Jump to content

Wikipedia talk:Revision deletion/Archive 3

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1Archive 2Archive 3Archive 4Archive 5

RD request example library

I wanted to suggest that in the interests of greater consistency and better guidance for both users and admins, we create a subpage (Wikipedia:Revision deletion/examples) listing examples of different RD criteria in action, with cases of "rev-deletable under this criterion" and "not quite rev-deletable". Obviously examples would need to be appropriately anonymised and presented without diffs or anything. I think this could be very helpful, but I suspect it's one of those things which won't happen unless I do it myself... and I'm trying to be on a long-term wikibreak. Still, I'm mentioning it in case somebody wants to step up... Rd232 talk 09:49, 21 December 2010 (UTC)

Lovely and useful idea. Any volunteers?--SarekOfVulcan (talk) 03:49, 22 December 2010 (UTC)
It's an interesting idea but there are going to be some problems. Right off the bat with the first criterion, how are we to show what a copyright violation looks like without repeating it? And do we really need to explain what grossly insulting or offensive material looks like? Maybe negative examples would work better. Show what we won't revdelete and explain why. Beeblebrox (talk) 04:23, 22 December 2010 (UTC)
I went ahead and created a framework to build the page on, we can continue this discussion at Wikipedia talk:Revision deletion/examples. Beeblebrox (talk) 04:37, 22 December 2010 (UTC)
I think it's a good idea, too. If we needs to show a copyvio (not sure we do), we can mock one up using pd material. --Moonriddengirl (talk) 11:35, 22 December 2010 (UTC)
Brilliant idea - I've been thinking about something similar for a while but haven't got round to suggesting it anywhere yet. What I was thinking off was a random sample of revision deletions (excepting copyright ones) suitably anonymised etc (including removing who done it) so as to give users a stronger idea that RevDel was being used as the community had decided and to allow discussion of what is and isn't appropaite for RevDel. Personally I think different admins are interpreting the guidelines quite differently although as a non-admin it is hard to know for sure. The idea suggested above would certainly allow discussion of what is and isn't appropriate although do less to give confidence to users that it is being used correctly. Obviously taking stuff out of the logs like this is likely to require a reasonable consensus given that it was hidden for a reason. Dpmuk (talk) 16:18, 6 January 2011 (UTC)

Environmentally sensitive locations

On the Lechuguilla Cave article, a revision was deleted that contained coordinates (presumably) of an environmentally sensitive cave. In this case, the coordinates are intentionally not published by the United States government and, as far as I know, the coordinates are not published by any reliable source (primary or secondary). However, the revision deletion policy doesn't seem to cover this type of edit.

Apparently this issue has come up elsewhere, with the location of a rare flower in the United Kingdom. That led to the failed proposal at Wikipedia talk:Sensitive wildlife locations.

There are two separate issues: what to do when there are (quality) reliable sources that publish the location, and what to do when there are no sources. In the former case, I don't think that there is anything Wikipedia can do: we are bound to follow the sources. In the second case, it isn't as clear.

Question: Should revision deletion apply to the locations of environmentally sensitive objects, when no high-quality reliable source can be found for the location?

I think it would be nice to clarify this. — Carl (CBM · talk) 02:28, 1 January 2011 (UTC)

  • Absolutely not. I can fully support removing unsourced coordinates from articles. WP:V is clear on that, if we cannot verify the location, the coordinates should be removed. However, there is no compelling reason to RevDelete the information. We don't revdelete every single inappropriate edit to any article, it is only used where the posted information is likely to be abused or is eggregiously inappropriate. A long buried edit of a faultily sourced geocoordinate is unlikely to be abused in any way. If there is evidence that it is being abused, then sure, perhaps in single instances we could enact a revdelete. (I can't actually fathom a way in which a removed geocoordinate, buried in the article history, could be "abused", but I can admit there are more imaginitive people than I) However, these instances are likely to be so unbelievably rare that there does not need to be any policy guidance on this. We don't use revdelete just to cover up routine WP:V removals. Remove the text and move on. --Jayron32 03:29, 1 January 2011 (UTC)
  • In the past, we have had instances where such information, which is being deliberately withheld by authorities, has been inserted into articles and then the links to those edits (if not revdeleted) have been distributed online with the express purpose of compromising environmentally sensitive locations. This goes beyond the lack of verifiability to ensuring that our project is not being used to wreak havoc. I would equate this with the removal of non-public address information of individuals, which is routinely revdeleted (and often suppressed as well). Risker (talk) 04:08, 1 January 2011 (UTC)
    • Well there you go. Still, I think such cases are clearly under the remit of WP:IAR and we get into a slippery slope when we start trying to define "environmentally sensitive" locations for ourselves. If something is being abused, we should deal with that as each case arises. We cannot try to forsee every way someone may use Wikipedia to "wreak havoc", and it would be a very bad idea (per WP:BEANS) to write every possible thing we could think of into policy merely to allow RevDelete to be allowed to be used. Instead of using an overly broad concept like "environmental sensitivity", which is easily open to wildly different interpretations, and thus actually increases (rather than decreases) potential conflicts over the use of tools like Oversight and RevDelete, why not just leave these individual cases up to individual judgement per WP:IAR, and just deal with them. Risker has already indicated that these cases have already been dealt with. Oddly enough, they seem to have been dealt with absent any such guidance and the Wiki, as far as I can tell, has not yet exploded. I think we are safe with allowing our ArbCom, OTRS, Foundation, and Oversight teams handle these issues as they come up, and we don't need specific policy guidance on this. --Jayron32 04:29, 1 January 2011 (UTC)
      • The difficulty is that the material is added, somewhat often, by well-intentioned editors or bots. So it would be nice to know what to do in these cases, so that editors can see whether to request revision deletion or not (and admins can see whether to do it). — Carl (CBM · talk) 04:41, 1 January 2011 (UTC)
        • OK, then, how do you define "environmentally sensitive"? If there is specific evidence of abuse, or a specific request from a government authority to oversight the information, then we act on that. But who gets to define "environmentally sensitive"? I mean, the phrase is so broad as to be meaningless... Is Mount Everest not a sensitive environment? Should we suppress its location? Who gets to decide these things? --Jayron32 04:46, 1 January 2011 (UTC)
          • Remember we are only talking about material that is not found in reliable sources (that excludes Everest). For the case of caves in the U.S., see [1]: "As further established by the FCRPA, specific locations of significant cave entrances may be kept confidential and exempted from FOIA requests." Here FOIA is the Freedom of information act. That's the type of material in question: it is deliberately not released by authorities, is not published anywhere, someone just went there with a GPS and recorded the coordinates. — Carl (CBM · talk) 04:52, 1 January 2011 (UTC)
            • So why not just remove the coordinates and not worry about revdeleting them? What is the compelling reason to revdelete in absense of evidence of abuse? I can concede that in the case of specific caves which are specifically named as FOIA exempt, and for which the entrance is not noted, it may, possibly, be OK to RevDelete those, but this must be a finite list we are working from, and narrowly defined as such. We don't want people running around revdeleting everything on their own opinion of how "environmentally sensitive" something is. Furthermore, where info is clearly publicly availible in reliable sources, we should neither be removing it NOR revdeleting it. The location of the Conkling Cavern dispute that precipitated this mess is reported by the USGS, and would be entirely outside of the remit of this discussion anyways. --Jayron32 05:01, 1 January 2011 (UTC)
              • "What is the compelling reason to revdelete in absense of evidence of abuse?" As Risker has stated, there is already evidence of abuse in past cases. In future cases, by the time there's evidence of abuse, it's too late. I think we all agree that we'll retain any information that's publicly availible in reliable sources. We also seem to agree that there are (few, limited) cases where it is appropriate to RevDelete information that authorities have recognised as sensitive and that has not been publicly released. The task now is to clarify as best as we can what those cases are so they can be codified in policy, thus stopping people revdeleting excessively based only on their own opinion. Adrian J. Hunter(talk•contribs) 13:20, 2 January 2011 (UTC)
  • The earlier proposal failed, so heck no. RevDel is overkill, normal deletion is sufficient; no good arguments have been presented for why it's not. There's no need for rules beyond WP:V+WP:RS; avoid WP:CREEP! Cf. WP:COMPREHENSIVE. --Cybercobra (talk) 05:16, 1 January 2011 (UTC)
  • A similar proposal was discussed at WT:Oversight#Proposal for new 'public interest' clause.
    If anyone wants to pick this up, I'd suggest:
Removal of non-public information that if published would tend to cause clear direct harm to a person, property, or part of the natural world, by identifying or locating it. Criteria: There is some form of formal and exceptional protection granted by society to ensure the wellbeing of the person or object at risk (for instance recognition as an endangered species, witness protection, court reporting blackout), it appears that this is for its own wellbeing rather than due to censorship, the information is not already readily available to the public or academia, and publication in Wikipedia would directly undermine or negate its protection. In such cases Wikipedia should take a conservative stance and avoid being the first "high profile" publisher.
There are a number of careful drafting issues - it should cover the very few exceptional cases where a specific matter gets specific legal protection, and not be used to censor articles such as the AACS encryption key controversy (widely publicized already) or rights related to intellectual property and the DMCA (too broad). It should set the bar as being "first high profile publisher" to protect, not censor. The information should be limited to identifying or locational data. Publication in Wikipedia should actually have a real effect of undermining its wellbeing. FT2 (Talk | email) 06:01, 27 March 2011 (UTC)

Conflict of interest

Is there consensus that admins should not RevDel where there is a conflict of interest? If so I think this needs to be made clearer in the guidelines. There was a recent RfC/U where an admin deleted their own edits and I've also come across at least one admin user talk page where they've deleted a lot of revisions. It seems wrong to me that the admin in question should act in either of these as they have a clear conflict of interest. As such I'd propose adding some to the guidelines making it clear that admins should not act when they have a conflict of interest and specifically mentioning the two case I give. Dpmuk (talk) 16:24, 6 January 2011 (UTC)

WP:INVOLVED applies to all admin actions. I don't see how hiding revisions on one's own talk page should necessarily cause a conflict of interest: when someone puts a RD2- or RD3-able rant on my talk page, I don't see why I should not be able to hide it myself. However, admins should certainly not use RevDel to hide their own mistakes (as happened in the RFC you alluded to), which is already stated in the guidelines. Ucucha 16:37, 6 January 2011 (UTC)
Could you point me to the relevant bit in the guidelines that states that as I looked before posting and couldn't find it (I may have missed it). If it's only in WP:INVOLVED I see no harm in reiterating here - both to remind admins and to reassure users that may not be aware of involved. My concern about admin using it on their own user talk pages (or a user page) is that, certainly under RD2, the rant is likely to be aimed at them and so they are involved. I can see where there would be cases where it wouldn't be aimed at them, but given that these would be in the minority, I think it would be better not to allow it full stop so as to make sure there isn't even the appearance of impropriety. Dpmuk (talk) 17:24, 6 January 2011 (UTC)

Why are many RD'ed edits so old?

Hi there, I was doing some statistics over the public deletion log and an interesting point came to my attention:

In the median case (that is, over 50% of the time), RevDel is used on an edit that is OVER 1 YEAR OLD. Could someone give me some intuition on why this might be the case? I understand that RD1 cases (copyright infringement) might be discovered long after they occur forcing back-deletion. However, even when I remove RD1's from the analysis -- the same statistic holds true. Doesn't this seem like a long time for defamation and libel to hang around in the median case?

Hope someone who actually uses the functionality might help me understand this. Thanks, West.andrew.g (talk) 04:41, 14 January 2011 (UTC)

I'd imagine a large chunk of those are old Grawp and Runtshit edits/edit summaries; Grawp pagemove vandalism often had grossly insulting page names and/or came with links to shock or malware sites. Runtshit stalks one particular user, and I know they RevDeled all of his 1000+ sockpuppets' edits. Also, in general I'd think it's much easier to RevDel something than it was to selectively delete (although not being an admin, I couldn't say for sure); now that it's easier, admins are removing those revisions just to be on the safe side. The Blade of the Northern Lights (話して下さい) 05:24, 14 January 2011 (UTC)
These are two problematic users in Wikipedia's history? I'll see how often their names show up when JOINED with public-log RIDs (assuming their user-names have not been redacted). FWIW, my analysis began in August 2010 and runs through January 2011. Thanks, West.andrew.g (talk) 15:13, 14 January 2011 (UTC)
I think Blade has hit the nail on the head. A lot of cleanup of old stuff was done the first few months. Over time that pattern should reverse itself and most RevDel will be very recent edits. Beeblebrox (talk) 20:57, 19 January 2011 (UTC)
In response to the first part, see the LTA pages for both users; you'd be shocked at the havoc people can wreak if they're bored enough. Years before I got an account (early 2007) I spent some time around ED, and although I never was one of those who participated in the /b/ attacks here one can't help but notice what they were doing; I saw a few of the revdeled edit summaries before they were deleted. The Blade of the Northern Lights (話して下さい) 05:28, 20 January 2011 (UTC)

When was RevDel enabled?

Hi everyone, I've done some digging but can't quite find the answer to this one: When was RevDel enabled on en.wiki? In particular (as I mention in the above thread), I have been parsing the public RD logs plus some stats about suppression. I've found that while use of oversight/suppression are pretty stable throughout 2010 -- almost no "standard admin" RevDelete was done prior to May 2010 (76 uses in april, 2800 in may). Was something switched over at that time? Was it opened up to admins instead of just oversighters? Thanks, West.andrew.g (talk) 18:31, 14 January 2011 (UTC)

Okay, a little more digging revealed this buzilla report. RevDelete was enabled for admins on approximately May 15, 2010. Sorry for wasting bandwidth. West.andrew.g (talk) 18:38, 14 January 2011 (UTC)

RFC about Criterion 3

There is an ongoing RFC about RD3 open here. All editors are welcome to contribute. --Shirik (Questions or Comments?) 20:31, 19 January 2011 (UTC)

This page does not make it clear how to request revision deletion

I just saw a revision which needed deleting, but after reading this page I wasn't sure where to request it, so ended up requesting oversight instead. It explains thoroughly what revision deletion is, but not what a user who sees a problematic revision is supposed to do. In future, should I have contacted an admin directly? Should I post a message on WP:ANI? Why isn't there a specific page for handling REVDEL requests? It seems there used to be one, at Wikipedia:Revision deletion/Noticeboard, but for some reason it isn't used; why not? Robofish (talk) 21:42, 1 February 2011 (UTC)

Lack of consensus for its creation, and the fact that discussing RevDel in too much detail on-wiki would lead to a Streisand effect. Best approach is to message an admin in Category:Wikipedia administrators willing to handle RevisionDelete requests, either on their talk page or by email if privacy is an issue. We should probably edit the page to reflect this. Beeblebrox (talk) 22:23, 1 February 2011 (UTC)
 Done. [2] Beeblebrox (talk) 22:44, 1 February 2011 (UTC)

Censorship?

