Wikipedia talk:Speedy deletion

This criterion strikes me as controversial, being vague and subjective on the lower end. It says " that does not indicate why its subject is important or significant." - But it is a matter of opinion. The explanatory essay gives a "Captain Obvious"-type examples: "John Doe is the President of the country of Wiki-Zeland". Sure thing. But what about "John Doe killed a mockingbird"? I say, for "tree-huggers" John Doe is a significant villain. Sages say "he who murders one man murders the entire human race" (Mishna Sanhedrin 4:5) but this would mean that "he who murders one mockingbird murders the entire mockingbird species", right? But what is your say about a mockingbird-killer?

My point is that the meaning of "wikt:important or wikt:significant" for the purpose of Wikipedia must be clarified on the lower end of the spectrum.

Any opinions?--Altenmann >talk 17:56, 9 February 2026 (UTC)

It's a two-part test:

1. Is it plausible that the claim is true?
2. Assuming the claim is true, it is plausible that someone would recommend keeping or merging the content if it came to AfD and/or that there might be sufficient coverage of them to confer notability?

If the answer to both question is yes then A7 does not apply. It is an intentionally low bar.

Assuming no other claims in the article, "John Doe killed a mockingbird" would get speedily deleted as while the claim is plausibly true, killing a single nonspecific bird is extremely unlikely to be notable. "John Doe killed the last mockingbird in Alabama" would not meet A7 though as this is the sort of thing that would likely get coverage. Thryduulf (talk) 18:17, 9 February 2026 (UTC)

I was thinking along the same lines and I intentionally gave this silly example to mark the "lower bound" for the lower end. However the explanatory essay says something different: "B. assuming that the claim were indeed true, could this (or something that "this" might plausibly imply) cause the subject (possibly with other plausible information added) to be notable? Or, does it give plausible indications that research might well discover notability?" I think your addition "...or merging the content" would be an important addition, because deletion and merging are drasticaly different outcomes.

@Thryduulf: That said, would you care to update the explanatory essay and the policy accordingly? IMO application of policies should not heavily rely on essays: essays are not policies.

Something into the policy along the lines: "A7 must be applied only if the content cannot be plausibly merged/redirected somewhere else" --Altenmann >talk 18:48, 9 February 2026 (UTC)

I think that would fall under "Before nominating a page for speedy deletion, consider whether it could be improved, reduced to a stub, merged or redirected elsewhere, reverted to a better previous revision, or handled in some other way (see Wikipedia:Deletion policy § Alternatives to deletion)", which is at the top of WP:CSD and applies to all the criteria. Extraordinary Writ (talk) 19:02, 9 February 2026 (UTC)

Right. I suspect nobody looks at the top when reading instructions for a specific tag, then next hour it is gone, and it is too late. Vague policies of this kind may be easily misused ad an extra caution in the description will not hurt. May be I am stupid, but I spent pretty much time in trying to figure it out. Not all of us are seasoned wikilawyers to know all books by heart. --Altenmann >talk 19:47, 9 February 2026 (UTC)

then next hour it is gone, and it is too late and then you ask an admin to undelete/userfy the page to keep working on it. I know your concern is the immediacy of the deletion (and that's not an invalid concern) but deletion is never the final stop. Primefac (talk) 21:38, 9 February 2026 (UTC)

Yes, that's my concern, because I suspect the number of really active wikipedians is dwindling. Even it is not, many active wikipedians I knew are long gone, so the numbers of watchful eyes is decreasing. In my watchlist 80% of edits are vandals and punctuation/formatting/typo-fixing wikignomes. I dont see why {{prod}} will not do. IMO speedy is for something detrimental. --Altenmann >talk 21:48, 9 February 2026 (UTC)

Adding something into a major policy like CSD is not something any one editor should do without first making sure that there consensus for the change first. In this case, as Extraordinary Writ notes, the change doesn't seem necessary. Thryduulf (talk) 19:40, 9 February 2026 (UTC)

