User talk:Tamzin/wild ideas/Unsucking unblocks

Common block concerns

Sock/meat puppeting

As a new admin reviewing unblocks, one of the bigger issues I encounter is how to deal with alleged sock/meat puppets. These typically fall into one of two camps:

1. I'm not a sock!
2. I'm done socking!

Either way, we have to review the information available through SPI and potentially call on CU. Unfortunately, many sockblocks occur without an official SPI and/or are CUblocks. As such, the evidence for whether the individual actually used multiple accounts is unclear. Other times, there may be an SPI, but to me, the evidence does not feel strong enough to have blocked. However, as someone who doesn't do much work in SPI, I also don't know exactly what I should look for.

Regardless, sock/meat puppet and CU unblock requests feel like they go cold because there isn't as much third-party admins can do. Significa liberdade (she/her) (talk) 13:38, 17 December 2024 (UTC)

So, not to pick on you, but I just randomly picked a sock block appeal off the list, at User talk:Wwew345t It appears you were sent evidence by the blocking admin seventeen days ago, but the request is still open. This suggests that if another admin wants to review this, they need the same evidence you have already seen, but it also begs the question of why, after being provided this evidence, the user has just been in limbo with an open request. If the evidence was compelling, a decline was in order, if it wasn't a rope unblock is in order, but neither of these things have happened. That neither has happened is just confusing to someone new to this particular situation, I have no idea how I'm supposed to review this appeal. This sort of thing is what makes other admins just walk away from this area. And the user has been asking as recently as yesterday what else they can do, and has been ignored. I don't know what to tell them as I don't have this evidence. El Beeblerino ^{if you're not into the whole brevity thing} 18:51, 17 December 2024 (UTC)

This is actually a perfect example of one of the issues I have with dealing with users blocked for sockpuppetry. I received evidence from an admin that showed "obvious" sockpuppeting, but when I looked at the evidence, I didn't see anything that would have even flagged my suspicions. As such, I'm left with two assertions: a) I haven't spent much time looking at socks and thus, don't know what to look for or b) this admin is very wrong. As such, I don't feel confident enough to block or unblock because I don't know what I'm looking at. Yes, I should be doing something, which I recognize I haven't, but I'm not sure what that something is except potentially sharing the evidence with other admins to see what they think. However, that also feels wrong because it was shared with me via email rather than shared publicly. This is just one specific scenario, but I feel like I face those two assertions a lot when dealing with sock puppets, and I'm not sure what to do with it. As such, I've recently decided not to handle sockpuppet unblocks. Significa liberdade (she/her) (talk) 21:30, 17 December 2024 (UTC)

I figured it was something along those lines. Sock blocks are kind of the worst as the reasons are often vague. And to be clear, the problem is clearly systemic, this was just literally the first thing I clicked on. Having looked at several more, some sock blocks and some not, I'm seeing this again and again. Discussion just fizzles out and nothing happens. I was going to object to Tamzin's proposal that this new board needs to keep appeals open for a week, but that would actually be considerably less time than most of these appeals are open. El Beeblerino ^{if you're not into the whole brevity thing} 21:35, 17 December 2024 (UTC)

Looking at the proposal again, it appears it would be a week for declined proposals, but an accept may be closed at any time (I think). This would address the issue of some unblock requests being immediately declined for non-serious reasons (e.g., formatting), which may result in editors either deciding to stop trying, sockpuppeting, and/or going through several unblock requests, which can also exhausting to review for those working with unblock requests. Significa liberdade (she/her) (talk) 21:58, 17 December 2024 (UTC)

I don't know, I don't think these are too bad to deal with, provided you get a CU who gives you literally anything whatsoever to go on. In both cases you mostly end up either handing out or accepting a WP:SO request. -- asilvering (talk) 18:57, 17 December 2024 (UTC)

That's the thing, evidence was asked for, was provided, and was reviewed, and then ... nothing happened. I've just done a few more and they were frankly worse. At User talk:Mr Nerd 96, there were three admins talking to the user, they were doing their best to reply, and again, nothing happened. At User talk:A1Cafel I count seven admins in the discussion, and the user was left hanging for over a month with no decision. I was expecting to find a lot of requests that were just being ignored, but instead I'm finding one after another where the matter was discussed at length but nobody actually did anything. This is clearly a systemic issue that needs .... well something as this just isn't working. El Beeblerino ^{if you're not into the whole brevity thing} 19:14, 17 December 2024 (UTC)

@Beeblebrox, I wasn't replying to you, as you can see from the indentation. -- asilvering (talk) 19:25, 17 December 2024 (UTC)

Meh. I think the point still stands. I'm still digging around and just finding more and more that are like this, a discussion is had but nobody does anything, either unblocking or declining. El Beeblerino ^{if you're not into the whole brevity thing} 19:34, 17 December 2024 (UTC)

I'm also seeing a lot of "you must tell me what specific edits you would make before I will even consider unblocking you" which has never been a rule as far as I know. El Beeblerino ^{if you're not into the whole brevity thing} 19:55, 17 December 2024 (UTC)

From my perspective, asking questions about the types of edits a user would make if unblocked (and trying to be more specific), helps me see what the user has learned. For some of these, it's wondering whether they have any interest in editing beyond their COI, which could cause more disruption. Other times, I want to see if they understand how to properly cite, for example. Significa liberdade (she/her) (talk) 21:36, 17 December 2024 (UTC)

I wouldn't have blocked them, but ...

Especially as a new admin who may be more tolerate to newbie behaviour, I regularly find myself thinking, "I wouldn't have blocked this person, but accepting an unblock request is different than not having blocked in the first place."

For example: UserX with 25 edits creates a promotional article in the mainspace. My tolerance for promo is quite high, so I might tag or draft it and send UserX a message about NPOV. Other admins might block. If I unblock and UserX continues making promotional content, that's on me.