Isn't this actually censorship? We should always be able to see previous edits, regardless of how offensive they are. —Preceding unsigned comment added by 160.83.57.41 (talk) 15:14, 23 March 2011 (UTC)

There is a line here. Our purpose and product is a neutral encyclopedia.Our norms and policies are not in a theoretical vacuum of perfect behavior, they are intended to help us achieve that in the real world and inevitably involve judgments. The open nature of editing is not a suicide pact. Edits outside that scope can and do harm the project. So within fairly narrow areas where harm is very likely to outweigh benefit, we can and do remove them from public (but not administrator or accredited researcher) view. FT2 (Talk | email) 16:39, 23 March 2011 (UTC)
To put it another way, if you were in a library and you saw a librarian erasing racial slurs and threats of violence that someone wrote into the works of Shakespeare, would you then accuse them of censorship? If you saw a man washing graffiti off his house that said his daughter was a slut and listed her cell number would you accuse him of censorship? I think not. This is for removing edits designed to harm or accidental/naive revealing of personal information. That's it. If that is censorship then so is our entire deletion process, which covers a lot more territory than just revdel. Beeblebrox (talk) 01:05, 24 March 2011 (UTC)
The problem is that this tool is now being used to remove material that isn't offensive. Rather, it is being used to censor gossippy material that you'd see on TMZ or read in People. Particularly, it appears admininstrators in England are using the RevDelete tool to censor gossipy material. Although this material may not be appropriate for the main entry, it is wrong to censor the material using revdelete. In theory revdelete sounds find, but it is being used for censorship by certain celebrities (and/or their fans) so when you look at the history of certain pages a large percentage of the historical pages have been censored. Historical pages are important and the use of revdelete should only be used for those egregious situations where something offensive was included (i.e., the actual policies for revdelete seem fair on the page). In practice, it is being used for censorship of non-offensive material where there would be no harm to simply leaving the history in tact.CavalierLion (talk) 20:02, 10 May 2011 (UTC)
Ah, you see, we're an encyclopedia, so we leave it to other people to write important "historical" material. We are not a repository of internet memes and gossip - I think that was elsewhere. --Demiurge1000 (talk) 20:18, 10 May 2011 (UTC)
I do not disagree with removing gossipy material from the main encyclopedia entries. But, the historical revisions should be left in tact. They are useful. This is the reason specific criterion were established for using revdelete in only the most egregious situations. Your argument is more directed to whether the criterion for use of revdelete should be expanded to censor non-offensive, but gossipy material that is not appropriate for the main page. I don't think the criterion should be expanded and I could expand on this further (e.g., the entries in March 2010 on Rothleiberger come to mind as an example where gossip was posted, removed from the entry, but the historical entries not censored). But, that is not important at present. Rather, the problem is that is that right now the criterion are not being followed and it is used to censor gossip - something that is not listed as one of the criterion for use of revdelete. Wikipedia (and its first amendment principles) suffers if the criterion are not followed. CavalierLion (talk) 21:30, 10 May 2011 (UTC)
Agree with OP/CavalierLion. The First Amendment may not apply to Wikipedia, but academic integrity should. And collegial respect for contributors and their differences of opinion. And the general idea that this "tool" needs to be kept on the leash. Wnt (talk) 16:13, 28 May 2011 (UTC)

Harassment

RD#3 covers purely disruptive material of little or no relevance or merit to the project, including allegations, threats and attacks serving no other valid purpose.

Not all harassment involves overt threats and attacks. For example some real-world stalkers engage in inappropriate obsessive expressions of admiration. We had a case like this earlier. "Purely disruptive material" already covers allegations, threats, attacks and disparagement of no project value. I was surprised to notice it didn't explicitly include material that is pure harassment. I'd like to add this so that admins can act more surely in future cases.

Proposed edit: RD#3. ... This includes allegations, harassment, grossly inappropriate threats or attacks, ... that serve no other valid purpose.

FT2 (Talk | email) 19:31, 23 March 2011 (UTC)

I think this is already being done in practice and the policy should reflect that. Go for it. Beeblebrox (talk)
Done. FT2 (Talk | email) 06:05, 27 March 2011 (UTC)

Proposed changes

Arising out of the discussion at Wikipedia:Administrators'_noticeboard#RevisionDelete_policy, there are two proposals - one to add a criterion, and one to delete a criterion.

  1. Add, as RD7, admin discretion to use RevisionDelete to DENY recognition to certain banned users. Such a criterion was proposed in June 2009 (Wikipedia_talk:Revision_deletion/Archive_1#Possible_G7) and rejected, but since then there has been some use of RevisionDelete for this purpose. Since there seems some community support for this approach, it should really be recorded appropriately in the policy (or else firmly disallowed, since a founding concept of RevisionDelete was sticking closely to the policy).
  2. Remove RD5, Valid deletion under Deletion Policy, executed using RevisionDelete. This serves no clear purpose, since Deletion Policy aside from RevisionDelete does not address selective deletion for reasons not already covered by other RevisionDelete criteria.

Rd232 talk 23:54, 9 May 2011 (UTC)

  • Support both, with some caveats. I would like it to be explicit that revision deletion should not be used to remove material inserted by banned users unless the revision deletion can be performed without compromising licensing, and should not be considered a routine solution for edits by ban and block evasion. I find RD5 wholly confusing, and could be persuaded to keep it if someone could clarify it well enough that I can really understand it.—Kww(talk) 00:45, 10 May 2011 (UTC)
Attribution/licensing takes a high priority in any deletion case, I think that's reflected in the policy already. Regarding edits by banned users, different admins draw the line differently on this with CSD, and always have done. Without rehashing that debate, wherever the line is drawn, RevDelete should only be used on blocked or banned users' edits to the extent that the administrator would have used selective or traditional deletion in the past under CSD, because RevDelete is always less damaging to the public record than CSD if a deletion is going to take place (revision doesn't "vanish" to non-admins and some fields can optionally be left undeleted). FT2 (Talk | email) 02:40, 10 May 2011 (UTC)
However, it's being argued to removing edits by banned users doesn't fall under the umbrella of RD5 incorporating CSD G5 at all. A Stop At Willoughby argues both at WP:AN and below that it does not, and Risker concurs with him at AN. If you think that they are wrong, you should explicitly say so.—Kww(talk) 03:05, 10 May 2011 (UTC)
Risker has said it well, and we're basically saying the same thing. Nobody is advocating using deletion rather than RBI as the usual approach to banned user editing. The usual and common best handling for banned user edits is RBI. But a number of the more tendentious ones, especially if harassment is involved, do have their edits deleted or redacted instead. These views are not inconsistent:
  • Risker: If there is another reason to revision delete, such as a BLP violation or personal attack or other form of harassment, please do use revision deletion; however, if it is just routine and has no serious effect other than to add a wasted edit, please use the standard RBI response. Deletion should always be a last option, not the preferred one, even when it comes to edits of banned or blocked editors
  • Self: [W]herever the line is drawn, RevDelete should only be used on blocked or banned users' edits to the extent that the administrator would have used selective or traditional deletion in the past under CSD.
In other words, REVDEL is not an excuse to replace reversion by deletion. It is an extra and better tool to minimize the impact in the minority of cases where deletion would have been appropriate anyway. RD#5 saves having to rehash CSD into REVDEL policies by simply saying "if deletion of an entire revision would be appropriate under CSD then redaction under REVDEL is also appropriate" -- which makes sense as REVDEL is inherently less destructive of the public record than selective deletion and leaves a better trail for non-admins. Hence if (and only if) deletion is appropriate and will be used, then of the two tools administrators have for the task, REVDEL is almost always technically preferable. FT2 (Talk | email) 05:00, 10 May 2011 (UTC)
That's all very well, but one of the problems here for me is that I can't find any deletion policy that actually states clearly when selection deletion of revision by banned users is acceptable, and when not (regardless of method). CSD#G5 doesn't cut it - it's just a shortcut to deleting entire pages which would otherwise be live, and has no relevance to deleting selected revisions which would not (having been reverted). Rd232 talk 13:34, 10 May 2011 (UTC)
Then the problem is that REVDEL is reflecting-accurately-a gap in CSD#G5. If this is a problem in G5 as well (as you imply) then removing RD#5 isn't going to fix it, whereas fixing G5 will resolve both G5 and RD#5. REVDEL will still be a less damaging means of removal when it's appropriate, under any criterion. I really think you're looking in the wrong place for the problem. FT2 (Talk | email) 14:13, 10 May 2011 (UTC)
Once again: IMO WP:CSD#G5 is irrelevant here. Pages created by banned or blocked users in violation of their ban or block, and which have no substantial edits by others. just does not translate to reverted edits. Rd232 talk 15:00, 10 May 2011 (UTC)
FT2, what made you change your mind on the question of whether RD5 allows admins to RevDel banned users' edits? Presumably you didn't think it did in June 2009 when you proposed a separate criterion to allow for the deletion of such edits. A Stop at Willoughby (talk) 20:14, 15 May 2011 (UTC)
  • Oppose addition of new criterion RD7. Essentially, I find this exceptionally hard to square with WP:BAN#Edits by and on behalf of banned editors. My reasoning, adapted from what I've written at AN, is as follows: We can all agree that banned users are banned and are not supposed to make edits, period, even helpful edits. We can all agree that, according to the policy, any user can revert any edit made in defiance of a ban. We can further agree that, under this policy, editors who independently verify the validity/helpfulness of the edits, and are willing to take responsibility for them, may restore them. I think we can agree that not all banned editors try to insert misinformation into articles, and that some try to evade their ban by making helpful edits. Surely, then, it follows that editors who are interested in the article in question and who are not banned should be able to look at the banned users' edits and determine whether they were helpful. If the edits were helpful, the article is improved; if not, oh well. Now, given the large number of non-admin editors relative to the number of admins, the interested, non-banned editors I mentioned are more likely to be non-admins than admins. If a revert-block-ignore process had been followed, no problem arises. But if a revert-RevDel-block-ignore process had been used, non-admin editors are completely incapable of viewing and judging the edits in question. That is where problems arise. RBI should suffice when the edits are innocuous except for the block/ban evasion.

    Neutral on removal of RD5 at this time. I think we should discuss possible situations in which this criterion might be applied before having a quasi-poll on its removal. I will say, however, that I'm pretty familiar with the criteria for speedy deletion, and I've had a genuinely hard time thinking of situations in which one of those criteria could be applied to an edit but RD1-3 could not, and RevDel would be necessary. I will ask FT2, who originally added this criterion to the draft of REVDEL back in May 2009, to provide some insight. A Stop at Willoughby (talk) 01:30, 10 May 2011 (UTC)

  • Comment. I drafted RD#5. The purpose is to avoid needlessly duplicating CSD criteria on this page and keeping this page's criteria short. Quite a few speedy deletion criteria cover situations where RevDelete has less impact than traditional deletion to the public record, eg if only certain fields need removing. In those cases it's commonsense to remove from public view only the minimum needed and not "vanish" the whole revision. This matches the aim of not using traditional deletion + selective undelete when RevisionDelete can do the job. Reasons why RevDelete is preferable and why it matters are stated at #RevisionDelete compared to traditional selective deletion.
Possible examples:
  • G5 (revisions added by blocked or banned users) - it may be preferable in some cases to show a redacted revision than an incomplete history (eg remove their attempted edits but leave the attempt itself visible).
  • G9 (office actions) - uncommon these days but could well apply to a specific field in a specific revision rather than a whole page.
  • U3 (non-free galleries in userspace) - a user who adds a non-free gallery to their otherwise valid user page must have it removed from current and history. It is disruptive to the record to delete their entire userpage or to delete the whole page and undelete all but the offending revision, rather than remove just the inappropriate revision text.
RD#5 also means that edits to WP:CSD are automatically incorporated into WP:REVDEL if needed, reducing BUREAUCRACY. FT2 (Talk | email) 01:59, 10 May 2011 (UTC)
I don't see that CSD G5 (deleting pages) transfers to deleting revisions from page history; it's too tortuous, and that's why if it's going to be permitted it should be a new criterion. CSD#G9 - easily added as a separate criterion, like the AC one. CSD#U3 - should be already covered by RD1. Anything else? (PS As per my comment on G5, I'm not persuaded that relevance of other deletion policy is enough to justify RD5.) Rd232 talk 03:57, 10 May 2011 (UTC)
The main advantage is that CSD's that by their nature don't translate, aren't relevant (you couldn't apply REVDEL to things like G4, A2, A7, F10, C1 even if you tried). So they aren't an issue. The advantage therefore is in matters where selective deletion is applicable ("delete page then undelete most revisions"). When those kinds of CSD are modified, REVDEL automatically keeps exact parity of requirement. Since REVDEL is in every possible case less destructive than selective revision (it leaves a clearer non-admin trail and visible placeholder and some fields may not be removed from public view) it makes sense that if the more destructive CSD would be proper and appropriate in some situation involving specific revisions, then REVDEL would handle it better (more transparently/less destructively to the public record).
That said if consensus prefers individual RD criteria then there's no reason not to. It just seems to be pointless extra work as it implies 2 sets of changes being needed and any change to CSD may potentially need subsequent tracking at RD. If G5 is being inappropriately used then my feeling is that CSD#G5 needs tightening and better guidance, RD#5 isn't affected. The latter merely reflects that what is allowed and appropriate to do using the original deletion tool may also be done with this one. FT2 (Talk | email) 05:25, 10 May 2011 (UTC)
First, you keep talking about CSD, but RD5 refers to "Deletion Policy", which is significantly wider in scope. Second, as I said above, I don't see that there's very much duplication involved because the relevance of CSD is limited. Third, meshing the policies in this way is untransparent and may have unintended consequences, particularly if other deletion policies are changed. If they are to meshed in this way, RD5 should at least specifically refer to CSD criteria X,Y,Z. Rd232 talk 13:40, 10 May 2011 (UTC)
Agree with you - RD#5 should read "speedy deletion policy", it isn't meaningful to apply to xFD or PROD. But if it were then it would be preferable to selective deletion so this is detail. I would not be averse to it specifying individual CSD criteria - the list of CSD's (rather than their wording) doesn't change that often so tracking and parity would still work.
On the issue of principle: - apart from log redaction, in essence REVDEL does nothing that traditional delete (selective delete, CSD) was not also worded to cover. Every deletion REVDELETE can do is done more destructively by CSD or other traditional deletion tools. So if a deletion action on a specific revision is permitted and appropriate to be done under CSD or any other deletion process, then it is inherently less destructive (hence more desirable) to use REVDEL as the tool of choice. So all RD policy needs to say is that if a deletion can be done under CSD then it can be done using REVDEL instead. Any cases where REVDEL would not apply (the page itself should cease to exist) it can't be used for anyway. FT2 (Talk | email) 14:13, 10 May 2011 (UTC)
The very fact that people think CSD#G5 can be taken to justify revision deletion via RD5 (including, apparently, you) proves that the vagueness of the link is harmful. It should be removed. If RD7 is added (or rejected, which surely means CSDG5+R5 shouldn't then be allowed the same effect), it seems like the only other base to cover from removing RD5 is adding an "O. Office actions" type criterion, like the existing AC Arbcom one. Rd232 talk 22:51, 10 May 2011 (UTC)

