:::Hi, I'm glad to answer your questions. I wasn't sure what was permissible by Wikipedias idea of not posting personal information. We are a team of three students working on a project as part of this class [https://web.stanford.edu/class/cs224v/]Conversational Virtual Assistants with Deep Learning. We are focused on a few things: Exploring if we can reduce LLM hallucination to negligible amounts, exploring if LLM can generate articles with the same high quality as Wikipedia, and exploring if article based information gathering will remain relevant with the advent of informational chatbots.
:::Hi, I'm glad to answer your questions. I wasn't sure what was permissible by Wikipedias idea of not posting personal information. We are a team of three students working on a project as part of this class [https://web.stanford.edu/class/cs224v/]Conversational Virtual Assistants with Deep Learning. We are focused on a few things: Exploring if we can reduce LLM hallucination to negligible amounts, exploring if LLM can generate articles with the same high quality as Wikipedia, and exploring if article based information gathering will remain relevant with the advent of informational chatbots.
:::Filling out the form is obviously not a binding agreement. We would appreciate you filling it out and we can contact you with more details. To be clear, the intent of this is not to build a model that will be used on or with Wikipedia. Rather it is in the scope of the class and of improving LLM models. [[User:Terribilis11|Terribilis11]] ([[User talk:Terribilis11|talk]]) 01:18, 9 November 2023 (UTC)
:::Filling out the form is obviously not a binding agreement. We would appreciate you filling it out and we can contact you with more details. To be clear, the intent of this is not to build a model that will be used on or with Wikipedia. Rather it is in the scope of the class and of improving LLM models. [[User:Terribilis11|Terribilis11]] ([[User talk:Terribilis11|talk]]) 01:18, 9 November 2023 (UTC)
::::Oh wow ok it is... significantly context-altering to know you are in a 2nd-year undergraduate course. New question: does your professor know that you are recruiting people who are not Stanford students as scorers? The way you've gone about this recruiting set off a lot of my "these people are acting completely without IRB oversight, as though they do not know they need it or for some reason hold it in contempt" alarm bells, which is understandable because you... are students, and indeed probably have never been told about ethics approval at all. What you are proposing is a study with human participants, and, barring some strictly defined cases, studies with human participants require ethics approval and data security standards. Since this is obviously a study with only extremely low risks for your human participants, you can probably get ethics approval expedited, but you probably need it and your professor can get into big trouble if they don't get it. (It's your supervisor's responsibility, so here that's your prof - you're not in trouble.) The compliance office at Stanford closes to new submissions on the 13th ([https://researchcompliance.stanford.edu/panels/hs]), so you need to work quickly to ensure you're ok here. -- [[User:Asilvering|asilvering]] ([[User talk:Asilvering|talk]]) 07:05, 9 November 2023 (UTC)
::The form appears to miss-understand the Good Article requirements. The zeroth GA requirement is that it's an article. The first sentence on the page ''A good article (GA) is an '''article''' that meets a core set of editorial standards[...]'' [[User:Stuartyeates|Stuartyeates]] ([[User talk:Stuartyeates|talk]]) 03:29, 9 November 2023 (UTC)
::The form appears to miss-understand the Good Article requirements. The zeroth GA requirement is that it's an article. The first sentence on the page ''A good article (GA) is an '''article''' that meets a core set of editorial standards[...]'' [[User:Stuartyeates|Stuartyeates]] ([[User talk:Stuartyeates|talk]]) 03:29, 9 November 2023 (UTC)
:::I'm not quite sure what you mean. The results of our model will be an article or rather in the format of an article. We won't be publishing the articles on Wikipedia.[[User:Terribilis11|Terribilis11]] ([[User talk:Terribilis11|talk]]) 05:40, 9 November 2023 (UTC)
:::I'm not quite sure what you mean. The results of our model will be an article or rather in the format of an article. We won't be publishing the articles on Wikipedia.[[User:Terribilis11|Terribilis11]] ([[User talk:Terribilis11|talk]]) 05:40, 9 November 2023 (UTC)
This page is used for the administration of the Articles for Creation or Files for Upload processes and is therefore within the scope of WikiProject Articles for Creation. Please direct any queries to the discussion page.Articles for creationWikipedia:WikiProject Articles for creationTemplate:WikiProject Articles for creation (admin)AfC project pages
Graphs are unavailable due to technical issues. There is more info on Phabricator and on MediaWiki.org.
Graphs are unavailable due to technical issues. There is more info on Phabricator and on MediaWiki.org.
List: Copyvios
Please see below the detected copyvios which have a 30%+ probability from Earwigs. Updated every 12 hours at midday and midnight UK time (and sometimes manually run by me). Each run a random 200 drafts are checked + the existing detected copyvios (to see if they have been fixed)
EDIT: I have adjusted RichBot to remove a check from any article which has been declined/rejected for whatever reason, but was still in the existing copyvios list. If the draft is re-submitted, it will be rechecked
Copyvios In Drafts(Beware Of Mirrors!) - Do You Use This Table? Please Let My Operator Know :)
As for me, I think it is best to have placed in a one of the steps/pages within the article wizard so that no single page is overloaded with text. Catalk to me!20:15, 16 October 2023 (UTC)[reply]
As you say "long of text can lead to banner blindness" I think the warning can be a lot shorter... I'm too tired to think deeply about it now but I'm thinking more like the banner of Wikipedia:Articles for creation/Scam warning. The only thing I would add is that if I was to read it I would assume it was actual real reviewers/admins are involved, so I think adding something like "the person who contacted you will often be impersonating a real reviewer or administrator". KylieTastic (talk) 20:57, 16 October 2023 (UTC)[reply]
@Ca: Not sure if this is still an ongoing issue, but I feel placing it at https://en.wikipedia.org/wiki/Wikipedia:Article_wizard/CreateDraft would be best. I've noticed you've moved up the scam warning on the AFC page, but other places linking to the article wizard (WP:YFA, for exmaple) don't have this warning. Regarding the content of the warning, I think the shorter the better, so a shortened version of your example would be fine. Maybe
WARNING! There is an ongoing scam designed to target new users on Wikipedia! Users impersonating volunteers will ask for payment in exchange for assistance, or to have a draft page published. The assistance provided to you by legitimate Wikipedia editors will always be completely free of charge. Real Wikipedia editors will never contact or solicit you for payment or compensation of any kind. See this scam warning for more information.
"Where should we place the scam warning?" How about in newspapers and business magazines? All over social media? On the Main Page, maybe in big, bold, blink text? The people who need to see this aren't the ones trying to write articles. WhatamIdoing (talk) 01:52, 24 October 2023 (UTC)[reply]
I see your point: a warning on the Article Wizard does not prevent scammers who contact victims in advance, just like what WikiPR did. It's impossible to prevent those scams without going into ridiculous lengths as you said. However, this scam warning is to prevent scammers who target editors who already had their draft declined. Scammer in actionCatalk to me!03:46, 24 October 2023 (UTC)[reply]
Okay. Then we need both, and possibly in the regular interface, too. Maybe a very brief version should also be added to a message like MediaWiki:Editpage-head-copy-warn, using CSS so that's it's visible only to new accounts (or maybe IPs). There could be some benefit to having your nephew know that articles are free. WhatamIdoing (talk) 15:13, 24 October 2023 (UTC)[reply]
I'm not so sure we should rush into adding warnings to the base MediaWiki interface. Sdkb put it quite succinctly earlier when we were discussing where to put the template warnings; new editors have a myriad things that we think they should know, but the more information we throw at them, the less likely they are to pay attention to any of it. It's also important that we exercise some restraint instead of making these kinds of changes as a knee-jerk reaction — unless evidence emerges that article creation scams are so widespread that editors need to be warned about them every single time they edit a page, I don't think the broader community is likely to accept these changes. —TechnoSquirrel69 (sigh) 18:07, 24 October 2023 (UTC)[reply]
Seconded. A warning on the article wizard should be sufficient. A more explicit mention on the decline warning might work as well. It's currently linked, but it's called "untoward behaviour" which isn't exactly very clear. ARandomName123 (talk)Ping me!18:19, 24 October 2023 (UTC)[reply]
"Untoward behaviour" not only isn't very explicit (for fluent English readers), I expect it means nothing at all for many less-than-fully-fluent readers of English. We should definitely improve that wording. -- asilvering (talk) 18:34, 24 October 2023 (UTC)[reply]
The character count isn't my concern here, I'm questioning whether this warning is needed for display in such a prominent and widespread view. This are article creation scams, and are most likely not relevant to most other editors who aren't engaging with AfC. —TechnoSquirrel69 (sigh) 20:10, 24 October 2023 (UTC)[reply]
Prominent, widespread, and, if you look at what I suggested, only for non-autoconfirmed editors. That means you'd see it 10 times per account/lifetime, not every time anyone edits. WhatamIdoing (talk) 20:19, 24 October 2023 (UTC)[reply]
How do I reject a submission?
