User talk:Balloonman/CSD Survey

In at least the first two surveys there are multiple places where the radio button text states G10 when what is meant is G1. Also, rationale is misspelled in multiple places.--Fuhghettaboutit (talk) 08:53, 27 December 2008 (UTC)

Thanks, I took care of the G10 issue as that is potentially key, I'll try to fix the spelling tomorrow... that's the problem with cut and paste... one mistake is repeated over and over and over again.---Balloonman ^PoppaBalloon 09:16, 27 December 2008 (UTC)

• Took all 4 surveys, nice idea, will be interesting to see the results. Davewild (talk) 09:31, 27 December 2008 (UTC)

Thanks... I'm interested too...---Balloonman ^PoppaBalloon 09:45, 27 December 2008 (UTC)

• Just a quick comment -- if none of the 4 options fit and you fill something in under "other", the survey still makes you go back and choose one of the four options.--Fabrictramp | talk to me 13:35, 27 December 2008 (UTC)

One of the four should fit more or less---eg you keep, delete per rationale, delete for another reason, or delete IAR.---Balloonman ^PoppaBalloon 17:45, 27 December 2008 (UTC)

Like Fabrictramp, I found a number of places where none fit—not even more or less—and where that occurred, I stated so in the place allowed for text in the other column. I imagine others will do likewise. The results will be skewed if the main use will be a simple graph showing who picked what for each question.--Fuhghettaboutit (talk) 21:03, 27 December 2008 (UTC)

speedy delete

You ought to more carefully check the word "delete" throughout the discussion , by which you generally mean "speedy delete" You are inadvertently perpetuating the impression that there is n difference between the two. "Disagree with rationale to speedy delete, but deletable by other criteria." should Disagree with rationale to speedy delete, but deletable by other criteria. "Disagree with rationale to speedy delete, but speedy- deletable by other criteria." DGG (talk) 21:27, 27 December 2008 (UTC)

On-wiki surveys are better.

This would be much more useful if it and the results were on-wiki. See User:Filll/AGF_Challenge_2_Directions for one way to do this. davidwr/(talk)/(contribs)/(e-mail) 15:36, 27 December 2008 (UTC)

I don't like that option. I've done a number of "surveys" (ok actually reviews) previously and asked for feed back. The problem is that people might change their position when they realize that they are alone. By having it in a true survey format, you ensure that people are giving their answer, not the answer of somebody else. (Now granted the question is, do you trust me to report the numbers accurately?) What amazes me is that there are only about 3 questions where everybody has answered the same thing! I won't say which one's because I don't want to influence the results.---Balloonman ^PoppaBalloon 17:43, 27 December 2008 (UTC)

The reason I like public surveys is precisely because it's easier to share the answers with everyone. BTW, your survey made it sound like if you filled in "other" you didn't need to click any of the options. You did. Consider adding an explicit "other:" option OR clarify that the comments apply to whatever option you actually choose, and are not an "other" option. davidwr/(talk)/(contribs)/(e-mail) 18:03, 27 December 2008 (UTC)

The problem with "sharing answers" in this manner is that it biases the results. There is a reason why the results are not released during presidential elections until the balloting is finished.---Balloonman ^PoppaBalloon 19:11, 27 December 2008 (UTC)

First off, Balloonman, I want to thank you for doing this. You've obviously gone to quite a bit of effort, and this is exactly the sort of thing that I encourage whenever anyone proposes a major policy change or review: understand the problem before looking for solutions! That said, I'm now going to poke some mean little holes in some of your hard work. (Sorry 'bout that.) While I appreciate the value of having the survey off-wiki, with confidential responses, there are also potential pitfalls and biases that may creep in. (For those who are curious – and in the interests of full disclosure – I didn't participate in the survey; I didn't get around to it.)

Others have noted above a preference for the opportunity to see the answers of other editors. Balloonman has noted that this might be a problem where "people might change their position when they realize that they are alone". I don't know if that is a bug or a feature; it depends on what one wishes to measure. Keeping the answers of respondents hidden means that you will get the 'first impression' of each respondent, but it also means that each respondent must 'reinvent the wheel'. One ends up testing not only whether or not the deletion criteria are applied correctly, but also how familiar each responder is with deletion (and other) policies and their correct application.