Initial thought on proposed RD#7. All DENY targets are blocked and probably banned. Edits by blocked and banned users are already covered by CSD#G5 and hence at present by RD#5. So we already have power to use RevDelete to deal with their contributions equally to traditional deletion or selective deletion, including full removal from the public wiki. (This is why RD#5 is useful, it keeps parity and consistency between deletion tools). RD#7 doesn't add any new criterion or change to that. At most it splits out "CSD#G5 via RevDelete" from "other CSD via RevDelete", if that's useful for scrutiny purposes. Beyond that I can't presently see any real gain by adding it. FT2 (Talk | email) 02:18, 10 May 2011 (UTC)

  • Oppose adding RD7. Revdel should not be a routine tool for every single edit made by every banned user. It should not be used where it compromises licensing. That includes cases where the banned editor's edit was reverted and someone else reinstated it – the user reinstating it is taking responsibility for the content of the edit, but they don't own the copyright to it. Given that, I really struggle to come up with any situations where we should delete the edits, but RD1-3 don't apply. I agree with A Stop at Willoughby, RBI is sufficient when there's nothing actually disruptive in the edit and deleting actually has some disadvantages. Mr.Z-man 03:31, 10 May 2011 (UTC)
    • I agree. One of the reasons I oppose RD7 (and the unofficial use some others have endorsed, even in the absence of it) is that the lack of transparency is not merely a philosophical issue, or a question of accountability; or even of saving potentially good edits. It also makes it harder (especially for non-admins, but for admins too) to track the problematic activity being dealt with. Rd232 talk 03:57, 10 May 2011 (UTC)
See my comment on this above, agreeing that this does not change the preference for RBI on such edits. FT2 (Talk | email) 05:00, 10 May 2011 (UTC)
  • I'm not going to oppose or support blanket rules specifically, however I think we need to draw a distinction between making RevDel a regular tool used against every banned editor (which I think is going to far) and allowing a certain narrow usage in cases where a banned editor is gaming the system to use reverted (but still visible) edits to get around their ban. There have specifically been a series of breaching experiments used by a banned editor who attempted exactly such a loophole to use proxies to edit for him using old revisions still visible in the edit history. He would basically make an edit, self-revert it, then post the diff in an open forum asking for proxies to reinsert it uncritically. It would be useful (under a highly limited WP:IAR-type criteria) to allow for admins to be able to use RevDel in ways to respond to such creative attempts at circumventing a ban, without going "all the way" in allowing RevDel to be used willy-nilly against every edit from every banned editor. I understand the need for judiscious use of RevDel, however there needs to be the ability for admins to use this tool under circumstances which may arise, but have not been forseen. I'm not sure that there's any need to ammend the current policy, but just to not freak out every time some novel disruption requires its use in ways that were not forseen by policy. --Jayron32 05:43, 10 May 2011 (UTC)
    • I'm happy to permit RevDel in the exceptional circumstances you describe (offwiki pointers to specific revisions), though it does seem to need IAR or RD5 to permit it. The issue is frequently using RevDel on banned user contributions without such exceptional circumstances. Rd232 talk 13:28, 10 May 2011 (UTC)
      • That's understating things a bit. Hopefully people would consider sockpuppeteers that have continued to edit for long periods of time after being banned, despite long-term campaigns of edit reversion and semi-protection, to be "exceptional".—Kww(talk) 13:47, 10 May 2011 (UTC)
  • Oppose. This RD7 is not correct, it's not made to DENY recognition, it's made to prevent disruption via old diffs. Removing RD5 would force admins to use the "deletion and selective undeletion" technique to achieve the same result, only with much more effort and much less transparency And, no, it's not being used as a blanket tool against all edits by a banned user, and nobody wants to use it in that way. --Enric Naval (talk) 07:38, 10 May 2011 (UTC)
    • "it's not made to DENY recognition, it's made to prevent disruption via old diffs." - OK, RD7 could be worded something like "where leaving old page revisions in place significantly facilitates exceptional disruption by banned users", that sort of idea. Rd232 talk 13:30, 10 May 2011 (UTC)
We might manage a few of these terms together ("in exceptional circumstances" or "facilitating exceptional disruption"), but a string like this is a recipe for disagreement on what it means. FT2 (Talk | email) 14:13, 10 May 2011 (UTC)
That's still better than not having it in the policy at all. At least if it's explicit, there's some handle on the relevant practice; without, it just flies under the radar. Rd232 talk 14:58, 10 May 2011 (UTC)
"significantly facilitates exceptional disruption by banned users". I also fear a lot of wikilawyering over "exceptional". Is it possible to remove "exceptional"? Or maybe add a couple of examples: where the banned user make off-wiki links to the old page revision to show his work, or uses that revision to make reverts, or asks other people to revert back to that reversion (not sure if these examples are 100% accurate). --Enric Naval (talk) 17:00, 10 May 2011 (UTC)
That's the sort of thing, which could be given in the RD examples. But (see below) I think some element of prior consensus can come in here; this should not be so frequent or so urgent as to require individual admins to decide that RD7 applies to a socking case (though once that's collaboratively decided, obviously individual admins can act on that). Rd232 talk 17:10, 10 May 2011 (UTC)
  • Oppose RD7 proposal as simply not needed. We revert banned editors (this is long standing practice) but there is no rational need to delete those revisions, unless the content or edit summary is disruptive or otherwise needs deleting. The specific issue identified by Jayron above is already covered in policy under deletion of purely disruptive edits. If this is the case with a banned user it should be identified and entered into the LTA log so that we can track and properly delete those attempts. --Errant (chat!) 14:49, 10 May 2011 (UTC)
  • Comment - clarifying somewhat how RD7 might actually work: use of RevDel to essentially deter socking really ought not be a decision made by individual admins. Which is to say, once a collective decision is made that a particular socker justifies it, then any admin can do it, citing RD7+SPI, but there needs to be an initial discussion that it's justified. Discussion would be at the relevant SPI. Rd232 talk 15:59, 10 May 2011 (UTC)
  • Oppose RD7 We need content, and content has to be judged as content. It's always been acceptable to reuse content from banned editors if a responsible editor is prepared to check it and take responsibility for it. This would essentially eliminate that possible. Much has been made about denying recognition. I think deletion is generally sufficient to do that, and in the exceptional cases that it is not, then this will not be either. The real problem we have with banned editors is very different--it's the problem that if they are reasonably clever, we won't find them. Nobody has a solution to that one, RevDel is a dangerous tool, and it's use as a device to control editors rather than remove harmful content is overkill. But I can not rule out the possibility that there might be special cases. At the very most, I'd accept it only for individual banned users with unambiguously clear prior consent at an/i, based on their known to be continuing to edit despite all other measures after being banned, and doing substantial harm by the editing apart from just being banned, and including agreement by one or more oversighters-- but certainly not at individual admin discretion. There's too wide a range of people's opinion among the 700 of us. I doubt whether the removals being done fall into this class, since the admin making them was not able to assert they were erroneous or harmful--just that it was a question of the individual being banned. In other words, even if we accepted this in principle, this was an incorrect use of admin discretion to do it. DGG ( talk ) 19:44, 10 May 2011 (UTC)
  • It's always been acceptable for people to reuse banned editors content, but it's never really been encouraged, either. Do you really want to encourage a review of a banned user's edits in order to incorporate the good bits on the grounds of "we need content"? Encouraging detailed review of a banned user's edits would actually encourage socking, because the banned editor would be getting more attention paid to his edits than a normal editor. The users I've applied this against have been socking for years. Not weeks, not months, years. I'm open to alternate solutions, but no one has suggested one yet.—Kww(talk) 20:42, 10 May 2011 (UTC)
  • Oppose There is little benefit in this and revision deletion is more likely to cause questions about the article history. For very disruptive edits there will already be reasons to remove. Our mission here is not to attack banned people but to build an encyclopedia. The other issue is adding more rules and beurocracy where it makes it harder to read and understand, outwaying any benefit. Graeme Bartlett (talk) 05:08, 11 May 2011 (UTC)
  • Oppose, ErrantX and Graeme Bartlett have summed up most of what I was going to say. Too much is being revdel'd at present; the proposed changes encourage revdel being used to censor. Have you all forgotten the feeling of being a new user .. able to learn the history of a page, the only limit being the time invested to follow the links. The more we hide, the less open we are. John Vandenberg (chat) 00:18, 13 May 2011 (UTC)
  • Oppose I see no reason to hand out the censor's black pens --Guerillero | My Talk 05:12, 14 May 2011 (UTC)
  • Oppose RD7, but OK to remove RD5. I've been in an interminable debate at the Refdesk talk page (which even briefly attempted to spill into ANI) about a supposed sock puppet taking up 1/16384th of the total IP address range. Some people there think they can recognize him, but at least as a group they don't seem to stick to their own criteria. Meanwhile, the puppet (?) claims he is somebody else. I think that trying to spot sockpuppets in an encyclopedia anyone can edit is a quixotic effort that will always be prone to debate, and hiding the diffs rudely puts members of the community out of the debate. What practical benefit could RD7 conceivably offer to make up for that? Wnt (talk) 16:18, 28 May 2011 (UTC)
  • Oppose RD7 in both forms. By my reading, the existing revdel policy makes an implicit distinction: edits that harm others can be revdeleted. Copyvios harm the copyright holder, outing harms the outee, threats harm the threatened, and virus/malicious pages/etc. cause "software damage," for lack of a better term. The exception is RD5 which is, of course, also under debate right now. I'm aware that revdel is occasionally used today for RBI purposes but I think that's a deviation from policy that should stop, rather than a valid argument for changing the policy. Orange Suede Sofa (talk) 18:16, 29 May 2011 (UTC)

RD7 proposed text

Maybe it would help to make the proposal more specific, taking into account various comments:

7. Combating exceptional disruption. Where leaving old page revisions in place significantly facilitates exceptional disruption by banned users, such revisions may be deleted, if there is prior consensus established at an appropriate venue that the disruption is sufficiently exceptional to require it. Disruptive edit summaries or usernames from banned users may be removed without prior consensus.

Thoughts? Rd232 talk 22:36, 10 May 2011 (UTC)

