Wikipedia talk:Revision deletion


 * Previous discussion at Wikipedia talk:Selective deletion

Revdel email in an edit summary?
I thought we revision-deleted email addresses given in edit summaries, particularly when it is from a throw-away IP such as Special:Contributions/2001:8F8:153D:447B:3DCD:6B3F:A44:C8C4. Before doing that I thought I would read the documentation to see what RD number I should use. Surely not RD4 Oversightable information? I'm not going to bother an oversighter for simple disruption like this? And if it were RD4, I can see the wise advice to not use RD4 but instead to just email oversight. If that is really intended, the documentation should say so. Johnuniq (talk) 06:06, 20 April 2023 (UTC)

Revdel and block logs

 * Full previous discussion. It's very long so I've tried to summarise the relevant bits here.
 * In the above block review discussion on AN/I, it emerged that EEng's been the victim of some bad blocks. Users were quoting the length of EEng's block log as grounds to indef him.  Ritchie333 remarked on how a high proportion of EEng's blocks weren't well founded, and I added that I wish we could edit people's block logs.  Ritchie333 tested and found out that in fact we can, using REVDEL.  On a technical level sysops can do this.Ivanvector challenged whether it would be appropriate because amending people's block logs would reduce sysop accountability, which is a fair and understandable concern, but I think that this use of REVDEL leaves the block in the performer's log.  It only erases it from the target's log.The Wordsmith pointed out that on a policy level, we can't do this.  WP:REVDEL says: My view is that many Wikipedia editors are living people, and WP:BLPDEL protects us.  For some, having bad blocks in their block log has caused measurable harm and distress.  I don't know whether it's harmed EEng because I haven't asked him; but I'm morally sure that it's harmed others.Therefore I'm minded to propose that we change WP:REVDEL to permit Arbcom, checkusers, and oversighters to modify block logs at their discretion, and sysops to amend block logs if there's consensus to do that at a community block review.  That kind of change would need a community RfC, but I thought it best to ask for your thoughts and advice here before I start one.—S Marshall T/C 15:49, 14 January 2024 (UTC)
 * Just noticed this discussion. Once again, the poor capabilities of MediaWiki are requiring masses of unnecessary community effort.
 * 11 years ago at I requested that block logs be upgraded to allow linking to discussions, instead of forcing admins to perform the incredibly crude step of an additional (un)block action to get a link into the log. This would allow us to leave the logs themselves inviolate but visibly corrected - imagine seeing a block entry accompanied by a link titled "Incorrect block of Username at 2024-01-01", or "Log of incorrect blocks/Username 2024-01-01", or whatever.
 * I'd welcome any public support for the idea. —  Scott  •  talk  12:20, 13 July 2024 (UTC)
 * A 1-second block does get the job done, but I agree that it's a kludge. Having a more elegant solution would be great, I'm just wary of allowing admins to edit logs at their discretion like that. It seems like more of a Functionary (Oversight) job, or possibly Bureaucrat. But yes, I'd endorse the general idea of upgrading software to make this possible. The Wordsmith Talk to me 16:52, 15 July 2024 (UTC)
 * Thanks for the ping. I'm against removing anything, ever, from a sysop's action log (i.e. the log[s] of actions performed by the sysop), but it maybe was shown that it's possible to revdelete a log entry in a way that it's hidden in the user's block log but visible in the sysop's. I say maybe because it seemed to leave behind a database error, and so I think the jury remains out on whether or not we technically can do this without causing problems, but the discussion became silly and I stopped paying attention before that was resolved. I am not strictly opposed generally to removing actions from user logs (i.e. the log[s] of actions on the user), but this isn't the time to have this discussion, framed as it is not around what's good for the community but around something that EEng's circle of friends doesn't like. Bad cases make bad policy, and all that. Ivanvector (Talk/Edits) 19:13, 14 January 2024 (UTC)
 * I am not now, nor have I ever been, a member of EEng's circle of friends.—S Marshall T/C 21:14, 14 January 2024 (UTC)
 * I can confirm that -- we're bitter enemies. While Ivanvector licks his wounds, let's continue the useful discussion. I don't think it's necessary (or desirable, actually) to remove/revdel entries for "bad" blocks; rather, they should be annotated somehow. You'd think that an adjacent unblock entry saying, "Unjustified block, see ANI diff" would be enough, but there's plenty of evidence that there are people who don't take the trouble to notice such things. (If the determination of unjustified-ness comes only after the block's expired, a dummy 1-second block can be enacted with a description referring to the prior block. But then we're still back to: there are people who just don't notice.) I think it would be enough to modify a given block's descriptive text -- if people overlook an all-caps THIS BLOCK OVERTURNED right there in the block itself, then there's nothing we can do for them -- but I don't know if that's technically possible. An analogy would be modification of edit summaries, and that's not possible either (though I don't know if that's for technical reasons, policy reasons, or just dev laziness.) EEng 00:07, 15 January 2024 (UTC)
 * I'm generally opposed to removing 'bad blocks', or at least I've never heard a solid proposal I'd agree with, and think that people should learn to properly contextualise them rather than just count them. I do just want to offer some insight on the technicals. I don't know what's going on with the error in ThisIsaTest's block log, however, I don't think it would work as it's described above, at least with the current software. If you remove the edit summary you'd still have a log entry for the user, and still a long list of blocks, but every non-admin will see no reason provided. And it will be exactly the same for the admin's log. For an example see the block log of 37.186.45.90 (the admin also shares this log). You'd have admin-only access to the reasons (via the revdel log). Limiting access to admins seems of questionable added value to me. And by the way, a not-well-enough-known fact: log redactions have their own logs. Go to Special:Log and enter Special:Log/block (or any other log) as the target. They can be very interesting on occasions. -- zzuuzz (talk) 01:59, 15 January 2024 (UTC)

