Jump to content

Wikipedia talk:Reflections on RfX/Archive 2

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

The criteria should change by consensus not by minority action

Some criteria are more easily measured than others, and in the grand tradition of giving most weight to things that are easiest measured; edit count and tenure both have enormous importance at RFA, with current de facto standards far in excess of the requirements when most admins were appointed. Part of the cause of this is that without any fixed criteria that would need consensus to change it only takes 30% of the community to oppose over one issue for that to become a criteria for adminship. Setting some of the less important criteria such as editcount, tenure and time since last RFA even at today's inflated levels would at least stabilise those criteria and prevent them changing without a consensus to alter them. It would also make RFA more attractive for potential candidates as they would know whether they met the arbitrary parts of the criteria or how soon they would meet them.

The important part of the criteria would remain judgmental things:

  • do the candidate's edits indicate they would use the tools correctly?
  • do the candidate's edits indicate they would be civil?
  • do the candidate's edits indicate they can explain things to others?

But the unimportant parts of the criteria could be codified (suggestions below for individual elements of the unimportant part of the criteria)

Support

  1. ϢereSpielChequers 09:08, 7 August 2015 (UTC)
  2. Kharkiv07 (T) 15:00, 7 August 2015 (UTC)
  3. Kudpung กุดผึ้ง (talk) 04:55, 8 August 2015 (UTC)
Comment

IMO, what is highly important is their apparent knowledge of Wikipedia policies and standards - thus a broad knowledge of such, preferably backed by their actions on appropriate noticeboards etc., is more important than "civility" which has exceedingly labile interpretations, or a belief that since they have made no comments about a policy or standard that they therefore clearly would not violate them. That, alas, is the huge leap on which a number of RfAs have failed. Collect (talk) 12:39, 9 August 2015 (UTC)


The unimportant part of the RFA criteria should include 12 months tenure

It has been a while since anyone has passed RFA within 12 months of creating their first account and starting to do logged in edits. A requirement in the unimportant part of the RFA criteria that RFA candidates need to have edited in 12 different calender months before submitting an RFA would fix that particular arbitrary part of the criteria, potential candidates would have a better idea where they stood and once frozen that part of the criteria could only change by consensus (declared alternate or prior accounts would count towards this, but not IP edits or undisclosed accounts).

Support
  1. ϢereSpielChequers 09:19, 7 August 2015 (UTC)
  2. Old accounts should count; IP editing should not. 12 months since the user's first edit as a strict criterion should eliminate most, if not all, NOTNOW candidates. Bilorv(talk)(c)(e) 09:37, 7 August 2015 (UTC)
  3. Kraxler (talk) 12:24, 7 August 2015 (UTC)
  4. I'm not sure what "unimportant part" means, but formalizing a 1 year of (substantive) editing requirement would probably avoid a lot of heartache, yeah. And, practically, I don't think anyone can learn the ins-and-outs of this place in less than a year. --IJBall (contribstalk) 15:50, 7 August 2015 (UTC)
  5. AliveFreeHappy (talk) 17:17, 7 August 2015 (UTC)
  6. As long as we aren't saying "consecutive", this is reasonable. Dennis Brown - 20:29, 7 August 2015 (UTC)
  7. But used with common sense. Many users, including me, created an account and then made fewer than 10 edits for the first two or three years. That can still include NOTNOW votes. Kudpung กุดผึ้ง (talk) 04:12, 8 August 2015 (UTC)
  8. Yes, however I don't like the word "unimportant" in this context, and I agree with the comments from Dennis and Kudpung re what "12 months" means. Callanecc (talkcontribslogs) 13:06, 11 August 2015 (UTC)
Comment

I'm uncomfortable with the concept of an "unimportant part", even though I understand the intent of the proposals here and just below. I could hypothetically have my own idiosyncratic criterion, such that I'm the only editor participating in an RfA who judges the candidate that way, and my !vote would still be counted. That would be a criterion, and clearly one that is not important overall, and yet not one that we could really determine a consensus about in this discussion here. I'm fine with advising potential candidates that they are likely to be judged by some editors based on certain criteria, but I think that we ought to be careful about designating those criteria with some sort of status. --Tryptofish (talk) 20:57, 8 August 2015 (UTC)

  • I feel bad about raising any reservation about the only thread on this whole page that until now had unanimous support, but I feel like I should say something because my own RfA, albeit awhile now, was a counterexample. And I hate to use my own RfA experience from several years ago as an instance of anything but I will, because I don't believe it's unique. I started serious editing in July 2006, was being asked to seek adminship by October, waited until I had six months' experience, and passed in January. I don't see how it would have done Wikipedia any good for me to have been forced, or even counseled in a way that people would oppose an RfA for defying the counseling, to wait another six months. Newyorkbrad (talk) 14:41, 9 August 2015 (UTC)
    • Those were better days at RFA, I'd be happy to go back to 6 months, but I don't think we could get consensus for that. If we can get consensus to set a 12 month criteria then my hop is that we can in so doing fossilise that part of the criteria so that it can only change by consensus. Otherwise in a few years it will have drifted to 24 months simply because 30% of the community can always impose a stricter standard on RFA. ϢereSpielChequers 20:36, 10 August 2015 (UTC)
      • I don't understand this hypothesis at all. If there are "fossilized" criteria written down somewhere, why would that make the voting population any less likely to use their own inflated unwritten criteria anyway? Seems much more likely that defined minimum criteria will attract candidates who just barely meet them and unexpectedly find themselves in a meatgrinder of unexpected opposition, while at the same time denying those voters who are comfortable with lower baselines the opportunity to advocate for their view. Opabinia regalis (talk) 01:15, 11 August 2015 (UTC)
        • Most userrights work that way, there is a criteria and you use good judgement to see if someone meets that criteria. I don't remember the criteria for Rollback ever changing, Autopatrolled changed from a recommended threshold of I think 70 articles to 50, and that was a consensus based change. I'm hoping that if we can agree some unimportant criteria such as editcount and tenure then the discussion can focus on more important criteria like based on actually looking at their edits are they clueful and responsible? But I agree there is a risk that people will simply say their personal criteria are different. ϢereSpielChequers 12:58, 11 August 2015 (UTC)
          • Most userrights are granted and removed by one person, often processing multiple requests at a time from unfamiliar users, so easily checkable criteria make sense. To make a lame analogy, this is sort of like saying "DYK has a hard minimum of 1500 characters in an article, so it makes sense for FAs to have a hard minimum too." Well, not really; the review is intended to be much more holistic, and hard minimums would be a distraction even if everyone agrees that a 1500-character article is not going to be an FA. Even with specific written criteria, there's a lot of variation in how closely people adhere to userright recommendations in any case. Opabinia regalis (talk) 20:44, 11 August 2015 (UTC)

