You're assuming car ownership will remain in place as it is now. If vehicles are shared instead of fully owned, you can be more efficient. Think of how many trips are simply transporting a car. For example, let's say I need to pick up a friend from the airport. I would waste a one-way trip going to the airport. I didn't need to go to the airport. The only reason it was necessary for me to go is that I needed to get a car to my friend.

If my fried could have just summoned the nearest self driving car, an entire one way trip wouldn't ever occurred.

If it becomes popular to share cars then this may work. Looking at reality I have my doubts though. If people wanted to share cars you would see neighbors sharing cars for example. In the US many families or friends not even share cars. When I go out with people it ends up with seven people going in 4 cars or similar.

A self driving car takes the pain out of driving so unless society changes a lot we will see more cars driving around (often empty) and people will have longer commutes while sleeping their car.

Sharing a vehicle is not fun when the other driver could: 1. wreck it, 2. not return it promptly, 3. trash it. A self-driving car solves problem 1, partially solves 2, and does nothing for 3.

I guess a self driving shared car needs also to be self cleaning. that may work.

Put together a amalgam self-driving Sanisette [1] and you implement an even more ridiculous aspect of Idiocracy than its La-Z-Boy with a built-in toilet [2]. And while I make a joke about it now, someones in the future are going to convince a line of VCs to incrementally enable people eventually gaming/Facebooking in VR while hurtling down the freeway doing Number Two, and get handsome exits on each step towards that, and it will all be counted as a net economic benefit in the end. I'm either kidding, or will have checked out well before that ever becomes mainstream.

[1] https://en.wikipedia.org/wiki/Sanisette

[2] http://www.eatmedaily.com/wordpress/wp-content/uploads/2009/...

However, there is a flaw in that system, in that most people tend to be on the road at the same time (rush hours).

I think that self driving cars will help with parking, but not with traffic. You won't have to park your car in the city, where space is at a premium, the car will go park itself at another location, or pick up another passenger.

I think self-driving cars will reduce congestion in two ways. On freeways, due to slow human reaction times, there is a 2 car-length space between cars, meaning cars only take up about a third of the space on the freeway. With self-driving cars, there is near instant reaction time, so the space between cars will be 2 feet. With that, you basically triple the amount of cars on the freeway, reducing congestion by two-thirds.

In addition to this, on surface streets, traffic lights meter traffic through intersections. With self-driving cars, the cars will be able to go through intersections without the need for a traffic light, eliminating that metering effect.

Highly unlikely. We'll probably be able to get twice the usage out of existing roads because the distance between cars will shrink dramatically.

In my cynical opinion SDCs will likely precipitate a ten year pause in new road construction as drive times drop dramatically. But at some point we'll catch up to the additional capacity and spend another five or six years suffering as roads are built and expanded.

But the newly constructed roads will be more efficiently utilized than roads at present, so it will be cheaper per passenger mile.