4 ms·

These kind of metrics should be mandatory for self-driving cars. Just because, say, you can produce a painstakingly detailed timeline should not free you from c

by spease 7y ago

These kind of metrics should be mandatory for self-driving cars. Just because, say, you can produce a painstakingly detailed timeline should not free you from consequences if that timeline indicates you acted with wanton disregard that any reasonable person could tell you would result in loss of life. If someone is so stupid that they don’t realize that a car needs to be able to avoid people in the road, even outside crosswalks, they have no business doing development of self-driving cars. They simply aren’t qualified. Ditto if that person simply doesn’t care. We’re not learning anything from this that we didn’t already know. Of course if you do nothing to avoid a person or a bicycle at 40 MPH, you’re gonna kill someone eventually. This kind of thinking reeks of the “it works with expected input, problem solved” that causes so many issues in critical software. This is like Boeing with the 737 Max. They didn’t want to deal with failure conditions, so they avoided detecting them so they could pretend like they don’t exist. Subsequently, those failure conditions then caused two airplanes to crash and kill the people aboard.

roenxi 7y ago
Aeroplanes have 50 years of global experience in how to fly planes safely. Self driving cars aren't even a full product yet. If (and it is a big if) there is one engineer who was grossly negligent then yes they shouldn't be working on self driving cars. But it is far more likely that this was totally normal corporate ineptitude and will be fixed in the due course of normal engineering processes. A safe culture is not built by people making wild guesses about what and why on forums. It is likely that the engineer responsible for this code is also going to be responsible on net for a very large number of lives saved and judgement of who may have been culpable for what should be left to the courts. Morally I'm happy to say that I believe not only is he or she in the clear but probably deserves a pat on the back for helping to push forward the technology most likely to promote save young lives that I've ever seen. I've had young friends who died from car crashes and ... not much else. Maybe drugs and diseases in a few rare cases. I want technology that gets people taking their their hands off the wheel and I want it ASAP. It doesn't need to be perfect; it just needs to be about as good as what we have now and consistently improving. Anyone halfway competent who is working on that as a goal has my near total support. This is not the time to be discouraging people who work on self driving cars. This is a time to do things properly, non-judgmentally and encouragingly while acknowledging that we can't have cars running people over if we can possibly avoid it and fixing what mistakes get found.
- josefx 7y ago
  > Aeroplanes have 50 years of global experience in how to fly planes safely. Self driving cars aren't even a full product yet. We have over a century on accident statistics for normal cars. Anyone pretending that pedestrians will not jaywalk on a well lit street with low traffic should not develop self driving cars and most certainly should not have a drivers license . > it is likely that the engineer responsible for this code is also going to be responsible on net for a very large number of lives saved and judgement of who may have been culpable for what should be left to the courts. Citation needed. Claims like that can be used to proclaim the most hardened serial killer a saint. I mean if we just let them continue they might start saving people at some point, point me to any "currently applicable" evidence that says otherwise. > I've had young friends who died from car crashes and ... not much else. In my area people die from old age and cancer. Maybe you should move to a place with sane traffic laws that doesn't let the mental equivalent of a three year old loose on the local population. > Anyone halfway competent who is working on that as a goal has my near total support. As a test obstacle on the road?
- SolaceQuantum 7y ago
  "But it is far more likely that this was totally normal corporate ineptitude and will be fixed in the due course of normal engineering processes." Normal engineering processes were built on bodies. We do not want to minimize outrage here. Public anger and pressure is the only way to keep engineering process going.

by spease 7y ago

roenxi 7y ago

Aeroplanes have 50 years of global experience in how to fly planes safely. Self driving cars aren't even a full product yet. If (and it is a big if) there is one engineer who was grossly negligent then yes they shouldn't be working on self driving cars. But it is far more likely that this was totally normal corporate ineptitude and will be fixed in the due course of normal engineering processes. A safe culture is not built by people making wild guesses about what and why on forums. It is likely that the engineer responsible for this code is also going to be responsible on net for a very large number of lives saved and judgement of who may have been culpable for what should be left to the courts. Morally I'm happy to say that I believe not only is he or she in the clear but probably deserves a pat on the back for helping to push forward the technology most likely to promote save young lives that I've ever seen. I've had young friends who died from car crashes and ... not much else. Maybe drugs and diseases in a few rare cases. I want technology that gets people taking their their hands off the wheel and I want it ASAP. It doesn't need to be perfect; it just needs to be about as good as what we have now and consistently improving. Anyone halfway competent who is working on that as a goal has my near total support. This is not the time to be discouraging people who work on self driving cars. This is a time to do things properly, non-judgmentally and encouragingly while acknowledging that we can't have cars running people over if we can possibly avoid it and fixing what mistakes get found.

josefx 7y ago
> Aeroplanes have 50 years of global experience in how to fly planes safely. Self driving cars aren't even a full product yet. We have over a century on accident statistics for normal cars. Anyone pretending that pedestrians will not jaywalk on a well lit street with low traffic should not develop self driving cars and most certainly should not have a drivers license . > it is likely that the engineer responsible for this code is also going to be responsible on net for a very large number of lives saved and judgement of who may have been culpable for what should be left to the courts. Citation needed. Claims like that can be used to proclaim the most hardened serial killer a saint. I mean if we just let them continue they might start saving people at some point, point me to any "currently applicable" evidence that says otherwise. > I've had young friends who died from car crashes and ... not much else. In my area people die from old age and cancer. Maybe you should move to a place with sane traffic laws that doesn't let the mental equivalent of a three year old loose on the local population. > Anyone halfway competent who is working on that as a goal has my near total support. As a test obstacle on the road?
SolaceQuantum 7y ago
"But it is far more likely that this was totally normal corporate ineptitude and will be fixed in the due course of normal engineering processes." Normal engineering processes were built on bodies. We do not want to minimize outrage here. Public anger and pressure is the only way to keep engineering process going.