6 ms·

All evidence so far suggests that self-driving cars are more dangerous than human drivers when similar conditions are taken into account. The comparison number

All evidence so far suggests that self-driving cars are more dangerous than human drivers when similar conditions are taken into account. The comparison numbers that Tesla parrots are between Teslas with modern safety features and in good weather conditions, and the entire rest of road fatalities, including motorcycles. Comparing similarly bodied cars in similar weather conditions, Tesla's self driving has more reported collisions per miles driven.

lol768 7y ago
>The comparison numbers that Tesla parrots are between Teslas with modern safety features and in good weather conditions, and the entire rest of road fatalities, including motorcycles. Comparing similarly bodied cars in similar weather conditions, Tesla's self driving has more reported collisions per miles driven. I'm interested in learning more about this, because I'd originally taken Tesla's figures at face value. Are there any sources you'd recommend?
- Jasper_ 7y ago
  The best data source I'm aware of is NHTSA's Traffic Safety Facts tables, but the data is a bit hard to filter by. https://cdan.nhtsa.gov/tsftables/tsfar.htm https://cdan.nhtsa.gov/tsftables/tsfar.htm The raw data is available in some other form available by one of the linked sites, but I cannot find it in the rat's nest of hyperlinks right now.
- gnicholas 7y ago
  I co-wrote an article [1] that covers some of the flaws in Tesla’s comparisons a few years back. Not sure if their stats have gotten any better in recent years, but the errors back then were pretty basic. 1: https://www.thedailybeast.com/how-tesla-and-elon-musk-exaggeraged-safety-claims-about-autopilot-and-cars?ref=author https://www.thedailybeast.com/how-tesla-and-elon-musk-exagge...
  - natch 7y ago
    One of the issues raised in the article was insufficient data. I guess there’s a lot more data now!
theferalrobot 7y ago
Do you have some sources on this? Every study I am finding even comparing like for like shows SDVs are about the same or safer. Also the facts I was able to find from Tesla do a lot of comparisons, not just the obscure ones you cite.
- masklinn 7y ago
  > Do you have some sources on this? Fatalities of human-driven cars is ~1.5 per 100 million miles driven, and that's across the entire range of driving conditions. The autonomous car industry (level 3) totaled under 10 million miles driven as of the Uber kill, putting them way beyond despite benefiting from only good conditions. Tesla just reached 1 billion miles on autopilot early this year and has 5 fatalities, that's slightly better than the human-driven rate but (again) one can assume this is in good conditions pretty much exclusively.
  - mentat 7y ago
    Only in good conditions? Have you seen the videos of the Uber kill? Come on... 95% of humans would have had the same outcome.
    - Jasper_ 7y ago
      The dashcam footage released by Uber was misleading; their cameras and sensors were far from as good as the human eye. The Tempe police concluded 85% of humans could have stopped the car in time: [0] > In simulating conditions of the crash and consulting their textbooks, detectives found that Herzberg could have been seen 143 feet down the road by 85 percent of motorists. [0] https://www.phoenixnewtimes.com/news/self-driving-uber-crash-avoidable-drivers-phone-playing-video-before-woman-struck-10543284 https://www.phoenixnewtimes.com/news/self-driving-uber-crash...
      - jmpman 7y ago
        I drove the section of road. If I wasn’t distracted by my phone, I would have stopped properly 100% of the time.
dkarl 7y ago
I agree; that's the kind of comparison I think we should be making. As I said in another comment, if self-driving cars replaced humans the day they became safer than human drivers, they would become our second most deadly consumer technology next to guns. Wringing our hands over fender benders isn't a good start to the conversation. I think instead of worrying that people aren't scared enough, we should think about how hard it will be for people to accept the idea of self-driving cars killing and mutilating people every day while at the same time being a safety upgrade over the status quo they're used to.
- jakeogh 7y ago
  They are not going to be fooled by a false equivilence between self killing hammers and self killing cars.
URSpider94 7y ago
That’s not accurate. The original Tesla AP safety study was done on the Tesla fleet before vs. after AP was deployed. It’s about as close to a clean, controlled experiment as you can get. Same cars, same drivers, same safety equipment, roughly the same road conditions. They also didn’t even take into account whether AP was driving, only whether it was available on the car, so they aren’t looking at just the safest miles driven (highway miles).
- mcguire 7y ago
  http://www.safetyresearch.net/Library/NHTSA_Autosteer_Safety_Claim.pdf http://www.safetyresearch.net/Library/NHTSA_Autosteer_Safety... "As a consequence, the overall 40 per-cent reduction in the crash rates reported by NHTSA following the installation of Autosteer is an artifact of the Agency’s treatment of mileage information that is actually missing in the underlying dataset."
  - AndrewBissell 7y ago
    HN discussion here: https://news.ycombinator.com/item?id=19127613 https://news.ycombinator.com/item?id=19127613 Of course, the original publication of NHTSA's botched study is still the number one HN search result for both "NHTSA" and "Autopilot," and it still routinely gets referenced in comments. As the saying goes, the lie made it halfway around the world before the truth had put its pants on.

lol768 7y ago

>The comparison numbers that Tesla parrots are between Teslas with modern safety features and in good weather conditions, and the entire rest of road fatalities, including motorcycles. Comparing similarly bodied cars in similar weather conditions, Tesla's self driving has more reported collisions per miles driven. I'm interested in learning more about this, because I'd originally taken Tesla's figures at face value. Are there any sources you'd recommend?