Where one observer might see a 'lone responder pressured into changing his vote', another observer sees 'someone who misread the criteria or was unfamiliar with the context and realized his mistake'. An article might be an 'obvious hoax' only to a particular admin who first saw it created on a forum somewhere. If you're worried about badgering of other editors, an alternative solution would be to forbid threaded discussion; a strictly-enforced word limit on a per-respondent basis would probably also do the trick. It also occurs to me that if a large enough preponderance of editors take a side that dissenters are intimidated, it may well be that the supermajority are in the right. TenOfAllTrades(talk) 16:34, 2 January 2009 (UTC)

Great survey

I just took all four of them; they were really good. Several of them made me think and involved some googling. The fact that these were real cases made me sad; I disagreed with the rationales or even the decision to Speedy Delete for far too many of them. NuclearWarfare ^{contact me}_{My work} 23:13, 27 December 2008 (UTC)

Those were real? That explains the "you can't make this stuff up" quality. I also frequently disagreed with the stated reason and frequently found the article did not qualify for CSD, but would die in a snowstorm at AFD if not improved. Many had more than one reason that could qualify depending on how strict you were in interpreting the criteria. davidwr/(talk)/(contribs)/(e-mail) 00:06, 28 December 2008 (UTC)

Agreed, very interesting survey (can't remember the last time I deliberately responded to one :). I hope you'll make the results public. Fvasconcellos (t·c) 01:14, 28 December 2008 (UTC)

Definitely, it'll probably be a week or two before I do so... but that is more because I want to wait until most people have had a chance to respond to the survey. EG once the respondants die down, I'll post at the same places where I announced the survey that the results are in... you can also watch this page.---Balloonman ^PoppaBalloon 01:22, 28 December 2008 (UTC)

Good balance so far...

So far we've had between 11-17 people taking each essay section. Of the people who have taken the survey with demographic questions, we've got: 3 people who are strong inclusionists, 3 people who are weak inclusionists, 3 people who are weak deletionists, and 2 people who are strong deletionists---surprisingly balanced.---Balloonman ^PoppaBalloon 23:38, 27 December 2008 (UTC)

Those may be deceptive: I marked "strong deletionist" for CSD, in that if something really qualifies for CSD and cannot be quickly edited so it no longer qualifies, I'm very unlikely to not try to delete it. But I'm much more of an inclusionist when it comes to non-CSD material. Others may have similarly restricted their demographic answers to the questions as written. davidwr/(talk)/(contribs)/(e-mail) 00:03, 28 December 2008 (UTC)

No, I'm sure you included me in there, but as I stated in the "other" section for that question, I literally played eeny, meeny, miny, moe to choose which to pick, since the inclusionist verses deletionist divide is a false dichotomy I don't recognize (again, this is why making people answer one of the four options for each question, instead of allowing "other" to serve as its own choice, introduces skew).--Fuhghettaboutit (talk) 01:22, 28 December 2008 (UTC)

How many times did you IAR?

I IARd once, possibly twice. How many times did you (plural) IAR? --Izno (talk) 06:29, 28 December 2008 (UTC)

Is this relative to the survey or to the number of times people IAR on CSD?---Balloonman ^PoppaBalloonTake the CSD Survey 06:35, 28 December 2008 (UTC)

The survey. --Izno (talk) 06:42, 28 December 2008 (UTC)

I did not IAR. I can think of nothing there, and in fact nothing I have ever seen in 2 yrs of checking CSD that needed speedy deletion and where IAR was necessary. I would much rather stretch the boundaries slightly than say IAR--there or anywhere. IAR is needed in 2 situations: a emergency, and a genuinely new situation for which we have no guidelines. I think I said IAR once in these two years, at AfD. The only emergencies in article content are covered by G10. Anything else which absolutely must be deleted speedy is covered by Vandalism. The use of IAR is by those without a knowledge of all the details of policy and inadequately developed imagination. DGG (talk) 17:55, 28 December 2008 (UTC)

In stretching the boundaries, you're inherently IARing to some extent... --Izno (talk) 18:47, 28 December 2008 (UTC)