I think this type of unblock may greatly benefit from a system where multiple people can review an editor's history and potential. It's not on one administrator to decide whether a user will eventually become a beneficial contributor, and thus, it's also not on one person if that user continues to be disruptive after being unblocked. Significa liberdade (she/her) (talk) 13:38, 17 December 2024 (UTC)

I do think we can reduce some of the backlog from the other end of the process, by prioritizing warning first as opposed to the numerous no-warning blocks that are handed out now.

Getting the admin corps as a whole to abandon that approach now when it has become entrenched will prove difficult, but I think it needs to happen. El Beeblerino ^{if you're not into the whole brevity thing} 18:37, 17 December 2024 (UTC)

I think this would be huge. An ounce of prevention, right? Significa liberdade (she/her) (talk) 21:41, 17 December 2024 (UTC)

More warning before blocks would also shift work elsewhere, though, onto the anti-vandal/anti-spam corps. That may well be worth doing under the circumstances, but the trade-offs merit consideration. —Compassionate727 ^(T·C) 01:39, 18 December 2024 (UTC)

I think we could do with a lot more education and a lot less blocking, but the mindset of the average admin would need to change dramatically. Generally when I've asked an admin why they blocked some editor I was trying to help, I have received either no response or a wtf-type response, in the face of which I don't feel comfortable unblocking. Espresso Addict (talk) 03:13, 18 December 2024 (UTC)

Community Ban for failed appeals?

Resolved

 – Please remove this tag if reöpening discussion.

One thing this proposal doesn't seemingly address (or maybe I missed it) is whether or not a failed appeal gets converted into a WP:CBAN. Is a failed appeal considered a community endorsement of the block if criterion UB2 cannot be established? –MJL ‐Talk‐^☖ 16:34, 17 December 2024 (UTC)

I hope not. That would make getting unblocked significantly more difficult, which I don't think is the intent of this proposal in general. -- asilvering (talk) 18:53, 17 December 2024 (UTC)

Nice to see you, MJL! No, as I say at one point, this is not a community consensus process, any more than the current unblock system. I'll clarify that that means a decline isn't a CBAN, though. I also meant to add a line about referring to AN in cases where community input would be desired. I'll do both when back at a computer. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 19:25, 17 December 2024 (UTC)

@MJL: How does User:Tamzin/wild ideas/Unsucking unblocks § Community bans look? -- Tamzin^{[cetacean needed]} (they|xe|🤷) 04:16, 18 December 2024 (UTC)

@Tamzin: I like it! :D –MJL ‐Talk‐^☖ 06:51, 18 December 2024 (UTC)

Doesn't address one of the major issues

There is a lack of admins (generally) which also leads to a lack of admins taking on the task of reviewing unblocks. There is an increase of burden here with the outlined clerking. Additional input from non-admins is a benefit, though. That's not to say this is a bad idea, but as long as there are too few admins available to handle unblock requests unblocks, regardless of venue, will suck. ScottishFinnishRadish (talk) 17:43, 17 December 2024 (UTC)

I think there are enough admins, it's just that, frankly, this area is depressing. and admins don't want to deal with it. Many appeals are walls of text, obvious lies, or pathetic begging without addressing the actual reasons for the block. Many admins just don't want to deal with it, and some of the admins that do seem awfully jaded, to the point where it is near-impossible to get unblocked if they happen to be the one to show up. El Beeblerino ^{if you're not into the whole brevity thing} 18:41, 17 December 2024 (UTC)

I expect at least part of that is owing to the format of the current system and have some hope that a shake-up like this would make a difference. My understanding from talking to many admins who avoid unblocks is that they find the experience stressful and depressing and thus avoid it.

The additional clerking does worry me though - not to mention that a lot of unblock requests are pretty badly messed up, to the point that I'm sure a lot of them go unanswered since they never end up in RFU in the first place. I'm not sure what the solution to these two related problems is. A really simple batsignal template (I'm thinking of the "help" ones) to draw a clerk to the user talk page in question would probably work best? -- asilvering (talk) 18:42, 17 December 2024 (UTC)

This is an excellent point. There are certainly a number of unblock requests that just get "Your response here" with something written above or below or you get five unblock requests at a time or someone replying to every question with a new unblock request or... Significa liberdade (she/her) (talk) 21:48, 17 December 2024 (UTC)

Timespan

One concern about the proposal seems to be related to the length of time the unblock request may sit at ANU. Based on Beeblerox's comment above, I'm guessing they think the weeklong discussion might be too long. My other concern is handling discussion with the blocked user. Sometimes, blocked editors are quick to review policies and guidelines, and answer relevant questions. Sometimes they aren't. What would happen, for instance, in the following scenario: EditorA is blocked for regularly using unreliable sources, then files an unblock request. We ask them to go read about reliable sources then share something to let us know they understand how to use reliable sources. However, five days go by, and we don't hear from EditorA. Does that mean they have to wait six months or so to apply again? Significa liberdade (she/her) (talk) 21:45, 17 December 2024 (UTC)

A week is the minimum, not the maximum. I've gone back and forth on how to word this, but the idea is that a request should never be declined just because someone hasn't responded. It can be declined because someone is responding but hasn't addressed the salient issue (like I say "Why did you call that editor a fuckface?" and they keep saying "I did not edit-war" in response), and I'm open to the idea of also declining if someone is active elsewhere on their talkpage or on other WMF wikis but not responding... but not sold on that either.

