2 ms·

> It's not acceptable to have FSD crashes same as it's not acceptable for people to crash. Not sure what you mean by that? We could prevent almost every death,

by IcePic 1y ago

> It's not acceptable to have FSD crashes same as it's not acceptable for people to crash. Not sure what you mean by that? We could prevent almost every death, for instance by limiting speeds to <10 for instance, that would remove SO many accidents and incidents. It would also make travel take a really long time, like as if it was 200 years ago and you rode your horse across the country. But it would prevent almost all serious accidents. So the statement "its not acceptable for people to crash" is not true in a general sense. We are already balancing a certain amount of risk versus convenience of riding at speed to get places fast, to get transports in a timely manner and so on. Some countries, like Germany has roads with free speeds and get really horrible crashes at times involving many cars. Others place maximum limits to try to strike some kind of balance between accident rate/impact and the convenience or speed people would like to have, but from what I can tell, no country has set the limits so low that it actually fulfills what you wrote, that it is not acceptable for people to crash. It is solvable today, just not popular to make that actual decision. But it could be done and people would stop dying on the roads.

IcePic 1y ago
And I was meaning to circle back again to the comparison of incidents and crashes with or without computer assistance but got derailed. =) If we then figure that we are in the situation where those who decide are balancing between X% fatalities per year when people drive like they do currently, why would it be controversial to allow FSD or other similar techniques to have the same or lower rate of crashes? I'm not buying into Teslas own statistics on crashes per km/mile because lying with statistics is kind of easy, but lets pretend one or all car manufacturers manage to get cars to have 0.4% crash rate per year when people have 0.5%, why would anyone then want to prevent people from using FSD? Why would we apply a different set of standards for one "type" of driver and not the other? And if we were to apply different standards, which might be reasonable, why would we not choose the "safer" one? Not claiming we necessarily are at a place where computer assisted driving is demonstrably safer, but when we actually can prove it is even just a tenth percent better, by that time I think we should have an answer for this question.
- TestingSpike1 1y ago
  We need to mandate FSD have 0 fatalities otherwise you put a price on human life. That means Tesla can just optimize to reach that target instead of 0. i.e: why should we pay for lidar or better training, it's good enough. Comparing to human accident rate is flawed. There we also expect 0 accidents. It's not acceptable for a human driver to do what FSD did here and the driver would have consequences as opposed to a company.
TestingSpike1 1y ago
[flagged]

by IcePic 1y ago

IcePic 1y ago

And I was meaning to circle back again to the comparison of incidents and crashes with or without computer assistance but got derailed. =) If we then figure that we are in the situation where those who decide are balancing between X% fatalities per year when people drive like they do currently, why would it be controversial to allow FSD or other similar techniques to have the same or lower rate of crashes? I'm not buying into Teslas own statistics on crashes per km/mile because lying with statistics is kind of easy, but lets pretend one or all car manufacturers manage to get cars to have 0.4% crash rate per year when people have 0.5%, why would anyone then want to prevent people from using FSD? Why would we apply a different set of standards for one "type" of driver and not the other? And if we were to apply different standards, which might be reasonable, why would we not choose the "safer" one? Not claiming we necessarily are at a place where computer assisted driving is demonstrably safer, but when we actually can prove it is even just a tenth percent better, by that time I think we should have an answer for this question.

TestingSpike1 1y ago
We need to mandate FSD have 0 fatalities otherwise you put a price on human life. That means Tesla can just optimize to reach that target instead of 0. i.e: why should we pay for lidar or better training, it's good enough. Comparing to human accident rate is flawed. There we also expect 0 accidents. It's not acceptable for a human driver to do what FSD did here and the driver would have consequences as opposed to a company.

TestingSpike1 1y ago

[flagged]