What we lost, what we gained

Deletion report

What we lost, what we gained

Most editors are familiar with the existence of Articles for Deletion (AfD), the process by which we determine the suitability of articles for inclusion in Wikipedia. Indeed, many of us have had direct experience with the process, whether we liked it or not: working your ass off on an article only to see it flushed down the drain is close to an official rite of passage around these parts. That said, there are plenty of articles that have no place on Wikipedia, and plenty of subjects that have no business getting an article written about them. Love it or hate it, AfD is one of the most publicly-known processes on Wikipedia; it's referenced often enough in mainstream publications for us to have an article about it. And it looks like the largest AfD of all time has graced us with its presence this month.

But what do we really know about it? Statistical analysis is rather hard to come by. Earlier in 2021, I wrote a piece of software that analyzes AfD logs, from which I was able to create a live dashboard of current deletion discussions. I was also able to analyze all 480,000 AfDs (statistics on which can be found here). There were some interesting revelations, including a sortable list of the longest AfDs of all time for the drama-minded.

The topic of this report, however, is what was going on with AfD in November 2021, including the monthly statistics and sortable tables of each AfD.

Overall statistics

These figures are current as of November 28. More detailed statistics can be seen at the monthly Oracle page for November 2021, including numbers and percentages for all closes in the month.

There were 1,767 AfDs listed in November, of which 1,172 have closed and 595 remain open. This is slightly below the 2021 average of 1,794 per month (and well below the 2005–2020 average of 2,400 per month).

This comes out to an average of 63.1 per day, with the least on the 5th (35) and the most on the 27th (92); the average for 2021 has been around 54 per day.

205 of the November AfDs were relists from October, meaning 1,562 new nominations have been made since the beginning of the month. Additionally, 15 were closed without a !vote being cast (one was withdrawn, one was deleted, four were speedily deleted, and the rest were closed as no consensus).

The most common outcome was "delete" (as has been the case for all months since August 2005); "delete"s and "speedy delete"s combined made up 61.8% of closes. Meanwhile, 19.2% closed "keep" or "speedy keep", slightly below the 2021 average of 20.5%. There was only one type of close that didn't happen a single time in November — the elusive unicorn of deletion, the "transwiki" close to move content to another wiki, has occurred only 324 times in nearly 500,000 AfDs.

Mass killings under communist regimes: The largest AfD of all time

The most !voted on AfD this month, and the largest by a wide margin, isn't even closed yet. The discussion for Mass killings under communist regimes is currently at a gargantuan 499 kilobytes, 172 !votes, and 1,096 total revisions by 204 editors. Even disregarding the 117 kilobyte talk page for the nomination, this is by far the largest AfD of all time.

Note: The discussion's final size was 510,874 bytes, with 174 !votes across 217 editors. It had the {{closing}} template added by Jo-Jo Eumerus on November 29; !votes continued to trickle in (reaching a maximum of 511,852 bytes) until post-close !votes were removed and the page was fully protected by Joe Roe five hours later to allow the closers to work. It was closed as "no consensus" on December 1, by a four-administrator panel consisting of Jo-Jo Eumerus, Joe Roe, Rosguill and Seraphimblade.

The previous champion (the third nomination of List of notable converts to Christianity in 2007) was a paltry 234 kilobytes. It is not the AfD with the most !votes, however: that would be the 60-article batch nomination at Esoteric programming languages in 2006, which garnered 301. Or at least, it isn't the one with the most !votes yet. It still has a few more days to run, and it's apparently been picked up by a few blogs and media outlets. It's had 65,696 pageviews so far, which probably puts it high in the running for the most-viewed AfD of all time.

This is the article's fourth stint at AfD: the first three discussions (one from 2009 and two from 2010) were no less contentious, weighing in at 31, 67, and 206 kilobytes respectively. Prior to that, the article was nominated twice in 2009 under its previous title, Communist genocide.

This Polyphemian (or, if you prefer, "huge-ass") discussion concerns a similarly massive article: currently 297 kilobytes, it's existed since 2009, and in that time has had over five hundred distinct editors. The talk page is festooned with twelve talk page headers, and links to 52 archive pages. In fact, the thread about the article at the dispute resolution noticeboard was so long it got moved to its own subpage (with its own shortcut, WP:DRNMKUCR, whose incomprehensibility seems apt for the situation). This subpage is, itself, 184 kilobytes. And it's on hold because the article is at AfD!

The current AfD, nominated on November 22 by cygnis insignis, has seen a variety of established editors coming down on both sides. In the interest of full disclosure, the writer of this article is one of them, and will do something smart for once in his life by refraining from giving further commentary.