Anyways, as to weeklong length in general, keep in mind, this would be a tradeöff for massively limiting the number of appeals people make. One week of pendency in exchange for three or six months with no more appeals. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 02:19, 18 December 2024 (UTC)

To avoid the board getting a huge and unmanageable backlog, I'd basically suggest posts get archived into two locations. You'd have the normal archive where the unblock request goes if it is fully resolved, and a secondary page for stale requests. Just because someone doesn't participate in a timely manner shouldn't result in a decline, but there should be some sort of "held" state for requests that do require a response to follow-up questions. Ideally, when the user returns, the post can be revived as if no time had passed, but how that could be accomplished is another challenge. –MJL ‐Talk‐^☖ 07:06, 18 December 2024 (UTC)

So what I was thinking, and probably should put in the draft, is that the archive would work like ArbCom's case request archive, just links to oldids. This seems wise because often unblock requests may contain sensitive information, and by switching to the AN/U model we'd be taking away people's ability to just blank that off their talkpages. So it'd be something like

• MJL siteblock appeal 2024-12-19 (declined 2024-12-28)
• MJL siteblock appeal 2025-03-31 (dormant 2025-04-21)

And then if you later return that gets removed from the archive, and at its conclusion becomes

• MJL siteblock appeal 2025-03-31 (accepted 2025-05-01)

-- Tamzin^{[cetacean needed]} (they|xe|🤷) 07:15, 18 December 2024 (UTC)

Pros/Cons of being off the talk page

A pro/con of being off the user's talk page: other editors may feel more/less comfortable stating something along the lines of "I don't know why this person was blocked" or stating something they wouldn't say directly to the blocked user. When discussing on the user's talk page, it often feels to me like we're sitting in their living room or something, which can make it awkward to discuss with other editors about the block--for better or worse. Significa liberdade (she/her) (talk) 21:52, 17 December 2024 (UTC)

An earlier version of this draft had everyone but the blocked user commenting at AN/U, with the blocked user's comments transcluded. That opens up several technical issues, though: Easier for the blocked user to fuck up the transclusion; easier for them not to realize they've been asked something; transclusion lag. So I think keeping discussions on talkpages, but deeming them for all intents and purposes part of the noticeboard, gets the job done better. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 02:21, 18 December 2024 (UTC)

But the idea is still that that portion of the talk page is transcluded to AN/U, right? Or have I misunderstood? -- asilvering (talk) 04:54, 18 December 2024 (UTC)

In both the old idea and the new one, the section is transcluded. The difference is that in the old idea, all comments other than the blockee's would be made at AN/U. See Special:Permalink/1191547912#Initiation. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 05:11, 18 December 2024 (UTC)

"Unblock request" or "appeal"?

Halfway through drafting this, I started getting picky about the word "appeal". In common parlance, that term is usually used for a request that implies some error in the underlying process. Seeking a presidential pardon, for instance, is not an appeal. Most unblock requests—or at least most successful ones—are not assertions of actual innocence, so you could make the case that the word "appeal" is misleading and skews people toward pleading innocence when they'd be better off admitting their mistake and requesting mercy. On the other hand, in practice we seem to often encourage the opposite, i.e. false confessions, as Nosebagbear Z''L would often point out. So I've been inconsistent about "appeal" versus "unblock request". Does anyone have thoughts about one being better than the other? Or should "appeal" be used but only in cases of UB2? Or just use them interchangeably? -- Tamzin^{[cetacean needed]} (they|xe|🤷) 04:21, 18 December 2024 (UTC)

We do call the category "requests for unblock", so there's that. In general, I think it's going to be impossible to stop people from using the word "appeal", much in the same way that AfC constantly has people using the word "reject" for the action that is actually called "decline". It just feels like an obvious word to use. Myself, I think I prefer "request" but if you audited my edit history I wouldn't be surprised if it turns out I say "appeal" more often. -- asilvering (talk) 04:46, 18 December 2024 (UTC)

In the spate of conversations I've been involved in in the last few days, I also find myself using them interchangeably and second-guessing which I should be using. For me, this could be partially because I was on the committee and they unfailingly call what they deal with "appeals".

WP calls a lot of things something other than what they really are, arbitration and oversight being the two glaring examples. And we use "admin" when you're requesting it and "desysop" when it's removed. I think both terms are broadly understood but consistency is desirable as well. El Beeblerino ^{if you're not into the whole brevity thing} 03:07, 19 December 2024 (UTC)

I'm in favor of changing our policy to allow checkusers to check those blocked for socking. It's something checkusers do on other language wikis, I think. Valereee (talk) 17:12, 19 December 2024 (UTC)

I always ask for CU checks on those, and they're always granted, so if it's against policy we have a lot of CUs who are going to be in trouble. -- asilvering (talk) 17:19, 19 December 2024 (UTC)

I thought it was against enwiki policy to CU to support a claim to innocence? Maybe I'm misremembering something? Valereee (talk) 17:44, 19 December 2024 (UTC)

CUs can't/won't run self-requested checks, but there's nothing stopping them from supporting a claim to innocence. The greater limitation is the inability to prove a negative, but I've definitely seen "there is no technical evidence of sockpuppetry" or even "based on technical evidence I find sockpuppetry very unlikely". -- Tamzin^{[cetacean needed]} (they|xe|🤷) 17:46, 19 December 2024 (UTC)

I'm sorry, clearly I'm stupid...what's the difference between "can't/won't run self-requested checks, but there's nothing stopping them from supporting a claim to innocence."? Valereee (talk) 17:47, 19 December 2024 (UTC)

The restriction is specifically on "Please check me to prove I'm innocent", and even there it's more an observation that CUs generally won't do it than a hard rule, AFAIK. (I've been checked because I had an issue with my password and feared compromise.) But there's nothing stopping a CU from checking someone who has an unblock request pending, to see if there is ongoing abuse, so long as there is valid reason to think that there might be abuse. At least that is my understanding. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 17:52, 19 December 2024 (UTC)