Well, I beg to differ. --Altenmann >talk 19:48, 9 February 2026 (UTC)

You did request opinions... BusterD (talk) 23:15, 9 February 2026 (UTC)

I did. But you phrased yours as a statement of fact, nail in the coffin, dismissed without contest. First you say, "seek consensus first", next you say "don't even bother". --Altenmann >talk 23:25, 9 February 2026 (UTC)

Personalizing my choice not to offer an opinion is an interesting way to gather information. BusterD (talk) 23:40, 9 February 2026 (UTC)

Sorry, my brain is dizzy and I thought that your comment was from Thryduulf right above.--Altenmann >talk 23:55, 9 February 2026 (UTC)

My first sentence is a statement of fact, making it clear that any change will not happen without consensus. My second sentence is my own opinion in the discussion that will determine whether or not there is a consensus to make a change. You are free to disagree with my opinion but you do not get to demand that I (or anyone else) makes a change to a policy based on that disagreement. Thryduulf (talk) 00:07, 10 February 2026 (UTC)

"Demand"??? Wow!. Goodbye to you all. --Altenmann >talk 00:26, 10 February 2026 (UTC)

This user was once an admin? BusterD (talk) 01:38, 10 February 2026 (UTC)

Let's not go there... The Blade of the Northern Lights (話して下さい) 03:27, 10 February 2026 (UTC)

Please consider joining the feedback request service. An editor has requested comments from other editors for this discussion. This page has been added to the following list: Wikipedia policies and guidelines When discussion has ended, remove this tag and it will be removed from the list. If this page is on additional lists, they will be noted below.

R3 is quite specific, designed for recent redirects that are not an obvious typo. In addition, interpretation of R3 doesn't seem to be quite crystal clear; I see only redirects with {{R from typo}} deleted under this criterion, and sometimes random other redirects are tagged for deletion as an "implausible typo" even when they are not trying to typo-correct. I suggest Redirects for discussion be used for questionable redirects instead.

Should R3 be retired, as RFD can be reasonably be used instead? TheTechie^[she/they] | talk? 17:52, 20 February 2026 (UTC)

• Of course not. Where's your WP:RFCBEFORE? —Cryptic 18:38, 20 February 2026 (UTC)
  • Quite literally on my way out the door, but here's some of the stats you should have, at minimum, collected before considering this. 6.1% error/recreation rate is only slightly over our target, and I guarantee you RFD does not want to deal with another 4.2 MInster Stakes->Minster Stakes's every day from now until forever. Let alone forbidding this rationale from being used for moves while suppressing the redirect, which is likely the overwhelming majority of both R3 usage and redirect suppression. —Cryptic 19:00, 20 February 2026 (UTC)
• Oppose I am sympathetic to this idea, because as I have said in the past, R3 is misused a lot. However, I'm not yet convinced that retiring the criterion entirely is necessary, perhaps an edit filter that warns on tagging/deleting old pages with R3 (and A10) could prevent most misuse? Also the opening statement is inaccurate. R3 is for deleting typos and misnomers. A redirect does not have to be misspelled to be eligible for R3. Warudo (talk) 18:52, 20 February 2026 (UTC) edited at 19:02, 20 February 2026 (UTC)
• Oppose. "[WP:R3] applies to recently created redirects from implausible typos or misnomers. However, redirects from common misspellings or misnomers are generally useful, as are some redirects in other languages." (This is part of the Speedy Deletion policy.)

"[RfD] is the place where potentially problematic redirects are discussed."