Some general comments on IAR: 1) Izno is right, stretching boundaries implies at least slight IAR, and 2) Most situations are at least slightly unique, it's just 99% of them are "close enough" to fitting the rules and guidelines nicely that nobody will notice the difference, or if they do notice, they won't care. "IAR" is a combination of the "necessary and proper" aka "elastic" clause of the United States constitution and an emergency "I'm willing to be blocked/desysopped for this or at least get yelled at a lot if I'm wrong" way to do the right thing if the rules don't provide a way to do the right thing. Having said that, when it is used in any way that won't have unanimous consent, it should be used either 1) very sparingly or 2) as a brake rather than an accelerator to processes. In other words, IAR and turning an obvious vandalism speedy into a PROD isn't harmful, IAR and speedy-deleting something because you think it needs to go, but there's no criteria that fits, should be done sparingly if ever and only with the understanding that you wiki-reputation will take a hit for it. Now, speedy-deleting something as A7-lack-of-notability when db-copyvio is a better fit, well, you'll have unanimous consent because nobody's going to care much about that. davidwr/(talk)/(contribs)/(e-mail) 19:17, 28 December 2008 (UTC)

I never used IAR and I think it shouldn't be used within CSD. If the whole point of the rules is to limit the cases something should be done, invoking IAR in those cases makes the rules completely moot. If we say "these are the only cases when admins are allowed to speedy delete an article", we cannot allow them to ignore that completely if they think it's a good idea. We could just reduce the criteria to "whenever an admin feels like something should be deleted" if we allow IAR. Regards SoWhy 19:38, 28 December 2008 (UTC)

Never. I'll occasionally use IAR as a "keep" argument – as on this AFD, for example – but never as a deletion reason. "IAR" as a delete reason is just a fancy way of saying "I don't like it". – iridescent 19:41, 28 December 2008 (UTC)

Re: "I think it shouldn't be used within CSD:" I occasionally have to resort to IAR when de-speedying something. I justify it by saying "yeah, it could be speedied, but it's best for the encyclopedia if it's improved in place, PRODded, userfied, or sent to AFD." I'm not an admin, but if I were, I would on rare occasions use it to combine several weak claims. If I may set up a contrived example: Take a new article that contained a border-line attack, was 75% copyvio, was a not-quit-obvious-hoax, and had only a marginally credible claim of notability. It might get me an IAR speedy rather than a rescue attempt or remove-copyvio-and-attack-then-PROD as I interpret the rules to call for in this situation. It probably depends on how good an article I think can be written about the subject and whether WP:GOODRIDDANCE, WP:STARTOVER, or WP:SOFIXIT is the better option. While I've never seen a speedy delete that was this convoluted, in the last 12+ months I have seen a very few where WP:SOFIXIT wouldn't work and deleting it now is better than the slow process of AFD, particularly where there is content that should be removed sooner rather than later. Cases like these are one of those "you are staking your reputation every time you do it" things, and nobody will fault you if you follow the rules, edit out and hide the copyvio/blp/attack material, and take what's left to PROD or AFD. davidwr/(talk)/(contribs)/(e-mail) 23:26, 28 December 2008 (UTC)

It did occur to me that "stretching boundaries" can be considered as IAR, but what I meant was using broad rather than narrow interpretations,and keeping them focussed on specific rules, rather than suitability in general. One cannot avoid all interpretation, and there will always be a fuzzy area around even the most carefully written rules. In some cases, i interpret the possibility of deletion rather broadly--in particular, "test page", which I will use to includes pages meant without any actual intent that they really be in the encyclopedia In some criteria, I work much more narrowly, such as G11, where I will not speedy a page that I think has any reasonable chance of rescue if the subject will actually be important. To a certain extent, one does go in the end by what one thinks will harm the encyclopedia. But what I will not do is take advantage of being able to review articles to keep articles in that I personally think ought to be in, but do not fit the consensus rules, and to remove ones that I think have no place in a proper encyclopedia , but where others disagree. If my view might be disagreed with, its not a speedy. My real problem with CSD is a few administrators who knowingly work otherwise, rather than relying on the community. DGG (talk) 01:53, 29 December 2008 (UTC)

