This is an essay.
It contains the advice or opinions of one or more Wikipedia contributors. This page is not an encyclopedia article, nor is it one of Wikipedia's policies or guidelines, as it has not been thoroughly vetted by the community. Some essays represent widespread norms; others only represent minority viewpoints.
In a recent bureaucrats' noticeboard thread, I made a comment on the current state of our administrator inactivity rules. In brief: if an admin goes 12 or months without a single edit or logged action, the +sysop flag is removed. While the original policy in 2011 allowed for restoration at any time by request to WP:BN, it has since been amended a few times to provide for cases where the flag cannot be restored via a simple noticeboard request. I think the crux of my comment at the time (while buried in a bunch of text) is that our current inactivity policy and their attendant rules, are a demonstration of the legal maxim that hard cases make bad law. What I hope to demonstrate (in this, a greatly expanded comment) is that as a community, we have inadvertently made a series of mistakes over the past decade when trying to quantify administrator inactivity, and that perhaps we have not had the discussion over what role administrators ought to have on the project.
No big deal
Through the 2000s decade, adminship was largely seen as no big deal. The term originates from a 2003 post by Jimbo Wales in an attempt to emphasize that adminship is meant as a technical role, instead of conferring a user any particular authority on the project. A perusal of the promotion statistics by month would suggest that we made administrators at a much higher rate than is likely possible now. The requirements were not nearly as stringent; in 2007, 3–6 months of sustained editing with a few thousand edits and a demonstrated need for the tools was a recipe for a successful RfA (provided the candidate hadn't been involved in disputes or otherwise pissed anyone off too much). I recall there discussions on WT:RFA at the time whether the requirements were too high—what a difference 13 years has made! Nowadays, a 18–24 month long editing record with perhaps closer to 10,000 edits, demonstrated experience in both article writing and maintenance activities, and not pissing too many people off is the more typical profile of a successful candidate. This change alone may be indicative that making (and being) an administrator nowadays is a much bigger deal than it used to be.
My conclusion in the previous paragraph is probably not an outlandish opinion. If we are to take this conclusion as true, then we are overdue to have a discussion on what role administrators have on the project. Is it purely technical, that is, some trusted users have additional access to certain tools and can perform certain actions under a role-based access control framework? Is it purely political, where RfA is used to elect users that are empowered to make certain decisions on behalf or with community, and that the fact such users have the technical means to do so being more of a by-product of the system? Arguably, the present system is an expression of a spectrum, where an administrator's role is some mixed parts technical and political; however, perhaps we select administrators based on a more political role, but our inactivity policies are much more shifted to the technical role.
In the distant past, admins were selected based on a combination of someone volunteering to do maintenance tasks and an implicit trust that someone entrusted with higher-level access wouldn't use it maliciously (i.e. a technical role). Since those times, community norms have been significantly more formalized (for a succinct explanation, see one of Risker's arbitration essays). In other words, admins used to be chosen because someone has to do the "janitorial" tasks, and many hands would make light work. Even if an administrator didn't do that much work, it was still something. Were such an approach applied today, we could probably safely make most users with 10,000+ edits administrators, with exceptions being the chronically disagreeable types such as frequently incivil editors, edit warriors, and others under active community sanctions (the list of exceptions isn't meant to be exhaustive).
Consider closing a contentious RfC, AfD, or community ban discussion; with limited exceptions for RfCs or AfDs, such closes are the domains of highly active and respected administrators. A legacy admin, that is, someone promoted over a decade ago and who hasn't been outwardly too active, would have a tough time making such a close, perhaps stemming from a lack of political capital. No one's very familiar with this admin, whether by long-term reputation or through recent actions. A recently (but perhaps not too recently) promoted admin would not have to face this issue. The modern RfA process is quite good at finding admins who are highly experienced and competent in most areas of the project (perhaps it over-aggressively filters out otherwise decent prospective candidates who find it too overwhelming).
If such legacy admins can't really have a political role (absent making more edits and actions!), perhaps there is the technical role. In terms of inactive or semi-active legacy admins, one ends with a sort of a circular problem, where these admins aren't exercising a technical role, by reason of inactivity. At any rate, the distribution of work done by administrator corps as whole largely follows the Pareto principle; were we to remove the sysop flag from a big chunk of the group (probably in excess of 50%), there won't be a significant increase in our administrative backlog. Over the past few years, I think I've made it less of a secret in discussions that I'm not a fan of our inactivity policy and I'm even less of a fan of trying to deal with the resysops. We try to quantify inactivity, but the only criterion that has consensus is so weak that it seems that most of its function is to make it look like we do something about administrator inactivity. Over time, the flaws of this criterion have led to amendments to the circumstances of when bureaucrats can restore the flag without RfA that I fear will lead to more amendments in attempt to tighten ever-spawning loopholes.