Process for requesting revision undeletion
I couldn't find a description on this page of how a non-admin can request revision undeletion, short of contacting an admin directly. Some images with deleted versions don't actually meet RD1 -- usually because the image was tagged as non-free at some point, but is now free (e.g., copyright expired recently). Should we have a centralized process/tag for these? Wikiacc (¶) 21:48, 17 April 2024 (UTC)
 * I assume you mean file revisions deleted per F5. Those can be requested at Requests for undeletion. —&thinsp;JJMC89&thinsp; (T·C) 22:20, 21 April 2024 (UTC)
 * Thanks . I'd like to add text to the main page (likely in the "Appeal and discussion of actions" section) to make this clear. Would the following text be an accurate description of existing policy?
 * To reverse uncontroversial revision-deletions made under RD5 or RD6 (including image revision-deletions under WP:F5), make a request at Requests for undeletion.
 * Wikiacc (¶) 01:40, 22 April 2024 (UTC)
 * That seems reasonable to me. (Other cases are best discussed with the admin that performed the deletion.) —&thinsp;JJMC89&thinsp; (T·C) 02:14, 22 April 2024 (UTC)
 * I’m not sure that revision deletions made under RD5/RD6 are currently eligible to be undone at WP:REFUND. I don’t recall seeing an RD5 revdel recently so I can’t comment fully on that specific case right now (although, if an admin has a valid reason for deleting something under the deletion policy, it seems like something that might need to be appealed in the normal way rather than being unilaterally undone by editor request). However, some of the revdels I’ve requested in the past have been to delete something that was missed when an admin performed an earlier revdel. Some of these were logged under RD6 (i.e., as correction of clear and obvious unintended mistakes in previous redactions), but these aren’t (imo) the type of redactions that should be able to be undone at WP:REFUND. All the best, &zwj;—&zwj; a smart kitten [  meow ] 06:55, 22 April 2024 (UTC)
 * Striking RD6. Since RD5 is supposed to incorporate the standard deletion policy, I'd think standard undeletion processes should apply (including REFUND when the deletion was uncontroversial). Wikiacc (¶) 21:58, 22 April 2024 (UTC)
 * I see where you’re coming from - my worry would be that the current wording might increase the scope of what revdels can be undeleted from what the current practices are, as could be read as implicitly stating that all revdels made under RD5 are uncontroversial and can be REFUNDed. My worry is that this is beyond what is current practice, and would therefore (in effect) represent a change to the policy. All the best, &zwj;—&zwj;  a smart kitten [  meow ] 07:08, 26 April 2024 (UTC)