The unimportant part of the RFA criteria should include >7,000 edits

Edit count is a bit of a joke as an RFA criteria, but some people seem to be concerned about it. It has been a while since anyone has passed RFA with less than 7,000 edits, and the recent successful candidate who had little more than that had some opposes for low edit count. Putting a requirement of 7,000 edits into the unimportant part of the RFA criteria would fix that particular arbitrary part of the criteria, potential candidates would have a better idea where they stood, and once frozen that part of the criteria could only change by consensus.


Support

  1. ϢereSpielChequers 09:19, 7 August 2015 (UTC)
  2. Kraxler (talk) 12:23, 7 August 2015 (UTC)
Oppose
  1. I nom'ed Basalisk with 6000 edits in 2012 Wikipedia:Requests for adminship/Basalisk and he passed and has been a great admin. 4-5000 I could support. Quality matters more than quantity, so I think we need to err on the low side if you have a criteria. Dennis Brown - 20:31, 7 August 2015 (UTC)
  2. Edit count is cetainly a joke - sometimes, but it's the most difficult of the numerical critera to qualify and quantify. My criteria only demand 6,000 edits - but it depends also what they are; I could eventually even go for less. Since unbundling some tools, it's easy for rollbackers, reviewers and other vandal patrolers to rack up thousands of semi automated minor edits using Twinkle, Huggle, or whatever else, and using AWB to rectify commas and spaces; it is also extremely easy and very fast for someone to mass create 100s of stubs (e.g. genera and species) by using a basic stub as a template. None of which demonstrate anything to do with the potential for admin work. Kudpung กุดผึ้ง (talk) 04:27, 8 August 2015 (UTC)
  3. Not all edits are created equal.--Staberinde (talk) 15:27, 11 August 2015 (UTC)
  4. I agree that, generally, candidates with fewer than 7,000 edits are not ready yet, but if a candidate's edits are major, than I'm willing to support at a lower bar. Esquivalience t 00:32, 14 August 2015 (UTC)
  5. We're going to have to start rethinking the "edit count" paradigm. An increasing number of editors is using the VisualEditor interface and people edit differently. I myself have found that I make about half as many edits when using VE than I would have if I had been using wikitext. Many longer-term editors also habitually save their edits very frequently, some because we learned to edit back when failed saves were commonplace, some because they want to see their edits in context, some because that's the way to build your edit counts, and some because they're making lots of minor changes. The edit numbers should start edging their way down, or VE edits count for two or three times as much as a wikitext edit. Risker (talk) 05:25, 14 August 2015 (UTC)
  6. 7000 is a very arbitrary number. Gizza (t)(c) 08:07, 15 August 2015 (UTC)

The unimportant part of the RFA criteria should include 4 months activity since last RFA

There used to be an unwritten rule that RFAs should be at least three months apart, like all arbitrary rules at RFA this has been subject to inflation. Putting a requirement into the unimportant part of the RFA criteria that RFA candidates need to have edited for four months since any previous RFA of theirs would fix that particular arbitrary part of the criteria, potential candidates would have a better idea where they stood, and once frozen that part of the criteria could only change by consensus.


Support

  1. ϢereSpielChequers 09:24, 7 August 2015 (UTC)
  2. Kraxler (talk) 12:22, 7 August 2015 (UTC)
  3. Kharkiv07 (T) 15:00, 7 August 2015 (UTC)
  4. Yes. But I don't think anyone is stupid eough to not wait at least 3 or 4 months. 6 months would be better. Kudpung กุดผึ้ง (talk) 04:29, 8 August 2015 (UTC)
  5. What Kudpung said. Dennis Brown - 13:46, 13 August 2015 (UTC)
Comment
Do we have data on this? All the "reruns" I've seen have been after a more sizable time-gap before their previous RfA (I think they've all been a year+), but I haven't been watching RfA long. How many people have tried to re-run again within 6 months? I think 6 months might be a more sensible suggestion. --IJBall (contribstalk) 15:59, 7 August 2015 (UTC)
Agree with IJBall, it's hard to see an editor having a successful RFA 4 months after an unsuccessful one. 6 months would seem reasonable and is standard practice anyway. Valenciano (talk) 11:00, 9 August 2015 (UTC)

Proposed improvements should be evidence-based

Wide latitude is given to individual participants to develop their own personal criteria for adminship. There is no requirement or social expectation that these have any basis in fact. Proposed improvements to the process, particularly those involving numerical thresholds, should be held to a higher standard and should be justified with evidence.

Support

  1. (as proposer) Opabinia regalis (talk) 18:37, 7 August 2015 (UTC)
  2. Not only should it be based on evidence, it should be based on the scientific interpretation of the evidence. Alakzi (talk) 10:06, 8 August 2015 (UTC)

