User:Useight/RFA Subjects/Statistics

Some charts (Archive 38)
I've been playing around with some data regarding RfA nominations from June 23, 2005 through October 28, 2005 (last nomination date of a closed nominaton, other than three withdrawals). I don't know if these will be of any use to any one, but I've uploaded three charts for your reviewing pleasure... I was originally setting out to see changes in success rate over time (as a result of a statement made earlier on this page). What I found was that over the last four months, there's been a small reduction in frequency of RfA success for nominees over 2,000 edits. I also found some other statistics that show the burden of the number of edits per day per admin has increased 38% since March 28, 2005. Unless there was a corresponding reduction in the frequency of vandalism, then we are losing capability to deal with vandalism. That doesn't mean we're losing ground, just that our admins that frequently fight vandalism are under increasing burden. We might be doing just as good a job of fighting vandalism as in March. Discerning that would be considerably more difficult (on par with the IBM study). --Durin 20:34, 4 November 2005 (UTC)
 * Image:EditsandRfASuccess.png: Chart showing number of edits vs. success rate of RfAs
 * Image:ActivityandRfASuccess.png: Chart showing average number of edits per day vs. success rate of RfAs.
 * Image:TimeandRfASuccess.png: Chart showing time on Wikipedia vs. success rate of RfAs.


 * Interesting charts. When you say frequency of RfA success for nominees over 2,000 edits, do you mean percetage of successful overall nominations or just the number? Would it be possible to compile statistics on the percent of successful nominations each month and do a graph for that? In addition, did you compile statistics on the number of support votes (perhaps percentages of nominations with 40+ support votes)? There seems to be a lot more nominations with 30, 40+ support votes... Thanks! Flcelloguy | A note? | Desk | WS 21:31, 4 November 2005 (UTC)


 * Yes, thank you for the charts, I am very impressed and appreciate that you put time into something like that for no other reason than to improve the community. Thank again.Gator(talk) 21:34, 4 November 2005 (UTC)


 * Your scales are confusing; not only are they in reverse order (from highest to lowest, rather than 0 upwards), but the scales are not linear either. Also your y-axis label is incorrect - it would suggest that there is a less than 1% success rate for the data scale shown. Would it be possible to fix this? Talrias (t | e | c) 21:38, 4 November 2005 (UTC)


 * Durin, this is painful. You're interpreting noise. --Tony Sidaway Talk  22:18, 4 November 2005 (UTC)


 * Feel free to collect your own data and do your own charts :) I too thought the charts were reversed. But, by the time I had done them (which was purely for my own viewing) I didn't want to bother with reversing them. Reversing them is relatively straight forward, but the charts convey the information anyways. As for the non-linearality of the x-axis...difficult. To attain the figures, I average over the last 20. Identifying those figures for specific target ranges is difficult; going with the average over the last 20 is a lot less time consuming. Sorry if the charts aren't perfect. I thought they might be useful for demonstrating trends (or lack thereof) so thought I'd share them.
 * Re: number frequency over 2,000 edits; % of total RfAs over 2,000 edits, not the number of RfAs (which, if memory serves, his risen significantly). I am keeping track of numbers of support/oppose/neutral votes. There's been a small rise # of overall votes.
 * Re: Tony: Your comments regarding the charts are so far a bit vague. Please clarify? --Durin 22:24, 4 November 2005 (UTC)


 * I'll field that one - there does not appear, from the brief look I've just given, to be sufficient discriminitory power on the variables you've chosen to draw any real conclusions. If you could perhaps make a cube available so that other people can run their own analysis?  Extracting the  data is the worst part. -  brenneman (t) (c)


 * Presumably you have your data in a spreadsheet; surely telling the program to plot the graphs linearly (and not with a back-to-front scale) is a quite simple step? It looks like Microsoft Excel to me. Talrias (t | e | c) 11:14, 5 November 2005 (UTC)


 * There's a fairly strong indication that, up to a point of 2500 or so edits, increasing edit count corresponds to increasing RFA success. Beyond that, it's all noise. --Carnildo 22:55, 4 November 2005 (UTC)


 * Confounds, confounds. You have three graphs each with similar slopes. What you need is a stepwise regression showing which of the three is the important one. The only thing that can be clearly seen from these graphs is the importance of length of time on Wikipedia, and in a very interesting way. If an editor has been on wikipedia a long time, any variation due to individual worth of candidate cancels out - with newer editors, individual considerations are taken more into account, as can be seen from the fluctuations on the right hand side (right-hand side? These scales are backwards!) of that particular chart. In which case, editcountitis is not a problem at all, at least not compared with "timeitis". Mind you, a further confound not quantified (and not quantifiable) has been totally ignored - the worth of the candidates. And that variable is the most important of them all. Grutness...  wha?  05:42, 5 November 2005 (UTC)

Chart showing growth of RfA nominations (Archive 40)
--Durin 21:07, 18 November 2005 (UTC)
 * Nice. Did you research the success ratio too? &mdash; David Remahl 21:10, 18 November 2005 (UTC)


 * Durin, where do you come up with that stuff? Excellent. Also, in addition to the possible comparison to success rate, how does it stack up to the volume increase of registered editors over time....more work for you:)--MONGO 21:16, 18 November 2005 (UTC)


 * Would it be possible to compare the number of successful nominations with 30+ support votes or find the average number of support voters per nomination per month? That would be some interesting data... Thanks! Flcelloguy (A note? ) 00:27, 19 November 2005 (UTC)
 * I agree, this would be even more interesting if you could compare the rate of nominations with the rate of Wikipedia's user growth. &middot; Katefan0(scribble) 00:31, 19 November 2005 (UTC)


 * Ok, here's the data I collect; Nominee, nominator, date of nomination, # of edits, # of article edits, date of first edit, # of support votes, # of oppose votes, # of neutral votes, close status as successful/unsuccessful/withdrawn. Most of this data from June 23 through October 6 is available. I don't maintain that page though; it's a real pain to generate automatically, and I don't want to do it by hand for obvious reasons. I do also maintain information over time on the number of admins, number of users, number of articles, the number of total pages, and number of edits. My data on this does not cover the same time period; I started keeping that set of data in August rather than June. I think doing comparisons based on admins vs. # of users is not a good idea. There is, I believe, a very significant subset of accounts which have been banned as socks, etc. One user could be responsible for a LOT of accounts. I think a better measure is admin growth compared to # of edits growth. Preliminary data on this, which I have been analyzing, suggests that we're maintaining the status quo; roughly 140-150 edits per day, per admin. Of course, we don't know how many admins are actually "active" in vandal fighting. I've thought about various ways in which to get a rough estimate of that, and may snapshot that variable from time to time. But, right now I don't have that data. I could make charts on all or most of the requests above. If you guys can kinda merge requests it'd be easier though :) What do you most want to see? --Durin 02:49, 19 November 2005 (UTC)

Actual stats on # of admins <3 months (Archive 44)
From time to time, I see people making generalizations about the changes in RfA. A recent one was that the average number of votes has gone up significantly, so getting 30 votes should be easy. My stats show the average number of votes has remained virtually unchanged over the last 400 RfA.

Similarly, I see a claim here that we are regularly promoting people with less than 3 months experience. The reality does not bear this out. Out of 285 successful RfAs since June 27th, just 10 of them have had less than 90 days since their first edit (or 3.5%), and only one had less than 2,000 edits:

Editor          Days        Edits (at time of nom) Phroziac         89         1083 JoanneB          89         6747 Redwolf24        87         2414 Jtkiefer         86         2601 Journalist       84         3230 Jkelly           82         3704 Flcelloguy       79         2886 NSLE             73         2071 Bmicomp          72         4904 Izehar           54         4547

Of these, only Redwolf24 had an RfC filed against him, but that was three months after becoming an admin. At least, no RfCs against these people where the RfC was titled rfc/.

Our users/admin ratio is the third highest among major wikipedias Administrators_of_various_Wikipedias. Only Japanese and Spanish wikipiedias have a worse ratio. This would seem to suggest that we need *more* admins, and not less.

Of note; if the bar had been 12 months for the last 285 successful admins, 162 would not have qualified. I wouldn't call you a jerk for such a voting standard, but you will receive a lot of flack for it from a number of people. I made attempts at getting people to stop attacking people on RfA, and was roundly attacked for it (I guess that's no surprise). So, the atmosphere of personal assault continues at RfA.

I also echo some of the comments made by Friday; quite a bit of the most egregious behavior has been exhibited by long time members of this community, apparently feeling they can act with impunity. The said thing is, they're right. To a person, they've gone unsanctioned.

The problem here isn't that the pool of admins is any worse than it was in the past. The problem here is that there are more bad admins because there are more admins in general. Furthermore, that there is no accountability loop, and no feedback mechanisms on admin behavior. And worse, (see section below on Wikipedia riots) ArbCom has essentially invalidated all policies except WP:IAR recently.

RfA does a good job of weeding out seriously bad candidates. I don't think there's anything wrong with the process in so far as how it approves candidates. --Durin 20:24, 6 January 2006 (UTC)
 * I agree, as I said above somewhere RFA hardly ever makes a huge mistake, I'm not sure we need any of these fixes people are proposing, at least at this point. I do think that we could use a mechanism to nudge admins that are consistantly making bad choices when deleting or blocking etc....some sort of temporary revocation of admin tools possibly. You mentioned a feedback mechanism above and I agree that we need one. Over all though I think RFA itself does a decent job and really doesn't need a fix currently. Rx StrangeLove 21:37, 6 January 2006 (UTC)
 * I was promoted with less than 90 days of experience? Hmm... Well, I sure haven't had a RfC against me... (knocks on wood). :-) Flcelloguy (A note? ) 23:09, 6 January 2006 (UTC)
 * Ve vill get to you ven ve are ready. You must be patient. Painful interrogation takes time to set up! --Durin 23:14, 6 January 2006 (UTC)

I had only been here for 54 days before been nominated??? I never thought of it that way! Izehar 23:58, 6 January 2006 (UTC)
 * Really? I agree with Colonel Klink. You need to have been here long enough to give us time to see what you are like; what kind of editor you are. -- §  Hurricane ERIC§ archive -- my dropsonde 07:42, 8 January 2006 (UTC)

What is going on? (Archive 47)
Why are so many people losing these RfAs so quickly and so often now...I definitely won't try to nominate myself or be posted here - the failure would be too humiliating. I guess Wikipedians have become tough cookies (or however that saying goes) ε  γκυκλοπ  αίδεια  *  20:06, 29 January 2006 (UTC)


 * There seems to be two groups of candidates failing:
 * The vandal fighters who come here before they have 3 months of solid activity (someone should spread the word that a premature RfA is not a good idea).
 * The elephant memory victims. If you have ever done something questionable or uncivil it will be brought up, no matter how far back it goes.  I don't know how to solve this one, we should have a statute of limitations on mistakes, but apparently some disagree.
 * NoSeptember  talk  20:22, 29 January 2006 (UTC)


 * Don't know. Seems we need a policy to to advise users to take awhile to get some experience in wikipedia for adminship nomination. Take it slow, get some edits, smell the fresh air and take a walk in policy. -ZeroTalk 20:26, 29 January 2006 (UTC)
 * There seems to also be a group of editors sailing through with hugely lopsided positive votes... these editors all have much longer histories, well balanced overall and solid projectspace contribution records (tools, policy work, etc) and good interactions with others. The elephant memory issue may be valid but the fact that these candidates are all sailing through gives me a lot of confidence in the process. I don't know if a POLICY is needed but maybe some enhancement to some of the "so you want to be an admin" pages and WP:GRFA would be helpful? ++Lar: t/c 20:40, 29 January 2006 (UTC)
 * I started WP:GRFA but gave up on it when two people hijacked it and revert warred anything they thought inappropriate, and blamed me for the edits when I hadn't edited in weeks prior to that. You are welcome to try to improve WP:GRFA, but don't be surprised if you run into a lot of resistance. Caveat; I haven't read it in more than a month (nor am I inclined to), so maybe it's improved. --Durin 15:19, 30 January 2006 (UTC)


