Wikipedia talk:Votes for undeletion/Deletion review proposal

Discussion on first part of proposal &mdash; the creation of Deletion Review.

Introductory remarks
Just to repeat the intent part:

The intent is to, without reducing Wikipedia to a simple democracy, provide a route for some action if most people are concerned about process, and a route for correctional'' action if a significant proportion of people mandate it. The clarity in the numbers is intended to prevent Deletion Review having the same interpretation disputes that will bring debates to the process in the first place. It is anticiapted that WP:AN/I be used to claim that someone has counted incorrectly; hopefully that will happen only extremely rarely.''

So we should talk about whether people think the suggestion for how we would work deletion review will be satisfactory or not. Hopefully we can formulate something that will allow us to create the Deletion Review mechanism as soon as possible.

Note that we still have the alternative of retaining the simple majoritarian system on VfU at present, but there were/are fears about this providing an "end run" around a keep result for those who would delete.

Additionally, several editors suggested taking speedies out of scope, and having them dealt with as they are at present on an ad-hoc basis. Should they be in scope to DR? -Splash talk 00:57, 29 September 2005 (UTC)

Caveat: Unaware of discussion

 * Note the section of the undeletion policy which says "...This may happen because they were not aware of the discussion on Wikipedia:Articles for deletion (AFD) or Wikipedia:Miscellaneous deletion (MD)...". Would this case still be included in DR (note that there is an example currently on VfU)? If so, where in the new purpose is this accomatated? do we need to edit to handle this case? DES (talk) 01:10, 29 September 2005 (UTC)
 * I don't think that entire section need change. The first box is intended to go on the VfU page, as it already is, and the second to replace the Wikipedia talk:Undeletion policy section. I should make that clearer. -Splash talk 01:14, 29 September 2005 (UTC)
 * I was thinking that "This process should not be used simply because you disagree with a deletion debate's reasoning" might need to indicate that "I didn't see it" nominations are an exception. DES (talk) 01:16, 29 September 2005 (UTC)
 * I added a clarifying bullet to the post-amble text saying that the scope is not changed except to include all kinds of deletion debates. That should avoid the same question being asked over each of the current list on Undel policy. -Splash talk 01:39, 29 September 2005 (UTC)
 * Unless they present new information, why are "I didn't see it" nominations allowed? That's me saying: I don't think that we should allow nominations of this variety.  brenneman (t) (c)  02:17, 29 September 2005 (UTC)
 * The contention, I presume, is that had the nom seen it his or her arguments would heve changed the result. (this might be particualrly persuasive when the nom is the creator or a primary editor of the page in question.) This may be new information, or perhaps a new argument. I suupose the contetion might also be that had the nom seen the XfD debate, that vote would have made a difference. This would IMO be plausible only in debates with borderline consensus and low participation. Are you suggesting that such nominations be removed from the undeletion policy, which now explicitly permits them? DES (talk) 02:32, 29 September 2005 (UTC)
 * To be frank, it's a terrible clause. Whomever inserted it in the first place either didn't fully think out the implications, or intended something other than what was written.  Unless an editor has something terribly compelling to say (e.g. new information) so what if they didn't get to vote.  Because that's what we're talking about here, "Oh, my 'keep, no reason provided' would have made it 65.99214% and thus no consensus."  (article undeleted) "But wait, our two "delete, just because" votes would have made it 66.02%." (article deleted) "Hold on..."  ad nauseam. Kill it with fire. -  brenneman (t) (c)  13:28, 8 October 2005 (UTC)


 * "I didn't see it" nominations are allowed, and always should be allowed, because of Assume good faith. Really if VFU actually did its job under the undeletion policy of looking at articles and deciding to undelete them because of an AfD screw-up, I don't think we'd be having this discussion. "So what if they didn't get to vote" misses the point entirely.  The article has been deleted and someone who thought it a worthwhile article wasn't around when the deletion discussion took place.  End of story.  If enough people agree, undelete and rerun AfD. Who knows what will emerge in an AfD discussion.  --Tony Sidaway Talk  21:33, 18 October 2005 (UTC)

Other types of restoration

 * Note that the curent procedure for temporary undeletion and history-only undeletion probably need not change. I think the proposal should make this clearer. DES (talk) 01:10, 29 September 2005 (UTC)
 * Yes, ok. -Splash talk 01:14, 29 September 2005 (UTC)

Wording
I find two illogicalities about the proposal.
 * Presumably the closer took some specific action. Therefore any support of any other action supports Overturning, although I agree that "Overturn and (action)" should be encouraged.
 * Note that the default action associated with an Overturn is to relist on xFD rather than to reverse. If someone wants to overturn and delete or overturn and keep, they should so specify. This could perhaps be better phrased, I agree. -Splash talk 02:01, 29 September 2005 (UTC)
 * I don't see why new information must be preceeded by an Overturn vote. Sometimes new information supports the closer; and sometimes an editor may have some information but still be undecided. Septentrionalis 01:52, 29 September 2005 (UTC)
 * The idea is that the primary vote is always either Overturn or endorse, and actions are an optional secondary vote as i understand it. To make things easier for the closers and avoid debates over interpretaion, each person should explicitly select either overturn or endorse, IMO DES (talk) 01:56, 29 September 2005 (UTC)
 * I think the idea here is that whn an overtun is based on new info, it should be a plain overturn, (which is implictly overturn and relist) not an overturn and (action). Perhaps the wording needs improvement, or perhaps i have misunderstood. DES (talk) 01:58, 29 September 2005 (UTC)
 * That was intended to allow for the current provision at VfU where if you have significant new information, you can quite reasonably request the restoration of an article deleted entirely in-line with consensus &mdash; at which point it will usually be expected to return to xFD with the new information inserted. -Splash talk 02:01, 29 September 2005 (UTC)

Perhaps then replace both shoulds by we recommend or should normally be. It sounds like Splash wants to encourage a default phrasing, but the present text can be read as a command (and Wikilawyers will so read it, if possible: "He didn't vote to Overturn the decision to delete; he just voted to Merge - so it doesn't count." Let's not go there.)
 * If this is to be a reccomedation let's make it a strong one. Personally, for clarity, i wouldn't object to a clearly stated rule that any comment without one of the standard terms is not to be counted in the numerical totals. DES (talk) 02:41, 29 September 2005 (UTC)

I also have qualms about the phrasing: Endorse will normally mean disagreeing with the DR nomination at the top of the discussion. I foresee honest confusion. How about Close (i.e.. as it was closed) and Relist?