Change it to "exceptional or long-term" or something similar, and I'll be happy. I wouldn't want the discussion to center on whether someone that had been socking for a year had been "exceptionally" disruptive, "excessively" disruptive, "excruciatingly" disruptive, or any other particular adjective. There's a time when you say "enough's enough". Alternatively, change the last sentence to "Disruptive edit summaries or usernames from banned users, or edits by banned users that persist over a year from the date of their final block, may be removed without prior consensus."—Kww(talk) 23:19, 10 May 2011 (UTC)
I would have thought we could allow "exceptional" to be defined through practice; length of disruption obviously features in judging that, I'm not sure it needs highlighting. And I definitely wouldn't add a specific time period or individual admin discretion for deleting revisions. Edit summaries and usernames are less of an issue to delete, and have more intrinsic potential for disruption because they appear in the page history. Rd232 talk 23:48, 10 May 2011 (UTC)
  • Without agreeing or disagreeing yet, it's worth focusing on the aim of any proposal. We already handle "purely disruptive material", so this presumably covers situations where the material per se is not "purely disruptive" and the focus is on the user rather than the material. I can think of 4 cases: - 1/ those few users sometimes described as "remarkably undesirable", where 2/ RBI is insufficient because the history version is being abused (eg by linkage), where 3/ we want to completely WP:DENY or deter a user from continuing to attempt to edit by removing all traces of their activity, or 4/ material by users where consensus agrees their overall purpose is harassment or disruption even if some edits are cleverly crafted to "skim under the barrier".
Drafting with that in mind:
7. Subject to prior consensus at an appropriate venue, revisions may be redacted to deny impact and recognition to banned users and non-users having an established pattern of disruptive, harassing or otherwise problematic activity, and to mitigate the effects of revisions which have been reverted but whose history link is being used externally in an inappropriate or harmful manner. This is intended for exceptional or persistent cases; in simple cases reversion is preferable. Purely disruptive or offensive edit summaries, usernames and log entries from banned users may be addressed under other criteria without prior consensus.
Is any of this redundant as covered by RD#2 or #3? Is it ever urgent enough that subsequent consensus (act then ask for eyeballs) would be needed? Are there any other cases that we may want to delete a revision rather than revert it? FT2 (Talk | email) 09:41, 11 May 2011 (UTC)
  • I'll bring over some text from AN, because the real impetus for this is the actions that I have taken against a handful of sockpuppeteers. In the case of Wiki-11233, I've been reverting each and every one of his edits since July, 2009. I've been semi-protecting his targets since May 15, 2010. I started revdeleting his edits Dec 11, 2010. With Brexx, I've been reverting every one of his edits since Jan 22, 2009, and began semi-protecting his targets in Aug, 2010. I started revdeleting his edits in Dec, 2010. For Wiki-11233, that was 8 months of RBI and 7 months of RBI+semi-protect. For Brexx, that was 16 months of RBI and and 4 months of RBI+semi-protection. Neither one stopped editing in the face of that, which led me to the conclusion that RBI simply wasn't working in those cases. Brexx problems dropped precipitously after adding revdelete. I haven't seen Wiki-11233 in a month. We aren't talkin a process that I started after a few weeks: I don't think anyone could look at those time periods and think that RBI was working. As for why I chose revdelete — I couldn't think of anything else to try. I don't mind fancy wording, as long as the wording recognizes that it's legitimate in cases like these two.—Kww(talk) 11:56, 11 May 2011 (UTC)
Agree, some (ex)users just dont stop and can take years to get the hint if ever. Making admins jump through hoops to deal with these users (who have no intention of reasonable editing) is pointless, equally I also agree with those who express concern, and we don't want to encourage a general migration from RBI to deletion for banned users other than case by case. I guess one question is, as someone who has dealt with these cases, do you feel that requiring prior consensus is too much of a burden, or fine? FT2 (Talk | email) 13:33, 11 May 2011 (UTC)
I worry that we have admins that are opposed to dealing strongly with socks on the simple basis of them being socks. Comments from DGG and Risker both show a very strong interest in analyzing the content of the banned user's edits, as opposed to simply evaluating the question of whether the banned user is continuing to edit. That can make it difficult to gain consensus. Brexx and Wiki-11233, who have become the focus of this discussion, are simply bad editors: sloppy about sourcing, sloppy about quality of the contributions, and unwilling to discuss the problems with their edits. That lead to indefinite blocks in both cases (and, in Wiki-11233's case, after I got him unblocked and attempted to mentor him). Probably 80% of their edits are OK ... it's the other 20% that lead to their bans. I can see having discussions about whether banning editors that are in the 80% range is the right thing to do, but I don't want the guideline written so that it encourages discussion about whether some banned editors are really banned, and others are just kind-of-banned-because-their-edits-aren't-100%-crap. In either case, the editing has to come to an end.—Kww(talk) 14:13, 11 May 2011 (UTC)
So what for you (personally) is the point where reverting isn't enough and redaction is needed? How would you suggest the line can be described? FT2 (Talk | email) 14:59, 11 May 2011 (UTC)
Timeframe, primarily, which is why I suggested a bright-line duration. A year of standard RBI is enough to demonstrate that normal techniques just aren't working.—Kww(talk) 15:04, 11 May 2011 (UTC)
A year will wear many good-faith editors out. Often they will have tried to work with a POV warrior or hostile user for 2-6 months, then dispute resolution (and having to rebut any tendentious claims made against them), then admin involvement, perhaps RFC/RFAR or a 2nd chance and another few months. So a ban can easily be after more than a year or 18 months of stress. Then when they hope they are finally rid of him, a bunch of RBI... I would say that for a banned user who is still undertaking significant activity, a year of that may be the straw that breaks editors' backs. A banned or ex-user who is given a "reasonable" chance to understand they are unwelcome, and whose actions continue to be disruptive, could switch from RBI to denial "once it becomes clear that RBI has been given a good chance" (and is not having a discouraging effect). Perhaps this captures the sentiment: "RevisionDelete should only be used after reasonable attempts to use RBI to deter the user have visibly failed"? FT2 (Talk | email) 16:01, 11 May 2011 (UTC)
I could go for wording like that. I agree that a year is a long time, but it unfortunately isn't rare at all. The only thing that is actually rare is for the RBI effort to actually continue that long.—Kww(talk) 16:58, 11 May 2011 (UTC)
I see no benefit whatsoever in mentioning a time frame in the policy. The issue is not time, it's disruption + failed attempts to combat it by the usual means. Sufficient disruption might occur in a couple of months in some cases; or in others it might continue at a low, occasional level for years, not being enough to justify RevDel. Each case needs examining on the merits, and practice allowed to evolve based on experience. Rd232 talk 00:55, 12 May 2011 (UTC)
  • Not sure if I should be commenting here or above. Support the general idea of usinfg RevDel as part of the RBI arsenal, oppose this proposed wording or any version that requires we talk about it and/or places a a specific time frame on it. Such ideas rather miss the point of denying recognition to trolls and banned users. They love generating long contentious discussions about their editing, and gaming the system. Why wait at all? Banned users are not permitted to be editing here under any circumstances. Nobody gets banned without significant discussion of why their editing is no longer welcome. Consensus that their edits are not welcome is already in place as of the moment they are banned. Quietly removing all trace of their edits is a better than just reverting them but leaving them viewable in the page history. Any edit they make after being banned is disruptive by definition and we should be able to remove it as "purely disruptive material." We have this awesome new tool that lets us zap vandalism off the face of Wikipedia, arguments that we should keep it to preserve the edit history of edits we manifestly do not want are nonsense. Suggesting we wait them out for a year before taking this extra step will just encourage these trolls, removing all trace of their edits on sight will discourage them. Beeblebrox (talk) 16:53, 11 May 2011 (UTC)
  • Philosophically agree, but disagree on pragmatic grounds. Revdelete is tedious and error-prone. Anything that encourages its routine use is probably a bad thing. This may be reasonable for formally banned users, but for the de facto case of "indefinitely blocked, and no one would ever unblock him" this would be overkill.—Kww(talk) 16:58, 11 May 2011 (UTC)
Use of the tool absolutely should not be routine for this purpose. A quick discussion at the relevant SPI, or at AN/ANI, is not a big deal for agreeing "OK, it's bad enough to use RD7". The sort of cases we're talking about for use, "recognition" isn't really an issue, since such users will be well up on efforts to combat them, and it won't make any difference to have some of the effort involve an RD7 discussion. (Recalling that it only needs to happen once for each case - once it's agreed, it can be applied by any admin for that case.) As for declaring harmless edits by banned users "purely disruptive material":NO, RD3 CANNOT cover that case. That it's occasionally used that way in practice merely strengthens the case for having RD7. Rd232 talk 00:52, 12 May 2011 (UTC)
  • Comment: it occurs to me that the purpose we're discussing here (deterring banned users) might be one of the few where the old selective deletion approach might be more effective than RevDel. Removing contributions entirely from the page history would, I think, have a greater deterrent effect than just putting a line through them, as it were. I'm not sure if that conclusion should mean we consider not using RevDel but selective deletion instead (authorised by a similar procedure as suggested above)? Rd232 talk 00:59, 12 May 2011 (UTC)
Selective deletion has a big problem - it's very easy indeed to undelete revisions by accident - you have to mark alll revisions of the page as deleted, then manually select all those to be undeleted. Too much risk for pages that already have significant selective deletion activity. Try this wording (v2), which attempts to cover the points made:
7 Exceptionally persistent abusers. RevisionDelete should not normally be needed to remove revision content by users whose edits are unwelcome, since revert, block, ignore is usually preferable, and any disruptive or offensive material can be summarily removed under RD#2 and RD#3. In exceptional cases, using RevisionDelete as an means of denying recognition may appropriately target a user with an established pattern of disruptive, harassing, persistent, or otherwise problematic activity, or gaming of the "revert block ignore" system (eg by externally linkage of the "reverted and ignored" revision in an inappropriate or harmful manner), but all of these conditions must be met: 1) If the responsible party is known to be a previous editor, then they must have been formally banned by community discussion or Arbcom; 2) reasonable attempts to counter their conduct using ordinary remedies were tried but failed to deter the activity - for ex-users this is subsequent to their ban; 3) These included reasonable good faith attempts to "revert, block, ignore" which have not deterred the conduct.
FT2 (Talk | email) 02:30, 12 May 2011 (UTC) (Updated 02:42, 12 May 2011 (UTC))
I'd have that as an excellent clarificatory footnote to my proposed text. Rd232 talk 02:35, 12 May 2011 (UTC)
As an alternative to introducing a new criterion, I would support FT2’s language above as a footnote to RD3. I disagree with the opinion that we need IAR or RD5 to permit the kind of RevDels that Kww has performed with Wiki11233 and Brexx. Their continued editing, after almost a year of RBI, has become “purely disruptive” and, thus, already covered. — Satori Son 15:17, 16 May 2011 (UTC)
  • Support a change to RevDel with FT2's language, or something similar. I think that Kww made a good case for some extreme situations where RevDel may be an effective way to enforce a ban. But it really must be stressed that this method should only be used in extreme situations. As others have pointed out, deleting revisions from banned editors could take away the long-standing policy that a banned editor's edits can be accepted if another editor in good standing independently supports the edits. I never want a case where we turn routine RBI into "RevDel-Block-Ignore". But allowing RevDel in the worst cases can curtail attempts of banned editors to game the system. -- Atama頭 18:06, 12 May 2011 (UTC)

FT2's text looks reasonable. Appears to be tight enough to avoid wikilawyering, and at the same time allowing admins to use RevDel for the most egregious cases of "look-I-evaded-my-ban-again-haha-look-at-them-run-in-circles-as-if-I-had-stomped-upon-an-anthill-Working-exactly-as-planned". --Enric Naval (talk) 11:16, 13 May 2011 (UTC)

  • This proposed text is an improvement because it is very narrow. However, it still doesn't address the fundamental problem with the idea of using RevDel to deter banned users. It still could potentially lead to helpful edits being removed from article histories, preventing non-admins editors who are interested in the article in question and who are not banned from looking at the banned users' edits and determining whether they were helpful. We should prioritize improvement of content above ban enforcement. I'm uncomfortable with RevDel being used to hide content that is not otherwise harmful. If you want to use RevDel to remove the edits of banned users who have persistently inserted misinformation or vandalized, that would be another thing entirely. A Stop at Willoughby (talk) 20:05, 15 May 2011 (UTC)
    • I once proposed that we provide a way for banned users to actually channel their energies into trying to contribute constructively, in a very limited way (possibly as part of a potential WP:STANDARDOFFER way back). For example, it could be done using a variation of {{Request edit}} which banned users can place on their talk pages (with a special category to distinguish their requests). This is at least as worth trying as RevDeletion of harmless edits. Rd232 talk 23:03, 15 May 2011 (UTC)
Banned users are precisely those where we have already tried to "channel their energies into trying to contribute constructively". As a rule, banned users have had restrictions, warnings, blocks, discussions, chances, clearly explained paths back to good editing, etc before or during a final ban discussion. At the point of community ban the community has already decided to give up on them, at least for a long period of time. That's what the ban discussion is. At the point of formal ban, we just want to forcibly disengage them and have them go away. If we had wanted to give them last-last-last-last chances under "very limited" permissions we would do that before a ban. Anyone could propose a restriction as you describe ("User will be restricted to posting proposed edits on their talk page only, provided they do not abuse it, for <period of time>"). But that isn't post-ban. It's pre-ban. Post-ban the aim is disengagement. See WP:BAN#Bans apply to all editing, good or bad. FT2 (Talk | email) 13:07, 16 May 2011 (UTC)
Well that's gigantically missing the point, isn't it? Yes, banned users are supposed to go away (and should have had plenty of chances pre-ban). But we're talking about those banned users who don't go away! For those, a means to channel their energies might be a sensible means of harm reduction (parallels with drugs policies here). Rd232 talk 13:35, 16 May 2011 (UTC)
It's far from missing the point, although this could be more a difference of philosophy than policy. I am against rewarding ultra-persistent banned users by establishing as a result of their persistent endeavors to ignore a ban, a "limited return". If there is any kind of limited right to edit then it comes before they are banned. After that, the priority is our right to manage our website with their absence per consensus. If we disagree on this (which is also longstanding policy) then the better venue is WT:BAN, it's a bit tangential here. Even the original post describes this as an idea which was "once proposed". If it's to go anywhere then WT:BAN is the place to propose it. FT2 (Talk | email) 14:39, 16 May 2011 (UTC)
Honestly, I don't really see this as particularly relevant to the point I've been making. I'm not contesting the banning policy; I'm just saying that any non-banned editor should be able to look at a banned users' edit and decide to restore it if it improved content. That's what the policy says already. A Stop at Willoughby (talk) 19:29, 19 May 2011 (UTC)
I oppose this despite the rewording. "Exceptional disruption" will mean many things to many people. And what on Earth is the point? What is more off-putting to editors, a bunch of old vandalism in the history, or a whole bunch of edits with lines drawn through them due to some kind of mystery censorship? Besides, it denies what actually happened - that the editors working on the article at the time had to deal with all that garbage. At worst you could end up with anomalies where it looks like a productive editor made vandal edits. My gut feeling is that people wouldn't try to RevDel edits by banned users unless they made some valid argument or gave some meaningful information that they were afraid other editors might agree with - and that's not how Wikipedia is supposed to work! Wnt (talk) 16:24, 28 May 2011 (UTC)
From experience of editors who have suffered such persistent attention, which is often deeply distressing, as well as the number of banned users who post them repeatedly to gain attention, I'm pretty certain that in these kinds of cases we should prioritize the stress levels of harassed good-faith users very highly. We have always done so, and should continue. The criterion proposed is very narrow since it requires the source of these posts to be a non-user or explicitly banned, prior attempts at prevention and usual attempts to ignore to have been tried, and all having been of little avail. In those cases we routinely do redact to deter posting and it is clearly seen as beneficial by most - not least by its targets. The traditional concepts of denial and avoiding allowing shrines to vandalism and abuse, also come to mind. The aim is been to narrow the criteria so it is usable on the few occasions it's appropriate. In most cases as described, RBI and removal of disruptive material does the job. FT2 (Talk | email) 11:31, 30 May 2011 (UTC)
Wnt, that's one of the reasons that this should be done only if it can be done early enough to demonstrate that there has been no impact on the article. My own uses of RevDel in this fashion typically have no follow-up edits, and sometimes a handful of unrelated edits. If anything like this ever gains consensus, I would strongly oppose using it if it creates licensing problems by RevDeleting a contribution but retaining text from it.—Kww(talk) 13:41, 30 May 2011 (UTC)

Proposed removal of RD5

I'm splitting this off so discussion about the potential removal of RD5 is separate from discussion about the potential addition of RD7. Earlier in the discussion, I asked FT2 (who drafted RD5) to provide some insight into how the criterion can be applied in practice. He said that it could be used to delete galleries of non-free images from userpage histories (but RD1 could already be applied for that purpose), to delete banned users' edits (this is controversial and is the subject of much of the above discussion), and for office actions (which might be better off as a separate criterion anyway). At this point, therefore, I would support removal of RD5, although I am still open to having my mind changed if good reasons for retaining this criterion are provided. A Stop at Willoughby (talk) 20:27, 15 May 2011 (UTC)

