Wikipedia talk:Reflections on RfX/Archive 1

The discretionary zone is 70–79%
As an alternative to the proposal just above.


 * Support
 * 1) As proposer. I think that this might better reflect community perceptions. --Tryptofish (talk) 21:53, 7 August 2015 (UTC)
 * Struck, per Kudpung, below. --Tryptofish (talk) 20:43, 8 August 2015 (UTC)


 * Oppose
 * 1) It should stay at 70 - 80. Lowering the discretionary zone is the wrong way to lower the bar. Lowering the bar should be achieved by encouraging votres to refrain from applying disproportionately high criteria for edit counts, and any other countable-type actions. --Kudpung กุดผึ้ง (talk) 04:42, 8 August 2015 (UTC)

The discretionary zone is 2/3rd to 3/4th
As an alternative to the two proposals above.

Based upon the history of passes/fails, I think that the range should be two out of three at the bottom (2/3, or 66.66...%) and three out of four at the top (3/4, or 75%). It appears that this will reflect the fact that historically the oppose comments have to be really good to fail someone who is a little under 3/4 and the oppose comments have to be really bad to pass someone who is a little above 2/3. Plus, 2/3rd to 4/5th is much more resistant to minor modifications, and feels less like a bright line with +/- 0.5% precision. --Guy Macon (talk) 01:04, 8 August 2015 (UTC)


 * Support
 * 1) As proposer. --Guy Macon (talk) 01:04, 8 August 2015 (UTC)


 * Oppose
 * 1) I take this to mean the same thing as 66.7–80%, which extends the range down too low. --Tryptofish (talk) 20:48, 8 August 2015 (UTC)
 * Typo. I meant 66.7-75, which of course is still extends the range too low in your opinion. How many RfAs have been between 66.7 and 70%, and what percentage passed/failed? --Guy Macon (talk) 00:07, 9 August 2015 (UTC)
 * As you correctly say, I think that places it too low. That said, I think the tendency has been for 75–80% to generally be successful, and (just from memory, not encumbered by actual facts) I suspect that 66.7–70% (as well as 66.0–66.69%!) have mostly been unsuccessful. --Tryptofish (talk) 00:42, 9 August 2015 (UTC)

Do you mean 3/4 (75%) or 4/5 (80%) as the top of your range? Right now both are used in the text. Happy Squirrel (talk) 19:40, 8 August 2015 (UTC)
 * Comment


 * Typo. I meant 3/4 (75%) at the top end. I fixed it. (The phone rang while I was editing, and I had a brain fart and forgot to prufreed my edit before saving.) --Guy Macon (talk) 00:07, 9 August 2015 (UTC)

RFAs that end in the discretionary zone should go to a crat chat
Many of our crats have been in post for over a decade, recent crat chats have shown significant variance in views amongst crats. Historically only a minority of RFAs end in or near the discretionary zone, but following recent precedents crat chats should be the norm for RFAs that end in the discretionary zone.


 * Support
 * 1)  Ϣere  Spiel  Chequers  08:44, 7 August 2015 (UTC)
 * 2) Kudpung กุดผึ้ง (talk) 10:36, 7 August 2015 (UTC)
 * 3) Kraxler (talk) 12:11, 7 August 2015 (UTC)
 * 4)  Kharkiv07  ( T ) 15:00, 7 August 2015 (UTC)
 * 5) AliveFreeHappy (talk) 17:12, 7 August 2015 (UTC)
 * 6) I guess I regard this as "best practice", rather than as something that is always mandatory. I can imagine (infrequent) circumstances when the RfA raw numbers are in the discretionary zone, but it is very obvious to any Crat that some significant amount of !votes on one side should automatically be discounted. --Tryptofish (talk) 18:17, 7 August 2015 (UTC)
 * 7) Fylbecatulous talk 20:22, 7 August 2015 (UTC)
 * 8) Valenciano (talk) 11:08, 9 August 2015 (UTC)


 * Half-Oppose
 * 1) Half-oppose; I don't think all marginal RfXes merit a full discussion to get an acceptable outcome. In some cases, you'll have one side much more confident in its stance than the other. Jo-Jo Eumerus (talk, contributions) 20:03, 7 August 2015 (UTC)
 * 2) I'm kind of agreeing with Jo-Jo, but at the same time, there are some that might go below the discretionary zone that still deserve a crat chat. The numbers 70-80% need to be vague, and have been for good reason: to allow for the fact that not all votes are the same.  I will say I agree that more crat chats are desirable, and agree with the goal of this idea, just not the exact method.  My example with Theo, linked above, is a good example, where 74.5% didn't pass, and I think some Crats would have passed, others would not, so it is the perfect example of where a chat should have been done.  I don't fault Maxim and the result could have been the same even with a Crat chat, but wished it had been done.  Dennis Brown - 2&cent; 20:18, 7 August 2015 (UTC)
 * The issue is not about percentages - but whether the person has a clear consensus that they will abide by Wikipedia policies and standards. If the "opposes" are sufficient in weight that there might be a problem,  then such opposes are of value in determining the will of the community.  If the "opposes" are heavily "I personally hate the fellow" then such opposes should be of much lesser weight.   We should stick to whether the person is shown to follow Wikipedia policies and guidelines, or whether they have been shown to possibly not follow them.  Collect (talk) 12:24, 9 August 2015 (UTC)


 * Oppose

The discretionary zone is 70–80%

 * Support
 * 1) I actually chuckled at myself when I saw your oppose above, because what you are saying is really what I intended all along. Although I'll admit that a single percentage point is probably too trivial to be worth this much discussion, I entirely agree with you that the existing perception of the community is not the 70–75% proposed above, and I agree with you that we at this discussion ought not to be lowering the zone. --Tryptofish (talk) 20:43, 8 August 2015 (UTC)
 * 2) Agree, with the caveat of the following comment by Tryptofish; bureaucrats always have discretion to review any RfA that appears unusual. Fylbecatulous talk 21:27, 9 August 2015 (UTC)


 * Oppose

It occurs to me that a very good case can be made that the community has always entrusted the Bureaucrats to exercise discretion from 0% to 100%. Although in practice it is very uncommon to fail an RfA when the raw numbers are in the low 80s or to pass one that is in the low 60s, in principle the Crats are entrusted to consider unusual circumstances regardless of where this discussion might locate "the discretionary zone". The concept is still useful to the extent that it indicates the range in which there very likely needs to be a close examination of the RfA discussion, but we may be gilding the lily by trying to define precise numerical boundaries. --Tryptofish (talk) 00:48, 9 August 2015 (UTC)
 * Comment
 * Agree - IIRC, there have been cases in the past where "off-wiki" canvassing was deemed to have tainted the process, etc. Collect (talk) 12:19, 9 August 2015 (UTC)

The discretionary zone is 70/75%
Crats have some leeway to decide on weight of argument factoring in things like voters who acknowledge their vote as weak or moral, but also discounting obvious trolls. RFAs that end in a "discretionary zone" may be closed as either successes or failures at the discretion of the crats.