 * Some data: The success rate of nominations from June 23, 2005 to November 23, 2005 was 70%. From November 24 to January 23 (last 2 months), it was 59.9%. From December 24 to January 23 (last month) it was 51.8%. The success rate of late has indeed dropped significantly. --Durin 14:00, 30 January 2006 (UTC)
 * I'll admit know that I have had an imperfect record here, so I don't plan to add myself to become an admin. here anytime soon. ε γκυκλοπ  αίδεια  *  21:25, 30 January 2006 (UTC)
 * And who could blame the statistics. Admistrative status is becoming a easier-abuised and lower quality position as of late. I beleive there is perfect justification in the voter's strict nomination attitude. We need people of high quality, and some users (I shall not cite names) have depicted actions very unsuitible of the position. Adminship is not a position to be trifled with, and it reflects a user's dedication and williness to accept more responsilbility for the good of the project. This is to be taken very seriously. -ZeroTalk 14:46, 30 January 2006 (UTC)
 * There has been and will be (I imagine) one side that views adminship as no big deal (backed up by a very old comment from Jimbo himself) and the other side that views it more like you do. --Durin 15:19, 30 January 2006 (UTC)
 * Funny, Durin, you just noted (during an edit conflict) the same point I did. Everyone hits Jimbo's old comment, but as I said long ago, it's as big a deal as the voters think it is. And see my comments below. - Cecropia 15:36, 30 January 2006 (UTC)


 * This is really an extension of the old Wikiargumennt: "Adminship, big deal or no big deal?" My impression is that we may have come to the end of a very liberal cycle in granting adminship as "no big deal" where we were adding a dozen or more admins a week. Two years ago the community was smaller and there were about 200 admins total. Most of the admins then were familiar names: the most active trusted users. Even though I was new at Wikipedia then, almost all the admins were familiar names and the smaller pool of voters knew what these people were about, so "consensus" really was more like a consensus in the way most people would understand it. Though it was talked about "rogue" admins were really rare, and "wheel warring" wasn't even anticipated. The rare admins who really messed up were jumped on by other admins and editors. Another issue is voting. With anywhere from no candidacies to six or seven at a time and typically between 10 and 30 votes per nomination most of the voters took the process seriously and, if they didn't know the candidate, took the trouble to follow criticisms and comments about the candidates to make informed choices (and also add their own observations). Now we have many candidates who are unknown to almost anyone but the nominator (if there is a nominator) and, with upwards of 130 votes on some nominations, the candidates get the most cursory examination. So my point, if it's not already obvious, is that the increasing proportion of objections may be at least partially a growing feeling that too many editors were made admin too uncritically, and the reaction is to reject marginal candidates who may have made it a month or so before. This is just my observation. Bureaucrats are bound to follow the rules set down over the two years that Bcratship has existed and interpret them when necessary. I believe Linuxbeak is attempting some kind of overall reform in RfA and editors have been arguing out a lot of issues. Will this have some comprehensive effect. I have no idea, but it seems that, in the meantime, RfA voters are reacting in the only way that's really noticed. -- Cecropia 15:36, 30 January 2006 (UTC)
 * Personally I'm in two minds about it. On the one hand, it's for the community to decide what their standards are for their new admins - this is a community project after all.  On the other hand though I do think there's a danger in being too harsh in the search for the perfect user to become a new admin and that at some point we're going to out-standardise ourselves in the quest for that perfect user.  However we are still promoting people and so there doesn't appear to be too much of a problem at the moment - we just need to make sure we don't become too bureaucratic in a process that was originally supposed to be "no big deal", even if it is a big deal now. -- Francs2000 [[Image:Flag of the United Kingdom.svg|25px|  ]] 16:07, 30 January 2006 (UTC)
 * Francs, the community decides how big a deal it is to be or not. Judging by the number of admins made in the last year I don't think the problem now is people looking for "that perfect user." Sorry to put it this way, but the job of the bureaucrat is to be bureaucratish in attempting to interpret the desires of the community as consistently, fairly and impartially as possible. Each Bcrat cannot have a personal standard of promotion different from the others, otherwise there is no point in having humans do the job at all. Just program a bot and let it roar! -- Cecropia 19:12, 30 January 2006 (UTC)


 * I personally think the process is antiquated, and me should strive to find suitable admins. some other way (watch disagreement and naysayers below :-) ). ε  γκυκλοπ  αίδεια  *  21:25, 30 January 2006 (UTC)


 * I think that people have simply raised their standards. Last week happened to have a flood of apparently unqualified candidates; the week before had a flood of good ones. There has been some talk of setting a minimum standard for nominating (e.g. 1000 edits and two months of account age) because anyone below that doesn't stand a chance anyway. But if people are enthousiastic, that's laudable. Just don't pile-up the nastiness. Radiant_ &gt;|&lt; 13:54, 31 January 2006 (UTC)

I just did my little chart for January and I get 44 approvals and 44 rejections for the month of January, down from 67 approvals and 32 rejections in December. NoSeptember  talk  15:07, 1 February 2006 (UTC)


 * When it comes to granting adminship, I personally am in the bigger-deal-than-Jimbo-makes-it-to-be school. Maybe calling adminship a "big deal" is a bit much, but the nomination process and the authority granted should be treated as more than just unlocking the rollback feature.  Also, I don't really see any problems with those who have been rejected, with the exception of CSCWEM perhaps. -- tomf688 {talk} 03:12, 3 February 2006 (UTC)


 * You know what could make everything easier on RfA? Clearly established rules.  Things need to be spelled out more clearly.  Discretion is necessary, or even unavoidable, in some aspects, namely the voting of each person and the degree of discretion allowed to Bcrats in closing an RfA.  However, this should be kept to a very minimal, and right now, it isn't.  You go through the GRFA and you see a lot of discretion, there's a rule, but it is applied at someone's discretion.  That worked for a smaller, more "coherent", for lack of a better word, community, which we used to have, as noted by Cecropia, but not anymore.  Clear rules would make it simpler, and more previsible: if a candidate doesn't meet the objective criteria, (s)he will be removed (forcibly, not at someone's discretion); if the candidate does meet the criteria and people still vote to oppose on grounds of "insuficient this or that", that vote will be discarded (again, no discretion). And on the front of making everything clearer: all votes should be justified, or else not count.  And if someone says anything like oppose, the candidate is uncivil/rude/ill-tempered (or whatever), a diff should be mandatory, or the vote should not count. This would make the process not only clearer, but also much more "resistent" to the interference of vandals, trolls and cliques.  Just my two cents.  Regards, Redux 01:02, 4 February 2006 (UTC)

Administrator activity for January (Archive 48)
Some of you may be interested in having a look at User:Durin/Admin activity. I started this as a trial balloon, and was rather surprised to find that the top twenty from each category were responsible for >50% of the activity. Enjoy. --Durin 19:42, 16 February 2006 (UTC)
 * A rider: - Care needs to be exercised in interpreting these statistics. For example, some one protecting/unprotecting main page images would score very high as in the case of Petaholmes or myself (I should be on 20th position along with Bishonen, dunno why I don't figure there) but I rarely protect disputed pages or vandalised pages (simply 'coz other admins beat me to it). The usefulness of this statistic lies in our ability to monitor a bulk of admin activities easily. For example, in the case of blocks, we can check if each of those blocks had a corresponding talkpage message or not. Great work, Durin. --Gurubrahma 10:59, 17 February 2006 (UTC)
 * It is interesting to note that because of User:Curps's automatic blocking bots, he has blocked more users than the next fifteen admins put together. J I P | Talk 09:25, 18 February 2006 (UTC)
 * Yes, I noticed that. I'm sure it's good faith but it also seems trigger-happy. Marskell 09:35, 18 February 2006 (UTC)
 * Well, look who he's blocking - hardy good faith editors... User:I love a block from JoanneB in the morning for instance. Or how about User:Fuckyouwhore? --Cel es tianpower háblame 09:42, 18 February 2006 (UTC)


 * The population statistics are interesting:
 * 2/3 of admins use their ability to delete or restore, but only 1/3 to protect and 1/2 to block;
 * the top 10 and top 20 in each case account for over a third and over half (respectively) of the relevant admin actions.
 * I wonder how many of those 200+ "inactive" admins (like me, I guess) are away, or just writing articles. -- ALoan (Talk) 22:06, 19 February 2006 (UTC)

RfA edit count standards going up? (Archive 51)
User:JoshuaZ mentioned above that he made a rough estimate that showed edit count standards at RfA are going up around 100 edits per month. There's a number of different ways in which we might determine the veracity of this assertion. One way is with the the chart I generated at right which shows the average edit count of successful RfAs and unsuccessful RfAs over time. It is interesting to note that there has been a considerable rise in the average edit counts of successful RfAs, but only a rather small rise in the same figure for unsuccessful RfAs. The approximate rise in edit count of successful candidates is about 171 edits per month. For unsuccessful, it's about 27. One possible conclusion of this is that it really isn't so much the standards that are going up as it is that the average successful candidate has a higher edit count, while the edit count rejection level has only slightly increased. Just some food for thought. --Durin 15:42, 7 April 2006 (UTC)

Wow. Thats a much better and systematic way of doing it than I did. Could you email me your data table for that? I'd be very interested in seeing it. By the way, what were the R values of these trends? Looking at your table it doesn't look like it is that strong a correlation. Also another thought as to why the rejection level is going up. We get a fair number of extremely unqualified self-noms who don't have any idea what is considered a minimaly reasonable edit count. Possibly new editors are becoming more quickly aware of the requirements to be an admin? JoshuaZ 15:48, 7 April 2006 (UTC)


 * Um. I think it is a little tricky to draw that line with confidence, really. For example, I could draw another line that goes down for the period Aug-05 - end Dec-05. -Splash talk 15:55, 7 April 2006 (UTC)
 * Sure, but that would be arbitrarily restricting your sample size to 5 months ending 4 months ago, as opposed to 10 months ending now - less sound, no? I don't like Durin' possible conclusion, that standards are not shown here - since standards are published, you would have to assume that a decent number of potential admins do not apply, knowing that they fall short of those established standards. Stevage 16:10, 7 April 2006 (UTC)
 * The bigger issue is that it includes the obvious pass and fails. We need to focus on the close calls only to see where the real line is (although edit counts is obviously not the key issue in most close calls either). NoSeptember   talk  16:00, 7 April 2006 (UTC)


 * You know, the amount of data being generated about RfAs can be put together to write a paper or two. To me, those curves look suspiciously like fading channel measurements...Anyway, I think there might be a very easy way to fix those "close calls" - why not institutionalize as standard practise, bureaucrats leaving a message on the candidate's talk page, simply asking what sort of circumstances they want to be promoted under? Do they want to wait a little longer for a clearer margin for promotion? Or, would they be willing to settle for something with more ambiguity and controversy? There is no reason why the candidate should not have any input into the circumstances of how they were promoted. The fact is, some users will take into account and value the social context in which they are perceived to be promoted - and for the users who care about it, there will be an inclination to decline, and simply wait for another time. If the candidate declines, the bureaucrat could even say something like, in X months, I'll come back and nominate you. The system does not have to be "cruel" to these candidates - it just needs to be adjusted slightly so that the problem is less pronounced, and possibly even fixed. If the candidate does not decline in a marginal case, then they will have to work on Wikipedia with those consequences - in this case, the bureaucrat does not have to carry the entire burden of the decision. Yes, this situation can be open to be "gamed" - there might be a canvassing of votes. This will probably be solved with some modifications to this idea. Without going into too much WikiPolitics and WikiHistory, this seems to be the most clear and straightforward way to rectify a problem that has been brewing chronically. Refinements? Comments? What do you think? Obviously needs some tweaking, but I have not seen this idea proposed anywhere yet...I do hope it would be useful to make new ideas. --HappyCamper 16:15, 7 April 2006 (UTC)
 * All candidates are certainly capable and welcome to withdraw their nomination at any time, right up to the second it's closed by a bureaucrat. If they are promoted, they may certainly withdraw their adminship at any time as well. I don't think we need to add another step to this. --Durin 17:04, 7 April 2006 (UTC)