Other discussions of note

The second most !voted-on AfD was the fiercely debated third nomination of ARS Public School, in which 30 !votes were cast in 51 kilobytes of discussion among 33 editors. Closer Daniel noted that "this debate obviously has caused fractures across the wider community, in other places than simply this discussion page". Indeed, this AfD featured in the latest in a series of content and conduct disputes over deletion at AN/I centering around the activity of the controversial Article Rescue Squadron. The article itself was about a school in Jharkhand, India. The third nomination for the page, submitted by Dronebogus, was made two days after a no-consensus close of the second. Ultimately, the ARS failed to save ARS; Daniel, in closing the debate on November 16, cited a clear consensus to delete, saying that "ultimately, it has come down to an interpretation of our relevant notability guidelines, and there is a strong consensus that this article does not meet those requirements".

The third most active AfD, with 23 !votes in 43 kilobytes, was Herbert V. Clark. The AfD, nominated by Mztourist on October 23, was closed as "no consensus" by Scottywong on November 11. Discussion focused on interpretations of what constituted independent and significant coverage, as well as whether Clark was independently notable in his own right or merely inherited notability from his status as a Tuskegee Airman.

Four discussions, each with 21 !votes, are tied for the spot of fourth most-active:

Irish actor Stephen Hogan had the second largest AfD of the month in terms of page size, weighing in at a hefty 66 kilobytes. Nominated by DGG on October 24th, this month-old discussion has so far spawned one very long thread at COIN, as well as a second and third at AN/I. With 21 !votes, it closed on the 26th as "no consensus" by Jo-Jo Eumerus, who said: "It seems like there are arguments on both sides of each side and no argument is clearly superior to the other. I note that the discussion was full of offtopic commentary, sockpuppetry and that some participants were sufficiently irritated by one participant that they struck out their !votes".

106 articles pertaining to the ongoing Tigray War in Ethiopia were batch-nominated at 2020 May Kado massacre, the third largest AfD of the month (at 51 kilobytes). The nomination was made by WMSR on November 8, and closed as "procedural keep" by Vanamonde93 on November 15. Two days later, a second nomination was made by Dawit S Gondaria, featuring 25 from the previous batch. This time, all 25 were deleted; closer Missvain cited "a variety of reasons presented - WP:NEVENT, failure to present WP:RS and WP:SIGCOV and the looking sockpuppet investigation taking place around link spamming".
Wikimedia UK trustee and marketer Monisha Shah had her article nominated for deletion on the 18th; her close relationship to the editing community made for a fairly contentious deletion discussion, with a long thread on COIN running in tandem with the AfD. It was closed as "delete per consensus" by User:Doczilla on the 26th.
The biography of unsuccessful Buffalo, New York mayoral candidate India Walton, on its second nomination, was closed "keep" by Daniel after 21 !votes. Retention arguments mentioned the "historic" nature of her loss, and Walton's status as a "prominent example of a divide within the Democratic Party in 2021".

The largest unanimous AfD was Washington Football Team vs. New York Giants Game September 2021 (boy, that's a mouthful). Of fifteen !votes, all were to delete; most arguments invoked the notability guidelines at WP:SPORTSEVENT, which says that "regular season games in professional and college leagues are not inherently notable [...] games should be extraordinary and have a lasting impact on the sport". The article's creator, Wiki Ed student editor Pippalenderking, was sanguine about the experience, saying:

"I just want to offer my thanks and appreciation for all of the advice on my article. This is the first article I've ever written for Wikipedia and I am grateful for all the suggestions you have provided! I am going to look into contributing to other related articles -- thank you for including me in this valuable discussion!"

Well, Pippalenderking, here's hoping you find something good to write about -- there's always geostubs!

The shortest AfD was the second nomination of Pulock Deb Roy, closed as speedy delete when the article was deleted just an hour and thirty-seven minutes after being listed. It was not the quickest-closed, however: that honor goes to Habibganj railway station, a pagemove request mistakenly filed as an AfD and non-admin closed after a whopping five minutes.

The oldest AfD closed in November was L'Oreal Hair Zone Mall Tour, opened on October 7 -- although it was by no means the longest-running, being closed by Barkeep49 as "redirect" on the second day of November. The oldest article nominated was Right Livelihood Award, which was created in August 2002.

The longest-running AfD from November was a tie between two nominations with remarkably similar stories: they were both open for 39 days, from October 10 to November 18, and closed by Scottywong within an hour of each other. Relationship anarchy, with sixteen !votes, was closed as "no consensus". Hearns Crossroads, Delaware, a batch nomination of four GNIS-based stubs of alleged unincorporated communities in Delaware, was closed as "delete all" after seventeen !votes.