There has been occasional debate as to whether the discretionary zone is 70-80% or 70-75%, convention seems to support the latter but the debate should probably be settled.


 * Support


 * 1)  Ϣere  Spiel  Chequers  08:44, 7 August 2015 (UTC)
 * There should be discretion when deciding whether something is discretionary (on the basis of IAR), but in general, the discretionary zone is informally 70-75% anyway and it makes sense to state this explicitly. seems to agree, having recently made this edit. — Bilorv(talk)(c)(e) 09:19, 7 August 2015 (UTC) See below. — Bilorv(talk)(c)(e) 17:50, 7 August 2015 (UTC)


 * Oppose
 * 1) Mine was 76.4% at the time of closing and went to a 'crat chat (for good reason) but was determined to be successful (also for good reason), so I would be very wary against imposing any hard number on what falls within the realm of bureaucrat discretion and what doesn't. ☺ ·  Salvidrim!   ·  &#9993;  16:24, 7 August 2015 (UTC)
 * 2) My edit was clearly not intended to place any hard bounds on the range. That was both explicitly stated in the edit summary of the diff you brought, as well as in the change of the text from "rule of thumb" to "historically". The former can be interpreted as proscriptive; the latter clearly is not. In accord with that edit, and my statement at WP:BN, I would continue to oppose any hard and fast definition of what the "discretionary" zone means, and allow burecrats to continue to use their judgment in each and every case. I understand we should not necessarily be opposing at this stage, but as my edits are being interpreted in a way diametrically opposed to what I meant, I felt that the clarification was necessary. Thank you. -- Avi (talk) 17:24, 7 August 2015 (UTC)
 * yes, I fully understand and agreed with your edit. I think the confusion is arising from this particular discussion: I read it as "this proposal is to clarify that the discretion of 'crats usually only affects the outcome when supports consist of between ~70-75% !votes, rather than the 70-80% often quoted" rather than "this proposal is imposing a strict limit: only 70-75% is within the discretion of 'crats". I think one of us is !voting the wrong way and after re-reading the text more thoroughly I'm pretty sure it's me in the wrong. I've stricken my comment and would note that the diff I gave already fixed the issue I thought needed addressing anyway. — Bilorv(talk)(c)(e) 17:50, 7 August 2015 (UTC)
 * I also seemed to miss the fact that it was actually this edit that changed "80" to "75". I am such an idiot. — Bilorv(talk)(c)(e) 18:03, 7 August 2015 (UTC)
 * Strike the above; you certainly aren't an idiot! You just happen to be a human being, like the rest of us [[file:face-smile.svg|28px]]. -- Avi (talk) 18:30, 7 August 2015 (UTC)
 * 1) I nom'ed an RFA that was just under 75% and there was no Crat chat, Requests for adminship/Theopolisme 2, and I can't remember the name, but at least once we have promoted someone with less than 70%. I don't think you can use percentages as a rule because that assumes all votes are equal.  They are not.  I can see scenarios where 65% might justify a crat chat, this hard limit would prevent that. I can also see 76% cases where there is obviously no consensus, depending on how you weigh the votes, canvassing, etc.  Rules based on numbers take away the ability to judge consensus.  I wonder if Maxim would have closed Theo's RfA the same today, or how others would have treated it? Dennis Brown - 2&cent; 20:07, 7 August 2015 (UTC)
 * 2) Consensus is determined by quality of arguments, not number of votes. While 70-75% is a useful guideline to determine when we're in a discretionary zone that warrants a full crat chat, there is no need to set this in stone and significant reason not to. As one example, momentum matters. Consider the following scenario: An RfA garners 100 supports and 5 opposes early on, but someone digs up a particularly convincing diff to oppose on the last day that leads to a final tally of 121 supports and 30 opposes. This means that the candidate finishes with a percentage above 80%, but the voting after the new oppose rationale was made available was 21 supports and 25 opposes. Is there consensus to promote? It depends quite a bit on the exact arguments made, and that's exactly what a crat chat is for. ~ RobTalk 20:33, 7 August 2015 (UTC)
 * 3) Now that there is an oppose section, per my comment in the discussion below. --Tryptofish (talk) 20:45, 8 August 2015 (UTC)


 * Discussion


 * Are there figures (or better yet, a graph) for how many RfAs have passed/failed at 69%, 70%, 71%, etc? --Guy Macon (talk) 11:15, 7 August 2015 (UTC)
 * I don't know, but I have seen it said that only former admins have been passed RFA with below 70% support in many years and that no-one or almost no-one has been failed at over 75% in many years. We could of course take out crat discretion and set the pass mark at a precise figure such as 72.5%, but I'm not seeing anyone calling for that.  Ϣere Spiel  Chequers  11:46, 7 August 2015 (UTC)
 * A non-reconfirmation example is Requests for adminship/Krimpet. Jo-Jo Eumerus (talk, contributions) 15:07, 7 August 2015 (UTC)
 * I don't think that the discretionary range can be expressed in such rigid absolute numbers. After all, RfA is not a headcount, or is it? Kraxler (talk) 15:18, 7 August 2015 (UTC)
 * How many RfA's have failed at over 75%? I could swear at least one of the ones that recently went to Crat Chat, and failed, was about 75% or above... --IJBall (contribs • talk) 16:10, 7 August 2015 (UTC)
 * I have this data back to 2008 on my other computer, but I can tell you the highest-scoring unsuccessful case was Requests for adminship/CWY2190 from 2008 at exactly 75%, and didn't even get a crat chat. I believe the most recent success under 70% was Requests for adminship/Davemeistermoab (68%, in 2009). Opabinia regalis (talk) 16:31, 7 August 2015 (UTC)
 * Procedurally, I'm concerned that the instructions say not to have an "oppose" section, so maybe there should be an alternative section for 70/80%. As for my own opinion, I'd prefer to see the range as being 70–79%, and my subjective reading of community sentiment is that's what the community thinks. Actually, I think that the community kind of goes by the color scheme used by User:Cyberpower678/RfX Report. --Tryptofish (talk) 18:14, 7 August 2015 (UTC)


