Over on #Bluesky there was a bit of a controversy over the suspension of Sarah Kendzior. I thought it was an interestingly nontrivial moderation problem, so I wrote up a little case study on how this would have been handled on Mastodon: https://www.tbray.org/ongoing/When/202x/2025/11/13/Kendzior-Case-Study
Discussion
Loading...
Post
@timbray Something that is somewhat implicit in the final section, but that I want to expand on: on the Fediverse, if you are on a small to medium sized server, there is a good chance that the moderator and reported user know each other to some extent, and are much more likely to have established some level of trust - this is why the moderator is more likely to understand the user's intention, and why the user is likely to take a warning seriously.
This is why it actually scales *better* than moderation on big instances or centralized platforms
that also makes some of the problems worse tho. Part of the problem with a figurative death threat on a global network is the impact it has on the people outside of the close circles of the person making the threat.
If the recipient of the death threat was an another server (like the writer referenced by Kendzior) there would be very little recourse already anyway, and the fact that violator and mods are buddies would not be great either
@ricci @timbray like, i dont feel great about that the pitch of the fediverse now is that its easier to make figurative death threats. my most normie straight-laced opinion remains that figurative death threats on open public networks are bad, actually, even (especially?) when people try to hide behind 'it was just a joke' or 'just venting'
I see your point here, but I also see it as not a matter of "where are death threats allowed" but "where are these statements considered actual condoning of violence and where are they not" - there are cultural contexts (and I don't just mean in an international sense, but in a "community" sense as well) where certain statements are seen as actual intent to do harm, or to encourage others to do so, and others where they are not.
A platform like today's Bluesky (I know, they aspire to be - and are - moving in a more decentralized direction) where one moderation team handles 99.5% of the network's users has to pick one standard and apply it everywhere (well, with exceptions for the authoritarian regimes, as we discussed elsewhere in this thread). And of course this standard is not at *all* neutral. It is a white American standard. That's fine, if that's the userbase they want.
But this is what I mean when I say that it's a good thing for moderators and reported users to have an established relationship and established trust. It's not just because "hey my buddy is not going to kick me off", it's because, in an online community that is closer in size to a natural human community, the social tools for establishing expected behavior are - certainly imperfect - but much more human. They have much more shared context. Whereas a distant, faceless moderator team is a necessity at very large scale, but it is also fundamentally more authoritarian.
Let's be clear, though: literal death threats are allowed on Bluesky. You are even allowed to post videos of you killing people along with the death threat, to prove that you are both serious about carrying out the death threat and capable of doing so.
https://bsky.app/profile/deptofwar.bsky.social/post/3m4z6dwopcc2d
bonfire.cafe
A space for Bonfire maintainers and contributors to communicate
Automatic federation enabled