Yes, it seems to be standard practice to check someone who was blocked for sockpuppetry or has a history of block evasion. @Tamzin, in the "Speedy declines" section below, you say As written, admins can speedily decline if there is admitted sockpuppetry from an account that is ineligible to appeal on some other account. Allowing anything beyond that, though, would defeat the point of having an appeal system - does this mean you disapprove of this kind of check? or that you disapprove of this kind of check, resulting in confirmed block evasion, being grounds to deny an unblock? -- asilvering (talk) 18:00, 19 December 2024 (UTC)

So, if someone denies they are a sock, and a CU says they in fact are, that will usually be enough basis to deny an actual-innocence appeal. (The only exception, I think, would be if they've acknowledged a shared IP and the question is whether there was intent to sock.) This gets back to my question in § Speedy declines about whether actual-innocence appeals should be exempt from the one-week limit if the evidence is clear. Still haven't decided my feelings on that. Maybe specifically making it that a CU decline on CU evidence is exempt? Since as a non-CU, I don't have the best perspective on whether "CU-confirmed sock" means "This is the same IP and user-agent as the sockmaster has been using since 2001" or "Eh they're both on Jio and they edited about the same thing using Chrome on Windows and I'm in a rush so let's call it confirmed". -- Tamzin^{[cetacean needed]} (they|xe|🤷) 18:06, 19 December 2024 (UTC)

I've spent a lot of time pestering CUs about this (sorry, CUs) and have come to understand that "confirmed" is really quite strong, meaning that it would be very unlikely for an actually innocent user to get caught up in a confirmed by mistake. -- asilvering (talk) 18:34, 19 December 2024 (UTC)

I can tell you, with a lot of confidence from two years as an SPI clerk, that that is not universally true of all CUs. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 19:20, 19 December 2024 (UTC)

Alas. Can you tell me which ones I need to adjust my CU-response translator guide downward for? -- asilvering (talk) 20:01, 19 December 2024 (UTC)

@Asilvering: Though implausible, I will point out that an actual case of WP:LITTLEBROTHER would result in a confirmed CU result. That still doesn't mean the older sibling is actually guilty of anything besides sharing a PC with a bad actor. –MJL ‐Talk‐^☖ 20:14, 19 December 2024 (UTC)

I mean, I'm not going to name names, but I would generally say that a) people who got CU through arbship are usually (with several major exceptions) trusted less by SPI clerks to give 100% accurate readings of CU data, and b) CUs who appear to not at all consider behavioral evidence should not be trusted to have, well, considered behavioral evidence. For instance, if CU suggests that a user who edits about U.S. history is actually a CIR sockmaster usually fixated on Philippine radio stations, that should raise serious suspicion. Good CUs will take the behavioral discrepancy into account. Some won't. Even more to the point, I will never fully trust a CU who I've seen call two account definitively unrelated based solely on CU evidence. There are known ways to evade CU completely undetected, and if a CU isn't considering that on the innocence side, I don't trust them on the guilty side, because then it seems like they think CU is WP:PIXIEDUST. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 20:20, 19 December 2024 (UTC)

I always read the CU results as completely distinct from behavior, as in behavior shouldn't be taken into account for the unlikely/likely/confirmed results, which are purely technical. Possibly not least because a CU result is often followed by "needs to be judged by behavior" or similar. CMD (talk) 02:42, 23 December 2024 (UTC)

This was my understanding as well. -- asilvering (talk) 04:13, 23 December 2024 (UTC)

Speedy declines

Should admins able to speedy decline unblock requests based on apparently block evasion (i.e. sock puppetery) in some recent timestamp (e.g. 90 days)? GZWDer (talk) 11:28, 18 December 2024 (UTC)

As written, admins can speedily decline if there is admitted sockpuppetry from an account that is ineligible to appeal on some other account. Allowing anything beyond that, though, would defeat the point of having an appeal system, because if I can decline someone's appeal as "You're ineligible to appeal because I say you're guilty," without actually doing a review of the evidence myself, then that's not really an appeal.

One thing I should maybe do, though, is lift the one-week limit if it's an actual-innocence appeal where the evidence is very straightforward, e.g. the latest in a long line of sockpuppets who all make the exact same edit to the exact same article that no other person would have cause to make. But there's a potential slippery slope there too, and letting things pend for a week doesn't have any inherent harm so 🤷. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 23:29, 18 December 2024 (UTC)

The issue is also that "apparently block evasion" also has different standards for different people, ranging from "knows too much for a newcomer" to actual CU evidence. That's way too wide of a range to justify speedy declines, and, as Tamzin already said, defeats the point of an appeal board. Chaotic Enby (talk · contribs) 23:05, 19 December 2024 (UTC)

Community bans

I will propose the user can also appeal community bans using this unblock process, with process be either of (1) the appeal will be transfered to WP:AN unless meeting one of speedily decline criteria; or (2) more prefered, also designate this page as one of venues (besides AN and ArbCom) to lift (but not impose) community bans, with #Unblocking having "if the user is banned by community, admin should only unblock the user if there is rough consensus to unblock in AN/U. In unclear cases, the request should be transfered to AN for further discussion". GZWDer (talk) 11:33, 18 December 2024 (UTC)

For this to have any chance of getting community consensus, it needs to not significantly change the current constitutional structure of bans and blocks. So I'm very much opposed to giving ANU any power to unban. As to letting a user request AN review over AN/U review, that's something I've been mulling over too. I could see a case for allowing it, or disallowing it, or something in between. What I'm most drawn to is allowing it after a first decline, without the three-month decline, but with a clear warning to the user first that they will be considered banned if the appeal is unsuccessful, and with some common-sense discretion for admins if the appeal is in obvious bad faith. What do others think? -- Tamzin^{[cetacean needed]} (they|xe|🤷) 23:25, 18 December 2024 (UTC)

