Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

You might be disappointed to learn that human drivers run into firetrucks way more often than Teslas.

The current rate as of October is:

Tesla Autopilot Accidents: 1 out of 4,530,000 Miles; US Average: 1 out of 479,000 Miles



Most accidents happen where Tesla autopilot isn’t being used like city driving and exiting and entering the freeway. Sitting in one lane on the freeway is the safest thing you can be doing and any car with lane centering and adaptive cruise control can do it. That’s why those numbers are a joke. I use autopilot everyday day on a 45 mile commute and my only accident is entering the freeway where my car and another car merged into the same lane.


Nothing better than A/B testing within the same population, right?

> In the 4th quarter, we registered one accident for every 3.45 million miles driven in which drivers had Autopilot engaged. For those driving without Autopilot but with our active safety features, we registered one accident for every 2.05 million miles driven. For those driving without Autopilot and without our active safety features, we registered one accident for every 1.27 million miles driven.

Even if you assume most accidents happen in that last group, city driving with AP off, that ratio (3.45:1.27) is better than the overall estimated proportion of city vs highway accidents for all vehicles, at 2.3:1. At a minimum, AP is making highway driving 10-20% safer, and obviously not causing any new city accidents when it’s off.


AP doesn't switch lanes. SO you turn it off during these events. It's not an AB test since AP is used in a very specific scenario where a person driving is used in situations where AP is incapable of handling like dealing with a stalled vehicle on the road.


How confident are we in the first number? Accidents usually do not happen just because on a straight road, they happen when something unexpected happens. How many unexpected cases were recorded in the first number vs the second? Because simply because of the number of teslas on the roads, I doubt many.


We're absolutely not, but Tesla fans love to repeat this quote, no matter how many times it is explained to them.

You cannot at all equate "subset of miles where AP is a possibility, be it good conditions/road/weather", where AP will turn off or not be available because it can't work, with "all the miles driven by humans in all conditions", and say in any way with a straight face, "Look, safer!". But that's what Elon, and a large number of people do.


That’s where the “per mile” part comes in.


Not sure how this factors in but autopilot primarily only works on the highway.

It’s certainly possible that city driving is safer and also more complex to be done autonomously but it’s something that needs to be considered.


The ratio of city vs highway accidents is 2.3 to 1. At worse the Tesla number is cut in half.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: