Motor Mouth: Who’s liable when a self-driving car crashes?

The manufacturers. They have to be. They're selling you something with the very strong implication and expectation that you don't have to pay attention, don't have to drive it. And so, many if not most people will not pay attention, despite any contract that you will undoubtedly sign that says you assume all responsibility.

If manufacturers are not held primarily responsible then they have no incentive to get these things as right as possible; they'll only need to get them "Pinto right." https://en.wikipedia.org/wiki/History_of_Ford_Motor_Company#...

And if they're not held primarily responsible, then that means you have to constantly pay attention, react in less than a heartbeat to take control and avoid an accident that you by law and definition are responsible for. It would be worse than mounting a DVD on the dashboard and watching while driving, worse than DUI.

And now, why are we excited about these things again? So we can read in our cars on the way to work? How irresponsible. And if you can't read while riding, then how wasteful.

EDIT: Watch for when these things begin to be marketed near-mainstream, and the industry calls them something other than "autonomous."

>> "... you, the driver, will have to be ready to take over should its operating system go all Microsoft 'blue screen of death' on you."

Driver re-engagement is a major problem we[1] are running into. Until cars can be fully autonomous on all roads (say 2025), the driver must keep the driver ready to re-engage as necessary. If the car sees (vision or radar) a problem at 200m that it doesn't want to handle autonomously, it has only a few seconds to get you to put down your tablet/coffee and take the wheel. Mercedes and Audi[2] are currently (2014 models) handling this by making you keep your hands physically on the wheel.

>> "...the head spins at the thought of being legally responsible for a car you’re not actually in..."

If your house explodes in your absence because you left the oven on, it's your fault. My uneducated guess it that you might see legal indemnity enacted or a variation on some U.S. states' "no-fault" laws.

>> "... if we want our cars to drive themselves ... you will never be exceeding 100 kilometres an hour again."

I repeatedly encounter this argument against self-driving cars: they will obey the law. If you want the legal speed limit to be higher, raise the limit.

>> "Why then, if autonomous cars offer little unchaining from everyday driving, are automakers pushing them so hard?"

Automakers make what they think people will buy. They see lots of valuable man-hours spent driving, hours partially recoverable by companies and individuals if the car is self-driving. Governments are pushing it because millions of people are dying annually. (WHO predicts 3% of fatalities will be automotive in 2030.) U.S. seems to be playing catch-up; EuroNCAP has a roadmap to require more autonomous safety features over the next several years to get 5-star safety rating.

[1] Automotive active safety engineers.

[2] https://news.ycombinator.com/item?id=7027018

While I agree we need to come up with an answer within the context of our legal system, I really see the bigger problem as the false responsibility claims our legal system encourages.

Many times there are negative results that should not result in wasteful legal proceedings but our system is so focused on pining the blame on someone to write a check that it accepts blaming x person because there isn't anyone else to blame. Often that is a bad outcome. I would welcome us examining the assumptions about everything needing to be blamed on specific parties.

I do agree there is a benefit to accessing pentalites to culpable parties. I think our current system needs to accept that complex systems often have bad results that should not be handled the way we do now.

The whole issue is very complex but I think there is a need for much better decisions about blame being made in our legal system. It requires much more understanding of science, engineering, statistics etc.. And less overly simplistic blaming based on over simplified models.

My father was a professor of statistics/industrial engineering/chemical engineering and worked some trying to get a better understanding of cause and effect into the legal system decades ago, for example http://www.tandfonline.com/doi/abs/10.1080/00031305.1983.104... I think we need to do much better in this area.

As long as all cars involved in a crash are insured, then the insurance companies can split the cost equally. If any additional liability can be pinned on one car or another, or one manufacturer or another, then they can pay additional costs proportional to that liability. In any case, the number of deaths, claims, and the cost of insurance will be much lower than it is today so likely few will have any complaints.