You start in the south where the weather is easy and work your way north. I think the hope is that by the time they get to Manhattan, global warming will have improved the weather.

It isn't just weather. Phoenix is probably the closest thing we have to a perfect city for this tech. In addition to the clear and precipitation free weather, its geographic location gives it the most daylight hours of any major city in the country. The roads are almost all big, straight, flat, relatively new, and easy to navigate. There is not much of a pedestrian culture. Drivers are not super aggressive like they are in some other highly congested cities. There is almost no native wildlife that would pose any threat to a car. And I am sure there are other things I am missing.

>There is not much of a pedestrian culture.

Arizona has the highest rate of pedestrian deaths in the US. Pedestrians have apparently been trained to keep off the streets.

It's also not immediately clear if this will be in the city of Phoenix itself. The earlier testing was in the suburb of Chandler.

The weather is the least of your worries in a place like Manhattan.

Pretty large chunks of the country are a lot like phoenix suburbs.

Indeed although many have much more inclement weather than Phoenix. To be clear, this is good progress. But expansion may be slower than many think.

And also lots that are an incremental step away. Gotta start somewhere.

A non significant chunk of trips are in places that don't look like them. If I still need a driven car for 10 or 20 percent of trips, I'm not going to ditch in favor of driverless cars.

It's a service, like Uber, you're not expected to ditch the car at this stage

If it gets cheap enough I at least could take the self driving car to the nearest train stop and go downtown via train. She got grocery shopping in the suburbs. However, shopping be better only a single stop. Maybe we could solve that problem though by having some kind of bin that gets automatically stored at a grocery store while I'm shopping, loaded with my groceries and then put into the next self driving car...?

Of course it's not the same, but it's also not incredibly easy. I drove around Phoenix for about 7 hours recently. Definitely not the easiest roads, nor the most considerate drivers. I'm curious to see how these cars will figure out all the weird lane closures, double-lane turn-right-on-red, blind intersections, etc. I almost got creamed turning left near the Heard museum, and there was a bunch of confusing construction downtown. And the highway merging can be... exciting. Not saying it's the most difficult driving in the states, but it's also not the easiest.

Manhattan also has a robust public transit system, making it feasible for a person without a car to get around. Driverless taxis are probably disproportionately likely to be used in suburbs and other places without that infrastructure.

Yet, that isn't where taxis are primarily used today. Why would driverless taxis be much different? Maybe they'll be a little cheaper (though Uber drivers are probably only making minimum wage--if that) but it's not clear why that would transform behavior.

> Yet, that isn't where taxis are primarily used today.

Huh? Taxis are for sure used way more in NYC than out in the suburbs. They benefit significantly from density. Out in the suburbs you can't even catch a taxi unless you call one to you.

The popular way of getting around in the suburbs is driving your own car. In cities mass transit and taxis are much more common, especially because parking is so expensive that it makes more financial sense to hail a taxi for the occasional trip.

That’s what I was trying to say. Dense urban areas are where taxis are used the most and those will be hardest for self driving.

Indeed. It's less urban areas that have poor public transport that need this the most. The bay area for example is extremely hard to get around without a car, but something like Seattle? Very different.

Environments like phoenix are also where cars are the most useful.