Point taken that my suggested wording is ambiguous in scope. How about this rewrite?
 * To contest or reverse revision-deletions, discuss with the deleting admin. For revision-deletions made under RD5, you may instead follow standard undeletion processes:
 * For speedy deletions or deletions that resulted from discussions, request at Deletion review.
 * For uncontroversial deletions (including image revisions deleted under F5), request at Requests for undeletion.

Wikiacc (¶) 16:15, 27 April 2024 (UTC)


 * That looks mostly okay to me - just a few things that came to mind:
 * I’d suggest potentially changing to something like : this takes out the term speedy deletions (which, as far as I’m aware, isn’t generally used to describe revdels), and also makes clear that a discussion with the deleting admin would generally be expected before filing a DRV.
 * Would this be incorporated into #Appeal and discussion of actions? If so, would the existing paragraph there need to be modified at all?
 * I’m not sure whether introducing DRV as a review forum for RD5 revdels would represent a change in policy or not — on the one hand, RD5 is for deletions made under the deletion policy; but on the other hand, WP:REVDEL currently states that revdel reviews should take place at WP:AN. I guess I’m neutral on this bit.
 * As a slight side note, do you mind if I drop a note at Wikipedia talk:Deletion review about this discussion?
 * All the best, &zwj;—&zwj; a smart kitten [  meow ] 20:01, 27 April 2024 (UTC)
 * Strikeouts added. And yes, please go ahead and drop that note. To your specific points:
 * I'd rather not codify that a discussion with the deleting admin must come before DRV. Pointing to WP:DRV allows the language there to take precedence -- as of now, discussing with the deleting admin is described as "good practice" but "not required".
 * This language could go in "Appeal and discussion of actions" (probably above the existing text) or in a different section. I don't think the existing language needs to change.
 * I'm also not sure; I'd just like the language to be consistent with current practice. If DRV is explicitly not for RD5 revdels then that should be made clear at WP:DRV.
 * Wikiacc (¶) 22:05, 27 April 2024 (UTC)
 * Re 1, fair point - in which case, it might be worth swapping the two bullet-points and replacing  with something like, as otherwise that sentence could potentially be read as implicitly excluding single-admin-action RD5 revdels from DRV.Re 3, I’ve dropped links to this discussion at WT:DRV & WT:DELPOL, but - if there aren’t any objections now - the wording can always be changed after-the-fact if editors don’t think DRV should be the review forum for RD5s. All the best, &zwj;—&zwj;  a smart kitten [  meow ] 07:27, 29 April 2024 (UTC)
 * Just noting that a policy page really shouldn't be changed if the only discussion is between two editors with tacit approval from a third after less than a fortnight. Primefac (talk) 09:39, 29 April 2024 (UTC)
 * Fair enough - I guess I just felt like I didn't want to object to the changes I'm personally neutral about solely due to the fact that other editors might object, as I'd worry about that potentially being slightly WP:POINTy. &zwj;—&zwj; a smart kitten [  meow ] 09:45, 29 April 2024 (UTC)

Thanks for the suggestions. Rewriting:
 * To contest or reverse revision-deletions, discuss with the deleting admin. For revision-deletions made under RD5, you may instead follow standard undeletion processes:
 * For uncontroversial deletions (including image revisions deleted under F5), you may request undeletion at Requests for undeletion.
 * For all other revision-deletions made under RD5, you may request review at Deletion review.

