Wikipedia talk:Requests for adminship/Inactive admin email

Discussion working up to RfC

Yes, the stewards tend to be rather particular that it must be communicated directly to them, either via private communication from the admin or via m:SRP, I can't see them bending the rules they apply to all 700+ wikis just for us. MBisanz ^talk 20:13, 20 January 2010 (UTC)[reply]

I largely agree with Xeno's changes that you reverted as well. One problem is that inactives probably won't know what a global account is or have an account on meta... Xeno can't we do a local page here just for this? Gigs (talk) 20:04, 20 January 2010 (UTC)[reply]

(edit conflict)I too prefer the revisions that Xeno and Gigs have made. My intent in propsing the draft was never that there would be a default to desysop unless the editor responds--an admin Dead man switch, if you will. The problem of inactive admins is not nearly so great that it requires such a measure, and this is by far the wrong place to claim some sort of consensus has arrisen supporting it. My preferred method of enacting this would be

To make desysopping voluntary
To start an RfC to determine if there is community support
To make a bot that handles the emailing automatically, and sends future emails as admins pass whatever activity cutoff we decide upon

Arcayne, if you'd like to get something done here, may I strongly suggest you drop the personal crusade angle? It isn't helping, and is only further dividing people. As far as the stewards are concerned, that's why we need to put the link in the email, as I don't see anything getting adopted as policyThrowaway85 (talk) 20:16, 20 January 2010 (UTC)[reply]

As long as its on en.wp the steward will probably act on it, but an edit to m:SRP is the best way to go about it. –xeno^talk 20:17, 20 January 2010 (UTC)[reply]

So long as the bot (if it is to be one) posts the reply somewhere in their userspace, or stewards have access to the raw emails (not forwarded, etc), then I could see that happening. Desysopping based on some editor's say-so seems unlikely. Throwaway85 (talk) 20:21, 20 January 2010 (UTC)[reply]

Stewards will want an edit on en.wp or meta. Period. –xeno^talk 20:23, 20 January 2010 (UTC)[reply]

Is there an en.wp equivalent to the meta link you posted? Also, if the global sysops deally goes through, as seems likely, then wouldn't they have to go to meta anyways? Throwaway85 (talk) 20:32, 20 January 2010 (UTC)[reply]

Global sysops is a red herring. We don't have an en.wiki equivalent for m:SRP, but the admin in question could edit their talk page and say something to the effect of "Please remove my administrator privileges effective immediately", and then someone could point a steward at m:SRP to that difflink. Though, it would be more efficient for the admin in question to post directly to m:SRP =) –xeno^talk 20:34, 20 January 2010 (UTC)[reply]

Works for me. If they don't have meta accounts, then that's fine. The goal shouldn't be to desysop every inactive admin or get them back editing, but rather to offer the opportunity to inactive admins to relinquish their bit and help us out on the housecleaning front. It's totally up to them whether they do so. Throwaway85 (talk) 20:39, 20 January 2010 (UTC)[reply]

I'm not sure need a formal RfC for what will be a relatively uncontroversial action, but it might be nice to break this conversation and the draft email off onto its own page because we are veering pretty far offtopic for this page and it looks like we do have resolve to actually get something accomplished here rather than just talk, and we just need some space to hammer out the details. Gigs (talk) 20:42, 20 January 2010 (UTC)[reply]

The RfC was so that we could legitimately sign it as "the community". What about making a bot for it? I sure don't feel like sending 800 emails manually. Anyways, shall we move things to Wikipedia talk:Requests for adminship/Inactive admin email?

Yeah if we want to sign it as the community and have a bot doing this on an ongoing basis then I agree we should do an RfC... probably just a simple talk page one though. I'm going to move this conversation over to the talk page there and leave a pointer on WT:RfA Gigs (talk) 21:05, 20 January 2010 (UTC)[reply]

My thoughts:

If it's a bot job, we should probably make it so it does not do the same person more than once ever. This ensures that no one ever gets more than one email from the bot. If they ignore it then we shouldn't keep harassing them, and if they no longer read the email then we aren't accomplishing anything by sending.
I think that even with a bot, we should just direct them to pages on en and meta where they can request bit removal. As has been pointed out, the idea of wiki-identity is based on making an edit here under that name (or solving a committed hash). We shouldn't create new security holes while trying to close them by trusting emails instead of edits. Gigs (talk) 21:19, 20 January 2010 (UTC)[reply]

