Talk:Deletion management redesign

From Meta, a Wikimedia project coordination wiki

look.

the first problem is that i cant tell within about 2 minutes of reading what exactly 'voted for deletion' means. and how the hell is it a 'vote' if there is no voting process, just a small group of people posting their opinion which has no bearing on anything because an admin can click delete if they feel like it, had a bad day at the office, whatever.

if you are going to have a 'government' of wikipedia, the first rule should be that it is transparent (ie easy to understand. no mumbo jumbo or complicated garbage that newbies have to wade through)

the second rule should be that it is democratic.


one poissbility is to have massive voting on deletion. put 'things to delete' on front page. or something. where masses of people can see it and vote.

but the details are not the important part. the important part is the principle. and the principle apparently is abandoned, because people think that oh, administrators dont need any checks or balances, there is nothing wrong with throwing the principle of democracy out the window, etc etc etc.

now , as far as i can tell, this 'administrator' thing is a stupid vanguardist elitism bull-caca crapshoot inserted into wikipedia by somebody who was acting on fear or anger.

actually you know what, i dont care. in the long run, wikipedia if it goes down in flames from tyrrany within itself, will just have someone else come up with a similar idea to wikipedia but more democratic. it doesnt matter that much. but wouldnt it be nice if that was now instead of 50 years from now?




One current deletion management[edit]

Currently, the deletion log displays in order

  • Time of deletion
  • Author of deletion (link available to the user page)
  • Title of the article
  • Comment

All types of pages are listed at the same place, whatever the space (except images)

The Undelete page list all articles, whatever the space, with a rather mysterious order (supposingly alphabetical) with

  • name of the article (with a link)
  • number of revision (interest ?)


When we click on the link, we see the

  • date of deletion
  • Author of deletion (link to the user)
  • Comment
  • history of the article (link to article version)
  • Restore button

To find an article, we need to first copy the link in the deletion log opened in the first window, open another window and wait a long time till the page is loaded (charge on the server), do a find this chain to find the article, click on the link, then check the article, undelete, do two backs to go back to the undelete page for another article.

There is also a query available to make this quicker



Maveric149 stated in a comment that "The first 150 characters of any deleted file is ... auto-inputed into every deletion comment". This triggered off the following tangential technical discussion, which I have separated out from the original thread to preserve the clarity of both: GrahamN 13 July '03
I initially created that page with the goal of it being wikipedia-wide. Not english centred. I would appreciate that it being written in english does not mean it is only for english, but rather to help communication, and that though basically all software changes are decided in the aim of solving the english wikipedia issues, the issues arising in small wikipedias are not totally forgotten.
At some point, I had the crazy dream that finally we were all gonna be on the same "software footing" :-) I see that not only is it not true (as some small international wikis are still in phase I), but we are regressing again, as even the 2nds "big" wikipedias are behind updates as well. Slowly drifting again.
I know it is not your fault Mav, I know it is nobody's fault, but time and energy, but please do try not to generalize everything on regards of what *is* currently available on the en.
For example, remember that it is extremely difficult for us to do anything on meta because we do not benefit the link w:. Similarly, the "first 150 characters auto-inputed..." are available *only* on the english wiki.
It is available in the most current version of the software. It isn't my fault a developer hasn't had time to upgrade the other wikis yet. The whole point of this page is to talk about a redesign. Well part of that redesign has already been implemented. --Maveric149
Like I said above, it is not your fault yes. No one fault. I think that part of the redesign was never mentionned on this page. My point is that sysops of different wikis don't see the same thing, since the deletion is not working the same way. And it is hard to discuss improving something which is not perceived the same way :-)
OIC. But at the same time everybody should know what the most current version of the software does - otherwise we will find solutions to problems that have already been solved. --mav
What does that mean OIC ??? :-)
You are quite right. However, unless I am wrong, the "paste" stuff was announced only on the english mailing list [1] and on the english announcement page. Not on the tech list and not on wikipedia main list. Hence, it was announced only as an english improvement. I know this announcement because I read the en list, but you cannot expect everyone (not even sysops) to read the en list. What do you suggest for everyone to know about the improvements if improvements are only announced on a language specific list ? Even if an improvement is initially planned to be accessible for english only, would not it make sense to announce it globally and to say "sorry, english only first, others will come as soon as we have time for update" ? That was what the global list was supposed to be for. Have a good day Mav :-)
OIC = "Oh, I see" (cute, eh?) It doesn't matter when and where it was announced. The only thing that matters is that it exists. Our discussion here makes the current state of the software clear and you could update this page, Wikipedia NEWS or the project-wide mailing lists in order to inform everyone. I don't know why the changes were only announced on WikiEN but I do know we only have one code base and as soon as the other wikis are updated the new features will be there too. --Maveric149
Very cute :-). Right. i'll update it when I have time.