Nota bene: I'm nitpicking because I think this is a sensible proposal. Septentrionalis 02:27, 29 September 2005 (UTC)


 * Fair enough. My idea when i suggeste those terms was that we should endorse or overturn the closer's action. If the nom includes an explicit overturn vote (as would be ususl), it will be clear that endorse does not mean endorse the nom. Once people start using this terminaology i think it will be clear. Close seems to mean to clsoe the current debate, adn relist is one of the possibe actions followign an overtunr (the default action) so for example "Overturn and merge" could not be expressed as "relist and merge". I am open to other suggested wording, however. DES (talk) 02:38, 29 September 2005 (UTC)
 * Yes, those are good points. One of the reasons votes are evil is because they constrain the way people can express themselves, so we should look for as flexible a jargon as possible. How about we make it explicit: overturn closure [and (action)], or endorse closure, with relist simply understood to be, well, relist. And yes, we should patch that other gap, too, I'll go an relax the phrasing. -Splash talk 02:48, 29 September 2005 (UTC)
 * How about Endorse decision or Endorse result then. I think that is clearer. Furthermore if speedies are in scope, they have a "decision", but not a "closure". DES (talk) 03:00, 29 September 2005 (UTC)
 * Yeah, but as stated above, there's a risk of some users wanting votes that don't read overturn and delete but just delete to be disqualified, even when overturn was implicitly stated within the delete. (Again, I'm playing devil's advocate.) Tito xd 04:24, 29 September 2005 (UTC)
 * The rule should be stated clarly one way or the other, to prvert arguments about it. Personally I would have no problems with a rule that says "you must use this exact word, or you are not counted in the numerical totals" nor with a rule that says "as long as the intent is clear, everyone is counted". Whichever it is to be, make it clear in advance, and clear to anyone who reads the page instructions. In any case, I don't see that this question has much to do with the words suggested as model exprtessions of choice. They can be whatever will be celarest, and then we either do or do not allow alternative phrasing if the intent is clear. DES (talk) 14:30, 29 September 2005 (UTC)

The existing wording already says that we're opting to Overturn or Endorse the decision, so it's redundant putting that word in again as part of the vote. One thing I'd like to avoid is this having too many extra words in it. The "biblical cannon" should be just a few lines, the rest can be from "tribal knowledge". Arguments about it are fine, play them out on the talk page the first few times, refer them to this discussion and that's it. For example, when I notice that someone has prefaced a clear opinion to delete on AfD with Redirect, I make a note underneath, "A redirect leaves the article in it's history, and thus may be counted as a keep, as opposed to "delete and redirect". That person never fails to amend their wording.  If the praxis is that only votes (I'm calling them that on purpose, please note) of the accepted form count, then that's what people will do.  If someone can't be bothered to follow those forms (especially if asked to do so) then I'm not that fussed if their opinion doesn't register.  In fact, I've changed "may" to "must" above! -  brenneman (t) (c)  23:04, 29 September 2005 (UTC)

Adding to "new information" clause. As I read the comments above, this is the intent. Septentrionalis 04:41, 30 September 2005 (UTC)
 * I've modified it a bit further. As I see it, the point is that when an overturn is justified by or supported on the basis of new info, it must be an overturn and relist, not overturn and some other action. new info which supports an endorsement of the closer's action, possibly to refute the arguments for overtuning, is of course a different matter. DES (talk) 18:17, 30 September 2005 (UTC)
 * I approve of that being made explicit. The section is starting to bloat, so I'd suggest we stop expansion of the section. -  brenneman (t) (c)  02:01, 3 October 2005 (UTC)

Rewording
Regarding the problem of the ambiguity of the word "endorse" (does this endorse the dispute or the original decision), I wonder if "maintain" might be a better word. Sjakkalle (Check!)  09:36, 4 October 2005 (UTC)

I have reworded the first two clauses of this in light of the discussion below. I haven't changed the operation of anything. The new wording makes explicit what endorse/overturn applies to, and creates a relist option to distinguish the langauge questions to do with what numbers were needed to change a result. Is this wording clear to everyone? -Splash talk 23:06, 11 October 2005 (UTC)