Other

I put a comment in 'other' for nearly every question, agreeing with one of the bubbles totally in only a few instances, and marked the option that was closest to what I wrote. How do you plan to treat comments in 'other' when you report the results? seresin ( ¡? )  07:21, 28 December 2008 (UTC)

• Yes, I was going to ask this, as I commented on nearly all of them too (usually to explain which CSD/deletion reason would have fitted, and why). Black Kite 10:48, 28 December 2008 (UTC)
  • Yeah, I commented on nearly all of them too, often just to say what criterion I would use but occasionally for more than that. ~ mazca ^t|c 14:15, 28 December 2008 (UTC)
• Am I allowed to say "me too" too? davidwr/(talk)/(contribs)/(e-mail) 19:19, 28 December 2008 (UTC)

I plan on posting the comments that others make... if it is just a short comment, I might summarize the comments, but if it is more than "This is an A7," I'll copy it.---Balloonman ^PoppaBalloonTake the CSD Survey 21:04, 28 December 2008 (UTC)

• I agreed with the bubbles more frequently, but did find some of the "don't agree" parts to be rather leading—instead of simply giving "Don't agree this should have been speedied" as an option, they seem (intentionally or otherwise) to say "Here's why this was a bad speedy, so pick this one." I think to provide balance, the options should be standardized, i.e. "Agree with speedy deletion under the criterion used", "Incorrect criterion used but valid speedy under a different criterion", "Not technically a valid speedy but IAR would apply", "Not a valid speedy, should have been handled in some other way." Seraphimblade ^{Talk to me} 08:36, 29 December 2008 (UTC)

Request add 3 columns

Please consider adding 3 columns to the results table:

• Proposed criteria
• Top 1 or 2 consensus suitable criteria w/ percents if you have them
• Top 1 or 2 criteria from Balloonman's analysis

The 1st and 2nd are important in interpreting the rest of the table. davidwr/(talk)/(contribs)/(e-mail) 03:00, 2 January 2009 (UTC)

Done ---Balloonman ^PoppaBalloonCSD Survey Results 04:13, 2 January 2009 (UTC)

Sorry I botched the repair. I see you fixed it now. Looks good. davidwr/(talk)/(contribs)/(e-mail) 04:27, 2 January 2009 (UTC)

No problem... I had it wrong and didn't realize it until you tried to fix it.---Balloonman ^PoppaBalloonCSD Survey Results 05:04, 2 January 2009 (UTC)

This is actually a *good* result

Looking at the raw numbers in the table, it seems there were only two cases (out of forty studied) where the plurality opinion was that appreciable amounts of legitimate content was likely lost. (The responders preferred 'cleanup' as the best course of action for just two articles.)

In almost all the other cases, the preferred course was deletion by some method (CSD, PROD, or AfD) or a redirect. One case was a content fork (3.1), while one case was an article posted on the wrong-language Wikipedia (4.4, the identical article was reposted shortly thereafter on yi-wiki).

Given that Balloonman explicitly chose for this survey cases which he thought were improperly deleted, having serious problems (loss of content, rather than simply procedural issues) with only five percent of the deletions probably means we're in very good shape indeed. TenOfAllTrades(talk) 16:41, 2 January 2009 (UTC)

5% is 50 articles a day, usually representing about 30 or so new contributors. If half of them never return, we are losing 150 potential editors a week, or 7,800 a year. if only 1/4 of them turn out to be regulars, that's about 2,000 a year. Considering we only have 10,000 or so really active contributors at present, that's the loss of a critically important element in the continuation and growth of Wikipedia. I look on the problem as being people, not articles.

But looking at it as a decision making process, of the 1000 or so speedies a day, probably 800 are truly unambiguous. So we we get 50 out of 200 actual decisions wrong, that's not a trivial error. it indicate in my opinion a significant number of very active problem administrators. DGG (talk) 16:53, 2 January 2009 (UTC)

Totally agree with DGG. See my point #5 here to save my typing it all out again. – iridescent 17:04, 2 January 2009 (UTC)

5% of all speedy deletions would be fifty articles per day, but that's not what we're looking at here. (I'm going to take your 1000 per day total as accurate.) The survey only includes deletions which used the default CSD criterion messages in their log entries. (Actually, it only looks at the small subset of CSD criteria which Balloonman feels are often misapplied: A1, A7, G1, and G10.) From that subset, most of the survey examples were deliberately drawn from deletions which Balloonman identified as problematic.

