Approxamatrix
Alexia Starling 
blog.cyrneko.eu/there-is-no-bsky-social-instance
New blogpost.
New blogpost.
Jesse 🇫🇷 
genMoji | 🟢@alexia alexia being unhinged as usual /pos
lalalalalalalala@alexia yay cool blogpost!!! idk about feditips or whatever is going on there but i think ur blog post actually identifies a problem with bluesky today? on fedi governance of users is split up among a ton of instances and any instance can get defederated so theres external pressure on instances to not invite bad people (jd vance) or theyll lose their social graph. as far as i can tell that doesnt exist on bluesky, because (i thought) most people are governed by the main PDS, but now it sounds like its also made difficult by the protocol itself because no defederation? and "user" seems like too small a unit for every individually moderated PDS to effectively keep up a blocklist for. so how does that work? if the main bluesky governance went really really really bad (like no moderation at all) what would a separately run fully independent pds do? block all their bad users but not bsky.social?
@760ceb3b9c0ba4872cadf3ce35a7a4
most people are governed by the main PDS,The PDS makes no moderation decisions, it only stores user data.
As the title implies, there's also no "bsky.social" instance. It's just usernames that end on bsky.social
If something went horribly bad, you'd just switch to a different moderation effort. That said no one hosts a full-stack bluesky "instance" as of right now, only individual components namely missing the AppView.
most people are governed by the main PDS,The PDS makes no moderation decisions, it only stores user data.
As the title implies, there's also no "bsky.social" instance. It's just usernames that end on bsky.social
If something went horribly bad, you'd just switch to a different moderation effort. That said no one hosts a full-stack bluesky "instance" as of right now, only individual components namely missing the AppView.
@alexia okay interesting ty for clarifying! so how do moderation decisions work then? like when a user is kicked off the platform what actual action is performed then? id have assumed that their actual account is terminated = their data removed from the PDS?
@760ceb3b9c0ba4872cadf3ce35a7a4 depends, actually!
There is ONE single moderation action that one can do on the PDS directly and that is indeed account termination. But of course that only works when Bluesky has access to the PDS the account is on
but everything else gets handled by a Moderation Service, most prominently @moderation.bsky.app (←link) which puts "labels" on to accounts, even "remote" ones
There are some labels, like
There are of course other labels, i.e a spam label and there is other moderation services which give out other, custom labels
There is ONE single moderation action that one can do on the PDS directly and that is indeed account termination. But of course that only works when Bluesky has access to the PDS the account is on
but everything else gets handled by a Moderation Service, most prominently @moderation.bsky.app (←link) which puts "labels" on to accounts, even "remote" ones
There are some labels, like
!hide which effectively make an account not exist to anyone that follows this moderation service. Because the default configuration auto-follows the Bluesky Moderation Service AND the bsky.app frontend enforces it (i.e can't be unfollowed easily), everyone is effectively siphoned into following the moderation decisions of Bluesky PBLLC.There are of course other labels, i.e a spam label and there is other moderation services which give out other, custom labels
@alexia ohh okay i get it! hmm .. so it seems like, unlike fedi , there is no network pressure to encourage moderation.bsky.app to make good moderation decisions because moderation is structurally centralized and theres no threat of fedi-style "defederation" or losing the rest of the social network. this seems like a flaw to me although it avoids all of the fediblock/blocklist stuff we have here