 * I've made several suggestions on this page and this is both the least popular and the only one that seems to have been enacted anyway. Happy to have it reworded to something like the core of the crat discretionary range is 70-75%, but if we have to go back seven years to get a fail at 75% then 75-80% is usually a pass. Of course canvassing and dramatic last minute trends could occasionally give RFAs where a bold crat could uncontentiously go outside that discretionary zone. But it hasn't happened in the last few hundred of RFAs  Ϣere Spiel  Chequers  06:58, 8 August 2015 (UTC)
 * I find it extremely unfortunate that the 'crats (yes, the 'crats) have slowly but surely pushed up the "discretionary range" to the point that it is as difficult to get a clear pass at adminship as it is at RFB. Both are now at 80%. Please stop that and revert to 75%; I'd actually prefer to see this dropped down to 65-70% but I know that's a pipe dream. Incidentally, "discretionary zone" does not mean the same thing as "must have a crat chat".  Risker (talk) 01:18, 9 August 2015 (UTC)
 * It seems to me that what you attribute to the crats is really the view of the community. There is a widely held perception (with which I agree) that admins who passed in "the good old days" were passed too easily, and now (insert here the standard request for a new method of desysopping) we are stuck with some admins who lack the trust of the community. I say this as someone who has recently supported candidates who also had significant opposition; even though I have sometimes disagreed with those who opposed, I agree with the broader concept that the standards, today, ought to be high. --Tryptofish (talk) 01:35, 9 August 2015 (UTC)
 * Lack of community trust today and successful RFA from some time ago are two different concepts. Some of the admins who have since been desysopped for cause had far, far higher support levels, and in some cases almost no opposition at all; there was no bureaucrat discretion involved on those RFAs. I don't know why anyone comes to the conclusion that "olden days" admins passed too easily; the fact of the matter is that the overwhelming majority of administrators who remain active and using their admin bits are the very people who are constantly being berated as having "passed too easily", no longer having the "trust of the community", and so on. If we are all such bad admins - and let me be clear, the comments at WT:RFA over the last few years really make it sound like there are a number of very vocal people who believe that we *are* all bad admins - then desysop us all. The encyclopedia will be full of spam, vandals won't get blocked, behavioural problems won't be addressed, moves won't be made, checkusers won't be run, suppression of abuse won't happen....but go ahead. It just plain gets insulting that everytime I come to a page about RFA, all I see is how horrible administrators are. Not once do I see "the overwhelming majority of active admins do their jobs just fine" - why is that? I still don't think that 'crat chats are necessary in a lot of the "discretionary" cases; however, the graphs below by demonstrate that my comment was incorrect, and I have struck it.  Risker (talk) 20:39, 9 August 2015 (UTC)
 * I take your point about how it is unpleasant when admins are all lumped together, and I hope that what I said did not come across that way. In fact, this discussion is reminding me of the discussion about WP:CDARFC, where I felt like users were simplistically divided into opposing camps of "entrenched admins" and a "mob with pitchforks". Speaking just for myself, my experience is that the large majority of admins, from all time periods of RfA, are very helpful and responsible, and often get abuse for their good work. And that ArbCom has gotten much better than it used to be at dealing with the occasional bad apple. But also that there are still a couple of long-time admins who do some stuff that makes me cringe. When newly-RfA'ed admins do something objectionable, it tends to get dealt with at ArbCom, but there are a few old-timers who are problems. That's not a valid reason to dump on all admins. But, based on what I see from the community, I think that it is reasonable, in principle, for the community to hold RfA candidates to a high standard. --Tryptofish (talk) 21:45, 9 August 2015 (UTC)

Flexibility is a feature, not a bug
Currently, the RfA process has very few formal rules for participation. Anyone may run and anyone may vote. The fact that this sometimes attracts participants too inexperienced to understand the process is a minor cost of allowing the community the flexibility to consider unusual cases. Proposals that advocate formalizing criteria for participation risk losing this existing benefit in search of improvements that are entirely hypothetical.

Support
 * 1) (as proposer) Opabinia regalis (talk) 18:37, 7 August 2015 (UTC)
 * 2) Well, sort of; it's not so much a feature, but a reflection of the diversity of opinion. Though people of all stripes can agree that bias in articles is a bad thing, for instance, voting and the criteria for voting at RfA rest on our various ideological convictions. Alakzi (talk) 09:58, 8 August 2015 (UTC)

Oppose "Comment"
 * 1) We already have WP:IAR to handle unusual situations. --Guy Macon (talk) 20:16, 7 August 2015 (UTC)
 * Yeah, but getting 100-200 people to IAR at once?
 * At one point in time, lurking was the standard way of joining a new online community. Under that system it's the lurker's job to self-evaluate and decide "I think I know enough about this to participate". Encouraging people to outsource that evaluation to some arbitrary predefined standard is not going to improve the quality of participation or make the environment more welcoming. Opabinia regalis (talk) 21:13, 7 August 2015 (UTC)
 * I am having trouble understanding "Yeah, but getting 100-200 people to IAR at once?". Are you saying that a hundred people will run for admin at a time? Or are you saying that this is not a requirement for running, as it says in the description, but rather a requirement for voting? I don't think anything we write will change the fact that people vote as they choose. --Guy Macon (talk) 02:00, 8 August 2015 (UTC)
 * I am having trouble understanding whether she means candidates or voters. Kudpung กุดผึ้ง (talk) 07:28, 8 August 2015 (UTC)
 * Both, of course. There are proposals on this page about instituting formal requirements for both voters and candidates. I think both of those things are bad ideas. Maybe opposing-by-counterproposal is too oblique :) Opabinia regalis (talk) 08:57, 8 August 2015 (UTC)
 * , there is no 'of course' about it - already two users were confused by your coments. And if they are such bad ideas, where is the foundation for your hypothesis? Again, the 'dated' but extremely serious research at WP:RFA2011 will tell you more. I think  has taken an excellent initiative in starting this page and I think it would be helpful to keep it on track. Kudpung กุดผึ้ง (talk) 10:43, 8 August 2015 (UTC)
 * I looked at this a couple of months ago and found no evidence of significant numbers of very inexperienced voters currently participating - I notice you sometimes gently shoo them off, which is good, but there's just not enough of them to matter. In order to validate proposed minimum standards for candidates, there would have to be evidence that new admins close to or under the thresholds are noticeably worse in some way, which there isn't. In any case the idea that statutory minimums will avoid inflation of unwritten criteria is not consistent with experience.
 * As for what's on track, this appears to be structured in a 'principle of deferred judgment' setup, in which case offering a positively framed alternative rather than objecting to existing proposals is explicitly the point. You could object to the kittens, I suppose, but then you'd have to follow Guy and go on the record calling kittens jerks. Even this one ;) Opabinia regalis (talk) 22:04, 8 August 2015 (UTC)
 * 1) Disagree, oppose, whatever you want to label it. Times change.  Wikipedia isn't the same website it was 5 or 10 years ago.  A principle that says "we can never change" is dead in the water around here.  Consensus can, and will, change.  Dennis Brown - 2&cent; 19:15, 10 August 2015 (UTC)