The actual reason I stated was:
"To avoid needlessly duplicating CSD criteria on this page and keeping this page's criteria short. Quite a few speedy deletion criteria cover situations where RevDelete has less impact than traditional deletion to the public record, eg if only certain fields need removing. In those cases it's commonsense to remove from public view only the minimum needed and not "vanish" the whole revision. This matches the aim of not using traditional deletion + selective undelete when RevisionDelete can do the job. Reasons why RevDelete is preferable and why it matters are stated at #RevisionDelete compared to traditional selective deletion."
"RD#5 also means that edits to WP:CSD are automatically incorporated into WP:REVDEL if needed, reducing BUREAUCRACY."
The items you are quoting were examples and not reasons, taken from current speedy delete criteria. The value of RD#5 is that any CSD where selective deletion (of one or more revisions) would make sense, REVDEL will always without exception be less destructive to the public record -- and hence REVDEL will always be preferable if a revision is to be deleted under any CSD criterion. RD#5 simply says that any new speedy deletion criteria or changes to existing speedy deletion criteria will allow REVDEL to be used instead of the more destructive CSD, and REVDEL policy automatically updates if CSD is updated. If a matter should not have REVDEL used then so much more it should not be allowed under CSD, and WP:CSD would then be the place to seek consensus. FT2 (Talk | email) 22:35, 15 May 2011 (UTC)
"any CSD where selective deletion (of one or more revisions) would make sense" - CSD is about the deletion of whole pages, so I do not see that it has any relevance here at all (and legitimate possible parallels are largely covered by existing criteria, in a stricter form - so let's not make a mockery of that strictness). Nor do I see the attraction of effectively using a RevDel criterion to transclude every other deletion policy in a way that clearly is at times being interpreted to go far beyond the bounds of both RevDel and the transcluded policy. (Specifically in the banned users case.) Basically, this is an extremely bad way to write policy, especially when the policy is supposed to be really really strict, as RevDel is. It effectively produces a situation where it's said "Warning! Only use RevDel strictly according to the RevDel criteria, or you might get desysopped! PS RD5 allows you to do whatever the hell you like." The argument about passing through updates and changes to policy doesn't wash at all - any relevant changes are likely to be infrequent, and should be examined for their applicability here; failing to do so may have unforeseen consequences. Finally, the "RevDel is better than selection deletion" argument is all very well, but since no other deletion policy (AFAIK) addresses selection deletion, that's irrelevant in regards to the usefulness of RD5. Rd232 talk 22:59, 15 May 2011 (UTC)
CSD also covers all summary deletion using traditional delete, although it doesn't say so explicitly no other policy exists saying when individual revisions can be deleted and so CSD was (and still is) used as guidance on it. I think we're slightly talking at cross-purposes here. Let me try and set out the logic step by step:
  1. Traditional deletion (delete all + selectively undelete) was for years the way that individual revisions were deleted, and still remains available for the purpose.
  2. There is no specific policy on its use this way; since selective deletion always involves full delete and selective undelete, it has always been covered by the criteria in WP:CSD.
  3. Therefore whether or not individual admins agree with the idea, or whether or not it's still the best way to do it, at this point in Wikipedia CSD is still the governing policy for selective deletion using traditional delete.
  4. Not all CSD criteria make sense applied to individual revisions but some do, and others may in future. So unless policy is explicitly changed to forbid it, we can see delete + selective undelete may probably still get used (as in the past) on specific revisions where it is applicable and permitted by CSD.
  5. In every last possible case that selective deletion could be used (other than page merge fixing), RevDelete is better from a damage and transparency viewpoint. For instance, it always leaves a public placeholder (traditional selective delete just "vanishes" the revision); it is less prone to error (you don't have to delete all and hope you undelete the right ones), and it can be field selective (certain fields only where traditional delete just deletes all of them).
  6. So in every case that traditional selective deletion could be used, RevDelete is better - it's less damaging to the public record and more transparent in every case.
  7. Insofar as CSD might be cited to delete a specific revision at present, RevisionDelete is less damaging and therefore more desirable as a tool.
  8. Insofar as CSD permits deletion of specific revisions on stated grounds, it is always more desirable (to reduce public record loss) that those policy grounds are covered by and the deletion effected using RevisionDelete.
Your argument says that CSD doesn't cover selective deletion (I disagree, for years it's been the tacit governing policy giving valid reasons for summary use of the deletion tool - is anyone claiming that all use of selective deletion was done under IAR?). You argue that it goes "far beyond the grounds of both RevDelete and CSD", which is a nonsense given that it permits only what is already permitted using past deletion tools. You claim it allows you to do "whatever the hell you want" although anyone reading it can see it allows only what CSD would equally allow.
So with respect, I don't agree with your non-argument. Perhaps what you mean is that you want traditional selective deletion to also have limits. If so, add them, but that's not a RevDelete problem really.
The other option I've favored in the past is a specific policy on "selective deletion" covering both traditional selective deletion and RevisionDelete (or else having CSD explicitly state that the criteria for use of traditional selective deletion are the same as those which apply to RevisionDelete). It doesn't make sense to have different criteria for essentially one wiki function. FT2 (Talk | email) 03:22, 16 May 2011 (UTC)
CSD doesn't mention selective deletion. If you say it was used as guidance for it in the past, well I'll believe you, though I've been an admin for 5 years and it's news to me [either that or my memory's playing tricks on me :P]. But this is in the past - nobody disputes that almost without exception, RevDel is better than selective undeletion. In other words, RevDel as a tool supersedes selective undeletion, which should not be used except where it is demonstrably better. Equally and concurrently, RevisionDeletion policy supersedes any tacit use of CSD for selective undeletion, and RD5 is therefore a useless, confusing, and potentially dangerous [as explained above] hangover. If the previous tacit CSD use for now no-longer-used selective undeletion requires some explicit repeal, then so be it: we just need to figure out how to do it with least fuss. Perhaps a guideline on selective deletion which essentially says "use RevDel (per RevDel policy) unless it's a history merge". Rd232 talk 04:18, 16 May 2011 (UTC)
I think we're now on about the same page here and saying the same things: - RevDelete improves selective deletion (nobody disputes this) and therefore the issue is what we do about selective deletion. Probably retire ("repeal") it with the sole exception of page merge fixes, and formalizing that the sole policy for deletion or redaction of individual revisions by administrators is via RevDelete. If that's done then CSD will no longer govern selective deletion and RD#5 will becomes redundant; the only question is whether there are any other provisions in CSD or otherwise, or times that redaction/deletion of a revision is desirable, to add to this policy.
If that fits with how you see it, then let's redraft to that effect and get some initial views here, and then RFC it properly, as it will be formalizing a significant change to the deletion tools. Below is a mock-up of how I'd approach it. FT2 (Talk | email) 05:53, 16 May 2011 (UTC)
Proposed cessation of selective delete (other than history merge fixes) and resulting changes to deletion policies

Following discussion at Wikipedia talk:Revision deletion, the following points came out which this proposal is intended to address:

  1. Before RevisionDelete the sole means of summarily deleting individual revisions was by deleting the whole page and undeleting valid revisions ("selective (un)deletion"). No specific policy existed for this. The use of selective deletion was governed by a mixture of IAR and CSD, the latter since selective deletion is based upon deleting the page (CSD) then undeleting selected revisions.
  2. RevisionDelete was introduced and its policy created in 2009. It is more transparent and less harmful to the public record in every situation and is more easily reversed. On the same day that WP:REVDEL became a policy, it was quickly updated to note that selective deletion was now "deprecated" and should not generally be used. This is still policy today. (See intro)
  3. To ensure migration of all tasks (present and future) from selective deletion to RevDelete was smooth, a criterion was added, that any selective deletion allowed under CSD could also be executed using RevDelete.
  4. It is the sense of many users now that we're past that point. Selective deletion has no real uses beyond fixing history merges and since CSD does not explicitly mention selective deletion, RD#5 can be vague in terms of what is allowed and desirable. It is time we cleaned up this situation.

The suggested clean-up is as follows and the community is asked to endorse this since it would be a significant change to deletion tool policy (though not their usage).

  1. Relevant deletion policies (CSD, RevDelete, DelPol) all to be explicit that selective deletion is deprecated, with the sole exception of fixing page merge issues, and that the only acceptable methods for administrators to remove improper content are CSD (for deletion of a page and all its history) or RevisionDelete (for deletion/redaction of specific revisions).
  2. Addition of suitable RevDelete criteria for any "extra uses" anticipated by RevDelete policy or previously handled under RD#5, CSD or selective deletion.
  3. Removal of then-redundant criterion RD#5.

The community is asked to consider any additional RevDelete criteria (a start has been made with a proposed RD#7 to replace the imported CSD#G5) and then to approve this change to policy.

FT2 (Talk | email) 06:25, 16 May 2011 (UTC)
Excellent summary. Let's do that. Rd232 talk 14:05, 16 May 2011 (UTC)

Of the CSD criteria I found only 3 that might need importing: - edits by persistent banned users (G5) is already under discussion as proposed RD#7, Office actions (G9) might need adding, and non-free galleries can be noted within RD#1 as the issue there is indeed, blatant copyright violation (as NF images may only be used within mainspace without breaching copyright policy). So..... are there other situations where RevDelete is or might validly be used (or are analogous to CSD) that could be worth adding? Are there any changes or additions needed related to log redaction? And given its tight limits of expectation and review, should we clarify that IAR applies to RevDelete as it does to all other actions, or is that self-evident, insofar as the policy was drafted tighter and with more warnings than almost any other? FT2 (Talk | email) 15:27, 16 May 2011 (UTC)

Also worth noting, misuse of links to history revisions as a basis for deletion is endorsed in principle by Arbcom at this 2007 case (bullet #5), in userspace at least. FT2 (Talk | email) 21:14, 16 May 2011 (UTC)

I agree that we're now on the same page on this particular issue. Let's hold a RfC on selective deletion, as proposed by FT2 above. A Stop at Willoughby (talk) 19:23, 19 May 2011 (UTC)
In which case, unless anyone says there's a problem, I'll de-RFC this section (as having done its job and got good discussion) and set up a new RFC as discussed, tomorrow or Saturday.
One question - since we need to solicit, consider and gain ratification for possible new criteria, is its better to do it in one stage (propose and solicit new criteria and ratify them, and ratify the severance of CSD/removal of RD#5, all at the same time) or two stages (solicit/agree new RD's first, with severance/RD#5 removal as almost a pro-forma procedural matter second)? May be simpler in 2 stages, especially as once new RD's are done, the rest is likely to be quite mundane. FT2 (Talk | email) 22:40, 19 May 2011 (UTC)
This discussion has been on WP:CENT a few days; I see no reason to expect much more input. The new criteria (RD7 aside, which is best left a separate issue) are mere formalities (non-controversial transpositions of relevant bits from CSD) really. So I'd tackle it in one go. Rd232 talk 01:12, 20 May 2011 (UTC)
Does that mean we seek consensus about proposed RD#7 separately from the rest, beforehand, or on the same page but in a different section? FT2 (Talk | email) 08:31, 20 May 2011 (UTC)
Proposed RD7 has been discussed pretty heavily above already. I don't see why we should throw out all of that commentary and re-start the discussion of that criterion in a different RfC. A Stop at Willoughby (talk) 16:30, 20 May 2011 (UTC)
Leave the proposed RD7 separate, and proceed with the RD5 and associated cleanup. I don't see any link between RD7 and RD5, and so the cleanup of the latter is independent of the former. Let's move this forward. Rd232 talk 06:31, 23 May 2011 (UTC)

Revision "deletion" (which is not actually deletion) was sold as a way for non-oversighters to do oversight-like things. If someone is not doing oversight-like things, they have always been told to just use normal deletion, which everyone is familiar with, especially because revision "deletion" has complicated and strict rules. So if the plan is to deprecate normal deletion, I agree this policy needs to be clear that the standards for revision "deletion" are not actually stricter than the historical requirements for deletion, and in particular they are much weaker than the requirements for oversight. — Carl (CBM · talk) 02:31, 20 May 2011 (UTC)

That's an interesting point. Can you be more specific about what that thought leads to practically, in terms of what we're trying to do here? Rd232 talk 03:47, 20 May 2011 (UTC)
Agree with Rd232. Also I was not aware of it ever being sold as a way to do oversight-like things (though that might be how some viewed it). It was always discussed as an improved and less disruptive way to do ordinary selective deletions where the page itself was generally ok but some revisions needed deleting by admins (copyvio, gross attacks, etc: stuff admins have always handled). See for example its early talk pages or early discussion on enabling the new tool (in 2009-2010). It was always agreed the tool replaced "selective deletion" (see its earliest policy page which stated this from the date it went "live" as policy). This proposal, tool, and policy doesn't affect CSD or page deletions at all, so normal deletion is not in any way being deprecated. Admins have for years taken action on individual revision issues and since RevDelete rolled out have been told to use RevDelete for this task (less destructive) so this isn't adding anything very new either. What is being deprecated is the confusing and vague dependence of this policy on other policies, specifically what may or may not be done, by importing into it any criteria needed and then severing the explicit link between them. Does that help? FT2 (Talk | email) 08:28, 20 May 2011 (UTC)
For a long time, the goal was to get a tool that would replace selective deletion by letting us actually delete individual revisions. That would have been very useful for page moves and history fixing, which is the main sort of deletion I do. Instead we got this tool, which can't replace selective deletion because it doesn't do selective deletion. It's true that some people though this was OK as an alternative, but once I realized this doesn't do what was supposed to to, I did my best to just ignore it, sticking with the normal deletion. Perhaps I should have commented at the time.
The proposal above says "Relevant deletion policies (CSD, RevDelete, DelPol) all to be explicit that selective deletion is deprecated," so I don't understand when you say "normal deletion is not in any way being deprecated." Admins have broad latitude to delete and undelete pages, but apparently less latitude to use revision "delete". If the goal is for this to replace normal deletion (of individual revisions) then this page should clarify that the rules are not more strict for this than for regular deletion.
For example, I can do many deletions from user pages at user request that don't seem to follow the rules for revision "delete". It's not even clear I can use revision "delete" on my own user page if I am the only editor. There may be other differences as well; my point is that everyone knows the best practices for deletion, while the rules for this have been portrayed as being more strict. — Carl (CBM · talk) 11:10, 20 May 2011 (UTC)
This is getting a bit off topic so once this thread's ended it is probably better collapsed.
RevisionDelete isn't intended for page move and copy-paste fixing, or total removal of pages. There's a long term proposal to use the "RevisionMove" extension to do the former but we don't have it yet. It is intended to replace all use of deletion tools where revisions are removed but not the entire page. A decision was made that in future, deleting a revision would not cause revisions to vanish completely. It would remove the content from non-administrator view but not the fact that the revision existed. (Mainly to improve transparency and attribution). RevisionDelete policy went "live" on 17 October 2009 - coming up 2 years ago. Since that date, policy has read:
"Selective undeletion: The older method of selective undeletion (i.e. delete the entire page then selectively restore revisions) as a method of deleting revisions should be considered deprecated in favor of this system. While this method does still have a few valid uses (such as complex history merges), due its relative lack of transparency and poor efficiency, it should not be used to remove revisions from the page history."
This wasn't just a matter of "some people thought", and it's unfortunate that you (apparently) spent 18 months "ignoring it", because in fact it's a much easier and more reader-friendly way to remove unacceptable revisions or edit summaries than selective deletion ever was, and in every way does less harm to the public record. To underline the point, and highlight what looks like a misunderstanding of RevDelete's intended use, consider this: There are only 3 categories of delete that administrators can do:
Situation Tool Comment
Page copypaste fixes (where the aim isn't deletion but deletion is used as an artificial way to bring the histories of two pages under one title for merging) Selective deletion At present selective deletion is used for this. RevisionMove would be better (since it directly and straightforwardly does the desired move in one step: choose revisions, choose target page, click "OK") but it isn't live yet.
Deletion of an entire page (so that it shows up as "page does not exist") Traditional CSD/xFD deletion Traditional deletion does all of these under Deletion Policy, Criteria for Speedy Deletion, and xFD policies. None of that changes. These have been and still are, the policies and processes to remove a complete page.
Removal of specific revisions within a page's history (but not the entire page history) - almost always due to unacceptable or problematic content in those specific revision(s). RevisionDelete This is usually or always due to problematic editing, copyvios to be removed, etc. In these cases RevisionDelete is easier, less damaging, more transparent/visible to non-admins, and easier to reverse without error on pages with past deletion. Because it has solid advantages and no disadvantages in these cases, selective/traditional delete has been deprecated since 2009 for this specific kind of admin action (although they remain valid for the other kinds of action above).
RevDelete was indeed portrayed as "strict", and that's a good thing. Only a small percentage of users are admins. To those who aren't, transparency, certainty of non-abuse, and clarity of appropriate use, are important. This puts on a firm footing the times when an admin can summarily remove specific revisions from the public record, much like CSD puts on a firm footing the times a whole page and its history can be summarily deleted. CSD is also very strict, but it works just fine. If we need to enlarge any criteria, then users will surely propose amendments as they do elsewhere. This proposal is purely to make its agreed use a bit clearer, it isn't changing anything procedurally.
I hope this explains. FT2 (Talk | email) 16:18, 20 May 2011 (UTC)
The content of this page has not, to my knowledge, previously been claimed to constrain the usual practices of admins who don't use revision "delete". That would be strange, because this page is only about revision "delete", of which Wikipedia:Deletion policy makes no mention apart from a "see also". That makes sense because this process is not deletion at all - it's just a more widely available form of oversight. There is only one form of deletion available at the moment, which is to delete a whole page. We have no selective deletion tool, and this tool is not deletion at all. So admins who delete pages have needed to look at the deletion policy; admins who use revision "delete" look at this policy. — Carl (CBM · talk) 02:49, 21 May 2011 (UTC)