Again, the purpose of this addition is to describe current practice, not to change policy. So comments are especially welcome on whether the text describes current practice. Wikiacc (¶) 22:02, 29 April 2024 (UTC)


 * I don't personally take any issue with this wording. F5 revdels are commonly undeleted at REFUND (with 5 completed F5-reversals on the current version of the page), so that describes current practice as far as I'm aware. Not that sure on the use of DRV as current practice for RD5s, so I'll leave that in case anyone else wants to make a comment on that front - however, I'll note that my found these possible previous revdel reviews: two in which an RD1 revdel appeared to be endorsed (the first speedily) [] [], one in which an RD1 revdel review was moved to WP:AN [], and one review endorsing a no consensus close to an AfD that proposed revdels under RD5 []. All the best, &zwj;—&zwj;  a smart kitten [  meow ] 12:35, 1 May 2024 (UTC)

Given the mixed DRV record above, I'll cut down the suggested language just to cover F5:
 * To contest or reverse revision-deletions, discuss with the deleting admin. For image revisions deleted under F5, you may instead request undeletion at Requests for undeletion.

If there are no objections I'll put this on the page soon. Wikiacc (¶) 21:43, 23 May 2024 (UTC)
 * Added.. Wikiacc (¶) 23:45, 30 June 2024 (UTC)

"Material must be grossly offensive"
Re: your revert: Congratulations! You've just switched off criteria 1, 3, 4, 5, 6, and 7. I hope for your sake that you never have to review any contracts. —  Scott  •  talk  01:35, 12 July 2024 (UTC)
 * You don't get to edit policy pages to try to get yourself out of a jam. Gonna quote RoySmith at the AN thread here "Editing WP:REVDEL to support his position was probably a troutable offense, but let's all just take a deep breath and move on to something productive." Also stopping with the hyperbole would be helpful. Other admins have done just fine with the previous wording. ♠ JCW555  (talk)  ♠ 01:59, 12 July 2024 (UTC)
 * I don't suspect you're correct about Scott's motives. Separate from the mind reading, do you really support the continued inclusion of "must be grossly offensive". It's plainly out of step with the rest of the policy, and it does not align with how revdel is used. If I revdel copyrighted material, or someone's phone number, or an IP address, am I violating this policy? Firefangledfeathers (talk / contribs) 02:04, 12 July 2024 (UTC)
 * I don't support or oppose it, I just oppose his unilateral editing of policy to even give the appearance of trying to get himself out of the jam he's currently in by changing the policy to fit his stance. To quote Scott himself "The "misuse" section of the preamble, not the criteria themselves which define the use of revision deletion, said that for revision deletion to be used, the "material must be grossly offensive". As I have pointed out multiple times now, that invalidates every single criterion except RD2. Somehow nobody noticed that since 2009 the revision deletion policy has been contradicting itself." Maybe the reason why people haven't noticed since 2009 is that Scott's view of this is idiosyncratic, and judging on the thread on AN, it seemingly is. ♠ JCW555  (talk)  ♠ 02:13, 12 July 2024 (UTC)
 * Would you object to my restoring the change? I promise my reason is not to get Scott out of a jam. Firefangledfeathers (talk / contribs) 02:18, 12 July 2024 (UTC)
 * Given that there's clearly some controversy about this, I think the best thing would be to propose a change here on the talk page and see if people object. RoySmith (talk) 02:20, 12 July 2024 (UTC)
 * Will do. Firefangledfeathers (talk / contribs) 02:24, 12 July 2024 (UTC)

The context here is that §Misuse has a line saying Scott and I feel that this contradicts most of the rest of the policy, which enumerates multiple reasons for revdel beyond grossly offensive material. Scott has proposed changing that line to say This retains the hyper-limiting spirit of the line while bringing it into alignment with the rest of the policy. I support the change. Firefangledfeathers (talk / contribs) 02:28, 12 July 2024 (UTC)