This HowTo repeatedly gives instructions/options to reject a submission but nowhere can I find a link to how this should or should not be done, never mind the actual instructions. Please can this be added (or, if it already somewhere and I mised it, given more prominence)? — Cheers, Steelpillow (Talk) 07:59, 18 October 2023 (UTC)[reply]
Steelpillow, are you talking about manually rejecting something, or using WP:AFCH? The "Decline/Reject" option in red in the middle of the AFCH window would (in my mind) be fairly obvious, though I do note the screenshot of the tool is outdated now (I'll work on replacing that). Primefac (talk) 08:03, 18 October 2023 (UTC)[reply]
Uh... so I also missed the bit about having to be an approved reviewer, my apologies. I'd still recommend a sentence explaining that you just need to follow the tool's instructions. (BTW, what brought me here was a submission which was unchanged from a previous rejection. Maybe that should be in the speedy reasons too? ) Anyway, I'll get my coat now. — Cheers, Steelpillow (Talk) 13:10, 18 October 2023 (UTC)[reply]
Hey Steelpillow, as you have good editing experience including a number of articles and a good AfD "matched result" rate you could just ask to be added to the participant list at WP:AFC/P as it looks like you easily pass the criteria (although that would be up to an admin to decide). Note that all admins and new page reviewers get access to the tool by default. Then you can review any submissions that overlap with the areas your interested/working in. You may also find Wikipedia:AfC sorting useful to find submissions by area, but if this is wikiproject based you probably have your own lists such as Wikipedia:WikiProject_Aviation/Article_alerts#AFC. Cheers KylieTastic (talk) 13:51, 18 October 2023 (UTC)[reply]
On reviewing the current state of affairs after coming across Wikipedia:Village_pump_(technical)#Purge_API_broken?, I discovered that there are 874 pages at the moment that should have been in Category:AfC G13 eligible soon submissions but are missing, and also realized that this db query provides a scalable solution to the problem which is not being used by the existing purge bot. (We just need to purge the pages returned by this query, rather than all pages in AfC categories - which are significantly more in number.)
I have tried to decline Draft:Moor but every time I go to do so, Wikipedia freezes. I have no problem declining other AfC articles but this one triggers a glitch that causes me to need to restart my browser. I have the same issue using Chrome and Firefox and have even tried Chrome in Incognito to no avail. Spiderone(Talk to Spider)11:17, 24 October 2023 (UTC)[reply]
I'm proposing to replace it with the following line:
{{hidden|Where to get help ({{Plain link|url={{fullurl:Wikipedia:WikiProject_Articles_for_creation/Help_desk/New question|withJS=MediaWiki:AFCHD-wizard.js&page={{FULLPAGENAMEE}}}}|name='''ask us a question'''}} - '''[[Wikipedia:IRC help disclaimer|live help]]''' - [[Talk:<includeonly>{{FULL</includeonly>PAGENAME}}|ask a WikiProject]])
Perhaps it will help more contributors discover the WikiProjects and contact them for assistance, as otherwise most of them don't know how to check the WikiProjects list at the talk page.
Proposed change 2
The template, if clicked the 'unhide' button, says the following text there among other things which I think is outdated because with the newer software the new drafts are already tagged with WikiProjects, and directing the users to the article talk page would be more appropriate:
If you need feedback on your draft, or if the review is taking a lot of time, you can try asking for help on the talk page of a relevant WikiProject. Some WikiProjects are more active than others so a speedy reply is not guaranteed.
This should say:
If you need feedback on your draft, or if the review is taking a lot of time, you can try asking for help on the talk page of a relevant WikiProject, which are listed at the draft talk page [this should link to the draft talk page]. Some WikiProjects are more active than others so a speedy reply is not guaranteed.
So just to be pedantic about the first proposal, this is what is in the hidden message:
{{hidden|Where to get help|
* If you need help '''editing or submitting your draft''', please {{Plain link|url={{fullurl:Wikipedia:WikiProject_Articles_for_creation/Help_desk/New question|withJS=MediaWiki:AFCHD-wizard.js&page={{FULLPAGENAMEE}}}}|name='''ask us a question'''}} at the AfC Help Desk or get '''[[Wikipedia:IRC help disclaimer|live help]]''' from experienced editors. These venues are only for help with editing and the submission process, not to get reviews.
* If you need '''feedback on your draft''', or if the review is taking a lot of time, you can try asking for help on the [[Help:Talk pages|talk page]] of a [[Wikipedia:WikiProject#Finding a project|relevant WikiProject]]. Some WikiProjects are more active than others so a speedy reply is not guaranteed.
}}
This expands to
Where to get help
If you need help editing or submitting your draft, please ask us a question at the AfC Help Desk or get live help from experienced editors. These venues are only for help with editing and the submission process, not to get reviews.
If you need feedback on your draft, or if the review is taking a lot of time, you can try asking for help on the talk page of a relevant WikiProject. Some WikiProjects are more active than others so a speedy reply is not guaranteed.
So I guess my question is, why do we need to put the links in the hidden message header when it's already in the hidden message? Primefac (talk) 13:03, 26 October 2023 (UTC)[reply]
Apologies for the delay, the message template has been updated: {{WPAFCDrive}}.
Hello WikiProject Articles for creation:
WikiProject Articles for creation is holding a month long Backlog Drive!
The goal of this drive is to eliminate the backlog of unreviewed articles with bonus points being given for reviewing drafts that have been waiting more than 30 days. The drive is running from 1 November 2023 through 30 November 2023.
Barnstars will be given out as awards at the end of the drive.
There is a backlog of over 1000 articles, so start reviewing articles. We're looking forward to your help!
active participants as at 08:06, 24 May 2023 (UTC), before the removal of participants with New Page Reviewer due to the enablement of this group of editors to access the AFCH script by default.
I don't feel like pinging all 780 NPR editors with the message as not all of them may be interested in doing AfC reviews. However, we can leave the above message on the NPR talk page with an additional line to encourage them to add themselves to a MMS list for future communications. – robertsky (talk) 03:37, 27 October 2023 (UTC)[reply]
An inspiration that just came to me after typing the message with respect to NPR editors. TODO: use Quarry to find out NPR editors and admins who have been using the AFCH script in the last 6 months. This eliminates the need for recompiling the list based on the removals of NPR and admins from the Participants list. – robertsky (talk) 03:41, 27 October 2023 (UTC)[reply]
Honestly, it might not be a bad idea to hit the entire NPR and entire admin lists. That could help recruit us some new AFC reviewers that wouldn't hear about this otherwise.
Maybe consider removing the "Note: If you are re-reviewing other's acceptances" paragraph at the bottom there. That little detail distracts from the main message.
Do we want to put in a request for a watchlist message? Might get declined because it requires a perm to AFC review, but if it's accepted it could help us recruit. MediaWiki talk:Watchlist-messages
I know I'm late to the party and missed out on the planning, but can we try to avoid indicating that we want to burn everyone out by nuking the entire backlog? Honestly I'd be happy to get down to <2mo on the timescale (i.e. shift the focus of the second sentence to encourage reviews of the oldest drafts). Primefac (talk) 08:24, 27 October 2023 (UTC)[reply]
@Primefac definitely, I have adjusted the text to be less aggressive on that front. We can afford to do so anyway as it seems that the number of pending drafts have been decreasing since the start of September.
For the admin newsletter will the following blurb be sufficient? The Articles for Creation backlog drive is happening in November 2023, with 3,400+ drafts pending reviews for in the last 4 months or so. In addition to the AfC participants, all administrators and New Page Patrollers can conduct reviews using the helper script, Yet Another AFC Helper Script, which can be enabled in the Gadgets settings. Sign up here to participate!
@Novem Linguae, good points. My initial hesitation to ping the NPR list was primarily due to a possible burnout of reviewers from there from this month's backlog drive on that side. For the admins, let's go the newsletter route, hopefully catching the next newsletter in time. – robertsky (talk) 08:39, 27 October 2023 (UTC)[reply]
I'd be happy with casting a wider net, and for the message to be sent whenever from now given how close we are to the start of November. Turnagra (talk) 17:49, 30 October 2023 (UTC)[reply]
I am part of a team at a University where we are building a LLM style model which will be given a topic and will generate different subtopics and then text in order to write an informative article. We are going to be using several different types of scoring mechanisms, but we would ideally like to have frequent wikipedia editors collaborate with scoring the articles.
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 also have an automatic scorer that will score 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 + Qualitiative comments). We would take a subset of our articles produced and score them by actual Wikipedia editors as a way to verify our scorer is within reason.
I have been dealing with some IRL issues the couple of weeks, and have been unable to devote the time and space to properly vet the new AFCH access requests. With the backlog drive coming up, as well as just the general idea that folks simply shouldn't be waiting that long to get looked at, I simply need to ask for help.
