[clock_tower]

I think there are physical reasons why driverless cars will never be as good as trains: they're rubber-on-asphalt instead of steel-on-steel, and you need a lot more engines, space, and transportation of dead weight per unit of people moved -- and they'll probably still need parking, too. As near as I can tell, driverless cars are horseless-carriage thinking... but trains have no element of novelty to them.

[vectorjohn]

Driverless cars can take you between train stations and destinations. No magical form of train will come to your house or go to your office.

In fact, this is a perfect use case for driverless cars. The problem with trains everywhere is the wasted time not on the train and the million stops along the way.

[vidarh]

Ultimately I think driverless technology can revolutionarise high density public transport too:

It gives us exactingly detailed information on what journeys people need. The dataset will be immensely valuable in allowing the companies that sit on them to first start doing quasi-bus services:

Order a bunch of minibuses. During peak hours, offer an option: Wait for the next dedicated car to be free, or ride share with quicker availability and a discount. Limit detours strictly - there'll be plenty of "Follow road X and pick up 6 people on the way to station Y" type stretches that will make people happy (little time lost; feels efficient if there's not lots of turning off).

Then you can see them partnering with bus providers to dynamically fill in during peak hours, or even bidding for bus franchises and proposing contract changes that would allow for more dynamic, demand-based scheduling.

Ultimately this can feed into planning train type services - companies offering these type of drive share will be able to e.g. let people order "end to end" journeys of the type "pick me up at address 1, get me to address 2" where they show journey options that include rail when it makes sense. The key beying that if they do so, they will know the entire desired journey, and would be able to offer insight into the most efficient interchange locations or other changes to train services would be most desirable.

[vectorjohn]

I really want to be in the business of writing the software that plans these things. The possibilities are crazy. Predicting where to have cars based on overall traffic flows, having cars ready for individuals who always leave at the same time. Think, a car is 30 seconds away every time you follow your routine. That's faster than getting into and starting your own car! Especially if you have a parking garage.

[vidarh]

If you're predictable enough, it might be worth it to simply arrive a minute or two ahead of schedule.

The raving reviews you'd get the first times people order a car and are told instantly "your car is already waiting right outside your door" would be rather interesting.

Get in, and your favorite radio station or music is playing.

And for colder climates: Your car is already pre-heated. I remember too many winter mornings during winter in Norway when even ensuring our car would actually start in the morning was an annoyance (even with a garage, space heaters in the garage or motor heaters is sometimes necessary).

[clock_tower]

> No magical form of train will come to your house or go to your office.

Yes they will; the magic trains you're thinking of are called streetcars. And, with the oddities in oil and the difficulties of maintaining suburbia, we'll be moving back to dense streetcar suburbs soon enough.

[vectorjohn]

Even buses don't take you door to door. I can't imagine a future where a streetcar is anywhere near as good as self driving cars (possibly paired with mass transit).

[clock_tower]

Are you familiar with streetcar suburbs? They'd have a streetcar running along a commercial street, with residential streets branching off from it; typically one would have a walk of a few blocks from streetcar stop to house. I don't know if a return to that pattern is the future I expect, but it's certainly the one I'm hoping for.