Hacker News Reader

>Planes are entirely valid solution to the problem of traveling between LA and SF

The problem is that LA and SF-area airports are almost at capacity and increasing that capacity through adding runways or building new airports would be a significant undertaking (several 10s of $billions) that would face a number of hurdles. Air travel is also affected by weather issues that don't affect HSR.

And air travel is less efficient, noisier, and dependent on fossil fuels than gliding over a maglev track.

As far as I know, the High Speed Rail system is not maglev and modern trains have roughly them same energy efficiency as modern planes.

That's interesting. I'd instinctively presumed that HSR would be much more energy efficient than flying, but the quick research I did indicates you are right. Electric HSR probably has the edge if you are willing to assume hydroelectric generation, but it mostly comes down to projected ridership. A full plane beats and empty train, and vice versa.

Actually, the robotic car will make for very cheap cabs that can be much smaller. Most will only need to carry a passenger and can have a much smaller engine without the need for speed that human drivers have.

So, smaller vehicles almost always moving and not parking will take up a lot less room per person-mile.

Why would it need a smaller engine? The load and cargo of the car is identical.

(i.e. one passenger, and one driver weigh exactly the same.)

The assumption is that autonomous cars would decrease the cost of taxi-style usage to the point where it competes with ownership, since you wouldn't have to pay for a driver.

When I buy a (non-autonomous) car, I size it for the maximum load I might want to carry. For the autonomous taxi, I can just rent the minimally-sized car for the trip that I'm currently on. If I'm by myself, the autonomous cab could even be a one-seater. Plus, I don't have to park it.

I still don't get what the difference autonomous makes. You can rent a car today. If one seaters were practical they would be rentable. Autonomous isn't going to change that.

Autonomous sounds great - but not because of car or engine size.

An autonomous single seater comes to you where you need it. A single seat, non-autonomous rental car doesn't come to you unless there is a driver, or a delivery vehicle with a driver, to bring it to you. The economics aren't going to be the same.

Its simply too costly to get the optimal size cab to a customer today. So cabs are usually much bigger than the average car. Flexibility is more important.

And cabs make up a small percentage of the overall market. And most of the cost of a cab ride is labor.

But if automated cabs become the norm, the cost of a cab ride will be governed by the size of the vehicle. And if a car costs 25 cents a mile to operate, a cab ride from my house to San Francisco will still cost me $11.

If a 1-seater with barely an engine and no trunk can do it for 10 cents a mile or $4.50, there will be demand. That's the commuter, the guy making 40 trips a month, who'd save $260 a month for giving up empty space.

Of course, we really don't know how this will all play out. But it will be fun to find out.

Automation will do the same thing for all kinds of things we rent.

Right now, the costs of finding items at the best price, going to get them and the manual paperwork involved, often make owning them easier.

But once all that is automated and you can rent the exact drill you need for a dollar an hour and it can be there in 15 minutes, why would you buy it?

And, yes, it will be delivered by a robotic vehicle, probably no bigger than a breadbox.

A robotic cab doesn't NEED a smaller engine. It can get by with a much smaller engine.

A robotic cab doesn't need a driver. Or a front seat for him. Or a powerful engine to suit his lead foot.

A personal robotic car will be the same size. But a lot fewer people will need them when they can call cabs for less money per mile. Especially in urban areas where storing cars is a big expense.

Frankly, I don't see how public transportation will be able to compete. Its too slow and inconvenient and completely unavailable after certain times.

In other words, the bullet train is doubling down on old technology that is highly inflexible and a big time waster.

But, hey, maybe people like going to Bakersfield.

Chicago isn't land constrained, but has a dense urban core (roughly comparable in overall density to London) and has great public transportation, both within the city and within the larger metro area (http://upload.wikimedia.org/wikipedia/commons/7/77/Metra-Sys...).

Proper urban planning can go a long way.

"Proper" urban planning is helped tremendously when the city was planned and reached critical mass in the days before the automobile and freeway systems really took hold.

> Whether the cars are automated or not won't drastically change that because there are still more cars than the roads can handle.

Automated taxi cabs + existing public transit would solve gridlock... go outside to your auto-cab, take it to the rail, get off the rail into a different auto-cab, arrive at work.

> But that still doesn't address the problem facing most large cities: gridlock. Whether the cars are automated or not won't drastically change that (although it may well reduce arterial traffic clogs due to accidents and rubbernecking) because there are still more cars than the roads can handle.

Humans are going to be a lot less efficient than computers in driving cars. With proper networking, you could have cars doing 60mph while only a few car lengths from each other, vastly increasing the capacity of any segment of road.

For example, the presence of an unexpected crash or obstruction down the road can be relayed back at the speed of light via networked cars, and the entire wave of cars behind the crash can slow down and reroute.

There have been some simulation papers on this, comparing humans vs. automatic driving algorithms. It's similar to automatic control in other areas.

>With proper networking, you could have cars doing 60mph while only a few car lengths from each other, vastly increasing the capacity of any segment of road.

That works if 100% of the cars are autonmous. I'm betting that even if automous cars were available tomorrow, it would take longer than 20 years for that to be true.

I think there's a threshold beyond which it becomes socially mandatory for all cars to be autonomous. The pressures will be enormous at a certain point. Endangering the lives of others when there's an option to not do so, etc etc.

What that threshold percentage is and how the transition will be accomplished in practical terms, I don't know. But I do believe the pressures will manifest themselves mightily.

>I think there's a threshold beyond which it becomes socially mandatory for all cars to be autonomous.

Economics are still going to be the driving factor (no pun intended) for the poor and working class. As long as beater non-automated cars are going to cost less than the (used but still newer) automated cars, and as long as cars are still going to be necessary in much of the USA, there are still going to be normal cars around.

At some point (say in 40-50 years, and if "classic" driver cars are outlawed) I think we'll reach that point, but then that weakens the argument of autonomous cars (and its benefits) getting here before the CA HSR project.

I would imagine it would start with restricted lanes just for autonomous cars - I would be fine doing the start and end of the drive, and letting the car take over for the long distance stretch.

I think you are right, there will be a tipping point at some stage. For something that is so thoroughly baked into the American way of life, I think 20 years isnt a bad estimate.

I'm not totally up on autonomous car tech, but while very promising, it still seems to be proof of concept stage. To me that means ~10 years from commercializarion, and then at least 10 years before it has penetrated enough of the fleet that you can even start to consider making it mandatory. Automotive engineering is conservative because of the scale and age of the industry, the cost of the goods sold, and the risks inherent in their failure. Automobile owners are conservative because cars are big purchases.

The thing about timeframes like that is that there are enough unknowns that estimates can easily be 2x too optimistic.

I tend to agree that rail (whether urban light rail, or region high speed rail) may not be the most cost effective way to move a lot of people from point a to point b given existing patterns of development, I know that part of the reason planners favor it is that its cost comes comes as part of a permanence that can shape the way communities develop around it over the long term.