I agree with both of those points. A slight modification to your first: Should not message the same person more than once for each period of inactivity. That is, if an editor is inactive for a year, comes back for a year, then is inactive for another year, it seems reasonable to resend the email. Throwaway85 (talk) 21:30, 20 January 2010 (UTC)[reply]

I'm not sure it's a critical point either way, but if we do it that way, we need to pick a longer interval so that people don't get mad about the spam if they are irregular editors who definitely don't want to give up the bit. Another question is whether we should even solicit a reply if they don't intend to give up the bit. If we did, would we track this and then use it as an opt-out list from further emails? Is this overcomplicating things? Should we just ask them to consider giving it up and ignore it if they don't want to? We will lose any data on reachability of inactive admins, but I'm not sure that it is worth the extra complication. Gigs (talk) 21:38, 20 January 2010 (UTC)[reply]

Doesn't hurt to solicit, then at least have it shown as received somewhere. Either way. Throwaway85 (talk) 21:50, 20 January 2010 (UTC)[reply]

I don't know that it's necessary to ever hit someone up with a second email; if they got the first and then came back, there is no reason to assume that their second (or third or fourth) period of inactivity will be any deviation from the pattern that they have sent. It's probably not harmful per se, it just seems a bit like overkill. Shereth 22:38, 20 January 2010 (UTC)[reply]

The purpose of the email is to assess who's intending on returning to the Project and who is not. We know what to do if they respond in the positive (yep, coming back "soon") or in the negative (nope, voluntary de-sysop), but so far, I have not heard about what we do if we receive no reply at all. Are they intending to come back? Are they dead? I'd like to hear a little more about what we do at that point, for I believe that in the absence of reply, we might need to assume that they do not plan to return. - Arcayne (^{cast a spell}) 01:56, 21 January 2010 (UTC)[reply]

We do nothing, of course. –xeno^talk 01:59, 21 January 2010 (UTC)[reply]

Let me see if I understand you, Xeno. The only way that you are willing to consider de-sysoping an editor is if they themselves choose it? How do you address the matter of dead admins or admins who have abandoned the project, either out of disinterest or disaffection? - Arcayne (^{cast a spell}) 02:31, 21 January 2010 (UTC)[reply]

For now that's all we have consensus for. I think it's important that you accept that as the scope of this mini-project at least for now. The numbers and data that come out of this endeavor might be useful for any future discussion regarding non-voluntary inactivity de-sysop. Gigs (talk) 02:39, 21 January 2010 (UTC)[reply]

Arcayne, anything more than what is proposed would require a change in policy, which would require a large RfC with much forum-shopping. This, on the other hand, can probably be dealt with here, with maybe a thread at AN to judge opinion, plus bot authorization should we choose to go that route. Throwaway85 (talk) 02:45, 21 January 2010 (UTC)[reply]

Fair enough. I'll be patient. I guess this is what pulling a freight train by one's teeth feels like. - Arcayne (^{cast a spell}) 02:54, 21 January 2010 (UTC)[reply]

A sound analogy - as the proposal was handily rejected, yet you tried to push it through like a freight train with this "you're inactive! email" Of course the answer is "do nothing" - until the community agrees that inactive administrators may be deadminned. –xeno^talk 19:12, 21 January 2010 (UTC)[reply]

I was proposing a means by which active admins are more appreciated. I'm terribly sorry you are unable/incapable of seeing that. Maybe you could stop trying to provoke me or trying to have the last word, Xeno. What are you so terrified of? - Arcayne (^{cast a spell}) 21:26, 21 January 2010 (UTC)[reply]

So, please tell this non-admin how we track this bot'd email to the inactives? - Arcayne (^{cast a spell}) 19:57, 22 January 2010 (UTC)[reply]

Someone would need the password for the email address registered to the bot, in order to login and check the replies. –xeno^talk 19:59, 22 January 2010 (UTC)[reply]

