You can’t compare FSD crashes / mile with human data for a couple of reasons. The primary one is: 1) FSD turns itself off whenever things get too difficult or

You can’t compare FSD crashes / mile with human data for a couple of reasons. The primary one is: 1) FSD turns itself off whenever things get too difficult or complex for it. Human drivers don’t get to do this. Recent crowdsourced data suggests a disengagment every 6 miles of driving: https://twitter.com/TaylorOgan/status/1602774341602639872 https://twitter.com/TaylorOgan/status/1602774341602639872 If you eliminate all the difficult bits of driving I bet you could eliminate a lot of human driver crashes from your statistics too! A secondary issue, but relevant if you care about honesty in statistical comparisons is 2) The average vehicle & driver in the non-Tesla population is completely different. Unless you correct for these differences somehow, any comparison you might make is suspect.