I have to say, with regret, that the new wording is not an improvement. I understand the concerns raised above, but we've said we want this process to be pretty "crisp". The new wording makes things less clear overall. - brenneman (t) (c)  00:54, 12 October 2005 (UTC)
 * I don't think so. Which part of it is unclear? I changed it more to deal with the concerns below than above. I really don't like the business of default actions associated with Overturning, because it will lead to newbie (and lawyer) confusion such as Encephalon illustrated below. We need either to say relist or overturn and relist. But relisting isn't necessarily overturning - if the article was kept, it's just a relist, there is no current change to the decision. Even if it has to be restored before relisting, we're not really overturning anything yet. -Splash talk 01:19, 12 October 2005 (UTC)
 * I can agree with what you're saying without liking the new version, can't I? ^_^ I'll try and knock up a re-word that I think meets in the middle. Quickly though, I'd still like to dictate the "vote" wording (e.g. must not should), and think that we could change to plainer speaking (e.g. Agree or Disagree). -  brenneman (t) (c)  01:40, 12 October 2005 (UTC)
 * Ok, see what you come up with, but do remember to allow for the 3rd way: relisting needs to be explicitly or not-very-implicitly available. I think that using words such as "must" is too dangerous on Wiki. It's one of the reasons votes are evil usually: they constrain the process to the options available when there may be other, better, options in a particular case. -Splash talk 01:51, 12 October 2005 (UTC)
 * An example of where a "must" wording would fail is the Ar15.com example currently on VfU. Clearly, the article should be kept deleted, but it is entirely within scope to bring it here on the nominator's grounds and for the result to be keep deleted, but unprotect &mdash; an option unavailable with the endorse/overturn phrasing when made compulsory. -Splash talk 02:00, 12 October 2005 (UTC)

Speedies
There is no reason speedies should not be subject to deletion review.
 * Pseudo-speedies {e.g. Deleted "Fred Flinstone" Content was: Fred's band The Rockboys have sold a lot of albums.) should come to DR.
 * Borderline speedies {e.g. Deleted "Fred Flinstone" as . Content was: Fred's band The Rockboys have sold a lot of albums.) should come to DR.
 * Clearly out-of-process speedies {e.g. Deleted "Fred Flinstone" as . Content was: Fred's band The Rockboys have three gold albums.) can still be restored, placed on AfD and a note placed on deleting party's talk page.  No need for DR at all.  brenneman (t) (c)  02:17, 29 September 2005 (UTC)
 * Speedies must come to DR; it's the only recourse (even RfC/admin won't get the article back). But I think even clearly out-of-process speedies must go first to the deleting admin's talk page and then to DR, for the sake of harmony. It might be possible to permit someone to undelete speedies while they are on DR, so that votes will be cast intelligently. (Note that claims of out-of-process speedying often do depend on article content.)Septentrionalis 02:27, 29 September 2005 (UTC)
 * I'm inclined to agree with this as my first choice. However, I can see the strength of the argument that speedies are about content over process and that AfD has better capability there. However, the content of speedies is frequently provided on VfU when it is needed. -Splash talk 02:48, 29 September 2005 (UTC)
 * There are enough admins roaming VfU to provide the content of the speedies. However, I'm worried that we might be overrun with users trying to undelete speedies, creating a de facto AfD at VfU. Maybe really improper DR requests (e.g. Deleted "Fred Flinstone" as . Content was "ASDFASDFASDFJKL'JKL'") should be speedy delisted? Tito xd  04:17, 29 September 2005 (UTC)
 * Reading that, I mean that we need to make sure nominators understand that the only way speedies can be listed here is if they DO NOT meet any CSD, not because they like the article or whatever. And to make it very clear. Maybe we should write it in the mechanics or instructions somewhere. Tito xd 04:19, 29 September 2005 (UTC)
 * I agree, speedies should only be in scope whn items were delted in a way that is defiantely out-of-process, becauae no CSD applies. Anyonme who brings an obviously valid speedy should be quyickly told to recreate it in an way that makes it not a speedy, and warned about WP:POINT just as usually happens now. (And just as now, there will be resistance to brigning here speedies that clearly don't fit any CSD, but equally clearly are bad pages that would never survive AfD. That shouldn't be a rule, but it is how people will act.) DES (talk) 14:36, 29 September 2005 (UTC)

If speedies were listed on Afd as a matter of course, there's your review, right? There's the advantage that people already use Afd. Anyone could browse and easily see a day's activities. I think this would get people to not worry as much about "common sense" speedies that don't fit a CSD, since the record would be plainly there for anyone to see. Things that turned out controversial could be undeleted and sent through a full Afd process. Friday (talk) 14:46, 29 September 2005 (UTC)
 * Of course, the main point of speedy deletion is to reduce the size of AfD. If you are suggesting that all speedy deletions should be automatically listed on AfD, I would object for this reason. If you are suggesting that speedies that someone considers invalid be listed on Afd instead of on DR, that is IMO more reasonable. I think it is asking AfD to do two rather different things, but that is possible. (Note that templates, categories, and images can bee speedied, as well as articles, but most problem speedies are articles). I don't see why this is an improvement over havign such come to DR, but if that is what people want, so be it. I note that they now come to VfU with some frequencey, and seem to be handled well enough there. DES (talk) 21:06, 29 September 2005 (UTC)

I remain concerned about the inclusion of speedies in the scope of DR as currently described. Speedy deletions are restricted only to the non-controversial and "bright-line" deletions. Any judgment calls must be handled through AFD. If someone oversteps on a speedy-deletion, their edit should be speedy-restored. We should not have to wade through 5-10 days of discussion to correct a mistake that took 5 seconds to commit.

If the article does not clearly meet one of the narrow speedy-deletion cases, then the article should have been nominated on AFD. The judgment call should be made on AFD and not here for two reasons. First, the discussion will consider content, not process. Second, the judgment call may well require expertise more widely than just admins. Remember that non-admins can't easily participate if we leave the article deleted during the discussion. If the speedy-deletion is being contested in good faith, the only way to give the article a fair hearing is to restore it and put it through the full deletion discussion.

