2 ms·

> Tesla isn't hiding data No they are. They won't break out FSD and they 100% could > their autopilot data suggests it's safer than human pilots on highways

> Tesla isn't hiding data No they are. They won't break out FSD and they 100% could > their autopilot data suggests it's safer than human pilots on highways Their autopilot data compared to every single type of car, with every single type of driver, only on highways. So we're comparing every 1994 Corolla with a teenage driver to every 2021 EV with a 30 yr old driver. The reality is AP should really be compared to modern cars which tend to come with LKAS and AEBm and stats should be adjusted to reflect the differing driver populations. But Tesla would rather throw up comparisons that make themselves look good, but only past muster if you don't think about it too much.

Volundr 4y ago
> Their autopilot data compared to every single type of car, with every single type of driver, only on highways. Is there new data I missed? Last I was aware Telsa had published data comparing autopilot to human drivers, but not adjusting for road type, weather, road condition, construction, and any number of other factors in addition to the ones you mentioned. So basically Telsa determined that autopilot on easy-mode driving (highway, clear enough conditions for autopilot to be willing to operate), with a human backup, has less crashes per-mile than humans in the full range of conditions (including those where autopilot will opt out) with no backup. If that's not an example of "lies, damn lies, and statistics" I don't know what is.
- BoorishBears 4y ago
  I think you're misunderstanding the quoted section, my exact point is they're comparing a narrow section of drivers on highways (people in Teslas who enabled AP) to all drivers on highways The narrow section excludes older vehicles, skews away from the most at risk demographics of drivers, removes the toughest highway situations from the equation since AP will disengage, the list goes on.

Volundr 4y ago

> Their autopilot data compared to every single type of car, with every single type of driver, only on highways. Is there new data I missed? Last I was aware Telsa had published data comparing autopilot to human drivers, but not adjusting for road type, weather, road condition, construction, and any number of other factors in addition to the ones you mentioned. So basically Telsa determined that autopilot on easy-mode driving (highway, clear enough conditions for autopilot to be willing to operate), with a human backup, has less crashes per-mile than humans in the full range of conditions (including those where autopilot will opt out) with no backup. If that's not an example of "lies, damn lies, and statistics" I don't know what is.

BoorishBears 4y ago
I think you're misunderstanding the quoted section, my exact point is they're comparing a narrow section of drivers on highways (people in Teslas who enabled AP) to all drivers on highways The narrow section excludes older vehicles, skews away from the most at risk demographics of drivers, removes the toughest highway situations from the equation since AP will disengage, the list goes on.