Currently 33,000 people die each year on American roads. Let's say that self-driving cars cut that number to 10,000, but that almost all that 10,000 is because of failures in the self-driving car technology. Will the American public accept that? They've behaved irrationally before, they could again.
Well, people believe that they can avoid being part of the traffic fatality statistics, and many of them can (by driving at appropriate speeds, not drinking and driving, etc.). This feeling of control has value and it does save the lives of people willing to drive responsibly.
I can understand people being very uncomfortable with knowing that no matter what they do on their way to work, the only thing keeping them from ending up on the other side of the windshield is statistical good luck.
They seem to accept that eventuality everytime they step onto a train, bus or plane.
Yes, but they accept that by choice in very small numbers, which suggests (in answer to Percy's question) that they wouldn't accept large-scale replacement of manually-operated cars with self-driving ones.
However, while on the topic it's worth pointing out that public transportation is the safest way to travel. Self-driving cars don't solve the safety problem to near the degree all-out public transportation would (though, imagine self-driving public transportation!) and they only worsen the environmental effects of easy transportation.
It's a little off-topic, but my own opinion is that the answer lies in increasing public transportation and reducing opportunities for private transportation (whether manually- or self-driven). Nothing else, I believe, can solve the health, safety, and environmental problems of transportation.
But with more efficient use of roadway space combined with automated routing maybe it never gets to this point.
But it does. Because more efficient use of roadway space and easier-to-use vehicles both make travel by car more attractive, and so more people do it; and the evidence tells us that the increase in people doing it offsets the gains in efficiency, which is why L.A. has mile-wide freeways that still clog with traffic every afternoon.
In the right picture is a street of cars, in front of which are standing their occupants. On the left is the single bus into which they all could easily fit. The horrible inefficiency of individual automobile travel is made very apparent by such illustrations. Add to this the facts, such as, when you are sitting in your car all alone, you're less than 5% of the load being moved—the rest is car—, and it's really hard to see how a company like Google, serious about making money, could be getting behind what is obviously such a bad idea.
Individual cars are the worst way to get people from one place to another, self-driving or otherwise. I think anyone developing technology meant to 'improve' individuals' driving experiences is betting on a losing horse.
Self-operated trollies (on tracks) are the best approach for now; even if we let them run on petroleum. The trolly just moves and stops where it's told to stop. If you want to get off at the next stop, hit the button by your seat and it will automatically stop—you get off and push the button on the door and send it on its way. If you are at the stop and want to get on, just push the button and it will stop for you. You wouldn't even need an operator, just a law-enforcement officer to maintain order and ensure the trolly isn't taking off too soon, etc.
Like this you don't even need roads.
Imagine how much smaller our cities could be if there weren't a parking lot between every building!