"it's a start... Musk argued that disclosing what amplifies or downranks tweets would reduce the risk of “behind the scenes manipulation.” [but] algorithms alone offer limited insights" because it's not just an algorithm that defines what you see: it's a huge dataset - "content that enters the platform, each user’s profile, the algorithms’ training data, moderation rules, and the code that trained the models... [so] You can’t simply open-source an ML model" which are constantly evolving as they are retrained.
Apart from being difficult, the "information could be copied by competitors, provide a tempting target for cybercriminals, violate user privacy... hinder Musk’s ambition to “defeat the spambots.”"
On the plus side, "open source offers new opportunities to find vulnerabilities and flaws... useful for research and evaluating biases"
Meanwhile, "Twitter cofounder Jack Dorsey ... suggested... creating an open marketplace of algorithm ... letting users choose which — if any — algorithm they use." Which is exactly what an AI-powered Fediverse should do.
More Stuff I Like