I have at least been able to go through and find that no one (currently) meets the quick-fail numerical criteria, but I need thoughts on whether someone is a) not meeting the "experience" criteria, b) is suitable for probationary membership, or c) should clearly be a full reviewer straight off. I'll put a subsection for those three options below, and if I don't hear anything before 1 Nov I will assume that there are no issues with the new applicants receiving a probationary membership (which, as a reminder, means nothing other than "any admin can remove the user for any reason") and will give them access. Thanks, Primefac (talk) 17:12, 29 October 2023 (UTC) And just as a note, I don't need to be pinged, I'm watching this page as well as subscribed to the thread.update: removing subthread, there are few enough left (thanks Spicy) that they can just be mentioned/discussed below easily enough. Primefac (talk) 07:24, 30 October 2023 (UTC)[reply]
Would there be any objection to other admins handling the requests? I should have the time to look at some of them today. Spicy (talk) 19:17, 29 October 2023 (UTC)[reply]
@Spicy just fyi, if you have another moment, there are a handful of people who have applied since the backlog drive started. To save anyone reading this the first round of checking: Styx & Stones (now Donnchadh4) does not have 500 mainspace edits, so can be declined easily. F.Alexsandr only barely meets minimum edit count. The other three (as I write this, Styx & Stones is the most recent) all easily meet the minimum edit count. -- asilvering (talk) 22:47, 4 November 2023 (UTC)[reply]
I declined the obvious one; of the four remaining:
R. S. Shaw: seems fine, good AfD stats, seemingly not as active recently as many years ago but no big red flags.
Micheal Kaluba: see #Query below (and details on his user talk); no AfDs that I can find.
F.Alexsandr: 506 mainspace edits, 50% on 2 AfDs. May have some experience at ru.wp.
LEvalyn: seems very strong all around, plenty of edits, good AfD stats, good interactions with new editors.
I plan on getting to these today at some point; my usual workflow is to deal with AFCH requests on Sundays. Thanks for the feedback on them :-) Primefac (talk) 07:41, 5 November 2023 (UTC)[reply]
Is AFC intentionally a higher standard?
I've been active at NPP for a while and saw your requests for help here and thought I might see if I could help. Both from my past observations and now from reading your instructions (including the later items on the flow chart) IMO the standards for passing AFC are higher than Wikipedia standards for existence of an article as implemented by AFD & NPP. By the latter, an article with significant problems / which needs significant work still passes if it meets wp:Not and wp:notability and doesn't have urgent problems like copyvio, whereas it would not pass AFC until further improved. If you agree, is this difference deliberate or does/did it sort of just happen? And if I were to pass AFC's more by the looser criteria described above, would that be considered to be doing the job improperly? Sincerely, North8000 (talk) 21:31, 30 October 2023 (UTC)[reply]
I posit that the majority of new articles are sub-standard and that AfC is a single barrier to maintain the standard. Having participating in this month's NPP backlog drive, some stuff is getting by NPP that probably ought not and the aggregate participants in AfDs are more inclusionist than they used to be. The only way to participate in a collective project is to follow the community's rules strictly constructed. To do otherwise is improper, and an arrogation of authority. Chris Troutman (talk)21:58, 30 October 2023 (UTC)[reply]
The core question is: What happens if an article is of low quality / substandard but the topic is suitable (per wp:not and wp:notability) for existence of an article? The latter is basically the passage standard at AFD and NPP. Sincerely, North8000 (talk) 23:02, 30 October 2023 (UTC)[reply]
@North8000 We should pass inherently notable drafts, but encouraging referencing is kind of important too.
@North8000 The broad premise is that an AFC reviewer should accept a draft that the reviewer feels has a better than 50% chance of surviving an immediate deletion process.
This almost certainly does not answer your question.
Early in their reviewing 'career' a new reviewer is likely to try for a higher quality that is strictly necessary. This may not be a bad thing, but it does tend to discourage new editors. That is not always a bad thing, but often is.
Sometimes experienced reviewers see the need to choose additional strictness on a case by case basis, often unwittingly. When I catch myself doing this I take a hard look at what I'm doing and why.
Drafts by paid editors to me fall into two camps. The good paid editor who is capable of producing mainspace ready material with one or fewer declines at AFC, and the incompetent paid editor who wishes to be spoon fed and gains a torrent of declines and whines about them. I make a conscious decision to hold these to a high standard, and, since they are bing paid, expect them to learn as part of their payment! I have digressed! 🇺🇦 FiddleTimtrentFaddleTalk to me 🇺🇦 23:13, 30 October 2023 (UTC)[reply]
It has often been said that AfC accepts articles with > 50% chance of surviving AfD based on the WP:AFCPURPOSE. In reality if someone took this literally I doubt they would last long. The trouble is there are many who think we should just act as a low bar junk filter and accept a lot more, but there are also many who will criticise a single accept they do not agree with and just re-dratify. In AfC we can review on what is presented and it's up to submitter to show notability so can look to have a higher standard, however any reviewer can choose to do a WP:BEFORE and accept on nobility. So North8000 you can review AfC as you would NPP by verifying notability considering maybe more than is presented or you can decline if the submitted sources do not yet show that notability. Personally I do both: sometimes decline and leave to the submitter to improve and sometimes if I think notable but the submission does not show I will accept but I personally choose to add the sources and improve before accepting, but you can just tag as you would in NPP. KylieTastic (talk) 23:37, 30 October 2023 (UTC)[reply]
OK, my point does depend on if someone only randomly selected submissions which most don't. Yes, if a reviewer has a general good review record you can (and should) afford the 51% chance accepts. As you have a stunning 51% accept rate (project average ~20%) and only ~6% deleted accepts (even before removing reasonable reasons for deletion) that would suggest you have selection criteria or maybe just your idea of 50% chance in one others would see as much higher... as it is very subjective. Regardless, your stats say your an AfC boss with that acceptance and quality rate plus doing this since 2014. Hats off to you. KylieTastic (talk) 19:22, 31 October 2023 (UTC)[reply]
It should be passing AfC if it could pass AfD (without new sources being added) and NPP would not draftify it. Basically, if the problems can be fixed "in normal editing", it goes through. If there's some kind of minor dealbreaker in the article (questionable unsourced paragraph, unsourced BLP info, etc) I tend to remove it myself and accept.
Everyone should keep in mind that having an article declined at AfC is incredibly frustrating and demoralizing for new editors. Though I confess I care a bit less about whether paid and coi editors are frustrated. -- asilvering (talk) 23:44, 30 October 2023 (UTC)[reply]
I would agree, and I think it's for a handful of reasons, which vary by editor - one being an intent to help teach people how to write better articles, one being (perhaps) a general distaste and resulting higher bar for coi editors (which you can see a few times in this thread), another the lack of BEFORE, and overall... I don't know quite the best way to word this, and I can't say for sure whether this is the case, but it seems like AfC reviewers, being fairly experienced editors, have seen many quality articles, and that familiarity can lead to disappointment in the reality of what gets written by those who aren't familiar with our policies or encyclopedic/formal writing in general. Again, not sure on that last point... but one last thing, there is a lot of junk that comes through AfC (especially if you look closer to the beginning of the queue) and that I think can color someone's perspective when looking at a draft - if 80% of what they look at is bad, it's easy to ABF and default to reviewing drafts from the view that they're proabably bad.
Thank you everyone for your expert advice which helps me gain a perspective. Of course I'll keep reading here but want to put in a thanks now. I've always assumed that there's another factor for the often defacto higher standard at AFC. Take for an example an article which would pass AFD/NPP but is otherwise in bad shape in many other areas. An AFC reviewer is sort of being asked to put their "stamp of approval" on an article (like "this is OK") but may be reluctant to do so for such an article. North8000 (talk) 14:29, 31 October 2023 (UTC)[reply]
@North8000 Honestly, I used to worry about NPP for this reason, since I frequently saw articles draftified that, as far as I could tell, met the basic AfC standard. I often leave a comment with suggestions rather than accepting outright if I notice something problematic that isn't an AfC decline reason. I'm glad to hear that NPP has relaxed somewhat. I did feel that I was seeing less of those draftified articles, but had simply assumed that was because I had switched my focus from the categories-by-topic lists to trying to get through the four-month-old articles instead. -- asilvering (talk) 17:24, 31 October 2023 (UTC)[reply]
I can't speak for NPP as a whole. Partially because it isn't a "whole" :-). Same with AFD. As a side note, the time limit for draftifying is shorter than the time of the que so that might explain the reduction. I think that the experienced NPP's evolve towards applying the standards properly by also following the norm, with a "typical OK AFD" being the guide as to being the norm. My comments about the typical AFC standard being higher than the typical NPP standard is from a lot of observations and involvement in situations. A lot of the AFC rejections are for article quality issues which would not be grounds for rejection under the NPP "should this article exist?" standard. Sincerely, North8000 (talk) 17:54, 31 October 2023 (UTC)[reply]
My 2¢, North8000, which you have quite possibly seen me voice before, is that I've had articles that were failed for having a single unreferenced sentence and I've had articles that were failed for claims of non-notability despite full length references about the subject being right there in the reference list, but not as formatted refs. Too much work to look at them, I suppose. It's because of repeated incidents like that that I avoid using AfC whenever possible and always recommend whenever anyone asks for them to not use AfC either. SilverserenC22:01, 31 October 2023 (UTC)[reply]
Thanks. I think that RFC is a great concept. However, if the AFC standard is higher than the AFD/NPP standard for existence of the article, then that sort of works against the viability of the AFC route. North8000 (talk) 23:58, 31 October 2023 (UTC)[reply]
Rejection
There has never been a consensus about Rejection, either about when a reviewer should reject a draft, or about should be done about the resubmission of a rejected draft. I think that there are three situations in which I think a draft should be rejected:
The draft is resubmitted tendentiously without improvement, or without significant improvement. The topic may not be notable, or the submitter may not know how to develop a good draft.
