RFA by year

WereSpielChequers,

I noticed your stats at RFA by month. I found this really educational, and I was hoping you would add a couple more stats... mainly because a bunch of us have been trying to understand why admin populations are down.

If you look at the chain of cause-and-effect, the lifecycle for an admin is:

1. User joins
2. User is active
3. User nominated for RFA
4. User passes the RFA, becomes admin
5. Admin is active
6. Admin eventually leaves

Right now, we have a lot of stats about the number of active admins. You also have some stats on RFAs. But it seems the issue is not that administrators are leaving, which is what I initially thought. A quick look at your stats shows that the problem is earlier in the lifecycle, because there are fewer RFAs now than at the peak in 2007.

I was hoping you could add one quick stat to that table:

• Number of admins to become inactive by year (= net promotions until that year - active admins that year)

Eventually, I want to compare that with stats like these to figure out what part of the lifecycle is the problem. Are not enough people being nominated? Or are there just not enough people, period?

This issue can't be taken in isolation with the problem that some editors may have received the wrong "hat". (see Senior Editors recommendation)

Sorry can you supply a diff for the wrong hat business? I'm personally of the view that most longterm editors could and should be made admins. Even if they don't anticipate using the tools there will be occasions when it helps to have them.

• Task force/Recommendations/Community health 6
• Task force/Recommendations/Wikipedia Quality 2
• Thread:Talk:Task force/Wikipedia Quality/Trusted/senior users (narrow focus)/reply (13)

Does this help?

I know that some current admins won't like a such evolution. However this change is warranted.

I think that could be part of the solution. There might be admins who never should have been admins... as well as other editors who could take on more responsibility but who are not interested in adminship. (I've never been an admin, and never wanted to be. I take it as a complement that a few people have asked me to be an admin, and many more just assumed I already was.)

The most interesting insight from looking at the numbers though: the admin attrition rate isn't especially high. It's hovered around 20%, and peaked in 2007 at 28%. If the numbers are correct, then the real problem is that nobody wants to be an admin anymore (or nobody wants to go through RFA).

Hi Randomran, I've done that, but I find the figures hard to believe. If I'm right we scarcely lost any admins till 07, then we started losing loads - if I've miscalculated something feel free to correct - I'm not precious about it being in my userspace, but it is transcluded into next weeks signpost.

This gives us two possible models for future loss of admins, one based on extrapolating the loss of editors per year of first editing, the other based on the year on year drops of the last three years. Both give similar results for the next couple of years, but then look dramatically different, in one model just under a third of each wikigenerations admins are around longterm, in the other practically no-one left during the dramatic growth phase, but subsequently we have been losing a fifth of our admins every year -

You also need to guesstimate the future successful RFAs. It is hard to believe that it could continue to fall, and I'm trying to turn it round or at least stabilise it. If you extrapolate from the last few months then we are stable with a new norm of half a dozen a month. If you extrapolate from past years then 2011 could see successful RFAs as rarely as 1 a week and 2012 will be worse.

This gives us two possible models for future loss of admins, one based on extrapolating the loss of editors per year of first editing, the other based on the year on year drops of the last three years. The first could see the active admin cadre stabilise out in a few years time to about three quarters the current number of admins, the second would tend to keep falling towards a figure of five times the number of admins appointed per year - and in this model if the RFA results continue to tank the admin cadre is heading for major problems.

I have no reason to believe your numbers are wrong. It's unfortunately consistent with 2007 as a turning point year for Wikipedia, with a lot of signals indicating a stagnation or even a decline. Unfortunately, I actually *started* to become inactive in late 2007. Maybe it's my fault. :)

I don't have any hard numbers about participation... but I'm not even sure that the decline in participation is proportional to the decline in admins.

1. User joins (no data)
2. User is active (20% decline in 2007)
3. User nominated for RFA (peak in 2007)
4. User passes the RFA, becomes admin (44% success in 2007)
5. Admin is active (peak in 2007)
6. Admin eventually leaves (28% decline in 2007)

Meaning, I'm leaning towards the conclusion that the sudden decline in administrators is NOT strictly about the slow decline in active users. It's most likely that there's some underlying community issue that froze the increase in participation, but had a really big impact on active users. I'd even go so far as to say that the admins were the canary in the coal mine. Admins (the most active and committed users) became frustrated with Wikipedia, and the overall community suffered soon after, maybe even as a direct result of there being not enough admins to handle things.