Why wouldn't you just have the bot post them to its userspace? I thought the issue was moot anyways due to the requirement of the editor themselves posting to meta. Throwaway85 (talk) 20:04, 22 January 2010 (UTC)[reply]

If it did that, it would need to advise the users it was going to reproduce private correspondence on wiki. I think Arcayne is more talking about the emails that said "returning" or "no" or something. –xeno^talk 20:05, 22 January 2010 (UTC)[reply]

Well the holder of the bot account would have access to that, but I'm not sure it's a good idea to concentrate access to information like that, even if they were a well-respected and highly trusted member of the community. Throwaway85 (talk) 05:50, 24 January 2010 (UTC)[reply]

I think we should just dodge that whole issue and only solicit resignations by telling people where and how they can do it. Then we don't need to worry about the return address. Gigs (talk) 03:34, 25 January 2010 (UTC)[reply]

Good plan. A "please do not reply to this email" should suffice. Throwaway85 (talk) 04:10, 25 January 2010 (UTC)[reply]

So what's our next step? I'm a programmer but I have pretty much no MW bot experience. Gigs (talk) 04:46, 25 January 2010 (UTC)[reply]

Shouldn't be too hard to whip something up in php, python, perl, etc, that grabs names from the inactive list, navigates to their userpage, hits "email this user", and sends them the email. I think we should make sure everyone agrees on the content of the email, then take it to WP:RFBOT. As far as the MW bot interface goes, there's plenty of opensource bots out there already. We can ~~steal~~ borrow their code. Throwaway85 (talk) 09:50, 25 January 2010 (UTC)[reply]

On that note, my programming skills are nowhere near where they should be. Gigs, if I grab you some source code for an already-running bot, do you think you can work with it? What language would you prefer? I've seen php and python bots kicking around, could probably find something else if need be. Throwaway85 (talk) 00:31, 26 January 2010 (UTC)[reply]

Local resignation page

Wikipedia:Resign created. Feel free to edit. Gigs (talk) 14:40, 25 January 2010 (UTC)[reply]

Not being programming-savvy, could you explain how this plays out? - Arcayne (^{cast a spell}) 17:36, 25 January 2010 (UTC)[reply]

There's really not much involved from a programming standpoint. Someone codes a bot that goes through our list of admins, determins which ones are "inactive" based up on the criteria we give it, and then it sends them an email pointing them to Wikipedia:Resign if they intend on resigning the bit. The user logs in, posts a request at that page, and stewards will remove the bit; pretty simple. Shereth 17:38, 25 January 2010 (UTC)[reply]

Someone will need to ping m:SRP when requests come in. –xeno^talk 17:39, 25 January 2010 (UTC)[reply]

Yeah, I figured one of us would clerk for them on the local page. It shouldn't be much work. I don't anticipate a huge response after the first week or two. Gigs (talk) 18:10, 25 January 2010 (UTC)[reply]

A few questions spring to mind. First, what inactivity criteria are we going to utilize? I opted earlier for 6 months, but am also amenable to including those who might have been absent for 3-4 months. As well, how do we calculate the number of non-responses? Some admins will see it as a hint to be active again (not a bad thing) and not respond, and others will not respond because they have no interest, or are not checking their email (the dead are so rude sometimes). Lastly, we are qualifying inactivity as being inactive on the wiki-en, right? If someone is a global admin on several meta-wikia, but has never acted within the wiki-en, does their resignation of admin here apply to their global admin status? Sorry, I heard talk of the global admin some months back, but heard that we had chosen to not participate in that; unlike many folk, my question is actually prefaced on not knowing the answer to the question before I am asking it. - Arcayne (^{cast a spell}) 05:27, 26 January 2010 (UTC)[reply]

Global admin is irrelevant really. Yes, we are only interested in en. If they aren't using these permissions then we want them to consider giving them up. They could consider giving up other adminship as well... but to be honest, most of the other smaller Wikis have a much more reasonable admin policy that says that all admins need to be reconfirmed periodically which leads to them expiring. I think en is mostly unique in making adminship until death. Gigs (talk) 14:46, 26 January 2010 (UTC)[reply]

Which other wikis have reconfirmation? I checked Meta, Mediawiki, Commons, Mediawiki, Frwiki, and Dewiki and all of them are admin for life unless the admin goes inactive. None of them use reconfirmations. MBisanz ^talk 14:53, 26 January 2010 (UTC)[reply]

Well, I guess I was mistaken. Many of them do have a process that removes the bit due to inactivity in some way:

Commons: An "inactive admin" is one who has made fewer than 5 admin actions on Commons in the past 6 months [1]
Meta: Any sysop inactive on Meta will be desysopped. "Inactivity" is normally defined as fewer than 10 logged actions in the past six months. Stewards do undergo reconfirmation each election.

