(1) Comparison against all other cars in US (and across all states) just introduces lots of sampling bias. Teslas are like much more common in states like CA, than say Tennessee. Tennessee accounts for much more accidents in general than California and that adds up in overall stats, but not in Tesla stats.
(2) What kind of variation do we have in Tesla drivers vs other car drivers. As comment above mentions, Tesla drivers may be more responsible drivers.
(3) What accounts as accident? Minor collisions (fender benders) vs Major collisions. More specifically, I want to know, how does Autopilot react when it encounters something it has never seen before (like the crash in article). Humans can handle things pretty well IMO given they are not way beyond speed-limit, and all kind of random shit happens on road.
(4) Autopilot vs Manual Tesla stats is promising IMO. However, a little more information would be nice to have, like proportion of 1-billion miles logged as autopilot miles.