalents for that and anyway we should strive to use the best sources we have, right? Szmenderowiecki (talk) 21:10, 25 November 2023 (UTC)Reply[reply]
    @Szmenderowiecki: Browsers are not search engines. Google's version of its web browser is getting worse and worse in terms of privacy violation, but that's an independent issue. Boud (talk) 16:31, 26 November 2023 (UTC)Reply[reply]
    Yeah, I mistook this word, but the arguments stand Szmenderowiecki (talk) 18:42, 26 November 2023 (UTC)Reply[reply]
  • Support for reduction of clutter. — Bilorv (talk) 20:25, 25 November 2023 (UTC)Reply[reply]
  • Support per proposer, reduces clutter Sohom (talk) 20:55, 25 November 2023 (UTC)Reply[reply]
  • Support strongly, especially since there's no reason to emphasize news sources by including many of them - for the vast majority of uses of {{find sources}} news sources might not even be right (let alone US based ones). Galobtter (talk) 21:00, 25 November 2023 (UTC)Reply[reply]
  • Neutral: ok to remove individual outlets but it doesn't help much when there's a link to Google News which contains all sort of trash. Also support removing JSTOR as another individual outlet that has no business being here alone. Support replacing the links to NYT and AP with a search engine which contains them (preferably with a small manual selection of outlets rather than an automated crawling of news-like websites). Nemo 07:21, 29 November 2023 (UTC)Reply[reply]
  • Support removing individual news outlets. Eddie891 Talk Work 15:32, 29 November 2023 (UTC)Reply[reply]
  • Support. Less is more. This is used very widely via templates, and what gets included in it should be of exceptionally broad and global applicability. There should only be tools to actually "find sources" and not any individual source or publisher. Adumbrativus (talk) 04:19, 30 November 2023 (UTC)Reply[reply]
  • Support per Sdkb. This is a slippery slope. Retswerb (talk) 01:31, 1 December 2023 (UTC)Reply[reply]
  • Support per nom. Ajpolino (talk) 01:54, 1 December 2023 (UTC)Reply[reply]
  • Support per nom Cgallagher2121 (talk) 08:59, 1 December 2023 (UTC)Reply[reply]
  • Strongly support per nom, Galobtter and particularly the desire to reduce banner bloat. Remagoxer (talk) 15:30, 1 December 2023 (UTC)Reply[reply]

Replace the generic link[edit]

The generic link to Google violates Wikipedians' privacy (storing detailed private data for the purpose of sales to advertisers), which is contrary to the spirit of UCOC; like any individual search engine, it is subject to search engine bias that biases our selection of sources, and it uses filter bubbles targeted to each individual, tending to amplify existing demographic biases in Wikipedia. We could either give a link to list of search engines or choose a meta-search engine that gives high-quality general search results while protecting user privacy, reducing the bias to any particular engine, and avoiding filter bubbles. Boud (talk) 13:48, 25 November 2023 (UTC) Clarify: this section is only for the generic link, not for the specific links for news, books or scholarly sources. Boud (talk) 16:10, 26 November 2023 (UTC)Reply[reply]

