Recent events do raise questions about the 'first principles' thinking Elon Musk is famous for. As far as Twitter is concerned it seems to mean following his hunches based on outdated Y2K era tech experience.
Maybe it's a different class of problem? Asking "why can't a rocket be reused" from first principles is different from thinking "why can't I block huge IP ranges in Asia to fight bots" and finding out you (predictably) messed up.
@paulg I plan to write a post about this... when Twitter users organically did "quote tweeting" before 2014, they used to write a reply to someone and move the '@' sign later in the text.
Like "I agree @ person: I love ice cream"
Since the @ was not at the start, such replies used to be visible to the replier's followers too
When Twitter implemented native QT they broke this concept. QTs were separate from replies which was their 'original sin' I think.
It's kinda cool that within months I've gone from being vaguely curious about how Stable Diffusion works to implementing ML-based tech. Credit is due to OpenAI and others in this space for taking this research out of the lab and bringing it to the people.
It's especially notable when you consider that people at Google invented much of this tech but the company didn't have the incentive to turn it into general-use products.