The top of the de facto discretionary zone is 75%
Since there is a lot of guessing in the above sections, here is some data. No RfA with over 75% support has failed since at least January 2008. "Discretion" that has not been used once in seven years, despite 36 opportunities to do so, is de facto acceptance of a 75% upper bound. Opabinia regalis (talk) 20:26, 9 August 2015 (UTC)


 * This is very helpful evidence, so thank you for it. At the same time, the fact of de facto practice does not mean that thought should not be put into making the close decision anyway. Please also see what I said under "Comment" at, above. --Tryptofish (talk) 21:49, 9 August 2015 (UTC)
 * In fact, I'll follow this empirical data with an example drawn from entirely hypothetical speculation. But please bear with me. Let's imagine a hypothetical RfA in which, up to about the last hour, there are 200 supports, and 0 opposes or neutrals. Then, about 1 hour before the scheduled close, an editor posts a thoughtful oppose with a well-documented illustration of a serious problem that had previously been overlooked. It's something very bad. In the remaining hour, several editors who had supported change their positions to oppose. The Crat comes, and decides to extend the RfA for a few more days. That's reasonable, in spite of the raw numbers being well above the discretionary zone. Then, after a few more days, many editors have changed their opinions to oppose, so that the percentage has become 81%. It hasn't gone lower, because there were so many supports to begin with, and many editors appear not to have come back; there are no comments from supporting editors saying that they looked at the new evidence and still support. It would be appropriate for the Crats to determine that there was "no consensus", even though they weren't in the discretionary zone. Admittedly, I'm creating quite a hypothetical there! But I think it shows that discretion always has to be used. --Tryptofish (talk) 21:58, 9 August 2015 (UTC)
 * You're right, but I think of the term "discretionary range" as being narrower than that. I think we (almost?) all agree that the crats have generalized discretion to figure out how to handle obvious abnormalities (late-stage revelation of copyvio or sockpuppeting, offsite harassment, mass canvassing issues, etc.). Your observation that discretion applies from 0 to 100% is right on that count. But the numerical "discretionary range" seems to be used mostly for otherwise routine cases that happen to be divisive for some reason (lack of content creation seems to be a recurring theme here; former admins/reconfirmation cases is another). Opabinia regalis (talk) 02:29, 10 August 2015 (UTC)
 * thank you once again for your excellent charts. 's theory is superb but such a situation is rare, if indeed it ever happened. An RfA that I once co-nom'ed came very close to this hypothetical example. It failed. More to the point, it failed due to somehing that wasn't even anything to do with Wikpedia and which normal research before voting or nominating could not have turned up. Fortunately that editor is still with us, but I'm sure their RfA left them somewhat bitter and disenchanted. Kudpung กุดผึ้ง (talk) 11:27, 10 August 2015 (UTC)
 * Thanks, and I agree that what I theorized is something exceedingly rare in practice. --Tryptofish (talk) 20:04, 10 August 2015 (UTC)

The percentages and "discretionary zone" are descriptions of what kind of outcome is expected at a given tally, not hard rules as to which outcome is allowed.
That N % is below the normal discretionary zone does not mean that anybody with N % support is automatically unfit to be an administrator; it just means that the usual expectation is that they don't get promoted. A promotion at this percentage may still be justified if the opposition is flimsy or not overly concerned, even if most cases won't result in promotion. Ditto for RfXes above the usual discretion percentage; if the support is heavily disclaimed or the concerns raised extremely serious, a non-promotion may be justified even if most cases would result in it. The assessment of such exceptional situations - on top of the usual discretionary cases - is delicate and comes with high expectations for fairness and judgment; hence the RfB process is restrictive.


 * Support
 * 1) (as proposer) Jo-Jo Eumerus (talk, contributions) 13:30, 9 August 2015 (UTC)
 * 2) Well said. --Tryptofish (talk) 22:06, 9 August 2015 (UTC)
 * Yes, but do we have a significant history of such RfAs that require this to be taken in to policy? --Kudpung กุดผึ้ง (talk) 23:56, 9 August 2015 (UTC)
 * My understanding is that this discussion page wasn't really started with the intention of changing policy, but more for the purpose of giving the Crats some community input about present-day community expectations. --Tryptofish (talk) 20:09, 10 August 2015 (UTC)

Bureaucrats who participate in an RFA should recuse from the ensuing 'crat chat II
Participation here means nominating a candidate, voting in support or in opposition, or otherwise expressing an opinion about the merits of their candidature at the RFA, or at another venue. Normal "clerking" of the RFA to ensure compliance with the rules, formatting conventions, or to maintain order does not introduce any COI that would require abstention from closing the request, or participating in the 'crat chat.

