Wikipedia talk:Requests for adminship/stats

From Wikipedia, the free encyclopedia

[edit] 1000 admins - some stats

The number of admin actions performed by each active admin in each month (on average)

The number of admins who performed at least one administrative action in each week

The number of admin actions performed per week, by number of admins

I was inspired when the admin clock recently ticked over 1000 to do some research into how many admins are actually active. As you'd expect, there are not really 1000 admins doing admin actions. The methodology was simple - get a list of admins from Special:Listusers/sysop and for each admin, retrieve and analyse all their admin logs (see me for the script or the data). For some reason, admin logs do not exist before 23/12/2004, and for ease I rounded the dates to actions between 01/01/2005 and 31/08/2006 inclusive - a period of 20 months. What follows is an analysis of active admins and their admin actions between those dates. It includes 1,388,273 admin actions. Admin actions are simply defined as: block, unblock, delete, restore, protect, and unprotect. It does not include other admin priviledges such as editing protected pages, nor non-admin actions such as commenting at AN/I or uploading images, nor additional rights such as bureaucrat actions. If a user has sysop rights and performed one of the admin actions then they are included here. See me for further methodological details.

This may not be the right place to describe this research, but since this is the main place where the number of admins is determined, I think it is appropriate to feed it into the other stats that have appeared here. Perhaps voters should be forced to read it before they reject applicants for "missing one edit summary" or "no featured article", and perhaps bureaucrats should read it before determining consensus in such cases. That is my hope anyway. Wikipedia needs more admins. This research however is objective and impartial (see me for details). Here are the prelimary results:

The number of admins active (have made at least one administrative action) since 1 Jan 2005 is 941. The number who have made an administrative action since 1 Jan 2006 is 873. The number who have made an action in the last two months is 705. In the last month (Aug 2006), 628 admins have made an administrative action. There were 232 admins active in January 2005. See graph for the weekly trend.

In terms of the number of admin actions, in January 2005 there were 19,101 actions (consisting mainly of 16,598 (87%) deletions), and in Aug 2006 there were 149,007 actions (consisting mainly of 123,413 (83%) deletions). Therefore in January 2005 each active admin performed on average some 82.33 actions, and in August 2006 the active admins performed some 237.27 actions each (per month). See graph for the trend.

As you'd imagine, the workload among admins is not spread evenly. Amazingly, the most prolific admin has performed 32,986 actions (mainly 31,952 deletions) since 22 June 2006. Only one admin has more actions to their name, at 33,500 (mainly 26,558 blocks, and 5,148 deletions) but this has taken over 74 weeks to achieve. Over half of all admin actions since Jan 2005 have been made by just 59 admins, and three-quarters of all admin actions have been made by just 169 admins. In order to take the length of time each user has been an admin into account we can calculate weekly and monthly rates. Only 147 admins average above 70 actions per week (the current average). Coincidentally, 147 admins average just one action per week or fewer, and 460 admins have averaged 10 or fewer actions per week since they were sysopped (this does not include admins who haven't done an action since Jan 2005 (inactive admins)). See graph for distribution.

I will leave it there for the moment. There is plenty more where this came from. Please ask me if you want the data or more specific analysis or details. Employ more admins. Discuss. Jim182 13:49, 24 September 2006 (UTC)

I personally think the last one is funny. ;) Highway ^Daytrippers 13:53, 24 September 2006 (UTC)

Note - I had to 'top' that graph at 1000 because of the single admin who averages 3000 per week (mainly image deletions) - the only one to average above 1000 actions. Without such adjustment the graph would look even funnier ;-) Jim182 14:14, 24 September 2006 (UTC)

• If easy to do, could you plop the data on a subpage in your user space?
  This data looks good to me. A lot of admin actions take thinking and consideration. It is a mistake to measure the value of admins by the number of actions. I loved Curps' bot when it was around, but it would have been a mistake to assume each action that bot took was of equal value to all other admin actions. These graphs show that we have 500 to 600 admins who are doing good work. Part time admins are just as needed as full time ones, this is a volunteer project afterall. This looks very good, almost all admins are doing work except the ones that are actually on semi-active or inactive status. Excellent. NoSeptember 14:18, 24 September 2006 (UTC)

I have dumped the relevant parts of the data here. I would also like to congratulate the active admins for a sterling job - the best indicator for that is the (low?) number of de-sysoppings or complaints against admin actions. This data has no relevance to that quality measure. As I said, this study was objective, but it does lead me to think what I thought before - that we need more admins. These stats really show one thing clearly - an increasing workload for each (of a select few) admin, and a quite-prominent levelling-off of the number of active admins depite ever-increasing articles and editors (and stuff to delete). Jim182 15:32, 24 September 2006 (UTC)

Is editing fully protected pages included in your counts? This is an admin action as well. --Ligulem 23:36, 24 September 2006 (UTC)

I mentioned above that it is not. The reasons are mainly methodological, since these edits are not recorded in the action logs, but equally they are likely to represent a small number as editing fully protected pages is in many cases discouraged. Additionally, it is unlikely to affect the number of active admins (since in most cases the editing admin is quite likely to have been involved in protecting the page). Nor have I included Wikimedia edits, viewing deleted histories, nor edits such as "do that again and I'll block you - I'm an admin don't you know", for similar reasons. It would have been nice to include them, but they are unlikely to affect the trends seen here. Jim182 00:12, 25 September 2006 (UTC)

Sorry. My bad. I didn't read carefully. Nice data presentation. Per the editing protected pages I was thinking about fully protected templates and editing in the MediaWiki namespace. These things tend to get overlooked. There are not only articles that must be maintained. More and more templates are getting used broadly and more of them are fully protected. --Ligulem 11:21, 25 September 2006 (UTC)

You could measure our use of rollback. Just look at our contributions and count them. hee hee hee ;). NoSeptember 02:56, 25 September 2006 (UTC)

This is really interesting data, Jim. Thanks for taking the time to compile all this. Firsfron of Ronchester 05:23, 25 September 2006 (UTC)

Looking at the second graph in particular, I'm reminded of a joje current in the UK in the '70s. "How many people work in British Leyland?" "About half." Alai 17:50, 12 October 2006 (UTC)