I'm pleased that the graph spurred discussion on this point. I thought that it would, and feel this is good thing. Quite a number of people insist standards are going up, but base such conclusions on anecdotal data. The graph above shows some hard data, but the data may certainly be interpreted in a number of ways. I find one conclusion, other people find others. But, I'd much rather be having these discussions using hard data than nebulous assertions based on anecdotal data. I added another chart above which is relevant as well. This chart ends at a different date because it was created some time ago, but it's still pertinent to the discussion I think. --Durin 17:04, 7 April 2006 (UTC)


 * That second graph looks like a much more substantial trend. I'd be very interested in what happens when one looks at both edit count numbers and percentage success as long term trends and be very interested in what the R values are for all three of the best fit lines. JoshuaZ 19:03, 7 April 2006 (UTC)

I think it is better to user the median. The mean will skew the stats upwards on weeks when there is a candidate like Naconkantari or CSCWEM because of all their rollback edits. Regards, Blnguyen | Have your say!!! 02:42, 10 April 2006 (UTC)


 * Actually, I think User:HappyCamper's suggestion there is a very intersting one. Perhaps all RfA's that fit into the 70-80% accept category could enter some second level of voting... Robdurbar 23:06, 10 April 2006 (UTC)

More graphs (Archive 51)
If people are interested, I created a page with the charts mentioned above and a few other charts relevant to RfA. The page is User:Durin/Admin charts. --Durin 19:54, 7 April 2006 (UTC)


 * Good work! Flcelloguy (A note? ) 20:28, 7 April 2006 (UTC)
 * Amazing. Many thanks.--Alabamaboy 20:33, 7 April 2006 (UTC)


 * Thanks. JoshuaZ 20:49, 7 April 2006 (UTC)

Admin activity report for March, 2006 (Archive 51)
To throw some raw statistics into the discussion, here's a summary of the top administrative users in March 2006. A total of 55 admins were responsible for 50% of the administrative activity in March (where "top 50%" has been computed independently for each type of administrative activity). The table shows, for each admin in the top 50%, their total count and rank in each category of action. Entries in roman font are in the top 50%; entries in italic font are not in the top 50% for that action.

138 admins were responsible for 75% of admin actions. 261 admins were responsible for 90% of admin actions. 468 admins were responsible for 99% of admin actions. 595 admins had at least one administrative action last month.

Raw data is available at, which I plan to update monthly (I should just automate it, cron is good for that). Kelly Martin (talk) 15:36, 8 April 2006 (UTC)
 * Hmm..I'm not in there. Time to get crackin' :P By the way, I don't know if I'd consider rights and renameuser administrative actions, seeing as they're more like bureaucratic actions &mdash; Ilyan  e  p  (Talk)  15:53, 8 April 2006 (UTC)
 * The complete rankings for all admin actions (but not bureaucrat/steward actions) can be found at User:Freakofnurture/Stats (discuss). Hope this helps. — Apr. 8, '06 [15:56] <[ freakofnurxture]|[ talk]>
 * Wow, how did I get in the top 20 for blocking? Must've been squidward and whatnot :) &mdash; Deckill e r 16:00, 8 April 2006 (UTC)
 * These statistics are great, but note that they don't cover all administrator activities. Less quantifiable things such as responding to requests to edit protected pages, review blocks (at least those which don't result in an unblock), move a page over a redirect, enforcing ArbCom rulings which don't require use of the listed buttons, et cetera can only be done by administrators and represent valuable contributions but are not listed in these reports. --CBDunkerson 16:28, 8 April 2006 (UTC)
 * Actually, a lot of those things can be done by nonadministrators. A nonadmin can review requests to unblock or unprotect a page (just can't actually hit the button).  Anybody can enforce an ArbCom ruling if the "shiny buttons" are not involved.  The ability to move a page over a redirect is not an administrative ability; any registered user (not recently registered) can do this.  It's moving a page over a non-redirect that is restricted to admins (and which is logged as a deletion followed by a move).  The main ability of administrators that is not reflected in these reports is the ability to examine deleted edits; since that action is not logged there's no way to report on it.  The notion that maintenance activities can only be done by administrators, while admittedly popular, is mistaken: admins have no special authority to do anything on wikipedia; they're just given access to extra buttons because they've (supposedly) demonstrated that we can trust them not to misuse them.  Nonadministrators are free (and encouraged) to perform maintenance activities, to the extent that they can do so within the capabilities allocated to them by their access level, and for those items which require access to facilities not available to them, request others who do have them to complete the task.  Kelly Martin (talk) 16:58, 8 April 2006 (UTC)
 * Hrrrm? How can anyone review unblock requests when only admins can actually perform an unblock? If everyone can enforce arbitration committee decisions why does WP:AER say that admins specifically are requested to do so? Do we really want to allow any user who feels like it to appoint themself as ArbCom 'cop'? Likewise, non-admins obviously can't edit protected pages when requested. I agree that there are alot of maintenance tasks which anyone can do, but these don't seem the type. --CBDunkerson 17:47, 8 April 2006 (UTC)
 * For instance, non-admins can remove the unblock from a user's page, or they can revert edits of banned users, etc. Christopher Parham (talk) 20:26, 8 April 2006 (UTC)
 * Interesting. Can you provide stats of "admins who should never have been voted in"? Alphax &tau;&epsilon;&chi; 16:46, 8 April 2006 (UTC)
 * How would you suggest that I define that category? Kelly Martin (talk) 16:58, 8 April 2006 (UTC)

As an administrator who uses the mop only occasionally, I must say I am impressed. We should institute an "admin of the month" or something. :) Oleg Alexandrov (talk) 17:11, 8 April 2006 (UTC)

What this says to me is that the semi-active admins are doing a lot more than people sometimes think. Let me summarize activity levels:
 * the top 55 admins do 50% of actions
 * the next 83 admins do 25% of actions
 * the next 123 admins do 15% of actions
 * the next 207 admins do 9% of actions
 * the next 137 admins do 1% of actions

Even looking at the least active two categories, a lot of actions are being done there. And we must consider that those admins are likely doing actions that are not easy. You can rack up a lot of actions by doing RC patrol, CSD patrol, or new user patrol, but when someone does a page merge they do but a single delete and undelete that represents important yet slow going work. Same with blocking and unblocking users involved in non-obvious content disputes, not as quick and easy as a username block. I imagine that the actions done by the bottom two groups are more heavy with complicated actions than the easy ones. And the complicated ones are very important to running the encyclopedia. There are not so many inactive and semi-active admins as I thought when you consider that, and there is a good argument for promoting users heavily involved in content areas, not just the visible vandal fighters, because those are the ones that will often be doing the complicated admin tasks that are still important here. NoSeptember  talk  18:12, 8 April 2006 (UTC)


 * I'm not sure. I mainly recall seeing the slow-but-important work being done by people who were doing lots of other tasks as well, not by people who specialise in complex page moves. However, the moral about relative utility of activity lists is a good one - how do we compare an admin who makes 995 uncontroversial blocks and 5 highly controversial ones with someone who makes 50 blocks? (BTW, I show up as a big fat zero in the figures, having been on wikibreak for the whole of March!) The Land 18:28, 8 April 2006 (UTC)
 * I'm not saying that they specialize in complex tasks, but they are just doing normal content editing most of the time and stumble across something that needs to be done with an admin action, as opposed to going on patrol looking to do blocks or deletes. NoSeptember   talk  18:37, 8 April 2006 (UTC)

I notice that the most active deleters are those that do the mundane image work. All of the top 5 deleters did work either deleting orphaned fair use images or unsourced/unlicensed images. By the way, is it possible to have an "all-time" list (not just monthly)? Something like List of Wikipedians by number of edits... perhaps a List of administrators by number of admin actions. The stats would certainly be interesting, and they might encourage admins to be a bit more active. :) Coffee 19:30, 8 April 2006 (UTC)


 * I can go back to January of 2005 (admin actions were not logged in the same way prior to that and I don't have access to those logs via the toolserver). I also note that NoSeptember's analysis is not exactly correct, because I misstated the import of the statistics.  How one defines "50% of admin activities" is kinda fuzzy.  You can just linearily add up across each categories, but then deletions rule because there are far more of them (over half a million in the past 15 months) than anything else.  My original report simply ignored this and computed the top 50% in each category and reported in the list (sorted by total sum of activities) for any admin who was in the top 50% of any category.  This overstates the number of admins required to get "50% of the activities".  I've modified my script to scale each activity by the total number of that activity over the selected time interval (which means the total scaled count of each activity totals to the same amount).  This means that each unprotection counts a lot more than each deletion toward one's rank.  Using this scoring approach, 50% of activity in March was accounted for by only 25 admins; over the past 15 months, by 34 admins.  When I have more time this evening I will generate the 50% list for each month for which I have data and generate a nice little grid, which I will post somewhere. Kelly Martin (talk) 20:18, 8 April 2006 (UTC)
 * I'm not all that concerned about the exact percentages or numbers. My point was that even those admins that are doing from 1 to 10 actions a month are making a worthwhile contribution to Wikipedia especially when you consider that there are hundreds of these people. NoSeptember   talk  20:37, 8 April 2006 (UTC)


 * These are interesting stats, Kelly. Thanks! Flcelloguy (A note? ) 23:41, 8 April 2006 (UTC)


 * Those of us that love stats are going to have a field day now! Fascinating stuff. Thanks Kelly! + +Lar: t/c 13:00, 9 April 2006 (UTC)

Interesting trend (Archive 53)
While considering the statistics regarding changes in edit count standards, I remembered another chart that I have at this location. I decided to update the table there, and replace the chart as both were out of date. I was frankly surprised to find that all categories of RfA candidates have dropped in success rate. I'd previously been under the faulty conclusion that we'd seen an upswing in the number of unqualified candidates (read in this context: low edit counts), which explained why the overall success rate has gone down significantly.

It can be observed from the table at the above linked location that all categories dropped in a range of 3-6% with the execption of 2001-3000 edits, which fell 9%. This got me wondering what the data looks like if we isolated only the data from this year. This showed something rather different. The drop in % for the categories for this year so far vs. all RfAs from June 27, 2005 to December 31, 2005: What really lept out at me was the 2001-3000 range dropping 40% in success rate. In general, for some reason, there are more failed RfAs across all ranges than last year. Why, I don't know. There's lots of potential reasons, and little data to show which reasons might relevant. I'm going to partially reverse myself here; this data does suggest that standards are going up. I think more analysis would help to highlight what's happening. --Durin 15:47, 21 April 2006 (UTC)
 * It could equally well suggest that, with the growing Wikipedia population, more unsuitable candidates come forward and are rejected, without standards having dramatically shifted. All these edit-building tools like AWB and Godmode-lite and things make it easy to rack-up edits and still fall well short of pretty basic admin standards, without those basics having fundamentally shifted. -Splash talk 15:54, 21 April 2006 (UTC)
 * Bingo. Any analysis that lumps all candidates together (including the obviously unsuitable ones) is flawed. We should focus on the pass/fail line, not the overall pass rate. I would be interested in seeing an analysis of RfAs that had between 50% and 90% support votes (so that we exclude the obvious failures and the obvious great candidates). On a separate point Splash brings up, is there a way to measure non-revert edit counts?  NoSeptember   talk  16:07, 21 April 2006 (UTC)
 * Do you have any data on the amount of votes Durin? Perhaps the fact that there are more voters around (more aspiring admins? At least that seems to be the case in my perception) has influenced it as well. The Minist  e  r of War  (Peace) 16:41, 21 April 2006 (UTC)
 * IMO this means that edit count standards have risen above 2000-3000. Candidates with such edit counts now pass much harder than before. Conscious 17:03, 21 April 2006 (UTC)
 * Apart from that simply not being borne out by the data, there are any number of reasons why a candidate with 2000-3000 edits might not have developed themselves sufficiently for people to support their RfA. There is nothing wrong with that. -Splash talk 17:11, 21 April 2006 (UTC)
 * To echo splash, there are many explanations other than the acceptable edit count having gone up. For example, we have recently had more POV-warriors trying to become admins. Many such candidates have many edits but will be voted down. Also, we have had more admins who have been desysopped by the Arb Com who have since run again. AFAICT Those cases lead almost uniformily to failure but will push the average edit count up. There are other possible explanations as well. JoshuaZ 17:57, 21 April 2006 (UTC)
 * I agree that there may be other explanations. Conscious 18:30, 21 April 2006 (UTC)