I don't know. We're close to a revelation here. I feel like we might need more numbers or percentages, but I'm really not sure.

I started editing in 2007, but only became active in projectspace in 08 and didn't succeed at RFA until early 09 - so some of this is history to me. But it won't be to some of the people I will meet at tomorrow's London meetup. I'll ask some of the oldhands how hard it used to be to get people to accept a nomination.

I've approached a lot of people about running in the last twelve months or so. Most turn me down, usually because they don't fancy RFA. Though I've had two successes - and three fails. I've also trawled through a lot of past RFAs to see if people are still around and might be persuaded to run again, I didn't quantify the pattern but it looked to me that a lot of unsuccessful RFA candidates leave the project immediately or fairly soon after rejection. Things may not be as bad now as when I first ran, in Sept 08 - my talkpage went dead for weeks after my first RFA failed. We've subsequently identified that rejected candidates need reassurance that the community hasn't rejected them, just declined to promote them yet, and I think we are better at that.

My suspicion is that becoming an admin often renews an editors interest in the project and extends their wiki career. I think this may also happen with Autopatrolled editors and Rollbackers, though probably to a lesser extent. I'm pretty sure it happens with crats.

BTW I wouldn't pay much heed to pass rates at RFA. We've put a lot of work into deterring the newbies who submit an application before they realise what a big deal it is, and the more clueful editors are postponing their runs later and later into their wiki careers to give themselves a chance of passing. Both of those factors dramatically skew the results.

Deterring newbies doesn't seem to be working. Yes, the number of nominations is lower. But if we were truly scaring away bad candidates, we'd actually see the pass rate go up as the number of bad nominations goes down.

Instead the pass rate has gotten lower and lower every year. It was 80% in 2004. 65% in 2005. Dipped below 40% in 2006, and has been around 33% ever since (except 2007, where it reached 45% again for a brief time). I wouldn't be surprised if the pass rate is a turnoff for a lot of editors. Part of that is being scared, but part of that is seeing RFAs as a battleground and thinking that being admin is kind of stupid.

I'm not sure if that's the main issue. But it's definitely an interesting trend: fewer people are trying to become admins. Has adminship become unattractive?

I think I can speak from experience about this, I considered going in for an RFA a few months ago on the english Wikipedia and Randomran is right, the low pass rate and what looked like a brutal RFA process for previous editors put me off to it. It seemed like every argument or disagreement an editor had was brought up with the involved parties voting against the candidate, not to mention random voters just opposing the Rfa for the most trivial of reasons. I think more people taking an interest in the process is a deterrent now than it ever was, like editors with grudges or previously failed RFA candidate being tougher on new editors.

Maybe there are stats that might show a co-relation between pass rate and the nomination drop-off or the number of voters involved.

Few remarks:

• Not sure how much of the English wiki situation can be globalized to all Wikimedia.
• We need the number of admin actions per active admins to see if the effective workload on remaining admins increased or not.
• Creation of an "Senior Editor" status will make de-facto "Admin" status even less a big deal.

• I think we'd need to look at the English wiki situation first, and then see if we can confirm our findings elsewhere.
• I think it almost goes without saying that the workload on the remaining admins is probably higher, but it couldn't hurt to do the math. We'd need to measure admin activity
• I agree on solutions... a lot of things end up running through the same issues: rewards, volunteer roles, and dispute resolution... but let's pin down the problem first.

What is the problem exactly? The attrition rate has actually improved! The pass rate is not great, but it's the same as it was a few years ago. The real problem is that nominations are down. Why?

My best guesses, with the data available:

1. Nominations are down because participation is down in general
2. Nominations are down because the RFA process is unattractive
3. Nominations are down because being an admin is unattractive

I think that covers all the possibilities. Are we on the same page so far?

I think that many experienced users are wiser about the adminship. Not that the adminship isn't attractive just that it's clearly not their cup of tea.

Do you think this might be a natural state of affairs? In 2004, a lot of people were excited about the idea of being admin... but by 2008, the reality set in, and people realized they'd just rather be editing?

I mean, that would be the fourth guess: Nominations are down because the initial excitement is gone.

