Community Wishlist Survey 2021/Anti-harassment

Add links
From Meta, a Wikimedia project coordination wiki
Community Wishlist Survey 2021
Anti-harassment
3 proposals, 186 contributors, 152 support votes
The survey has closed. Thanks for your participation :)



Partial block of IPs across categories

  • Problem: Editors evading a block/ban often go to the dynamic IP route to continue editing. Sometimes it is not possible to rangeblock that IP across the whole of WP due to collateral damage of blocking good editors.
  • Who would benefit: Anyone who has suffered harassment/hounding/stalking from an IP that continues to pop-up day after day.
  • Proposed solution: Partial blocks should be allowed across a group of categories for a range of IP addresses. For example, blocking range 123.245/ across Category:Black English sportspeople. I've raised this before at WP:VPT, with a reply stating that an article could be removed from the category, and the block wouldn't work. But once the block is in place, then the IP could not edit the page(s) in question to circumvent the partial block. Thanks.
  • More comments:
  • Phabricator tickets:
  • Proposer: Lugnuts (talk) 08:24, 17 November 2020 (UTC)[reply]

Discussion

  • It's also possible that such a feature would be used to provide technical implementations of topic bans, though I think on the English Wikipedia there might be disagreement to such an idea. In the IP case, how will this work under the new system of IP addresses being replaced with identifiers (see en:Wikipedia:Wikipedia Signpost/2020-11-01/Op-Ed)? I'd be worried about a tool that relies on the current system of rangeblocks if this may change soon. — Bilorv (talk) 17:42, 17 November 2020 (UTC)[reply]
  • One major concern raised when this was mooted on en-wiki was that articles could be added to a category incorrectly (including wilfully incorrectly) to cause problems for the user. Nosebagbear (talk) 14:43, 18 November 2020 (UTC)[reply]
    Indeed, category blocking was discussed when partials were first implemented and the concern for abuse was why it was not implemented (first). --Izno (talk) 01:04, 21 November 2020 (UTC)[reply]
Anyone gaming the system to add categories to pages to stop others can be dealt with if and when it happens, with continued distruption being stopped with blocks. I think the p/block of IP ranges across categories to stop long-term harassment far outweighs other concerns, and is something that this survey should take very seriously in relation to harassment. Lugnuts (talk) 08:44, 23 November 2020 (UTC)[reply]
Unless you know how to make it clear what is gaming and what is dubious adding (there is a lot of dubious adding to categories in the usual course of things), how to make it rapidly handled, who will do that vetting work and so on, then I'm rather concerned by your fairly sangfroid "can be dealt with if and when it happens" Nosebagbear (talk) 15:54, 2 December 2020 (UTC)[reply]
  • I could see this being used for abuse, as it effectively gives non-admins the ability to block partially blocked users from a page by adding a category. Although such attempts would probably be seen relatively quickly, it does still give the ability. On smaller wikis additions of the categories may go unnoticed, so a user could potentially block IPs from pages they don't want them editing. Dreamy Jazz talk to me | enwiki 00:00, 20 November 2020 (UTC)[reply]
  • Hi Lugnuts. I work with both the Anti-Harassment Tools team (who built Partial Blocks) and the Community Tech team. We are going to allow this wish to move to voting. However, because of the circular nature of categories on wikis, if implemented, this would work on a single category only and not extend to any sub-categories or sub-sub-categories. There are holes/workarounds here in that someone could remove a category, commit their bad-faith edit, and then move on or even add the category back. Still, it could be implemented so we are allowing it to be voted on with the single category caveat. --AEzell (WMF) (talk) 21:55, 2 December 2020 (UTC)[reply]
Yes, that sounds like a fair compromise. Thank you. Lugnuts (talk) 08:45, 3 December 2020 (UTC)[reply]
Would it be technically possible to prohibit it for certain users or IP ranges to add or remove categories in general? -- Aspiriniks (talk) 18:39, 8 December 2020 (UTC)[reply]
AEzell (WMF), most wikis would disallow circular nesting of categories. Is there not some existing API call that can check for supercategories of a page without choking on the rare cycle? Instead of limiting the depth to 1, you could set it to say 5 and still avoid an infinite loop. Pelagic (talk) 02:00, 16 December 2020 (UTC)[reply]