However, we do need a mechanism so that non-admins can request the restoration of a speedy-deletion. I recommend adding the following to the page on the model of the "history only" section or the "content review" section of the current VfU page, both of which I assume will also remain under the DR policy. Rossami (talk) 22:10, 30 September 2005 (UTC)

Speedy undeletion
There are a few limited cases where admins can delete Wikipedia pages or media "on sight" and without the consensus-seeking processes of WP:AFD or WP:MD. If a speedy-deletion is contested in good faith, the article should be speedy-restored by any admin, and a full deletion discussion (AFD, MD, etc.) immediately opened. Non-admins may request speedy-restoration below:
 * Please restore foo. user:me

I've edited this, because, unless bad faith can be proved, speedy restoration and XFD listing should be routine. If the only text ever in the speedied article obviously meets the speedy condition (the text of foo is foobar foobar foobar....) a comment to the effect can be appended to the request. Septentrionalis 21:04, 2 October 2005 (UTC)

I'd oppose this inclusion strongly. Per my above comments, we don't need to say anything here about clearly out-of-process speedies, a protocol already exists for these (although it isn't followed very well). In fact I'd even support that if the original speedier objected, they could re-delete and move the AfD here. The vast majority of speedies are uncontested, the ones that are contested with any serious merit should go here, and stay deleted until such time as the discussion completes. That's why there is a content review section! brenneman (t) (c) 02:25, 3 October 2005 (UTC)


 * I have a question then, what is the nature of the discussion once a (possibly) out of process speedy is presented to DR? Is it whether or not it was properly speedied? Or is it an AFD type keep/delete/merge discussion? Rx StrangeLove 00:36, 5 October 2005 (UTC)
 * Good evening, Aaron. Perhaps this is the source of the confusion.  I (and I suspect others) read the text on the project page to supercede the existing protocol for out-of-process speedies.  It is not at all clear that the current wording allows for both processes.  In fact, it appears to clearly say the opposite - that all challenges must be brought to DR.  In addition, I have a serious concern about the presumption of your proposed process.  You argue above that a speedy-deleted article stays deleted until the discussion completes.  Everywhere else, we assume "when in doubt, don't delete."  Your proposed standard is exactly opposite our practice and philosophy in every other area of the deletion process.  I agree that the vast majority of speedies are uncontested.  Those which are controversial should be immediately restored and decided by the community.  Your last comment also confuses me.  You wrote "That's why there is a content review section!"  Not only don't I see a content review section, I explicitly see the instruction that "This page is about process, not content."  Please help me understand your intent here.  Thanks.  Rossami (talk) 20:32, 5 October 2005 (UTC)

(Restoring to indent=0)—As Rossami suggested on my talk page, we need a mechanism to delist proper speedies. How about if a speedy is listed here, and if within 48 hours it doesn't get anything but Endorse votes, it is delisted? Tito xd (?!?) 23:45, 11 October 2005 (UTC)
 * Simple question: what's the rush? Admin "A" goes drinks some bad pumpkin juice and deletes a crappy but salvagable stub.  Admin "B" brings it to content review and we have to wait five whole days to get this stub back. What's the harm there?  How many five day units have passed since wikipedia was born?
 * As to the "If in doubt" chestnut we aren't talking about deleting something, so that's a red herring. We're talking about restoring something, which has no such clause.
 * Admins make mistakes occasionally. Admins think other admins make mistakes all the time.  If there is a difference of opinon, let us have a deliberate and open forum in which to decide.  Or we could just let the wheel wars continue.
 * Thus bring all but the most egregious speedies to DR for the few days it takes to have something more than two sets of eyes on it. brenneman <sup style="color:#2f4f4f;">(t) <sup style="color:#2f4f4f;">(c)  13:58, 8 October 2005 (UTC)
 * Actually, that's exactly my point. What's the rush?  If it's a contested speedy, let the regular deletion process run it's course.  If it's really deletable, then the article is gone in five days.  There should not be a rush to delete.  Furthermore, we specifically need the article visible during the discussion because the review of a contested speedy is a content discussion, not a process discussion.  Content discussions are very likely to need the expertise and knowledge of the community at large, not merely the few admins who happen to frequent this page.  And, yes, it is specifically limited to those admins who are here.  Non-admins can't contribute to the content evaluation because they can't see the deleted text.  I agree that differences of opinion should be discussed in a deliberate and open forum.  It's a question of which forum is best.  I feel strongly that the better forum for a contested speedy is the AFD process, not this new deletion review process.  Rossami (talk) 04:03, 9 October 2005 (UTC)
 * The problem is that we speedy stuff so we can take some load off AFD'. Do we want to reintroduce articles that don't have a snowball's chance in hell to survive, just to clog AFD more? The other day we had over 200 articles on the thing, it's a nightmare just to try to load that page... Tito xd (?!?) 04:27, 9 October 2005 (UTC)
 * Good afternoon, Titoxd. You lost me on this one.  If the speedy is contested in good faith, the fate of the article must be discussed in open forum.  Whether that discussion occurs on AFD or DR makes no difference to the server load or, more importantly, to the burden on reader/editors.  I'm not sure how this decision makes any difference to "taking some load off AFD".  On further reflection, I'm going to revise that statement.  If we leave the article deleted during the discussion, we will actually increase the burden on servers and readers since it will take each person several more clicks and page-loads to investigate the deleted page than if the page is temporarily restored. I'm having trouble understanding the source of the insistance that speedies be evaluated through DR.  No, we don't "want to reintroduce articles that don't have a snowball's chance" but that's not the case with a speedy that's being contested in good faith.  No do I think that we will be creating a situation where we will see a sudden increase in frivilous requests to unspeedy.  The current process is that contested speedies go to AFD.  Very few of the total nominations on AFD on any given day appear to be the result of challenged speedies.  I trust the judgment of our admins.  They are not going to process a challenge to a speedy unless they think that the request is being made in good faith.  That's the way the system works today and all evidence seems to indicate that it works quite well.  The difference that I do see is in who can participate in the discussion.  If the speedy is temporarily restored and the discussion occurs on AFD, we can draw on all the expertise of the community.  If the speedy is not restored, then only a few admins can participate.  Rossami (talk) 20:47, 9 October 2005 (UTC)
 * And I'm not sure how it'll save time or take load off AFD, if a speedy is contested in good faith and brought here, the only discussion can be whether it was a proper application of speedy criteria. And since the default overturn result is to list it at AFD, they'll end up there anyway. I think it's quicker just list them there in the first place and let the whole community decide. Rx StrangeLove 21:59, 9 October 2005 (UTC)
 * Hmm. Sorry, that was a knee-jerk reaction on my part, because I've seen a lot of speedies brought here lately that were deleted clearly within the scope of the various CSD and they're starting to annoy me. I guess that what I don't want is users bringing here their speedied articles and clogging VFU/DR to the point where it becomes similar to AFD. However, that can be solved if we define what an improper or bad-faith listing is, and then speedy-delist them (and in more egregious cases, protect against recreation). Tito xd (?!?) 04:26, 10 October 2005 (UTC)
 * My concern is that we're still left with having to decide on speedy nominations twice (here and at AFD after legit, good faith listings are given the default overturn result.) and involving a limited part of the community. I think it makes more sense to streamline the process and just list them at AFD without them having to come here first. I think everyone agrees that a note to the admin that speedied the article in the first place should come before everything. Rx StrangeLove 15:58, 12 October 2005 (UTC)
 * The problem lies in the term "improper speedy" - it is POV. If an admin speedily deletes something, then obviously he believes it is in fact a proper speedy. Of course, he may be mistaken. But similarly, the person calling it an "improper speedy" may be mistaken. That means that in effect we have a disagreement between two admins regarding the speedy deletion criteria. This doesn't happen all that often, but when it does it would be best to list it here, not on AFD, so that we can form consensus on whether or not it was a proper speedy, and learn from it.
 * Basically, whenever an admin undoes another admin's action, one of them is mistaken. And it would be hybris to assume that it's always the other party who is mistaken. The proper way to go is either discussion or finding a third opinion - not simply overriding the other party, and not having the article disappear in the vastness of AFD. Radiant_ &gt;|&lt; 22:58, 18 October 2005 (UTC)
 * I don't see the term "improper speedy" used anywhere in this discussion, but in any case your comments don't speak to the two problems I have, as stated above. First we're deciding on some speedy nominations twice, at DR and then at AFD after articles are given the default overturn result. Secondly, given the fact that the pages are deleted, the discussion can only involve a limited subset of the editing community. And, as I said in the last sentence in my last set of comments, a note to the admin that speedied the page should come first. A discussion at that level should result in an agreement between the two parties in most cases, but we do need a process that makes sense for the times where the editors involved don't come to an agreement. Of course the proper thing to do is to discuss the issue and find outside opinions, the question is where it makes sense to do that. No one is automatically assuming someone is wrong or simply overriding a deletion. Rx StrangeLove 04:12, 20 October 2005 (UTC)
 * Indeed we can use the term out of process deletion to describe a deletion on which a single admin made a decision and another admin disagrees enough to make an exception and undelete and perhaps take it to AfD. I agree that AfD is a much, much better forum for this in every way. Speedies undeleted under the out-of-process exception are too rare to put a serious load on AfD. --Tony Sidaway Talk  04:43, 20 October 2005 (UTC)

