The Signpost

Special report

Thirteen years later, why are most administrators still from 2005?

Contribute   —  
Share this
By WereSpielChequers

Thirteen years ago today, I wrote a Signpost article, which among other things lamented that a Wikigeneration gap was emerging. At that time, over 90% of our administrators had made their first edit more than three and a half years earlier.

Things have not gotten better over time. Actually, they have gotten worse! In 2010, 90% of admins had made their first edit more than three and a half years prior. In 2023, 99% of our admins made their first edits over four and a half years ago. As for that 90% threshold I used in 2010? Over 90% of current admins made their first edit before I wrote that article. As an editor who started in 2007, I'm still relatively new compared to most of our admins — a majority of whom joined the project in 2005 or earlier.

In thirteen years, the Wikigeneration gap has widened by twelve years.

Admins by year they started editing, comparing 2010 with 2023
Year Year that
Aug 2010 admins
started editing
(as of 2010)
Year that
Aug 2023 admins
started editing
(as of 2023)
Ratio,
2023
vs.
2010
2001 32 8 25%
2002 109 38 35%
2003 223 75 34%
2004 404 168 42%
2005 481 198 41%
2006 326 184 56%
2007 115 63 55%
2008 43 43 100%
2009 13 27 208%
2010 0 14
2011 12
2012 10
2013 6
2014 5
2015 9
2016 2
2017 5
2018 7
2019 4
2020 2
2021 2
2022 0
2023 0
Total
1746
881

This is a comparison of the admins of August 2010 versus the admins of August 2023, by the year they created their account on the English Wikipedia. Note that in some cases (many in more recent years) it won't be the same editors — this is when people started editing, not when they became admins. The 2010–11 study is here.

Admins by year started editing. As at Aug 2023

  2001 (1%)
  2002 (4%)
  2003 (9%)
  2004 (19%)
  2005 (22%)
  2006 (21%)
  2007 (7%)
  2008 (5%)
  2009 (3%)
  2010 (2%)
  2011 (1%)
  2012/21 (6%)

I can understand why we don't yet have any admins who started editing in 2022 or 2023: few candidates now succeed without two years' experience in the community, and candidates with only one year of experience are very rare indeed. But I'm surprised at how few admins we have who joined the community in the entire decade of the 2010s, and especially with the class of 2016. Why do we only have two admins who started editing in that year?

This study looks at admins not by when they became admins, but by when they joined the community. If this inspires someone to go off and analyse things by when people became admins, then I'd be interested to see the result; I think both approaches are potentially interesting (to be honest, I suspect that I used account creation date for the 2010 study because it was easier for me to get that data). As for the 2023 study, there is an advantage in repeating the same analysis (on the same benchmark) thirteen years later. But the results are starker, as many of the three hundred new admins we've had since I published that article thirteen years ago were already editing at that point.

On the flip side, I doubt if anyone imagined thirteen years ago that so many of us would still be adminning on this site thirteen years later. My hope is that we can persuade some Wikipedians who joined the community in the 2010s to become admins; I'm sure there are many of you who would pass easily. But given the fact that we have kept Wikipedia supplied with admins through the last decade, less by recruiting more of them than by retaining the ones we had, I'm confident that if Wikipedia is still here in 2036, many of our current admins will still be around.

I just hope they are outnumbered by new recruits.

S
In this issue
+ Add a comment

Discuss this story

These comments are automatically transcluded from this article's talk page. To follow comments, add the page to your watchlist. If your comment has not appeared here, you can try purging the cache.

"Joined the community"

@WereSpielChequers, is measuring from first edit maybe a little misleading as a way to assign a "class" or generation? Look at the editing history of FireFangledFeathers, a brand-new admin. They created an account in 2009, but it wasn't until 2021 that they made their 100th logged-in edit. So are they the class of 2009, or the class of 2021? I'd say 2021, myself. Is there a way to capture what year people actually started actively editing?

I also wonder if we can even assign any meaning to length of time (whether between first edit and RfA or between "becoming active" and RfA) because of course the average time is going to be longer in year 22 than it was in year 5. In year 5 it couldn't have been more than four years. Valereee (talk) 10:18, 11 August 2023 (UTC)[reply]

Hi Valereee, yes there are people who create an account several years before becoming active with that same account. I think they are rare, but yes they exist, and if someone comes up with a way to start the clock at when people become very active rather than first edit, then their study will likely be better than mine. Also I accept that we are now almost a 22 year old organisation rather nearly nine. So averages would be expected to stretch, but that isn't my focus. I'm looking at the gap between when we recruited our admins and people are joining the community. Of course to make that clearer it would make sense if we had the same data on currently very active editors, and compared it to currently very active admins. It is possible that the real problem is that we aren't recruiting new editors, and the very few we recruit are mostly becoming admins, I'm pretty sure that's not the case, but I haven't currently got the data to rule that out. ϢereSpielChequers 12:55, 11 August 2023 (UTC)[reply]