Another idea would be to make page categories available as abuse filter variables. --Count Count (talk) 19:14, 8 December 2020 (UTC)[reply]

  • Make this "list of pages, generated from a snapshot-in-time category membership list" instead of "categories" and I'll support it. This would solve the "any user can add or remove a page from a category, thereby page-blocking/unblocking an IP address from that page, possibly unintentionally" issue. Davidwr/talk 15:44, 21 December 2020 (UTC)[reply]
  • One problem with this is that the search engine does not reliably search within a category using with the incat function. There are often error messages such that the category is too deep. Before something like this is implemented, the search engine should be fixed in order to allow people under the partial block to quickly see which pages they can and cannot edit.--76.4.9.147 01:15, 28 December 2020 (UTC)[reply]

Voting

Just to clarify, the range-block doesn't work, as it is too big a range, with a lot of collateral damage. Lugnuts (talk) 12:22, 9 December 2020 (UTC)[reply]

UserBlind mode

  • Problem: Certain discussions, particularly those dealing with individual user conduct issues, are difficult to participate in in an unbiased manner and without generating side-effects like occasional animosity between editors. The fact that users know that particular other users have passed negative judgement on them can make it difficult for those users to work together in the future. The concern over this can also cause people to avoid dealing with conduct issues entirely.
  • Who would benefit: All contributors, but more directly those dealing with reports of conduct issues (including Arbitration Committees and users active in fora like enwiki's ANI, AE, etc). Also contributors to wikis with "unblockable" users which the community has difficulty fairly judging.
  • Proposed solution: A [selectable] "UserBlind" mode, in which [for you only] all visible usernames would replaced by tokens, eg "[USER #23]", effectively anonymizing other users while the mode is enabled. This would allow fair assessments of conduct reports in a impersonal manner. Not only would the person passing judgement on the actions not know whose actions they are (thus evading bias), the person whose actions are being judged would not have reason to think that the commenting user has anything against them in particular, thus avoiding inter-user tension.
  • More comments: Links: Proof of concept, VPM post
  • Phabricator tickets:
  • Proposer: Yair rand (talk) 05:51, 30 November 2020 (UTC)[reply]

Discussion

  • @Yair rand: - how would individuals handle verifying evidence and doing further investigation with this process? Neither AN or ArbCom can function like judges where evidence is just provided - it needs to be cross-checked (in ANI, the closer doesn't do investigation, but everyone else participating would follow it down the rabbit hole). Additionally, when I'm reviewing people for certain roles, such as at RfA, then I need to be able to review their ANI participation to know the quality of their judgement. Nosebagbear (talk) 16:02, 2 December 2020 (UTC)[reply]

Jo-Jo Eumerus, MarioSuperstar77, Braveheidi, Firestar464, Tyrekecorrea, KlauRau, Lollipoplollipoplollipop, I-Bin-A-Bibi, The Hand That Feeds You, WikiFer, valepert, Gereon K., Lostinlodos, SarahSV, Klaas, Shushugah, HAL333, I am concerned that you might not understand what you've voted against, so here's my attempt at an explanation. If you had a significantly different idea, then you might want to remove your comment ('oppose votes' aren't counted anyway).

Current situation:

  • The policy says you are supposed to w:en:Wikipedia:Comment on content, not on the contributor.
  • Some editors find it difficult to follow the policy if they can see who said what. Instead, they find themselves evaluating comments based on the reputation of the editor who posted them. For example, if there's a snarky comment, but you see that "I" posted it, then you just know that it was kindly meant or at least not meant to be mean-spirited, but if That Other Editor posted the same words, then you know that it's terrible behavior.