Next "Deletion report" →

In this issue

29 November 2021 (all comments)

In the media

WikiCup report

Deletion report

From a Wikipedia reader

+ Add a comment

Discuss this story

These comments are automatically transcluded from this article's talk page. To follow comments, add the page to your watchlist. If your comment has not appeared here, you can try purging the cache.

"But what do we really know about it? Statistical analysis is rather hard to come by." - Not to diminish the value of the author's work, but there have been quite a few academic papers that have studied AfD discussions, often producing detailed state. See for example the June Signpost issue: "Analysis of two million AfD (Article for Deletions) discussions". Regards, HaeB (talk) 02:22, 29 November 2021 (UTC)[reply]

I'll have to take a look at those when I get home. That one about analyzing 2 million AfDs is very strange if true (there have been a little under 500,000 from 2005 to 2021, and even if you assume that some pre-2005 discussions were held without their own separate pages, I don't think there should be 1.5 million of them). Were they using multiple projects? jp×g 03:43, 29 November 2021 (UTC)[reply]

Good point - the headline there was relying on the paper's abstract ("an analysis of 1,967,768 AfD discussions between 2005 and 2018"), but in the "Data Collection" they explain more precisely that this means "1,967,768 recommendations" in the sense of "votes"...

Anyway, the underlying corpus was published (by other authors) here: https://github.com/emayfield/AFD_Decision_Corpus . It might be interesting to compare the metadata they extracted with your own.

Regards, HaeB (talk) 04:57, 29 November 2021 (UTC)[reply]

The Oracle data is very cool, thank you for putting that and this report together. Levivich 03:48, 29 November 2021 (UTC)[reply]
Another rather nice visualisation is http://notabilia.net with left spiralling lies indicating 'keep' votes and right spiralling lines indicating 'delete'. T.Shafee(Evo&Evo)^talk 04:15, 29 November 2021 (UTC)[reply]
I remember the Michael Z. Williamson afd, I see it got an honorable 22:d place.
working your ass off on an article only to see it flushed down the drain is close to an official rite of passage around these parts Is that really the case? I've created around 120 articles and never had one deleted and only once or twice had to defend one at AfD. Is that unusual? Am I naïve to believe that if you pick a notable subject, make sure you've got enough material to write a few hundred words, and cite your sources, that your new article will be largely immune from AfD? HJ Mitchell | Penny for your thoughts? 21:21, 1 December 2021 (UTC)[reply]
- I'm at about 30 articles but my experience is the same. Gråbergs Gråa Sång (talk) 08:39, 2 December 2021 (UTC)[reply]

- Only one of my 156 articles has been deleted, which was probably me flying too close to the sun on a current-events BLP (I was banking on continued coverage which never materialized). Sure, it deserved to go, but I'll be damned if I didn't feel a little regretful about the whole thing. I remember hearing a maxim once along the lines of "if you've never missed the bus before, you're showing up to the bus stop too early" -- maybe the same is true here. Well, I've missed the WP:GNG bus once, so now I know how long to wait :) jp×g 03:51, 3 December 2021 (UTC)[reply]
  Old Swedish saying, translated: Better to hear the sound of a bowstring breaking, than to never attempt to draw a bow. Gråbergs Gråa Sång (talk) 11:42, 3 December 2021 (UTC)[reply]

AfD

The lack of mention of immense canvassing across the internet and even by media articles being written seems rather wrong. Not only 4chan, but also people like Larry Sanger have been canvassing to get the article kept. Isn't all of that something that would be appropriate to mention, especially as a reason on why the AfD is so big? Silver seren^C 18:16, 29 November 2021 (UTC)[reply]

Well, I did mention that it'd been picked up by a few blogs (the Fox story came after this article was published). I think it's a stretch to call the 4chan thread "canvassing" -- maybe there was another one that nobody's linked to, but the one I see on the talk page header for that AfD had only six posts, all within one hour, after which point it dropped off the board into oblivion. A typical thread on a high-traffic board will get hundreds of replies, so that would have been a very unsuccessful thread even by >2011 standards. jp×g 23:57, 29 November 2021 (UTC)[reply]

Is media coverage considered canvassing? Are critics of WP considered canvassing if they're not contributing to the project? I may be mistaken but I don't think that was the intent of WP:CANVASSING. When a controversial topic is up for deletion, it will naturally garner public attention and people will talk about it. Regardless, the AfD is in the process of being closed as I type this reply. Atsme 💬 📧 20:41, 29 November 2021 (UTC)[reply]