Carl, to be honest, I'm not sure why you think "admins have broad latitude to delete and undelete pages, but apparently less latitude to use revision 'delete.'" I think there's a pretty clear parallel between the criteria for redaction in this policy and the criteria for speedy deletion; it's supposed to be very unusual for administrators to deviate from the approved reasons for summarily deleting pages. Your example, user requests within one's own userspace, does not really clarify the problem. Why couldn't we simply propose a RevDel criterion for "revisions in one's own userspace – not including user talk pages," an adaptation of CSD U1? A Stop at Willoughby (talk) 16:23, 20 May 2011 (UTC)

Well we could, and whilst such proposal would be unlikely to succeed, a proposal would illuminate the arguments for and against allowing it. Doing such deletions because it "ought to be allowed" when it isn't and the policy is supposed to be strict is bad. Rd232 talk 20:26, 20 May 2011 (UTC)
Agreeing with Willoughby and Rd232. The point is we could propose or discuss such a criterion, it would be illuminating if we did, I suspect (like Rd232) it wouldn't pass but that doesn't make it "wrong" to propose it, and his final sentence is spot on, could not be put better: — "Doing such deletions because it 'ought to be allowed' when it isn't and the policy is supposed to be strict is bad". Summary deletion should have strict criteria, which doesn't prevent them being used or improved (eg CSD) or at times IAR'ed, but this level of care in definition and expectation does greatly reduce abuse and wikilawyering. FT2 (Talk | email) 23:16, 20 May 2011 (UTC)
I can delete my own user pages for essentially any reason - the whole thing, or just some revisions. If someone else asks me to delete their user pages, the same goes. This is all within policy. Are you saying that also applies to revision "delete"? — Carl (CBM · talk) 02:49, 21 May 2011 (UTC)
There are plenty of good reasons to delete userspace pages. There are no good reasons (that I can think of...) to delete userspace page revisions which are not already covered in the RevDel policy. Rd232 talk 03:48, 21 May 2011 (UTC)
Similar from me. At present RevDelete would arguably allow you to, purtely because traditional delete does. Should either really allow it? Probably not, because there just aren't really very many circumstances where removing a selective revision of your userspace page is of benefit other than disruption, offensiveness, privacy, or defamation related issues. By contrast entire deletion can be of benefit. So this is a case where policy is arguably slack -- it allows something that on reflection, perhaps we should seek consensus whether it should be allowed or not. FT2 (Talk | email) 03:57, 21 May 2011 (UTC)
Nothing wrong with right here, it's not that big a deal and I doubt there'll be much more debate than we've already had. FT2 has done such a good job of summarising the proposal, I don't anticipate much resistance. Rd232 talk 21:53, 27 May 2011 (UTC)

RD2 log summary

I'm slightly bothered by the RD2 log summary: "Grossly insulting, degrading, or offensive material". While this accurately mirrors criterion 2—which I do not suggest changing—I am concerned that as a log entry, it in some way confirms the success of the attack. In particular, "degrading" suggests that the victim has in fact been "degraded". To a lesser extent "insulting" has the same issue. I suggest changing the default RD2 edit summary to something like simply: "Grossly offensive material". It may or may not be desirable to also change the title of the corresponding criterion, while preserving its present meaning intact. Feezo (send a signal | watch the sky) 22:52, 22 May 2011 (UTC)

Does anyone know the original thinking behind that wording? The meanings are pretty closely related. "Offensive" would probably do for the log entry, if anyone shares Feezo's concern, which I'm not entirely convinced of because of the word "or" in there (so a reader of the log doesn't know which was meant). Rd232 talk 06:34, 23 May 2011 (UTC)
As the default wording, whatever is there will be used "as is" on a significant number of RevDelete actions, and user contributions. It's important that it doesn't misrepresent the action. Two issues if it does, 1/ we get wikilawyers or claims that users have had redaction or blocks based on incorrect reasons; 2/ users will appear to be the authors of a type of problematic edit which technically they are not. Same reason why some CSD's have "X or Y" as their default narrative. It's precise as to the reason for deleting, and that has considerable value.
The RevDelete rationale doesn't appear in the target page's history, and doesn't appear in the author's contributions. You have to look up the logs for a page and you'll just get a list of "2 revisions deleted - grossly insulting, degrading, or offensive material" ... "1 revision deleted - grossly insulting, degrading, or offensive material" ... etc. That's not really going to be a trophy for anyone as it doesn't show up anywhere except the log showing deletion (with no further details of the material visible to non-admins).
My $0.02. FT2 (Talk | email) 22:35, 23 May 2011 (UTC)
The latter paragraph is a good point (although I believe people can seek trophies anywhere) but can you give an example of "grossly insulting" or "degrading" material that couldn't be considered "offensive"? Feezo (send a signal | watch the sky) 00:21, 24 May 2011 (UTC)
Words can often be subtly different for different people. If you used "offensive" and nothing else as a criterion you would fi8nd a number of people arguing that post X wasn't really "offensive" (meaning, they personally wouldn't be offended and don't think others should be, perhaps they have thick skin or come from a culture or country where it's common). Ditto if you used "insulting". We might agree what it means, you may be sure others won't. If you use 2 or 3 terms rather than just one, people across multiple cultures and skin thickness start to be able to agree it makes sense, enough not to edit war over it so much. In some cultures and languages, and to some individuals, words like these may carry distinct shades of meaning which as a whole, indicate the kind of post we're talking about. It seems to work well - better than a single word which we then hope everyone will interpret similarly. If you want a specific example, then you need to think like this: something you feel is "offensive", someone else might feel "it's not offensive. Insulting yes but offensive? No way! It's not covered!" Hope that helps explain. In the end, gut feel that "covering all bases" helps to minimize or avoid significant wikilawyering and drama. FT2 (Talk | email) 04:12, 24 May 2011 (UTC)
If no one else is bothered by this, then I suppose it's fine, but I get a kind of "icky" feeling when I use RD2. Feezo (send a signal | watch the sky) 04:15, 24 May 2011 (UTC)
I see where you're coming from, but overall it's probably best to leave it as is. Unless anyone likes the idea of removing the description altogether, leaving just "RD2" (with a link to the policy)... Rd232 talk 12:35, 24 May 2011 (UTC)
I think there should be more emphasis on the "little/no encyclopedic value". Thus I think a summary "unencyclopedic slurs" would be a step in the right direction. (By saying this I do not mean to condone the existence of RD2, and 'encyclopedic' is still a term abused by some) Wnt (talk) 19:21, 30 May 2011 (UTC)

RD5 removal redux

Prior discussion in section above resulted in this proposal:

Proposed cessation of selective delete (other than history merge fixes) and resulting changes to deletion policies

Following discussion at Wikipedia talk:Revision deletion, the following conclusions came out which this proposal is intended to address.

Background and summary to date:

  1. Before RevisionDelete the sole means of summarily deleting individual revisions was by deleting the whole page and undeleting valid revisions ("selective (un)deletion"). No specific policy existed for this. The use of selective deletion was governed by a mixture of IAR and CSD, the latter since selective deletion is based upon deleting the page (CSD) then undeleting selected revisions. (Wikipedia:Selective deletion existed but was merely a howto guide, not policy or guideline.)
  2. RevisionDelete was introduced and its policy created in 2009. RevisionDelete is more transparent and less harmful to the public record in every situation, more easily reversed, and prevents the silent "vanishing" of revisions outside public awareness.
  3. By consensus, where the target page was not to be removed completely, inappropriate content would henceforth be redacted rather than deleted. Oversight also changed to the same approach (deprecating its old tool) so that a visible placeholder would be left there as well, with the breaching material being inaccessible.
  4. On the same day that WP:REVDEL became a policy, it was quickly updated to note that selective deletion was now "deprecated" and should not generally be used. This is still policy today. (See intro)
  5. To ensure migration of all tasks (present and future) from selective deletion to RevDelete was smooth, a criterion was added, that any selective deletion allowed under CSD could also be executed using RevDelete.
  6. It is the sense of many users now that we're past that point. Selective deletion has no real uses beyond fixing history merges and since CSD does not explicitly mention selective deletion, RD#5 can be vague in terms of what is allowed and desirable. It is time we cleaned up this situation.

The suggested clean-up is as follows.

  1. Relevant deletion policies (CSD, RevDelete, DelPol) all to be explicit that selective deletion is deprecated, with the sole exception of fixing page merge issues, meaning that the only acceptable methods for administrators to remove unsuitable content are CSD and other deletion processes specified by policy (when deleting a page and all its history) or RevisionDelete (when deleting/redacting text within specific revisions or logs).
  2. Addition of RevDelete criteria for any "extra uses" anticipated by RevDelete policy or usage, or previously handled under RD#5, CSD or selective deletion, or otherwise agreed upon. At present these are:
    1. Display of non-free images on an unacceptable page (such as userspace) - touched on by CSD#U3.
      Proposed RevDelete criterion: - (to discuss, see below)
    2. Office actions - a rare but important criterion covered under CSD#G9.
      Proposed RevDelete criterion: - The Wikimedia Foundation reserves the right to authorize staff to remove or redact specific revisions temporarily in cases of exceptional circumstances. Deletions of this type should not be reversed without permission from the Foundation.
    3. Use of RevisionDelete to deny or remove posts by persistent abusers and harassers when all else including bans, blocks and RBI fail - discussed in a separate RFC, and not part of this proposal. The separate RFC considers whether a criterion of this kind would be useful and ought to be retained for any cases under RevisionDelete.
  3. Removal of then-redundant criterion RD#5.

The community is asked to endorse this since it would be a significant change to deletion tool policy (though not their usage), and is also invited to suggest and consider any additional RevDelete criteria which may be needed, at the same time.

OK, this looks about ready to go live as an RFC. Any changes needed beforehand? Rd232 talk 19:38, 30 May 2011 (UTC)

Isn't the proposed RD7 already rejected from the conversation above? Your basis in WP:CSD#G5 makes it clearer why: "Pages created by banned or blocked users in violation of their ban or block, and which have no substantial edits by others." Clearly revision deletion never applies to this situation, because if there are no substantial edits by others you can do a normal deletion. Wnt (talk) 22:52, 30 May 2011 (UTC)
Yes, RD7 looks rejected, but I was reluctant to bring that conclusion in here before it's formalised, and that can't really be done before 30 days is up, and I didn't want to wait with this RFC. Rd232 talk 23:21, 30 May 2011 (UTC)
I've done some rewording which hopefully doesn't change the spirit but does improve the text. Major points:
  1. New bullet point in background (#3) to better explain the background and that "deletion was agreed to be replaced by redaction - Carl was unsure whether redaction = deletion, others may also be.
  2. Accuracy fix: CSD is a policy or process, not a "method". Also it's not the only policy/process.
  3. Better present the proposed new criteria - give them proper titles or descriptions, remove criterion numbers (not salient to whether we agree them and numbering depends on which ones are endorsed), reword proposed "office actions" based on the CSD wording ("temporarily remove"), and remove a lot of the text on proposed RD#7 (not needed here).
  4. Non-free content criterion - brief discussion requested, see below.
Otherwise it's pretty much as it was.
Last and for the record, I don't see a consensus that RD#7 (or something of the kind discussed) is "rejected" from the narrow discussion above, but that doesn't affect this discussion so it's moot. FT2 (Talk | email) 04:33, 31 May 2011 (UTC)
Fine, except, per my remark below, I'm not sure if the NFCC thing can't be squeezed into RD1. Also, the numbering I used sought to keep parallels with CSD - I thought that would be helpful. Rd232 talk 09:47, 31 May 2011 (UTC)

This looks fine to me. Uncontroversial, even. Dcoetzee 05:26, 31 May 2011 (UTC)

Wording of non-free content criterion

NFCC policy says nonfree images must not be used in a range of pages, not just userspace and not just when part of a "gallery". So we need to endorse their possible removal/redaction from a wider range of namespaces and pages to prevent display where it's forbidden. The policy wording must also be very clear (as CSD is). It probably needs its own criterion. Possible wording:

Display of non-free images or video on forbidden pages: - Wikipedia's non-free content policy forbids the display of non-free content anywhere except in live mainspace articles, with very limited exemptions (broadly, pages where such images are managed and discussed). The text field (only) of such revisions may be redacted. This applies even to content that the user themselves uploaded and even if the content is initially hidden by "collapsible" sections or otherwise. Mere links to non-free content are not covered by this criterion, and non-free image policy and clear consensus on a recognized noticeboard take precedence over this criterion.

Thoughts? FT2 (Talk | email) 06:08, 31 May 2011 (UTC)

Why can't that be part of (an extended) RD1? Rd232 talk 09:44, 31 May 2011 (UTC)
It can, but struck me as too long - people need to be able to read clear simple statements for summary deletion criteria. RD#1 by itself is short and simple. So is this. Put together, I'm not sure they would be. The points stated in each of them can't really be cut down, in order to ensure correct use. It seems simpler to just have it as a 2nd criterion. We aren't likely to have that many RevDelete criteria, if we have only found 7 or 8 occasions a revision might need redacting in all the history of Wikipedia so far. So having this as a stand-alone item seems easier. How would you do a single joint criterion? FT2 (Talk | email) 11:58, 31 May 2011 (UTC)
I don't see why it couldn't just be an explanatory footnote to RD1. It's not really saying anything different, to my mind, just being clearer on some specific cases. Rd232 talk 12:08, 31 May 2011 (UTC)
How to decide whether deletion is valid shouldn't really be a footnote. Footnotes just elaborate or explain details and edge-cases, or provide cites for past decisions, they aren't a way to actually remove central parts of a criterion from the body. See CSD. I've had a go at merging them but they are just too different. They cover different situations and the tests for allowing deletion would be completely different. Is there any real problem with having them as two criteria? FT2 (Talk | email) 12:56, 31 May 2011 (UTC)
OK, fine. For me it doesn't seem different enough from RD1 to justify a separate criterion, but I don't feel that strongly about it. Rd232 talk 13:07, 31 May 2011 (UTC)
This is the best I got, for what it's worth. On reflection and with a bit of extra work, the merged version isn't that bad. I think it might even be preferable. Which do you prefer:
Merged version:

  1. Copyright and non-free image violations. This covers two cases:

    • (a) Blatant copyright violations that can be redacted without removing attribution to non-infringing contributors. If redacting a revision would remove any contributor's attribution, this criterion can not be used;
    • (b) Removal of non-free images where they are forbidden. Non-free content display is forbidden anywhere except in live mainspace articles, with very limited exemptions (broadly, pages where such images are managed and discussed). The text field (only) may be redacted.