I seem to recall at least several of the smaller language wikis doing something similar with reconfirmation/inactivity as well. I can't read the languages so it's hard to check. Anyway my point was, none of them have a massive backlog of inactive administrators like en does. And besides, we can't speak on their behalf anyway. Gigs (talk) 16:02, 26 January 2010 (UTC)[reply]

Might want to refocus the discussion before we get off-track; this isn't really supposed to be a discussion on the merits of reconfirmation or what other projects do it. It's not really relevant to the question posed - specifically, what do we want to define as our criteria for inactivity for the purposes of the mailing? And how to we calculate non-responses? To the first question, I would say that the "6 months with no edits" rule being mentioned earlier makes sense. To the second, I would say that it does not matter; as it stands there is nothing that can be done in the case of non-responses and therefore no value in calculating them. Shereth 16:18, 26 January 2010 (UTC)[reply]

Right. It looks like 6 months is a good standard. Gigs (talk) 16:23, 26 January 2010 (UTC)[reply]

(ec) I recall that MBisanz made that same point elsewhere. Like you, I'm not up on all the languages of the other wikis, but I appreciate Gigs' greater breadth of knowledge of other wikis.

However, I still have an outstanding question. How do we calculate the number of non-responses? Calculating who writes back in to voluntarily release the bit is easy. Harder to calculate (but not impossible, I'd wager) is the number of inactive admins that suddenly become active. What I think we also need to be able to watch those who do not respond at all.

A question from the T:RfA discussion as well: how long do we wait before looking at the results of the email? 7 days? 14 days? 30 days? - Arcayne (^{cast a spell}) 16:24, 26 January 2010 (UTC)[reply]

I think we need to calculate the number of non-responses as well, Gigs. While you are correct in noting that there is nothing that can be done currently regarding admins who have simply abandoned the project (via death, disinterest or whatnot), that may not always be the case, and among the goals of this email is to gather information about those admins who do not respond. Indeed, I think that information is actually critical. - Arcayne (^{cast a spell}) 16:30, 26 January 2010 (UTC)[reply]

I think the only relevant data we can make sure to preserve right now is the last edited date for each editor before the email went out. This gives us the baseline. All other data can be collected later or calculated after the fact. Gigs (talk) 16:40, 26 January 2010 (UTC)[reply]

The math really isn't that hard; Take the number of emails you sent out, subtract the number of responses, subtract the number of admins who are still inactive after the email push, and you have the number of admins who became active again without replying to the email. Shereth 16:48, 26 January 2010 (UTC)[reply]

Well, that would be obvious if - and this might only be a small 'if' - we are able to calculate, say, 30 days out, who was inactive and is now active. Do we have ways of determining that, or is there not going to be any way to determine that? A follow-up question would be whether the list of inactive admins is a manually maintained or bot'd one. - Arcayne (^{cast a spell}) 17:30, 26 January 2010 (UTC)[reply]

A bot would make the most sense unless you can find someone who really enjoys going through 1700+ editors to determine the time of their last edit. A bot could also store the activity state for each account and pretty easily make a comparison from one month to the next to determine who has gone inactive or who has become active, but there's really no need for that. It doesn't really matter which admins became active, all you really need to know is how many did (or did not). Shereth 17:38, 26 January 2010 (UTC)[reply]

I agree completely. Are we going to wait 30 days before discussing the results of the emailing? As I mentioned before, those who are going to respond/react will likely do so by that time. - Arcayne (^{cast a spell}) 12:53, 27 January 2010 (UTC)[reply]