Support
 * 1) Offering as alternative to above proposal, since I am not comfortable with the commentary accompanying that one. Abecedare (talk) 21:41, 7 August 2015 (UTC)
 * 2) Something like this is standard practice already, isn't it? Jo-Jo Eumerus (talk, contributions) 21:43, 7 August 2015 (UTC)
 * 3) I support this, and I think that it is correct in ways that the first proposal is mistaken. --Tryptofish (talk) 21:50, 7 August 2015 (UTC)
 * 4)  Ϣere  Spiel  Chequers  23:02, 7 August 2015 (UTC)
 * 5) For procedural reasons, and to remove any appearance of impropriety, this is obvious. Dennis Brown - 2&cent; 19:12, 10 August 2015 (UTC)
 * 6) With Collect's caveat, but I don't think there's a need to oppose because of that. Callanecc (talk • contribs • logs) 03:03, 11 August 2015 (UTC)
 * 7) Per  with 's caveat. What we want is a balance between trust and the protection against an appearance (let alone actuality) of impropriety. -- Avi (talk) 04:23, 11 August 2015 (UTC)

Oppose
 * 1) "At another venue" is ill-constructed here.  I trust that  a bureaucrat who told George Gnarph to "stand for sysop because you are pretty good on Wikipedia" would not use that word of encouragement to disqualify himself or herself from being a true impartial viewer of the results of a later RfA, and if the wording means "on IRC, on any other site, etc." then it is really outside our reasonable purview or ambit here. Collect (talk) 12:46, 9 August 2015 (UTC)

Negative assertions about candidates should have diffs
Supports:
 * 1) The diffs can be elsewhere on the page, in another oppose, a question, comment or neutral vote. But without diffs it is hard for an oppose not to be a personal attack.  Ϣere  Spiel  Chequers  07:26, 7 August 2015 (UTC)
 * Yes, but as WSC says above, there's no need for every one to repeat them. --Kudpung กุดผึ้ง (talk) 08:22, 7 August 2015 (UTC)
 * 1) I remember this exactly question being posed years ago (and I'm sure it's been posed many times before and after). The objection was that supporters are never asked to back their opinions with diffs, but I think that's actually the way it should work. I very much still hold to the belief that adminship is no big deal, and that most editors should be promoted absent a good reason not to. Someguy1221 (talk) 09:23, 7 August 2015 (UTC)
 * 2) If there isn't any evidence it didn't happen.  Kharkiv07  ( T ) 15:00, 7 August 2015 (UTC)
 * 3) I support this with the caveat that the diffs don't have to be repeated, and also that a discussion link instead of a diff per se is just as useful. Also, I would not go so far as to say that a diff-less rationale is automatically a reason to discount an opinion – it's kind of like WP:V, in that the need for evidence (like the need for an inline cite) increases as a function of it being non-obvious. An example of what I mean by that is that "oppose, because the candidate has only been editing for one day" is a negative assertion, but one that hardly needs a diff. --Tryptofish (talk) 18:10, 7 August 2015 (UTC)
 * 4) Andrevan@ 21:10, 7 August 2015 (UTC)
 * 5) Agree, but other types of links can be sufficient depending on the circumstances (talk pages, the history of articles, etc.). And they don't have to be repeated. Gizza  ( t )( c ) 10:40, 9 August 2015 (UTC)
 * 6) Yes in spirit, but "diffs" is too restrictive, e.g. it would prevent referring to a section from an ArbCom case. Callanecc (talk • contribs • logs) 04:09, 11 August 2015 (UTC)
 * 7) Agreed, but would suggest that they be limited to "exemplar" diffs to avoid lists of 10+ diffs being placed in a comment. – Philosopher Let us reason together. 23:30, 11 August 2015 (UTC)

Discussion Not entirely supporting here; I do think some patterns can be described best by a clear description without the need of diffs; I also worry about a diff requirement leading to cherry-picking or excessive weight being given to a few one-off diffs that do not establish a long term pattern. I do agree with the concept that the more personal remarks need proof, as well as strong accusations. Jo-Jo Eumerus (talk, contributions) 15:05, 7 August 2015 (UTC)
 * Should is not must. "Oppose per block log, happy to reconsider in future if you can go 12 months without a block" would not require a diff.  Ϣere Spiel  Chequers  19:13, 7 August 2015 (UTC)
 * That is of course quite logical and proves that the prposal should be approached with common sense. The proposal also covers instances where unlinked snippets of text are quoted, being specially cherry-picked and taken deliberately out of context to impart a negative opinion. Pile-on opposers are often influenced by such 'quotes' and take them on face value. Kudpung กุดผึ้ง (talk) 02:33, 8 August 2015 (UTC)

Clearly statements made about specific edits (ones "quoted") should have references to those edits to allow others to view context. Other claims are not something for which "diffs" per se may exist - people who "oppose" should not have to find "diffs" to support their position any more than those who "support" need diffs to support their !vote. Gander - goose applies here. Collect (talk) 12:06, 9 August 2015 (UTC)

Having an unwritten rule about voting requirements does more harm than good
Newbie editors who turn up at RFA risk being dismissed as the socks that they may well be. Goodfaith newbies may well be being deterred by the presence of an unwritten rule as to who can participate at RFA. Having a clear rule with a lowish threshold (>200 mainspace edits, >50 in last year, account at least two months old) could well be the reason why the German Wikipedia despite having a smaller community has higher RFA participation than we do. We should copy that rule from DE wiki. Higher participation could lead to more candidates and would be a greater protection against trolling from offsite.


 * Support


 * 1)  Ϣere  Spiel  Chequers  07:56, 7 August 2015 (UTC)
 * 2) I've been saying his for years - at least since WP:RFA2011 was started. --Kudpung กุดผึ้ง (talk) 08:25, 7 August 2015 (UTC)
 * 3) Kraxler (talk) 12:15, 7 August 2015 (UTC)
 * 4) Beyond the fact that they could be socks, I don't trust somebody who's only had a few weeks of experience judging how a candidate responds to and treats policy.  Kharkiv07  ( T ) 15:00, 7 August 2015 (UTC)
 * 5) Yes. A minimum threshold is a good idea, I think. --IJBall (contribs • talk) 16:36, 7 August 2015 (UTC)
 * 6) I've turned a corner on this and believe that a small fixed requirement is more helpful than the current vague situation. AliveFreeHappy (talk) 17:13, 7 August 2015 (UTC)
 * 7) --Tryptofish (talk) 18:22, 7 August 2015 (UTC)
 * 8) Since policy is just a reflection of consensus and practice, it seems obvious this should be in policy. Dennis Brown - 2&cent; 20:27, 7 August 2015 (UTC)
 * 9) Fylbecatulous talk 20:33, 7 August 2015 (UTC)
 * 10) This is just obvious, and it makes it significantly more difficult to influence RfAs via off-site canvassing as well. ~ RobTalk 09:42, 8 August 2015 (UTC)
 * 11) I've read RfA and the talk page for a while but only started participating a few days ago largely because of this problem. It seemed that new editors to RfA were viewed as socks, canvassers, or generally dismissed so I held my tongue for a while not wanting to cause trouble for myself. I'm not sure where the threshold should be (and isn't really important at this phase) but thought maybe a comment from an actual newbie might be helpful. Wugapodes (talk) 03:25, 9 August 2015 (UTC)
 * 12) Primarily per Kharkiv07. Callanecc (talk • contribs • logs) 02:34, 11 August 2015 (UTC)
 * 13) I agree, though I believe this is mainly the problem for NOTNOW applicants, and this is clearly not the most important RfA issue.--Ymblanter (talk) 09:53, 12 August 2015 (UTC)


