Requests for comment/Adding abuse filter rights to global rollbackers

From Meta, a Wikimedia project coordination wiki

The following request for comments is closed.
Support: 9 Oppose: 7 Neutral: 0; closed as unsuccessful
No consensus at this point in time to modify the global rollback toolset to include the abusefilter-log-private permission. Should a later request for comment on this matter be opened, it is worth highlighting TonyBallioni's comment that the addition of this permission could "circumvent local processes and local consensus for access to these rights" which would most likely require notification of all impacted communities where a edit filter helper or equivalent user right exists to ensure the request for comment was a global discussion involving all shareholders. ~riley
(talk) 05:36, 15 January 2020 (UTC)
[reply]


Proposal[edit]

I'm proposing that abusefilter-log-private be added to the Global Rollback toolset to allow them to view the details of triggers of abuse filters that have been marked as private, which is the majority of vandalism-related filters. This would give Global Rollbackers the ability to more efficiently respond to and identify vandalism via filter triggers, while still maintaining the privacy of the text of the filters, as the abusefilter-view-private (which allows people to view the filters themselves) right would not be given. Vermont (talk) 20:31, 28 October 2018 (UTC)[reply]

Discussion[edit]

To see how prominent private filters are, I picked a random wiki in the SWMT IRC channel, et-wikipedia. Of the 50 most recent filter triggers, 19 are public. In th-wikipedia, 20 of the 50 most recent filter triggers are public. In hr-wikipedia, 20 of the 50 most recent triggers are public. In lt-wikipedia, 21 of the 50 most recent filter triggers are public. Most of these public filters are not related to vandalism, or are pointed at very obvious vandalism (i.e. this public filter, which is blanking). Vermont (talk) 10:52, 7 November 2018 (UTC)[reply]
Also, being able to view private filters would allow easier tracking of x-wiki LTAs, so Support Support. SITH (talk) 16:43, 18 January 2019 (UTC)[reply]
  • The only thing Global rollbackers can really do is to revert revisions. I wonder whether the Global Sysop toolkit is better for such users? Not sure why users mainly reverting vandalism need to do with abuse filters. Leaderboard (talk) 11:58, 27 January 2019 (UTC)[reply]
Abuse Filters have everything to do with preventing and reverting vandalism. I frequently use them as an assistance to identify potentially harmful edits and actions, and it would be useful for global rollbackers to be able to see abuse filters themselves, to identify why the edit was flagged, rather than just seeing it was. Vermont (talk) 15:38, 30 April 2019 (UTC)[reply]
I don't disagree, but wouldn't it still make more sense in global sysop rather than rollback? At the very least it erodes the meaning of "private." ~ Amory (utc) 15:26, 6 May 2019 (UTC)[reply]
Global sysops already have that permission on Global Sysop wikis. With the issue of privacy, private filters are used so that vandals, spambots, and LTA's wont know how to get around them. Global Rollbackers are essentially confirmed not to be in their community discussion. Vermont (talk) 18:04, 6 May 2019 (UTC)[reply]
  • Support Support I did not know anything about this, but seeing this evidence so I support Owen912 (talk) 13:17, 30 April 2019 (UTC)[reply]
  • Support Support per this zh:Wikipedia_talk:持续出没的破坏者/User:Πrate#提議Πrate破壞的應對措施 and zh.wp Rollbacker can do abusefilter-view-private, see AbuseFilter.--AldnonymousBicara? 16:21, 30 April 2019 (UTC)[reply]
    Adjusted wikilinks twice as the thread get archived and moved. Hope it's fine with you Aldnonymous. --Cohaf (talk) 16:12, 27 May 2019 (UTC)[reply]
  • Support Support. This is a fairly rational proposal - with this we can give more tools to those that need them. Hiàn (talk)/editing on mobile account 18:16, 2 May 2019 (UTC)[reply]
  • Support Support. I think it will be helpful against LTAs.--BRP ever 09:22, 11 May 2019 (UTC)[reply]
  • Support Support. This proposal makes sense. Vulphere 21:49, 12 May 2019 (UTC)[reply]
  • Support Support- good to have, useful at times but I really seldom go about combating spam using filters, as I encountered many false positives which then I switched to Recent Changes. --Cohaf (talk) 16:00, 27 May 2019 (UTC)[reply]
  • Oppose Oppose While filter rights are useful to combat LTA´s (Long term abusers), it takes more than reverts to stop them. Any well established rollbacker who wants to combat LTA´s should apply for global sysop. When it comes to regular vandalism, ORES is more helpful than the abuse filter. On top of that, most edits cought by the abusefilter are either going to be flagged with a tag or denied outright. Abusefilters that do not meet that criteria are either test filters, or in my opinion wrongly configured.--Snaevar (talk) 12:34, 28 June 2019 (UTC)[reply]
  • Strong oppose This makes global rollback a significantly bigger deal than it actually is. It is a tool that is largely intended to be used on small wikis where requesting the individual rights associated with it would be unproblematic. The issue here is that global rollback is truly global: it applies on every project, just not small ones. I don't trust many global rollbackers to have access to some of the more sensitive filters on large wikis, which will often include material that eventually needs to be suppressed. Global rollback is currently very much not a big deal and is handed out as needed to users where it would help them. This changes the right to something where users who would be declined for edit filter helper or it's equivalent on large projects can circumvent local processes and local consensus for access to these rights. The big irony here is that by adding this to global rollback, GR would in my view become a bigger deal than global sysop, which is at least limited by the GS wiki set. TonyBallioni (talk) 01:17, 11 July 2019 (UTC)[reply]
    • That's a reasonable argument I mostly agree with; after review, I see that the ability to view private abuse filters opens up a plethora of content, specifically on large wikis, that would necessitate a significant degree of trust not presently judged for in Global Rollback requests. In order to add this right to this user group, it would require a review of people who would receive these abilities, which is infeasible. Vermont (talk) 03:35, 12 July 2019 (UTC)[reply]
      • Just a note, GR itself circumvent local policies at that very start, for instance in zhwp one need to have 90/1000 to be even able to be considered for rollbacker. Many of GR dont even have 50 but still did rollback on zhwp. I'm not saying that since we circumvent on one part, we circumvent more. I know that some wikis have local policy that you cant use GR if you have rollback revoked like en or you cant use GR at all except for spam like in mrwiki. I will say that if this access is given, vetting needs to be stricter but I dont see how some of the GR cannot be trusted with the logs of private filters. I wont see the logs unless otherwise necessarily for sure. And to the suggestion of asking GR to run for GS if they want to see filters, its different in scope as GS is more restricted. Even if they run for GS they wont have such an access. However, I'll say if this doesn't go through, theres still a global EFH group, so I think it might be good for GR needing the access to apply separately.Regards,--Cohaf (talk) 04:09, 12 July 2019 (UTC)[reply]
        • Cohaf, your last sentence is another reason this isn't needed: there is already a specific vetting process for global edit filter helper, so if someone actually needs this, they can just ask for that.