Proposal:

  • Do not change anything whatsoever about your own editing or what you can see in the page history.
  • Do not change anything whatsoever about what's stored in the database. Every edit gets recorded with each editor's permanent username.
  • Offer, as an option, for people who happen to recognize that they're interpreting comments in certain ways, or voting in RFCs (and maybe even in this wishlist proposal) on the basis of "who" said something, rather than focusing on "what" the editor said, a button that would replace all the usernames on their screens only, and not affecting anyone else at all with a number.
    • Set this up so that it's easy to tell which comments on the same page came from the same editor (e.g., every comment from the first editor named on that page is from "User 1", every comment from the second editor named on that page is from "User 2", etc.).
    • Set this up so you can turn it on or off whenever you want.

I see in the oppose votes below that there are a lot of IMO misplaced concerns about transparency and conflicts of interest. What I don't see is any explanation of how transparency suffers by simply not requiring me to see your username at the end of your comments. Nobody's being prevented from seeing the usernames; the proposal here is merely to stop forcing me to see your lovely usernames if (and only if) I don't happen to want to see them. WhatamIdoing (talk) 19:19, 14 December 2020 (UTC)[reply]

WhatamIdoing, there's no benefit in cutting myself off from "this user who is proposing A is the same as the user who proposed the very damaging B last week, so I need to look closely at all this person's ideas." How is this proposal related to anti-harassment, by the way? SarahSV talk 23:08, 14 December 2020 (UTC)[reply]
SlimVirgin, is there any benefit in you requiring all editors to see everyone's usernames? Some people think there is a moral imperative to ignore an editor's reputation. They could choose to use this feature; you could choose not to. You said that letting people voluntarily choose not to display usernames on their own screens would reduce transparency. That's not the same as just saying that you wouldn't choose to use it yourself.
(I assume that this is in anti-harassment because not knowing editors' identities means that jerks won't know who's female. The "r u female?" jerks didn't cease to exist when AOL chat rooms died, and some communities have a fairly significant problem that way.) WhatamIdoing (talk) 02:42, 15 December 2020 (UTC)[reply]
WAID, there is a script that does this at de:Benutzer:Martin Kraft/hideUserComments.js. If it's voluntary, the jerks will see the names. I can't think of a way this would help anti-harassment. As for reducing transparency, will this remove names and histories? Because if you click on a user name, you'll see the other issues that person has commented on, and it will often become clear who they are. SarahSV talk 06:48, 16 December 2020 (UTC)[reply]
I'm sorry, I didn't realize there was a script for enwiki too. Why does it need help from the WMF if we can already do it? SarahSV talk 06:51, 16 December 2020 (UTC)[reply]
I suspect few people would opt in to this and leave it on for themselves wiki-wide, especially if they perceive that nobody else is doing so. An interesting variation would be to mark specific pages with a keyword that turns on BlindMode by default, with a button for the user to unblind at will. (Button could toggle off-on-off, save state in a session cookie so that it works for anons as well as logged-in users.) Pelagic (talk) 02:44, 16 December 2020 (UTC)[reply]
If a person clicks Edit, will the names still be aliased? Then they would need to be translated back on publish/save/commit. It would be hard to catch instances where people mention others by name in plain-text without wikilinking, doubly-so if the mention isn't their exact username, like this: Yair, WAID, SarahSV. Still, I guess you could start with the easiest case (sigs, read mode only) then progress to the harder ones (mentions, editing). Pelagic (talk) 02:44, 16 December 2020 (UTC)[reply]
If I reply to or @-mention "User24" and it turns out to be "User:WhoImInteractionBannedWith" is that going to go badly for me? How would I prove that I'm interacting blindly with them? What if I criticise a bunch of different User65, User31, etc. across multiple pages and it turns out to be the same person, then I get accused of harassment and stalking? Pelagic (talk) 02:44, 16 December 2020 (UTC)[reply]