 * Comment


 * 1) While not disagreeing totally with the idea of minimum requirements, I have serious doubts that the reason more people don't take part is that we're allowing people with less than 200 edits to participate. I can also think of one commonsense exception to it from my own experience, if an editor has made substantial good faith contributions to another language Wikipedia and decides to become more involved here, I don't see why we should prohibit them from taking part, since they have a clear verifiable history of non disruptive edits. Valenciano (talk) 10:56, 9 August 2015 (UTC)
 * It isn't as simple as that. The current situation is not that we always allow people with less than 200 edits to participate, but that we have vague unwritten rules as to how new a new voter at RFA can be before they are dismissed as a sock. Replacing that uncertainty with a clear message that once you meet a set criteria you are welcome to !vote is going to be more welcoming to newbies than the current cliquey set up.  Ϣere Spiel  Chequers  20:42, 10 August 2015 (UTC)

RfA reform has been tried before, and current ideas are reincarnations of past ones (some dating back to 2005)
RfA reform is far from a new movement on Wikipedia - there were major attempts in 2007, 2011, and 2012, and even more attempts since 2005. Many of the current ideas have been proposed in the past, such as:


 * A de-adminship process was implemented in 2004.
 * Admin mentorship and a probational period, even with a new "request for temporary adminship", was proposed in 2005.
 * Even obscure ideas actually proposed today such as giving arbitrators the responsibility for promoting admins have often been proposed before; the aforementioned idea was proposed in 2007.

The above pages often contain useful statistics and historical facets of the RfA reform movement, useful for this movement that is yet again spawning. Esquivalience t 23:36, 8 August 2015 (UTC)

Concur
 * 1) Indeed, on all points. And I would hasten to recommend using the vast amount of information in WP:RFA2011 as a reference work, because in her charts below has just disproven the claims that it is out-dated. Kudpung กุดผึ้ง (talk) 00:04, 10 August 2015 (UTC)
 * Some of the tables and graphs in this section are interesting comparison points. But please skip over the polynomials; not everything Excel can do is a good idea. The observation that outcome distributions are stable despite declining numbers of candidates doesn't by itself say anything about the rest (voter behavior and so on). Opabinia regalis (talk) 17:30, 10 August 2015 (UTC)
 * 1) I concur as well; however, the pattern since WP:RFA2011 has demonstrated ever-decreasing numbers of administrators overall, and ever-fewer RFA candidates. What I keep coming back to is that there was a watershed at that point that seriously changed the effectiveness of RFA. I do not think it was the fact that we've split off various tools; nobody ran for RFA to gain access to those tools that have been split off. Bottom line, people have *always* run for RFA to have access to blocking and deletion tools with few exceptions, which are two "tools" I hope are never devolved from requiring a full RFA to access. Risker (talk) 19:22, 11 August 2015 (UTC)
 * Question: Do you include deleting redirects in that? In order for there to be the equivalent of "Article mover" user rights, deleting redirects (but not "regular" articles) would almost certainly have to be included in that. But forcing someone who wants "Article mover" rights to go through a full-blow RfA (or its equivalent), basically just to gain the ability to delete redirects, rather defeats the purpose of splitting off a role like that in the first place... --IJBall (contribs • talk) 19:49, 11 August 2015 (UTC)
 * The watershed year was 2007 - all statistics clearly demonstrate this. All the work at WP:RFA2011 was based on the years leading up to the dates in 2011 whe the various charts and tables were compiled, and not on 2011 alone; thus the slope began its exponetial fall already in 2007 gaining impetus with the figures roughly halving each successive year until it now seems to be bottoming out. WP:2011 confirms a trend that was already known and documented by  and is indeed the reason why the WP:RFA2011 project was srarted.  As I've said elwhere on this page, we need to dig deeper - what exactly happened in 2007 that caused this drop-off in so many areas of Wikipedia and whty was the start of it so dramatically abrupt. We need to do some archeological forensics to know the answers. Kudpung กุดผึ้ง (talk) 13:10, 12 August 2015 (UTC)
 * The biggest change came at the end of the first quarter of 2008, just after the unbundling of rollback. There was a big drop in the number of successful RFAs and not coincidentally "good vandalfighter" stopped being a sufficient qualification to pass RFA. Unfortunately Rollback of its own isn't enough for vandalfighters, they really need to be able to block IPs and at least new accounts. If you look at the combined numbers of Rollbackers and Admins things don't look as bad as they otherwise might. But if you give someone Rollback it is just one tool and today of limited value (in early 2008 you needed Rollback to make Huggle work). Admins can keep on learning new tools but rollbackers need an RFA before they get the rest of the toolset.  Ϣere  Spiel  Chequers  14:38, 12 August 2015 (UTC)
 * I don't think it was rollback that did it; there simply isn't a big enough usage of that tool to explain it. Edit filters, however, could have done it, since they immediately prevented a large amount of the vandalism from ever making it into articles, and thus the "need" for RC patrollers was significantly decreased, thus limiting the pool of potential candidates. They went live in March 2009, which is pretty much when RFA started shutting down. Risker (talk) 12:22, 13 August 2015 (UTC)
 * Yeah, I actually think Rollback is pretty useless, at least on its own. (Which is why I've never bothered with it...) I suppose Rollback probably has more utility in concert with Twinkle or Huggle (which I don't use). But, compared to Reviewer, and even Autopatrolled, Rollback's utility seems limited. I imagine including Rollback in with a broader "Vandal fighter" unbundled user right would make it more useful... --IJBall (contribs • talk) 21:12, 12 August 2015 (UTC)
 * Rollback has the advantage that it is fast. Wikipedia is a high-traffic website and vandalism will be seen by many people. Getting rid of it as fast as possible is important. Also, at least on enwiki Huggle requires Rollback. Jo-Jo Eumerus (talk, contributions) 21:18, 12 August 2015 (UTC)

Comments


 * Related:
 * Wikipedia talk:Requests for adminship
 * Wikipedia talk:Requests for adminship
 * --Guy Macon (talk)

Bureaucrats who participate in an RFA should recuse from the ensuing 'crat chat
WJBscribe's involvement in the Liz RFA 'crat chat was probably ill-advised, considering he objected to some votes on both sides during the RFA. xeno's involvement in the 'crat chat was definitely ill-advised after he lobbied for a softening of a reasonable oppose vote during the RFA and participated in subsequent discussion. The result of the chat was defensible, but the process for getting there was tainted. If the 'crats want their process to be respected, they should make it a respectable process by installing basic separation of duties standards.

