Paul Cantrell
Just to state the obvious:
Any moderation system that regularly lets through the kind of sewage that @KimCrayton1 has been highlighting is a moderation system that is failing.
Failing.
Full stop. https://dair-community.social/@KimCrayton1/112872020308883967
Stu
@inthehands @KimCrayton1 just posted about this myself, about someone else getting threats for having the temerity to criticize the platform.
Beyond unacceptable. The people at the helm have all the wrong priorities.
Paul Cantrell
Spare me the lectures about moving instances, the philosophical underpinnings of federation, “just ignore it,” “I don’t see it,” or whatever excuse you’re halfway through typing.
There’s an old engineering saying: all it’s gotta do is work.
It doesn’t work.
Paul Cantrell
The account in that screenshot appears to be on a self-hosted instance. That is one of the recurring sources of moderation troubles: some new actor can create a pop-up harassment node. Since there’s no instance-level moderation, every •other• instance has to block the account •individually•, which takes time.
(Of course, self-hosted instances are also more vulnerable to attacks of various kinds, and are paying for traffic out of pocket if there’s a DDoS, so…real tradeoff for the trolls there.)
Stu
@inthehands I think I read Plemora (for example) is easier to stand up to make throwaway harassment comments. This is all so very broken.
Paul Cantrell
Several replies either asking what the solution is, or proposing one.
Some important context for the posts upthread: We are still at the point where many people are denying that there even •is• a problem with moderation on Mastodon.
The immediate next step here is simply to say, loud and clear, “This is unacceptable.” Let’s not skip that step. It matters.
Paul Cantrell
Beyond that…
…to those proposing solutions: Thanks, keep brainstorming, seek the thoughts of others. There are people who’ve already been thinking about this for a while.
…to those replying in a “How do I help?” spirit: Awesome, please mind the previous post upthread.
…to those inching toward “Don’t complain if you don’t have a solution:” The first step of finding solutions is identifying problems. Denying problems is a way of preventing solutions. Piss off and thx for coming to my TED talk.
Paul Cantrell
My own quick brainstorm:
- Reply controls already, like yesterday
- Default to allowlisting for new/unknown instances instead of denylisting
- Federated moderation: option to publish & subscribe to moderation decisions of other trusted instances. Instance owner could (1) fully subscribe to one (e.g. “mirror all of hachyderm’s moderation”) or (2) automatically apply decisions to own instance that multiple other instances made, past some consensus threshold
Paul Cantrell
[Minor note on last idea, since it got boosted: the system would to need either distinguish originating from federated mod decisions, or employ some other mechanism to prevent graph cycles and thus feedback loops where a since mod decision achieves an arbitrarily high level of trust by circulating. Solvable. Just not a completely obvious idea; needs a little thought.]
Paul Cantrell
In light of all the above, I want to give a shout-out to the #Hachyderm moderation team. In my view, they are doing just about the best job any instance mod team could reasonably do under the current system: proactive, transparent, and often ahead of the curve on the block when these toxic trolls surface.
We can’t expect the average instance to do better — or even nearly as well. They're your gold standard for instance-level work. Further improvements must happen at the larger system level.
Zack Cerza
@inthehands I also feel like users ought to be able to block entire domains. Heck, I can't even find a way to _report_ a domain via the Mastodon UI
Paul Cantrell
@zack You can. It’s buried, and not in all clients. Try the web interface & search for instructions; they’re out there! (Don’t have it off the top of my head, sorry)
Paul Cantrell
@tehstu @kims
Yeah, to be clear, when I say I think they’re doing the best job we could expect, I mean “and there are still problems!” It just can’t all be on instance owners alone. We need better tools.
I also am familiar with some of the bad experiences Gebru has had, and I know that she’s received a lot of the sort of obnoxious, splainy replies that don’t (and probably shouldn’t) result in an outright ban, but still really, really suck. For things like that…
Kim Scheinberg
@inthehands
In light of calling out what is unacceptable... I thought hachyderm was well-moderated too until I saw this:
Followups confirm that others have had problems as well
https://mas.to/@timnitGebru@dair-community.social/112856164658443486
Stu
@kims @inthehands yeah, I was really disappointed when I saw that earlier. I specifically joined here, after shopping around as it were, because the rules specifically make this a safe space.
I need to bring it with the team, somehow.
Paul Cantrell
@tehstu @kims
…I’ve wondered about a sort of “personal block to taste” subscribe list system, where if (say) x% of the people I follow block someone, they’re blocked for me too. Because I certainly block lots of people who (1) shouldn’t be banned entirely, not really and (2) I wish I’d never interacted with at all.
Stu
@inthehands @kims that's an interesting concept. I certainly block for vibes, if not outright rule breaking. I could see a benefit of sharing other/trusted people's vibes blocks.
Mind, if you could just set your post to "only people I follow can reply", essentially allow listing, that also reins in what trolls can do (particularly sans QTs). I dunno, very much don't want to come across as having any expertise, or that "tech will save us" kinda thing.
draNgNon
@inthehands many people on Mastodon, when I first joined, were hostile to quote-posts becuase of the potential to trigger pile-ons.
but I've seen over & over again underrepresented persons getting harrassed... nigh invisibly to everyone else. then, there's some weird exchange from reply-guys "well I didn't see it" that sounds achingly similar to "works on my machine".