The former is about erroneously written Redirects. The latter about "problematic" Redirects, which is evidently rather beyond simple errors. One is lower court, the other higher court. When one of the two courts has, for whatever reason, fewer cases than the other, you do not integrate them into one court. It's just confusing. -The Gnome (talk) 22:32, 20 February 2026 (UTC)

• (Not treating this is as a formal RfC for now.) From looking at R3 deletions on and off, I seriously think that one third to one half of deletions under R3 are incorrect as the criterion is worded. A large chunk of those would be kept at RfD. An additional proportion would be deleted anyway under G7 or some other CSD (G6, I guess). That would suggest the criterion is harmful as things stand. Incorrect deletions is a problem that plagues CSD. But it's particularly worth noting in an instance with (a) such a high error rate and (b) when the value from deletion is negligible compared to most speedy criteria. J947 ‡ ^edits 06:14, 21 February 2026 (UTC)

  J947, where did you get one third to one half of deletions under R3 are incorrect from? Cryptic gave some stats that imply that it's <10%. (please do not ping on reply) Primefac (talk) 12:34, 21 February 2026 (UTC)

  Cryptic's stats imply no such thing because they assume that just because a page stayed deleted, deleting it with R3 is correct. This is not the case as one can see with Pianie which was incorrectly deleted per R3 even though it was not recently created, undeleted and then deleted at RfD. A better way to see how often R3 is misused is to count the R3 entries in Wikipedia:Database reports/Possibly out-of-process deletions#Not recently created (various criteria) which still severely undercounts because it only considers timing and not if the redirect is an implausible typo/misnomer. Warudo (talk) 14:24, 21 February 2026 (UTC)

  About 6% that are restored/recreated, anyway, which is going to have errors both ways. There's certainly more that are incorrect and still redlinks. 0.9999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999->0.999... was certainly plausible, for example (though also a G4 several times over, most recently at Wikipedia:Redirects for discussion/Log/2019 February 7#0.99999999999999999999999999999999 that I can immediately find), as was DRIVER IRQL NOT LESS OR EQUAL->Blue screen of death. But a third to a half? Nowhere near. —Cryptic 14:29, 21 February 2026 (UTC)

  I note the only deletion discussion linking that 255-byte title was Wikipedia:Redirects for discussion/Log/2025 April 5#0.99999...., which closed as no consensus (and didn't consider the max-length redirect specifically). Wikipedia:Redirects for discussion/Log/2019 February 7#0.99999999999999999999999999999999 considers only a much shorter title. In contrast, Wikipedia:Redirects for discussion/Log/2025 September 22#Redirect to pi shows precedent for a similar max-length title redirect to exist. Anomie⚔ 20:13, 21 February 2026 (UTC)

  It's anecdata. The vast majority of incorrect deletions are unchallenged. I'll try to give a sample, but I can't easily check the target of the redirect, or who nominated it for deletion (was it the page creator?), or sometimes its creation date (normally accessible via log entries, but occasionally the creation doesn't appear in the logs for whatever reason).

  Here's the last 20 deletions in 2025. Not using the most recent deletions, because there's been a lot deleted in the past day largely deleted by the same admin and created by the same editor, so that sample is unrepresentative. (Though there are a fair few incorrect deletions in that lot anyway...)

