Post AtBUICIDoNp2D3ExWq by JessTheUnstill@infosec.exchange
(DIR) More posts by JessTheUnstill@infosec.exchange
(DIR) Post #AtBIt7YPC9OumCeMsK by Daojoan@mastodon.social
2025-04-17T11:51:15Z
0 likes, 0 repeats
The "marketplace of ideas" was always a flawed metaphor. Ideas don't compete on merit; they compete on memetic fitness—how easily they spread, how good they make believers feel, and how effectively they suppress competing ideas.
(DIR) Post #AtBIt8WJbPK5m0gDFw by wjmaggos@liberal.city
2025-04-17T12:01:29Z
0 likes, 0 repeats
@Daojoan what's a better one? fedi is the best tool ever for the people collectively to decide what info, ideas and art should go most viral worldwide, without the influence of wealth and unearned status. we just have to get everyone here. #AttentionDemocracy
(DIR) Post #AtBR4s8L9X0i3jZZPE by JessTheUnstill@infosec.exchange
2025-04-17T13:33:17Z
0 likes, 0 repeats
Fediverse is still edited and moderated. There are absolute cesspool instances full of proud Nazis and pedophiles and such. Moderators have to do work every day to block or limit these instances and users. Volunteer moderation doesn't scale, and is subject to all the same biases humans have.The only reason this place feels as comfortable as it does to you is because:You're a similar alignment to the others in your feed and your moderatorsThe place is still a (barely) manageable size that volunteer moderators can (barely) keep up with filtering out the worst of the worst.The technology and platform of the fediverse won't save us. It's the work of thousands of people who make it sorta liveable for people.@wjmaggos @Daojoan
(DIR) Post #AtBUIAwuo7BW2gGQgC by ottomate@noc.social
2025-04-17T13:41:00Z
0 likes, 0 repeats
@JessTheUnstill @wjmaggos @Daojoan I have not yet seen any successful example of automating moderation. I am grateful for the quality of the toots that now flow to me so its good to understand to whom that gratitude belongs.
(DIR) Post #AtBUIBiPxVBMQ0UMim by JessTheUnstill@infosec.exchange
2025-04-17T14:00:04Z
0 likes, 0 repeats
Because it's impossible to do. If you do word filters, people just use dog whistles and euphemisms. Or it hits minorities trying to discuss the harassment they experience. A natural language model or an LLM still can't catch nuance and ever changing language. Even human mods have to constantly learn what is and isn't deserving of moderating, and try and consider the context in which the reported post was made.@ottomate @wjmaggos @Daojoan
(DIR) Post #AtBUICIDoNp2D3ExWq by JessTheUnstill@infosec.exchange
2025-04-17T14:01:21Z
0 likes, 0 repeats
Moderating at scale is and probably always will be one of the unsolved problems of social media. At best, it's a game of limiting it to not be TOO toxic.@ottomate @wjmaggos @Daojoan
(DIR) Post #AtBUICsNdwkI1C9ptA by wjmaggos@liberal.city
2025-04-17T14:09:15Z
0 likes, 0 repeats
@JessTheUnstill @ottomate @Daojoan isn't it a game of keeping the instances small and getting the moderators paid? I think this will happen more as people come to accept the other models don't work and not paying leads to burnout.my more controversial take is to focus the moderation primarily on unwanted tagging and hashtag pollution, and less on misinfo and shitty opinions where only the people who follow them have to see it. and any coordination of offsite abuse and illegality, of course.