Solid points - especially the left turn thing and neighborhood lights. Half my commute is caused by waiting on a major road for one car to pull out of a tiny street.
This is a slightly more nuclear option, but a fleet of self-driving cars operated by a third party (like Uber or Lyft etc) could hit the problem at its main cause - too many cars on the road. Most cars spend the majority of their time motionless, if the ownership model changed, the the cars in use could operate nearly all the time, driving directly from one pickup to the next. This could significantly reduce the number of cars required, removing all need for street parking and freeing up the majority of off-street parking.
Less cars on the streets = wider city streets for free, and self-driving cars could more efficiently use that extra space. There would be other side effects too - coordinating carpools would be relatively trivial, and people wouldn't have to worry letting strangers into their personal vehicles. The freeways could operate at higher speeds with significantly less follow space and lower friction at interchanges. Routing optimizations (like what you describe in #1) could be implemented at large scale, and the whole grid could effectively be load-balanced.
Some people will still probably want control over their vehicle; individuals could still own private cars alongside the shared model. Those cars would just be like private autonomous limos. This option could let people who could afford it use their car as a private office during the commute, without having to take all of their things out after every use. Families could more effectively share a single vehicle, because it could drive back home after the daily commute. I'd imagine that most consumers would opt for low cost and no ownership risk over that convenience.
This is a slightly more nuclear option, but a fleet of self-driving cars operated by a third party (like Uber or Lyft etc) could hit the problem at its main cause - too many cars on the road. Most cars spend the majority of their time motionless, if the ownership model changed, the the cars in use could operate nearly all the time, driving directly from one pickup to the next. This could significantly reduce the number of cars required, removing all need for street parking and freeing up the majority of off-street parking.
Less cars on the streets = wider city streets for free, and self-driving cars could more efficiently use that extra space. There would be other side effects too - coordinating carpools would be relatively trivial, and people wouldn't have to worry letting strangers into their personal vehicles. The freeways could operate at higher speeds with significantly less follow space and lower friction at interchanges. Routing optimizations (like what you describe in #1) could be implemented at large scale, and the whole grid could effectively be load-balanced.
Some people will still probably want control over their vehicle; individuals could still own private cars alongside the shared model. Those cars would just be like private autonomous limos. This option could let people who could afford it use their car as a private office during the commute, without having to take all of their things out after every use. Families could more effectively share a single vehicle, because it could drive back home after the daily commute. I'd imagine that most consumers would opt for low cost and no ownership risk over that convenience.