In other words, the sample included in this survey is absolutely not representative of all deletions (or even speedy deletions) as a class. Balloonman has explicitly acknowledged this. Indeed, the procedure used to choose these questions would seem to be designed to give a sample as highly-enriched as possible in iffy deletion calls.

The 5% number is beyond the absolute worst case, based on deliberately biased sampling. I dare not guess what the actual overall error rate is without someone actually compiling statistics on all deletions, but it will be much, much lower. In other words, if one grants your assumption that two hundred of the thousand speedies per day are in the difficult-calls category, we're getting 5% of those blown, or 10 per day — not 5% of all speedies. I think that even 200 is a high estimate, but we obviously don't know for certain until someone harvests all the numbers. TenOfAllTrades(talk) 17:50, 2 January 2009 (UTC)

Ten, I think you are under estimating the problem. First, you might want to take a look at the specific surveys that lead to this particular survey. Those surveys were ones where I simply looked at the deletion rationale and cited them. It didn't invovle any selection criteria except looking for deletion criteria AND trying to avoid using the same admins over and over again (especially two that I consider to be careless.)

It would also be significant if it took me a significant amount of time to find these 38 cases---it didn't. Give me an hour and I can guarantee that I can create another survey or two---that's including the formatting! But yes, we can't rely too strongly on these numbers, it was not a scientific study... but I would guess that about 10-30 speedies are done every day that are wrong.

You also pointed to the fact that only two were labeled as cleanup and that the rest were suggested to be deleted otherwise. What you failed to consider is why we send things to AFD/PROD. By sending them to AFD/PROD, we are encouraging others to write an article that meets WP criteria. My first article was tagged for CSD and then sent to AFD. Because it was at AFD I got pissed and decided to get said article to FA status. I would not be here today if my article was deleted outright. So by sending an article to AFD/PROD, we do two things: First, we give the author a chance to salvage the article. Who knows, in the article that you see absolutely zero redeeming value, there might in fact be a featured article in the making. Second, we give the author a sense of justice. If you tell me that my article is garbage, it doesn't mean much to me. And I might get upset if you delete it. If it goes through due process and gets deleted, then that is a different story. Even if an article is deleted through PROD/AFD, the author has a positive experience. S/he leave, but they now know what they have to do to keep an article here.---Balloonman ^PoppaBalloonCSD Survey Results 18:30, 2 January 2009 (UTC)

Also, yes, there are some CSD criteria that are generally utilized correctly... the problem isn't the user requested CSD or other clearly defined criteria. The problem is the ones where judgment is required and the fact that there are admins (13% of the respondants) who beleive that it is acceptable to routinely IAR when it comes to CSD---and my guess is that those who are more likely to support IAR are also the ones who are the most active at CSD'ing articles.---Balloonman ^PoppaBalloonCSD Survey Results 18:43, 2 January 2009 (UTC)

"all speedies -- you are quite right about it not including the other classes; for example, it does not include G11, promotional. I think about 3/4 of these are at least debatable, and about 1/3 the decisions unambiguously wrong--even more than the others, its the real problem area. . Even copyvio -- I observe many instances of not checking for a noncopyvio version, or not stubbifying. G10 -- you'd think that would be clear--but except for the rare obvious, everyone has their own standard there. and G4 -- obviously the non-admins tagging cannot see if something is substantially identical to the previously deleted version, so it all falls on the admins, and they sometimes don't check either. I think the same proportions would hold for the total.

and I wouldnt say that prod or afd are necessarily much better. I' think you;'d get the same audit figures, for the same underlying reason.

But there is another factor making the error rate higher than estimated: the type II error, not recognizing something as suitable for speedy when in fact it is, and sometimes leaving it in for long periods.To observe these, look at any days prod or afd listing. Here, I think the biggest cause of really harmful error is to neglect to check for copyvio.