I suspect so few editors would ask for AN review over AN/U review that each incident is probably best dealt with on a case-by-case basis. Sounds like the sort of thing that's probably better to come up with policy for once the main idea is actually working in practice? -- asilvering (talk) 18:29, 19 December 2024 (UTC)

Maybe it would be smart to include a list of things that are intentionally not being resolved by this policy, but rather would be decided once/if it's implemented and some time has passed to get a sense of things. Off the top of my head:

• Procedure for requesting community unblock/unban
• Whether p-blocked users can still request unblock directly at AN

-- Tamzin^{[cetacean needed]} (they|xe|🤷) 04:39, 20 December 2024 (UTC)

This is much less lightweight than the current version

Ok, I keep not having the time/space to write out the thoughts I have bouncing around in my head, so I'm giving up on "thoughtful" and will just post a few thoughts/ideas here in the hopes that someone can run with them:

• the lightweight nature of unblocks right now is a big strength in a lot of ways, and this version is much less lightweight
  • it's obvious to me that at least some of the problems with unblocks right now are also because they are lightweight
  • if we can balance these pros/cons, we'll have a much improved system (thanks so much to Tamzin for all this thought and effort so far)
• things to think about while designing said much improved but less lightweight system:
  • loads of editors put in an RFU immediately after being blocked, with predictably poor outcomes
  • AN/U has the possibility to become "ANI, part 2" in a way that will increase rather than decrease frustration/drama
  • three months will feel like a long time to a lot of first-time-blocked editors
  • "quick and easy on the talk page" becoming "weighty and serious in a more public location" might be embarrassing or otherwise uncomfortable for blocked editors
  • I definitely had more thoughts, but have forgotten them

-- asilvering (talk) 17:33, 19 December 2024 (UTC)

For the avoidance of doubt, these are just things I think we ought to keep in mind, not necessarily "things that demand a change to this proposal". Though I do think there is a real problem waiting to happen with the "editor blocked in the middle of acrimonious interpersonal event, immediately appeals" scenario. In that mindset, I don't think blocked editors are likely to do much more than dig themselves a deeper hole. And I don't really want someone to be able to dig themselves a three-month-deep hole if walking away for a week or a couple of days could have saved them from themselves. -- asilvering (talk) 17:37, 19 December 2024 (UTC)

Just a thought on length, I'm not opposed to making the first window like a month, and/or letting the declining admin set a shorter ineligibility window. The big thing to remember, though, is that no one would be having their appeals declined before they've made the best case they're able to, except people who just refuse to engage in dialogue. So there wouldn't be any of the "Hi I changed username" / "Decline. Doesn't address the promotionality" / "Hi I know what spamming is and won't do it" / "Decline. Is this old account also you?" / "Yes that was me, I'll stick to one account now" / "Decline for sockpuppetry" / etc. etc. etc. cycle. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 18:09, 19 December 2024 (UTC)

I think having the first window be a month is probably a good improvement regardless. With the no one would be having their appeals declined before they've made the best case they're able to, what I'm trying to get at here is that I don't think we're setting someone up to be able to make that best case. If you start out your unblock appeal in a rage haze, I don't think you're really setting yourself up for success, and if we then try to engage with them for a week I think it's really quite possible that we're just helping them dig their own grave. What I've been doing in this kind of case in the current system is suggesting that they withdraw their open appeal and explaining why I don't think it's likely to work (I think I may have given out at least one "please for the love of G-d stop digging this hole" decline, too). I don't think that really works all that well either, but at least they're able to make a new appeal without restrictions.

What do you think would happen if we had some kind of RFU+AN/U setup, where if you want to request an unblock in the first, say, 48 hours, it goes to RFU, but if it's past that (or you're still in RFU at that point) you get directed to AN/U? I realize this is kind of "AfC, but for unblocks", but the queue can't build up too badly if it gets bounced to AN/U in the end. The tendentious unblocks (MAB, etc) get taken out before anyone at AN/U has to deal with them (hopefully), the "holy shit please stop digging" ones have a chance to be seen before they hit the higher-stakes venue, the obvious mistakes hopefully get speedy-accepted before 48 hours goes by, etc. We could even count those 48 hours towards the first week at AN/U if we thought that would be a good idea. -- asilvering (talk) 18:26, 19 December 2024 (UTC)

I estimate we get between 100 and 200 unblock requests a day, between CAT:UNBLOCK and UTRS. That's only accurate within an order of magnitude. Now, many of those could just be speedily handled, but I'm concerned that this new process would buckle under the load. My concerns are specifically around Asilvering's "this version is much less lightweight" point, above. Any ideas how to ensure the new process wouldn't buckle under the load (or arguments that it'd be fine, shut up)? --Yamla (talk) 23:42, 19 December 2024 (UTC)

Do you have any idea how many of those requests are re-requests, vs new ones? In principle, the re-requests would be happening much less often, since they'd have to take the 1/3/6/whatever-month time out in between requests. -- asilvering (talk) 00:16, 20 December 2024 (UTC)

Tough call. I mean, a lot of them are immediately-obvious declines. For example, threats to murder us account for ten or more per day by themselves. And there are a lot of re-requests. I have to imagine new requests that aren't obvious declines are less than 50 per day. But that's still a lot... --Yamla (talk) 00:49, 20 December 2024 (UTC)

Other venues like AfD can handle around the same quantity of nominations. I'd support a trial period to see if there is enough editors to handle to this process. Ca ^{talk to me!} 12:40, 20 December 2024 (UTC)

Closing AfDs, however, is a job lots of admins like... -- asilvering (talk) 13:56, 20 December 2024 (UTC)