Is a free-for-all, pure wiki deletion system practical or desirable?[edit]

Is there actually a need ever to delete anything?. Couldn't things be moved to into some kind of perpetual WikiRubbishBin, where any user can still see them? I suppose copyright paranoia might militate against this being applied to absolutely everything, but I can't see the harm in keeping copies of uncopyrighted things. GrahamN 14:56 18 Jun 2003 (UTC)

The same reason most people clean their bedroom once in a while instead of constantly throwing trash in a corner. Not completely removing graffiti, vandalism and other garbage only encourages the creation of more of the same - even it is otherwise de-linked from the normal database. The first 150 characters of any deleted file is already auto-inputed into every deletion comment. IMO that is enough. --Maveric149

The reason most people clean their bedroom once in a while instead of constantly throwing trash in a corner is because after a while the pile of trash gets in the way and starts to smell and make the place generally unpleasant. I can't see how anything analogous to that would happen here. What it could do is reduce the amount of work that administrators would have to do, as well as making the whole process more transparent. I'm no computer expert, so please bear with me here. Suppose that articles could be marked by any contributor as being "Rubbish" (or "Trash", to you). Links to rubbish would appear like links to non-existent articles, except that when you followed the link, the message that appeared would tell you that the article used to exist but that it has been binned. All users would be able to access the article's full history and talk page, etc., and they would have the right to remove it from the bin again. Any "binning wars" would be resolved in the same way that edit wars generally are. "Binned" articles would not disappear from people's watch-lists, but show up as having been binned. I don't know if this is possible, but it seems to my simple non-technical mind hardly more complicated than many of the apparently miraculous features that have been developed and implemented here already. What do you think, mav? I do respect and value your opinion, honest, despite our recent differences. GrahamN 15:04 20 Jun 2003 (UTC)

Or we can do as we are now trying to do and have a whole bunch of Admins - basically anybody who edits Wikipedia regularly (all we need from you is your consent and you will be "one of us" :). It is still important to make sure random junk is wiped from the database - otherwise a great many edit links will have useless histories that dozens of people will check and recheck just to make sure some vandal hadn't "binned" an actual article. However, I am not in opposition to changing the behavior of how blanked pages are displayed so that they appear as edit links but Admins should still have the ability to delete page histories that have truly useless content. There are a whole bunch of 0 length pages that have no business showing up as valid article links so you are right in saying that your proposed system would reduce Admin work since history deletion will no longer be that important of a job (we could even, perhaps, have the database automatically delete edit page histories after say 90 days of being an edit page...). I respect your opinions too - if I didn't I would not have nominated you to become an Admin. --mav

But what you are now trying to do results in bitterness, suspicion, divisiveness and potential abuse of power. Your vision seems to me to be of an elite corps of autocratic vigilantes. That dozens of people will check and recheck just to make sure some vandal hadn't "binned" an actual article seems to me to be an excellent thing. GrahamN 25 June '03

I'm with you Graham, at least in principle... :) --mrd

I've just understood what mav meant about "changing the behavior of how blanked pages are displayed". This is an excellent idea. It simplifies the whole thing. There would be no need for a separate "binning" operation - all a user would have to do to "bin" an article would be to blank it. If links to blank pages looked like edit links, but otherwise blank pages behaved exactly as they do now, that would be all that would be needed, wouldn't it? GrahamN 18 July '03