          The other issue you bring up about the circumventing being inherently part of GR is a bit different here. Despite what many in the community think, read-only user rights are often more sensitive than rights that let you do things. It is impossible for me to tell if someone has been going through private filter logs looking for people's credit card numbers before suppression happens, because, there is no trace that has been given. This is just one example of the type of thing that can happen with a read-only log.

          I also kinda question if this is really needed for the group as a whole. I know I've never used filters extensively: I find them cumbersome and and I only look at them when someone else points it out to me. I think I've done a pretty decent job dealing with LTAs and socks basically never using the read-access I have. Now, I know that it can be helpful for some, but I think the overwhelming majority of GRs can be effective without this. If it is something they need and they are trusted enough, they can request GS or GEFH (or both.)

          The tl;dr is that this can be fairly sensitive on large projects, isn't really needed for global rollbackers to be effective, and duplicates an existing vetting process. I know that GR tends to focus on small wikis, but the issue here is that private filters on large wikis are typically more sensitive, and this would give them access to those without any oversight, which is something I am very much not comfortable with. TonyBallioni (talk) 19:11, 14 July 2019 (UTC)[reply]

          • @TonyBallioni:. Yes, I knew, I am saying that if this proposal fails, theres still a way to gain the access just as a note. Yeah, I agree on read only issue, the same is per deleted content, theres just no log to record who read what. I understood your concerns. Thanks for sharing, the potentially oversightable content is a danger, I wont want to see it either. Regards,--Cohaf (talk) 03:43, 15 July 2019 (UTC)[reply]
  • Oppose Oppose Users who need editfilter access can specifically apply for that, which is a much better idea than expanding global rights even further from their intended scope. * Pppery * it has begun 23:28, 31 July 2019 (UTC)[reply]
  • Oppose Oppose User Group global rollback should not have access to abuse filters. This is the user group abusefilter-helpers (Abuse filter helpers)— The preceding unsigned comment was added by MrJaroslavik (talk)
  • Oppose Oppose Based on the discussion above, it will be helpful against LTA, but If this happen then the value of EFH will be nothing. Regards, ZI Jony (Talk) 14:29, 17 September 2019 (UTC)[reply]
  • Oppose Oppose Per Tony and pppery/MrJaroslavik. This would be a huge change in the GR toolset, one which I think would lead to far-greater scrutiny of GRs, likely making them fewer in number and thus less effective. We have a group for this, we should use it. ~ Amory (utc) 14:59, 13 October 2019 (UTC)[reply]
  • Oppose per Pppery and Snaevar. This could be useful to global rollbackers to help gain a little more information x-wiki vandalism, but they really couldn't do much with the information they gain from it. Those who could benefit from this should apply for GS or EFH. Waggie (talk) 22:40, 23 November 2019 (UTC)[reply]
  •  Very very weak support--SRuizR ¡Pure life! 18:56, 9 December 2019 (UTC)[reply]