Hacker News Reader

I think a lot of this depends on how often drivers use the AP feature, and how many close calls have been caused by the AP. It's also pretty important to get an idea of the timeline of those crashes. If there was a spike in crashes after a software patch, then that would definitely warrant an investigation.

Also the accidents they have caused are pretty significant. There does appear to be a bug that causes a Tesla to slam on the breaks for no reason, and this has caused multi-car pileups.

I thought full self driving would be better than it is. My intuition tells me that not treating these self driving crashes like a regular manufacturing defect is wise.

Perhaps they believe (or know) that 36 is not the total.

The NHTSA opened an investigation on sticky pedals in Toyota vehicles after three crashes. 36 crashes caused by a defect is comparatively enormous.