  • Wikipedia:Stupid-to-Use, Panicking, Idea-Destroyer's Deleted Article with a Freaky Title → Wikipedia:Deleted articles with freaky titles – good deletion, also G1
  • Wish List (so → Wish List (song) – good deletion
  • Fent (george floyd) (disambiguation) → ? – good deletion, possibly also G14
  • Wikipedia:Gillingham Borough Council → Gillingham Borough Council – incorrect, namespace errors aren't what R3 is meant for; use G6-error instead
  • Kijevo (Klina ( → Kijevë – good deletion, also G6-error
  • Draft:Green Alchemy: Harvesting Nickel Through Hyperaccumulator Plants → Draft:Metalplant – iffy, but the target was deleted right after anyway
  • User:Omarjonesfilm/sandbox → Draft:Omar Spencer Jones – incorrect, created by page move from long-standing title and not a typo/other misnomer, loads of these sorts of redirects exist and would be kept nineteen times in twenty at RfD
  • Windows Three Point One → Windows 3.1 – incorrect, not implausible (see Wikipedia:Redirects for discussion/Log/2026 January 4 § Windows Ninety Five)
  • Ganda language) → Luganda – good deletion
  • Harambee, North Side, Milwaukee (Second Article) → Neighborhoods of Milwaukee#Harambee – good deletion, if an atypical R3
  • Harambee, North Side, Milwaukee (First Article) → Harambee (neighborhood) – same as above
  • Microsoft.live → Microsoft Live – iffy, seems plausible enough to me, but it was G6-error and G7 anyway
  • Draft:Lost in Harmony → Lost in Harmony – incorrect, not an implausible misnomer; should be kept per WP:RDRAFT
  • Shell-replacing state → Puppet state – incorrect, alternative names and related topics aren't misnomers (this one looks very obscure at best, but that's RfD fodder not R3's job)
  • Wikipedia:Lost in Harmony → Lost in Harmony – incorrect, namespace errors aren't what R3 is meant for; use G6-error instead (G7 also in this instance)
  • Draft:Cosmology → Cosmology – incorrect. This isn't a draft, it was just created as a redirect in draftspace. Occasionally these redirects show up at RfD, and opinions are split between leaving them alone or deleting them. I guess speedy deletion might be appropriate than RfD deletion, but it seems a bit harsh for what are placeholder or experimental redirects that will be eventually G13ed anyway. Regardless, it's not really within R3's remit.
  • Wikipedia:Wikipedia:WikiProject Articles for creation/December 2025 Backlog Drive/Participants/Quinntropy → Wikipedia:WikiProject Articles for creation/December 2025 Backlog Drive/Participants/Quinntropy – good deletion, but it was G7 anyway
  • Draft:Move/KVGS → KVGS – incorrect, technically ineligible due to being created by page move from a non–recently created page, just use G6 instead
  • Vishnu Siva Temple → Sri Siva Vishnu Temple – incorrect, not recently created and This is a vague and ambiguous term that could refer to multiple topics is an RfD argument, not anywhere near an R3 argument (this was undeleted by Pppery)
  • Jeremy Jahns ( YouTuber) → Jeremy Jahns (YouTuber) – good deletion but the target was deleted moments after anyway

  So that's 9 in 20 that were bad R3s. Maybe more – I couldn't check creation dates for all of them. This set had what felt like an unusually high number of non-mainspace redirects and redirects where R3 was a secondary rationale for deletion. It's certainly not big enough a sample size to draw definite conclusions from, but it should impress on the doubtful that there's a lot of incorrect R3 deletions going around, and of the correct ones, a very high amount qualify for another criterion anyway. J947 ‡ ^edits 23:21, 21 February 2026 (UTC)

  Re the "recently created" criterion, I've been monitoring that at WP:Database reports/Possibly out-of-process deletions#Not recently created (various criteria). And these days if something that's years old pops up there I will undelete on-sight (as happened with Vishnu Siva Temple) * Pppery * _{it has begun...} 00:32, 22 February 2026 (UTC)

• I would strongly encourage supporters of change to provide more concrete evidence of misuse or ambiguity in this criterion rather than simply stating claims. Mz7 (talk) 20:47, 21 February 2026 (UTC)
• Also noting that redirects lacking a space before a parenthetical disambiguation (previous dealt with through WP:X3) fall under R3 if newly created, which was one of the arguments for deprecating X3, and is in my opinion a reason to keep it. Chaotic Enby (talk · contribs) 23:13, 1 March 2026 (UTC)

• Opppose, I've editors are of the belief that the criterion is not worded sufficiently well then address that. I see no good argument to move more work onto RfD. TarnishedPath^talk 02:17, 2 March 2026 (UTC)
• Oppose No convincing reason has been floated above for the removal of R3. ^Thanks,L3X1 ◊distænt write◊ 18:22, 3 March 2026 (UTC)
• Oppose. In my experience, while not frequently-used, this criterion is still useful. The analysis data provided above does not make a compelling case that R3 should be retired (but kudos to the user who compiled it for us). Arcticocean ■ 10:34, 7 March 2026 (UTC)

G2, which is a test draft, please delete this i didn't know another template existed, this falls under G2. IsHorse, the Khan of the Universe and Ungulates (Please don't click this) 11:27, 6 March 2026 (UTC)

Done, but for future reference you don't need to post here, placing a speedy deletion tag on the pages is sufficient. Also, WP:CSD#G7 is the best tag to use when you want to delete a page you are the only significant contributor to. Thryduulf (talk) 11:53, 6 March 2026 (UTC)

I'm trying to get a feel on how other admins/users view G15 being used when the concern is not that the sources being cited are Implausible, non-existent, or nonsensical, but that they do not support the text they are being cited to support.