 * The proposed change seems like a clear improvement; I support it. DanCherek (talk) 02:30, 12 July 2024 (UTC)
 * That sentence specifically contextualize the previous sentence, but the paragraph break separates it in an undesirable way. Rather than removing that sentence, it would be better to combine the two as clarification regarding the point being made: Especially, RevisionDelete does not exist to remove "ordinary" offensive comments and incivility, or unwise choices of wording between users, nor to redact block log entries; such material must be grossly offensive, with little likelihood of significant dissent about its removal. Grandpallama (talk) 02:34, 12 July 2024 (UTC)
 * I like that combination. -- SarekOfVulcan (talk) 02:41, 12 July 2024 (UTC)
 * This proposal sounds good to me too. Firefangledfeathers (talk / contribs) 02:42, 12 July 2024 (UTC)
 * +1. The next section could nevertheless begin with "Material must strictly fall under one of the following criteria:". Levivich (talk) 18:26, 13 July 2024 (UTC)
 * Maybe replacing grossly offensive with grossly disruptive could still make the point while providing latitude for other criteria than RD2? Chaotic Enby   (talk · contribs) 02:37, 12 July 2024 (UTC)
 * Maybe both? grossly offensive or disruptive? -- SarekOfVulcan (talk) 02:41, 12 July 2024 (UTC)
 * This would be an incremental improvement, but there are still RD criteria that apply to more than just disruptive content. Firefangledfeathers (talk / contribs) 02:42, 12 July 2024 (UTC)
 * I suggest deleting the "misuse" section in its entirety. Other sections indicate why this is a high-risk action, and the giant green box below duplicates guidance. introduced for administrators in 2010 can be moved somewhere else. Ca talk to me!  06:23, 12 July 2024 (UTC)
 * Best option, and the only one so far to see the bigger picture as we've been focusing on minutiae. The section is very characteristic of much of the policy-adjacent material written circa 2008 - verbose, lecturing, and redundant. Everyone benefits from cleaner, tighter policy pages. —  Scott  •  talk  08:53, 12 July 2024 (UTC)
 * I think the green box should be dismantled and merged with the "Misuse" section, which should be retained. —Alalch E. 10:01, 12 July 2024 (UTC)
 * For sure. One, the other, or a hybrid, just not both. —  Scott  •  talk  10:43, 12 July 2024 (UTC)

I'd recommend reading the talkpage archives,  specifically the first, when this ability was being rolled out. The "grossly offensive" language wasn't casually chosen, and was discussed at length, so we shouldn't be too cavalier about removing it if it can just be better clarified or incorporated. Grandpallama (talk) 02:48, 12 July 2024 (UTC)
 * Support Scott's edit per the comments made above. And I favor that over the copyedit proposed by Grandpallama that retains "must be grossly offensive", as the material does not in fact have to be grossly offensive to fall under some of the criteria under some of the circumstances. In addition, I propose changing "RD3. Purely disruptive material" to "RD3. Substantially disruptive material", as almost all vandalism is purely disruptive (vandalism is editing (or other behavior) deliberately intended to obstruct or defeat the project's purpose—how can such edits be anything other than purely disruptive, except in some unusual instances or under extraordinary circumstances), but the intention is not to cover all or almost all vandalism, so that name for the concerned criterion isn't descriptive, which is evident from the following text that defines the criterion which really does not describe most vandalism: harassment, grossly inappropriate threats or attacks, browser-crashing or malicious HTML or CSS, shock pages, phishing pages, known virus-proliferating pages is rather bad vandalism, significantly worse than average vandalism, and unlike much of vandalism that is only notionally or mildly disruptive since it is easy to detect and undo—while still being exclusively disruptive and nothing but disruptive ergo "purely disruptive"—the listed forms of vandalism are substantially disruptive, and need more drastic mitigation.—Alalch E. 09:57, 12 July 2024 (UTC)
 * Great analysis of why the current wording of RD3 is imprecise. That would be an excellent refinement. —  Scott  •  talk  10:48, 12 July 2024 (UTC)
 * I think an intensifier would be good, but "substantially" can also mean "true down to the very substance/essence of the thing, not just superficially true". As you say, all vandalism is disruptive, and substantially so, by that meaning. Could we just use "grossly"? Firefangledfeathers (talk / contribs) 12:07, 12 July 2024 (UTC)
 * Maybe "severely disruptive" would be better to have some variation because "Grossly insulting, degrading, or offensive material" uses "grossly" and is also "grossly disruptive"; actually, since R3 is effectively a bit of a catch-all relative to R2, and however we phrase it: "severly disruptive", "grossly disruptive", "substantially disruptive", its label would encompass R2, so I suggest "RD3. Other severely disruptive material". —Alalch E. 12:45, 12 July 2024 (UTC)
 * Looks good to me. Firefangledfeathers (talk / contribs) 12:50, 12 July 2024 (UTC)
 * Good catch. Although none of these intensifiers are quantifiable per se - how many vandals dancing on the head of a pin does it take to be "severely" disruptive? 😉 Still a definite improvement. —  Scott  •  talk  12:57, 12 July 2024 (UTC)