Over the past many years, I've had a keen interest in looking up editing stats of many administrators in the context of inactivity. I have resisted giving examples, because I'm leery of singling individual administrators in a manner that could be seen by others as prejudicial. That said, I think that if there is consensus that adminship encompasses both technical and political roles, it would be useful to significantly tighten up our inactivity policy, and it would be instructive to consider actual examples of administrator (in)activity. Full disclosure: I proposed 50 edits/year with RfA always required, about 3 years ago, which failed.
At one point or another, all of these individuals were highly active and respected editors; that they have largely but not entirely moved on from the project should not make them any less respected. My very unscientific estimate of how long someone is highly active on Wikipedia is, as a median, 18+/-6 months, with continuous high activity of 24+ months much more common than <12 months. These examples are picked from administrators listed at WP:ADMINLIST, WP:INACTIVE, or by looking at Special:ListUsers; again, this was not done particularly scientifically. The common idea here is that if these individuals were to have their flag was removed, and come to WP:BN to request restoration, there would be at least a heated discussion.
- Camembert : last 50 edits go back 8 years. Last logged action 2010. Less than 50 total actions. Former arbitrator. Mostly active mid 2002 to mid 2004, with a diminished level of activity through late 2005.
- XJaM : last 50 edits go back 7 years. Last logged action 2010. Less than 50 total actions. Active on slwiki. Mostly active early 2002 to somewhere between 2010 to 2012.
- Oliver Pereira : last 50 edits go back 5 years. Last logged action 2012. Less than 50 total actions. Mostly active late 2002 to early 2004.
- Thehelpfulone : last 50 edits go back 5 years. Last logged action 2018. Last 50 logged actions go back 7 years. Highly active late 2007 to mid 2009.
- Tim Ivorson : last 50 edits go back 2 years. Last logged admin action 2016. Less than 50 total actions. Particularly active mid 2004 to mid 2006 with gap in late 2005.
- Stemonitis : last 50 edits go back 3 years. Last logged admin action 2017. Last 50 logged actions go back 4 years. Highly active mid 2005 to mid 2007 and mid 2009 to mid 2011 and then in occasional bursts to 2016.
- Staxringold : last 50 edits go back 3 years. Last logged admin action 2019. Last 50 logged actions go back 10 years. Highly active mid 2005 to mid 2007 and mid 2009 to mid 2011 .
- HappyCamper : last 50 edits go back 7 years. Last logged admin action 2019. Last 50 logged actions go back 9 years. Highly active early 2005 to mid 2007.
- Xymmax : last 50 edits go back 1 year. Last logged admin action 2020. Last 50 logged actions go back 1 year. Highly active late 2007 slowly phasing out into the mid-2010s. Spurt of activity late 2018 to mid 2019, and almost inactive since.
- Wwwwolf : last 50 edits go back 7 years. Last logged admin action 2019. Last 50 logged actions go back 9 years. Highly active starting 2005, peaking in 2006, and exponentially decaying over about a dozen years.
It is up to you, dear reader, to consider who, of these examples, could be considered meaningfully active, and how that may relate to a tightened inactivity policy.
In a perfect world, we could encourage all these valued administrators to return to active editing, but I don't think our encouragement would necessarily successfully. In many cases, over 15 years have passed, and a lot will change in a person's life in that time. Interests change, free time deceases (or increases), and so on. These aforementioned examples aren't particularly unusual, and I would argue these administrator are not meaningfully active. But we seem to spend more time discussing how to enforce the present mess of when a desysoped inactive admin (after a year with no edits) can or cannot request resysop at WP:BN. It seems to be a yearly or twice-yearly tradition.
Someone else has floated an "administrator emeritus" flag for inactive admins. I think that's a nice way to recognize our now-inactive admins' contributions in the past. As unfortunate as it is, we have plenty of volunteers with technical access to a role that has evolved to be much more political, and I don't think that's ideal.
Personally, my first choice would be a system where consensus is reached to remove the sysop flag for inactivity (but based on roughly codified guidelines; inactivity to me has always been equal parts quantitative as qualitative; I'm sure there plenty of exceptions that can be made in cases of low activity or even activity that appears to be high); my second choice would be to not bother desysoping inactive admins; and lastly to keep our current system.