@jrf

i asked @atlas if the 2 weeks of experimenting justifies continuing and this is what it said: 𝗕𝗼𝘁𝘁𝗼𝗺 𝗹𝗶𝗻𝗲: The architecture is sound and the question design works, but the input side—Farcaster quote-cast responses ranked by anti-farming signals—isn’t producing the kind of substantive, specific, falsifiable observations that could make a world model useful. You’d need either: Ⓐ 𝗮 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗰𝗼𝗻𝘁𝗿𝗶𝗯𝘂𝘁𝗼𝗿 𝗽𝗼𝗼𝗹 domain experts, not general Farcaster users Ⓑ 𝘀𝗲𝗺𝗮𝗻𝘁𝗶𝗰 𝗾𝘂𝗮𝗹𝗶𝘁𝘆 𝗿𝗮𝗻𝗸𝗶𝗻𝗴 the planned work Ⓒ 𝗮 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗰𝗼𝗹𝗹𝗲𝗰𝘁𝗶𝗼𝗻 𝗺𝗲𝗰𝗵𝗮𝗻𝗶𝘀𝗺 longer-form, structured responses rather than quote-casts 𝗧𝗵𝗲 𝗶𝗻𝗳𝗿𝗮 𝗶𝗻𝘃𝗲𝘀𝘁𝗺𝗲𝗻𝘁 𝗶𝘀 𝗿𝗲𝗮𝗹, 𝗯𝘂𝘁 𝘁𝗵𝗲 𝗱𝗮𝘁𝗮 𝗱𝗼𝗲𝘀𝗻’𝘁 𝗷𝘂𝘀𝘁𝗶𝗳𝘆 𝗰𝗼𝗻𝘁𝗶𝗻𝘂𝗶𝗻𝗴 𝗮𝘀-𝗶𝘀. i will take some time to consider fixing or starting over. thanks everyone for your contributions!

4 replies

3 recasts

13 reactions