I wonder if actually having quote-posts would help. Certainly seeing @timnitGebru post some screenshots was eye-opening for me.
Paul Cantrell
@fulanigirl @KimCrayton1
To clarify: when I say the moderation system is failing, I mean •all• of Mastodon, not just dair-community. This is a whole-fedi problem, not something instances can fix on their own. I’m sorry for my bad writing that made that ambiguous. I’ve edited the post to fix, and I apologize.
fulanigirl
@inthehands @KimCrayton1 Not sure why you chose to repost hateful language Kim was experiencing. This is her story to tell and the that of the BIPOC community. Also, perhaps you should catch up with Dair-communities posts so you can understand what is happening. And by the way...read some of the comments about automatically assuming Dair failed. Your own instance is not free from racism...so there's that too.
Paul Cantrell
@fulanigirl @KimCrayton1
Re reposting hateful content: My post only linked to Kim’s. Did I mishandle that? I specifically do •not• want to repost hateful language; I want to direct people to pay attention to Kim. Is there a better approach I should have used, instead of linking to her post?
Luka Rubinjoni
@inthehands @KimCrayton1 I notice some weird behavior of the "stock" Mastodon app, while being on mastodon.social: I don't see / can't access the original post from @KimCrayton1 . Once I click on the link, it opens in the browser. I wasn't able to see it before @inthehands posted it. If this is a Garagon/mastodon.social thing, it's an extreme failure of moderation - I'm gaslit into not seeing racism, and thus not believing people who object to it.
Paul Cantrell
@rubinjoni
Bizarre. Could that be because of a moderation action by mastodon.social?
BeAware :fediverse:
@inthehands I've been screaming to the rooftops for some sort of automated moderation for almost a year now. I was told that it wasn't a priority.
Eugen doesn't see things like this as an issue. I suggest moving to a different federated software ASAP, but then you'd have to find out which ones do have more automated moderation and trust that the admins of the instance are using it efficiently.
Emelia 👸🏻
@inthehands I'm working on some of the parts required for "federated moderation" at the moment, through my NLNet-funded FIRES project.
Hachyderm
@inthehands We just wanted to say thank you :hachyderm:
The moderation team works very hard and we appreciate you. We do our best to research harmful content before it reaches the server so that our users aren't exposed to it. We are always open to discuss our process to explain how we do things and how to make improvements.
Paul Cantrell
@hachyderm
I do appreciate you all a whole lot. You have a super tough job.
Like I said in the replies to this discussion, I don’t think it’s reasonable to expect •any• moderation team to do a better job than you all do at the instance level; further improvements need to be systemic and structural, at the level of Mastodon and/or the whole fedi.
Hachyderm
We agree and have a lot of thoughts on moderation tooling and the things that we do to try and make it work regardless. As has been pointed out by others, essentially all Fedi instances (including ours) are volunteer driven and almost all software being developed in the Fedi space is likely also free.
In the past we've focused on "how you can help us" information:
https://community.hachyderm.io/blog/2023/04/05/a-minute-from-the-moderators/
1/2
Hachyderm
That said, we are working on a couple of blog posts right now, and explaining at least some of the tooling could be a fit for the piece, if people find it helpful. Our thinking thus far was that people would mostly find it a lot of information they don't use ever (like learning how to change a tire, never changing a tire, and then forgetting how to change a tire).
That all said, if people find it useful we can queue up a post about tools work.
2/2
Paul Cantrell
@fulanigirl
So after some investigating, it looks like this is a Masto client behavior: if you link to a post, many clients will display the post attached (like a quote tweet), and if the linked post contains a media attachment, some clients will show the attached media in addition to (and in some cases instead of!!) the content of the linked post.
I’m not sure there’s a way to disable this. Investigating….
fulanigirl
@inthehands I don't know what method you used or didn't use but the hateful language was in your post. That made me sad because as I general rule I appreciate what you normally post. The fact that you are on hachyderm and you repeated the ugliness made me worry as some attacks are coming from your server. So yeah.... I hope it was an over sight on your part.
Paul Cantrell
@fulanigirl
Bah. It looks like if I attach an image to •my• post, it stops showing the image from the •linked• post in some clients (web, Ivory) but not others (Mona). I’ve done that with my post linking to Kim above, since it helps with at least some clients…and that seems to be the best I can do. 😤😞
Paul Cantrell
Here’s more from @mekkaokereke on #Hachyderm being well-moderated, expanding on many of the points upthread: what “well-moderated instance” means in the current fedi, how hard that is, how limited it is, how it does •not• mean “free from racism,” and how current tools and current systems limit the ability of individuals instances to make the fedi safe and inclusive:
GNU social JP is a social network, courtesy of GNU social JP管理人. It runs on GNU social, version 2.0.2-dev, available under the GNU Affero General Public License.
All GNU social JP content and data are available under the Creative Commons Attribution 3.0 license.