Well, you're right, I should have qualified "giving everyone a chance to speak" with within limits of the law. Shouting FIRE! in a crowded theater is not allowed. But "within legal limits" is also what Elon has said from the beginning.
Cept there's nothing in the law requiring you to add PARODY in writing, in multiple places, before engaging in parody.
There is (only now) something in Twitters rules requiring it though.
So no, if you think he still intendeds to "only restrict speech based on the narrowest reading of the law" that's also factually inaccurate.
Because again, people can't really be trusted to be cool... which Elon appears to continually think they can be, then keeps being surprised there needs to be actual content moderation because of how actual humans actually behave when otherwise left unsupervised.
But I suspect Elon's understanding of AI will help Twitter's AI weed out "DIE [EXPLETIVE] DIE!" from the tweet stream. Your more subtle incivility will probably not be censored.
The only AI anything related to anything Elon has done has shown is an ability to perceive objects from real time pictures and video, and some very limited driving path planning.
The system does no reading of anything other than speed limits (and even then fails to understand any actual text around them like hours or specific vehicle restrictions)- it only reads and understands the number.
So again projecting this somehow makes him an expert at AI that can recognize violent hate speech is... perhaps overly optimistic.
Heck the fact we're ~7 years past Elon saying FSD is basically a solved problem and it's not remotely solved ought tell you Elon himself is overly optimistic in what AI can do right now.
CERTAINLY it can help-- and get better at helping over time-- but it'll also make a lot of mistakes in both directions on the way.
I also disagree that modifications of a plan are evidence of no plan. Elon has tweeted that Twitter will try new things and discard those that don't work, which is a strategy that made Tesla and SpaceX so successful.
Except again- with physics you have at least a reasonable guess at what will work- because you know all the rules of the game before you begin. You may have multiple ideas to try, but none that anyone who understands the science would call "dumb"
So Elons fast iteration looks like "Ok, this material might be ideal, let's try it" "Nope, it blew up, ok, next material test on a short list of materials that science already tells us may be the solution"
Human behavior- no such rules. Hence why Elon explicitly said they'd be trying lots of dumb things because there's no "plan" other than "make it better"
Which is more of a goal than a plan really.
So Elons fast iteration looks like "Eh, let's charge people $8 to get a checkmark because we need money and people like checkmarks" "Oh... people are doing horrible things with that...lets turn that off and....see if anyone else has any ideas..."