There still would be no need to have junk or copyright violations in the edit page history. Simply having the history exist is going to cause people to check it to make sure somebody else didn't "delete" valid content. That is wasted effort and bad database design. --mav
You might say the same for any past revision of any page. The onus is on the person who blanked the page to supply a decent summary (eg "blanked - previous version was Spanish, moved to Spanish wikipedia"). If the blanking person is trustworthy, then it's very unlikely to be checked. If the blanking person is a known vandal, then it's going to be checked carefully. All quite normal, and the standard way these things work.
The "wasted effort" you fear will be counter-balanced by efficiency savings because:
* no need to list a page on VfD before blanking
* no need to list a page on VfU if it is incorrectly blanked
* no need to list a page on VfU if the author wants to retrieve the text
* no need to move content elsewhere in case the author wants to retrieve the text
* anyone can make a page behave as if it's been deleted, not just admins (IE, show up as red, not appear in searches, etc)
* Anyone can make a page behave as if it's been undeleted, not just admins
It's actually very good design: it follows the design principle of avoiding special cases - by treating deletion as just another edit, you substantially simplify the software. --mrd
? What? User A blanks a valid page and it is missed in the sea of 5,000 edits for day. It shows up as a red link. User B either sees it is a red link from another article and ignores it (all the effort of the people who worked on that article is wasted), follows the link and creates a "new" article (not noticing the history link; more wasted effort) or user B checks the history (and if the only content in the history is complete rubbish then that action is wasted effort).
The same applies if User A takes a good article and delete it all except the first paragraph. User B sees that it is a stub link and ignores it. User C wastes time creating a "new" article from the stub. Etc. But that's not a problem. I don't worry that my articles will be replaced by stubs, and I don't worry that my articles will be blanked. The Wiki process takes care of it. There should be a D in recent changes/watchlists when an article is deleted/blanked in this way as for N and M - that's all that would be required. --mrd
Blanking and unblanking can already be done by anybody. Admins (sysop is a user account switch not a person) simply have the added ability to de-link the history of an article from the main database.
I've edited my list to be clearer on these points. --mrd
I'm beginning to see negative aspects of the blank=red link idea and hereby withdraw my support for the idea. Making things more transparent is a good thing though; regular users should have the ability to see the content/history of deleted articles via links on the deletion log for say, a week. Then they can vote to have things undeleted. But really, all this is an attempt to cure an illness that doesn't exist - where is the deletion abuse? --Maveric149 04:08 19 Jul 2003 (UTC)
There are instances of mistaken or incorrect deletion. There are instances of controversial deletion. They're not "abuse", but they do happen. But I'm more interested in the efficiency, scalability, and decentralising benefits of scrapping votes for deletion. --mrd
So instead of having cases of mistaken or incorrect deletion only occur amoung our most veteran of users who are trusted we allow anybody the change to make these mistakes? How is that efficient? VfD is very important because it is centralized place where we, as a community, can decide just what is and is not deleted (whenever there is any question). --mav
I beg to differ. Centralised control seems to me be completely contrary to the spirit of the project. Wikis are naturally de-centralised. Anybody can make changes. Anybody can revert those changes. If you are not convinced that this is an efficient system (with other highly desirable benefits too), may I suggest you take a look at Nupedia. They employ the bureaucratic centralised system you are advocating. All decisions are made by committees of "trusted" "veteran" users, not just deletion decisions. As a result, Nupedia seems to me to be considerably less successful than Wikipedia. Frankly it also seems rather dull. GrahamN 25 July '03
the red link could lead to an empty page, which would be different from a regular new empty page. With an indicator the page has history for example, as a warning. (anon)

Discussion moved from en:Wikipedia talk:Deletion policy:

It seems to me that if you deliberately tried to design a system that would make consensus decisions as difficult as possible, you could hardly do better than our curent system. I consider the "Votes for Deletion" system to be an anachronism. We don't have a "Votes for Re-naming" page, a "Votes for Blanking" page, or a "Votes for This Particular Edit That I Would Like To Make to The Manic Street Preachers Article" page. Consensus happens everywhere else on Wikipedia using that rather clever thing, the "Wiki". It may not be a perfect form of consensus - bullies tend to get their way rather more often than we might wish - but it is how Wikipedia works. I can't understand why we treat deletion decisions differently from any other kind of decision. A simple software tweak would make links to blanked articles appear the same as links to non-existent articles. Anybody could then "delete" or "undelete" any page by simply blanking it or reverting it. With a few other minor software tweaks, deletion decisions could be then made by exactly the same brand of consensus that is used for every other decision here - the brand of consensus that has made Wikipedia such a success. Please see meta:Talk:Deletion management redesign. GrahamN 20:36, 7 Oct 2003 (UTC)

