Light rail versus self-driving cars

This is Joseph

Consider:

“We are definitely going to have pushback,” said Brad Templeton, a longtime Silicon Valley software architect who preaches the potential of “robocars.” (He believes the subway paved over in concrete for autonomous vehicles could transport more passengers than rail can.) “I regularly run into people who even when they see the efficiency numbers just believe there is something pure and good about riding together, that it must be the right answer.”

and

“Don’t build a light rail system now. Please, please, please, please don’t,” said Frank Chen, a partner with the venture capital firm Andreessen Horowitz. “We don’t understand the economics of self-driving cars because we haven’t experienced them yet. Let’s see how it plays out.”

Now I typically love Brad Templeton's analysis.  Look at this great article, for example. But I do think there are some issues with this approach.  I certainly don't think my like for public transit is about being with other people -- that is actually one of the major downsides of a light rail system is dealing with other passengers who are sick or unaware of personal hygiene standards.  There are other issues.

One is the geometry one:

Highways today can carry about 2,000 cars per lane per hour. Autonomous vehicles might quadruple that. The best rail systems can carry more than 50,000 passengers per lane per hour. They move the most people, using the least space. No technology can overcome that geometry, said Jarrett Walker, a Portland-based transportation consultant. 

Which is brought up in the same article.

Another issue is the average age of cars on the road, which right now is 11.5 years. This means, in the absence of subsidies or very aggressive regulation, it is going to be something like 30 years before we could have all self-driving cars.  Many of the efficiency gains come from the self driving cars being the dominant type of car on the road.

Also, cars that make 2 trips (to pick up and then return) are going to use more road space that a car that is driven and then parked.  Now it might be the case that you can get it down to a reasonably small percentage of total miles -- technology is cool.  But that also weighs into the geometry argument.

So I hope that self-driving vehicles are coming.  But I don't see that they necessarily solve issues of congestion -- you make driving more pleasant and people will be willing to do more of it.  They might radically improve safety (which would be awesome) and they might make commuting time much, much more pleasant.  These are goods, in and of themselves, which don't need a pure robocar future to harvest.

But it seems like a tough gamble to stop current transportation planning until the new technology is mature enough that my type of objections can be answered.