  • Support as proposer. Suggest either (1) list of search engines, or (2) the Searx generic link at, and/or, or if a pseudo-random generator can be linked up to the module (should not be difficult with e.g. /dev/urandom, which is fast), use a round-robin selection from a list of e.g. 10 of the best-rated Searx instances listed at, or (3) The round-robin solution with Searx would keep the link compact (5 characters) and would statistically reduce the bias of any individual Searx instance implicit in the way it is configured and run. Boud (talk) 13:48, 25 November 2023 (UTC)Reply[reply]
  • Oppose. Google Search has plenty of problems that make it non-ideal, but it has over 90% market share. At that level, it's what editors expect, so providing anything else would go against WP:ASTONISH. Google also leverages its market dominance to provide better results in some cases, and editors' familiarity with it makes it easier for them to use. The metasearch engine idea is intriguing, but I wasn't impressed when I tested it just now. Searching for "Wikipedia" and navigating to the news tab produced results like this. Linking to the list of search engines is a nonstarter. The entire point of these links is to make searching for sources easier (to encourage more people to do it or just to add convenience). The current setup goes instantly to the Google results for the topic, whereas linking to the article would then require people to navigate to the search engine they want, click through to its article, click the external link to its site, and then re-enter the search term. At that level of inconvenience, people are just going to type the search query into their browser instead, bypassing the template and making it useless as a convenience aid. {{u|Sdkb}}talk 17:34, 25 November 2023 (UTC)Reply[reply]
    • Just to check what the actual arguments presented here are against the Searx proposal. They seem to be: (1) Google is dominant and it's what people expect; (2) anecdotal evidence. Boud (talk) 16:26, 26 November 2023 (UTC)Reply[reply]
      As of September, Searx's github repo is no longer maintained. I'm not real familiar with the project, but surely that's bound to be an issue moving forward if we were to use it as the sole replacement for Google search? Any replacement for the sole link should be something with staying power. Retswerb (talk) 01:29, 1 December 2023 (UTC)Reply[reply]
  • Oppose. Privacy isn't a goal in and of itself. I want to see the balance we trade between utility and privacy. If otherwise equal, obviously switch from Google but prefer utility in the balancing equation. Szmenderowiecki (talk) 20:24, 25 November 2023 (UTC)Reply[reply]
    • If we balance between utility and privacy, then privacy is a goal (among others). Boud (talk) 16:33, 26 November 2023 (UTC)Reply[reply]
      A goal that I believe is secondary to utility, which is the primary goal (find as many sources as you can). You can believe we should prioritise privacy even to the detriment of utility, which is fine but I think few people will share your view.
      Also, {{search for}} already includes a couple of search engines outside Google. You can suggest a couple more there. Theoretically if there is independent confirmation that is equivalent to Google but is more privacy-friendly, why not? We can change the link.
      But my testing of relatively obscure topics I mentioned (e.g. reasons why few people live in Tasmania) showed that most alternatives simply performed worse. Now whether Google (or Microsoft, Apple or whatever) abused its market dominance is basically irrelevant for me, because the point is to find data and (hopefully) let users exercise best judgment in choosing.
      I need to see that any of the SearX, Mojeek or other search engines are good enough to use. This also applies to searches in languages other than English, so if the engine is optimised for English but sucks in Russian, it's not good enough. Szmenderowiecki (talk) 18:41, 26 November 2023 (UTC)Reply[reply]
      In principle, OK to let users exercise best judgment in choosing, but the search engine bias in the results that the users have to choose from only makes it easy for them to use judgment within that biased selection. A mix of biases (via a meta-search engine) should tend to reduce the overall bias.
      Searx is not a search engine, it is a metasearch engine. Mojeek is a search engine.
      As for other languages, given that in English a very sort of notorious example is when the Google search engine was categorizing black people as monkeys per a Princeton engineering interview, the case of Timnit Gebru's exit from Google, and the Santa Clara University advice Search engines and artificial intelligence are neither neutral nor free from human judgment, I see no reason to trust Google to be better in non-English languages than in English. Financial reasons suggest the opposite: paying fluent speakers of small-population languages of poor countries won't contribute much to Google/Alphabet advertising revenue. Boud (talk) 19:35, 26 November 2023 (UTC)Reply[reply]