The draft would be subject to one of the speedy deletion criteria in article space, especially A7.
The draft is not an article draft, but is either a bad joke or some other sort of nonsense.
That is what I think.
I still don't know if there is a consensus about what to do about resubmission of rejected drafts. I think that some reviewers think that there is a consensus, but they have different ideas as to what the consensus is, which is no consensus at all. Robert McClenon (talk) 21:39, 31 October 2023 (UTC)[reply]
November Articles for creation backlog drive
Hello WikiProject Articles for creation:
WikiProject Articles for creation is holding a month long Backlog Drive!
The goal of this drive is to reduce the backlog of unreviewed drafts to less than 2 months outstanding reviews from the current 4+ months. Bonus points will be given for reviewing drafts that have been waiting more than 30 days. The drive is running from 1 November 2023 through 30 November 2023.
I see that there will be awards given at the end of the drive. That implies that there may be a way of scoring reviewers. In previous backlog drives, I have found the method of scoring reviewers to be arcane. Will the criteria for scoring be easy to understand this time? (Sometimes there have been additional activities in the backlog, such as reviews of reviewing, that I have found especially obscure. Fortunately I don't see that mentioned this time.) Do we get different numbers of points for acceptance and for decline? Do we get different numbers of points for 4-month-old drafts and for 2-month-old drafts?
Robert McClenon (talk) 21:21, 31 October 2023 (UTC)[reply]
I see that there are zero points for a review that is found to be incorrect. I can see several ways that a review could be said to be found to be incorrect. Obviously, that includes speedy deletion or deletion after AFD. What about acceptance after a decline? Robert McClenon (talk) 21:50, 31 October 2023 (UTC)[reply]
User:Asilvering - Yes. That is why I asked. That is the way that the process is supposed to work. An author submits a draft. A reviewer declines it, and says why they are declining it. The author edits the draft, possibly adding text and sources, and resubmits the draft. Another reviewer accepts it. That is the way that the process is supposed to work. So I wanted to be sure that that isn't considered to be a case of the first review being found to be incorrect. Robert McClenon (talk) 03:50, 1 November 2023 (UTC)[reply]
@Robert McClenon I think you may have found the answer to this already, but just to write it here for anyone else who is wondering: "review that is found to be incorrect" refers to the re-review process only. -- asilvering (talk) 19:24, 1 November 2023 (UTC)[reply]
To add a question, who is supposed to be doing "re-reviews"? Anyone at all? Or just the co-ordinator? I was alarmed to see a 100% decline rate on the top reviewer (currently GraziePrego), and the two declines I spot-checked look completely out of line to me. One decline is for unsourced statements (there are none), and another apparently for not including a birthday. -- asilvering (talk) 03:25, 1 November 2023 (UTC)[reply]
@Novem Linguae & @Ingenuity, for the re-reviews, I indicated as participants of the backlog drive are eligible to conduct the reviews. But from the bot codes, it seems that it doesn't discriminate on who can do the rereviews. Shall I indicate that anyone can do the rereviews? – robertsky (talk) 11:19, 1 November 2023 (UTC)[reply]
I think it would probably screw up the bot that is doing the counting, as well. But I'm not sure what it says about AfC if we agree a review is incorrect and then don't do anything about it. I don't want to resubmit and accept, because then the accept notice will go to me. And when dealing with the months-old backlog, the original editor may be long gone and never come back to resubmit it either. -- asilvering (talk) 18:07, 1 November 2023 (UTC)[reply]
I suppose a compromise might be to contact the first reviewer and have them undo their own work (?), though I can see how that might be impractical in many situations. It just wouldn't sit right with me to remove another editor's good-faith review, even if I believe they were misguided in their reasoning. —TechnoSquirrel69 (sigh) 18:15, 1 November 2023 (UTC)[reply]
If you think an article is acceptable no reason to wait/hope the submitter re-submits.... they have been put off they may not come back. fix the issue with a re-submit and accept. The only time I would not submit and accept is if an article has not yet been submitted and the author(s) have recently worked on it as they should be able to work in piece till ready. KylieTastic (talk) 19:58, 1 November 2023 (UTC)[reply]
If you think the review was definitely wrong deal with it. If it's a bad accept: AfD/draftify as appropriate; If it's a bad decline you think should have been accepted, re-submit as last submitter and accept. For instance on Draft:Ferdinand Jan Ormeling Sr. it was commented "Subject clearly passes WP:NPROF and should be accepted"... so why is it still not accepted? If I was the submitter I would be confused. General principle here is if any of us find and draft we clearly think is in an acceptable state (and has been submitted at some point) then accept it. KylieTastic (talk) 19:48, 1 November 2023 (UTC)[reply]
I didn't accept it because I was waiting on a response here for what is intended to be done with them as part of the backlog drive. How do you re-submit as last submitter? Hit the submit button and then edit the template by hand to have their username? -- asilvering (talk) 19:51, 1 November 2023 (UTC)[reply]
In the last AFC backlog drive I participated in, we told reviewers that re-reviews were mandatory but in practice we didn't penalize if they didn't do it. That worked fine to put a little pressure on folks to do re-reviews so that the re-review system wouldn't just be ignored/collapse. The days/week after the drive was re-review week. Points were recalculated to include re-review points, then barnstars went out.
I think re-reviews should be limited to drive participants. The idea is that they get extra points for re-reviews.
If a review is failed, I think it's fine for the failing reviewer to go fix it if they're confident that they're right and it's not just a borderline case. For example, if an AFC reviewer erroneously declines a draft on a professor because they don't pass GNG, and the re-reviewer finds they pass NPROF, then the draft should be accepted. It is good to fix incorrect reviews. –Novem Linguae (talk) 20:00, 1 November 2023 (UTC)[reply]
I shouldn't be here at all and I'm sorry I'm being intrusive but isn't doing declines at high speed the easiest way to win your award? Thincat (talk) 20:13, 1 November 2023 (UTC)[reply]
Invalid declines do not get you points if/when caught. More importantly any reviewer who was found to be doing such a bad job would be removed from the project and/or have NPP permission removed. KylieTastic (talk) 20:20, 1 November 2023 (UTC)[reply]
Yeah, I was confused by this as well, as I don't remember ever seeing an option to submit on another user's behalf. It was my understanding that the message is sent to the user provided in the {{AfC submission}} template, which could be manually edited in source if necessary. —TechnoSquirrel69 (sigh) 22:37, 1 November 2023 (UTC)[reply]
Theroadislong, you have to submit from the AFHC tool not from the submit button on the decline. The AFCH one lets you choose: 'most recent submitter', 'page creator', 'Yourself' and 'Someone else' KylieTastic (talk)
Thincat it is early days but acceptance rate is actually higher that the normal ~20% - Day one was ~25.8% and so far day 2 23.3%. Also you and anyone are welcome to be here and part of discussions on AfC (we are not a cabal). You raised a valid concern in the correct place, no intrusion at all. Cheers KylieTastic (talk) 15:27, 2 November 2023 (UTC)[reply]
I presume this has to do with more attention going to the end of the queue. All the really obvious blink-and-decline-it ones are dealt with on day 0. -- asilvering (talk) 22:50, 4 November 2023 (UTC)[reply]
"article is improperly sourced" decline text
The text of the WP:V decline notice currently reads "This submission is not adequately supported by reliable sources." Can we get some form of reference to WP:PRIMARY in there? The kind of thing I have in mind is when people cite, say, a birth certificate or court case - these are unimpeachably reliable but fall under "this article is improperly sourced". If we really want to keep the "improperly sourced" decline to specifically refer to problems of unreliable sources only, can we get a new decline for "overly reliant on primary sources"? (Or does this exist already and I'm misinterpreting it?) I know that a standard "non-notable" decline is probably intended to cover this kind of case (ie, the subject is not shown to be notable because there are not enough secondary sources to show it), but I often really want to avoid that one because the "non-notable" language is often very confusing or even offensive to new editors, and I often don't think it's appropriate to use it. That is, the text of the article makes a very good case that the subject is a notable person, but the sources are all or mostly primary. Imagine, for example, an article on Anne Frank sourced entirely to her diary, Karl Marx sourced only to Das Kapital, Douglas MacArthur sourced only to army records, or something. -- asilvering (talk) 17:20, 31 October 2023 (UTC)[reply]
Honestly, it might be worth it to add another decline reason so that reviewers can make the distinction between unsourced or nearly unsourced submissions and submissions that are sourced to unreliable sources, including primary sources. (That was a lot of "source" words.) Notability is a related but conceptually different decline reason. What do other reviewers think about this? —TechnoSquirrel69 (sigh) 02:05, 1 November 2023 (UTC)[reply]