So, where are we?
Has all work stopped on this proposal? Do we have any significant objections to it? Do we subject it to a vote (no, they're not evil, if consensus is acquired first and then a vote is done to validate it)?

On another subject, if we do the change, are we going to use subpages a la TFD? Because looking through the history is awful... Tito xd (?!?) 01:02, 4 October 2005 (UTC)
 * No, I've just been elsewhere. I mean to ping the talk pages of the originally involved users and ask them for their input here but it's a tedious job... -Splash talk 01:07, 4 October 2005 (UTC)

Just for the record, Titoxd and I have just pung the relevant talk pages. -Splash talk 02:21, 4 October 2005 (UTC)


 * How about we put the mechcnics as they are now up, call it a "test case" and only use it for two days before moving the whole lot here for dissection? At the end, all articles discussed get replaced on the main page as per usual.  Downsides are having to (potentially) state opinions twice and having tow run two systems for the last three days of the test. -  brenneman <sup style="color:#2f4f4f;">(t) <sup style="color:#2f4f4f;">(c)  01:21, 4 October 2005 (UTC)


 * We don't need to do that: we can just analyse the presently running cases, I think? It'd be annoying to send people round the block twice. Plus, short of actually instituting the new procedure wholesale, we'd be unlikely to attract any keep-challenges which are the really interesting ones. -Splash talk 02:21, 4 October 2005 (UTC)
 * Or we could just use the new mechanics, and keep their results as binding instead of relisting. Tito xd (?!?) 02:23, 4 October 2005 (UTC)


 * I always seem to swim against the tide. First I'm too bold, now I'm too timid! ^_^ I'm happy to have mechanics kick in, I put them on the main page ages ago.  brenneman <sup style="color:#2f4f4f;">(t) <sup style="color:#2f4f4f;">(c)  02:29, 4 October 2005 (UTC)