Having said this, why don't I routinely check the logs and revert or ask the admin for reversal or take to Deletion Review everything I see thats really wrong? well, sometimes its more effective to counsel the new author into writing a better article. But the real reason is that I think it essential to stay on relatively good terms with my fellow admins, and I do not want the reputation of someone who is always complaining. Human relations with the people one works with are even more important than always doing the job right. I was a slow learner about emotions, and this is one of the things I should have but did not learn in kindergarden--but I did learn it eventually. When people complain about an admin cabal, they are wrong if it means anything concerted--they are right if it means this. DGG (talk) 21:23, 3 January 2009 (UTC)

Section break: Problematic Admins

I have to repeat and highlight something DGG just said: But the real reason is that I think it essential to stay on relatively good terms with my fellow admins, and I do not want the reputation of someone who is always complaining. When I first started looking into CSD, I was thinking, "I'll talk to the admins whom I perceive as having problems and see if I can educate them." I confronted one, whom in my opinion, is a chronic offender and realized that the frontal approach won't work... even thought I tried to use tack and a subtle approach, it didn't go over that well. But I can name 4 admins whom I consider to be problematic in their deletion patterns. 4 admins whom I can pretty much guarantee will make significant mistakes on a daily basis. The problem I have is, what to do about those admins?---Balloonman ^PoppaBalloonCSD Survey Results 23:36, 3 January 2009 (UTC)

If you see that someone's made a clear-cut error, why are you not requesting DRVs? And if DRV won't overturn their decision, why are you calling it an error? If they're really making errors on a daily basis, start getting them overturned on a daily basis. That will get the point across (or if they won't change their behavior, make it clear that we may need to change it for them). On the other hand, if they're making the right decision through the wrong process, that's not an error in any meaningful sense. Seraphimblade ^{Talk to me} 00:13, 4 January 2009 (UTC)

That option is limited and won't work in the long run. Problematic deleters need to be addressed in some meaningful way in a direct method. I say this because it would take scores of DVR requests to make a difference on one problematic editor. If there are 4 problematic editors, the person raising the concerns is going to get the label of being problematic before the issue is resolved.---Balloonman ^PoppaBalloonCSD Survey Results 00:30, 4 January 2009 (UTC) EDIT: The other part of the problem is that just because an article that shouldn't be deleted per CSD doesn't mean that it should be recreated via DVR. The problem is in deleting articles prematurely is the problem---it doesn't give the author the chance to salvage the article or prove its value. DVR isn't the solution in these cases, especially when the verdict of "endorse incorrect deletion" is returned. This only emboldens garbage deletions, without addressing the problem. It's like the CSD'er who inappropriately tags an article. The tag is removed and an admin explains why the tag is wrong. The CSD'er replaces the tag. Another admin removes it saying it is wrong. The CSD'er replaces it again, but this time an admin agrees and deletes the article. The CSD'er sits back and says, "I knew it would be deleted eventually." They have learned that there are admins out there who will delete articles and they just have to catch those admins attention. Those admins have taught CSD'ers sloppy habits and do the project a disservice. DVR would be the same thing for these admins. Speedy deleting an article that doesn't meet WP:N but clearly passes A7, does not mean that the admin is justified in IAR. A7 was explicitly written with a lower bar than WP:N, but some seem to have forgotten that.---Balloonman ^PoppaBalloonCSD Survey Results 07:42, 4 January 2009 (UTC)

I do see your concern there, and the behavior you're describing is certainly problematic. A CSD tag is like a prod in that regard, if removed by someone besides the original page author (who is disallowed from removing it and should use {{hangon}} to express objection), it shouldn't be reinstated. I see that as more a behavior issue than a deletion policy issue, though. If discussion with those engaging in such behavior has failed to stop it, and it's happening routinely, I don't think a user conduct RfC would be out of order. Seraphimblade ^{Talk to me} 10:31, 4 January 2009 (UTC)

