How will self-driving cars affect insurance?

[Zenmervolt]

This is a good point. However I would say it's debatable that most of the way things on a car to fail are the result of improper maintenance. If that was the case, there would not be such a huge disparity between reliability between manufacturers. While proper maintenance can help a lot for less reliable cars, it also draws an ambiguity between faults due to design defects and proper maintenance.

Even assuming, arguendo, that such disparities exist, there is a difference between a design defect, or even a quality defect, and something simply not lasting as long as something else.

There's a reason why you can't sue a car company for the power steering failing on a 20-year-old car with 180,000 miles.

If your theory on manufacturer liability were to come true, no-one would manufacture these self-driving cars at all. The legal risk would be too great. Frankly, that's one of the reasons I think that self-driving cars aren't really practical in the real world.

ZV

 

[yhelothar]

What makes you think there is a "huge disparity" of reliability?

Most "studies" are badly done but even then you can see most cars fall into a very small medium of number of problems. Now if you said many cost more to repair when they have a problem then yea I might agree. European cars still seem to cost more on average to repairs vs most American and Asian. But the number of issues are not as extreme, just how much it cost.

http://autos.jdpower.com/ratings/2014-Vehicle-Dependability-Study-Press-Release.htm


When I was a tech many times large issues were due to the owners, not the cars. This was mostly true 2-3years after the car was made. A recent example was my BiLs timing belt idler bolt broke. Instead of stopping he kept driving and burned up some valves. So instead of a sub $100 timing belt kit it will not be close to $1000. That was the owners fault, not the cars.

Everyone knows Lexus could easily hit 200,000 miles with minimal maintenance while with a VW, you'd be spending more on maintenance to keep the car running past 100,000 than it'd cost to lease a new one.

You linked a study that looked at 2011 cars in a 3 year dependability study. Most of those are under 50,000 miles. It's hardly surprising that there isn't a huge variance of problems given that time frame.

Even assuming, arguendo, that such disparities exist, there is a difference between a design defect, or even a quality defect, and something simply not lasting as long as something else.

There's a reason why you can't sue a car company for the power steering failing on a 20-year-old car with 180,000 miles.

If your theory on manufacturer liability were to come true, no-one would manufacture these self-driving cars at all. The legal risk would be too great. Frankly, that's one of the reasons I think that self-driving cars aren't really practical in the real world.

It does seem that the liability issue would be complex with this. One way around this might be to require drivers to manual override in the event of a system error/failure thus accidents would still be the fault of the driver. Alternatively, perhaps there could be a smog check like system to ensure the automatic driving system is running properly in order for those cars to legally be on the road.

 

[Zenmervolt]

One way around this might be to require drivers to manual override in the event of a system error/failure thus accidents would still be the fault of the driver.

That won't work. If the system fails at the wrong time, forcing manual control won't automatically make it the driver's fault. There are plenty of situations where it's not reasonable to expect a driver of average skill to be able to recover from. Just because a system has returned manual control to the driver doesn't mean that it returns control at a point where manual recovery is even possible with normal skill.

Plus, even if a driver of average skill could have recovered from the situation if he or she had been in control of the vehicle the whole time, that's still not accounting for the shock of suddenly having manual control when automatic control is expected. Remember, these cars are being promoted as allowing people to nap on the way to a destination or to read a book. If you create a situation where people need to keep their hands on the wheel, their feet on the pedals, and their eyes on the road in order to be ready for manual control in case the automated driver fails, you haven't actually made things more convenient.

Alternatively, perhaps there could be a smog check like system to ensure the automatic driving system is running properly in order for those cars to legally be on the road.

And they're going to perform this check how often? Because, technically, not even a daily check of every car would catch all the failures. Smog checks only ensure that the emissions are within spec at the time of the check. They're designed to catch cars that have already broken, not to prevent cars already in good shape from getting worse. They won't prevent catastrophic failures.

The only way they could reasonably prevent a decent number of failures would be to enforce a preventative maintenance schedule similar to the one that the FAA imposes on airlines. They'd need to mandate that still-functioning parts get replaced before they wear out and would need to make the cars disable themselves if the replacement intervals were exceeded.

ZV

 

[Midwayman]

That won't work. If the system fails at the wrong time, forcing manual control won't automatically make it the driver's fault. There are plenty of situations where it's not reasonable to expect a driver of average skill to be able to recover from. Just because a system has returned manual control to the driver doesn't mean that it returns control at a point where manual recovery is even possible with normal skill.

Plus, even if a driver of average skill could have recovered from the situation if he or she had been in control of the vehicle the whole time, that's still not accounting for the shock of suddenly having manual control when automatic control is expected. Remember, these cars are being promoted as allowing people to nap on the way to a destination or to read a book. If you create a situation where people need to keep their hands on the wheel, their feet on the pedals, and their eyes on the road in order to be ready for manual control in case the automated driver fails, you haven't actually made things more convenient.

This is a perfect example of why google wants to go direct to level 4 SDC and skip the level 3 cars. Their research show it takes an impractically long time for someone not actively driving to take over. That means in an emergency you've already hit something if the car's response it to hand it over to a human.

And they're going to perform this check how often? Because, technically, not even a daily check of every car would catch all the failures. Smog checks only ensure that the emissions are within spec at the time of the check. They're designed to catch cars that have already broken, not to prevent cars already in good shape from getting worse. They won't prevent catastrophic failures.

The only way they could reasonably prevent a decent number of failures would be to enforce a preventative maintenance schedule similar to the one that the FAA imposes on airlines. They'd need to mandate that still-functioning parts get replaced before they wear out and would need to make the cars disable themselves if the replacement intervals were exceeded.

ZV

This is actually why I almost expect SDC to be available as a service or lease. Buying them doesn't really make sense. A fleet car will just send itself to the yard at the first sign of sensor trouble. A lease car would be bundled with a service contract. Something happens and your car goes to the shop. A loaner shows up at your house on its own. Traditional ownership would be the place where bad maintenance practices could creep in.