Support
 * 1) (as proposer) Townlake (talk) 15:19, 7 August 2015 (UTC)
 * Yes, those who vote should recuse. Happy to have those who clerk to participate in crat chats. Struck as the option below is better than signing this with a caveat.  Ϣere Spiel  Chequers  18:25, 7 August 2015 (UTC)
 * 1) Support in part. Those voting or expressing an opinion should recuse, but I think this is mostly done already. Clerking is inseparable from participating in the crat chat, as you must clerk to be able to have an accurate crat chat. ~ RobTalk 21:22, 7 August 2015 (UTC)
 * 2) Yes - with the clear understanding that 'crats who vote, clerk, or otherwise comment in any way must recuse themselves from the 'crat chat. There have been suggestions for community clerking of RfA in the past, but as with all sugestions for improvement, while bitterly complaining about the state of RfA, the community threw them out. (BTW recuse in this sense is a reflexive verb) . Kudpung กุดผึ้ง (talk) 05:24, 8 August 2015 (UTC)
 * At the present time, there are not enough active bureaucrats to 'clerk' RFA (to try to make it less harsh and daunting for candidates) if they are expected to recuse from bureaucrat chats where they have done so. I made exactly two comments on Liz's RFA and the result was the participant expanding their position to provide constructive commentary to the candidate and attending bureaucrats (so no,, I wasn't asking for a softening of the oppose, but a rewording or expanding to be more constructive); this is what is expected of participants in RFA, this was my motivation for comment. I would be interested to know if there are others that feel these two comments should have required my recusing from the bureaucrat discussion ( perhaps the "II" position immediately below is better for your comments?), for if this belief is widely-held I will probably be much less likely to clerk RFAs and that task should be given to a different group so that 'separation of duties' can occur. I would bet would feel much the same (since any user can call out an unconstructive comment, only bureaucrats can participate in 'crat chats).  may have useful comments here, as she has been quite vocal in pushing bureaucrats to take a more active role in RFAs. The alternative would be a cage match where the officials don't step in until the bell has rung.  –xeno<sup style="color:#000">talk  11:29, 8 August 2015 (UTC)
 * I agree with Xeno. If the community decides this is a conflict and warrants recusal so be it, but I would agree that it will simply lead to bureaucrats having no involvement in RfAs until the close. <strong style="font-variant:small-caps">WJBscribe (talk) 13:02, 8 August 2015 (UTC)
 * As we have learnt through BARC (but which we actually knew full well already, but it was worth a try), and other discussions elsewhere, most current active Bureaucrats (and their supporters) appear to be extremely opposed to any new activities for which they were not expressly elected - whether they have time to do them or not. Kudpung กุดผึ้ง (talk) 16:10, 8 August 2015 (UTC)

Oppose
 * 1) I'm going for an oppose here as the proposal below is better. If they vote they should recuse obviously. But if they clerk they should still be able to participate in the 'crat chat as they are still acting as bureaucrats (since community clerking has been rejected). Callanecc (talk • contribs • logs) 03:02, 11 August 2015 (UTC)
 * 2) Clerking does not lead to an appearance of impropriety, IMO. -- Avi (talk) 04:21, 11 August 2015 (UTC)
 * 3) Version below is best. Clerking should not disqualify, and in fact, I would want to encourage more clerking since no one knows the thresholds for clerking better than the Crats themselves.  Dennis Brown - 2&cent; 13:56, 13 August 2015 (UTC)

Increased emphasis on questions weakens RFA
To spot candidates who are editing in breach of policy you really need to check their edits, and the drift from judging candidates on their actual edits to judging them on arbitrary statistics and generic questions is probably making RFA less effective at sorting good candidates from bad. The Q&A section is useful, but only if questions are tailored for the candidate and include a diff of their edits. Would you still do this four month old edit today, and if not what would you do differently? can be a very useful question, What is the policy on a random area that the candidate hasn't been involved in and hasn't said they would use the tools in is sadly too common and rather less useful. One of the side effects of the dramatic decline in the number of candidates is that while the number of questions per month has fallen nearly as sharply, questions per RFA have risen and often exceed 20. This could lead to a situation where people read the RFA and vote without really checking the candidate themselves.

A requirement that all optional questions include a diff showing their relevance to that particular candidate would at least ensure that those who ask questions look at the candidate's edits.

Support

Comments
 * 1)  Ϣere  Spiel  Chequers  08:20, 7 August 2015 (UTC)
 * 2) Kudpung กุดผึ้ง (talk) 08:28, 7 August 2015 (UTC)
 * 3) I suspect this trend is partly thanks to the edit count requirements; few people are going to sit down and inspect 7000+ edits, or even a chunk of them. Thus, the focus shifts on statistics that may not mean what one think they mean, and questions. Jo-Jo Eumerus (talk, contributions) 09:26, 7 August 2015 (UTC)
 * 4) I'm quite tired of the textbook answers to generic questions which in practice are not followed. Kraxler (talk) 12:17, 7 August 2015 (UTC)
 * 5)  Kharkiv07  ( T ) 15:00, 7 August 2015 (UTC)
 * 6) Oh my, yes! It has very much gotten out of hand. --Tryptofish (talk) 18:23, 7 August 2015 (UTC)
 * 7) Opabinia regalis (talk) 18:38, 7 August 2015 (UTC)
 * 8) <b style="color:#595454">Fylbecatulous</b> <b style="color:#DB7093">talk</b> 20:34, 7 August 2015 (UTC)
 * 9) Anyone can provide a satisfactory answer to an RfA question when they have unlimited time to research policy, look at previous RfA questions, or even get help from someone in crafting the answer. Actual decisions made on-wiki before they knew that they would be under a magnifying glass are a far better prediction of future behavior. --Guy Macon (talk) 01:19, 8 August 2015 (UTC)
 * 10)  Collect (talk) 12:35, 9 August 2015 (UTC)
 * 11) I'm waiting for someone to oppose a candidate because if they're willing to spend that much time researching dozens of questions to figure out what the 'trick' is so their answer will be textbook perfect, it must mean they want it too much.  valereee (talk) 12:00, 11 August 2015 (UTC)
 * +1! --Tryptofish (talk) 23:41, 11 August 2015 (UTC)
 * 1) Staberinde (talk) 15:30, 11 August 2015 (UTC)
 * 2) Indeed.  Gizza  <sup style="color:teal;">( t )( c ) 07:55, 15 August 2015 (UTC)


