[baldfat]

In the near age of driverless cars I think this might be its biggest demise. If cars are automated people wouldn't need a train. It will be cheaper and less confining to a schedule and to a destination.

I think air travel for short to medium flights will fall off. By the time I leave my house and get to someplace that is 8 hours by car it would take 5 hours to check in and check out and you still need to rent a car.

[phinnaeus]

Autonomous vehicles do not obviate grade-separated transit options. Even if we imagine a road network populated only by highly efficient and inter-connected autonomous vehicles, the increased capacity allowed in such a situation will only create more demand, which will scale up traffic problems.

Now, other than traffic, I agree. When deciding whether to drive somewhere or take the train, my second factor is whether or not I feel like actually driving a car. That answer is usually no, so I'm very excited about driverless cars.

[vidarh]

The real benefit, in my opinion, from autonomous vehicles is that they remove a major barrier for a lot of people to not owning one. By taking drivers out of the equation, you can cut cost or increase vehicle availability or both to the extent that being able to get vehicles urgently, or cheaper, depending on need, can give most people in densely populated areas an experience close to as good as owning a car. Better if multiple people need to go different places.

And while you're right that this will probably drive up demand, it achieves something else too: Availability of extensive route information. Now imagine a shared system for such cars to negotiate routing, and such cars can know which routes will be less congested, and can know which cares it is most effective to re-route because of where they're going.

E.g. there's a chokehold near me for Southbound traffic that it's tremendously beneficial to bypass if you're heading South-East, because you can take back-roads with lower speed limits but that are shorter, and that will end up being much faster when the chokehold is congested. Which it "always" is between certain hours. But if you're going South West, it's better to suffer through it.

With vehicles able to negotiate route information, vehicles going South East could be automatically re-routed when there's congestion, resulting in average speed improving for everyone.

In fact, you could even imagine a system where cars autonomously negotiate a fee for other cars to let it get in front (so you could e.g. pay a 30% premium for urgent trips and have the car company spend, say 25% to buy its way in front of others on congested routes).

This may not entirely solve the problem, but it can make congestion more about total throughput than maximizing throughput on a small number of routes.

End-to-end route information also opens the door for services to mediate multi-step routes with e.g. cars feeding public transport much more efficiently (e.g. in my case I live almost exactly halfway between two major rail stations, with two more small ones not much further away. I almost always opt for the largest of the four because it's at a central junction that gives me the best option for almost every destination so I rarely need to think about it, and it also means I know exactly how long it takes me to get there.

But if I can tell an app where I want to go, and get told to either go to the bus stop or wait for a car, get taken to whichever station is the best option and told which track/train to take, and get picked up by a car on the other end, there might be substantial time savings and often a better shot at a seat. The only thing stopping me from "shopping around" like that today is that it's extremely inconvenient.

Again combine that with pervasive end-to-end route information, and you can make it even better by ensuring not too many people get routed onto each train if two options are relatively equal and one is less loaded.