Gregory's Serverlook i’m with you on bluesky being bad. but there are lessons to be learned and postmortems to be had. and anyone claiming there isn’t is in denial. you all were told by me and others all the way back in the early days of ActivityPub that we needed to take the abuse and trust problems seriously.
|
6 comments
 for the output of a protocol design group to be effective, the work has to be centered around solving real problems for the consumers of that protocol. or, in the case of BlueSky, at least faking the appearance of that. in SocialCG we had demos of people reinventing Grindr but with ActivityPub. cool, I guess, but what about abuse? what about building systems of consent? little energy for those things. if the goal of the fediverse is actually to build a platform for techies to shitpost, fine, but own it. stop talking about it going mainstream, because it will never go mainstream without some sort of real trust and safety story. it just won’t.  @ariadne so, the current story is per-instance mod teams plus between-server blocks. We probably need better ootb support for shared blocklists, so new sites don't get brigaded by Nazis. We probably also need better shared metrics, and I'd like to see more Bayesian filtering to catch harassing content before it happens. Anyway, do you think there are other structures we need? @evan we don’t need any of that. what we need is the ability for users to *opt in* to interactions with others, rather than having everything thrown in your face until you smack the block button. diaspora gets this right. |
why do the bluesky advocates talk about decentralization not being important? it’s simple: decentralization is hard to get right from a security POV. it is the decentralized nature that allows trolls to spin up throwaway instances just to send death threats and other content to people they want to harass. you might disagree on the solution to the problem, but it’s a problem.