Yeah, it is more of a behavior issue than deletion issue. It is why I am opposed to the IAR/SNOW deletions that are frequently appealed to by some CSD admins. Are there possibly cases that might fit an IAR/SNOW the best? Of course. There are always scenarios that people didn't consider or can't categorize, but if you (generic) are appealing to IAR/SNOW on a daily basis, then that indicates that either the criteria/guidelines are incorrect OR you (generic) are deleting too many articles incorrectly. I mean, last night I reviewed an editor who nomed an article with the following in the lead: Currently the lead singer and keyboardist for "Ernie And the Automatics" which features original and former members of the band "Boston" Barry Goudreau on guitar and Sib Hashian on Drums. A7 explicitly sets a lower threshold than WP:N. This article clearly failed WP:N and WP:MUSIC, but it did meet the threshold to avoid A7. A7 only asks for a claim to significance or importance. Being the lead singer and keyboardist for a band that includes two former members of the group Boston is a claim to significance. Goudreau and Hashian's membership in a group might give the group "Ernie and the Automatics" enough clout to be notable per number 6, and being the lead singer of such a group might in turn be enough give the articles subject notability. I'm not saying that it is, but it is clearly enough to avoid A7. (Which it didn't, but the fact that it was deleted does not mean that the admin's actions were correct.) And yes, there are 3 or 4 admins that I think are chronic in speedy deletions.---Balloonman ^PoppaBalloonCSD Survey Results 17:16, 4 January 2009 (UTC)

8 have significant minorities of "do not speedy"

Some interesting results: 8 of these hand significant minorities of "do not speedy." This indicates these particular cases lack consensus that speedy is the way to go on these articles.

Item, % saying don't speedy-delete, my recommendation after reviewing item and/or my prediction at AFD. Note that only 1 or 2 of these have any hope of remaining as articles.

• 3.4, 31.7% no-speedy, "The house of Cedars is a block in Broadgate Park..." easy unencyclopedic AFD SNOW
• 1.8, 27.6% no-speedy, "Uglau is the german word for 'telephone'." Almost certainly hoax/AFD SNOW
• 4.2, 27% no-speedy, "pata nahi hai. jisko pata ho to bhej do", if language can be determined tag for translation or transwiki otherwise userfy and alert user of other-language wikis.
• 2.4, 25.7% no-speedy, possibly foreign-language, possibly gibberish, "ASSALAM ALIKUM L.A. IQBAL HE WAS A GREAT PLAYER" possibly a notable person, web search came up dry. Misspelled name? Likely AFD SNOW
• 1.7, 25.5% no-speedy, Creative is a Brisbne based Website and Multimedia development company. May be a notable company.
• 4.5, 22.2% no-speedy, Maher Diab, also known as GOD for many civilizations, likely fail at AFD as hoax, but could be notable fictional character
• 2.9, 17.1% no-speedy, "TEXT: Son of [Sam XXX], he created a religion called Samuelism," very likely hoax/AFD SNOW
• 1.6, 13.3% no-speedy, "Nankeyman: A filming business responsable" - could be notable, check, if no references then AFD likely snow
• 2.6, 11.4% no-speedy, "The great butterfly migration takes place in the fall/winter time. It starts in" Incomplete article, wait a few hours/userfy

davidwr/(talk)/(contribs)/(e-mail) 18:34, 2 January 2009 (UTC)

in general, I agree. that some people wont speedy when justified is also a problem. /That some newpage patrollers wont nominate when justified but merely tag is also a problem. There's going to be human error is any process like this--the only way or decreasing error is through work--a multistep regular review process. Or education, which is also work. or deadmining the worst, which might be even more work. the question is how much work is worthwhile, and where can it best be applied. I know one point where it could well be efficiently applied, and that is prohibiting single-handed admin deletions. We have enough admins to check everything quickly, even copyvio. But beyond that, though we should clean and tighten what we can, how should we best proceed on a practical basis? Changes that the community will not agree to are not really in the class of practical. I emphasise the practical DGG (talk) 21:33, 3 January 2009 (UTC)

Since the vast majority of these would snow-close at AFD, the practical thing to do is train admins that it's perfectly OK to change a speedy into a PROD or AFD, teach users, esp. page-patrollers, to not speedy if in doubt, and to take care of the mistaken speedies at DRV. It would also help if there were well-publicized tools with well-publicized instructions patrollers could use to fast-create an AFD. Speedy and prod are trivial to do, an AFD takes at least 4 steps if you do it by hand, 3 if you don't count notifying the editor. davidwr/(talk)/(contribs)/(e-mail) 22:20, 3 January 2009 (UTC)