Comments

Guy: Yes. All of the above.
Kudpung: Ironic that I'm the one saying this, but four-year-old data is dated. Especially the statistics. There have been over 250 RfAs since then. Opabinia regalis (talk) 09:39, 8 August 2015 (UTC)
Opabinia regalis, Yes, one of them was mine and I've voted on all the others. It's not dated, it's just that you haven't taken a look at what's there. You would see that most of the talk and revival talk ever since on WT:RfA is pure perenial recycling of what was said there and for two years before it. The only things that are not up to date are some of the tables regarding voters and their habits, and bringing them up to date would almost certanly reflect the same trends except perhaps that the process is somewhat less of an ordeal now than it was then. In fact, FWIW, I've spent most of today reading through 2009 and 2010 discussions on WT:RfA and although most of the participants have changed, the talk is identical to that of today. If you need proof, beware of an avalanche of diffs, but you really just need to get yourself up to date and help us to take these 7-year-old suggestions finally to a conclusion. Kudpung กุดผึ้ง (talk) 10:07, 8 August 2015 (UTC)
2009? There are threads from 2006 that you could replace the numbers and dates in and re-post the same posts now and nobody would notice. Trying to reach a conclusion about what's wrong with RfA is like trying conclusively settle the question about whether the local sports team actually sucks. I've actually read a good chunk of the RFA2011 stuff (hey, I have a long commute and don't have to drive) but the problem is, if you want to show that a four-year-old dataset is still relevant, then you have to collect new data and show that they're similar. For example, the total number of RfAs dropped by a factor of 10 from 2008 to 2014, yet the proportions of outcomes are almost identical. A process that is losing candidates uniformly is qualitatively different from one that is filtering out candidates likely to have a certain result. Throughput is currently so low that some of the historical infrastructure built around deterring early candidacies seems overdue for a rethink, and yet we have (perennial) proposals aimed at rigidifying participation criteria. Opabinia regalis (talk) 22:05, 8 August 2015 (UTC)
Well, thank you for concurring. When I read you above it sounded as if your were disagreeing. Not only the successful RfA, but if yu were to draw a graph with three lines: Pass, fail, and participation in talk at WT:RfA, you will find that the falling lines are exactly parallel. I've done theexercise a few weeks ago, but I don't know how to produce the graphs (not that it needs proving anyway). Kudpung กุดผึ้ง (talk) 04:13, 9 August 2015 (UTC)
I was disagreeing with your methods, not (necessarily) your conclusions. Opabinia regalis (talk) 20:57, 9 August 2015 (UTC)
  1. This makes no sense and is not supported in policy. A "change" that benefits Wikipedia may have no "evidence" to be presented, it might just be a better idea. All this seems to be is a barrier to dissuade others from proposing new and novel ideas to improve the place. Not a viable idea. Dennis Brown - 19:17, 10 August 2015 (UTC)
Well, none of the changes that have occurred since 2008, and particularly since 2010, have been useful in meeting the goal of having enough hands on deck to accomplish the goals of maintaining the project, even as it has doubled in size. We've added tons of tools, we've broken up chunks of the toolkit, and in 2015 fewer edits than ever are reviewed contemporaneously, vandals take longer to identify, we have few people developing the skills to manage the tools above the admin level, there are fewer admins who carry an ever-heavier burden of work (and a much higher percentage of admins don't do any logged admin actions at all). So having some evidence-based criteria, instead of just allowing RFA to continue its downward spiral, is worth trying. Risker (talk) 05:45, 14 August 2015 (UTC)
@Risker: Just FYI: there was discussion about the distribution of admin actions a couple of months ago during the last cycle of "OMG we need to fix RfA" in this AN thread, where I posted these graphs comparing 2014-15 to 2006-07.
Distribution of the fraction of total admins taking a given number of actions in one year.
Distribution of the fraction of total admin actions taken by each admin in one year.
Almost exactly the same percentage of admins (~35%) are totally inactive in a year, but the 2015 set has a higher percentage taking very few actions. In other words, the burden used to be more broadly distributed, but is now concentrated in fewer hands. (There are various caveats in the linked thread but I think the point stands.) It's not that there are more who do nothing, it's that there are more who do less.
I agree with the general point that the problem is pipeline-related though. Opabinia regalis (talk) 03:12, 16 August 2015 (UTC)

Kittens are great

Apparently you're not supposed to oppose things on this page. Here's something unopposable. Kittens are just the best.

Support
  1. (as proposer) Opabinia regalis (talk) 18:37, 7 August 2015 (UTC)
  2. Well of course. ('eow...) They grow up to be cats that guard the tuna casserole. Fylbecatulous talk 20:25, 7 August 2015 (UTC)
  3. Of course, although cats don't make the best tuna casserole guards. Someguy1221 (talk) 23:50, 7 August 2015 (UTC)
  4. Just see my sig.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  11:26, 9 August 2015 (UTC)
  5. ...with catsup (or ketchup, if you prefer : ) - jc37 14:48, 15 August 2015 (UTC)
    I prefer cat-sup. Or prepared as kits and chips. --Tryptofish (talk) 18:49, 15 August 2015 (UTC)
Oppose
You can lead a horse to water but you cannot tuna fish.
  1. (as a fish) --Tryptofish (talk) 19:36, 7 August 2015 (UTC)
    The solution is clearly to be an organism that's already extinct ;) Opabinia regalis (talk) 20:58, 7 August 2015 (UTC)
    Come to think of it, I haven't been feeling that great lately. :-) --Tryptofish (talk) 21:05, 7 August 2015 (UTC)
  2. ...simply to get on Tryptofish's good side and possibly obtain a position in his Conquering Horde. I am looking for a position as a minion; I have a lot of experience in various henchman and sycophant positions and want to move up. --Guy Macon (talk) 20:21, 7 August 2015 (UTC) Guy Macon (talkcontribs) has made few or no other edits outside this topic. --Tryptofish (talk) 20:26, 7 August 2015 (UTC)
  3. Allergic, although Morbo finds them delicious. Dennis Brown - 13:40, 13 August 2015 (UTC)
Neutral
RfA candidate who is likely to go to Cat Chat
  1. I only like Kill Kittens. --IJBall (contribstalk) 20:48, 7 August 2015 (UTC)
http://www.explodedsoda.com/2011/01/kittens-are-assholes.html
http://www.cracked.com/article_19007_6-reasons-kittens-suck-learned-while-raising-them.html
There. Now if I ever get tempted to run for admin, there will be a nice diff for the oppose camp to post. --Guy Macon (talk) 02:13, 8 August 2015 (UTC)
1. Opppose. Per edit summary at [1], Guy Macon clearly fails the kitten litmus test and is most probably a robot and/or sociopath. Someguy1221 (talk) 08:30, 8 August 2015 (UTC)
My cat vandalized an article once. I tried sticking a warning template on him, but, well, you've seen those "put tape on your cat" youtube videos. Opabinia regalis (talk) 08:49, 8 August 2015 (UTC)

Oops. It appears that the jig is up. I have been exposed. Tell me, what is the first thing that comes to mind when you combine "robot" and "sociopath"? That's right, Dalek. Now look at these diffs:[2][3][4][5]. --Guy Macon (talk) 12:53, 9 August 2015 (UTC)

See also User:Radiant!/Classification of admins. Newyorkbrad (talk) 14:48, 9 August 2015 (UTC)

Well some of us apparently are too unfunny to be listed there... (mopes...<sniffle>) - jc37 14:20, 16 August 2015 (UTC)

Participants in "RfA reform" discussions are a self-selected group

Discussions like this one (and all the other RfA-related threads currently ongoing) attract participation by people who like to write big piles of words in projectspace. There is no evidence that the best candidates are people who like to write big piles of words in projectspace, and there is a reasonable argument for the exact opposite. There is no evidence that people who like to write big piles of words in projectspace are good at identifying good candidates or bad criteria, and there is a reasonable argument for the exact opposite. There is real risk of overinterpreting opinions posted in these discussions due to their relative salience.