I thought the concern was that there would not be enough commentors, not closers. In any case, consensus-gauging is a very different job than appeal reviewing. Ca ^{talk to me!} 16:37, 20 December 2024 (UTC)

Checkuser blocks

Under policy, checkuser blocks cannot be overturned without consulting a checkuser. But since they can be discussed publicly I understand why they would fall under this proposal. However, it's not at all clear to me how that works in reality under this proposal. I will say in general I found this very hard to parse and understand but I got there, I think, in the end. Best, Barkeep49 (talk) 18:20, 19 December 2024 (UTC)

I think it'd be pretty similar to the status quo? Use of {{checkuser needed}} would be replaced with |awaiting=checkuser, but as with the current unblock system, whether and when you request CU would depend on what the person's claim is (actual innocence vs. yes-I-socked-but vs. shared-IP etc.), or whether the admin has their own suspicions they want investigated or want to trust but verify for a ROPE unblock. But if you want to give a specific example, I'm happy to walk through how I think it'd play out in this system. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 05:23, 20 December 2024 (UTC)

AI requests

This is going to be contentious. I think we should speedily decline any unblock request that is clearly written by an AI. I think Tamzin should modify "There is no coherent rationale given (not just a bad reason, but garbled text, no reason at all, a complete non sequitur, etc.)" to include this ("... a complete non sequitur, written by AI, etc.)"). I think the primary point of contention will be that there's no 100% certain way of determining if an unblock request was written by AI. To that, I'll argue that it's frequently blatantly obvious, and AI chatbots are an existential threat to Wikipedia. Please understand my proposal is serious, but is said with a smile on my face and an understanding that this suggestion may be unworkable or unacceptable, and that reasonable people may disagree strongly. --Yamla (talk) 18:36, 19 December 2024 (UTC)

Given that you're then SOL for the next three months under the current proposal, I have to disagree, though I have been declining AI-written unblocks myself without a second thought under the present system. If they keep using AI after being told not to, well. I for one won't be unblocking; I don't really have any interest other than morbid curiosity in hearing about what chatGPT thinks our policies are. -- asilvering (talk) 18:40, 19 December 2024 (UTC)

I'm thinking a speedy decline for AI requests. They'd be immediately eligible to make a new request that could be taken up for discussion, so long as it wasn't written by AI. But, good point. I wasn't clear on that. --Yamla (talk) 18:45, 19 December 2024 (UTC)

I'm open to a separate speedy decline criterion that says something like "The request is obviously written by a large language model and does not address the specifics of the block reason. (But be aware that some users may use LLMs to touch up requests they wrote themselves.)" I feel like those caveats are important to avoid misapplication... In practice I think they would come up very rarely, since usually LLM requests are utter slop. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 19:43, 19 December 2024 (UTC)

@Yamla: I'd been meaning to write Wikipedia:Identifying LLM unblock requests anyways, so I've done that now; such a criterion could link to it. Please feel free to improve what I've put there. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 21:01, 19 December 2024 (UTC)

Another one I see a lot is the "praising wikipedia" variety. Can't ad lib one off the top of my head but it's one to keep an eye out for if you're checking the queue. Also, it's very keen on telling us that it's committed to verifiability and neutrality, whether that's a nonsequitor or not. -- asilvering (talk) 21:06, 19 December 2024 (UTC)

Also pings to @Chaotic Enby and @JPxG on this one. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 21:10, 19 December 2024 (UTC)

Yep, definitely agree with this as a speedy decline criterion. As Asilvering just said, they're often very formulaic and mention being "committed" to generic Wikipedia principles, without actually going in-depth into anything. Chaotic Enby (talk · contribs) 23:00, 19 December 2024 (UTC)

 Added -- Tamzin^{[cetacean needed]} (they|xe|🤷) 05:24, 20 December 2024 (UTC)

🦈 jp×g🗯️ 21:37, 19 December 2024 (UTC)

Ah, yes. In that case, endorse. -- asilvering (talk) 21:07, 19 December 2024 (UTC)

A blanket "NO AI" that excludes LLMs also excludes the current implementation of much of machine translation (see [1]). As outlawing translated requests from a non-English-first user probably wasn't by intent, I'm reopening this question. ☆ Bri (talk) 19:07, 24 December 2024 (UTC)

A good way to word it would be to make a translation between AI-generated content and AI-translated, human-written content. The latter is much more natural in writing and hardly distinguishable from untranslated human writing, especially compared to a ChatGPT output. Using AI to "retouch" a human-written appeal is definitely a grey area between the two, and the line should be drawn at AI adding actual content of its own, although this would be difficult to evaluate in practice and editors should still err on the side of caution. Chaotic Enby (talk · contribs) 19:12, 24 December 2024 (UTC)

@Chaotic Enby: Did you mean to say "make a distinction" in the first sentence of your response? ☆ Bri (talk) 19:17, 24 December 2024 (UTC)

Yes, my bad! I was already thinking about the rest of the sentence and mixed my words up. Chaotic Enby (talk · contribs) 19:20, 24 December 2024 (UTC)

This is obviously not about that kind of AI use, and Tamzin's earlier comment makes that clear. -- asilvering (talk) 19:43, 24 December 2024 (UTC)

Very true, although the question is more about how to word it on the page so that this intent is just as clear. Chaotic Enby (talk · contribs) 19:50, 24 December 2024 (UTC)

IPBE, partial blocks, speedy accepts?