    These apply even to content that the user themselves uploaded and even if the content is initially hidden by "collapsible" sections or otherwise. Mere links are not covered by this criterion. Copyright policy, non-free image policy, best practices for copyrighted text removal (see: Wikipedia:Copyright problems) and clear consensus on a recognized noticeboard, all take precedence over this criterion.

Unmerged version:
  1. Blatant copyright violations that can be redacted without removing attribution to non-infringing contributors. If redacting a revision would remove any contributor's attribution, this criterion can not be used. Best practices for copyrighted text removal can be found at Wikipedia:Copyright problems and should take precedence over this criterion.
  2. Non-free images on forbidden pages. Non-free content display is forbidden anywhere except in live mainspace articles, with very limited exemptions (broadly, pages where such images are managed and discussed). The text field (only) of such revisions may be redacted. This applies even to content that the user themselves uploaded and even if the content is initially hidden by "collapsible" sections or otherwise. Mere links to non-free content are not covered by this criterion, and non-free image policy and clear consensus on a recognized noticeboard take precedence over this criterion.
FT2 (Talk | email) 13:22, 31 May 2011 (UTC)
Hm, thanks for your efforts. I think either version acceptable, but I slightly lean to the 2-criteria version as being clearer and more maintainable (less likely to lose clarity in future). Rd232 talk 14:56, 31 May 2011 (UTC)
Argh. This is clear abuse of the wiki collaborative model ! After a serious discussion, how dare you reverse yourself and agree to my view, just when I've reversed myself and agreed to yours! Now what on earth shall we do? :) Do any other users have a preference? FT2 (Talk | email) 15:43, 31 May 2011 (UTC)
Well you've done so much work on this, I'm happy to leave the choice to you (or else the RFC could mention both possibilities). Let's just get the show on the road. Rd232 talk 19:28, 31 May 2011 (UTC)
This is a huge stretch from the original mission of suppressing "blatant copyright violations". Whatever policy Wikipedia may have about serving up Fair Use images, we allow people to click on the image and get the image by itself, outside of the context of the intended article. If you copy the URL from any page containing a Fair Use image, you can even get the image without any kind of text or annotation; you could embed that to use in your own (offsite) web page if you want to, though certain web etiquette people are snooty about that. These things don't seem to be regarded as problems. So it is unreasonable to use RevDel to suppress a history version of some discussion or article ... even if it was not hidden in a show/hide box. The sense of legal urgency in the original RD1 is completely absent in this case. It sounds like mission creep - like someone just wants to be able to go through the history and black out things they don't like. Wnt (talk) 17:46, 31 May 2011 (UTC)
It's not a "huge" stretch - or any stretch at all. In fact it's been part of Wikipedia:Copyright problems/Advice for admins#Handling copyright violations for ages, that Delete and also RevDelete may be used or is recommended for use on copyvios even if reverted. In some cases it goes so far as to state that it should be used, or is an admin's judgment call to use. It is important to remember that non-free content is not "copyright free". It's copyrighted material we serve under a legal exemption. If that exemption doesn't exist and there isn't a valid non-free rationale, then the material is not just an image on the wrong page. It's a copyright violation. So an admin needs the right to prevent it being visible in certain cases. Again to quote NFCC policy: "Restrictions on location - Non-free content is allowed only in articles... and only in article namespace, subject to exemptions". CSD adds that deletion may be used to remove "galleries" of non-free images in userspace. Selective deletion has been used for these. RevDelete will be used in future. So admins need the policy to cover their scope for discretion in such cases if CSD and RD are severed.
Of course nobody's advocating going back to year zero redacting all revisions with possible NFCC content. We don't do that for other copyvios either. But in those cases where NFCC content appears on an improper page, it constitutes a blatant copyright violation and has the same handling as a blatant copyright violation, including potentially either Deletion or RevDeletion at discretion. If the concern is that the wording is too strong, a minor edit will emphasize the current norms.
Possible edit: "... The text field (only) of such revisions may be redacted at admin discretion, in line with existing policy and advice on handling copyright violations..."
(And the collapse box point is to prevent the most likely kind of gaming: "It's not displaying non free images, they are collapsed!" Fail.)
FT2 (Talk | email) 19:10, 31 May 2011 (UTC)

If it's not acceptable to revdel content linking to non-free images, it should not be acceptable to revdel content showing them.--SarekOfVulcan (talk) 19:42, 31 May 2011 (UTC)

There's a big difference. A link to a non-free image is not itself a copyright violation. For example adding "We should use this image" to a user talk page discussion is not a breach of non-free content (because it doesn't cause the actual content to be displayed), while actually including the image would be a breach. That's why RevDelete isn't needed for non-free content links. FT2 (Talk | email) 20:19, 31 May 2011 (UTC)
Yes, but you can change a display to a link by adding a colon. So why should the entire revision be deleted, when adding a colon can solve the problem? --SarekOfVulcan (talk) 21:06, 31 May 2011 (UTC)
Because if it's posted on-wiki without a colon, you can't add the colon in afterwards to that revision, and admins may need to exercise their discretion in some cases to remove copyright material from public view even in a history revision - see explanation above (19:10). FT2 (Talk | email) 21:51, 31 May 2011 (UTC)
Well, let me come at this from another angle. Earlier today, I revdeled a lengthy copyvio explaining exactly what the hidden Jewish caballistic meaning of a grainy photograph was. There was nothing I could have added to that to make it less a copyvio -- it had to come out. On the other hand, I can change that Olympic bid logo above to not-a-violation by adding a colon. I don't have to remove anything that's already there, I merely add wikimarkup to make it acceptable. So, why should it be necessary to revdel it when it's trivial to fix it? Actually, I think I'm starting to see scenarios where this makes sense, but let's go with my original argument for a bit.--SarekOfVulcan (talk) 22:23, 31 May 2011 (UTC)
I still don't accept this. There's a difference between a blatant copyright violation and a trivial, incidental, debatable copyright violation almost nobody even sees. Even in the old days of manual typography there was always a chance a Fair Use figure would end up being found in a defective or defaced book in the library while the reasoned discussion on the facing page had parted company. And there's a difference between even such a trivial copyright violation and the Wikipedia policy broadly drawn to ensure copyright violation doesn't happen at all. Therefore, just because a Fair Use image appears in a place that violates the Wikipedia policy does not make it a blatant copyright violation. In fact, the fact it appeared in the history at all shows that someone likely thought it was appropriate even within the bounds of careful policy.
Furthermore, the Fair Use criteria include consideration of the effect on the market for the copyrighted work. You wouldn't be questioning an item if it weren't already being used, legally, on some current Wikipedia article - with no image file, there's no image. How much does it affect the market for a copyrighted work to use an existing Fair Use image somewhere deep in the history of another article? Less than in the article out in the open, I would say. Which means that the rationale for using the article in the history revision should also need to be less. Thus, even a use that failed to reach consensus for appropriate Fair Use out in the open should quite likely be alright to use in the article history. Wnt (talk) 22:58, 31 May 2011 (UTC)

Suppression by administrators

Please see WP:VPI#Suppression by administrators. — Kudu ~I/O~ 20:55, 2 September 2011 (UTC)

RfC: Proposed changes to deletion policies, including removal of RD5

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
The one problem I see with the discussion and how it flowed out was the inadequate discussion regarding each option. If any of these were to look for a new consensus in the future, separation of the proposal would be an idea. That said, regarding the concerns shown with removing selective deletion in the way of copyright cleanup, a full discussion did not entail, and in general there was no consensus to remove it, so I would wrap it up as the community leaning towards opposing the removal of selective deletion. The proposed RD7 criterion did not find a consensus, and it was indicated that a rewording or modification of it might be more appropriate. For RD8, there was no specific opposition to it, though the was very little general opposition to it, therefore the community supports adding RD8 to the current revision delete process. There was no consensus at this time to remove RD5 from the options for revision delete. -- DQ (ʞlɐʇ) 02:15, 20 April 2012 (UTC)

A multi-point proposal on adjusting the policies around selective deletion to more accurately reflect current practices. 16:18, 10 February 2012 (UTC)

Proposed cessation of selective delete (other than history merge fixes) and resulting changes to deletion policies

Following discussion at Wikipedia talk:Revision deletion, the following conclusions came out which this proposal is intended to address.

Background and summary to date:

  1. Before RevisionDelete the sole means of summarily deleting individual revisions was by deleting the whole page and undeleting valid revisions ("selective (un)deletion"). No specific policy existed for this. The use of selective deletion was governed by a mixture of IAR and CSD, the latter since selective deletion is based upon deleting the page (CSD) then undeleting selected revisions. (Wikipedia:Selective deletion existed but was merely a howto guide, not policy or guideline.)
  2. RevisionDelete was introduced and its policy created in 2009. RevisionDelete is more transparent and less harmful to the public record in every situation, more easily reversed, and prevents the silent "vanishing" of revisions outside public awareness.
  3. By consensus, where the target page was not to be removed completely, inappropriate content would henceforth be redacted rather than deleted. Oversight also changed to the same approach (deprecating its old tool) so that a visible placeholder would be left there as well, with the breaching material being inaccessible.
  4. On the same day that WP:REVDEL became a policy, it was quickly updated to note that selective deletion was now "deprecated" and should not generally be used. This is still policy today. (See intro)
  5. To ensure migration of all tasks (present and future) from selective deletion to RevDelete was smooth, a criterion was added, that any selective deletion allowed under CSD could also be executed using RevDelete.
  6. It is the sense of many users now that we're past that point. Selective deletion has no real uses beyond fixing history merges and since CSD does not explicitly mention selective deletion, RD#5 can be vague in terms of what is allowed and desirable. It is time we cleaned up this situation.

The suggested clean-up is as follows.

  1. Relevant deletion policies (CSD, RevDelete, DelPol) all to be explicit that selective deletion is deprecated, with the sole exception of fixing page merge issues, meaning that the only acceptable methods for administrators to remove unsuitable content are CSD and other deletion processes specified by policy (when deleting a page and all its history) or RevisionDelete (when deleting/redacting text within specific revisions or logs).
  2. Addition of RevDelete criteria for any "extra uses" anticipated by RevDelete policy or usage, or previously handled under RD#5, CSD or selective deletion, or otherwise agreed upon. At present these are:
    1. RD7: Display of non-free images on an unacceptable page (such as userspace) - touched on by CSD#U3.
      Previously discussed above at #Wording of non-free content criterion where two editors supported it and two editors opposed it. Further opinions are sought.
    2. RD8: Office actions - a rare but important criterion covered under CSD#G9.
      Proposed RevDelete criterion: - The Wikimedia Foundation reserves the right to authorize staff to remove or redact specific revisions temporarily in cases of exceptional circumstances. Deletions of this type should not be reversed without permission from the Foundation.
  3. Removal of then-redundant criterion RD#5.

The community is asked to endorse the above changes, as they would comprise a significant change to deletion tool policy (though not their usage), and is also invited to consider, at the same time, the additional RevDelete criteria which may be needed, which are listed above.

This proposal was largely formulated by Rd232, in collaboration with FT2, as a result of discussions on this page (#Proposed changes and #RD5 removal redux) last May. I don't think that most of this proposal will be controversial at all, because it essentially codifies what is already practiced. That includes the proposed "office actions" RevDel criterion, which would simply make explicit what is already implicit under WP:OFFICE policy.

The only issue that I expect to be controversial, then, is the proposed RevDel criterion which would, filling the void that would be left by RD5, explicitly authorize administrators to delete revisions outside of mainspace which contain non-free images – a parallel of CSD#U3. You can get a sense of the arguments for and against this addition in the #Wording of non-free content criterion section above. A Stop at Willoughby (talk) 16:18, 10 February 2012 (UTC)