 * I agree with Splash's comments in general, that standards (not standards regarding edit counts, but standards) may have not changed at all. There are other factors at play here. There are many different ways of looking at the same data. Minister of war: Yes, I have data on the number of sup/opp/neu votes for every RfA from June 2005 through current. NoSeptember (see, I knew you'd drool over the data ;)) Yes, I can generate the same data only for candidates in the 50%-90% support range. I don't have the time right now to do that though. If I don't get back to you, remind me about this sometime next week. --Durin 19:14, 21 April 2006 (UTC)


 * This only goes to show that the distribution of edits is more important than the raw number of edits. But I guess we already knew that.  Or did we?  Redux 20:16, 21 April 2006 (UTC)
 * PAGAN SINNER! IT'S THE RAW NUMBER OF EDITS YOU FOOL! Anyone beneath 10,000 edits just isn't worthy! wait...that includes me...hmm... --Durin 22:23, 21 April 2006 (UTC)


 * Good job, Durin. Just a reminder, though, whenever we look at these statistics: correlation is not causation, and there may be many other lurking variables that influence these data. (In fact, the decreases may not even be statistically significant; the appropriate test must be performed to determined that.) All caveats aside, though, these numbers are quite interesting. Thanks again! Flcelloguy (A note? ) 23:21, 21 April 2006 (UTC)
 * Absolutely. What Flcelloguy said...right...totally.  We cannot assume causation from correlation...wait, no... [check dictionary and thesaurus]... I mean, yes!  Oh, I'll just go count edits ;).  Redux 04:02, 22 April 2006 (UTC)

Stats question (Archive 55)
This is minor, no pun intended...but I programmed by edit summary tool to mark any edit it though was minor as minor, in addition to ones the user marked. The problem is that for edit summary use, do only look at the ratios of based on what the user marked as minor or not, or do I look at all edits that likely where minor as minor (even if it was not marked by the user)? I can program it either way, but I can't decide what to do. Right now, it counts edits it sees as minor as minor for edit summaries. Voice -of-  All T 22:19, 8 May 2006 (UTC)
 * It's all totally arbitrary, but what people probably expect "minor" to mean is when they tick the "minor" flag. If you want to have some other meaning, use a different term lik "significant edits" or something. Stevage 22:34, 8 May 2006 (UTC)
 * Can it do both? I.e., show both minor edits and "non-significant" edits (minor edits according to your tool's algorithm). Although I like the idea of the tool considering edits not marked minor for insignificance, I agree with Steveage that calling these "minor" will probably confuse some people. Kimchi.sg 09:17, 9 May 2006 (UTC)
 * How does the tool determine significance? Regards, Ben Aveling 10:18, 9 May 2006 (UTC)

Another monthly decline (Archive 64)
For those keeping track, we've promoted 21 admins so far this month (with 8 RfAs still open, 6 of which are currently above 75%). So it looks like yet another monthly decline in our promotion rate (admin stats), and the second month with under 1 promotion per day average (previously, for 10 straight months we promoted at or above the 1 per day rate, often well above). Draw your own conclusions. NoSeptember 10:19, 26 July 2006 (UTC)
 * People don't edit as much in the summer (at least for those of us in the norther hemisphere), since there are better things to do. David D. (Talk) 10:28, 26 July 2006 (UTC)
 * And yet last year, we were trending strongly in the other direction during these same summer months. NoSeptember  10:32, 26 July 2006 (UTC)


 * The great tragedy of Wikipedia–the slaying of a beautiful POV by an ugly fact. With apologies to Thomas Henry Huxley David D. (Talk) 10:42, 26 July 2006 (UTC)


 * If I wasn't an admin already, I would not apply for adminship now, it just looks too combative (I don't even comment on other RFAs any more). I am sure that is the reason less people want to go for it, which is a shame, because these are the kind of people who would make good admins. Martin 10:48, 26 July 2006 (UTC)


 * I agree with you regarding the high combativity, Martin. The RFA process is now a headache, and I have to wonder if I would even be able to re-qualify for adminship considering the rigid standards of many users. --tomf688 (talk - email) 11:05, 26 July 2006 (UTC)


 * The biggest problem isn't the standards. If the community feels there should be higher standards to promote admins so be it. Now you could also argue that too high of standards means we're not promoting enough admins and that's a problem given how much backlog we have in things like handling copyright violations. But I still feel the bigger problem is that so many voters don't even look into the candidates at all or make any effort to judge the intangibles. Instead they vote based on edit counts, edit count distribution, and other nonsense. - Taxman Talk 11:55, 26 July 2006 (UTC)


 * We put too much statistical data on RfAs, and too many people expect candidates to fit into some sort of ideal statistical profile. Maybe we should vote in a new way. Is this person likely to misuse the tools? Yes, No, Too early to tell. NoSeptember  12:04, 26 July 2006 (UTC)


 * One of the most annoying things, "you don't sound like you need the tools". Granted, they have some points, but the more admins we have the more people who may feel inclined to help. DYK is now a minefield, with new admins too scared to do it, because only admins could do it, and with the high sdtandards, things are getting worse. Much worse, H ig hway Return to Oz... 12:09, 26 July 2006 (UTC)


 * I would agree completely with a system like this. It is exactly how I "vote" on RfAs currently. Most people are likely to use the tools in the correct manner, so they get my support, and I generally am neutral (occasionally I oppose) for users which have not made enough contributions to tell if they will abuse the tools or not (around 1000 edits is generally a good enough indicator). I oppose the occasional few that are incivil and extreme POV-pushers who perhaps could abuse the tools. Adminship in my opinion is not a big deal, and there are only three issues which really matter (in my opinion again): will they abuse the tools, will they be helpful and courteous to other users and are they experienced enough (with policy) to use the tools in the correct manner. I would much prefer to see this system suggested by NoSeptember instead of the current RfA system. DarthVad e r 12:25, 26 July 2006 (UTC)
 * I wouldn't mind that system as long as it was misuse (as you specified) instead of abuse. If I believe that the tools won't be used properly due to inexperience or lack of familiarity with policy, I'm going to oppose. I may be 100% confident that the candidate won't willfully abuse the tools, but I'll still oppose if I believe the tools will be misused. Aren&#39;t I Obscure? 12:35, 26 July 2006 (UTC)
 * I agree that there may be a problem with the way in which standards are currently applied. However, I don't think that any of the suggestions here will rectify that situation.  In the case of is this person likely to misuse the tools then standards can equally be applied and nothing will have changed.  Editors who believe in a certain minimum of edit count or experience will have no reason to change those standards as they can equally be applied in this case.  It appears to me that standards are applied because the project is too large for editors to get to all of the potential candidates so there will not be any shared history or trust built up.  The standards are proxies for history and trust but have taken on their own significance to the point where they almost appear to be badges of honour in my view.  MLA 14:45, 26 July 2006 (UTC)
 * The idea is not to force people to vote with a different standard, but to encourage them to vote on relevant issues. Often you get people new to RfA voting, the quicker you clue them in to what matters in adminship, the quicker they will make good voting decisions, all voluntarily. NoSeptember  18:42, 26 July 2006 (UTC)


 * "backlog.....in things like handling copyright violations" Do administrators really need to be heavily involved in cleaning up copyright violations? If there is a backlog in getting rid of copyright violations maybe we should put an article about this problem on the main page and let the whole community pitch in to fix the problem. Is there a list of objective criteria by which we can judge if there are enough administrators? --JWSchmidt 15:19, 26 July 2006 (UTC)

Well yes but what about all of the copyvio's that need deleting? To be honest I think we can never have enough admins --Errant Tmorton166(Talk)(Review me) 19:53, 26 July 2006 (UTC)
 * Anyone can remove material from Wikipedia...I suspect that if we Avoid Copyright Paranoia then there are very few copyvios that require admin action. If there really is a shortage of administrators, why not start using a bot to find more candidates? Establish some basic criteria such as 500 edits, 6 months of editing and no past blocks or other obvious problems. Let the bot send all such candidates an email asking if they want to be an adminstrator. If they say yes, then they automatically go to requests for adminship. Also, Wikipedia is large enough that we could have many parallel rfa channels. The bot could assign each candidate to one of the rfa channels: Arts, Biography, Geography, History, Mathematics, Science, Society, Technology, etc according to the kinds of articles that each candidate has edited. We could also have a list of existing administrators willing to serve as mentors for new administrators. The bot could automatically assign each new administrator to an existing administrator who is on the mentor list. --JWSchmidt 22:53, 26 July 2006 (UTC)


 * You want to create a bot paired versoin of Admin Coaching, but with the pupils already having the tools? *hides under desk at destruction waiting to happen* H ig hway Return to Oz... 22:57, 26 July 2006 (UTC)

How about me just give 'crats the power to appoint admins if they truely believe in them? This wouldn't be a common thing, but it would pass the RfA shithole for people who really deserve it. --mboverload @ 22:59, 26 July 2006 (UTC)
 * It would potentially seem reasonable, but it breaches the idea that admins/'crats have extra tools but no extra authority (thence various processes by which the opinions of non-admins are solicited, or else rigidly-defined rules like CSD). Not everyone fully agrees with that principle, of course, but there are enough who do that it would be unlikely to attain consensus.  I can't think of any way to allow consensus-building on RFAs without the stress levels we have now, though, unfortunately.  Any other suggestions? —Simetrical (talk • contribs) 23:25, 26 July 2006 (UTC)

One possibility to consider is that Wikipedia, as a project and a community, may simply be maturing. The semi-exponential growth in traffic that Wikipedia had been experiencing turned into more of a plateau earlier this year. That may simply mean that fewer people have been joining the community since then and as a result there may be fewer potential candidates for adminship now that it is several months later. Unfortunately, the official internal statistics haven't been updated since February (technical difficulties), so it is hard to know how the number of active Wikipedians has actually evolved. Dragons flight 23:51, 26 July 2006 (UTC)


 * Is it possible that the administrative burden is growing, requiring “super-administrators?” All one needs do is page back a year or two on most any WikiProject page & you see growth in administrative requirements and detail. As requirements and expectations grow, it becomes harder to do as a part time activity; folks must specialize (at least a bit) to do it well. When new folks are recruited to an admin role, they must pass a litmus test that they intend to be at least semi-dedicated administrators. And as more administrators pass through that qualification, having passed it, they may think it must be the appropriate test so they continue to apply a rising standard of adequacy. The more administrators must be specialists, the more you’re looking for a different type of person than your typical dedicated contributor. Williamborg 00:56, 27 July 2006 (UTC)

Admin canidates with less than 1000 edits used to be the commonplace, no? The ikiroid (talk·desk·Advise me) 00:59, 27 July 2006 (UTC)
 * Of course they were. There were admins before anyone had 1000 edits, and there were admins when 1000 edits were a rarity.  Now that so many voters have exceeded that threshold, 1000 edits is suddenly too low.  Fact is, though, anyone who's made 1000 edits has shown enough commitment to the project that they aren't going to go crazy when given extra tools, and if you're worried the candidate will be ignorant of policy, grill them on policy — knowledge of policy is only weakly correlated with edit count. —Simetrical (talk • contribs) 00:42, 28 July 2006 (UTC)

It's all those damn RFA standards. Gahhhh. -- Cyde↔Weys 01:08, 27 July 2006 (UTC)