Partly resolved

 – Softblock speedy accept added

Unblock requests are sometimes used to request WP:IPBE. Those clearly shouldn't go through this process. Similarly, if a user is unable to edit because of a proxy block or an autoblock, I think they should just be speedily accepted or rejected in-situ rather than going through this process. Partial blocks can generally be contested straight to the blocking admin; perhaps we should speedily decline such requests until they've first discussed it with the blocking admin. It would seem pointless to go through the week-long process otherwise. I also wonder if there's any sort of "speedy accept" criteria. Certainly, a soft block due to the username should be speedily accepted if the username has changed (or if the user commits to putting in a change request). That's the intention of the soft block, after all. --Yamla (talk) 18:43, 19 December 2024 (UTC)

Speedy accepts for softblocks should definitely be in here. Thoughts on the rest later. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 19:48, 19 December 2024 (UTC)

Two problems

1. The term "de facto ban" is both vague and unnecessary. It's not at all clear to me what types of blocks "no reasonable administrator" would lift unilaterally that could actually have a chance at being lifted with community consensus; Wikipedia:Administrators' noticeboard/Archive358 § Koavf unblocked is a case where a veteran admin made an unblock that would have been SNOW-declined had it been presented at AN first. Since the term is used to allow preventing requests that could never succeed due to preempting circumstances, it would make much more sense to add "the block should only be removed with consensus" as a speedy decline reason. I know you want to limit the speedy decline reasons to procedural grounds only, but this would be a relatively unproblematic addition since it doesn't limit the filer's ability to have their case heard, only where their case can be heard.
2. Why are there three bullet points under UB2? There is no to assess relative (mis)behavior beyond the level of if it justifies a block.

Mach61 20:16, 19 December 2024 (UTC)

1. "de facto ban" and the "no reasonable administrator" definition come from WP:BLOCKBANDIFF, which is established policy. But, thinking about the rest!
2. The bullet points are there because those are three different ways that a block could be found unjustified, each with their own considerations. In particular I think it's important to separate exoneration, because if someone is truly blocked for completely incorrect reasons (examples: [2] [3] [4] [5] [6] [7]), this should be made very clear. My strong feelings about this may or may not come from my own block log, and there's a good chance I wouldn't be an admin if Coren hadn't made that distinction when unblocking me.

-- Tamzin^{[cetacean needed]} (they|xe|🤷) 20:28, 19 December 2024 (UTC)

@Tamzin Re (2): I was wrong to say that there is no reason to note how a block is unjustified, but that doesn't mean that classifications need to be codified in the notice board rules; admins can make such notes on an ad hoc basis right now if they think it's appropriate. Mach61 21:36, 19 December 2024 (UTC)

I can see a case for keeping it prose to avoid the impression that it's a strict ternary of options. How does Special:Diff/1264055211 look to you? -- Tamzin^{[cetacean needed]} (they|xe|🤷) 04:56, 20 December 2024 (UTC)

An acceptable compromise Mach61 05:02, 20 December 2024 (UTC)

And yes you're completely right on de facto ban handling. See [8]. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 05:09, 20 December 2024 (UTC)

Net benefit?

What's not clear to me is that this provides a net benefit. As we've recently seen at AN there are some number of editors who would get unblocked with wider discussion who are currently getting declined. This seems reasonably crafted to help those editors. It's not clear to me how many editors who are currently unblocked write a bad first appeal improve on the second (or subsequent appeal) and then have it accepted in a way that this new method would prohibit by imposing a waiting period. I'm willing to do a little legwork on this but didn't see an easy way to even find successful unblock requests. And even once that's established we need to then compare this against the added costs that this new process filled system would entail, especially as I expect many/most unblock requests to continue to be declined. Best, Barkeep49 (talk) 18:42, 20 December 2024 (UTC)

@Barkeep49, do you think the time-tiered system I describe here would help at all with any of those concerns? I'm wondering about those same editors.

(One thing I do quite like about this proposal is that it would be much easier to see past unblock requests. Is there a log of what pages are added/removed to categories anywhere? If so, one could check that against Cat:RFU.) -- asilvering (talk) 18:51, 20 December 2024 (UTC)

Maybe? I had obviously read that section as I linked to it, but it would all depend on how its implemented (and I don't see any implementation yet). But even beyond that I think right now this proposal is asking people to go on vibes that the extra cost is worth it and I would hope we could do better than that. Best, Barkeep49 (talk) 18:57, 20 December 2024 (UTC)

System pages shouldn't name a particular editor

While I really miss Nosebagbear a lot, if this moves to the proposal phase, I think naming him should be removed. I can't think of another process/guideline/policy that name checks a particular editor. Best, Barkeep49 (talk) 18:44, 20 December 2024 (UTC)

That bit would probably be moved to an /Administrator instructions subpage or something, which are more essay-class. -- Tamzin^{[cetacean needed]} (they|xe|🤷) 18:49, 20 December 2024 (UTC)

Clarifying "No coherent rationale"

One of the items in Speedy declines currently reads:

There is no coherent rationale given (not just a bad reason, but garbled text, no reason at all, a complete non sequitur, etc.)

I'm afraid that the limit between "a bad reason" and "a complete non-sequitur" might be blurry, and that an overzealous administrator might speedily decline a reasoning they find "incoherent". To avoid this, I would propose rewording it as:

There is no rationale given (not just a bad reason, but garbled text or no reason at all)

Making it closer in essence to Wikipedia:Patent nonsense than to "any sufficiently bad reason". Chaotic Enby (talk · contribs) 19:19, 24 December 2024 (UTC)

You vill answer the following questions or vill not be unblocked

Currently a very large percentage portion of unblock requests have and admin asking "what type of edits will you make if unblocked" or "describe exactly what edits you would make if unblocked" or even more stringent demands that boil down to "write me an essay about what you've done wrong."

The users who do this say they are doing it because they don't want the user to just get blocked again. I say that may be their intent but it quite often doesn't have the desired effect.