Support

  1. (as proposer) Opabinia regalis (talk) 18:37, 7 August 2015 (UTC)
  2. Unfortunately yes. At best, a record at sound argument there might help with a RfB. Jo-Jo Eumerus (talk, contributions) 18:57, 7 August 2015 (UTC)
  3. That. – iridescent 19:00, 7 August 2015 (UTC)
  4. Except for me. --Tryptofish (talk) 19:35, 7 August 2015 (UTC)
  5. Wikipedians often talk too much and say too little. Alakzi (talk) 10:08, 8 August 2015 (UTC)
  6. Self-selection biases are a serious and perhaps unavoidable problem on a non-compulsory volunteer website. There are solutions but none which would not introduce more problems than they solve. Bilorv(talk)(c)(e) 09:28, 17 August 2015 (UTC)

Comment: Serious question – is there even a solution to this problem? --IJBall (contribstalk) 20:50, 7 August 2015 (UTC)

Lots, but none will ever happen. An obvious one would be a once-only all-users spam (or even a popup with a checkbox) asking "Do you consider the current RFA process to be a problem: yes/no/don't care". I imagine the Jeremiahs who haunt WT:RFA would be shocked to see "yes" a distant third and "don't care" a runaway winner; the only people who make comments about a problem tend to be those who think there is a problem. – iridescent 20:55, 7 August 2015 (UTC)
You seem to know a lot about the Jeremiahs who haunt WT:RfA - yes and I'm not ashamed to be a very regular contributor there, so I do know what gets discusse there. In fact I would not be surprised at all at your surmise, most RfA get more attention than serious debates on policy changes. But then, policy change discussions are not popularity contests or reality shows. Kudpung กุดผึ้ง (talk) 16:24, 8 August 2015 (UTC)
Oh, I'd include myself among the Jeremiahs—as you presumably know, I've been banging the "Wikipedia's barriers to entry have reduced it to a shell being propelled forward by a hardcore of old-timers" drum for years now—but it doesn't mean I think most editors, let alone readers, particularly care. ‑ iridescent 18:05, 9 August 2015 (UTC)
Participants in any discussion, with the exception of the Arbcom deliberations and of crat chats, are a self-selected group.--Ymblanter (talk) 10:06, 12 August 2015 (UTC)

We need to unbundle more of the tools, so that usage is based on competence/merit, not popularity or a politicized notion of trust

People keep opposing this idea, but every time a tool is unbundled, the results are positive. The template-editor bit is recent-ish example in a long line of unbundlings.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  11:24, 9 August 2015 (UTC)

  1. Support This is how to make it no big deal. valereee (talk) 12:32, 9 August 2015 (UTC)
  2. Though I would not call it "unbundling" but more a matter of "this is the area the person wishes to work in, and these are the tools useful for his or her purposes." Collect (talk) 13:00, 9 August 2015 (UTC)
  3. Sure, I'll support this, but it's true that people will keep on opposing it. --Tryptofish (talk) 22:05, 9 August 2015 (UTC)
  4. Certainly probably for WereSpielChequers' idea to allow some highly experienced patrollers the limited facility to block some kinds of vandals. The downside of of minor maintenance rights however, is that they attract hat collectors as any admin knows who has worked at WP:PERM. Kudpung กุดผึ้ง (talk) 23:51, 9 August 2015 (UTC)
  5. Weak support for a limited amount of tools for the editors who need them and can show clue and competence, particularly deletion (an RfA-like process?) and limited blocking capabilities. I would not support giving them to the norm, just the exception; for instance, the quality of new page patrols now is obscene. The only problem is hat collecting - 90% of WP:RFPERM is unhelpful "non-admin observations" and wannabe semi-admins applying for rights beyond their ken. Esquivalience t 00:51, 10 August 2015 (UTC)
    The quality of NPP always was obscene. With RfA reform it has been my other 7-year crusade for clean-up. Indeed my pressure on the WMF got them to create the new suite of NPP tools, but the community didn't play ball with getting some minumum qalifications established for using them. Thus paradoxically, the quality is low because it doesn't confer a 'privileged' hat to wear and hence does not attract users to the task.Kudpung กุดผึ้ง (talk) 10:03, 10 August 2015 (UTC)
  6. Absolutely. --IJBall (contribstalk) 06:34, 11 August 2015 (UTC)
  7. We've had oodles of proposed unbundlings only a handful have actually happened, but they have all been successful. If a tool can be treated independently of other tools and is needed by a group of people who can no longer get through RFA then we should unbundle it. Next would be block IP/Newbie for vandalfighters who haven't done enough content to pass RFA. ϢereSpielChequers 12:01, 11 August 2015 (UTC)
  8. Strong support given a loose definition of "tool". My first post to Wikipedia talk:Requests for adminship#Split up the powers lists the 33 unique group rights given to administrators, and other groups which are above sysops in the hierarchy (e.g. 'crat). I support the unbundling of (suppressredirect) and (unwatchedpages), both of which could either be given automatically to users that meet certain criteria (e.g. 1 year editing, 5000 edits) or on request with more lenient rules like rollback, reviewing etc have. But I doubt anyone supporting this proposal wasn't thinking about those two rights. I am open to any more important unbundling ideas like 'vandal fighter' or 'page protector', but would note that the greater the number of tools that are unbundled, the more confused everyone is. On a separate note, seriously, what is (jsonconfig-flush)? Bilorv(talk)(c)(e) 09:53, 17 August 2015 (UTC)
I think a block is a big deal, so I'm quite nervous about the thought of giving the block tool to vandal fighters. I do see the limitation to IP/newbies, but it still worries me. However, in the interest of contributing positively, I thought about what I would need to be more comfortable with the idea. Two things would help. First, I'd like a time limit on the length of the block. Vandalfighters are often dealing with board schoolchildren who if blocked for 24 hours might move on. Some will return, but for those it may be worth getting an admin involved. So my first suggestion is that the likability granted to non-admin vandal fighters be limited in time. I don't want indefs or even one-year blocks issued in the heat of the moment. I'm picking 24 hours for convenience, but a short period of time might even do the trick.
My second suggested modification arises from my concern about block logs. Someone running for RFA who has a block on their record will inevitably be asked about it. Perhaps an explanation that an overzealous vandal fighter overreacted really on will be an adequate explanation, but there's an easier solution. Let's agree that blocks issued by vandal fighters rather than admins do not get recorded in the block log. This would obviously require a technical change the software, but if we're going to grant her right which can only be used against certain classes of editors (IP/newbie) and possibly limited in time, our tech people might find it easier to create it as a brand-new concept rather than an expansion of the block function with some limitations.
Create a brand-new concept to be granted to vandal fighters who meet certain criteria applicable only to IP's in newbies with a time limitation, it would be easy enough to keep track of them in a separate place other than the block log. We might even call it something other than a block. Call it a timeout. Vandal fighters who meet certain criteria have the ability to give IP's in newbies 24 hour timeout. There could be a timeout log, but someone running for RFA a few years later who has an entry in the timeout log will be viewed very differently than someone who has to explain a block.--S Philbrick(Talk) 13:42, 11 August 2015 (UTC)
  • I quite strongly disagree with the thinking here. The two admin tools that are almost uniformly discussed in every single RFA are blocking and deleting. We already know that the WMF has quite categorically said that only people who succeed in an RFA-identical or RFA-like path can have access to deleted content, so that isn't something that *can* be devolved. And blocking is very serious business; I genuinely believe that anyone who has access to blocking tools needs to have undergone significant community scrutiny. The same issue arises with the ability to edit the MediaWiki interface (there is genuine harm that can come from that), and speaking as a checkuser I'm not too thrilled with the way some admins hand out IPBE pretty routinely, as we've seen it abused, so I'd really not want to see that further devolved. There are things in the admin package that are there because *someone* in the community needs to be able to do it, not so much because only admins should do it (e.g., all of the stuff related to the Education programs, which are also available by separate rights, but sometimes admins have had to step in and address). So...if you can't devolve blocking and anything related to deletion, what would you devolve, and to whom would you devolve it? Risker (talk) 19:32, 11 August 2015 (UTC)
I'm not sure anyone is seriously talking about doing away with the Admin position entirely. And, in general, the type of "blocking" that is discussed in the "unbundling" concepts is the so-called "Vandal fighter" stuff, which had always been designed to be restricted to IPs and non-autoconfirmed accounts (possibly in combination with only strict time-limiting of blocking allowed). So, basically, that leaves what I think of as "AfD expert" rights – I guess if that kind of role is ever "unbundled" out, it would require an "AfD-like" process, though hopefully not one as rigorous as the current AfD process. Aside from that, "Article protector", for one, seems like it could be unbundled and granted to trusted editors. (I discuss "Article mover", separately, above...) But stuff like revdel, etc., should always be reserved to just Admin-Crats, IMO... --IJBall (contribstalk) 19:58, 11 August 2015 (UTC)
Hi IJBall. Blocking is blocking, and I don't think there is technically a way to do what has been proposed; I'm actually quite unconvinced it is needed or desirable. The target groups are also the same group of editors from whom all our new commuity members derive, and we already know that it's a lot easier to block or template someone than it is to talk to them and try to correct behaviours early. Lots of AIV reports are turned down for precisely these reasons. The "Article protector" button might possibly be worth considering, although given how very very few pages are actually protected, I am seriously concerned that we'd see it overused. Do we have reason to believe that things that should be protected aren't being protected? ( We have had a lot of overuse of rollback since was devolved, and it's improperly used in at least 30% of the instances I've seen on reviewing a lot of edits over the last few weeks. It only seems to ever be taken away from people who are otherwise not editing well, though.) The biggest problem right now at XfD is not that there are significant delays in deletion, it's that there aren't enough people participating in the discussions to make good decisions about deletion. Risker (talk) 21:12, 11 August 2015 (UTC)
I am not proposing this, but technically block, delete + protect, and edit mediawiki buttons can be unbundled in such a way that any candidate who wants to have access to any of them, must pass RfA indicated which buttons they are interested in. (If such system existed during my RfA, I would certainly indicate I am not interested in mediawiki edit button). This would pe perfectly in line with WMF requirements.--Ymblanter (talk) 10:15, 12 August 2015 (UTC)


risker I know that "unbundling" often sounds attractive in the abstract, but many of the tools are so interlinked that the concept falls apart when one tries to make a specific proposal. I also know, and sympathize with the Foundation's concern about viewing deleted material. That said, I think it is worth exploring the the possibility of some very limited tools available to vandal fighters. I do take your point that we may already be too quick to use templates or resort to blocks when conversation might be the better approach, so anything we do has to be treated as a pilot and monitored.
I was originally thinking of a time-limited (24 hour) limit to the block, but I am now thinking that it should be even shorter. I suspect there are some vandals might become bored and move on if blocked for two hours. (If they persist one can always get an admin involved.) Of course, a newbie caught up in a mistaken block will be unhappy, but I don't think the two hour block is likely to permanently turn away an otherwise capable new editor. Obviously, there are risks but I think it's worth an experiment to see if very limited blocking ability and very limited page protection tools might be appropriate short of a full RFA.--S Philbrick(Talk) 14:33, 12 August 2015 (UTC)
  • Block tool is very useful when a vandal fighter wants to block an IP address, after they vandalize after final warning, instantly, instead of reporting to AIV. Blockers should be able to block indefinitely but there should be very strict rules like block indefinitely only if the username is against the policy, if the account is vandlism only account. An experienced vandal fighter will know how and when admin blocks users in different situations. Blockers can also help out in AIV, ANEW and UAA. I have seen many instances where block tool would have been useful to stop further disruption to wikipedia. There are lot of non-admins who will be able to use this tool for the benefit of wikipedia. Users should get this tool if they are experienced in vandal fighting and have good AIV record. If blocker misuses his tools admin can easily remove the rights. Supdiop (Talk🔹Contribs) 17:03, 12 August 2015 (UTC)
Remember, you can't "indef" block an IP. And the usual way "Vandal fighter" is usually conceived is that there would be a limit on the length of the block a "Vandal fighter" could impose – IOW, the idea would be that a "vandal fighter" could immediately block a problem IP or unconfirmed account for 31 hours, which would then allow enough time for the "Vandal fighter" to check with an Admin or a group of Admins to see if the block should be lengthened by an Admin. --IJBall (contribstalk) 21:19, 12 August 2015 (UTC)
I am talking about a different version of it. The group I am talking about will have ability to block users indefinitely. They fight vandals and they also work with administrators in ANEW and UAA. The limited time block tool will restrict vandal fighters from helping ANEW and UAA, even though they are capable of doing it. I am supporting both, the vandal fighter and indefinite-period block right ideas. I am supporting indefinite block right because we give block tool for very trusted users. If we can trust them with a block tool why can't we trust them that they can block obvious username violations and vandalism only accounts (most common reasons for indefinite blocks). Of course they will go through the mini-hell to get the tool and I don't think they will go out on a blocking spree, just like someone did in 2006. Supdiop (Talk🔹Contribs) 15:36, 13 August 2015 (UTC)
Ah, I see. The issue with what you are proposing is that anyone granted the ability to block "indef" will almost certainly have to go through an RfA (or a similar process roughly as rigorous), and we're right back to where we started. I'm not saying it's not a good idea. I'm just saying that the hurdle to getting the kind of tools you are talking about are likely to be about as steep as an RfA. --IJBall (contribstalk) 16:22, 13 August 2015 (UTC)
Not necessarily. If this RfA would only give access to a block tool to block new editors (say below 100 edits and below 1 month tenure) and no access to the delete tool, I could reasonably expect that such light RfA would be much easier to pass.--Ymblanter (talk) 16:35, 13 August 2015 (UTC)
Yes, it would be much easier to pass if they can only block new users and ips. This proposal should get consensus from wider community. This will be a major step in making the RFA, no big deal. Actually IMO, this type of change is necessary because very few admins will be promoted in future, if we continue with no change. Supdiop (Talk🔹Contribs) 08:47, 14 August 2015 (UTC)

The distribution of RfA outcomes is stable over time

No need to support this, really; just entering some data into evidence. (Oppose if you think I goofed ;)

Distribution of RfA outcomes by number of RfAs.
Distribution of RfA outcomes by fraction of RfAs.
Historical average outcomes
Not now Withdrawn Unsuccessful No consensus Successful
28.5% 25.8% 3.1% 8.2% 34.4%

The distribution of outcomes has been quite stable since 2008 despite a drop in total candidacies by an order of magnitude. The decline of course is not news; the distributions may be. (For example, "not now" cases have some fluctuation but are back to historical average; early withdrawal rate hasn't changed.) That the process is losing candidates uniformly, rather than filtering out a certain type of candidate, suggests that the primary problem to solve is throughput. Opabinia regalis (talk) 20:34, 9 August 2015 (UTC)

Interesting data, Opabinia regalis. Thanks for pulling this together. Liz Read! Talk! 21:51, 9 August 2015 (UTC)
Yes, this is very interesting indeed. I would not have predicted the extent of consistency. We are losing candidates at all levels, but the overall pattern suggests that a perception of high expectations is equally discouraging the highly qualified and the poorly qualified – or alternatively, we could be seeing fewer highly qualified candidates because it is now more difficult to actually be considered highly qualified. --Tryptofish (talk) 22:13, 9 August 2015 (UTC)

Opabinia, thank you very much for producing these invaluable charts. They put into graphics what I have been saying for a long time (at least since WP:RFA2011) and demonstrate that the decline in candidacies is not neccessarily due to the environment of RfAs. If I'm not mistaken, even the volume of edits to the talk page at WT:RfA follows the same slope.

@Tryptofish:, I don't believe these graphs necessarily contribute to the theory that RfA has become more difficult to pass (I think this is a conclusio put about by us having occasionally stumbled on some isolated votes from newer voters who have misguidedly set their expectations disproportionately high, such as 20,000 edits or demanding FA and GA. Suc votes are not helpfull and fortunately don't appear to sway the opinions of other voters.

IMO we should be digging deeper for the cause of the decline, especially where I have offered to nominate many mature editors whose edting history long already surpasses even the most exagerated requirements but who have declined to run. Many cited the humiliating experience as a reason, while others (mainly with a background in traditional scholarly areas) were simply not interested in the drama that goes with admin work, or having to do such work at all. Kudpung กุดผึ้ง (talk) 23:40, 9 August 2015 (UTC)

The decline in numbers of words on the talk page is hard to really validate without a comparison set of projectspace talk pages unrelated to RfA, though. It'd be interesting to normalize these numbers by number of active editors. The fact that "not nows" track the overall trend fairly well surprises me - these are people otherwise oblivious to internal community dynamics, opinions of the RfA environment, recent successes or failures, perceived need for admins, etc. Opabinia regalis (talk) 02:44, 10 August 2015 (UTC)
It's not the number of words or bytes that matters. It's the number of posted mssages/contributions. I was able to extract this before someone messed up the tool that does it. So in fact although I can't represent it as graph now even if I could, the decline in intrest in participation on WT:RfA does in fact parallel the excellent charts you produced. Kudpung กุดผึ้ง (talk) 09:54, 10 August 2015 (UTC)
Kudpung, everything you said here makes good sense to me. I'm inclined to modify what I said above, along the lines of "suggests that various factors, some of which might include a perception of high expectations, a perception of RfA as becoming increasingly unpleasant, or a lack of interest in using administrative tools...." --Tryptofish (talk) 20:14, 10 August 2015 (UTC)
If one ignores the standards at RFA and the decline in the number of RFAs the results seem stable. But in reality a candidate who was considered well qualified a few years ago might now struggle to get through. Our trickle of candidates still includes some people who pass by acclamation and some are so close they need a crat chat. But the participants and potential participants are often aware of this and hold back if they don't think they meet the latest criteria. ϢereSpielChequers 20:29, 10 August 2015 (UTC)
This feels true, but is it really? You've been looking at this for a long time; any ideas on how the "not now" candidates who aren't paying attention to anything could stay so consistent a fraction of the total? That makes me think that whatever is going on is a lot less closely coupled to internal community dramas than we assume. Opabinia regalis (talk) 01:53, 11 August 2015 (UTC)
I'm pretty sure it is. We made some efforts a while back to embed messages in the process to deter people from standing who were the sort of NotNows we used to have - dozens of edits and days of experience. I fear we've overkilled and deterred some people who should qualify. But also the community has changed, we are now a fourteen year old project with a core of editors many of whom have been around quite a while - the average tenure of the core editors goes up 6 months a year. So my experience is that the tenure and edit count of the RFA candidates is steadily rising to reflect the community of active editors who are not admins, as an example this recent RFA was a candidate with ten months experience and nearly three thousand edits. I doubt that would have got you through RFA in recent years, but if you go back a few years it would have been taken more seriously and people wouldn't have opposed per NotNow, go back to the peak years for RFA and people became admins with less tenure or edit count - though in that case people would actually have looked at the edits rather than just oppose based on statistics. So what I think is happening is that the goalposts are changing and candidates are getting more cautious based on seeing other RFAs. Yes we get the occasional unprepared candidate who runs without looking at other recent RFAs, and we get people running on the basis that an unsuccessful RFA resets the clock and subsequent RFA 6 months later will be much more based on the 6 months since the unsuccessful RFA. ϢereSpielChequers 11:53, 11 August 2015 (UTC)
Changes in RfAs per highly active editor (>100 edits/month, blue) and WT:RFA posts per RfA (green).

Here you go, Kudpung: you're right that raw number of posts to WT:RFA has declined substantially over time, but the number of comments per individual RfA has gone up. The number of candidacies per highly active editor is an interesting way of expressing the data that gives a different sense of scale than the plain counts, and is maybe more to the point if your hypothesis is 'interest in being a candidate has dropped'. Opabinia regalis (talk) 01:53, 11 August 2015 (UTC)

Awesome. So the odds of passing an RfA are only 1-in-3 – no wonder nobody wants to volunteer for one... --IJBall (contribstalk) 06:41, 11 August 2015 (UTC)

Not really; that assumes all nominations start with an equal chance of success. The "not now" cases have effectively zero, so a viable candidate has over 40% chance starting out - probably more because the withdrawals are a diverse group and many of them would also be in the no-hope category. Opabinia regalis (talk) 07:56, 11 August 2015 (UTC)
I would say it is a little more complex than that. Odds are better for people with GAs or FAs, near hopeless for anyone whose answer to question 1 doesn't include between one and three things they would do with the tools and a rationale as to why their edits qualify them. It is also pretty difficult for anyone who wants to use the deletion button but has a personal non standard deletion policy (I expect the same would happen re blocking but it is much rarer to have a candidate who would block differently to other admins). Then there's the issue of skeletons in the closet, sometimes RFA finds things that weren't obvious, and sometimes the unexpected happens like when the nominator accidentally did an edit from his nominee/girlfriend's PC and the RFA imploded. ϢereSpielChequers 17:37, 11 August 2015 (UTC)
This is the problem with actually analyzing RfA: it's way too easy to get way too deep into the weeds. If you know nothing about a candidate beyond "not a 'not now' case", then they have a baseline ~40-45% chance of passing. Then you ask what else is most informative about the end result. Your comment about 'account age' makes me think it's worth forgetting about absolute 'standards' for a bit and instead looking at where candidates stand relative to the distribution of the same features among the core contributors at the time of their RfA. One way to get constant distributions despite declining candidacies is if the community always wants people who are perceived as 'above average' by a certain amount. That's a lot more effort than magic data fairy dust and a few minutes making the graph look pretty, though. Opabinia regalis (talk) 20:24, 11 August 2015 (UTC)
Could the green line in the newest graph just be illustrating an increase in unproductive discussion at WT:RFA? --Tryptofish (talk) 23:53, 11 August 2015 (UTC)
There's any other kind of discussion there? ;) "Posts per RfA" just means the units; it's not posts specifically about each particular candidacy. So the increase could be that each RfA generates more discussion now, or it could be unproductive dead-air filler. Opabinia regalis (talk) 00:25, 12 August 2015 (UTC)
Once again, Opabinia, I thank you most sicerely for producing a visual display that confirms my findings. Taken together now, your graphs clearly demonstrate that the overall interest in general in all things adminship has waned significantly. The increase in participation on each individual RfA is probably due to the fact that there are far fewer opportunities to vote on an RfA these days. I think it's not a great help to try and extrapolate artificial stats such as the rate of chance of passing an RfA, the variables are far too numerous and in any case, most candidates pass who should , and most of them fail who sho should not pass, and as demostrated, both types have been falling at the same rate.Thst's all we really need to know. Stats reveal tends but thet do not present solutions and it's up to us to find out what is really the cause of this general downward trend rather than clutching at unreliable speculation. Kudpung กุดผึ้ง (talk) 12:23, 12 August 2015 (UTC) .
Minor comment: the green line here isn't participation (votes or comments) on the RfA itself, but posts at WT:RFA. There's no doubt that votes per RfA have gone up, but conversations about 'adminship in general' are more ambiguous. (I suspect there's a feedback effect here: it is definitely different being the only candidate than being just one out of the 10-15 running at a time. Nobody in the pre-rollback RfA era had time to ask or read 20+ questions on 10+ RfAs, and nobody running had to make time to respond to them.)
Saying that the candidates who should pass do and those who shouldn't don't is assuming the conclusion ;) The idea that every one of the ~4500 on-wiki RfAs to date is its own unique special snowflake is IMO overinterpreting noise, but I don't have time to actually do per-candidate analysis.
A question: if you believe that interest in being a candidate is declining, and that this is a problem, why support this stuff up above about hard minimum requirements? Those are great solutions to the opposite of the problem we have. Opabinia regalis (talk) 18:55, 12 August 2015 (UTC)
That section above does not enumerate any minimum requirements. Such an issue is dealt with at each individual RfA depending on the criteria of anyone who turns up to vote. I do believe it has been firmly established however that the majority of voters now and even as far back as WP:RFA2011 and earlier are not RfA regulars - for a great many it is indeed the only RfA vote they are likely to make and there is no way we can speculate on their motives. This leans towards the hypothesis that RfA is populated by voters who may not fully understand that RfA is a due process that should be kept as equitable and friendly as possible whatever the outcome and how their votes impact not only on the RfA but on the reputation and purpose of adminship in general. The answer therefore is to do what the other Wikipedias do: introduce minimum qualifications for the voters. However, this is a bit lop-sided in a situation where no qualifications are demanded from the candidates. Kudpung กุดผึ้ง (talk) 02:43, 13 August 2015 (UTC)
I linked the top-level section for convenience's sake, but I mean the proposed minimum requirements in the subsections. Which subset any one person prefers doesn't matter much; if the problem is a lack of people willing to be candidates, then it's hard to see how adding more restrictions on who can be a candidate is a solution to that problem.
From a couple of months ago: [6]. 2015 voters participated in an average of 2.7 RfAs out of 10 completed ones at the time, just under half voted in just one, the singleton voters were almost all identifiably experienced users, and only a handful could meaningfully be described as new. The 'new standards every time' problem does not seem to be a current one, and there are just not enough supposedly unqualified voters to warrant more rules about them. We certainly can speculate about one subset of singleton voters: they take little interest in RfA in general but have an opinion about a particular candidate. This is useful input, at least as much as comments from the 'regulars' are. Opabinia regalis (talk) 08:46, 13 August 2015 (UTC)
@Opabinia regalis Re: "Supporting hard minimum requirements? Those are great solutions to the opposite of the problem we have." If our problem is that there are hundreds of editors who would qualify under current de facto criteria but who are deterred by RFA's reputation, then codifying parts of that criteria should encourage some of them to stand. That's a similar issue to the Germans getting more editors to vote in RFA by having written as opposed to unwritten minimum edit requirements for voters. But more importantly, RFA has a problem of standards inflation, it has shifted its focus from do this editor's edits indicate they would make a good admin to arbitrary, barely related but easy to use metrics such as edit count and tenure. Fixing such criteria so they can only change by consensus will stop a longstanding and damaging process whereby if 30% of the community oppose over something then that de facto is an admin criteria, even if 70% thought the change unhelpful. ϢereSpielChequers 14:21, 14 August 2015 (UTC)

If hard criteria for candidates were to be established, they must nevertheless not be allowed to become a bright line between pass and fail. Whereas the support section is basically a popularity vote, the oppose section is where serious research has been made that may prove that in spite of being on the bright side of a bright line, other objective reationale can be introduced to demonstrate the candidate's unsuitability. This is the way it actually works and still works quite well without codified criteria. There is however, IMO, still a very stong argument for qualifications for voters. Kudpung กุดผึ้ง (talk) 07:04, 16 August 2015 (UTC)

WSC, the fact that de.wiki has both minimum qualifications and higher participation in no way implies that one causes the other, and the hypothetical causal explanation strikes me as wishful at best. Things that are objectively not problems with the en.wiki RfA process: 1) too many obviously unqualified candidates; 2) too many obviously unqualified voters. I forget now if I already posted this or just intended to, but codifying minimum criteria is just inviting people holding the top ~70% of edit count/tenure/etc. standards to vote the lower ~30% out of consideration, which has obvious ratchet problems considering that nothing stops anyone from using higher standards in their votes. Opabinia regalis (talk) 23:15, 19 August 2015 (UTC)
  • The high turnout on German RfAs is due to a feature of German culture and has little to do with the way their RfAs are structured.Kudpung กุดผึ้ง (talk) 21:47, 21 August 2015 (UTC)
    • That's an interesting theory, and I don't know German culture well enough to dispute that. But it doesn't mean that having written as opposed to unwritten rules doesn't increase turnout, it just means we now have two possible reasons for that higher turnout. It would be interesting to get a study done of RFA turnouts across different wikis - assuming there are enough where there is a clear but modest voting threshold to do a meaningful comparison. ϢereSpielChequers 12:19, 25 August 2015 (UTC)
      • I have to say that I also have serious doubts that having minimum standards for voters, worthy idea though it is, has any connection to RFA turnout. That would mean that there are a significant number of voters with more than 200 edits who don't take part in RFAs because voters with less edits than them are allowed to take part and I don't see how one follows from the other. Spanish Wikipedia, for example, has minimum criteria which is: registered account, at least 30 days old, with at least 100 mainspace edits. They've only had 5 RFAs this year and all of them saw less than 100 people participate. Minimum standards is a good idea, but I don't see the evidence for a causal link between that and turnout, though it would be good to have a look at how other Wikis do it. Maybe Kudpung or someone else has already done similar research? Valenciano (talk) 12:53, 25 August 2015 (UTC)
I made a complete comparison table of many Wikis' RfA methods in WP:RFA2011/VOTING and they haven't changed, that's why I keep saying it is wrong to dismiss that enormous resource of research as being outdated. That said, comparing them is like comparing apples with oranges. The Germans have a distictly different cuture, I know because I spent 20 of my most formative years there - and I'm not going to provide RS for it, you'll have to read some (German) books on sociology; such a comparison will only lead to a false dichotomy. The wierdest system is the Italian one - theirs is as chaotic as the traffic in Rome.
There is absolutely no proof that the introduction of rules for voters would have a negative effect - its all about us wanting to stay within our comfort zone of better the devil we know than the devil we don't, and that's why all attempts at reform are doomed to fail, and if the opposers can't find lany objective rationaes they just resort to shooting the messengers. Kudpung กุดผึ้ง (talk) 13:35, 25 August 2015 (UTC)
Thanks Kudpung, that is useful data and I doubt it's changed much. Other Wikis will probably have similarities to ours but will reflect the local culture at times. For example Spanish ones are open for 15 days and only one user in the last 3 years has passed with less than 10,000 edits, which seems to be the de facto minimum there. I was looking at the Swedish Wiki's rules today and they simply require a registered account and a user to be "established and fairly well-known to other editors" which is more or less how we do it, and I definitely agree that having explicit minimum standards is better all round, whether it will increase turnout is another issue, but can't see any negatives in doing it and it will cut out sockpuppetry and the like. Valenciano (talk) 18:29, 25 August 2015 (UTC)