That's probably about the right amount of time, but we'll know for sure based on the rate that replies come in. We don't need to plan everything up front. Gigs (talk) 14:21, 27 January 2010 (UTC)[reply]

When are we to send the email out? - Arcayne (^{cast a spell}) 18:30, 27 January 2010 (UTC)[reply]

I suppose after someone writes the bot and files a WP:BRFA for approval. Maybe ask at WP:BOTREQ if there's anyone available to write the bot. –xeno^talk 20:24, 27 January 2010 (UTC)[reply]

I submitted the request. Someone has suggested that "API has a module for e-mailing users. Should be simple enough to write and I think bots are (or should be) excluded from the usual rate limits". Thoughts? - Arcayne (^{cast a spell}) 18:16, 30 January 2010 (UTC)[reply]

To do as of now

Feel free to edit:

Decide on a bot name -InactiveAdminEmailBot
Decide whether return email should be a noreply or not - en.wiki.inactivemailerbot@gmail.com
Decide how to sign the email - No signature
Decide whether we need an RfC or what. - Looks like we should.

Comments

I don't think we need a full 30 day RfC on this. There's nothing that says an RfC can't run for less time, and 7 days seems customary for giving people notice of things. So I propose a 7 day RfCtag on this page. Gigs (talk) 16:49, 2 February 2010 (UTC)[reply]

Agreed on the RfC part. As for what to sign it with, let's just sign it with the bot's name. I propose "InactiveAdminEmailBot". Esoteric, I know. As for the email address, let's just ask a dev to assign <botname>@en.wikipedia.org, then forward it to the email addy of whoever's going to oversee things. Throwaway85 (talk) 01:10, 3 February 2010 (UTC)[reply]

When the blast goes out, we need to make sure that user is around so they don't quota out. Gmail might be best. Gigs (talk) 05:23, 3 February 2010 (UTC)[reply]

I've registered en.wiki.inactivemailerbot@gmail.com, and will hand it over to whoever decides to oversee it. Does that sound good?

Works for me. Listing the RfC now. Gigs (talk) 22:04, 8 February 2010 (UTC)[reply]

Boldly taking out the signature line entirely. We don't really need a signature. Gigs (talk) 22:12, 8 February 2010 (UTC)[reply]

Heresy! But yeah, I'm fine with that. Throwaway85 (talk) 07:51, 11 February 2010 (UTC)[reply]

RfC - Inactive administrator resignation solicitation email

This RfC is regarding the above discussion and the project page associated with this talk page. Our current plans are a bot named "InactiveAdminEmailBot" which will email the contents of the project page to every administrator that has an email on file and hasn't edited in 6 months or more. The mails with have a From address of en.wiki.inactivemailerbot@gmail.com, which will be monitored. It will not solicit replies to the email, but will direct them to edit a page on enwiki or meta to request lowering of access level. This short RfC is to make sure that there is indeed community consensus for what we are about to do. This RfC will close on or about February 15th assuming there is not any vigorous objections that need to be addressed. The rough consensus seemed to be that we should never bug the same administrator twice, even if we do this again in the future, so the most any administrator will ever get is one email in their lifetime. Gigs (talk) 22:17, 8 February 2010 (UTC)[reply]

Well, I guess no one cares that much. Closing the RfC and removing the RfCtag. Gigs (talk) 19:03, 22 February 2010 (UTC)[reply]

InactiveAdminEmailBot is blacklisted because it has the word Admin in it. I'm going to use a InactivityEmailBot instead Gigs (talk) 14:08, 23 February 2010 (UTC)[reply]

wording

I'm not happy with the currently proposed wording. Remember the message needs to be that we miss you and wonder when you'll be back. - Something like the Strategy wiki's survey. Alternatively we could do a newsflash - an admin account was recently compromised and we'd like to suggest that you revisit EN wiki and make sure your admin account is protected with a strong password. Ϣere SpielChequers 17:56, 24 February 2010 (UTC)[reply]

Please do feel free to make your suggestions by directly editing the project page (and I agree the message should start with some sweetness and then get down to business =). –xeno^talk 18:05, 24 February 2010 (UTC)[reply]