Phroziac's concerns
I like this proposal. But, it's rediculous (and has been already) to restrict it to process. Tony Sidaway has pointed out several times that the undeletion policy already says that it's not just about process. Additionally, I don't think we should put undeleted articles back on AFD. I feel this is unneccessary, unless there is a simple majority to undelete. If there's a rough consensus to undelete, just undelete it. --Phroziac(talk) 02:23, 4 October 2005 (UTC)
 * The direct undelete would happen if a supermajority voted to undelete (which is the case for most articles that come here anyway). Tito xd (?!?) 02:26, 4 October 2005 (UTC)
 * Ok. Sounds good to me. I will have to support this proposal. ;) --Phroziac(talk)[[Image:Flag_of_Phyzech_Republic.svg|25px]] 15:32, 4 October 2005 (UTC)
 * Ok. How would you handle decisions to delete a kept article by a simple majority? VfU works pretty well on process mainly. Of course, the article has to be considered to make sure the arguments have some substance, so it's not a completely bureaucratic process by any means. -Splash talk 02:33, 4 October 2005 (UTC)
 * Well, the system should always be biased to keep an article, unless there is significant consensus to delete it. --Phroziac(talk)[[Image:Flag_of_Phyzech_Republic.svg|25px]] 15:32, 4 October 2005 (UTC)
 * Yes, that's what I thought you'd say. I think Titoxd has basically dealt with your question already, much better than I did. The overturning to deletion wouldn't happen unless at least three-quarters of people were behind it &mdash; and as Titoxd observes most present VfU debates are pretty clear cut one way or the other. -Splash talk 15:57, 4 October 2005 (UTC)

DR mechanics: the third point
Hi all. From mechanics #3:
 * If there is a three-quarters supermajority, then the decision may be overturned directly and the consensus result applied. Failing that, if there is simple majority to overturn a previous decision the item in question will be listed on XfD as above.

I do not think this is a good idea. If I understand correctly, if in a given discussion 8 people say Overturn and 2 say Endorse, not only is the prior decision overturned, the reverse decision is applied. Consider the following, then:
 * WP:AFD/X
 * X kept. User disputes, brings matter to DR.
 * 10 participants, 8 Overturn, 2 Endorse.
 * Result: X is deleted (not relisted on AfD)


 * WP:AFD/Y
 * Imagine same scenario, similar article on closely related subject Y, similar dispute, but this time,
 * 10 participants, 7 Overturn, 3 Endorse.
 * Result: Y relisted on AfD.

What we're doing here is causing a qualitative difference in the disposition of articles, based on how many people think there was a process problem in the original AfD. This is a fundamental weakness: remember that DR examines only issues of process with respect to AfDs. We should not cause articles to suffer widely disparate fates based on a discussion that by its nature is constrained not to discuss the articles' merits, but rather the lawfulness of the articles' AFDs.

There will also be problems in practice. Lets say I happen upon a DR discussion where 70% of participants have already voted Overturn. I too agree that there was a procedural weakness in the original AfD; but I also think it was close, the consensus was not really clear, and the closer was wrong to make that decision, so I want it to go back to AfD. But I can't vote on that DR discussion without getting the article deleted. It's even worse if I happened to have voted early: would I have to stop the person putting in the vote that puts it over 75%, or pull out my vote when that mark is reached? I think we need to improve this. The problem is essentially arising because in this scheme, we're not fully respecting the intent of a vote.

I'm not expressing this very clearly, but what I mean is, if 80% vote just to Overturn, then the result should be Overturn (+default) (which is relist). If a User feels the article must be deleted directly, he must specifically say Overturn and delete (similarly for other specific actions, like merge [merges should also state a target article]). If an editor simply votes Overturn, but we later interpret this to allow Overturn and delete just because 10 other guys also voted Overturn, we're screwing with the process. None of them may actually want a direct delete, but AFD 2.

I also think that specific calls (e.g. Overturn and delete) must ideally be based on a clearly identified consensus for that action in the original AfD; the only reason that action wasn't executed in the first place was a process failure of some kind occured (e.g. a rouge admin ignoring a 9:1 consensus to delete). The reason for this is that I think we should be very conscious of the role of DR as an appeals court for process. By its very nature, DR is ill suited to examining content issues, because:
 * the participants are generally a relatively more select bunch, generally better informed about policy, but certainly not a good sample of the User community, and
 * we cannot see the articles (unless we happen to be part of the infamous evil cabal). (This, incidentally, is a good thing: I think it should be maintained because it forces editors to address process rather than peek at the article and perhaps get biased by the content). enceph  alon  15:45, 8 October 2005 (UTC)


 * I'm going to answer you in reverse. The concerns you express after the initial AfD/X and AfD/Y examples are rooted I think in the yellow box wording more than intent.
 * The 70% of participants example. If you vote to simply overturn, then you are asking for a relist, rather than any other result having not specified another result. This is one reason I don't like the exact wording of the yellow box too much since it leads to conditional thinking like this. However, the point remains. Unless you say "overturn and delete", you are not mandating the deletion of the article and will in fact be moving things further from the 75% threshold you would prefer not to be reached. At the smae time, you are completely expressing your desire to have the article return to AfD. I don't think there's a problem with this example. -Splash talk 16:59, 8 October 2005 (UTC)
 * The 80% example. Same answer as above. The wording says that to mandate an alternative outcome, the editor must say so. If 80% say "overturn", it gets relisted unless they say "overturn and delete". We should change the wording in the yellow box to reflect this fact. -Splash talk 16:59, 8 October 2005 (UTC)
 * Yes, you are right on the final point. It is currently the state of play in VfU and I don't think we're seeking to change that. It is not a way to second-bite the cherry, and any responsible admin ending a DR debate should not include votes that are plainly dealing outside DR's remit. We frequently get such comments from irate webcomic owners here, and they do not at present disrupt the outcome, even if they do entangle the debate. -Splash talk 16:59, 8 October 2005 (UTC)
 * I don't honestly see the problem in the X-Y example. If Y returns to AfD and is kept, then so be it. Articles stand or fall on their own merits, usually, comparisons to Pokemon aside. I think, however, that when Y returned to AfD it would get soundly deleted since actions such as keeping on 80% reasonabley-argued delete does not generally go down well in the relisting debate. In the end, the outcome would probably be the same. If it weren't the same, I'm not sure that anything is necessarily broken. -Splash talk 16:59, 8 October 2005 (UTC)
 * Thanks for replying, Splash. I knew that X Y example wasn't the best. Yes, I quite agree with you that articles are judged independently. My concern here is this: if we had a DR for a previously kept article that went
 * Overturn Splash
 * Overturn Aaron
 * Overturn Tony
 * Overturn Dpbsmith
 * Endorse Encephalon Tito :)
 * this should not lead to the article getting deleted. The wording "If there is a three-quarters supermajority, then the decision may be overturned directly and the consensus result applied" seems to imply that as long as >75% of participants vote for an Overturn, not only is the previous decision overturned, but the opposite action is directly applied. This is worrisome, because if the first four editors wanted it to be relisted, they'll be disenfranchised, as it were. There is no possible way to know in advance what the final number of voters would be and what the percentages would be like; and in any case the principle objection is simply that their votes are being used in ways not specified by them, purely depending on how others voted. We cannot change the intent of a vote in that way. The X-Y example was intended to illustrate a related problem: the disposition of an article could depend on a random interpretation of votes.