Highly subjective

"Things have not got better over time, actually they have got worse. Instead of 90% of admins having made their first edit more than three and a half years ago, we now have 99% of our admins having made their first edit over four and a half years ago." Lots of us would not consider that any kind of problem, as we expect admin candidates to have several years of experience, and of showing constructive work here. The fact that zero of our current admins first started editing as recently as 2022–2023 is perfectly fine by me (nor am alone in feeling that way about it). The majority of our admin corps being people with deep institutional memory is a good thing. "Over 90% of all our current admins made their first edit before I wrote that article [13 years ago]": Well, we retain admins as active editors (and as admins) at a higher rate than we retain editors in general, so this makes sense. Yhere has been an uptick in requests for adminship in the last maybe two years, after a several-year slump, and most of these requests have been successful The real problem with RfA is that it is legendarily a nightmare to go through, so there is little incentive for people to do it, especially as we don't seem to have a "we're running out of admins" emergency. I think the years-long RfA slump we had (largely because of how awful RfA was getting) is the answer to your "surprised at how few admins we have who joined the community in the decade of the 2010s" wonder. (As for 2016 in particular, it's just a statistical blip in a small sample size.) Another factor is that many formerly-admin tools have been unbundled from the admin bit (page-mover, file-mover, template-editor, etc.), thus fewer people actually need the admin bit. Is there really a pressing need for a bunch more people running around with the ban-hammer? As you say yourself: "three hundred new admins we have had since I published that [2010] article". That's an awful lot of admins.  — SMcCandlish ¢ 😼  17:51, 11 August 2023 (UTC)[reply]

As I said "I can understand why we don't yet have any admins who started editing in 2022 or 2023: few candidates now succeed without two years' experience in the community, and candidates with only one year of experience are very rare indeed". So that bit is fine by both of us. My concern in this article is about the Wikigeneration gap between the admins and the current community. Yes we expect experience at RFA, but in 2023 people argue whether new admins need one, two or three years experience, it is now 2023, there aren't many members of the community who would baulk at an admin who started editing in 2019, letalone 2010. Hence my question, why don't we have more admins who stated in the 2010s. As for the idea that there has been an uptick in RFAs in the last two years after a several year slump; 2021 had 7 successful RFAs and 2022 14. Looking at Wikipedia:RFA by month the last three years were three of the five lowest years ever for new admins, and this year is looking similar. I'm seeing a downtick, but I ignored that in the article as possibly COVID related and maybe not statistically meaningful. ϢereSpielChequers 06:33, 22 August 2023 (UTC)[reply]

Misc. post-publication comments

I also seem to remember (I'm sure we have stats somewhere) that that was around the time when editorship was also growing rapidly as Wikipedia became more mainstream.  — Chris Woodrich (talk) 16:38, 15 August 2023 (UTC)[reply]
this shows an exponential growth until 2007. then decline from 2007 to 2014, and then the 2015 rally, which since we are now in 2023 we should accept as more than a blip. ϢereSpielChequers 19:19, 15 August 2023 (UTC)[reply]
  • Thanks for that. It's a shame that this data only starts in 2006, as the rise in active editors was quite precipitous. And it does make sense that a lot of the early adopters - maybe not ground floor, but close to it - would be more committed to the long-term success of the project. Not to use cliches, but there are a lot of "true believers" in that crop. — Chris Woodrich (talk) 19:39, 15 August 2023 (UTC)[reply]
  • I'm not surprised that so many of the people who started editing in the early years are still around. I'm just surprised at how few of our newer members have become admins. I'm pretty sure we have a lot of active editors who started in the 2010s. ϢereSpielChequers 21:01, 15 August 2023 (UTC)[reply]

"class" year

This is something WSC and I discussed a bit at the previous location's talk, but just as a for instance, I opened my account in 2006 but didn't make my 200th logged-in edit until 2014. So are we counting me as class of 2006? FireFangledFeathers' editing history is even more stark; if we're counting them as class of 2009, I'd argue they should be counted as class of 2021 as that's when they actually started actively editing. I don't know how we get at this data.