OK I've changed the approach, I think this might do the job and doubt if any reasonable person would object to being asked to use a strong password. Ϣere SpielChequers 20:54, 25 February 2010 (UTC)[reply]

Well, I think it's a better basis to work from than what we started with, but I think the emphasis should be on giving up the bit if they don't need it. Strong passwords aren't much of a concern because the devs already run cracking against the admin's accounts. The primary goal here was to implement least privilege principles as much was possible with a voluntary program. I'll give it a go and we can go back and forth a little until we can get something we are both happy with. Gigs (talk) 03:32, 26 February 2010 (UTC)[reply]

BTW- can't use wiki links since the contents of this will be the email, not a link to this page. I kept most of what you wrote but reworked it a little to try to retain more emphasis on the sysop part. Gigs (talk) 03:38, 26 February 2010 (UTC)[reply]

I agree that the main aim is to get inactive admins who have no intent to return to give up the bit, though I'm also concerned there could be admins out there who haven't changed their password for years and such passwords might have become compromised. But I think that a more diplomatic approach might work better and avoid people responding along the lines of "can't someone spend a couple of years doing a degree/having a baby" etc. Whilst I hope that those who are irritated by getting a security warning about something they have long finished with respond by resigning their adminship. As for the wiki link - yes they can only use that by going to Wikipedia. Ϣere SpielChequers 09:16, 26 February 2010 (UTC)[reply]

Interesting, I didn't think about accounts that may have already been compromised. Gigs (talk) 13:04, 26 February 2010 (UTC)[reply]

I was actually thinking of accounts which were still under the admins control but where they knew the password to have been compromised. However there is a risk that you are emailing a completely different person. For example if someone used a work account for their Wikipedia Email and they have subsequently changed jobs, I know of companies where mailboxes of former employees are monitored, or worse if someone was foolish enough to use a role account such as sales@ or support@ that account might now be in the possession of a completely different person. Ϣere SpielChequers 08:52, 12 March 2010 (UTC)[reply]

Done

195 sent 76 with emails disabled. Wikipedia_talk:Requests_for_adminship/Inactive_admin_email/results Gigs (talk) 17:59, 23 March 2010 (UTC)[reply]

No apparent replies thus far. I guess if we've proven anything, solicitation is not an effective way to eliminate inactive administrator bits. Gigs (talk) 15:08, 30 March 2010 (UTC)[reply]

If they're anything like me, they have a separate email just for wiki business, and are probably not checking it along with not logging in to the 'ped. –xeno^talk 15:18, 30 March 2010 (UTC)[reply]

I suspect a lot of them are very unlike you. We have hundreds of admins who only made a handful of admin actions and /or never made more than a few thousand edits. I had >10,000 edits before I got round to creating a separate account for Wikipedia, but yes I'd agree that a fair few of those emails will be sitting in accounts that are no longer used. However I wouldn't assume that no replies has to mean no result. At the beginning of March we had 1720 admins, we've appointed two, had one unretire, gained one admin bot and lost another and had one admin desysopped by arbcom, but instead of a net increase of two we have a drop of three. Now its entirely possible that there are other explanations for our drop of five admins, but there could be up to five resignations resulting from that email. Ϣere SpielChequers 15:59, 30 March 2010 (UTC)[reply]

So let's troll the logs and see what happen! –xeno^talk 16:07, 30 March 2010 (UTC)[reply]

39 admins listed on WP:LOA/I as of March 1 are no longer there today: [2] (list), though all of them still have +ops except for the one decedent desysopped by arbcom mentioned above. So perhaps we rustled them out of their slumber? Still a net benefit! –xeno^talk 16:15, 30 March 2010 (UTC)[reply]

I'm sure a few changed their password and updated their email as well, though that's harder to measure. It wasn't all for naught. Gigs (talk) 16:45, 30 March 2010 (UTC)[reply]

Damn freedom fighters and their thunder-stealing! [3]. –xeno^talk 18:10, 30 March 2010 (UTC)[reply]

If you e-mail all the inactive admins, you'll find there's more than 76 you can't reach. All the old admins who haven't edited since before 2007 cannot be reached by e-mail. Back then, it was different as far as having an e-mail address on record. Enigma^msg 19:37, 7 April 2010 (UTC)[reply]