It's around 10 times the cost of a car with a driver.

Driverless cars will not be as cheap as cars which require drivers.

There will be ongoing operational and maintenance costs which cars with drivers do not have.

Liability insurance will likely be lower for driverless cars, but damage and theft insurance will likely be higher.

Not every cab needs to be operated 24/7, there are clear peak and non-peak times which allow for fewer than #Cabs x 5 drivers for full practical coverage.

And, of course, there's no guarantee that any savings in operation would be passed on to the consumer (see electronic publishing).

So I don't think it's automatic to say driverless cars will result in cheaper transportation costs. It's possible , but it's far from guaranteed.

>And, of course, there's no guarantee that any savings in operation would be passed on to the consumer (see electronic publishing).

In this case it's likely that they would be. Book publishing is a rather unique case in that it's a million natural monopolies; if you want to read The Hunger Games, you're not going to be satisfied with a close substitute. Taxi services are fairly commoditized- excluding medallions, pretty much any automated taxi to show up at your door and bring you where you need to go is going to be acceptable. Thus you've got a much higher elasticity of demand thanks to this substitutability, so consumers capture more of the surplus. There's a low barrier to entry, so consumers can and will just switch to any cheaper option that comes along.

There's no reason to assume the number of driverless cars won't be artificially restricted by regulation in the same way that taxi medallions are now.

Maybe so, but I'm struggling to imagine a scenario where that comes true. As we generally think of them now, they're for the suburban commuter to replace their current car. As such, they can't be limited much, lest only a tiny elite be able to purchase them- probably not tenable politically. Furthermore, politicians don't seem farsighted enough to even set quotas if they wanted to; any limits set would likely be at something resembling estimated real-world demand, and since we're talking about a shift that would radically reduce the total number of cars in service, politicians would likely overestimate the need.

Plus, driverless cars can't form taxi associations with reasonably large voter blocs.

Yeah, I don't mean driverless cars as personal property would be limited. Rather driverless cars as a service will likely be regulated much like taxis are now. Corporations owning the cars and transporting people for money, that will be regulated and restricted.

Labor is by far the largest cost here.

I have a pretty fair knowledge of the Google car and the Stanford cars that came before it. The equipment at scale isn't really that expensive and can be fairly easily retrofitted onto existing cars.

Even if it ends up costing 50k and an extra 20k per year in maintenance (which is probably too high for a computer and a few actuators and sensors) taking driver labor out of the equation results will result in vast cost savings as a percentage of total operation costs.

>theft insurance will likely be higher

Theft will be nearly impossible for an average criminal. Remove human controls and you've instantly defeated 99% of car thieves.

But as I mentioned, 5 drivers per car is overestimating considerably the actual number of drivers you need for practical coverage. You don't need your full fleet 24/7, the demand just isn't there for it around the clock.

Call it three drivers for practical coverage. Now we're at $120k for labor (3 x40k). But we've added 50k to operations in other areas. So now we're saving 70k in overall costs. Which is good, but a long way from 200k.

If you just needed 1 driver for the car, then there's a net loss with the driverless version (with these theoretical numbers anyway).

Perhaps the most efficient method would be a base fleet of driverless cars that operated 24/7, with peak times adding in cheaper cars with human drivers....