 * I'd say we already have too many admins, not too few. Kelly put some statistics up on the mailing list a few weeks ago showing that the bulk of the work is done by quite a small number, so I actually wonder why we're continuing to promote people. The sensible thing would be to promote people as others drop out, either by choice or by failing to use the admin tools regularly. SlimVirgin (talk) 01:09, 27 July 2006 (UTC)
 * An inactive admin is not an admin. --mboverload @ 01:16, 27 July 2006 (UTC)


 * Those are true stats, but your conclusion is a non sequitur. The bulk of editing is done by relatively few people, but that doesn't mean we don't need more talented editors. - Taxman Talk 02:57, 27 July 2006 (UTC)

Some people don't meet RFA criteria until the setting sun of their career on wikipedia, maybe this is why so many admins are inactive. You see some really active users become really inactive admins after a small amount of time. The ikiroid (talk·desk·Advise me) 01:11, 27 July 2006 (UTC)
 * Well, they finally levelled up! And then there's nothing left to achieve in the Wikipedia RPG. If b'cratship was more attainable, I think we'd whip a bunch of admins into working harder. --W.marsh 01:22, 27 July 2006 (UTC)
 * I don't agree that an inactive admin's not an admin, mboverload. They still have access to the tools and are therefore able to cause problems. For example, I've seen very troublesome blocked users unblocked by a semi-active admin, because the blocked user e-mailed them, and when you check their block log you see they have no experience and corresponding poor judgment. There are hundreds of such admins around, and every week we create more. I can't see the point of it, except to store up trouble for the future. SlimVirgin (talk) 01:25, 27 July 2006 (UTC)

All I can say is that I just got through the adminship process, and on a self-nom. The thing I found: moost of the people who opposed me were other admins! All of my oppose votes were because I didn't have enough experience. At the time I ran for adminship, I had been regitered for 18 months, active for five, had over 2,200 edits (2,500 by the end of my candidacy), had translated, a job no one seems to want to do and which has a huge backlog, twenty articles, written twenty-five additional articles, voted in FAC regularly for over a year, and never had any official actions taken against me, including any angry messages in my talk page, etc. You get the point! Admin standards, of late, are absolutely ridiculous. RyanG e rbil10 (The people rejoice!) 01:31, 27 July 2006 (UTC)


 * Yeah, but most things in the world that are evil or ridiculous are built on logic and/or good intentions. The ikiroid (talk·desk·Advise me) 01:34, 27 July 2006 (UTC)


 * And you passed with 83% support. If good candidates are being promoted and inadequate ones are being rejected, is the system really that broken? Aren&#39;t I Obscure? 01:37, 27 July 2006 (UTC)


 * Yes, but unfortunately, some people felt I wouldn't be a good admin. The defintion of a bad admin, at least in the past, was someone who causes damge to the encyclopedia. Unfortunately, after my contributions, 17% of people still felt I might damage the encyclopedia. RyanG e rbil10 (The people rejoice!) 01:40, 27 July 2006 (UTC)


 * They didn't think you'd sabotage wikipedia, they thought you'd make mistakes because of your lack of experience, and they would be harmful mistakes. Few opposes are because of suspected ulterior motives. The ikiroid (talk·desk·Advise me) 02:28, 27 July 2006 (UTC)


 * Aaargh! We'll never solve the debate over admin standards. I was criticized on my RFA for not participating in discussions, and I'm not going to continue here. No offense to anyone, but I'm going to stand back, watch everyone reach no consensus, and then start the debate again. I'm going to go close AfDs :) . RyanG e rbil10 (The people rejoice!) 02:41, 27 July 2006 (UTC)

90% of admin actions performed by 170 admins
I found Kelly's figures. Apparently, 50% of admin actions in March 2006 were performed by 29 out of nearly 900 admins; 90% of admin actions were performed by 170. SlimVirgin (talk) 02:56, 27 July 2006 (UTC)
 * If 90% of edits were made by 20% of users... would that be a good reason to not let any new accounts be created? --W.marsh 03:02, 27 July 2006 (UTC)
 * I'm not sure the two are comparable. Inexperienced admins can do a lot of harm, and people who don't use their tools much are going to remain inexperienced. It's almost certainly the case that most of the ones we're currently promoting won't use their tools a lot, just as the ones already promoted don't, so all we're doing is creating a large pool of people who don't know how to do what they've been promoted to do. I've lost count of how often I've found admins who don't understand the content policies, don't understand 3RR, haven't read the blocking or protection policies, and so on. SlimVirgin (talk) 03:51, 27 July 2006 (UTC)
 * Would you mind linking to that thread? Dragons flight 03:41, 27 July 2006 (UTC)
 * The post is here. Kelly wrote that we could probably lose 500 admins without having an effect on our administrative actions. I think she was being conversative. SlimVirgin (talk) 03:51, 27 July 2006 (UTC)
 * Even if 90% of admin activities are being handled by the top 170 admins, that is still a large number of actions being performed by other admins. Furthermore, the occasions where admins do in fact do anything damaging are rare, and wheel wars when they occur seem to involved often the most involved admins. I therefore see nothing wrong with having many admins. JoshuaZ 04:02, 27 July 2006 (UTC)
 * Again it's a non sequitur. For one, it's confusing hindsight with foresight. No one could have properly predicted which would do the 90% of the work. (And I dispute that 90% figure is correctly measuring the real work that needs to get done). For another, it's easy to demonstrate that a lot of tasks go undone for too long. Unless you've found a way to motivate people or give them more time to work on admin tasks, the only solution is to promote more admins (or reduce the need for the work). - Taxman Talk 04:10, 27 July 2006 (UTC)
 * Thanks. I was kinda hoping to learn how much activity was required for the top 90%, but I guess it's not there. I know I don't use the admin bit nearly as much as some, but I certainly wouldn't consider myself inexperienced or a liability, so I guess I am curious what degree of inactivity we are really talking about. Dragons flight 04:25, 27 July 2006 (UTC)
 * Durin had done some studies but I can find only this one for January 2006. Tintin (talk) 04:11, 27 July 2006 (UTC)

There seem to be hundreds of people who want to be admins (see Category:Wikipedia administrator hopefuls). Perhaps an initiative should be created to scan through that category for minimum requirements (# of edits, relatively active, no recent history of being blocked/warned for disruption, etc.), and then we just pick 5 or 10 a week and start nominating them? We could even perhaps do something interesting with subcategories, like Category:Admin hopeful with more than 1000 edits, or Category:Admin hopeful who has edited several times per week for the last 90 days, or Category:Admin hopeful with a record of civility for the last 90 days, etc., and then we look for whoever's name is in all of the appropriate subcategories, and nominate them. :) Granted, some of these subcategories are a bit on the facetious side and might be difficult to nail down in terms of guidelines, but it's a potential way to move forward.  We could also make subcategories like Category:Admin hopeful who wants to help with Vandal patrol, Category:Admin hopeful who wants to help with RM, etc., and then when needs in a particular category of adminship are identified, it makes it easier to find interested candidates, and push them through the RfA process.  :) --Elonka 04:39, 27 July 2006 (UTC)

I find SlimVirgin's arguments that we have plenty (or more than enough) admins to be rather unconvincing. So what if sometimes admins can be clueless about a thing or two? Teach them. :) That any reasonably calm and knowledgeable user can in due time get access to more tools is a long Wikipedia tradition which has worked admirably well for the project. Oleg Alexandrov (talk) 04:33, 27 July 2006 (UTC)


 * Is there an explanation somewhere of how these numbers were determined or what data set was used? Interesting stuff- I'm sure more people would like to do their own analysis.  Friday (talk) 04:42, 27 July 2006 (UTC)

You can find Kelly Martin's March list here. It's deceptive to say 170 admins do 90% of all actions. It is just a snapshot of current activity, you can identify a lot of knowledgable admins who had little activity that month, just look at the semiactive and inactive lists on WP:LA for starters. Should I have considered myself twice as experienced as Kelly Martin because I did twice as many actions as she did that month (even though that was only my second full month as admin)? Jimbo doesn't make the 170 most active list, so obviously I'm much more experienced than him ;). I also see several admins on that list that have hundreds or even thousands of deletions, but not a single block, specialist admins may be inexperienced outside of their specialty. No, admincountitis is just not a valid measure, and if it were, then a bot (Curpsbot) would be more experienced than all of us. NoSeptember  05:04, 27 July 2006 (UTC)
 * Note: a reformatted version of the same data, plus derived statistics, can be found here. — freak([ talk]) 13:51, Jul. 27, 2006 (UTC)
 * Nope, I'm not on the list. Oh well.  Dragons flight 05:14, 27 July 2006 (UTC)


 * Kelly Martin only just made the list. The trick is to make your own list ;-) Stephen B Streater 08:19, 27 July 2006 (UTC)


 * Is a list available for the activity for other months ? If the % of very active admins is relatively stable at 20%, it is a strong case for promoting more people. Tintin (talk) 05:08, 27 July 2006 (UTC)
 * I had thought so, but User:Freakofnurture/Stats is also for March. NoSeptember  05:12, 27 July 2006 (UTC)


 * The issue with that is that the people willing to do the job keep getting not promoted due to a lack of consensus (see posts above on the varying standards on voting). Pegasus1138 Talk 05:13, 27 July 2006 (UTC)


 * No-one's advocating admincountitis, NoSeptember. My argument is that lots of people become admins and then do little or nothing with the tools, so it's not clear why they were given them in the first place, or why they're allowed to hang onto them. It's a strawman argument to say that this observation necessarily leads to judging admins solely on the basis of how many admin actions they carry out, but a failure to carry out more than a perfunctory number necessarily means the admin is inexperienced. My point is that we have hundreds of admins who don't do much now, while we promote hundreds more who won't do much in the future, which seems fruitless at best, and potentially troublesome. Taxman's point is a good one, of course &mdash; that we can't predict in advance which of the 90 per cent we promote won't do much &mdash; but we could perhaps try to. SlimVirgin <sup style="color:purple;">(talk) 05:19, 27 July 2006 (UTC)
 * My argument is that lots of people become admins and then do little or nothing with the tools, so it's not clear why they were given them in the first place, or why they're allowed to hang onto them.
 * Because being an admin is supposed to be "no big deal". That's obviously not true anymore, if it ever was, but that's why. SchmuckyTheCat 05:29, 27 July 2006 (UTC)
 * Yes, I agree. I don't think it was ever true, nor should it be. SlimVirgin <sup style="color:purple;">(talk) 05:34, 27 July 2006 (UTC)
 * Then we should focus on those specific actions that have potential for harm (causing disagreements and frustration), such as unblocking not knowing the full circumstances or previous discussion. The answer is making new admins aware of key policies so they don't make these mistakes. Other admins actions can just be reversed without a big deal when a mistake is made. You focused on statistics at the start of this thread, so I pointed out that that focus is not the issue. It's not inactivity that does the harm, it's a bad decision that does harm, and bad decisions can be made by both active and inactive admins. NoSeptember  05:37, 27 July 2006 (UTC)
 * I agree, but how do we do it? We promote users who, in some cases, don't have a clue (about editing or adminning) and they often continue that way unabated. How do we get admins to read the key policies, and how do we check that they've understood them? SlimVirgin <sup style="color:purple;">(talk) 05:48, 27 July 2006 (UTC)
 * The same way we educate any editor who makes a mistake, we leave a note explaining their mistake and pointing to the relevant aolicy. Sometimes when people say admin A is not following policy, however, it is more a matter of a disagreement about interpretation of policy than one of the admins not understanding it. We should not lump those cases into the discussion about inexperienced admins making mistakes, and we have WP:ANI to resolve it. Honest mistakes of inexperience can be solved by a bit of helpful instruction. We should not slam the brakes on promoting new admins because of mistakes, everyone was new and inexperienced at some point. NoSeptember  06:32, 27 July 2006 (UTC)
 * I don't vote for people unless I see evidence that they've understood the key policies. I have to wonder why someone would stand for adminship without having read them, and why anyone would vote for people who haven't, but I see it happening all the time.  I agree with you about education, but people have to want to be educated, and if they wanted to read the policies, all they have to do is, you know, read them. But they don't. :-) SlimVirgin <sup style="color:purple;">(talk)  06:40, 27 July 2006 (UTC)
 * Some have used the optional questions as a way to tell if someone understands policy. Understanding policy is part of what matters when I ask the question Is this person likely to misuse the tools?, the key question I ask myself when considering an RfA. NoSeptember  06:47, 27 July 2006 (UTC)

