[greenyoda]

"Almost-free on-demand point-to-point public transit. Combine UberPool and self-driving cars and this is what you get. It could be actually free with ads..."

I don't see how self-driving car services could be "almost free". Self-driving cars cost money to manufacture. Like all mechanical equipment, they have finite lifetimes, so their purchase cost needs to be paid back over that lifetime. They need to be maintained, e.g., tires and shocks don't last forever, and metal corrodes over time (especially in environments where roads are salted for de-icing). Someone will need to clean the interiors every day (or more often), since people tend to leave messes (sometimes really disgusting ones).

The power for them isn't free. Even if they're solar powered, solar panels have to be manufactured, installed and maintained (e.g., cleaned off periodically to maintain maximum efficiency), and they need to be replaced after their finite lifetimes expire. Rechargeable batteries need to be replaced after a certain number of recharge cycles.

Roads, bridges and tunnels cost as much to maintain for self-driving cars as for conventional cars. There will still be tolls and taxes. As gas and diesel vehicles die off, the revenue from the taxes on their fuel will need to be replaced by other taxes, such as per-mile taxes on electric vehicles.

I doesn't seem that merely showing ads to passengers could pay for all these capital, labor and tax expenses. And if the cars are owned by a for-profit company such as Uber, there has to be enough income for the company to have a profit after paying their operating costs and taxes.

[hugh4]

Let's make some generous assumptions -- a taxi costs $20K, lasts 300,000 miles, gets 50 miles to the gallon and fuel costs $2.00 a gallon. (Or equivalent, electrically, I'm easy.) Then fuel is four cents a mile and depreciation is seven cents a mile for a total cost of eleven cents a mile.

In the future, running an ultra-efficient electric car we can imagine the fuel price could be lower and the range before replacement even higher.

Is eleven cents a mile "almost free"? It's certainly a lot cheaper than the vast majority of public transport trips I've taken in recent years. Could it be ad-supported? Well, maybe. How much would advertisers be willing to pay to captively monopolise the attention of someone whose individual tastes and spending habits were already well profiled? Quite possibly more than eleven cents a mile (a few cents a minute in traffic).

(Personally I'm more than willing to outbid the advertisers in order to get peace and quiet.)

[jkldotio]

The advertising is going to be extremely targeted too. If person is going to the shopping mall to see an optometrist then the stores at the same mall could bid to pitch to them. They would literally be showing their ads to someone who is about to be a short walk away and trump even Amazon on how quickly a product can be in the hands of the consumer: "press purchase now and we'll have it waiting for you at the front desk for pickup". Doing a late evening airport to hotel journey? Ads for takeaway food to be delivered to your room or ads for tours and events in that city. Not on the same level of annoying as audio ads for something I don't want playing while I am trying to read an article. I am sure they will figure out a way to be extremely annoying though; I can easily imagine poorly targeted ones that play automatically and so loud you can't have a conversation in the taxi.

[petra]

This is pretty close. Benedict evans did a more accurate calculation , and got ,for electric self driving cars, 14-19 cents/mile:

https://twitter.com/benedictevans/status/636684338352951297

One thing he forgot - what about shared transport - say sharing with 3 people could offer ride time relatively close to a car. That gets us to 4.5 cents/mile. That comes about to $50 a month/american-person(avg 13.5K miles/year) , less for a kid, surely less for people in dense cities(say $25) .

EDIT:and if you're willing to ride an on demand bus/minibus , maybe it could go to 1/2-1/3 of that , so maybe your monthly transport demands could be met by less than $10.

[mdorazio]

I did my own calculations last week and came to about 2x his numbers (https://www.linkedin.com/pulse/how-cheap-can-uber-self-drivi...). He makes a few faulty assumptions (ex. maintenance costs and depreciation are tied to miles, not years) and in my opinion underestimates costs significantly in areas like insurance, and over-estimates billable miles that will be driven per day. Still amazingly cheap, though, especially if carpooling and/or using public transit part of the time. For comparison, CalTrain runs between 17 and 25 cents per mile if you take it more than one zone.

[hugh4]

Sharing rides with strangers is precisely the kind of thing that I'd hope to avoid by replacing public transport with self-driving cars.

But anyway, yes, it does appear that the economics of the self-driving taxi beat the economics of the bus or the train, especially given that buses and trains generally operate at a large loss and still charge people, on average, a higher fare than what they'd pay for a self-driving taxi.

A self-driving bus that operates a route could be even cheaper, but why bother when you can hail a cab to take you straight to your destination without stops for just a few tens of cents more?

[petra]

I can imagine a self-driving taxi , for shared rides, that's designed for privacy , so you don't feel like riding with others, except for a bit of an extended trip.

I wonder why nobody have build such a car yet, but there's good likelihood it will happen in a huge self driving market.

[kps]

When all cars are self-driving, small single-person cars are a realistic option.