I do, of course, but that doesn't help with the the "non-notable" language is often very confusing or even offensive to new editors, and I often don't think it's appropriate to use it bit, obviously. -- asilvering (talk) 03:36, 1 November 2023 (UTC)[reply]
Because many submitted drafts have the same deficiencies, I use canned explanations in templates, such as {{compsays}}, which is short for "the company says" (but we are more interested in what third parties say). Robert McClenon (talk) 03:45, 1 November 2023 (UTC)[reply]
I think something like "non-notable" would be better since that response is really confusing. I think "article is improperly sourced", or needs reliable sources often are unclear about whether source quality is the issue (article is perfectly fine but find better sources), vs not established notability with claims, etc. Immanuelle ❤️💚💙 (talk to the cutest Wikipedian)03:46, 1 November 2023 (UTC)[reply]
Very true, and I also sympathize with Robert McClenon's approach above, but I feel like this particular situation with sources is common enough that having a WikiProject-approved blurb could be beneficial. —TechnoSquirrel69 (sigh) 04:18, 1 November 2023 (UTC)[reply]
Backlog progress chart
As the graph module is still dead jim, I've created a basic progress chart that will need daily manual updates. Added to the top on this page and the backlog page. KylieTastic (talk) 15:07, 1 November 2023 (UTC)things you do while board in a Teams meeting[reply]
Our backlog is around 4,000, and the month long NPP backlog drive that just finished lowered their backlog by 4,000. Zero backlog here we come! –Novem Linguae (talk) 17:31, 1 November 2023 (UTC)[reply]
fyi, many of these are translations of zh-wiki articles by the Taiwan 1000 project or related editors and aren't described as such in the edit history. Reviewers should double-check to be sure they're not looking at an improperly attributed translation. -- asilvering (talk) 18:09, 1 November 2023 (UTC)[reply]
The problem is when it is dependent on non-english sources to establish wp:notability, you really need someone fluent in the language to evaluate them for GNG in a reasonable amount of time. North8000 (talk) 19:44, 1 November 2023 (UTC)[reply]
I just went through all of the pages that have been added to Wikipedia:Articles for creation/recent since the drive started and I found 25(!!!) articles that had no categories whatsoever, with at least twice as many BLPs with only 1-2 categories. Many of the entirely uncategorized articles moved to main space were moved by users with the autopatrolled right. For those unaware, if you have AP and you move a page from draft space to the main space, it's automatically marked as reviewed and does not enter the Special:NewPagesFeed. This means a member of the WP:NPP team does not catch these mistakes and that means these articles can sit with these issues for quite a while. I understand that we can't all know every category in existence, but there is no excuse for leaving articles entirely uncategorized. Let's all get it together (not naming names). Hey man im josh (talk) 19:41, 1 November 2023 (UTC)[reply]
@Hey man im josh Are these ones that have been up for a while or are you catching them as they come in? If I add categories, I always add them after accepting because I find hotcat a lot easier to deal with than the AFC review script. -- asilvering (talk) 19:48, 1 November 2023 (UTC)[reply]
I would change Josh's request to "Categorize or tag the pages you accept". I try to remember to tag the pages that I accept as {{improve categories}}. I think that the gnomes who categorize pages as needing categorizing can do a better job with the often-arcane system of categories. The reminder is a good idea, especially since the number of articles accepted will be higher than normal for the next few weeks as the backlog drive succeeds. Robert McClenon (talk) 20:09, 1 November 2023 (UTC)[reply]
Categorizing used to be a mandatory part of the NPP flowchart, but was made optional a year or two ago. Categorizing has never been mandatory for AFC. I think there are some gnomes that focus on categorizing and that these articles all eventually get fixed, so this may not be particularly urgent. –Novem Linguae (talk) 20:11, 1 November 2023 (UTC)[reply]
I understand it's not mandatory, but not adding any categories is lazy and half-assing things in a way that we shouldn't be. I think I was most annoyed by the instances of autopatrolled users doing it because we typically expect autopatrolled users to work on articles in a way that don't need a second set of eyes. Hey man im josh (talk) 20:15, 1 November 2023 (UTC)[reply]
I am an unapologetic half-ass then because more often then not I tag drafts I accept with the "uncategorized" or "Improve categories" tag. I have at times spent much more time trying to find appropriate categories than I did reviewing a draft so gave up and figured it's best left to specialists. S0091 (talk) 20:25, 1 November 2023 (UTC)[reply]
I disagree that even one or two cats should be a hard requirement. The hard requirement should be either two cats or a tag as needing category improvement. The accepting reviewer should be allowed to defer categorization to gnomes, providing that they mark the article as needing gnome review. Robert McClenon (talk) 04:04, 3 November 2023 (UTC)[reply]
AFCH seems to work with the Minerva mobile skin. I was able to open the accept menu. Can you go into a bit more detail about what's not working? –Novem Linguae (talk) 20:26, 1 November 2023 (UTC)[reply]
See screenshot. Is that the bug you're trying to report? (Workaround: tick "Desktop site" / "View in desktop mode" in your browser, which should turn off auto scaling) –Novem Linguae (talk) 21:57, 1 November 2023 (UTC)[reply]
I just ran into an editor, User:Micheal Kaluba, who is reviewing drafts and is not listed as a participant on the AFC team. He declined a draft that had already been rejected six months ago and hadn't been resubmitted by an editor. So, I don't know if they are just wandering through Draft space, commenting on random draft articles. I suggested that they come here and submit their name as a reviewer and I reverted their actions, one of which was to move a draft to main space (that action was undone by another editor). I guess my question is, I spend a lot of time in Draft space, reviewing drafts and sandboxes that are nearing their expiration date, what is the typical action that is done when an editor who doesn't work with AFC conducts reviews on their own? I know this isn't common but I have run into it before although typically it is editors approving their own drafts which wasn't happening in this case. Thanks. LizRead!Talk!02:43, 2 November 2023 (UTC)[reply]
If someone decides to move a AfC declined to mainspace with the intention of improving it there or as part of WP:DRAFTOBJECT, then that's allowed (even though often not optimal), with the new mainspace article obviously subject to deletion processes or BLARs. However, an editor pretending to be an AfC reviewer (like they did so here) is definitely disallowed. I am also not confident that they would be granted the AfC pseudo-perm given that they have no AfD or CSD experience. VickKiang(talk)02:53, 2 November 2023 (UTC)[reply]
You brought up another editor recently who was acting as if they had access to the AFCH script while not actually a reviewer. I think most people would agree that AfC-exclusive processes should be left exclusively to AfC reviewers and patrollers. I'm sure it's also quite confusing for the draft creator to have to deal with a situation where they're getting official-looking reviews from someone who's not been vetted to do so. I'd say that the editor should warned on their talk page and requested to stop, which should hopefully be enough if it was just a good-faith misunderstanding on their part. —TechnoSquirrel69 (sigh) 02:54, 2 November 2023 (UTC)[reply]
Thank you for the feedback, it's appreciated. Since they don't have access to the AFCH script, I'm not sure how they declined a draft but they could have copied and pasted the code from a previous decline on the draft article and signed their name. Thanks again. LizRead!Talk!03:28, 3 November 2023 (UTC)[reply]
Pretty sure the creation of user talk pages with semi-automated tools adds {{Talk header}} to the top by default. I'm not sure if this is the scripts doing it, or if MediaWiki throws that in when it detects you're performing the action with a script. —TechnoSquirrel69 (sigh) 03:32, 3 November 2023 (UTC)[reply]
AFCH does it, not MediaWiki. There's a talk page discussion about whether this should be added to all pages somewhere. And I discovered in that discussion that there's definitely a faction of editors that don't like this template. But I like it and AFCH adds it and it doesn't seem to be causing any problems, so I recommend no action at this time. –Novem Linguae (talk) 19:34, 3 November 2023 (UTC)[reply]
Yeah, that's about what I thought, because I just remembered that ClueBot doesn't add this template when it issues user warnings, so it can't be MediaWiki doing it. I'd agree that nothing about the current system seems to be causing problems, so don't fix what ain't broke. —TechnoSquirrel69 (sigh) 19:46, 3 November 2023 (UTC)[reply]
Process for wp:AFC accept when the target is a redirect?
What is the process for accepting an article when the the target is currently a redirect? The example that raised the question for me is Draft:Economy of the Qing Dynasty but I'm also trying to learn in general. The script instructions said to just request a general move but wouldn't that leave the draft open at wp:AFC? Sincerely, North8000 (talk) 02:53, 3 November 2023 (UTC)[reply]
I think the correct procedure is first indicate you will accept the draft by a comment (this step is optional), then tag the redirect with this template (mandatory), and finally move the AfC draft to mainspace after the redirect has been deleted under G6.