 * Having said this, it's quite clear to me from the answer you provided in the next section that you're not interpreting #3 in this way; ie. we're both interpreting it the same way. But I don't think we can say that for every one of the discussants, so a clarification of the intent of #3 is in order. Methinks, anyway. :) enceph  alon  17:53, 8 October 2005 (UTC)
 * You probably won't see a decision like that because I spend a whole lot of time at VFU and I try to drop my two cents in in every discussion I see... :P But aside of that, you make a very valid point. The only way that a keep should be deleted outright is if the participants vote Overturn and delete. Or maybe, we could clarify it a little bit more: You could have Overturn votes for those who really want the decision to be reversed (as many would want if a 9-1 consensus were ignored), Relist votes for those who feel that the discussion merits relisting due to procedural problems, and Endorse for those who, well, endorse the decision. Tito xd (?!?) 18:03, 8 October 2005 (UTC)
 * Why yes, Tito, you're absolutely right! I've made a change to reflect that :). As to the votes, basically I think it should be:
 * >50% Endorse → endorse.
 * >50% Overturn (whether or not overturn votes are coupled with a more specific action)→ overturn & relist
 * >75% Overturn & (action) → perform action
 * Regards enceph  alon  08:19, 9 October 2005 (UTC)
 * That's what I think too about the numbers, so it seems we all agree on this. Tito xd (?!?) 08:26, 9 October 2005 (UTC)

DR mechanics: the numbers
What will we do if the DR result is We need to think about these things. I can see ways of making it work, but would love to hear suggestions. I am a bit worried that the system might have to be a bit complex to work well. I want it to be as simple as possible, so that any editor with a problem with an AFD result can walk in here and participate to the fullest. enceph alon  15:45, 8 October 2005 (UTC)
 * 1) 50% Overturn, 50% Overturn and delete?
 * 2) 45% Overturn, 55% Overturn and delete?
 * 3) 33% Overturn, 33% Overturn and delete, 33% Endorse?
 * 4) 25% Overturn, 25% Overturn and delete, 50% Endorse?
 * Relist, majority to overturn, not 3/4 to change result
 * Relist, majority to overturn, not 3/4 to change result
 * Relist, majority to overturn, not 3/4 to change result
 * Nothing, no majority to overturn.


 * These are exactly the scenarios in which it is intended to relist because there is a feeling that something was wrong, but no clear feeling that it was so wrong as to directly reverse it. This does highlight a problem with the wording of the votes though; we shouldn't insist on a particular phrasing as has been intimated in places above. -Splash talk 16:25, 8 October 2005 (UTC)
 * Those are my answers too. So #3 in DR mechanics needs to be changed, yes? enceph  alon  17:53, 8 October 2005 (UTC)

DR mechanics: the new world
What's the one great big change we're making by dissolving VfU and creating DR?

We're accepting disputes over an entire new class of close. We're accepting disputed keeps. I haven't fully thought about the ramifications of this, and although I believe it's a good and fair idea, I think we should all think about a class of problem that has arisen in the past and how we might handle it fairly.

WP:AFD/B What will you say? enceph alon  15:45, 8 October 2005 (UTC)
 * 8 Del, 7 Mer, 2 Keep. (all merges agreed on one target article)
 * Closer says: Debate closed as Keep.
 * Disgruntled User brings it to DR. It's definitely not a keep, he says. A total of 15 people said it should not be kept. This is a clear violation of process, he says.


 * This I presume arises from the fairly detailed 'study' that took place in Tony Sidaway's RfC. There are two schools of thought:
 * AfD is binary in its mission to either delete or not and that is all that is binding on the closer. Anything that does not speak to those two options is guidance on what should be done with a kept article, but is not a required action.
 * AfD is M-ary where M is the number of different options AfD editors come up with. Any outcome they mandate is as binding on the closer as delete or keep.