  • Personally, I support this entire proposal, which, as I said, I think is largely uncontroversial. I only weakly support proposed criterion RD7, which would allow administrators to delete revisions outside of mainspace that contained non-free images; it seems like it would be overkill in most situations, but in certain cases (such as the inclusion of many non-free images on an old revision of a userpage) I can see the use of RevDel being appropriate. Like CSD#U3 itself, of course, I think this criterion would be used relatively rarely. A Stop at Willoughby (talk) 16:18, 10 February 2012 (UTC)
  • Support everything except RD7. U3 covers the deletion of a page in userspace that is just a gallery of non-free images. Per WP:UP#NOTSUITED, if the individual just has one or two images, or has a gallery and text, U3 does not apply, but the images can be removed from the page by an administrator. There is nothing in that provision that would suggest any need to revdelete edits where the image is visible on the page. It is worth noting that it would be possible for an editor to use a non-free image in userspace and not be in breach of the US understanding of fair use, so it is not universally a question of copyvio. Elen of the Roads (talk) 16:53, 10 February 2012 (UTC)
  • Oppose termination of selective deletion - I've had to use selective deletion a couple times because a long term abuser used autoconfirmed socks to add over 150 malicious revisions to List of indigenous peoples and to several related articles. Reaper Eternal (talk) 17:02, 10 February 2012 (UTC)
Looking at the delete history, I'm not sure I see the logic in this. Not only it's pretty unusual, but it doesn't seem to have been any easier selecting revisions to restore than selecting revisions to delete. At most the RevDelete interface might benefit from clear "select all/none" buttons ("select/unselect all revisions in a range" is already included: shift-click a checkbox). If those were added would that solve it? FT2 (Talk | email) 23:16, 10 February 2012 (UTC)
I can't speak for Reaper Eternal, but I think the point there was to unclog the revision history, not the ease of the hiding.Jasper Deng (talk) 23:27, 28 February 2012 (UTC)
  • Comment RevDelete is not a form of "deletion", it is a form of "edit suppression" and a variation of oversight. "Deletion" refers to the process through which revisions end up in the "deleted revisions" list for an article (visible to admins from the article's history page). It is still the case that the only way to delete individual revisions is to delete the entire page and to the undelete the desired revisions; RevDelete did not change that. — Carl (CBM · talk) 18:25, 10 February 2012 (UTC)
  • Weak oppose as stated. There are at least two legitimate reasons to delete (not suppress) individual revisions: (1) As part of complicated page merges, when certain revisions are intentionally left deleted. (2) On users' own userpace pages (not talk pages) they have always been free to delete and undelete revisions as they please. My general feeling is that the current system is functioning well, and I don't see any evidence of widespread abuse of either RevDelete or selective deletion, so I don't think there is a need to change the existing policies. — Carl (CBM · talk) 18:25, 10 February 2012 (UTC)
Page merges are already excluded from this - they are the one time selective delete/undelete is still needed. Any ability users have to delete or undelete individual revisions in their userspace, or to ask someone else to do so, would need to be stated under RevDelete criteria if it were to continue. Is your concern here that an additional criterion is needed for this? If it had concensus would that solve this concern? FT2 (Talk | email) 23:16, 10 February 2012 (UTC)
We don't need an additional criterion for suppression - I may really want to delete revisions, for example to keep the history of my sandbox clean\, which has always been permissible under CSD. The suppression system leaves the revisions in the history, which misses the point. The RevDelete suppression system was not intended for this sort of housekeeping work. — Carl (CBM · talk) 00:35, 11 February 2012 (UTC)
Now that I see that this would also extend the CSD criterion so that any non-mainspace page that ever showed a nonfree image could be suppressed, I am leaning more towards strong oppose. — Carl (CBM · talk) 00:47, 11 February 2012 (UTC)
See WP:NFCC policy summary on this a few lines below. FT2 (Talk | email) 01:31, 11 February 2012 (UTC)
  • Bolded text I don't support suppressing userspace revisions simply because they have a link to a non-free image. The pages don't "contain" a non-free image. I'm not going to vote here because there is a lot of non-controversial stuff here. Not sure why you wanted to bundle something controversial with some fairly non-controversial things. Gigs (talk) 18:54, 10 February 2012 (UTC)
Wording is pages that "display" a non-free image, not pages that "contain a link to one" (and nobody's saying "suppress" (=oversight)). The non-free image policy is that NF images must not be used (for any reason) outside mainspace, with a few very strict and limited exceptions. Our history pages are still pages in the encyclopedia and may at times need to be deleted/redacted like any other revision found to contain significant copyvios or copyright policy breaches. So this simply says RevDelete can be used for that purpose like other deletion tools have previously been. Does that reassure? FT2 (Talk | email) 23:16, 10 February 2012 (UTC)
I don't think there is a need to delete or suppress old revisions of a page that happen to show a non-free image (here supress includes RevDelete which is not deletion). The NFCC requirement is only that such images should be removed; there has never been a CSD criterion for such things, we have generally just removed the image and moved forward. I think it opens a can of worms to start hiding revisions of other users' pages just because a non-free image was once used. — Carl (CBM · talk) 00:38, 11 February 2012 (UTC)
Explanation/comment on NFCC image removal (redundant in light of update below)
See WP:NFCC policy: "Non-free content ... may be used on the English Wikipedia only where all 10 of the following criteria are met", and specifically policy #9: Restrictions on location: Non-free content is allowed only in articles ... and only in article namespace, subject to exemptions" ... "Exemptions ... are made for ... certain administrative ... pages as necessary to creating or managing the encyclopedia, specifically for those that are used to manage questionable non-free content".
History pages are a core part of Wikipedia and its public pages, and technically can't be modified. When a user edits a revision to prevent display of a non-free image, the software creates a new history page leaving the old one intact. In some cases redaction to not display the history revision may be needed. Do you have an alternative solution or feel that NFCC shouldn't apply to history pages? NFCC is unambiguous and strict on this - we must not display non-free on Wikipedia outside content and at times a few admin pages, full stop. History pages are part of Wikipedia. We can't fix everything but if administrators agree some case should be fixed, REVDEL criteria should support their using the only tool that can so.
How to fix this? I think NFCC should say under what circumstances redaction is reasonable, to prevent pile-on activity and fanatical removal of old revisions, but the issue lies with NFCC not REVDELETE. I don't think anyone will disagree that sometimes redaction might be needed to prevent non-free image display. The most elegant solution is for NFCC to provide guidance when it's appropriate, not for REVDELETE to disallow use. Even where CSD#U3 applies, REVDELETE would be preferable. FT2 (Talk | email) 01:31, 11 February 2012 (UTC)
UPDATE and related RFC: This needs eyeballs at WT:NFCC, so I've opened an RFC there on the specific issue. Would it be agreeable to consider this issue effectively resolved, in that whatever's decided at NFCC gets reflected in deletion policies? (I have no great view either way, my concern is simply that if NFCC implies a need for deletion then deletion policy has to allow it). FT2 (Talk | email) 03:05, 11 February 2012 (UTC)
Most users aren't admins and selective delete removes their ability to at least see what's happened, that something was deleted, and those parts of the history (time, date, etc) that didn't need deletion. So for non-admins (= most users) it's extremely important and provides transparency. Arguably even for a page with much deletion, the user has the right to see there has been deletion and its history if they want. Seelctive delete means they never can. What this does suggest is a "hide deleted edits" button or script, much like Special:RecentChanges' "hide bots", "hide minor edits" etc. Is anything else crucial that's missing for you? FT2 (Talk | email)
There are times when we don't want transparency.Jasper Deng (talk) 01:09, 11 February 2012 (UTC)
    • We already have a "hide edits" system, which is to move the revisions to the deleted edits list where they belong. The feature that is most missing from RevDelete is the option to actually delete individual revisions. — Carl (CBM · talk) 00:35, 11 February 2012 (UTC)
  • Support - As was mentioned above, most editors aren't admins. I also disagree that it will "clog history" for the articles. As someone who cannot see deleted edits, it would be far more informative if 20 individual edits were redacted than one group of "several". This level of visibility would also make me more comfortable that other admins could see what was going on and ensure that the process was not being abused. Obviously it is still important for hist-merges. I couldn't care less if Admins want to use it on their own user pages (not talk pages), that won't effect anyone else. ▫ JohnnyMrNinja 00:13, 11 February 2012 (UTC)
  • Weak whatever support on most of it: most of it is uncontroversial enough, so I'm mostly okay with it. But I don't quite see the point of the RD7 component of this proposal and therefore oppose that for the reasons given by Elen of the Roads. I'm willing to change my mind if we can have reasons why occasional non-free images in userspace need revision deletion rather than just an edit/revert to remove them. Don't get it. —Tom Morris (talk) 11:17, 11 February 2012 (UTC)
  • Support all changes, but reword RD7 to explicitly make it clear that revision deletion is only acceptable when one or more of the non free images in question still is being used elsewhere in full compliance with NFCC. There is no need to hide links to dead files.— Train2104 (talk • contribs) 21:02, 11 February 2012 (UTC)
  • Comment Selective deletion still has potential value in copyright cleanup, if only in those cases where extensive copyright issues exist in too many versions of an article to make RevDeletion practicable. Prior to the enabling of RevDeletion, we routinely used selective deletion in copyright work to move articles, delete edits containing extensive issues, and then move back the edits in which that material did not appear. This is beneficial to help avoid inadvertent restoration of content. These days, RevDeletion almost always serves that purpose, but I am not sure that we should eliminate selective deletion entirely from the toolbox in copyright work. Currently, the RevDeletion criterion for copyright work notes that "Best practices for copyrighted text removal can be found at Wikipedia:Copyright problems and should take precedence over this criterion", and I would be concerned about any chance to policy that would completely eliminate that option. Otherwise, I generally support the idea that revision deletion should take precedence wherever it can for the reasons expressed above. --Moonriddengirl (talk) 16:29, 13 February 2012 (UTC)
  • Oppose removal of RD5. It is the (relatively) low-drama way to remove something pending oversight. Franamax (talk) 22:31, 13 February 2012 (UTC)
  • Oppose phasing out selective deletion, as it is still useful in combating extreme cases of vandalism. After a coordinated vandalism attack or similar on an article, users should not be forced to wade through pages upon pages in the edit history for any constructive edits (not to mention not keep their badges of honor documented). If possible, I would support a feature which would send individual edits to the "deletion side" instead of merely hiding them. However, I acknowledge fighting an uphill battle on this one. That said, I maintain that it should still be only used sparingly and in those extreme cases that I am referring to. --MuZemike 01:47, 17 February 2012 (UTC)
  • Support this measure, which would help to increase transparency by sometimes preventing administrators from removing edits without leaving a trace. Basically, administrators shouldn't use the "delete" button unless there's a community consensus to permit deletion, and a proper paper trail helps keep them honest.—S Marshall T/C 12:26, 20 February 2012 (UTC)
  • Oppose as to selective deletion, per Reaper Eternal, Moonriddengirl, and MuZemike. No opinion on the rest. T. Canens (talk) 04:30, 23 February 2012 (UTC)
  • Weak overall support, I haven't studied the issue enough to look for holes. BUT I think we should have a policy statement that if ignore all rules is invoked to do a selective delete for reasons not approved by the then-current policy, that a strong justification for using selective delete be logged to a place where checkusers or preferably all administrators could see it. The reasons for this are twofold: 1) if we see the same reason over and over again and it's a good one, it can be added to the policy, and 2) it provides a feedback mechanism for cases when revision delete is preferable. davidwr/(talk)/(contribs)/(e-mail) 20:26, 24 February 2012 (UTC)
    Interesting idea. With page deletion, there's plenty of feedback because there are lots of eyes on each level of the deletion process, from new-page patrollers to PROD-monitors to XfD-participants to deleting administrators. Over time, participants in each level pick up on trends and discuss them at the appropriate talk pages; many times, they alter policy as a result. But with revision deletion, my impression is there isn't much monitoring going on, as admins usually act alone in situations where RevDel is appropriate. The result is that admins don't pick up on trends that could lead them to modify policy in various ways. So, I think you have a point here. Discussion about RevDel among administrators who use the tool is something that should be regularly facilitated. A Stop at Willoughby (talk) 22:40, 24 February 2012 (UTC)
  • Oppose all. I would support some well-crafted guidelines as to when to use selective deletion; its function is generally better handled by RevDel, but there are some important exceptions where selective deletion is a much better tool, so I can't support (further) deprecating all its uses. I don't see RD7 as necessary at all; allowing such a criterion could have the effect of wiping out huge chunks of user page or user talk page history during the time a non-free image was displayed, with no apparent benefit to anyone. 28bytes (talk) 03:58, 28 February 2012 (UTC)
  • Oppose removing RD5. The proposed two criteria to replace RD5 are not the only reasons under the header (quoting the deletion drop down option) "other valid deletion under deletion policy". Ks0stm (T•C•G•E) 00:40, 8 March 2012 (UTC)
  • Oppose - maybe I'm missing it, but what is this attempting to solve? Also, as noted above, revdel, and selective deletion are two different things, and as such are likely to be used for different things. 28bytes above said it clearly enough, so i won't further repeat. I also think that, since some of the various parts are starting to sound controversial, that those should be broken out of this into a separate discussion. And finally, my experience: With revdel is once it's gone, I, as an admin, couldn't bring it back. I tried to revdelete someone redirecting a page, with the idea to merge the page history elsewhere (it was a copy/paste move) and then restore the person's redirect. Well, it didn't work. I can see that there was a deleted revision, but apparently I couldn't restore it. (Basically I was trying to use revdel the way we had used selective deletion in the past.) I ended up manually creating the redirect. Perhaps the whole thing isn't a big deal, but I think I should have been able to restore the editor's redirect. - jc37 18:27, 10 March 2012 (UTC)
    Just tested RevDel, and restoring worked fine: find the revision in the history that has been RevDeled, click the "del/undel" link, uncheck the appropriate check boxes (or use the appropriate radio buttons, if you're adjusting multiple revisions at once) on the revdel screen, and submit. But note that fixing a cut-and-paste move is one of the things we would still need selective deletion for (as noted in the proposal), as RevDel can't be used for that. Anomie⚔ 17:53, 12 March 2012 (UTC)
    Struck that part of my comments. I was attempting to use revdel as part of a history merge, which, as noted, it isn't intended for. Thanks for your help : ) - jc37 22:53, 13 March 2012 (UTC)
  • Oppose - per MRG and MuZemike. Selective deletion has practical uses beyond histmerges, in particular in copyvio cleanup. MLauba (Talk) 10:25, 22 March 2012 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Inline-twin engine

Can I ask, has Inline-twin engine been "revision deleted" and if so, for what reason? Thank you.

The pages was just deleted by User:RHaworth but does not appear here [4]

"09:43, 4 July 2012 RHaworth (talk | contribs) deleted page Talk:Inline-twin engine (G8: Talk page of a deleted page)"

I notice that my own contributions to it have also been deleted [5] as have others. [6].

What does one do if one wants to question a Revision deletion?

--Bridge Boy (talk) 10:58, 4 July 2012 (UTC)

No, Inline-twin engine has not been "revision deleted". For a short time it was entirely deleted as a duplicate of Straight-two engine, but then it was restored and listed for discussion at Wikipedia:Articles for deletion/Inline-twin engine instead.
If you want to question a revision deletion (or an ordinary deletion), you should contact the administrator who performed the action on their talk page; if that fails you could raise the issue at WP:DELREV, WP:AN, or WP:ANI. Anomie⚔ 17:43, 4 July 2012 (UTC)
Could you point me to a page which clarifies the difference between deletion, Revision deletion and "entirely deleted"?
What was strange about this was that, firstly, it was deleted but appeared in the logs, then the logs disappeared, and so did all mention in any users' contribution logs all in a matter of hours of the tagging and without discussion.
Yes, I see it has now been restored. --Bridge Boy (talk) 09:42, 5 July 2012 (UTC)

Userbox?

Is there a userbox incorporating the "Wikipedia administrators willing to handle RevisionDelete requests" category? — Richwales 02:25, 7 September 2012 (UTC)

Spam in edit summaries

I understand that RD3 doesn't cover "mere spam links", for obvious enough reasons. However, what about spam in edit summaries? This is a form of spam that is comparatively highly visible, at least when it falls within recent enough revisions to be shown on the first page of edit history. I feel like it would be right to redact the edit summary only in these cases. — Hex (❝?!❞) 16:45, 10 December 2012 (UTC)

I saw briefly a page listing Revision Deletions for a particular article talk page -how can I call this up again  ?

I saw briefly when editing a page listing only Revision Deletions for a particular article talk page -how can I call this up again  ?--— ⦿⨦⨀Tumadoireacht Talk/Stalk 06:45, 24 December 2012 (UTC)

Revision deletions are part of the deletion log. For example, here are those for the page Talk:Circumcision. Generally speaking, if you go to any page and click on its "View history" tab, there should be a link "View logs for this page" immediately below the page title. This lists all page logs, but you can filter it down to the deletions (which include the revdels) by selecting "Deletion log" from the first drop-down menu.
If you have administrator rights, you can view the deleted revisions at Special:Undelete, for example Special:Undelete/Talk:Circumcision. --Redrose64 (talk) 15:17, 24 December 2012 (UTC)
Archive 1Archive 2Archive 3Archive 4Archive 5
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:Revision_deletion/Archive_3&oldid=1137452171"