Hopefully the facts about computing speed mentioned here are correct. Most of the other stuff is very questionable. First of all HW4 can not at all do 'unsupervised FSD'. Elon has stated that they will do this tomorrow, but there is not a single bit of evidence that he will accomplish that. HW4 can do 'supervised not-full self driving', and makes life-endangering mistakes within 10 miles if not 'supervised'. The Austin 'robotaxi' promise has still to fullfill itself and will be postponed and faked by remote controlling the vehicles once they're able to implement that. How many 'parameters' are needed to get to the real full self driving promised by the vision of Elon in 2017,... nobody knows. HW5 might be enough, maybe not. But we know that Waymo can do it NOW without such technology. Tesla trying to accomplish something similar in an equally geofenced area doesn't make me enthusiastic. Elon is trying to package a catching-up situation into something more fancy, but who still believes him now anyway?