One thing to consider is also the scope of articles an admin is covering. Does 90 of the vandalism take place in 20% of the articles, and are the most active admins found there? If someone is an admin, but covers a area that is not in many watchlists, and performs less frequent admin tasks there, it is till usefull to have someone with admin right who know what to do in those area's. -- Kim van der Linde at venus 05:23, 27 July 2006 (UTC)


 * And it's not as if we have only a finite supply of adminships to go around. Maurreen 05:53, 27 July 2006 (UTC)
 * You've obviously never tried to contact an admin for immediate help. Many admins are too afraid to even do page protection.  --mboverload @  06:01, 27 July 2006 (UTC)


 * If you have difficulty getting a page protected, there's probably a reason it would be better not to protect it. SlimVirgin <sup style="color:purple;">(talk) 06:05, 27 July 2006 (UTC)


 * No, as in they won't even look at it. "I only do unblocking or unprotection" --mboverload @  06:44, 27 July 2006 (UTC)
 * Then find another admin. Maybe it's a good thing that an admin that is unsure about making mistakes declines doing something that may be a mistake. Having specialist admins is a good thing, we shouldn't force them to be generalists against their will. NoSeptember  07:02, 27 July 2006 (UTC)
 * I'll keep that in mind when I run through the only 3 admins on IRC. --mboverload @ 07:08, 27 July 2006 (UTC)
 * Why limit yourself to IRC? We can't have perfection. Not all admins are expert in all areas, it is a voluntary project, and if they make a mistake they get hammered. Not all admins are actively seeking to become rouge admins ;). NoSeptember  07:17, 27 July 2006 (UTC)
 * How the hell else are you supposed to find admins to do something within 15 seconds? If you're an admin not on IRC, you don't exist. >=| --mboverload @  08:51, 27 July 2006 (UTC)
 * Of course, now it all makes perfect sense... that's the reason I don't do enough to make the list. ;-)  Dragons flight 08:54, 27 July 2006 (UTC)


 * For the purposes of these stats, admin actions are only delete/undelete, block/unblock, protect/unprotect. For the most part, these things have relatively little to do with one's editting interests and more to do with willingness to RC patrol, close *fDs, etc.  Dragons flight 06:00, 27 July 2006 (UTC)


 * I'm not sure that's true. I very rarely do RC patrol, and I never close AfDs, but I'm still on the list. SlimVirgin <sup style="color:purple;">(talk) 06:04, 27 July 2006 (UTC)


 * Did you also get to #16 in protections without going out of your way? I'm not being critical, as I assume they were beneficial, but protection almost never comes up in normal editting for me.  Dragons flight 06:09, 27 July 2006 (UTC)
 * Hear hear. Deletions related to page history merges are more significant than a routine speedy. A block in a complicated 3RR is more significant than a routine username block, and so on. The raw number count tells us little about what an admin's value to the project is. You've got to analyse their actions to know that. NoSeptember  06:43, 27 July 2006 (UTC)

Take a look at how many of the top 40 admins are no longer active or no longer around at all. Obviously, we need to keep finding more replacements for these people ;). NoSeptember  07:08, 27 July 2006 (UTC)
 * Or maybe we need to encourage some people to enjoy editing more and get less caught up in admin matters. I suspect that the wiki world might be a happier place if we spread the burden around a little more.  Dragons flight 07:47, 27 July 2006 (UTC)

I assume you folks here are talking about. This table laks edits of protected pages. Of which for example are MediaWiki:Common.css or high profile templates. And don't forget that wearing a big stick doesn't mean you actually have to use it. So admin bean counting may be as silly as edit counting. --Ligulem 08:41, 27 July 2006 (UTC)
 * If you're not going to use your stick why did we even admin them at all? --mboverload @  08:52, 27 July 2006 (UTC)
 * Do you count actions of police officers by how many times they shoot? Would you take away guns from police officers because they never shoot? :) --Ligulem 08:56, 27 July 2006 (UTC)
 * Exactly. I don't block 50 vandals every day, nor do I delete 50 pages. Does that mean the admin actions I do perform are any less important? I'm certainly not as productive as the admins on Kelly's list, but I totally reject the idea that that means I don't do useful work. Raven4x4x 11:34, 27 July 2006 (UTC)
 * This attitude is exactly why I haven't stood for adminship yet. I don't use automated tools or do RC/NP Patrol, and mainly would continue working on pages in my watchlist which are purposefully not in the main cadre of articles (those have enough eyes, whereas something like Domestic goat can have vandalism for days), so I'm sure I'll get opposes that I "don't need the tools". Except that having the ability to see deleted edits/pages, rollback vandalism, and continue to help clear out CAT:CSD would help me be a better "trusted editor". I don't necessary want to administrate the whole wiki, but there's nothing between the two levels. -- nae'blis (talk) 15:08, 27 July 2006 (UTC)

Ok so only 20% of admins are highly active, that doesn't have an effect on promoting more people surely - we obviously need more active admins, there are still backlogs regardless. What we really need to work out is how to make the inactive admins active again or how to desysop them, surely only when we have 80 odd percent of admins working can we discuss whether we need any more!! --Errant Tmorton166(Talk)(Review me) 09:39, 27 July 2006 (UTC)


 * I'd be more interested in quality than quantity. 100 admins each making one excellent action a day is much better than one overworked admin making 100 dubious actions a day. I see no need to de sysop regular but low level admins. Stephen B Streater 11:53, 27 July 2006 (UTC)


 * Re: the general point of "If you're not going to use your admin abilities, why be an admin?". There are multiple views with regards to one's role as an admin. Some are highly active admins, frequently involved in deletions, blocks, and protections. Some are only interested in editing the main page. Others view it as being granted a fire extinguisher for their particular knowledge areas to be used only on the odd occasion when it is needed. There is no one-size-fits-all role of adminship. The criteria here should not be how active an admin will be in using admin tools. --Durin 13:14, 27 July 2006 (UTC)
 * Yes, the old adage of "adminship is no big deal" certainly applies to admins who don't use the adminship tools but are still editing, I think. --Deathphoenix ʕ 14:21, 27 July 2006 (UTC)
 * The apparently low usage average of the "hot" admin tools (block/protect/delete) is a sign to me that we actually do have different kinds of admins. There have been proposals to split the admin features. But as we all know that has not gained consensus in the past. I believe the first level of trust would be to give entry level admins the right to edit fully protected pages and to do rollback. Blocking users/IP's isn't done by the average admin. --Ligulem 16:47, 27 July 2006 (UTC)
 * I'll note that the stats don't include use of rollback. I suspect that a lot of admins make regular use of the one-click rollback, particularly if they do RC patrol.  (They may make few blocks doing RC patrol, as most vandals are one-off tests and stop after – or even without – a warning.)  It's nice to be able to rollback without having to use one of the hacks.  TenOfAllTrades(talk) 16:13, 27 July 2006 (UTC)


 * Rollback is one of the most annoying things imaginable. Admins often use it to revert your edits, which are often style problems, while the tool is meant for only simple vandalism. It also beats me to the post of reverting vandalism every time, which I am sick of. There is no point trying to revert unless it's on a dea page at 1am. H ig hway <sup style="color:#FF917F;">Return to Oz... 16:30, 27 July 2006 (UTC)

"Openness and inclusiveness are ..... our radical means to our radical ends."<BR> In my view, it is a dark day for Wikipedia when existing administrators start trying to make a big deal out of adminship and they start playing the game of deciding which Wikipedians are important and they suggest that inclusiveness (more administrators) might be bad for the project. I'll be blunt: I think administrators that play such games might need to take a break and try to regain personal perspective on what a wiki is. I'd be willing to see Wikipedia carry out an experiment. Let's first make a list of existing administrators who are willing to be "admin mentors". Then let's randomly select an equal number of Wikipedians who have never been blocked from editing and who have 1,000 or more edits during the past year and make them administrators. Match each new administrator to a mentor. Then we can watch and see how much damage is done to the project. --JWSchmidt 16:33, 27 July 2006 (UTC)

I'm in favor of the general sentiment that adminship should be no big deal. This is a wiki; the entire idea is that the more the merrier, and nothing admins do is irreversible. There's no reason we need backlogs in things like CAT:CSD or any similar category. But losing adminship must be no big deal too if adminship is going to be no big deal, since the abilities are certainly abusable. Either set some criteria for immediate desysopping by a steward or maybe bureaucrat (e.g., blocking a user one is in a personal dispute with) in addition to ArbCom ruling, or have the ArbCom be more aggressive in desysopping admins who ignore policies other than wheel-warring and outright caprice. —Simetrical (talk • contribs) 00:42, 28 July 2006 (UTC)

Statistics:
 * If we cut our number of admins by 500 by selecting 500 people at random, we would get exactly the same percentages.
 * If we promote twice as many admins, then we would STILL get exactly the same percentages (ie the number of most active admins by the same criteria as now would also double)
 * I am unsure if the current admin criteria or changing those criteria would change much. It would be interesting to see who the most active people are per year promoted.

Kim Bruning 08:37, 28 July 2006 (UTC)
 * Unless I'm missing something here, these statistics would mean that we need many, many more admins than we might think. Since apparently only about 20% of our candidates are going to be especially active as admins, we need to promote five times as many people as our gut instinct would tell us is necessary in order to keep the admin pool growing apace with Wikipedia's user base. --Aquillion 23:08, 31 July 2006 (UTC)

Quality verse quantity. --Masssiveego 08:16, 2 August 2006 (UTC)

July 2006 WP:RFA in Review (Archive 65)
1 - July data is preliminary and assumes success of Ambuj.Saxena, Phaedriel, Sir Nicholas de Mimsy-Porpington, Crazycomputers, MisfitToys, GHe and failure of SynergeticMaggot.

--Durin 21:27, 2 August 2006 (UTC)


 * I don't know what's worse, the statistics, or the fact I'm not suprised. H ig hway <sup style="color:#FFCBDB;">Return to Oz...  21:54, 2 August 2006 (UTC)
 * Not surprised at what? The one thing that surprises me is that there is a large increase of self noms. Do most unsuccessful RfA's come from this group? David D. (Talk) 22:13, 2 August 2006 (UTC)
 * Probably. In the past, I imagine nominating someone was a nice thing, now nominating 90% of editors is generally tying a noose around their neck. H ig hway <sup style="color:#FFCBDB;">Return to Oz...  22:19, 2 August 2006 (UTC)
 * Yes, don't they just show the world collapsing around us? We had better throw RfA in the bin, and begin the weekly hue and cry about it. Are you volunteering for this week? Where did you pull 90% from? -Splash - tk 22:25, 2 August 2006 (UTC)

-Splash - tk 22:27, 2 August 2006 (UTC) 22:25, 2 August 2006 (UTC)
 * 1) There has been no particular decline in the number of candidates.
 * 2) People are more inclined to withdraw a failing RfA recently than in February.
 * 3) Probably relatedly, there have been a sharply higher proportion of self nominations.
 * 4) Also probably relatedly, in this pair of months with sharply higher self-nominations than in February, more have failed. Perhap self-nominators are less able to judge when they are ready, or perhaps they are less qualified candidates, or perhaps people are more likely to oppose self noms. There could also be generally worse candidates, more choosy people participating, a spate of patently unacceptable candidates, the oft-reported but rarely-sighted 'rising standards', an unusually well-established crop in the arbitrary 4 week period selected for comparison.....
 * 5) The average edit count requirements are essentially unchanged.
 * 6) The proportion of success based on edit count fluctuates wildly from month to month and no particular conclusions can be drawn, particularly considering that one month is compared with an arbitrary other month, rather than the interverning averages.
 * 7) The proportion of successful RfAs went up this month.
 * 8) There are more people participating in the average RfA (unsurprisingly; Wikipedia grows).
 * 9) Accounts of all ages are affected approximately equally by the lower success-rate this month, and so there is little discrimination on account-age basis apart from "don't be really new" (4 and 6 months are just as likely to be ok, but 2 is bad).