I know source-text integrity is low when LLMs are used; but humans also misuse and misquote sources all the time, and besides, that is not what G15 says (nor is it even listed as a Wikipedia:Signs of AI writing. However, on Draft talk:1300 New York Ave NW the nominator asserted that "Source to text issues like this are normally accepted as meeting the G15 criteria."

Are other admins deleting for G15 when, for example, the page in question asserts that a building has "capacity for 1,200 automobiles" but the source says 1,300, or various other statements where the source does not directly support the text it is being cited for? I've seen human editors do that forever - sometimes it's sloppy, sometimes it's malicious, and sometimes it's a result of the nature of collaborative editing. If other admins really are treating it as proof of unreviewed LLM generated content, I think we've got a BIG problem. ~ ONUnicorn^{(Talk|Contribs)}problem solving 19:13, 12 March 2026 (UTC)

I agree with you, we should not be deleting such things as G15 - indeed we should never be speedily deleting things for reasons not listed in one or more speedy deletion criteria. In the example you give, the difference could easily just be a typo. If there are things people think should be speedily deleteable but are not currently then the correct course of action is get a consensus for a new or amended criterion that meets all the NEWCSD requirements. Thryduulf (talk) 19:18, 12 March 2026 (UTC)

Although not a G15, check to see if the article is a hoax. Graeme Bartlett (talk) 07:55, 13 March 2026 (UTC)

With that caveat that it's been a few months since I regularly patrolled CSDs: Small errors like 1,200 versus 1,300, no, I wouldn't delete. Large, blatant, and repeated errors, though, are the gold standard in identifying LLM writing. Absence of source-to-text integrity is stronger evidence than any stylistic tell ever can be, and is evidence of a real problem with the content. I think I've deleted for that before. Though usually there were other issues as well, I think it's reasonable to consider such issues to fall under "implausible, non-existent, or nonsensical references" – just as a reference about a completely unrelated topic is implausible or nonsensical, a reference about a related topic that supports little or none of the text is implausible or nonsensical. Toadspike [Talk] 12:56, 13 March 2026 (UTC)

Looking at the article, I wouldn't call that a G15. Ref3 doesn't qualify as part of the criteria, and ref6 supports the entire sentence that it's next to which is sufficient. If the citation was something like The structure included an Amoco gas station, automobile repair shop, car wash, and retail space.[6], then that is something that could be be part of a G15 case (but not alone) I also spot checked the one newspaper source that was also on newspapers.com and confirmed the paper in question did talk about a fire at "The Rink". Jumpytoo Talk 05:02, 14 March 2026 (UTC)

See Wikipedia:Miscellany for deletion/JeiAllenYes's user pages. I'm not sure if anything could or should change to accommodate a situation like this, because it's fairly unusual, but I'm just putting this link out there in case others have further comments. Graham87 (talk) 10:20, 15 March 2026 (UTC)

One could plausibly argue that altfic would be a good WP:G3 candidate. Primefac (talk) 12:52, 15 March 2026 (UTC)

The need for an MFD (or waiting a few months) in some cases was intentional in the replacement of U5 with U6/U7. Wikipedia:Replacement of CSD U5 FAQ has details. Wikipedia:Village pump (miscellaneous)/Archive 87#Alternate History in Sandboxen and Wikipedia:Village pump (miscellaneous)/Archive 87#Alternate History also seem relevant to this particular case. Anomie⚔ 15:00, 15 March 2026 (UTC)