  • Oppose This proposal feels a bit wikilawyery especially bringing up the UCOC which has nothing to do with this proposal. Given everything being equal, I would definitely support using alternative engines (or atleast giving users the options to do so). However, this is not the case, instead results from other engines tend to be inferior or outright non-existent for certain search terms. Additionally, a geographical bias can actually sometimes help editors who live in specific regions find better sources (Annecdotally, I have a easier time digging up sourcing about Indian topics if I switch to my Indian registered internet connection when I am in other countries.) -- Sohom (talk) 21:09, 25 November 2023 (UTC)Reply[reply]
    If there's consensus to continue violating UCOC - and in fact I expect that there will be consensus to continue to at least partially violate UCOC by keeping at least one or two Google links - then we'll find that out. WMF will have to fight us if it wishes to fully implement UCOC. Recommending that other Wikipedians violate their privacy is disrespectful and risky, e.g. it's inconsistent with ensuring that the Wikimedia projects are productive, pleasant and safe spaces, and contribute to the Wikimedia mission. It's not safe to be encouraged to violate your personal privacy. To make an analogy: suppose you're invited to a Wikimedia community face-to-face workshop and on entry to the room, there's someone at the door who asks you to take off all your clothes. Whether Google's privacy invasion into your mind - your browser history and browser parameters - is worse or better than a violation of your bodily privacy is a matter of judgment, but both cases are privacy violations. Boud (talk) 19:57, 26 November 2023 (UTC)Reply[reply]
    @Boud You've lost me at the analogy. Giving up ones privacy is not akin to sexual harrasment, linking to the UCOC's harrasment clause and giving examples of sexual harrasment as a analogy is not going to change that fact.
    The {{find sources}} template links are just suggestions/convinience links for good places to find sourcing, you are welcomed to ignore it completely and use your own search engines (in fact I do so myself most of the times, even when using Google). I would liken it more to being offered a milk-coffee at a Wikimedia event, when I myself am lactose intolerant (I am not, but hypothetically speaking). Sohom (talk) 14:07, 30 November 2023 (UTC)Reply[reply]
  • Oppose Google search/news/scholar/books are very good for finding sources and any replacement has to have similar quality. Galobtter (talk) 21:11, 25 November 2023 (UTC)Reply[reply]
  • Oppose. A pretty normal WP:BEFORE would probably include Google news, Google books, and Google scholar. Removing links to Google would make this workflow inefficient. At a minimum, equivalent search engines that search news articles, books, and academic papers should be proposed. A general "let's get rid of Google" with no suggested replacement, in my opinion, is not the way to go. –Novem Linguae (talk) 02:00, 26 November 2023 (UTC)Reply[reply]
    • This proposal is only for the generic link. That is independent of the specific links for news, books and scholarly sources. The above section (so far) seems to be only about news links. Boud (talk) 16:10, 26 November 2023 (UTC)Reply[reply]
  • Support replacing Google search with (pretty much) anything else. Linking Google means we're letting advertisers decide what ends up used as source in Wikipedia, an obvious source of systemic bias. If people think a direct link to a web search is needed for people's workflows, DuckDuckGo would be an improvement on the current state. DuckDuckGo introduces less systemic bias because it generally doesn't personalise results based on user fingerprinting and doesn't serve automatically generated prose or other non-sources into the "search results" page. Nemo 07:12, 29 November 2023 (UTC)Reply[reply]
  • Support the idea of replacing the generic google link with some sort of list of search engines. We already do this when providing links to ISBNs and co-ords. Rather than picking what search engine people use, we can present them with a list of options. I'm not convinced any Google alternative is actually good enough to fully replace the link to google, especially recognizing its sheer market dominance (it will be what many people expect to find and use). Eddie891 Talk Work 15:36, 29 November 2023 (UTC)Reply[reply]
  • Oppose replacing Google: The top searches on Bing, and likely most other search engines is "Google". Meet the reader where they're at. Mach61 (talk) 05:41, 30 November 2023 (UTC)Reply[reply]
  • Oppose As much as I dislike the corporation, it is the best overall source of sources. O3000, Ret. (talk) 19:38, 1 December 2023 (UTC)Reply[reply]
  • Oppose otherwise WP:V is compromised. ~~ AirshipJungleman29 (talk) 21:06, 1 December 2023 (UTC)Reply[reply]

