User:UninvitedCompany/sandbox

Inactivity policy

There have been a number of proposals in the last year to amend the inactivity policy. In general, these have either failed to attract consensus, or have resulted in relatively minor policy changes that do not address the needs of the project. This proposal is grounded in an attempt to learn from the comments made in other discussions.

Brief history

The first administrators were appointed in 2002. Administrator activity was not seen as a problem when Wikipedia policies were first formalized in 2005, and administrative rights could be held indefinitely. There was an awareness that some administrators had become inactive, but they were far outnumbered by the rapid growth in the number of new administrators. In June 2011, the first inactivity policy was adopted. Adopted in the wake of vandalism from a compromised, dormant administrator's account, ti was primarily intended as a security measure to remove admin status for editors who no longer had access to their accounts or who had left Wikipedia with no intention of returning. The mechanics of the policy were chosen to avoid upsetting any administrators affected by the policy who still hoped to contribute to the project. The original policy allowed administrators who had been desysopped for inactivity to receive administrative rights upon request with no time limit.

252 users who were inactive at the time the policy was adopted were desysopped.

In 2012 and again in 2018 the policy was amended slightly so that former administrators could no longer obtain the automatic return of the administrative rights (without making a request and achieving consensus support at WP:RFA) after certain time periods elapsed. These provisions have had a practical effect in only a handful of cases.

The number of inactive administrators

There are presently about 1200 accounts with administrative rights. Of these, about 500 of them have averaged more than half an edit a day over the last two months while 700 have not (lists are split [1], [2] by degree of inactivity). By manually checking a sample of the contributions and logs for some of the 700 less active accounts, it is possible to infer that:

1. Typically the recent contribution history shows routine edits in a relatively narrow topic area, rather than patrolling for vandalism, deleting pages, blocking users, protecting pages, etc.
2. Most joined the project in 2005, 2006, or 2007. (These are also the years when the largest number of users became administrators).
3. Most of these accounts have not been regularly active in administrative matters for several years or more.

While the "half an edit a day" standard is not perfect, a manual review of a sample of the contributions of logs shows that it provides a reasonable indication of the rough number of admins involved in the project on a regular basis. The administrators in the "active" list typically make at least several hundred edits a year.

Effects of the current inactivity policy

For 2018, 49 administrators had their administrative rights removed under the existing inactivity policy. An approximately equal number made edits in response to the notification messages, and therefore retained their administrative rights.

For 2017, 55 administrators were affected.

Resysop policy

It is relatively uncommon for inactive administrators to return and request their tools back. Fewer than half of those who request the tools back use them regularly after receiving them.

• 9 users resysopped in 2018 (4 had been inactive, 5 had resigned)
• 20 in 2017 (9 had been inactive, 11 had resigned)
• 15 in 2016
• 28 in 2015

All users resysopped in 2018 after losing adminship due to inactivity

1. Al Ameer son (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - active
2. Liz (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - active
3. Gogo Dodo (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - intensively active at times, gone for several months now
4. KaisaL (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - not active after resysopping

All users resysopped in 2017 after losing adminship due to inactivity

1. Longhair (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - active
2. Dlohcierekim (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - active
3. Berean Hunter (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - active
4. Anetode (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - not active after resysopping
5. Doug Bell (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - not meaningfully active after resysopping, no logged admin actions other than restoring and editing own article after deletion at AfD)
6. Karanacs (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - not meaningfully active after resysopping, one logged admin action
7. Kim Dent-Brown (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - very brief period of activity after resysopping
8. Deckiller (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - made a dozen logged admin actions in one day and hasn't used the tools since
9. Happyme22 (talk · contribs · blocks · protections · deletions · page moves · rights · RfA) - occasionally contributes to articles but did not use admin tools after resysopping

Problems posed

Accounts with administrative rights that are not regularly used to serve the project pose a number of problems:

1. They are susceptible to compromise and use for vandalism. While accounts in active use are also susceptible to compromise, the sheer number of inactive accounts means that we are perhaps doubling the exposure we might otherwise have.
2. There is some risk that the accounts will be sold, transferred to friends, used for paid editing, or otherwise used in a way that is at odds with the goals of the project. While there is only one such case that has come to the community's attention, it would be naive to believe that there are no undiscovered cases.
3. Though rare, a handful of these users have returned to edit disruptively without due regard for community norms. This sort of thing is not unique to Wikipedia, see meatball:VestedContributor.
4. The occasional content changes characteristic of mostly-dormant administrator accounts may be given undue weight for a user who is not currently closely involved with the project. This is due both to whitelisting and to social factors.
5. The presence of very large numbers of inactive accounts distorts perceptions of the number of administrators, and the change in that number over time, complicating policy discussions
6. Some more recently arrived Wikipedians may perceive a lack of fairness, particularly among long-inactive administrators who would not meet the community's expectations to become administrators under current standards.
7. Some approaches to making accounts more secure pose dollar costs to the WMF per account (such as security tokens) or require additional WMF staff proportional to the number of accounts (such as 2FA reset requests), or both. The presence of large numbers of inactive accounts inflates these costs and limits the number of approaches that are feasible.

Voluntary resignation

In December 2018, there was an effort to encourage long-inactive administrators to either return to the project or voluntarily resign their adminship. This effort was largely unsuccessful:

• Individually written messages were sent to 15 long-term inactive administrators who make only occasional edits. Each of these editors had been inactive for over five years and averaged under 10 edits a year during their period of inactivity. Messages were left on talk pages and sent via email. Particular effort was made to individualize the content, emphasize the value of each editor's past contributions, and encourage re-engagement. Editors were encouraged to resign their adminship if they did not intend to rejoin the project. Of the 15 editors contacted, 14 did not respond in any way, and 1 resigned.

• The bot-delivered inactivity messages were modified to emphasize that inactive editors were encouraged to rejoin the project in earnest, or resign if they did not intend to do so. These changes first took effect for the second notices sent out December 25 (for the January 1 round), and for the first notices sent out on January 1 (for the February 1 round).

1. For the January 1 round, there were initially nine inactive administrators. None resigned, and six made a handful of edits. None has returned to active editing.
2. For the February 1 round, there were initially eleven inactive administrators. As of January 4, one has resigned, and none have made edits.