once again I’m calling for relay actions to be publicly broadcast as a note kind.
Probably doesn’t need to be complicated, someone gets kicked you see they get kicked, you notice someone kicked and there is no record of an action, it’s a bad actor.
Discussion
I too want to see moderation events. But they need a lot of thought to get right.
start with something simple I can see who you muted, let me see who the relay muted.
I’m thinking Kind 1985, since that’s what comes right after a Kind 1984.
Read this… https://s3x.social/nostr-content-moderation It’s how #[3] & I see things working. And #[4] (who works for Rabble) pointed me to an awesome research paper that parallels what I laid out.
Are you on Git? Have you seen the issue I raised and the PR Rabble & I started?
Ignore my “are you on Git?” question. Duh. We’ve had that conversation.
Issa a yellow aim icon there.
but here I’m thinking more from the pov of wanting to see a lack of actions by operators who claims that.
I’m no cryptographer, maybe it needs more complications, maybe those complications could be tied to some proof that the operator is in a v4v share with the writer of the implementation, and the receipt key is some derivative, because I’m mostly looking for the absence of action to verify that sales pitch is true.
but maybe it’s as simple as being familiar with how the implementation handles white and black lists and then format those so clients or relay explorers like .watch could surface them in a standard way. Maybe there is a way I can ping an implementation to return the list that I’m not familiar with. But I’m not speaking about making it easy to interope this data else where.
Thinking about it (casually)…
There needs to be a a block list for a relay that could be published under the pubkey of the account designated by the relay. I’d need to investigate the best way. I’d like to see things like a `reason` code that uses the “NIP-69" coding #[2] & I have proposed plus an `evidence` option that’s a comma-delimited list of nostr IDs and URLs that support the decision.
I also want `severity` and `confidence` options. Severity would be scales that define the level of “offense” within a defined category. And confidence would mostly be used by bots (e.g. they think they see nudity but they’re not sure).
Thoughts?