Discussion (Module:Find sources)[edit]

There are two orthogonal issues here, reliability and coverage of sources (which is core encyclopedia stuff) and search engines' respect for privacy (which is a user preference). This is unlikely to lead to a productive conversation. Personally I think we should recommend source-finding techniques on a per-wikiproject basis. We need to look in different places for sources for a contemporary American biography, a 20C New Zealand law, a 19C Malay biography, an 18C German political scandal, a 17C book and an ancient Middle Eastern location. Seems likely to me that we can come up with a per-user preference for generic search engine privacy and then parameters to help find specific content (bot-assisted based on cats and wikiprojects). Stuartyeates (talk) 19:14, 25 November 2023 (UTC)Reply[reply]

It's true that there are two orthogonal motivations, and you are probably right that a tech solution may be able satisfy both. The idea of a per-user preference parameter for search engine privacy, that would be used by the module to switch between privacy-violating and privacy-respecting search engines and meta-search engines, is good. This would need techie willingness to implement it. There would also be the question of which setting should be the default: should selecting privacy be opt-in or opt-out? Boud (talk) 20:13, 26 November 2023 (UTC)Reply[reply]

If we think this is going to be a popular subject (~50 editors?), please copy/paste it to a subpage before adding an RFC tag. This page was recently nearly a million bytes long, and that makes it very difficult for people to read (especially on smart phones). The most popular title is Wikipedia:Requests_for_comment/YOUR-THING-HERE (see examples), but it's okay to do Wikipedia:Village_pump_(proposals)/YOUR-THING-HERE if you prefer. WhatamIdoing (talk) 18:59, 24 November 2023 (UTC)Reply[reply]

I think this first should have an RfC tag before we move it to a subpage of VPR. People won't know of this discussion if we hide it in a subpage. Szmenderowiecki (talk) 21:12, 25 November 2023 (UTC)Reply[reply]
If it's an RFC, people will know about it no matter where it happens. The Wikipedia:Feedback request service posts personalized messages to editors' talk pages about RFCs in subject areas that interest them. But I agree (and, more importantly, so does WP:RFC) that it would be good to keep a link here, especially if the discussion starts here and gets moved. WhatamIdoing (talk) 23:20, 26 November 2023 (UTC)Reply[reply]
Maybe this should be done reactively rather than proactively. That is, wait for sections to get big, then put them on a subpage. –Novem Linguae (talk) 02:05, 26 November 2023 (UTC)Reply[reply]
@Novem Linguae, do you have a preferred number for "big"? There are already more than 50 comments in this section. WhatamIdoing (talk) 23:22, 26 November 2023 (UTC)Reply[reply]

Please consider transcluding or posting a link to what we're discussing here. Visiting Module:Find sources/templates/Find general sources doesn't show much, just some code. How does that code render? –Novem Linguae (talk) 02:04, 26 November 2023 (UTC)Reply[reply]

{{Find sources}} Szmenderowiecki (talk) 09:08, 26 November 2023 (UTC)Reply[reply]
The template renders as "Find sources: Google (books · news · scholar · free images · WP refs· FENS · JSTOR · NYT · AP · TWL". Boud (talk) 19:36, 26 November 2023 (UTC)Reply[reply]

Proposal for change to copy of "has been changed" emails[edit]

Currently, the message sent to anyone who subscribes to edits on a page reads as below, emphasis mine:

There will be no other notifications in case of further activity unless

you visit this page while logged in. You could also reset the

notification flags for all your watched pages on your watchlist.

I propose it be changed to : You can also reset the

notification flags for all your watched pages on your watchlist.