Much of the media attention was solicited by the groups that were indeed trying to canvass and were the reason why so many newly made accounts showed up. Not to mention dormant accounts that hadn't edited for months or years. Again, isn't all of that something that should be mentioned in regards to the AfD and why it became so big? Silver seren^C 20:48, 29 November 2021 (UTC)[reply]

Dormant accounts probably have more to do with socking than canvassing. Publicity and the opinions that are shared off-wiki are just that, which is not a violation of canvassing to my knowledge, unless an involved editor is behind it; i.e., an actual WP editor has to be the "canvasser" and I'm not aware of that happening. Publicity brings people in and I don't see a work-around for that, much less consider it good reason to dismiss the AfD. In fact, it would be great if more people showed up at AfDs instead of just 3 or 4 involved editors making a final determination to keep or delete an article. Atsme 💬 📧 21:10, 29 November 2021 (UTC)[reply]
In regards to publicity bringing in people, I definitely agree. For me, Rama's preliminary statement in the arbitration case about them seems worth mentioning as an example. –MJL ‐Talk‐^☖ 07:26, 2 December 2021 (UTC)[reply]

- - I keep hearing about all these canvassed votes, but it is not easy to find one. I did see an IP vote, but it appears to have been removed. A couple of days ago I checked the article stats and there were a total of 11 IP *edits*, but I don't see them on the page now (it's a bit hard to say for sure on a page this size). So I just checked 15 keep votes (willy-nilly, or "random"). All of them had recently edited, *before the AfD started* I didn't count how many edits they had before the AfD this year, but it was generally a couple of screen-fuls+, i.e. over 200. Well there are about 150 keep votes. So can anybody find one of these canvassed keep voters (user names please)? Smallbones_(smalltalk) 23:37, 29 November 2021 (UTC)[reply]
    - Sure, @Smallbones:, we can start with the person who voted right before you did: User:Gumaaaa and then move on to others including User:Volatile.memory, User:ChronosTruth, User:Johnathan 2001, User:Delhatch, User:Gnuthomson, User:Redderc, User:Rocket in the sky, and User:Cedric Contra. Just as an example. Silver seren^C 23:51, 29 November 2021 (UTC)[reply]
    - I just had some time to review the !votes. My rough count is 143 keep, 39 delete. It doesn't seem to be going anywhere. I did check out *all* the !votes with red-linked signatures on the page. You managed to get all one side but not the other. My count on these 16 "!votes" was 2 comments (not !votes), 8 (most of the ones above) look like their first edits, and 6 who have many edits before the AfD and look like regular editors who just don't have a user page. So you've reduced the margin to 135 keep, 39 delete." I trust the closers will check these all out in much more detail. But I very much doubt the deletes will ever overcome the deficit. Smallbones_(smalltalk) 04:33, 1 December 2021 (UTC)[reply]

My personal opinion is that admin corps is generally overburdened, sometimes leading to actions that are hastier than they ideally might be. Scores of AfDs must be closed every day, some requiring an hour or some few hours to consider (which in a few cases may not be provided -- a symptom of the overburden, perhaps). Surely this is a significant timesink for our fine but very busy admin corps. We do have Wikipedia:Non-admin closure, but this is only useful in a few cases.

There is an open discussion (started by yours truly) of an idea to address this, at Wikipedia:Village pump (idea lab)#Adding "AfD closer" status, in which participation is here invited. Herostratus (talk) 22:05, 29 November 2021 (UTC)[reply]

A major problem may have to do with fairness. This was the fourth AfD for the page, at least one more than maybe should be allowed. If a page has been saved two times, and then goes for broke again, with all three keeping the page either by Keep or No Consensus, then a "let's really get it this time!" fourth AfD ignores that enough editors have spoken or commented with silence. When articles have successfully gone through the gauntlet numerous times, triple-jeopardy seems as far as we should take it. Randy Kryn (talk) 22:10, 29 November 2021 (UTC)[reply]
Never mind canvassing, call it what it is: fake news. There's no such thing as a "moderator". Anybody [not banned/blocked] can nominate a page for deletion, the omission of which (whether due to malice or lack of basic journalistic research) renders all implications by these news sources completely wrong. Other overtly misleading nonsense permeates the other content.
The news here is that an unremarkable discussion by a direct democracy over how best to arrange freely available information that the same community actually wrote resulted in... nothing (at least not yet, and probably not at all). That a University of Cambridge historian compares compares this to Holocaust denial is rather unsurprising to me given my experience of such people. It reminds me of a right-wing backlash to undergraduates at an Oxford college choosing to rearrange the portraits in their private common room that made the front page of the Daily Mail: in both cases, the idea is that a conspiracy of censorship emerges when, after a group of people choose to take an action, the very same group of people choose to change or update their decision. However, it is one of the stupidest comments I have read in recent months. (Another strong contender is a tweet by Larry Sanger, co-founder of Wikipedia, that implies that he is hearing of WP:SYNTH for the first time in this AfD.) Actually stupider than anything in a far-right reddit subreddit that weighed in on the issue, which is quite the achievement. — Bilorv (talk) 22:30, 29 November 2021 (UTC)[reply]