 * Personally, and it's personal, I prefer the second approach since if someone has said "this" is what should be done, I see no reason not to do if the consensus follows them. Clearly, in your example there is no consensus on a single action. But it's not at all difficlt to see that almost everyone didn't want the article to stand alone any longer but that there was not sufficient support for discarding the content outright. Consensus indicates, as does common sense, that consensus is, in part, those who didn't get their way being able to stomach the result they did get. So we can please, to some extent, 15/17 by merging. We can alternatively please, to some extent, 9/17 by keeping, although it is entirely reasonable to suppose the mergers are happier with a merge than a keep, seeing as they said as much. I would prefer to please 15/17 reasonably well, than 9/17 people less than reasonably well. Anyway, this is off-topic, but:


 * So what should DR do? Depends on the DR editors' interpretation of AfD. I personally would advise User:Disgruntled Editor to seek a merge result for the reasons I described above, which DR will only give effect to as far as amending the result on the face of the AfD and adding a merge tag to the articles in question. -Splash talk 16:40, 8 October 2005 (UTC)
 * I interpret it that way too (ie. approach #2). This question was meant to stimulate some thought over the sort of discussions we might have to have if DR goes through. I can forsee a User:Disgruntled getting his merge decision from DR, and Closing Admin telling us all to go bugger off please, because DR has no right to impose its opinions against one possible interpretation of closing procedure. I guess I'm trying to foresee problems we may face down the road; if we can produce something that minimizes the unhappy conflicts we wikipedians sometimes face, all the better. :) enceph  alon  17:53, 8 October 2005 (UTC)
 * This example is not a change in policy; but an instance of needing to be absolutely clear:
 * It is current policy to merge in this example. Some admins choose (correctly, strictly speaking) to call this Keep, and do not  do the merge; because the merge is an ordinary edit and it ought to be done by someone who understands the subject, like the AfD voters. If this is changed, AfD-talk is the place to do it. Septentrionalis 18:01, 8 October 2005 (UTC)
 * Being a devil's advocate is fine, Encephalon, I've been known to do that a lot myself. In this case, I agree with Splash. All I would do is go to the AFD and amend the result to merge to Article2 and add the respective merge tags. I would also explain that a merge result is technically a keep. Tito xd (?!?) 18:13, 8 October 2005 (UTC)

I'm such a pest, aren't I? Advocating for devils. :) The point of this section was to serve as a heads up to the kinds of discussions we might be having with the new DR. There appear to be two broad interpretations of how an admin can (correctly) close the above AfD scenario. That there appears to be more than one way to do it is a reflection of the current deletion policy: it does not specify how an admin must close the above. There are some admins who'll simply disagree with Pmanderson's statement that "it is current policy to merge in this example", and take the view that they are required only to not delete; everything else that is done is the province of the article's editors. Some disagree with that view, and encourage a merge decision to be explicitly stated in the close; they think that the way the decision is often written—KEEP—is, if anything, a misleading summary of the AfD discussion, and furthermore that the decision to not recognise the consensus not to keep (which is actually the only real consensus in such debates) does a disservice to editors who have, rightly or wrongly, used AfD as a platform to form a consensus over the article's future.

When we open DR to disputed keeps, this is one species of AfD closures that we can expect to see—my intent in posting this was simply to make this apparent. As DR is a venue for decisions about process, it may legitimately be held that we should not return a decision to change the AFD closing decision from Keep to Merge in instances such as the above, since that is apparently one way to legitimately close these AfDs. (Just as, for example, we will not vote to undelete a 7D-3K decision, despite some editors' belief that a delete consensus is only formed at 80%, because we recognise that a delete consensus at 67% is also legitimate way to close AfDs). Am I making any sense? enceph alon  08:48, 9 October 2005 (UTC)


 * In fact, I endorse the "binary" version (with slight modifications, "transwiki" for example doesn't fit in nicely), and I think that if anyone makes a dispute that the result should have been "merge" rather than "keep", they can be answered by saying "Look, if you want to merge this, just be bold and do it".
 * The fact that I endorse this "binary school of thought" does not mean that I always close debates as "keep" or "delete", but that I view the choice between keeping and merging as an editorial one, i.e. one where WP:BOLD does apply. (For example: "I see 10 votes to keep, 3 votes to delete but I will be bold and delete this anyway because I think the article is garbage" is usually not acceptable, but "I see 10 votes to keep, 3 votes to merge, but I see that this article is a stub and that all other such articles in the series have been merged, so I will be bold and merge this one too" is usually quite acceptable.)
 * In general, I think we might well respond to disputes regarding merge vs. keep the same way we handle VFU nominations (usually by newbies) which protest that an insertion of three paragraphs into an article was reverted away, just politely point out that this is the incorrect forum for such complaints and that such discussions, if neccesary, can take place on the article's regular talkpages. Sjakkalle (Check!)  10:15, 9 October 2005 (UTC)

I have added text encouraging nominators to vote (usually Overturn). The nominator vote is sometimes ambiguous; but the chief benfit of this is making even clearer that Overturn = agree with nominator. Splash's new language should be a great help. Septentrionalis 19:43, 12 October 2005 (UTC)

Templates
Here I go again, but we need to figure out what we're going to do with the templates for VFU/DR. I'm picking on this because I got into a small revert war with Mysidia over templates and CSD pages with just templates...
 * Ok, we have Deletedpage, Vfu and TempUndelete. So what I propose is to use TempUndelete for speedies under review, Deletedpage for pages with significant recreation/redeletion histories, and Vfu for deletion discussions that are revisited.
 * On a separate point, we need subpages. The TFD method seems to work best for us, since we don't need to do so many archivals (hopefully) as AFD does. So, a hierarchical system is best: /log/overturn/kept, /log/overturn/deleted, /log/relist, /log/endorse/kept, /log/endorse/deleted and /log/others (in case any other permutation escapes my mind) would be a decent approach. Any thoughts? Because if we're going to do a massive change (when are we going to do it anyways?) then we might as well do it all at once. Tito xd (?!?) 05:59, 12 October 2005 (UTC)
 * I don't see a need to parse them. I recommend putting them all in one log for simplicity.  If/when the log gets too big, archive it by time.  As long as we're going to sub-pages for discussions, the log can be a simple list of links, not transclusions.  Rossami (talk) 13:29, 12 October 2005 (UTC)