Dreameditsbrooklyn (talk) 01:08, 28 November 2023 (UTC)Reply[reply]

Who can change this? Dreameditsbrooklyn (talk) 01:08, 29 November 2023 (UTC)Reply[reply]
The text is in MediaWiki:Enotif body. Changing it should be done in the MediaWiki code, not locally at the English Wikipedia. You can ask at Wikipedia:Village pump (technical) for help filing the technical request on phab:
@Trizek (WMF), I suspect that one of the teams you deal with would normally triage such requests. WhatamIdoing (talk) 04:52, 30 November 2023 (UTC)Reply[reply]
@WhatamIdoing Probably not one of mine, as it seems to be Mediawiki's code. Documenting it on Phabricator will be followed by a triage to the right team. Trizek_(WMF) (talk) 14:12, 30 November 2023 (UTC)Reply[reply]

I have came up with new consistency in some articles of chemical compounds. Gain consensus from most other readers and helpful volunteers staying.[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
There's enough WP:SNOW in here to build a very large snowman, so I'm closing this as snow oppose to avoid wasting any more editor time. (non-admin closure) Edward-Woodrow (talk) 22:50, 29 November 2023 (UTC)Reply[reply]

Hi, I have brought up two clean-up changes to fit the consistency of whole non-organic compounds articles including: Hydrogen fluoride, Hydrogen chloride, Hydrogen bromide and Hydrogen iodide. I'm keen to keep the lead sentence of each context mentioned to follow the form: [Chemical compound name] is an inorganic chemical compound; it has the formula [Chemical symbol], linking formula and bolden the symbol. I've made two significant changes to Hydrogen iodide and both were restored to last revision before me. Yes, I've referenced WP:CHEMNAME MOS policy and I have known and declared that the formula should be styled in template in accordance. Any thoughts from you? 2001:EE0:4BC2:C190:74A5:B633:51B2:265D (talk) 03:14, 29 November 2023 (UTC)Reply[reply]