Never fear - WP always makes room for the far left's counterbalance. Perhaps a paragraph from this perspective should be included in the article subject of this discussion, and vice versa. Just a thought...Atsme 💬 📧 23:57, 29 November 2021 (UTC)[reply]

So far as I can discern, this bears no relation to my comment. — Bilorv (talk) 21:22, 30 November 2021 (UTC)[reply]

I don't think it's fair to hold mainstream news outlets to lofty Signpost standards... understanding of the project seems to be in pretty short supply across the wider Internet. For example, there are about ten thousand Google news results for wikipedia moderators (and 31 for "Wikipedia moderators" in quotes). Heck, even the New York Times references Wikipedia having "moderators". If anything, I appreciate stuff like this, because it inches us a little closer to breaking our Gell-Mann amnesia...

jp×g 00:07, 30 November 2021 (UTC)[reply]

Good journalism is in short supply in the modern day, and generally found in obscure gems within the field of non-mainstream or specialist news sources. Such as The Signpost, of course. — Bilorv (talk) 21:22, 30 November 2021 (UTC)[reply]

One one hand, I emailed The Daily Caller regarding their story to ask for a correction; they seem to have additionally gotten the date on when the article was tagged for deletion wrong. While I included the link to the diff where the deletion notice was added, I've yet to see anything corrected. On the other hand, to lump in the report from The Telegraph as "fake news" seems like a stretch-and-a-half. — Mhawk10 (talk) 06:25, 2 December 2021 (UTC)[reply]

Fiddling around with a wrong date in a Daily Caller article is moving deckchairs on the Titanic. I've had The Guardian and similarly reputed publications refuse to address simple corrections I've emailed to notify them of, like dates and numbers, so I'm not surprised at the lack of response there. The Daily Caller article doesn't seem worse (if anything, it's better) than the corresponding Telegraph article—and that is not meant as a compliment to the former. — Bilorv (talk) 21:14, 2 December 2021 (UTC)[reply]

Is it expected that when someone is mentioned in a 'report' that they be notified? ~ cygnis insignis 06:47, 30 November 2021 (UTC)[reply]
- No. Our policy is to get the facts right. If you are mentioned, e.g. in "The current AfD, nominated on November 22 by cygnis insignis" and we are certain that you indeed did nominate the article for AfD (see link in that section), then we likely won't contact you. If there is some controversy involved about what the facts are, we likely will ask you for a comment. Of course anybody who does have a comment (even if they aren't mentioned) can comment here, but please don't use that as an excuse to just keep a controversy going here that has been addressed ad nausem in other places. So, first question: did The Signpost get the facts right? If so you probably don't want to complain to us. Smallbones_(smalltalk) 20:13, 1 December 2021 (UTC)[reply]

@Cygnis insignis: This is the first article I've written for the Signpost, and I ended up having to throw the part about that AfD together at the last minute (since it wasn't showing up in AfD logs until a day or two before the publication deadline). If it weren't so rushed, I probably would have thrown a ping on the talk page or something. But as Smallbones says above, there wasn't a whole lot of commentary on your specific role in it (just that you were the one who made the nomination), so I'm not sure there would have been much to dispute or amend. jp×g 06:53, 2 December 2021 (UTC)[reply]

JPxG. Thanks for the insightful article. Quick question. It says that 62% of AFDs close delete and 20% close keep. What does the rest of the pie chart look like? What percent are redirect, merge, draftify, no consensus? I ask because I want to calculate the % of articles that are deleted including alternatives to deletion. I suspect that number is much higher than 62%. Thanks. –Novem Linguae (talk) 01:21, 8 December 2021 (UTC)[reply]
JPxG. The prose you quote from WP:SPORTSEVENT is a result of the students' efforts as we both found the previous prose confusing. So a beneficial consequence of this incident has been that I think the notability guideline is now more clear. -Reagle (talk) 17:35, 8 December 2021 (UTC)[reply]

The Signpost is looking for new talent.

Home

About