I don't understand the "2 trips for one passenger". The car will pick up the closest human that needs a ride and the operator will optimize the trips globally b

I don't understand the "2 trips for one passenger". The car will pick up the closest human that needs a ride and the operator will optimize the trips globally based on where cars are, where humans are and where humans what to go. It's about as optimal as it gets. Unless commute patterns change dramatically, in the morning most people will be going downtown to jobs, there will be a few that live in downtown and go to jobs elsewhere, but by and large, all of the cars will be bringing in commuters from outside of down and dropping them off in downtown centers -- there will be no other human looking for a ride from that car. Without changing cars, we can put 4 people in a car, which is 2-4x better than the common commute of 1 person in a car. We can do that today, but we don't. It's not clear why a self-driving car would change that. How often do people take uber-pool instead of having a car to themselves? (I don't know the answer to that question). People don't want to share cars? Maybe they won't want to, but as autonomous cars become more popular, only the affluent will make that choice - a car is expensive, especially when it sits in a garage at home and work (that you pay for) unused for 22 out of every 24 hours.