Twinkle and also even Huggle,which a lot of Patrollers use, have systems which make it very easy to AfD articles. It's certainly no more difficult than typing in a rationale for Prod. NuclearWarfare ^{contact me}_{My work} 22:54, 3 January 2009 (UTC)

It would be better to build in a system that makes it automatic for those using any interface. If it can be programmed for TW, it can be programmed for the main interface. Personally, I think another step would be to encourage most patrollers not to use eith TW or HG until they know better what they are doing, for it encourages tagging without checking. DGG (talk) 23:11, 3 January 2009 (UTC)

I can't find it now, but I'm 99% positive that User:Werdna was developing a "deletion queue," a more natural way to include AfD/Prod/SD articles. NuclearWarfare ^{contact me}_{My work} 23:35, 3 January 2009 (UTC)

DGG, I think David's point isn't that these cases should all be CSD'd, but rather that CSD'ing them isn't as clear as some might think. CSD should be used in cases that are clear cut, if 20% of the admins won't CSD, then they are not as clear cut as others might think.---Balloonman ^PoppaBalloonCSD Survey Results 23:39, 3 January 2009 (UTC)

If 20% of the admins won't CSD in a given case, 20% of the admins might be wrong. Just as likely as instances where only 20% would CSD. Excessive timidity does not help the encyclopedia. Throwing yet more cases into AfD does not help consideration of the ones that are there--the rough guide used to be that 100 per day is the maximum reasonable, but that's usually far exceeded. What we need more use of in general is PROD. Rational deletion does not mean minimal deletion. DGG (talk) 23:25, 5 January 2009 (UTC)

Inherent bias in CSD patrollers

There's a divide between those who feel deletion (as rapid as possible) is the best response to a poor article and those who feel that improvement is often the way to go. The problem is that the former will tend to hang out at CSD, while the latter will tend to avoid it, resulting in a bias towards speedies being looked at by admins of a deletionist stance. Espresso Addict (talk) 17:29, 4 January 2009 (UTC)

Which is partially why I advocate a stricter set of standards... people who work in that arena are more likely to be just as you say, have a bias towards IAR'ing articles they see as problematic--especially with immature/inexperienced CSD'ers/admins. It is also a problem that is liable to get worse, if there aren't forces pushing the other way. EG I've been told by non-admins, "It doesn't matter, the article will be deleted." Then a few minutes later, somebody deletes said article, despite its speedy deletion having been rejected by two other admins! The admins that do that are not teaching people how to properly delete articles, but rather encouraging adminshopping. "Keep nominating the article, eventually, I'll get to it and delete it for you." They might as well take adds out saying, "If you have an article nobody else will speedy, see me, I'll do it for you."---Balloonman ^PoppaBalloonCSD Survey Results 01:48, 5 January 2009 (UTC)

Indeed. It might be improved if a single decline from an administrator were sufficient to force the longer consideration provided by the PROD process -- it certainly might make CSD taggers think more carefully about which rationale to select. Your results show that a surprisingly high proportion of speedies are controversial and might benefit from longer community consideration. Espresso Addict (talk) 01:57, 5 January 2009 (UTC)

Here I have to echo Tenofallspades to a certain degree... I would argue that A1,G1, A7 are often misapplied. The other criteria I haven't looked at too closely. We can't extrapolate these results to cover ALL CSD's because *I* haven't looked at them and the survey didn't include them. (At the same time, we can't say they are done correctly---because they weren't reviewed.)---Balloonman ^PoppaBalloonCSD Survey Results 16:05, 5 January 2009 (UTC)

"criterion"!!

Please. "Criterion" is singular; "criteria" is plural. Write "This criterion is..." or "These criteria are..."; do not write "This criteria is..." or "another criteria". Michael Hardy (talk) 03:20, 16 January 2009 (UTC)

For what it's worth, the Merriam-Webster says "[S]ingular criteria is not uncommon in edited prose, and its use both in speech and writing seems to be increasing. Only time will tell whether it will reach the unquestioned acceptability of agenda." From my personal experience I'd say in very many people's dialects it is an acceptable singular. Olaf Davis (talk) 18:18, 16 January 2009 (UTC)