I think the best figure for looking at the "big picture" is the success rate of candidates with over 1000 edits (those with less are almost always rejected). In July 47% of these candidates were successful, up from 45% in June and down slightly from six months ago. Doesn't seem like there's too much to panic about here. Aren&#39;t I Obscure? 22:37, 2 August 2006 (UTC)

I just did a review of all nominations since June 27 of 2005. In that time, self nominations of all candidates with >2000 edits have had a 49% success rate. Non-self noms, 75%. At 26%, that's a big difference. --Durin 22:43, 2 August 2006 (UTC)


 * Well, this is my edit 3163, so perhaps I'd better avoid a self-nom ;-) Stephen B Streater 22:53, 2 August 2006 (UTC)
 * Perhaps, yes :) --Durin 22:55, 2 August 2006 (UTC)
 * Apparently, I'm a statistical anomaly. My self-nom with only 2200 edits passed, although I had been here for 18 months, and only 54 votes were cast in my entire RFA. If you think you're ready, go for it. RyanG e rbil10 (The people rejoice!) 08:24, 3 August 2006 (UTC)


 * There might be a correlation between the increase in number of self nominations vs. perception that RfA standards have risen. Maybe the most significant factor in the "rise" is not in edit counts or time here, but in the rise in number of self noms. --Durin 22:55, 2 August 2006 (UTC)
 * Would you happen to have any stats on the average edit counts and time for self noms vs. non-self noms? Great work, BTW. Aren&#39;t I Obscure? 23:01, 2 August 2006 (UTC)

Durin, you wouldn't have to use preliminary numbers if you used the date the RfA ends or is withdrawn (like on my chart) instead of the date it is started ;). NoSeptember  23:31, 2 August 2006 (UTC)
 * Patience young grasshopper :). It won't be preliminary in a few days, and results are skewed per month by using a less than firm date of end or early withdraw. I prefer to use the incept date for each nom. --Durin 01:21, 3 August 2006 (UTC)

So can you chart how many admin have been removed from their position , for what reasons, and how many complaints on admin in the recent days? --Masssiveego 08:15, 3 August 2006 (UTC)
 * For your first question, have a look here; and concerning the complaints: do you mean complaints in general, or founded complaints? Lectonar 08:57, 3 August 2006 (UTC)
 * Yes. both Requests for de-adminship and User:NoSeptember/Desysop can help you track desysopped admins. With complaints, there are so many bogus complaints mixed in with the legitimate ones, we can't quantify it objectively. NoSeptember  10:51, 3 August 2006 (UTC)
 * One could get some general idea by going through the admin user WP:RFCs. Some may be spurious, but the effort of an RfC suggests that they're largely not "bogus".  Jkelly 20:35, 3 August 2006 (UTC)

Durin has posted some other charts in the past here at WT:RFA, I collected those I could find here. If you find more, please add them. NoSeptember 10:54, 3 August 2006 (UTC)

Admin Growth chart (Archive 68)
Would anyone be interested in creating a chart showing admin growth, similar to this one from German WP? NoSeptember 13:47, 24 August 2006 (UTC)


 * I'd be glad to. Do you know where I (or others) could find out how many articles existed on the English Wikipedia at a certain point in time? --  tariq abjotu  14:36, 24 August 2006 (UTC)
 * There is a magic word for it,, and I'm sure some people were tracking it. I think the place to ask would be the Village Pump. It may be in the history of List of Wikipedias. There should be a page tracking the historical number and I'm sure there must be one. In June I started a page to track the admin count number because we need to track that trend too, not just have the current number.  NoSeptember  15:39, 24 August 2006 (UTC)
 * Multilingual statistics has the monthly numbers for each year. As a note to the topic, if there is some good to come out of this section, that might be finding and nominating more good candidates. The admin to be promoted when the count of current admins passes 1000 (the same way we counted the millionth article) has not been nominated yet. Go find the best candidates now. :) - Taxman Talk 16:48, 24 August 2006 (UTC)


 * My best attempt to create a similar graph is at Image:En-admin-growth.png. I could not figure out how to get the line superimposed in Microsoft Excel; I'm doubting the creator of the German image even used Excel. --  tariq abjotu  16:57, 24 August 2006 (UTC)
 * [[Image:En-admin-growth.png|100px]] Well done! NoSeptember  18:11, 24 August 2006 (UTC)


 * [[Image:En-admin-vs-article-growth.png|100px]]. I've updated the above graph to include a series for article count (it is possible in Excel, just not obvious!) If someone has got a count of users per month I'll happily product a similar version for that. Petros471 14:16, 25 August 2006 (UTC)
 * This is interesting, because it appears to show that the ratio of articles to active admins is increasing -- surely a bad thing.  Mike Christie (talk) 14:22, 25 August 2006 (UTC)
 * Well we used to routinely promote more than 1 admin per day, but now that rate has gone way down, so the fact that article growth outstrips admin growth is no surprise. NoSeptember  14:56, 25 August 2006 (UTC)
 * Great piece of work. Well done! --Guinnog 14:21, 25 August 2006 (UTC)
 * P.S. The number of users per month is at http://stats.wikimedia.org/EN/TablesWikipediaEN.htm -- nae'blis 15:47, 25 August 2006 (UTC)

These are indeed excellent, thanks. This seems to show that we're starting to fall behind in our admin:article ratio and have been seeing a significant and widening gap since around the beginning of this year. It would also be interesting to chart the number of admins applying during the same period; that might shed some light on how appropriate the standards we use are, at least against this metric. Gwernol 14:24, 25 August 2006 (UTC)
 * How would you compare it, though? There's Image:RfAsPerWeek.png from November 2005, but it's only a small snapshot... maybe you could combine the data here with User:NoSeptember/List of failed RfAs (Chronological) somehow? It's such a small sample size, though... -- nae'blis 14:54, 25 August 2006 (UTC)
 * My admin stats chart includes RfAs rejected, which I get from my failed RfAs list. NoSeptember  15:02, 25 August 2006 (UTC)




 * [[Image:En-admin-vs-user-growth.png|100px]]
 * And, as Petros requested, an admin vs. "users with 10+ contributions" graph. I didn't spend too much time looking for the total user count, but I suppose that's for the best as we've gone from one million registered users to two million registered users in six months (hmm... that's a bit strange and that certainly wouldn't make for a lovely graph). However, as Petros pointed out, that data might still be useful in that the creation of frivolous accounts can lead to more work for admins (if they're vandalism-only accounts or inappropriately named).
 * [[Image:En-admin-vs-edits.png|100px]]
 * So, here's an admin vs. "number of edits per month" graph; that should be more useful than a total registered account graph. Anyway, I suppose I'll step away from the graphs now... --  tariq abjotu  16:12, 25 August 2006 (UTC)
 * }
 * }


 * Very interesting, thanks. Now I can say 'we need more admins' with a bit more confidence... (As if the current at least 8 admin backlogs wasn't enough say so!) Even if you take into account bots doing more work (both increasing the edits per month count and taking *some* of the load off admin RC patrollers, it still seems pretty clear that we do need more admins promoted per month than is happening now. Petros471 16:20, 25 August 2006 (UTC)

The admins vs articles graph is interesting. It seems the natural number of articles per admin is around 1,100. We've held steady on that number for most of the history shown on the graph - the one time we strayed from it (the 2nd half of 2005), we had a sudden increase in promotions to bring it back to that ratio. If that's anything to go by, we're due for an increase in promotions now. --Tango 16:47, 25 August 2006 (UTC)
 * How about this for a graph... a 4 line graph with the lines 1)articles, 2)edits, 3)users, 4)admins, and start it at January 2004 for better detail. NoSeptember  17:41, 25 August 2006 (UTC)

This was the best I could do on that one. The problem I came across is that those four data series have different scales. To force them to be close enough for comparison purposes required me to apply a different scale factor to each of the article, user, and edit series. Therefore don't try and obtain numbers from this graph- it exists purely as a trend comparison. Hope that makes sense (if not just use the other ones above!) Petros471 19:37, 26 August 2006 (UTC)

Have you noticed how growth in number of admins is roughly linear, while growth in number of edits appears exponential? Ut oh. Kim Bruning 10:12, 26 August 2006 (UTC)
 * That was exactly my reaction, too, Kim. If RfA continues as it's going, there is no way the Admins will ever be able to keep up with demand. The people !voting on RFA just don't understand that their !votes are causing larger and larger backlogs. Every day we reject another person who wants to help, we're just making the few Admins work harder, and more junk will get through. How do we know it's not getting through already? Firsfron of Ronchester  10:47, 26 August 2006 (UTC)

Something else to think about: On Jan 1st of this year we listed 42 inactive and 23 former admins (total: 65), today we list 83 inactive and 40 former admins (total: 123). So two of the eight months of this year, we have been merely promoting admins to replace those who have left. Attrition rate increasing, promotion rate decreasing, while edits keep exploding. NoSeptember 12:58, 26 August 2006 (UTC)

How about a graph of amount of RfAs in general per week/month? There are only 7 RfAs now, but I remember there often being about 14 at once several months ago. Is the amount of people applying going down? --Rory096 02:28, 27 August 2006 (UTC)




 * [[Image:NumberofRfAsPerWeek.png|100px]]
 * There's this graph, but it needs some updating. However, based on the data already in the graph, I wouldn't be surprised if you were correct. --  tariq abjotu  02:52, 27 August 2006 (UTC)
 * }


 * Any chance you can do a graph of number of policies (and if possible, kb of policy) & number of administrators over time? It grows ever harder to learn everything one ought to learn to become an effective administrator.  I suspect policy is outstripping the guardians of policy implementation. Williamborg (Bill) 21:50, 27 August 2006 (UTC)
 * Yes one more graph idea, but it is useful! All this is great but moreuseful might be a graph of number of articles per active admins. That would inconclusively show the need for more admins. I agree though with hwhat has been said above, we are currently only replacing admins that have retired or become inactive yet the article count is still increasing!It looks like my RFA is going to failk (well there wasnt much hope really), I'm not bitter just sad i cant help out in another way. I am certain I will be able to in th future but for now... I also think that alot of experienced and rewgualr editors are not getting nominated for adminship - all of whom would pass with flying colours. And they don't dare self-nom as people dont seem to like that. It's a tricky situation all round and evantually I am sure it will work out, but for now things are in the balance. --Errant Tmorton166(Talk)(Review me) 19:49, 28 August 2006 (UTC)

Interesting discussion. The point about the amount of policy to get to grips with is a good point. At some point it will take users longer to become experienced enough to be an admin, simply because they will take longer to read through a reasonable amount of policy and get to know the ropes in certain areas. As an example of specialisation away from certain areas, I've personally always avoided this thing called RC Patrol, mainly because I've never liked the idea of stalking vandals and swatting them, necessary though it is. Also, the more policy and guidelines bloat out of control, the more likely it is that current admins will start to become out of touch with what is happening 'on the ground'. Though the anarchic aspects of Wikipedia are there, it is possible for people to feel that they know it all, when in fact they very rarely do. I've often seen even obviously experienced admins taken aback by things they hadn't come across before. Also, as the number of ordinary users outstrips admins by a greater amount, they will find and tag things for admins to do, increasing the backlogs. So yes, more admins are needed it seems. Carcharoth 03:26, 1 September 2006 (UTC)
 * I started a discussion at the Village Pump about whether some separation of deletion capability from other admin powers might be useful. One response was that a generally finer granularity might be a good approach.  This could complicate RfA dramatically, of course, so there's a big downside.  However, it does seem to address the specialization issue, and might speed up the ability of certain editors to acquire certain bits.  What if Carnildo, whose current RfA is clearly going to fail, had been able to request just the deletion bit?  There might well have been a consensus to give him that.  If vandalfighters had the rollback bit, and more experienced vandalfighters had the blocking bit, and deletion reviewers had the deletion viewing bit -- would that help with keeping admin growth up?  I would think RfA would remain the primary venue, with the alternatives routes only taken for those whose experience or history did not make a successful RfA likely.  Mike Christie (talk) 13:41, 2 September 2006 (UTC)
 * Feel free to revive Requests for rollback privileges. Personally, I think the atmosphere here at RFA has changed since it was first proposed, so with a bit of tweaking, it may have a shot this time. Tito xd (?!?) 04:26, 3 September 2006 (UTC)

August 2006 WP:RFA in Review (Archive 68)

 * 1 - Unsuccessful nominations only includes those nominations that were not withdrawn early and were not successful. This is the smallest subset of RfAs, typically comprising less than 20% of all RfAs.