And for many of us there are multiple reasons we might see such a pattern. In 2006, when I created my account, I had a 13-yo and a 10-yo. I was flipping busy. I created the account for a single purpose: to create a missing article for a prominent author of lesbian pulp fiction. Then I went back to spending my time and energy on real life. It wasn't really until my youngest went off to college that I started becoming active. Valereee (talk) 15:49, 17 August 2023 (UTC)[reply]

I agree that that would be a better way to do things, but these are the stats I could create. If someone else can do the same thing but count from the first month that someone made 50 edits in that month then I suspect the figures will be a bit different, One excuse for my doing things this way is that I started by doing the same sort of study as I did thirteen years ago and then compared them. My belief is that many years ago accounts where people had made the odd edit or two and then become more active after many years were relatively rare. Now it is common. If someone were to analyse things that way then the more recent classes would be larger - but I'm confident they would still show a wikigeneration gap between our admins and our active editors. — Preceding unsigned comment added by WereSpielChequers (talkcontribs) 20:32, 17 August 2023 (UTC)[reply]
Fascinating discussion, all! I've taken the liberty of history-merging in the old discussion to here (so the Signpost article and talk page histories are in the same place) and pasting the old comments to the top of this page. Graham87 08:28, 18 August 2023 (UTC)[reply]
I don't count in the above table, I'm sure, but my account was created 2017 and I consider myself "class of 2020". (Although really, using the high school metaphor, I'd be class of 2023 and my freshman year was 2020.) theleekycauldron (talkcontribs) (she/her) 20:16, 18 August 2023 (UTC)[reply]
You're right, the data was drawn before your RFA completed. Hopefully if I rerun this in another thirteen years (or maybe less) you'll be in then. I agree that it would be better to treat class as the time you became active rather than first edit. But there are advantages to using the same criteria as before when you make historic comparisons, and sometimes you have to work with the data you can access - my rusty IT skills weren't up to doing this the way that many people would prefer. Perhaps someone else will? ϢereSpielChequers 06:06, 22 August 2023 (UTC)[reply]
One could use median age of edits. Thus, an account that was sleepy in its first few years would be aged from the much more recent time it got fire in the belly. Not that this is would be a commonly found pattern. Jim.henderson (talk) 12:28, 25 August 2023 (UTC)[reply]
And an editor who starts using AWB would suddenly become more "recent". I used the term class because of the connotation of a group of people who started out at about the same time, and my concern at the gap between when most of the admins started and much of the current community. Perhaps a better way to do it would be to measure from the hundredth or even the thousandth edit. But I don't have an efficient way to do that, also I can see it becoming contentious if people start saying that their first x years and y edits in the community are being ignored. ϢereSpielChequers 19:37, 26 August 2023 (UTC)[reply]

Adminship as "no big deal"

The main cause of all of this is the fantastically over-complex and excessively rigorous RfA process. Back in the day, adminship was said to be "no big deal"; admins were editors like any other only with a few more powers than others, and the requirement to use those powers only according to the site's rules (with a bit of occasional WP:IAR where absolutely necessary). They were basically janitors, not demigods.

My solution for this is to go back to something closer to the previous state of affairs. Adminship should be far more easily granted to any well-behaved well-established editor who shows the willingness to do so, with a probationary period of say six months during which new admins have their admin bit put on hold, or removed entirely if they misuse their powers, based on something as simple as consensus in a discussion on WP:ANI. After that, they can keep adminship indefinitely unless they misbehave, with a somewhat higher burden of proof required for de-adminiship. A list of well-behaved productive editors could quite easily be maintained by a bot. — The Anome (talk) 15:53, 18 August 2023 (UTC)[reply]