 * While it sounds good in theory, sometimes it's because there isn't an edit to demonstrate something that a question is asked. For example, without the two questions I asked Cyclonebiskit I would have opposed as I hadn't seen anything to demonstrate that they knew and could apply the policy. Callanecc (talk • contribs • logs) 02:39, 11 August 2015 (UTC)
 * Another example is "Do you ever edit under another username or as an IP (not logged-in)?" which doesn't need to be tailored for the candidate. Arguably this should be one of the mandatory questions anyway. - Pointillist (talk) 10:38, 11 August 2015 (UTC)
 * I'm not sure I agree with this. There are valid reasons for legitimate secondary accounts, some of which would be adversely affected if the candidate were forced to disclose their existence (outside of ArbCom, perhaps). -- Avi (talk) 12:46, 11 August 2015 (UTC)
 * So try to get consensus to make something like that a standard question 4, get the wording right and I might even support. However I'm sure we've all had accidental IP edits and apparently many of us started as IP editors, I have no problem with the world and her auntie knowing which of the biggest UK ISPs I subscribe to, but for others it is much more personal info. And then there is the issue of WP:CLEANSTART. Would you be OK with please list all accounts you have edited this wiki from in the last 24 months?  Ϣere Spiel  Chequers  12:43, 11 August 2015 (UTC)
 * I don't have any hard-and-fast rules worked out, I don't expect this would be a flawless forensic tool, but there could be a lot of value in seeing the candidate's responses. If you would like to polish the question my instinct is some sort of libertarian/transparency combination:
 * If you keep the wording vague and open, the candidate can respond in a way that makes sense to them. This should give straightforward candidates an advantage.
 * Of course, we shouldn't tempt candidates to out themselves by accident, and the question doesn't ask for specific details of any other accounts, just whether they exist.
 * The community probably wouldn't think it bad faith for a candidate not to mention a CLEANSTART. However where an editor has publicly changed their username it would be useful if they could point us to the earlier name(s) associated with their opinions/actions.
 * The IP editing angle might produce interesting answers too. AFAICS some well-established contributors make a lot of (minor) changes while not logged-in e.g. at work or on a university network. - Pointillist (talk) 21:38, 11 August 2015 (UTC)
 * @PPointillist & Avi, I occasionally do some IP edits, usually just typo fixes, I'm pretty sure those edits stick. I doubt that many of us deliberately do that nowadays, but we all get edits where the system logs you out just before you save and it saves as an IP edit. I think my two year wording would be a good compromise with the cleanstart policy, however we also have people who run two accounts one for where they are an expert and one for where they are an enthusiast. Though I don't know how common that is. "Apart from this account and the inevitable accidentally logged out IP edits, are there any other accounts that you have edited from in the last two years?" Would be an interesting wording, could you live with that?  Ϣere Spiel  Chequers  09:28, 12 August 2015 (UTC)
 * As someone who has been dealing with the effects of private information (not very active now due to being on the Ombudsman commission), I am still leery of this as one of the legitimate uses of an alternate account is for privacy: editing in areas where having a link between the primary account and the subject may prove harmful. Similarly, consider someone who successfully underwent WP:CLEANSTART. By asking this question, you are either forcing them to lie, or to put themselves in harms way. Moreover, what is the purpose? A recidivist socker is going to prevaricate, and someone who isn't socking, it doesn't matter. I see very little benefit outweighed by harm. Thank you. -- Avi (talk) 14:27, 12 August 2015 (UTC)
 * Cleanstart wouldn't be compromised, other than that there would be a requirement of two years with the new account before running RFA. I accept the point re privacy, though I don't know how common it is. I would be tempted to say get an Arb to give you a dispensation to run, but after one particularly memorable case I doubt the community would buy that. As I said earlier on in the thread "So try to get consensus to make something like that a standard question 4, get the wording right and I might even support." I think my suggested wording would resolve both the accidentally logged out scenario and Cleanstart, not sureabout the issue of legit but undisclosed alt accounts.  Ϣere Spiel  Chequers  14:47, 12 August 2015 (UTC)

Off wiki canvassing and publicity is a problem
We are a big high profile site and a target for trolls of all sorts. Whether or not the most recent RFA with its very high turnout and slide in support after being discussed on hostile sites is an example of this is disputed. But the potential for trolling sites to troll us via RFA is real.

Supports


 * 1)  Ϣere  Spiel  Chequers  07:43, 7 August 2015 (UTC)
 * 2) Kudpung กุดผึ้ง (talk) 10:38, 7 August 2015 (UTC)
 * 3) AliveFreeHappy (talk) 17:13, 7 August 2015 (UTC)
 * 4) A growing problem. --Tryptofish (talk) 18:21, 7 August 2015 (UTC)
 * 5) Dennis Brown - 2&cent; 20:24, 7 August 2015 (UTC)
 * 6) Actually a problem since the beginning of this process becoming contentious, alas.  And not just from "trolling sites" but also from "well-intentioned POV sites", IRC chats, and so on.  Any canvassing by whatever means has a possibility of resulting in an improper result. Collect (talk) 12:30, 9 August 2015 (UTC)
 * 7) This is why I am not on IRC and prefer all communications to be on-wiki and open.--Ymblanter (talk) 09:52, 12 August 2015 (UTC)
 * 8) This has been a problem for many parts of Wikipedia and for a long time (at least since 2007).  Gizza  <sup style="color:teal;">( t )( c ) 07:56, 15 August 2015 (UTC)