Of course, for contentious and contested redirects this approach won't work, but for the case you mentioned, doing the above procedure is fine. VickKiang(talk)02:59, 3 November 2023 (UTC)[reply]
Always worth checking the history on the redirect as well. I remember a situation some time ago where the draft was a recreation of the article that had been taken to AfD and closed as redirect. Also, are page movers able to move drafts over redirects using the AFCH script?(Never mind, just read WP:PMR again.)—TechnoSquirrel69 (sigh) 03:06, 3 November 2023 (UTC)[reply]
Not directly. You can move the current redirect as part of WP:SWAP to a temporary
"Draft:Move/Location X" first, while disabling the leave a redirect option unchecked while performing the move. Then do your acceptance using the AFCH script, and finish off with moving the redirect at the temporary title into the draft title overwriting the new redirect. – robertsky (talk) 15:45, 3 November 2023 (UTC)[reply]
I'd like to build detecting this situation and placing the {{Db-afc-move}} template on the redirect into the AFCH script someday, since questions about this get asked several times a year. This workflow is confusing and building it into the software would help. –Novem Linguae (talk) 19:38, 3 November 2023 (UTC)[reply]
The link created by AFCH directing comments to user talk pages doesn't leave signatures
The substitution template at Template:AfC decline/HD preload doesn't include the four tildes when used by the link clicked on at the declination template left by the AFCH script on the submitter's talk page, and the result is that any talk page messages left by the submitter at the reviewer's talk page goes unsigned, and there is no [reply] button following the message. microbiologyMarcus(petri dish•growths)15:52, 3 November 2023 (UTC)[reply]
I think the user just edited the response to remove the auto sign.... you can lead an editor to a talk page but you can;t make them sign :) KylieTastic (talk) 16:15, 3 November 2023 (UTC)[reply]
The other possibility is their sig has no link. I thought they removed that option a couple of years ago (i.e. it forces you to have at least one link) but you might be right that they could have manually typed in their sig. Primefac (talk) 16:18, 3 November 2023 (UTC)[reply]
There is the outstanding problem, though, that when a user replies to the talk page messages, we never get notified about it. Very often I come across people who have replied to past declines but have been shouting into the void. -- asilvering (talk) 16:18, 3 November 2023 (UTC)[reply]
Any chance of implementing that wording change to the templates we discussed a few weeks ago? Not sure we got consensus, purely because no one really interacted with the suggestion, but....! Qcne(talk)17:55, 3 November 2023 (UTC)[reply]
MicrobiologyMarcus are you talking about when a user selects to leave a message on the "reviewer's talk page" from a message like you left here as I see the reply they left you did not look properly signed. However if you try the link yourself it does have four tildes (or at least it does when I just tried it). KylieTastic (talk) 16:12, 3 November 2023 (UTC)[reply]
If I click on that same link, the pre-filled text for me is:
== Request on {{SAFESUBST:#time:H:i:s, j F Y}} for assistance on [[Wikipedia:Articles for creation|AfC]] submission by {{SAFESUBST:REVISIONUSER}} == ::{{anchor|{{SAFESUBST:#time:H:i:s, j F Y}} review of submission by {{SAFESUBST:REVISIONUSER}}}} ::{{Lafc|username={{SAFESUBST:REVISIONUSER}}|ts={{SAFESUBST:#time:H:i:s, j F Y}}|declinedtalk=Draft:Maher_Jarmakani}} ::{{SAFESUBST:Void| ::<!-- First, tell us why you are requesting assistance. Take as many lines as you need. --> }} ::<!-- Start of message --> ::<!-- End of message -->~~~~{{SAFESUBST:Void|<!-- ::Finally, make sure to click the "Publish changes" button below or your request will be lost!-->}} ::
So this is an instance of people just removing those tildes in their message before hitting submit? I'm curious why I never had a review button herepromise to start rereading things fully before responding, answered abovemicrobiologyMarcus(petri dish•growths)16:21, 3 November 2023 (UTC)[reply]
Incidentally, this is why we've tried to move more towards the script-based wizards - try as we might to say "DON'T REMOVE THIS LINE" people, invariably, do. Primefac (talk) 16:44, 3 November 2023 (UTC)[reply]
You can see all your draft creations here. If you're looking to find just the drafts you've submitted for review, you can head over to {{AfC statistics}}, and then simply press Ctrl+F to search for your username without the 'User' prefix. – DreamRimmer (talk) 02:08, 4 November 2023 (UTC)[reply]
@Immanuelle the quickest way to see your active submissions (99% accurate) is with a search like this. Which show you have 45 submissions which is ~2.5% of all submissions and climbing (yesterday it was 2%). Can you please take a break from submitting for a while for the sake of both the reviewers and other submitters and let us catch up. Currently the top submitters list is: Immanuelle: 45, Dogloverr16: 18, 126yt: 10, Liuchinghuang: 9, FloridaArmy: 8, Shan-Chen Lu: 7, Winnie0510: 5. You can also find your declined submission with this. Cheers KylieTastic (talk) 10:59, 4 November 2023 (UTC)[reply]
@KylieTastic I think I'll just take a general break from Wikipedia. I managed to reorganize my drafts a lot reducing my draft count from 3,946 drafts to 3,522 drafts, after being criticized for having too many drafts. I think I will just take a break for a while. I submitted the drafts I thought were good (although a few weren't that good), and now it's my time to relax for a bit and decompress. I've been acting a bit from a perceived pressure for performance lately, now my drafts are all ones someone else can realistically contribute to if they find them. Might be gone for a week or two Immanuelle ❤️💚💙 (talk to the cutest Wikipedian)01:47, 5 November 2023 (UTC)[reply]
@Immanuelle wow that's a lot of drafts, which means a lot of work, so yes you need to take care not to get overloaded and burnout. Over the years I've seen a number of good editors push themselves then disappear. It's good to recognise when you need to take a break and actually take one for your own health. Have a good chill. Cheers KylieTastic (talk) 10:15, 5 November 2023 (UTC)[reply]
Ancient Backlog Cleared
At this time, we do not have any drafts that are 3 months or more old. Does this mean that the backlog drive has been a success so far? (I think so, but that is only my opinion.)
Robert McClenon (talk) 14:31, 4 November 2023 (UTC)[reply]
Good start, but no one should really have to wait more than a month for a review (preferably a week), so more to go for a real success on !queue reduction. Success will also be judged by quality which we will probably not tell for a while. At least accept rates are up. KylieTastic (talk) 14:44, 4 November 2023 (UTC)[reply]
I think that means the drive has been a runaway success! Extrapolating from the rate the backlog has been reducing, we're due to hit zero before November 8. (Not saying that will happen, only that it might be expected given the current pace.) —TechnoSquirrel69 (sigh) 14:45, 4 November 2023 (UTC)[reply]
I know we're getting close to being "done" a scant week into the drive, but I would ask that folks start focusing on re-reviews, in particular those with maybe 50+ reviews (and or the top 10 reviewers). It's one thing to get rid of the backlog, but if folks are getting sloppy to do so then we've probably done more harm than good. Thanks! Primefac (talk) 13:29, 7 November 2023 (UTC)[reply]
I've done some spot-checks of reviewers with abnormally high decline rates and the results haven't been terribly encouraging, but it's easy enough to fix if we work quickly. (ie, so far I've been resubmitting failed declines before the original author has done so) -- asilvering (talk) 15:15, 7 November 2023 (UTC)[reply]
@Joe Roe I've been spotchecking people whose decline rates approach 100%. I'm not sure exactly where the bottom end of "abnormally high" would be - maybe about 80% declines? The project average is something like 60%, but it really depends on how you approach the queue. My decline rate was dead average or a bit low for a long time, but when I switched to prioritizing the four-month-old backlog, it went up quite a bit. The easy declines are all at the front of the queue. My ratio early in the drive (before I started doing re-reviews and going back to clearing out the Books and Literature categories) was something like 30% declines. So I would expect that someone who exclusively makes quick calls at the front of the queue, leaving anything that requires further investigation, to approach something like 80% declines without making inappropriate declines. But that's pretty much all based on feel. -- asilvering (talk) 15:24, 7 November 2023 (UTC)[reply]
Thanks. I sometimes look at AfC decline rate when evaluating WP:PERM/NPP requests and generally find >80% to be a red flag too. It's nice to have a broader basis for comparison. I've never thought to check my own... 11%, I guess that puts me on the other abnormal end! – Joe (talk) 15:35, 7 November 2023 (UTC)[reply]
@Joe Roe there is a huge variance and frankly I don't think it's a good indication of review quality. yes we certainly have had those doing bad declines, but also those doing bad accepts, and worst of all bad rejects. I just run the last 7 days stats this and we have 24 reviewers at 100% accepts 49 at 0% accepts. My accept rate is always low because I spend the most time playing Whac-A-Mole with spam, copy vio, promo etc in 0 day. Some focus on good articles or old end of queue where you then expect a higher accept rate. KylieTastic (talk) 15:45, 7 November 2023 (UTC)[reply]
For sure there are good reasons why some reviewers might fall on the extreme ends of the distribution. But if you're just reviewing drafts at random and decide that less than 1 in 5 would pass AfD, I think it's at least worth a closer look. – Joe (talk) 16:06, 7 November 2023 (UTC)[reply]
I've been thinking on this and looking at the data wondering if there is a way to spot possible problematic review areas.... but failed to come up with anything :/ I think we are going to have to rely on good old random checks unless anyone has some bright ideas. KylieTastic (talk) 17:55, 7 November 2023 (UTC)[reply]