 * Well, here is my screed: revision deletion fundamentally breaks the model of open collaboration on which Wikipedia is built, it makes us unauditable by the public and forcibly inserts administrators as an unaccountable class of kommissar intermediaries between people and the provenance of their knowledge resources. My firm opinion is that it should, generally speaking, not be done unless absolutely necessary, such as text or images that have been made literally illegal to view. Using it for, or even  , is like using nerve gas on rowdy schoolchildren: not worth it. <b style="font-family: monospace; color:#E35BD8"><b style="color:#029D74">jp</b>×<b style="color: #029D74">g</b>🗯️</b> 07:55, 13 July 2024 (UTC)
 * JPxG, I don't think anyone is proposing that here. Did mean to comment at the AN thread? W ADroughtOfVowels P 10:29, 13 July 2024 (UTC)
 * Wheo! That's one spicy comparison JPxG is making and I can't endorse it. But it is worth me reposting and elaborating on something I said at the tail end of that thread because this is the right place for it.
 * The whole difference of opinion over whether it's right to revdel "normal" vandalism is a subset of the ancient debate of inclusionism and deletionism. It's been forced to extremes, because when you boil down the issue what's left is that our software is simply not good enough. I think that vandalism should be hidden away (not "deleted" - the fact of the system being called "revision deletion" is just one of many issues with it) for two reasons: to deny recognition to vandals, and to minimize disruption to people reading page history. Revdel as it exists today is a crude and blunt tool with almost no nuance whatsoever. You have two options for a given unit of data (actor, revision text, edit summary) - leave it on public view, or punt it into the Phantom Zone where only admins can read it. There's no gradation that can be applied when hiding revisions.
 * By contrast, consider a system where applying, for example, RD2 would make an item inaccessible to anyone without elevated privileges, but RD3 would hide it behind an additional interface element until a button is clicked. Minimal visibility for disruptive material without presenting any barrier to edit filter managers, recent changes patrollers, etc. This isn't a new idea at all: various social media apps have had it for a long time in the form of "hidden replies". It's a middle ground which raises no hard barriers but also keeps the house clean.
 * Incidentally... we are already an unaccountable class of kommissar intermediaries between people and the provenance of their knowledge resources, because no effort has been made at the system level to have it otherwise. 99.9% of Wikipedia readers don't even know that talk pages exist, or that there are fora in which they can seek redress for perceived issues - let alone that there are page histories, logs of when admins renamed things, etc. The WMF, for all its posturing as an organization "making knowledge available to everyone", has made precisely zero effort on developing an accountable knowledge environment for the general public through accessible design. This project is still as much of a technocracy as it was twenty years ago.
 * The problem is also being worsened by parasitic third parties extracting our content, such as the snippets on Google's search results which isolate people from knowledge provenance even further - not to mention the current "AI" trash craze. While those are out of our control, the WMF has had the opportunity to increase fairness in our own precincts for decades but chooses to ignore it and focus on technicalities instead. Accessibility improvements are obviously to be welcomed but we need acknowledgement of ethical matters as well, and commitment to do better. —  Scott  •  talk  11:37, 13 July 2024 (UTC)