Jasper_ 7y ago
The best data source I'm aware of is NHTSA's Traffic Safety Facts tables, but the data is a bit hard to filter by. https://cdan.nhtsa.gov/tsftables/tsfar.htm https://cdan.nhtsa.gov/tsftables/tsfar.htm The raw data is available in some other form available by one of the linked sites, but I cannot find it in the rat's nest of hyperlinks right now.
gnicholas 7y ago
I co-wrote an article [1] that covers some of the flaws in Tesla’s comparisons a few years back. Not sure if their stats have gotten any better in recent years, but the errors back then were pretty basic. 1: https://www.thedailybeast.com/how-tesla-and-elon-musk-exaggeraged-safety-claims-about-autopilot-and-cars?ref=author https://www.thedailybeast.com/how-tesla-and-elon-musk-exagge...
- natch 7y ago
  One of the issues raised in the article was insufficient data. I guess there’s a lot more data now!

theferalrobot 7y ago

Do you have some sources on this? Every study I am finding even comparing like for like shows SDVs are about the same or safer. Also the facts I was able to find from Tesla do a lot of comparisons, not just the obscure ones you cite.

masklinn 7y ago
> Do you have some sources on this? Fatalities of human-driven cars is ~1.5 per 100 million miles driven, and that's across the entire range of driving conditions. The autonomous car industry (level 3) totaled under 10 million miles driven as of the Uber kill, putting them way beyond despite benefiting from only good conditions. Tesla just reached 1 billion miles on autopilot early this year and has 5 fatalities, that's slightly better than the human-driven rate but (again) one can assume this is in good conditions pretty much exclusively.
- mentat 7y ago
  Only in good conditions? Have you seen the videos of the Uber kill? Come on... 95% of humans would have had the same outcome.
  - Jasper_ 7y ago
    The dashcam footage released by Uber was misleading; their cameras and sensors were far from as good as the human eye. The Tempe police concluded 85% of humans could have stopped the car in time: [0] > In simulating conditions of the crash and consulting their textbooks, detectives found that Herzberg could have been seen 143 feet down the road by 85 percent of motorists. [0] https://www.phoenixnewtimes.com/news/self-driving-uber-crash-avoidable-drivers-phone-playing-video-before-woman-struck-10543284 https://www.phoenixnewtimes.com/news/self-driving-uber-crash...
    - jmpman 7y ago
      I drove the section of road. If I wasn’t distracted by my phone, I would have stopped properly 100% of the time.

dkarl 7y ago

I agree; that's the kind of comparison I think we should be making. As I said in another comment, if self-driving cars replaced humans the day they became safer than human drivers, they would become our second most deadly consumer technology next to guns. Wringing our hands over fender benders isn't a good start to the conversation. I think instead of worrying that people aren't scared enough, we should think about how hard it will be for people to accept the idea of self-driving cars killing and mutilating people every day while at the same time being a safety upgrade over the status quo they're used to.

jakeogh 7y ago
They are not going to be fooled by a false equivilence between self killing hammers and self killing cars.

URSpider94 7y ago

That’s not accurate. The original Tesla AP safety study was done on the Tesla fleet before vs. after AP was deployed. It’s about as close to a clean, controlled experiment as you can get. Same cars, same drivers, same safety equipment, roughly the same road conditions. They also didn’t even take into account whether AP was driving, only whether it was available on the car, so they aren’t looking at just the safest miles driven (highway miles).

mcguire 7y ago
http://www.safetyresearch.net/Library/NHTSA_Autosteer_Safety_Claim.pdf http://www.safetyresearch.net/Library/NHTSA_Autosteer_Safety... "As a consequence, the overall 40 per-cent reduction in the crash rates reported by NHTSA following the installation of Autosteer is an artifact of the Agency’s treatment of mileage information that is actually missing in the underlying dataset."
- AndrewBissell 7y ago
  HN discussion here: https://news.ycombinator.com/item?id=19127613 https://news.ycombinator.com/item?id=19127613 Of course, the original publication of NHTSA's botched study is still the number one HN search result for both "NHTSA" and "Autopilot," and it still routinely gets referenced in comments. As the saying goes, the lie made it halfway around the world before the truth had put its pants on.