I think the re-review system will probably work well enough, but it does leave me wondering about what the project's standards for reviewers are. For instance, right now Vanderwaalforces has 6 failed reviews. Is that a lot? Well, no one has failed more. But that's also only six fails (so far) out of over 600 reviews. A fail rate of 1% seems pretty good to me. GraziePrego has also six failed re-reviews, this time out of about 120 reviews. That's 5%, which is a lot higher. But is it "too high"? It's still not very high (for the ttrpg players, it's how often you expect to critfail, so it's a number I'm used to treating as negligible). To be perfectly clear, I don't think enough re-reviews have been done in any case to make any statements about any particular reviewer's overall review accuracy. But I don't really have a sense of what "enough reviews to determine whether someone is reviewing acceptably" would be, either. -- asilvering (talk) 18:28, 7 November 2023 (UTC)[reply]
I don't think it's so much "failed reviews out of the total", but "failed reviews out of the re-reviews". If someone has six failed reviews out of 6 re-reviews, that's bad. If they have six failed reviews out of 10, 20, 30, that gets increasingly better. Someone who is failing a majority of their re-reviews should get more re-reviews, while someone who is failing almost none likely does not need many more re-reviews. It's a rolling-snowball metric more than anything. Primefac (talk) 18:36, 7 November 2023 (UTC)[reply]
I agree, it's a "rolling-snowball metric". As re-reviews are done the pass/fail rate should determine if more are needed or not. I would aim for 5% re-reviewed at first, if all pass then probably good if not aim for 10% re-reviewed and re-evaluate... it's not just pass/fail but are they opinion/nuanced fails or WTF fails. KylieTastic (talk) 18:45, 7 November 2023 (UTC)[reply]
@Primefac yes, that's a good point. Though I think it's important to point out that re-reviews aren't neutral (as in, they are not selected completely at random). I've been paying particular attention to rejections, for example. I like KylieTastic's suggestion of 5% re-reviewed before we try to make any further conclusions. Worth noting also that someone who declines articles 100% of the time, without even reading them, would have a re-review pass rate of about 60%, so it seems to me that whatever our hoped-for pass rate is, it should be rather higher than that. -- asilvering (talk) 19:08, 7 November 2023 (UTC)[reply]
I agree that re-reviews are far from neutral or random. We all probably pick, consciously or unconsciously, first the reviewer we want to check, and then the draft. And depending on whether we like (consciously or un-) to pass or fail reviews, we might skate over 'problematic' ones, or make a point of picking up on them. For a better idea of review quality, the re-reviews should be allocated truly randomly, and ideally be anonymised. Which is probably far more hassle to organise than it's worth, but still. -- DoubleGrazing (talk) 19:54, 7 November 2023 (UTC)[reply]
I assumed the idea was that every draft in the backlog drive was going to be re-reviewed. I read the discussion in the previous few comments as "when to raise the alarm early", not "when to stop re-reviewing". I don't think re-reviews should be anonymized. If we block someone for gross incompetence, for example, we'd want to check their re-reviews. And I think we should be open to the basic standards of being available for comment on our re-reviews, just as we are on our reviews. -- asilvering (talk) 19:57, 7 November 2023 (UTC)[reply]
Per WP:CANCER, I'm not accepting any drafts. I only decline/fail those that merit and just pass by the ones which ought to be accepted. Please don't think that the overall acceptance rate should be represented by each individual reviewer. Also, while I admit to being a deletionist, we should avoid bringing inclusionist opinions to judging the work of other editors. Chris Troutman (talk)15:51, 7 November 2023 (UTC)[reply]
I don't think that it should be represented by each individual reviewer, and I think it's quite clear from my posts (and Joe's) that this isn't what we're intending. -- asilvering (talk) 16:45, 7 November 2023 (UTC)[reply]
I have included re-reviews as part of the points awarding mechanism. The re-reviews will be counted at the end of drive at the very least if the bot isn't updated. – robertsky (talk) 16:04, 7 November 2023 (UTC)[reply]
Usually I try to wait at least a few minutes before reviewing (in case they're still making changes), but I didn't check on this one. At least KylieTastic and I are in agreement :P.
In mainspace yes wait as we don't know if they are still changing stuff, but if they hit submit here before ready that is really their issue not ours. Also all they have to do is finish and resubmit so it's not a big issue. KylieTastic (talk) 18:10, 4 November 2023 (UTC)[reply]
Also, this reminds me of an issue I had been meaning to bring up earlier: the AFCH script doesn't seem to have a way of dealing with edit conflicts and just overwrites the previous review if two reviewers happen to decline a submission at the same time. It's usually pretty easy to catch and fix, but it'd be nice if it could warn the user if it detects that an edit has been made to the page. —TechnoSquirrel69 (sigh) 18:17, 4 November 2023 (UTC)[reply]
What Wikipedia calls an Edit Conflict is known more generally in electrical engineering and computer science as a race condition. It is not easy to design systems to handle race conditions optimally. I think that most Wikipedia processes handle race conditions at least as well as could be expected. Most but not all Wikipedia editors understand that race conditions happen, and that results are sometimes unpredictable. The unpredictability is a characteristic of race conditions, and is what makes them difficult to design for. Robert McClenon (talk) 00:24, 5 November 2023 (UTC)[reply]
One way to deal with such race conditions is to have the script to check if there was a prior review done in the last hour, or any other suitable duration and warn the reviewer if so. – robertsky (talk) 03:43, 5 November 2023 (UTC)[reply]
I've had two AfC declines overwritten in a row just now, as can be seen by my edit history. I don't mind being pre-empted by someone else (though whether I agree with the decision is another matter), but I do mind that it glitches out and forces you to revert everything manually, in effect punishing you for being too slow. Hopefully the bug gets fixed. ᴢxᴄᴠʙɴᴍ (ᴛ) 09:01, 8 November 2023 (UTC)[reply]
If I'm remembering correctly, drafts on topics that are under ArbCom sanctions that only have substantial edits by non-extended-confirmed users are subject to deletion, right? I wanted to run this by someone else before moving forward; I've marked it as under review for now so it hopefully stays in draftspace until we figure something out. —TechnoSquirrel69 (sigh) 20:54, 5 November 2023 (UTC)[reply]
Technically, can probably tag it csd WP:G5 (in violation of topic ban) and get it deleted. How bad is it? If its good, you could instead "accept" it and basically adopt it and vouch for it. Up to you! –Novem Linguae (talk) 21:41, 5 November 2023 (UTC)[reply]
I'd say it's in an acceptable state, and I would have accepted it if not for the fact that it could be merged and that it's in a sanctioned topic. I wouldn't really feel too good about throwing away good-faith contributions from this editor, as it's unlikely that they knew about the extended-confirmed restriction. I was thinking about overseeing the merge myself, since the article creator can't edit the destination page either, and I could use a second (or higher!) pair of eyes on that as well. I think I'll drop a template on their talk page explaining that they've stumbled into an ArbCom remedy, and give them a heads-up that the draft will probably have to be deleted once I've merged over what I can. —TechnoSquirrel69 (sigh) 22:22, 5 November 2023 (UTC)[reply]
I saw his comment about it being "potentially notable, but 2023 Israel–Hamas war protests already covers the topic," but I would say that logic supports a merger. Admins reviewing an AfC do not need to agree. It seems you have already started a merger, from your other comment, which is good. Historyday01 (talk) 14:26, 6 November 2023 (UTC)[reply]
My initial thought is to treat something going through AfC on a contentious/sanctioned subject like that sort of like an edit request - I don't have the wording in front of me, but it's not that new editors can't in any way contribute to these topics, it's that they can't directly edit an article and instead must go through a more experienced editor, right? Is that a reasonable way to treat it, or am I just making stuff up? LittlePuppers (talk) 03:03, 6 November 2023 (UTC)[reply]
From WP:ARBECR: Non-extended-confirmed editors may not create new articles, but administrators may exercise discretion when deciding how to enforce this remedy on article creations. In other words, admins are welcome to delete these drafts, but they are not required to, and a "good" draft can probably be accepted if it meets our criteria. Primefac (talk) 09:23, 6 November 2023 (UTC)[reply]
IMO it's a bit of wp:gaming but it's probably best to deal with the articles individually rather than acting based on this tactic between the two. If they just tried moving the material over, the new article it would have an even weaker case North8000 (talk) 18:08, 6 November 2023 (UTC)[reply]
I was debating a history split, but since the topics are somewhat intertwined ("the founder" -> "the founder's company") I think it's acceptable to leave the history alone. I don't think we should keep decline notices for the first iteration, though, if the draft has been essentially rewritten. Primefac (talk) 20:08, 6 November 2023 (UTC)[reply]
I don't think this is concerning unless there is some other shenanigans. I have at times suggested a draft should be a about different topic in my reviews, for example an author where sources are mostly about a book, companies where sources are mostly about a specific product, etc. S0091 (talk) 20:21, 6 November 2023 (UTC)[reply]
Are you sure it's not a bug? Steven Jackson (politician) looks like it was only created on 22 October 2023 and submitted that same day, but shows as being pending for 359 days. The one you mentioned also has that it was submitted in its edit history. Turnagra (talk) 09:17, 7 November 2023 (UTC)[reply]
The template does not go off the actual page history, it goes off the timestamp added to the page. In this case, it is {{AfC submission|t||ts=20221107185100|u=MoviesandTelevisionFan|ns=118|demo=}}, which gives a time stamp of 7 Nov 2022. Primefac (talk) 09:26, 7 November 2023 (UTC)[reply]
I don't think there is an easy way to fix the scoring for this situation, unless @Ingenuity thinks otherwise. What we can do is to check for any drafts that're more than 150 days pending and adjust the scoring manually at the end of the backlog drive. – robertsky (talk) 13:59, 7 November 2023 (UTC)[reply]
I think that'd be the best duct-tape solution for now. Since it's been ascertained that the bot is not malfunctioning, there's no urgent need to fix anything at the moment. The AFCH maintainers will probably look into the issue of the script's timestamps when marking as under review, but that could take a minute. —TechnoSquirrel69 (sigh) 14:05, 7 November 2023 (UTC)[reply]
hey man im josh, could I ask why you have accepted at least 9 pages created by MoviesandTelevisionFan, who is CIR-blocked from editing the article space, without any of those drafts having been submitted? Primefac (talk) 09:30, 7 November 2023 (UTC) Apologies for the ping, I guess I'm just being overly cynical this morning. Primefac (talk) 10:26, 7 November 2023 (UTC)[reply]
The block was because of poor English skills but MaTF appears to understand nobility so I see no problems with getting these into article-space once reviewed. I do think this is a bug in the fact that if a draft is not submitted and you mark under review it should use the current date. If do a submit on behalf of the creator first then mark under review this is what would have happened. In normal times this would not matter to anyone, but this oddity makes it look a bit like gaming the drive points system rather than saving lost drafts which I'm sure was the actual motivation. I assume this is easily fixed with manual timestamp manipulation. KylieTastic (talk) 10:09, 7 November 2023 (UTC)[reply]
An unsubmitted draft doesn't need a timestamp (i.e. there's nothing that uses it) so really what needs to happen is it should be ignored (or better yet, not even added in the first place) until it is either submitted or marked for review. Primefac (talk) 10:20, 7 November 2023 (UTC)[reply]
Is this a bug in AFCH? When marking an unsubmitted draft as under review (such as here), what should it write instead of {{AFC submission|r||u=MoviesandTelevisionFan|ns=118|demo=|reviewer=Hey man im josh|reviewts=20231101022236|ts=20221107185100}}<!-- Do not remove this line! -->? Should ts= the current timestamp, or maybe be left blank? –Novem Linguae (talk) 10:42, 7 November 2023 (UTC)[reply]
I don't think it's a bug so much as a situation which was not really planned for. If a draft is being pushed straight from /draft to /reviewing the time stamp ideally should be when it was marked as under review, as seen here. That being said, we can likely hash this out in a new section to avoid cluttering up this one, which is definitely not about the template itself. Primefac (talk) 11:03, 7 November 2023 (UTC)[reply]
Fwiw, if you look at my AfC log, I was going through their drafts before this drive even started. It's a continuation of what I was doing, no gaming. No objection from me if my points are adjusted at the end. Hey man im josh (talk) 10:37, 7 November 2023 (UTC)[reply]
@Primefac: Just to follow up a bit more now that I'm at a computer, I mark them as under review so I can take advantage of the automation portion of the AFCH tool to add appropriate tags, adjust commented out categories, and to add it to my AfC log so that I can continue to track what drafts I've accepted or declined. Thank you @KylieTastic, you're correct in my intentions. I just wanted to get more drafts from draft space into main space and MaTF has a niche and pattern to the articles they create, so I've looked through their current drafts (found here). In short, they've figured out the formula for creating stubs that pass WP:NPOL and are written well enough to be accepted, so I wanted to clear those away instead of them getting to a point where they get close to expiring. In October I accepted 21 of their drafts, some of which had not been submitted. Thus far, in November, I've accepted 14 of their drafts, none of which had been submitted. When looking through their drafts in progress I made a bookmarks folder of drafts of theirs that I intended to accept, which currently has 4 drafts left in it. I spaced out when I was accepting their articles because I didn't want to flood their talk page with mass accepts all at once.
I want to be absolutely clear I had no intention of gaming the system. I also have no issue whatsoever with removing the entries from the my backlog drive log. With that said, perhaps we should instead consider explicitly allowing drafts that were not submitted to count towards scoring. With the backlog dwindling, and the reviewers on a roll, I think it could be a good idea to help reduce the total number of drafts and save some content from eventually getting G13 deleted. For those who aren't aware, there are roughly ~150-250 articles that appear on User:SDZeroBot/G13 soon each day. Hey man im josh (talk) 13:06, 7 November 2023 (UTC)[reply]
Thanks for the expanded explanation. Honestly, I wasn't even thinking about the backlog drive or points or anything like that, I was not thinking broader and just saw "AFC reviewer accepting a bunch of unsubmitted drafts". I do apologise for jumping to nefarious conclusions (I obviously shouldn't have) and do feel like a right twit for doing so. I think going through drafts like that is probably a good thing (I know Prax is doing it for FA's creations) since it clears out the unnecessary wait times for good pages. Primefac (talk) 13:14, 7 November 2023 (UTC)[reply]
No worries @Primefac, I absolutely see how you reached the conclusion you did and why you felt the way you did at first, so don't be too hard on yourself. Star Mississippi is doing so as well with FA's drafts, which is actually what gave me the idea! Hey man im josh (talk) 13:54, 7 November 2023 (UTC)[reply]
@Hey man im josh thanks for that link, it hadn't occurred to me to go through these. How do you mark them as under review? I've never bothered to do this before and I don't see an option when I run the AFCH script. -- asilvering (talk) 15:13, 7 November 2023 (UTC)[reply]
@Asilvering if you fist use the submit from within AFCH, then optionally mark as under review if you need more time, then accept it will get around the date that started all this. KylieTastic (talk) 15:24, 7 November 2023 (UTC)[reply]
Scoring for Wikipedia type Articles Generated by LLM
Our research team is 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.
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.
We will be providing monetary compensation for work provided. This was posted earlier, but now with next steps. We hope to begin the review process in a few weeks.
This is still extremely low on answers to important questions like "who are you", "what is the purpose of your educational research", and "where are your standards of research ethics". -- asilvering (talk) 23:44, 8 November 2023 (UTC)[reply]
Hi, I'm glad to answer your questions. I wasn't sure what was permissible by Wikipedias idea of not posting personal information. We are a team of three students working on a project as part of this class [3]Conversational Virtual Assistants with Deep Learning. We are focused on a few things: Exploring if we can reduce LLM hallucination to negligible amounts, exploring if LLM can generate articles with the same high quality as Wikipedia, and exploring if article based information gathering will remain relevant with the advent of informational chatbots.
Filling out the form is obviously not a binding agreement. We would appreciate you filling it out and we can contact you with more details. To be clear, the intent of this is not to build a model that will be used on or with Wikipedia. Rather it is in the scope of the class and of improving LLM models. Terribilis11 (talk) 01:18, 9 November 2023 (UTC)[reply]
Oh wow ok it is... significantly context-altering to know you are in a 2nd-year undergraduate course. New question: does your professor know that you are recruiting people who are not Stanford students as scorers? The way you've gone about this recruiting set off a lot of my "these people are acting completely without IRB oversight, as though they do not know they need it or for some reason hold it in contempt" alarm bells, which is understandable because you... are students, and indeed probably have never been told about ethics approval at all. What you are proposing is a study with human participants, and, barring some strictly defined cases, studies with human participants require ethics approval and data security standards. Since this is obviously a study with only extremely low risks for your human participants, you can probably get ethics approval expedited, but you probably need it and your professor can get into big trouble if they don't get it. (It's your supervisor's responsibility, so here that's your prof - you're not in trouble.) The compliance office at Stanford closes to new submissions on the 13th ([4]), so you need to work quickly to ensure you're ok here. -- asilvering (talk) 07:05, 9 November 2023 (UTC)[reply]
The form appears to miss-understand the Good Article requirements. The zeroth GA requirement is that it's an article. The first sentence on the page A good article (GA) is an article that meets a core set of editorial standards[...]Stuartyeates (talk) 03:29, 9 November 2023 (UTC)[reply]
I'm not quite sure what you mean. The results of our model will be an article or rather in the format of an article. We won't be publishing the articles on Wikipedia.Terribilis11 (talk) 05:40, 9 November 2023 (UTC)[reply]
Done Ca talk to me! 00:50, 17 October 2023 (UTC)
- 🔥𝑰𝒍𝒍𝒖𝒔𝒊𝒐𝒏 𝑭𝒍𝒂𝒎𝒆 (𝒕𝒂𝒍𝒌)🔥 16:31, 30 October 2023 (UTC)