What response did you get when you asked at WT:CHEMISTRY or WT:CHEMICALS, the projects that have lots of experts in the subject and editors interested in its articles? You will need to clarify how this proposal is different than how articles generally are currently, and why the bolding of the formula. DMacks (talk) 04:19, 29 November 2023 (UTC)Reply[reply]
Looks like there is some discussion at Talk:Hydrogen iodide#Consistent grammar wording where 2001:EE0:4BC2:C190:74A5:B633:51B2:265D was told to bring the question/suggestion here. Jo-Jo Eumerus (talk) 09:46, 29 November 2023 (UTC)Reply[reply]
So, give me more perspectives. I want to clarify something you are encouraged to say. 2001:EE0:4BC3:23D0:74A5:B633:51B2:265D (talk) 10:44, 29 November 2023 (UTC)Reply[reply]
So I want to make sure: what kind of formula would you like to use? In inorganic chemistry or for simple organic chemistry compounds that's often pretty straightforward and no one really writes otherwise (CuSO4·5H2O, CH3CHO), but I would like to see how you apply this to chlorhexidine, amanitin, heme or methylene blue, or probably a simpler case, choline.
Besides, all formulas are already in the infobox, and, for more complicated cases, it's the molecular structure that matters, molecular formulas often tell little in terms of isomers and functional groups.
If your idea is to make simple formulas only according to the template, where's the cutoff? I dread the prosect of some chemistry fans editwarring over the cases when (not to) use the template, or over the preferred types of formula.
I may consider supporting it if you give clear rules about using this formula. Szmenderowiecki (talk) 12:12, 29 November 2023 (UTC)Reply[reply]
The proposal states this is for "non-organic", so of your examples only CuSO4·5H2O would be in-scope. DMacks (talk) 12:59, 29 November 2023 (UTC)Reply[reply]
Oh, didn't notice it, thanks for turning my attention to this word. Szmenderowiecki (talk) 13:23, 29 November 2023 (UTC)Reply[reply]
In my view, bolding the formula symbol to prevent the readers from losing attention and might quickly get confusion about the chemical formula of the chemical compound we are directing the user to. 2001:EE0:4BC3:23D0:B58E:E5C:90B0:54BE (talk) 13:54, 29 November 2023 (UTC)Reply[reply]
Oppose this proposal. It is not consistent with the Wikipedia:Manual of Style/Chemistry/Chemicals#Introductory paragraph essay written as a supplement to MOS guidelines and with support of the WTCHEM folks because it forces lumping of all non-organics together as a monolithic single class rather than the "or more specifically" option. That essay seems to give better flexibility than this proposal when a chemical can be more specifically classified without getting too technical. So it seems to force being overly superficial (not well in keeping with MOS:LEADSENTENCE). More importantly, "non-organic" is not even a univeral or well-defined term (per numerous cited refs and discussions on associated articles). Experts have agreed to disagree, and all agree that there are numerous edge cases, so Szmenderowiecki's concern about "where's the cutoff" is actually a verifiable one via WP:RS. I will alert WT:CHEMS for you, so this discussion does not appear as an attempted end-run around them. DMacks (talk) 13:17, 29 November 2023 (UTC)Reply[reply]
Strong oppose as a policy, we don't strive for uniformity among articles as each article in a category will have different things that should be highlighted in the lead. Carbon and Silicon belong to the same group, but the information of each is drastically different and, as such, the opening paragraphs reflect that. Important details respective to each compound would be lost if we chose to adopt a uniform opening structure, and that directly contradicts basic policy set out in MOS:LEAD. microbiologyMarcus (petri dishgrowths) 13:37, 29 November 2023 (UTC)Reply[reply]
(EC) Upon further consideration, I understand that you want to change the manual of style in this respect. In fact, if the point is that you want to show the chemical formula, it's already in the infobox and there is in general no need to repeat it. Other than that, I fully agree with DMacks, flexibility is often a needed thing when making a proper lead. I don't think I can ever recall some rigid template for any category of articles I edited, and I see no compelling reason to introduce it here. Szmenderowiecki (talk) 13:50, 29 November 2023 (UTC)Reply[reply]
I would care to receive your comments more before I jump in the arguments. 2001:EE0:4BC3:23D0:B58E:E5C:90B0:54BE (talk) 13:52, 29 November 2023 (UTC)Reply[reply]
@Szmenderowiecki: hang on, I'm not proposing any MOS changes. I want differences between articles (not to get artsy, but I think it's beautiful). My response to the IP opener in the article talk page linked above was in regard to one article they proposed changes to, not to a broad swath of articles—my suggestion there was a compromise from the revision they had made, wanting the chemical formula in the lead. Once I realized they had proposed mass changes, I informed them that they would need to seek wider consensus as it didn't pertain to one article. microbiologyMarcus (petri dishgrowths) 14:08, 29 November 2023 (UTC)Reply[reply]
My response was to the IP rather than to you. It's that your indent is a little confusing (as it was obviously an answer to IP's proposal). FTFY. Szmenderowiecki (talk) 14:16, 29 November 2023 (UTC)Reply[reply]
Understood, mea culpa. Thank you kindly! microbiologyMarcus (petri dishgrowths) 14:38, 29 November 2023 (UTC)Reply[reply]
Oppose. The Chemicals subproject of the Chemistry project is doing fine without a "big plan" from someone with negligible experience in chem topics and article creation. Chemistry is NOT as simple as HX (X = halide) and black/white demarkation between various compounds types. So the idea is doubtless well intentioned, but I recommend doing a few thousand edits in the area and then reporting back.--Smokefoot (talk) 14:08, 29 November 2023 (UTC)Reply[reply]
Strong oppose. This IP editor from Vietnam seems to be same one whose earlier proposal about consistency in element articles did lead to a useful consensus (see this closed discussion). However, all the actual work of implementing that consensus was done by others (myself and User:Praseodymium-141 mainly). I am opposed to arbitrary attempts to create further cross-article consistency for topics where the effort of doing so, including the effort of gaining consensus, is much more than the benefit to readers. This specific proposal is particularly poor, as the molecular formula of simple inorganic compounds is not always needed in the lede sentence. Mike Turnbull (talk) 17:18, 29 November 2023 (UTC)