This effectively tells other admins that the inquisitor is now "holding" the block and it cannot be actioned until the questions have been answered. What doesn't seem to have occurred to them is that the user on the receiving end of this may be offended at being treated like a child and may just give up on Wikipedia. Even when that is not the case, they may not be checking back every day, and another admin may come along and find that the unblock requests they have already posted is sufficient to make a decision to either accept or decline, but seeing that the inquisitor is interrogating them, they are apparently supposed to feel that is some sacrosanct procedure that may not be interrupted. This is not helping the backlog, it is making it worse and it needs to go. Unblock requests are supposed to be pass/fail system, not a trial. El Beeblerino ^{if you're not into the whole brevity thing} 21:18, 29 December 2024 (UTC)

I think the natural result of that would be all such unblock requests would be declined. Do you think that's correct? If so (and you may well disagree), do you think that's better than the current de-facto system? --Yamla (talk) 22:13, 29 December 2024 (UTC)

Because you can't read my tone, I can well imagine this being better in the medium term (and possibly even resulting in more unblocks). So I'm trying to figure out exactly your position before thinking through all of the consequences. I'm not trying to passively-aggressively indicate disapproval or anything. --Yamla (talk) 22:14, 29 December 2024 (UTC)

I've seen it used often when the request says that the user regrets their actions and have read the relevant policies cited in the block notice and/or in previous warnings. To me, for a first offense block after only a few edits, that's enough to AGF and give them another chance. If they are not being truthful it usually comes out right away and they get reblocked. I'd much rather go that route than just sit there for a week waiting to see if they reply to the quiz. El Beeblerino ^{if you're not into the whole brevity thing} 22:43, 29 December 2024 (UTC)

I am with Yamla. If the rules say "the only thing you can consider is what's in the appeal box" the result will be to end up with more declines. It seems that your higher priority is that you want more admins willing to accept appeals than what admins are currently willing to do. The solution to that is different than limiting the current iterative discussion process. Best, Barkeep49 (talk) 19:14, 3 January 2025 (UTC)

Sometimes those discussions lead to results, we just had one at User talk:82.44.247.44. In this case the user had several dozen edits and had gotten some previous warning. I'd been watching the progress for several days, like six users were involved and the blocked user eventually agreed to a topic ban as an unblock condition. I'd call that worth the effort.

Taking the exact same approach with someone who made like two edits and was blocked with no warning whatsoever is not the same thing.

However, I am seeing some admins who use it as their first response to nearly every single unblock requests regardless of the content of the request. Personally, if I see a bad unblock request that offers no detail, I decline it while clearly explaining why and what they need to do better in their next request.This very clearly puts the ball back in their court, and has the added affect of removing it from the backlog. I'm not suggesting that asking a question or two is always the wrong approach, but I am suggesting it is not always right approach either. Beeblebrox ^Beebletalks 19:59, 3 January 2025 (UTC)

All of this seems neither here nor there for what Tamzin is brainstorming here though? Best, Barkeep49 (talk) 20:24, 3 January 2025 (UTC)

Maybe not the proposed process per se, but if the overall topic of making the unblock process in general suck less.

I've just gotten done dealing with SpeedyHaste's unblock appeal. This was originally a CU block but was downgraded, with the SU noting it was probably a MEAT situation. The blocked user readily confirmed this and said it was his friend thinking it would be funny to deliberately get him blocked, which sounds exactly like something a teenage boy would do. For some reason three admins spent nineteen days demanding that they somehow provide concrete proof that they did not operate the other two accounts, with the bewildered blocked user asking what such proof they could possibly provide, and not being given any details. One would assume the reason the admins in question did not provide any such details is because they don't know what it would be either, so what was the point? Beeblebrox ^Beebletalks 20:30, 3 January 2025 (UTC)

@Beeblebrox, I agree more or less with everything you said in your unblock rationale, but I still don't think this is a good unblock. Can I believe that was my friend trying to make me mad and get me banned for no reason? Absolutely. But TechnoSquirrel69's SPI request that led to these blocks shows that these accounts have been involved in disruption that goes back for a few months. None of that appears to have been addressed in the course of this block appeal. You'll note that the unblocked editor has now responded to you, asking for further unblocks and saying they and their friends "didn't do anything wrong". They haven't learned. -- asilvering (talk) 00:02, 4 January 2025 (UTC)

@Beeblebrox: In your unblock statement, you wrote assuming good faith that, what[e]ver happened, it will not happen again. Did you actually examine what had happened here? Also, FWIW, I never asked for proof, I merely pointed out how deficient said proof was. I was also never inclined to unblock someone who'd been collaborating with confirmed socks of a sockmaster who is now 3X-banned. I don't see this unblock as being a net positive and I hope you'll keep an eye out for the inevitable disruption this editor will cause. voorts (talk/contributions) 00:05, 4 January 2025 (UTC)

I've never agreed with the perspective that if you take one administrative action regarding a specific user, you are then responsible for them and must monitor their behavior. If they start making disruptive edits or whatever any admin can block them, I gave them a second chance, not an absolution and a get-of-jail-free card. Beeblebrox ^Beebletalks 01:44, 4 January 2025 (UTC)

I suppose it's very easy to disagree with that perspective, when you're not the one cleaning up after the disruption. -- asilvering (talk) 03:15, 4 January 2025 (UTC)

It's an absolutely ridiculous standard to hold anyone to. I've been involved in thousands of blocks as well as granting user rights to those who request them at perm. I've had people make this absurd argument before, it has no basis in policy, no precedent, and is utterly unsustainable if admins are to actually get anything done. If we actually had to operate that way it would be impossible to administrate this website. It's only brought up when people have an axe to grind, as your needlessly personalized and accusatory remark implies you may. Beeblebrox ^Beebletalks 03:30, 4 January 2025 (UTC)