Voting

Global semiprotection of user talk pages

  • Problem: There are certain crosswiki vandals who will go to a user's talk page and leave harassing comments. When they get blocked from that wiki, they will just go onto another wiki where an account exists for that user and leave more harassing comments. And on and on until a steward intervenes. Since these are done with throwaway accounts and proxies in many cases, this just repeats itself over and over.
  • Who would benefit: Those being harassed crosswiki
  • Proposed solution: Allow for a user's talk pages on every wiki to require autoconfirmed permission to be created/edited, using some sort of list hosted on Meta that only stewards can edit (similar to what we have for spam and titles).
  • More comments:
  • Phabricator tickets:
  • Proposer: Rschen7754 01:47, 17 November 2020 (UTC)[reply]

Discussion

  • Can't this just be done using a bot? (& if so, that'd be easier to do, too). It'd have to run under the account of someone authorised to semi-protect pages globally (a steward, I guess?). It'd be a diversion from the global bot policy and the steward policy, I guess, so maybe it needs an RfC to be permissible or whatever, dunno. But I think technically this may be more feasible than a software edit. ProcrastinatingReader (talk) 02:43, 17 November 2020 (UTC)[reply]
  • Even better, a user should be able to activate semi protection on their own pages themselves, and do it globally as well. No need to bother stewards or others with this - if I want to enable semi for my user talk pages on one or more wikis, I should be able to do it, easily. Just like we now have some enhanced options for who can send email to you. --Piotrus (talk) 04:12, 17 November 2020 (UTC)[reply]
  • I'm quite new on Wikipedia, I'm not sure whether crosswiki bots are common. I tend to agree with Piotrus' suggestion. --Himbeerbläuling (talk) 06:40, 17 November 2020 (UTC)[reply]
    • They are not, afaik there is no steward crosswiki bot that performs actions on local wikis (the proxy one takes actions on meta I think), but community consensus can probably change that for this limited usecase. I disagree with the suggestion that users can protect their own pages. Different wikis have different protection policies on this, so users shouldn't be able to protect their talk pages without having valid reason, and some wikis may only allow it for limited times. With a protected talk page, it's harder to get in touch with someone (give or take Echo notifications). ProcrastinatingReader (talk) 15:58, 19 November 2020 (UTC)[reply]
  • Can't you also have individual users block certain comments? I mean, I don't see a problem in making that happen.
    • Hi. User:Bédévore/Harassment I'm a target of LTA, made by vandalism-only accounts who leave me their insults and threats cross-wiki, including on wikis where I have never written a single word. I don't really see why an IP user or single-purpose account would contact me on hu. or ko. or az.: I never write there--I don't know the language. If I don't edit at all these wikis, why on earth would an IP choose to leave me a relevant message in russian or persian? Especially when my meta page is clear about the stuff. - Bédévore (talk) 23:03, 8 December 2020 (UTC)[reply]
  • By the proposed solution, will a new user still be able to edit their user page (since they won't be autoconfirmed)? Adding basic info on their user page by the new user is a common task in Wikipedia training events. RXerself (talk) 00:24, 9 December 2020 (UTC)[reply]
  • I suggest renaming this from "Global semiprotection of user talk pages" to "Global semiprotection of specific user talk pages" to avoid the implication that this would be applied to all user pages. Thanks. Mike Peel (talk) 20:56, 9 December 2020 (UTC)[reply]
    Does "Allow for a user's talk pages on every wiki" not imply that this is for all user talk pages? —Atcovi (Talk - Contribs) 21:56, 9 December 2020 (UTC)[reply]
    @Atcovi: Why would you semi-protect my userpage on all wikis for no reason? Mike Peel (talk) 22:33, 9 December 2020 (UTC)[reply]
    Fair point - I believe Rschen7754 has made it clear. —Atcovi (Talk - Contribs) 19:52, 10 December 2020 (UTC)[reply]
  • This could be done by a global abuse filter. Updating a user list inside an abuse filter is not a great experience; I wonder if some generic mechanism for having abuse filters which are restricted to some list of pages (with a mechanism for storing that list outside the filter logic) could fulfill this use case while being useful for other issues as well. --Tgr (talk) 04:08, 13 December 2020 (UTC)[reply]

Voting

Retrieved from "https://meta.wikimedia.org/w/index.php?title=Community_Wishlist_Survey_2021/Anti-harassment&oldid=20877411"