Wouldn’t distributed moderation kind of remove distribution in a sense? Your impl sounds like it’s distributing compute, but more central in control. E.g. Lemmy mods can only mod their own instances, so power is distributed more than if they could mod any instance.
Lemmy is decentralized, but not distributed. I’m essentially saying that there are no instances, everyone is just a client, just like BitTorrent. The only servers that would exist are relays to help connect people behind firewalls.
There are no mods of a community, at least not mods everyone agrees on. The concept of moderation is entirely local and based on the peers you trust, so something like this:
User A flags a post as problematic
User B explicitly trusts A and hides the post
User C implicitly trusts User B and may hide the post
Each step here is based on statistics, so you never completely trust anyone, and the moderation decisions would be completely public just like on Lemmy (I’m thinking of a “discovery queue” like Steam has, but for reviewing trust). Explicit trust (you’ve clicked a button) would give a high starting trust factor, whereas implicit trust (you’ve upvoted their posts and made similar mod decisions) would have a smaller impact.
At least that’s the gist. I’ll need to test it out to see what performance looks like, but I think it should work well in theory. I also have some privacy issues to work out, like hiding IP addresses (may need to rely more on relays or something like Tor).
Wouldn’t distributed moderation kind of remove distribution in a sense? Your impl sounds like it’s distributing compute, but more central in control. E.g. Lemmy mods can only mod their own instances, so power is distributed more than if they could mod any instance.
Lemmy is decentralized, but not distributed. I’m essentially saying that there are no instances, everyone is just a client, just like BitTorrent. The only servers that would exist are relays to help connect people behind firewalls.
There are no mods of a community, at least not mods everyone agrees on. The concept of moderation is entirely local and based on the peers you trust, so something like this:
Each step here is based on statistics, so you never completely trust anyone, and the moderation decisions would be completely public just like on Lemmy (I’m thinking of a “discovery queue” like Steam has, but for reviewing trust). Explicit trust (you’ve clicked a button) would give a high starting trust factor, whereas implicit trust (you’ve upvoted their posts and made similar mod decisions) would have a smaller impact.
At least that’s the gist. I’ll need to test it out to see what performance looks like, but I think it should work well in theory. I also have some privacy issues to work out, like hiding IP addresses (may need to rely more on relays or something like Tor).