User:SamuelTheGhost/Re-electing admins

The proposal

 * 1) Admins or ex-admins to be permitted to stand for election or re-election at any time, using the current procedures.
 * 2) A process (bot or other program) to be introduced which will once per day select the admin whose most recent successful RFA or RFB was longest ago; providing that RFA or RFB was at least two years earlier, send an expiry of term message by email and as a note on the admin's user page. If the selected admin does not successfully stand for re-election within six weeks, admin status to be honourably withdrawn.

Intended consequences
There are two desired and probable consequences of this proposal.
 * 1) A large minority of the current admins can be described as "inactive", that is, they seem to have lost interest in administrative action or even in wikipedia altogether. Presumably those admins would not seek re-election, and the number of nominal admins would be pruned down to the active ones.
 * 2) A small minority of the current admins have lost the confidence of the wikipedia community, even though they have not done anything bad enough to warrant arbcom or steward action as in Removing administrator rights. This proposal provides a fair (though slow) means of removing them.

Quantitative aspects
In round figures, there are currently 1,700 admins, 900 of them active, 500 semi-active and 300 inactive. About half the inactive ones have not made any edit at all within the last year. The potential re-election load would be 7 per week, maximum, but if the admins who are not fully active retire gracefully, in practice this might reduce to about 5 per week. This does not seem excessive.

Since 365 admins per year would be prompted for re-election, it would take nearly 5 years before today's newly elected would come up for re-election. Assuming that the inactive were gradually eliminated, the number of admins might eventually stabilise at around 1000, and when that happened the expected time to re-election would be a little under three years. The proviso in the proposal that at least two years of admin status are guaranteed might never come into play, but it would have the consequence of preventing the number of admins dropping below about 730.

Tweakings

 * The most important parameter which might need alteration in the light of experience is the rate of one retirement per day, which was chosen because it's simple and about the right magnitude. It could easily be altered to be once per 18 hours, or once per 30 hours, for example. The aim is to make it as fast as possible without creating an excessive re-election load.


 * The figure of six weeks notice could easily be altered if necessary. Its purpose is just a courtesy to retiring admins. It was chosen so as to allow ample warning time, even in holiday periods, while not so long that the matter would be forgotten.


 * As noted above, the two-year minimum would quite likely never be used.


 * Exceptions to the general requirement for re-election might be made for privileged people like arbcom members and Jimbo.

Comparison with previous proposals
There is a massive list of previous proposals at Requests for de-adminship. The current proposal differs from all of them in at least one of the following respects.
 * 1) Most of the proposals on the list are concerned with de-sysopping of targetted admins who are felt to be unsuitable, on some criteria and by some process. Identitfying the criteria and process then fails to achieve consensus. This proposal is completely even-handed, requiring confirmation of admin status without any prejudice.
 * 2) A few of the proposals, such as Adminship renewal, suggest periodic re-election, but using unrealistically short re-election periods, and without taking into account the burden of repeated RFAs. This proposal takes the limiting of that burden as its starting-point.

Useful links

 * List of administrators
 * Perennial proposals
 * Requests for de-adminship

Comments
Please comment here or on the talk page. Maybe I'll add a FAQ section if that seems necessary.