A simple software tweak would make links to blanked articles appear the same as links to non-existent articles.
I think this is the heart of a very good point. There might be a wrinkle to sort out with regard to copyrights (i.e. sometimes we might have to 'hard-delete' to stay legal, need to ask the legally minded about this). I think we should also check that carrying a certain amount of blanked junk around is not going to add more strain on the server than is currently the case. With these points in mind, I support GrahamN's proposal. Pete 12:47, 9 Oct 2003 (UTC)

Pros and Cons of Graham's proposal (please add/edit this) Terminology: 'Soft delete': Blanking of page, links to page appears as anon existent page link 'Hard delete': Deletion as it is carried out now, only sysops can delete/undelete

Pros:

  1. Because soft deletion can be trivially reverted, it is less of a big deal. All users could do it, dramatically increasing openness and democratic nature of WP.
  2. Decisions to soft delete could be made the same way as all other wikipedia decisions... unilaterally if the change is 'obvious' or through discussion on the talk page if more tricky.
  3. No need for an overlong Votes for Deletion page that often provides more heat than light.

Cons:

  1. More potential for damage? Vandals who decide to blank pages wreak slightly more havoc as they cause red links to appear on other pages.
Potential mitigating factor: Vandals rarely get away with their dastardly work.
  1. More strain on server, with more pages for it remember/linkify
Potential mitigating factor: This may be a non-issue as the server has to remember hard-deleted pages already in case they need to be undeleted. Expert technical advice required.
  1. The disappearance of VfD would amount to the loss of a real community page where lots of users interact, and get to hear about recent developments.
  2. May add extra strain to another oft-edited page, the Village Pump, as people rush there to say... blanking of X is being considered... please come and support/deny it at the talk page... maybe we'd end up with a pages where blanking is being considered page which could (given how built-in the VfD way is) become a de-facto VfD
  3. Some developer work required.
Mitigating factor: Might not be that much work through. Given that we have a Wikipedia:Blank pages page it looks like the facility for detecting blank pages is already in code.

Pete 13:57, 9 Oct 2003 (UTC)


Two more cons:

  1. I believe blanked pages show up on searches (though I haven't tried it). We don't want this anyway, but also if the consensus is that a page should not exist, having it show up in searches encourages people to recreate it.
  2. To add to the community argument, a lot of discussion on VfD gives people a good idea of what should or should not appear in Wikipedia.

207.236.234.180 14:34, 9 Oct 2003 (UTC)

Did I mention I agree with Graham? What's needed is someone to just implement it. I'd do it myself, but I have other itches to scratch at the moment... --mrd


I think the "pure wiki" deletion system discussed above is not a case of deletion management redesign, because if it works it will require no management. (There is no need for "editing management" or "re-naming management", so I don't see why we need "deletion management"). The proposal is for a re-design of the whole deletion system rather than of the way it is managed. It's also not very visible here, so I'd like to set up a separate page for this proposal, at Pure wiki deletion system (proposal). I don't have much free time, however, so I'd welcome constructive assistance from any quarter. (Including sensible suggestions for a snappier title!) Thank you. GrahamN 15:41, 14 Oct 2003 (UTC)

I have posted a critique of GrahamN's proposal on the same page. -- Cyan 17:37, 14 Oct 2003 (UTC)

Risks of "big bang" change[edit]

I'm very uncomfortable with the concept of making a big change to the deletion system, for a couple of reasons. For one, there's what I call the "grass is greener" problem - we are (painfully) familiar with the weaknesses of the current system, but can only guess as the problems a new system would have - and in any case we haven't had to endure them. Second, I think there are probably other things that are a lot more important for us to be dealing with now, such as the en:Wikipedia:Forum for Encyclopedic Standards stuff. We can't afford to divert our focus on this - and moreover, we don't really need to, because the current deletion process has, after some tweaking, gotten reasonably good. It's basically democratic (anyone can vote on VfD); admins basically don't abuse their powers and do rogue deletes (and can be hauled in front of ArbComm if they do); we've got VfD working somewhat better; etc, etc. However, in the spirit of "incremental improvement", there are some changes we could make, ones I think most people support. In particular, I think we should allow everyone to see deleted pages, except those deleted for some serious reason (copyright, defamation, etc). Perhaps there are some other tweaks we can borrow from the various proposals, but for the reasons I started with I believe any change should be incrememental. Jnc 02:50, 13 Dec 2004 (UTC)