--Durin 13:59, 13 September 2006 (UTC)


 * I suspect that perhaps experience > x months might actually reduce the chance of an RFA succeeding again. I don't know, but at a guess, try x~=12? Could you take a look? Kim Bruning 10:18, 14 September 2006 (UTC)
 * Over the last 6 months, there have been 156 nominations for people with >1 year of experience. Of those, 77 were successful for a rate of 49%. During the same time period, there were the same number of nominations for people with >.5 year and <1 year of experience. 74 were successful, for a rate of 47%. Others have noted a perceived drop off in success rates as experience/edit counts become high. Every analysis of the data that I've ever done has not supported this perception. Instead, what happens is past a certain point, the success rates plateau. --Durin 12:39, 14 September 2006 (UTC)
 * I think the most interesting statistic there is the drop in average votes. Could you put together statistics of average number of votes in a successful nom and average for an unsuccessful nom? It looks like less people voting means more promotions, although it could easilly be the converse that is true, or a combination of both. --Tango 12:19, 14 September 2006 (UTC)
 * I've added the data to the table. See the note as well. Also note that doing average number of votes per withdrawn RfA would not be useful as the number of votes on withdrawn nominations varies significantly based on how rapidly they were withdrawn. --Durin 12:39, 14 September 2006 (UTC)
 * Thanks! Interestingly, it's the opposite of what I was expecting (although it might be skewed by the withdrawals). Successful RFA's get more votes, so it would seem the increase in promotions and the drop in average votes are unrelated. Maybe lots of RFA-regulars were away on holiday in August and it's nothing more interesting than that... --Tango 15:22, 14 September 2006 (UTC)
 * Thank you, Durin, for posting these very interesting stats. --Guinnog 16:22, 14 September 2006 (UTC)

Another request (thanks for doing all this): Can you add "<1000 edits" and "<2 months" rows? And if you've got nothing better to do, how about "number of RFAs that fit this category" columns? These stats are really interesting - please keep them coming. --Tango 16:37, 14 September 2006 (UTC)
 * I took a quick look. All nominees in the last six months with less than 1000 edits failed. Same for all nominees with less than 2 months experience. It's been a long time since we had a nom with <1000 edits succeed. The last one was User:Orioane who had 981 edits at time of nom. That nom went live December 5, 2005. Over the last 1000 RfAs, there have been just four successful noms with less than 1000 edits. Orioane as noted, User:Extreme Unction, User:Hermione1980, and User:NicholasTurnbull. --Durin 17:12, 14 September 2006 (UTC)

RfAs per week (Archive 69)
Some weeks ago, a graph I made back in April (Image:NumberofRfAsPerWeek.png) was referenced in a discussion here on WT:RFA. At the time, it was voiced that it might be useful to have that graph update. Here's the updated version: --Durin 21:20, 19 September 2006 (UTC)
 * The single trend line doesn't seem to fit. Just visually, it looks like an upward trend for the first 1/3, followed by a flat or slightly downward trend for the last 2/3rds. Isn't there a minimizing the square roots of deviations formula that can be applied for a non-straight trend line? NoSeptember  21:28, 19 September 2006 (UTC)
 * Sure, there's a variety of ways that you can do trend forecasts. I added this trendline the same as I added the one in the last graph for consistency. It's the same function as the last graph. Note that this trendline is considerably shallower than the last one. --Durin 21:31, 19 September 2006 (UTC)
 * How about a moving average (3 or 5 week, I'd say)? That would get rid of the weekly variations and show a trend without having to decide what shape the trend is going to be first. --Tango 22:49, 20 September 2006 (UTC)
 * I already tried it. It's unrevealing. The trend line's r^2 value is low and may be entirely useless because of that. --Durin 23:01, 20 September 2006 (UTC)

RfA success rate over time (Archive 69)
In February of 2006, I created a chart (Image:RfASuccessRate705-206.png) showing RfA success rate over time in support of a discussion occurring on here on WT:RFA at the time. It being six months since then, I thought it might be interesting to revisit the data and see where we stand now. The chart below shows the RfA success rate over time from July 2005 through August 2006. Nominations for editors with less than 1000 edits at time of nomination have been removed from the data set.

Observation: The success rate over the last several months has plateaued at 45%. --Durin 17:47, 20 September 2006 (UTC)
 * I don't know that those red curves can reasonably be called a plateau, but I'm wondering what evidently localised effects cause the sharp transitions? Presumably some highly controversial candidate(s) come along one week, crank up the edit count averages and then don't get promoted... or something like that. One trend would seem to be a lower threshold for success in the last few months but again, with such sharp variations, it's hard to see reliable trends. Interesting, nevertheless, though. -Splash - tk 17:52, 20 September 2006 (UTC)
 * The plateau at 45% is based on the average over the last two months. The curve has been dancing around that 45%, whereas the prior several months showed a decline. --Durin 18:04, 20 September 2006 (UTC)
 * What jumps out at me is the dark blue line. In 2005, the average % support plateaus near 90%, then it declines in early 2006, and now it seems to have plateaued at 80% since April. So we used to have a support/oppose ratio of about 8 to 1, now it is only 4 to 1. NoSeptember  18:08, 20 September 2006 (UTC)


 * This may have already been posted recently, but I'm curious what the decline in total nominations has been. Might success rates go down when there are fewer to look at? Marskell 18:14, 20 September 2006 (UTC)
 * If you add the promotions to the rejections from this chart, you may get an idea. Total nominations haven't declined it appears, but that is without excluding the <1000 candidates as Durin has done. NoSeptember  18:25, 20 September 2006 (UTC)
 * Each data point on the red lines in the above chart is based on the prior 20 RfAs for that data point. So, each data point is looking at the same number of nominations. This helps stabilize the data, at the cost of having some RfAs having an effect on more than one data point. --Durin 18:33, 20 September 2006 (UTC)
 * Thanks, and interesting data as always, Durin. I suspect the declining percentage of support may also be due to several anomalies, such as a number of near-unanimous RfAs a few months ago, combined with some unusually low promotions more recently (Carnildo, etc.), though I think that it (the line) is an accurate indicator of the overall trend. Thanks! Flcelloguy (A note? ) 20:37, 20 September 2006 (UTC)
 * Perhaps a look at the median instead of the mean would be interesting. NoSeptember  20:57, 20 September 2006 (UTC)
 * Median of what value? The blue line? Want stdevs too? :) --Durin 21:00, 20 September 2006 (UTC)
 * I don't know, I was just thinking out loud, wondering if other figures would confirm what Flcelloguy was speculating about. NoSeptember  21:11, 20 September 2006 (UTC)

1000 admins - some stats (Archive 71)
Wikipedia talk:Requests for adminship/stats

Recently failed RfAs? (Archive 75)
I was looking for a list of recently-failed RfAs, like there is for recently successful RfAs, and I couldn't find it. Did this exist, or am I just imagining it? I was looking for this because I wanted to see which RfAs had been closed recently, without digging through the page history. I only managed to find an alphabetical list. I found exvicious at Unsuccessful_adminship_candidacies/E, but only because I remembered the name. I eventually found a chronological list at User:NoSeptember/List of failed RfAs (Chronological), but this seems to be updated less frequently. Would it be possible to make this all clearer, rather than spread between 'official' pages and 'user' pages? Carcharoth 16:54, 23 November 2006 (UTC)


 * User:NoSeptember/List of failed RfAs (Chronological) is updated frequently; it appears just about up-to-date to me. If you're looking for a list of failed RfAs, sans the chronological order, you should take a look at Unsuccessful adminship candidacies, which is linked from the top of the Requests for adminship. --  tariq abjotu  16:59, 23 November 2006 (UTC)


 * Thanks for the info about the update frequency. I'll try and be a bit more patient. I was already aware of Unsuccessful adminship candidacies, as that was where I found Unsuccessful adminship candidacies/E... but thanks for that link as well. :-) What do you think about the material being split across Wikipedia and User space. Is that a problem or not? Carcharoth 17:40, 23 November 2006 (UTC)
 * I think NoSeptember's page should be moved to the Wikipedia namespace. It doesn't pose as a problem right now on NoSeptember's userpage, but I just think moving to the Wikipedia namespace would be more logical.  Nish kid 64  18:45, 24 November 2006 (UTC)
 * I have no problems with any of the pages in the The NoSeptember Admin Project being moved to WP space if people want them there, but I also don't want to create a bunch of WP space pages that others don't want. I think the amount of updating that other people do now may be an indication of how interested people are in a particular page. User:NoSeptember/Desysop and User:NoSeptember/RfA voting records get updated a bit more often by people other than me, so I usually consider these the most popular pages. Although I include User:NoSeptember/Admin stats and User:NoSeptember/List of Administrators as my favorites. Anyone who wants to update any page is welcome to do so. Curiously, now when someone gets desysopped, there are 4 (sometimes 5) pages to be updated (1 2 3 4 5) :). NoSeptember  14:52, 28 November 2006 (UTC)
 * Thanks for the reply. Actually, the one that I was interested in (and which hasn't been updated) is User:NoSeptember/List of failed RfAs (Chronological). Basically, what I am after is an expanded version of the weekly list that the Signpost does of all those promoted to admin during the previous week. Expanded to include all submitted RfAs and what happened to them. ie. RfAs submitted this week, RfAs that succeeded, and RfAs that didn't succeed. At the moment, updates seem to be made to long alphabetical lists. I would update User:NoSeptember/List of failed RfAs (Chronological) myself, but I'm not sure what the best way is to do this? Go through the history of WP:RFA and watch for the nominations that go up and note the result when they are removed? The two in particular that dropped off the list when I wasn't around, and I didn't have watchlisted, I needed to dig around to find them: Requests_for_adminship/exvicious (I had forgotten how to spell this, and the lack of a capital 'E' meant I had trouble finding this one using Special:Prefix - I was loking under "Ex"...); and Requests_for_adminship/The_Transhumanist. Admittedly, both of these are listed at Unsuccessful_adminship_candidacies, but you have to remember the name of the candidate. Currently open RfAs can be seen on the page, but is it not worth having a section to place the "results" in. This is currently done for successful RfAs, so I can look at Recently_created_admins to see the recent successful RfAs, but there is nothing equivalent for the whole RfA traffic (ie. a list of the RfAs submitted in the last week). The unsuccessful ones are organised at the alphabetical page noted above, but that is useless for anyone wanting a list of the RfAs that failed in the last week. Carcharoth 16:14, 1 December 2006 (UTC)
 * The quick updating is an issue as you say if you want to find recently unsuccessful ones. Before WP:RFAF was split into subpages, the way to find them would be to look at a diff comparing a week old version to the current version, which would highlight the added entries of the past week. Now you have to do it with 27 subpages (unless you know the letter). I set up this page to make it easier. NoSeptember  23:47, 7 December 2006 (UTC)