Yea, about the fourth guess. The "hype" around adminship dead down within Wikipedia while the admin bashing remained at the same level or increased.

The reality check about the admin function and responsibilities hitting hard.

From outside adminship looks like one of the shiniest achievement Wikipedia has to offer. Once you are inside Wiki for a while, you get quickly whatever you are fit or not for the tools. I should note the numerous failed RfA by very unexperienced users.

I've approached a number of potential RFA candidates this year. Of those who bothered to reply by far the most common answer was that they didn't fancy the RFA process. The second most common answer was agreeing to run, - I've nominated four times in the last 12months and have two more who will probably run in the next two months. Other reasons have been as varied as wanting to stick to content work and being about to retire - but the empirical information I have is that fear of RFA difficulty is a serious problem on EN wiki. One concern I have is that this could be inbuilt in the system and that every project will hit this if and when they reach EN wiki's size.

Yeah, a lot of problems start to come after the honeymoon period is over and things "mature". At first it's wide open possibilities. But then once things crystalize, people start to fight for territory. It's like that for anything.

So yes, I think problems at RFA might be a symptom of a deeper problem: hostility, infighting, factionalism, politicking, crusading...

The wiki process is predicated on having very weak rules and organization. At first, there are a few people operating far enough away from each other that they never clash with one another, and the lack of rules actually inspires people to take initiative and express themselves. But as the population becomes larger, more people start interacting, and you have fights between people with different viewpoints. And with no official leader or policy to decide how those fights should be resolved, you create a power vacuum. The power vacuum is officially sanctioned too: ignore all rules, admins are not special, you must build consensus with everyone equally. So people who want power learn to create power by creating obstructions. They obstruct articles, they obstruct policies, and they obstruct RFAs.

There may be elements of that, though ignore all rules is not as widely used as it once was. Also the population isn't growing, it is either stable or gently declining - though not as fast as the admins are declining. However I agree that consensus is difficult at times and that RfA can be a battle ground between different factions - though oddly that may not be the snarkiest side to it. Opposing because a candidate lacks audited content or has been making errors at speedy deletion tends to be a "not yet" type of oppose that focusses on their contributions. The more personal opposes of the "I don't trust you" or "terrible contributions" variety are what I think poisons RFA and are very difficult to counter unless there is a diff that explains why someone has lost people's trust or made poor contributions.

Yeah, I wasn't really trying to focus on the officially sanctioned power vacuum, with ignore all rules or the way admins are described. I was just making the overall point that there's an unfortunate side effect to a community with no power structure. In theory, there is no power, and every decision is made purely on its own merits. But in practice, power comes from persistent obstruction. If you find a group of 5-10 stubborn editors, they can jam most articles, policies, every RFAs from moving forward. The rest of the community can't move on without their consent (consensus).

Which ties into your point... in an RFA, if someone just says "I don't trust this guy" or "his contributions aren't good enough", you're stuck. There is no consensus to be found there. People who are there to discuss have no power, because it's those fringe voters (in a culture that supposedly is against voting) who have a veto.

Well RFA is not blackballing. If one person says they don't like someone or their contributions but they don't say why or give examples it won't derail the RFA - but the resulting Drama can make the RFA look a little poisonous.

Where we really have contention is where they give reasons which 75% of the community deems weak.

I think we're talking about the same thing. It only takes 30% of people to derail an RFA. Sometimes less for other RFCs. The power to obstruct really is the only added power you can find on Wikipedia. If you can make two friends who will watch your contribution history and follow you to a discussion, you can pretty much obstruct and any discussion smaller than 10 people. (Same thing if you make three enemies.) It's a more fundamental issue. People have learned that gumming up Wikipedia helps them retain power.

I think you're right that drama is the real reason people shy away from RFAs. But I'm more trying to point out where the drama comes from. In an encyclopedia with no systematic decision making process or power structure, drama is power.

The people just say no without proposing alternative or remedy to the situation.

De-facto let's do "nothing" until something or someone outside Wikimedia create an incident big enough that we have to act.

ON EN wiki we've managed to get things going a bit by publicising the problem. August 2010 is now the best month so far in 2010, and one of the best months in the last two years. But we are still in drought and the recent admins have mostly been oldtimers - all 13 successful candidates in August have been editors for more than two years and 11 of 13 were from 2005/6 and 7.