I've long thought we need some form of assistant/junior/apprentice Adminship. Application would be based on fixed criteria, e.g. experience in multiple areas, lack of bad behavior, etc. Powers would be limited, e.g. no more than 24 hour blocs. A full admin would be assigned to mentor. Terms would be limited, with renewal based on performance in the role. Promotion to full admin would be based on their record. We need to open the process up.--agr (talk) 16:14, 18 August 2023 (UTC)[reply]
This has been repeatedly proposed, and repeatedly rejected. And I certainly don't want a bot deciding what counts as a "well-behaved productive editor!" There's a famous story on how statistics can be misleading: often times senior doctors at a hospital have worse casualty rates from operations than junior doctors. Are they senile / out-of-touch? No, the very best doctors also take the hardest cases, while the trainee still in residence is given the easy cases. Basically, if an admin wades into a radioactively controversial area - which are often exactly the areas that *need* admin action the most - they're much more likely to do something that will be considered "misbehavior", or to make "enemies", or whatever. The bot wouldn't realize this and would think an incompetent admin in a low-stress area is better than a very good admin in a high-stress area. It's much better for admins to be admins, and not have any special probationary period where they have to worry about taking action in controversial areas.
Now, that said, Arbcom should be a little freer on pressing the de-admin button for admins who have proven just to not be very good at the job, or have become detached from the community. But that's a different solution than two tiers of adminship. (And we'd need that ability even if we did decide to have two tiers - the admin who just did non-controversial janitorial stuff during their junior adminship and then starts going rogue after getting a full adminship.) SnowFire (talk) 05:26, 19 August 2023 (UTC)[reply]
Several years ago I did a trawl of the admins who had had the bit removed "for cause". Instead of a bunch of newbie admins screwing up in ways that a probationary system might have helped, what I saw was a bunch of longserving admins who after three years or more had drifted away from community norms. I think that the admin newsletter was set up in response to concerns about longterm admins and community norms drifting apart. We could do more along those lines, including periodic retraining. I'm not convinced that we have an unresolved problem with new admins screwing up, or that when it happens we don't notice it or get new admins to learn from it. But we do have issues with longstanding admins and it would be good to either retrain or get them to arbcom earlier. ϢereSpielChequers 05:51, 22 August 2023 (UTC)[reply]
WSC, did your research seem to show any patterns about admins who've been desysopped for cause for reasons that are temperament-related vs. operating outside current community norms? (Wow, which brings us right back to the idea of "not a jerk, has a clue"!) Valereee (talk) 15:48, 22 August 2023 (UTC)[reply]
Sorry Valereee, I'm not sure where I documented that trawl, but the one thing I remember is that three year peak. My suspicion is that part of our "temperament" problem is that people who learn to communicate with problem editors by dealing with and blocking vandals don't sufficiently up their empathy level when they move to dealing with edit warring and articles on insufficiently notable people. But I haven't done the work to test that theory or quantify it; however it does fit with the phenomenon that it is the experienced admins that get desysopped, not the new ones. ϢereSpielChequers 11:34, 28 August 2023 (UTC)[reply]
Following User:Maxim/ArbCom and desysops, in the past several years, the typical admin who gets mixed up in an arbitration case has been around for a solid decade or more and has been an admin most of that time. The major exception in that list is RexxS, and he lasted almost two years as an admin. I think it's tough to paint a picture of an "average" admin who ends up at ArbCom, but it's reasonably clear to me that it's not a newbie admin. Maxim (talk) 13:04, 28 August 2023 (UTC)[reply]
There was a point a couple of years ago when we had lost three of my fellow Brits in a row, all slightly older than me, and all people I had met multiple times in real life. Looking at the rest of that list I suspect that there is a bit of an Atlantic skew in there. There are four other former British admins who I have met in real life on that list, and those seven are definitely not the only non Americans on the list. True I have met a lot of current and former admins in real life, and most of those I've met are my fellow Brits. But it is chastening to go through that list and see the names I recognise and can put a face to. I don't know whether there are some norms of American culture that have become unwritten rules for adminship as enforced by Arbcom, or whether we Brits are more likely to have the free time to become just a bit too active on this site. Or its just that give us Brits a bit of power and eventually it goes to our heads. But I'm seeing a bit of a pattern in that list of desysops and it feels quite personal. ϢereSpielChequers 15:47, 30 August 2023 (UTC)[reply]

A positive sign or collusion?

That many admins have stayed with the project could also be viewed as a positive, unless they have been secretly conspiring to keep out as many newer candidates as possible, although I don't see a clear indication of that in the numbers provided! CurryCity (talk) 19:22, 26 August 2023 (UTC)[reply]

If there is one clear sign from the figures it is that most admins don't take part in the typical RFA. I suspect that most admins haven't !voted in an RFA for more than a year. There are nearly 900 admins, nearly 500 active admins, and the most recent RFA is only the fourth ever to get over 300 supports (many, perhaps most of which will not have been admins). ϢereSpielChequers 19:40, 26 August 2023 (UTC)[reply]
This should be straightforward to check if one of the bot owners would agree to help. Ymblanter (talk) 02:50, 27 August 2023 (UTC)[reply]
I have just gone again through the oppose section of the RFA from the beginning of this month. I counted 14 admins and 57 non admins in the oppose section, of course that one passed and it is possible that there are others where admins are piling into the oppose section. But somehow I doubt it. My impression is that the people in the oppose section are usually not admins. ϢereSpielChequers 12:39, 28 August 2023 (UTC)[reply]
I believe I opposed two or three candidates in my entire life. Usually if I am not sure I prefer not to vote. Ymblanter (talk) 18:26, 28 August 2023 (UTC)[reply]



       

The Signpost · written by many · served by Sinepost V0.9 · 🄯 CC-BY-SA 4.0