48 Comments
User's avatar
Jason Paris's avatar

There are certainly some downsides, but considering how many regional rail stations—and even some subway stops—sit in seas of parking with few nearby amenities or housing, I can see this helping with that “last kilometre” or two. It could even help more suburban residents go car-free, since many keep a car mainly to reach the GO station because they’d rather avoid taking the bus.

Myrtonos's avatar

Yes, cheaper taxis could help more suburban residents go without owning their own cars and make it easier to get to and from GO stations and get around suburbia without a driver's license.

Calvin P's avatar

I'm one of those people who doesn't have a car and uses Uber occasionally. I see Waymos starting to test in Boston and I can't wait for them to go live and scale up. Uber drivers are some of the worst drivers on the road, and will sometimes do things like make you wait half an hour because they want you to cancel so they can get an airport fare.

Reese's avatar

I've tried Waymo a couple of times out of curiosity, and it was far smoother and safer than any Uber ride I've ever had. I once had an Uber driver announce that when he wasn't driving for Uber, he was drinking because his wife left him. This was unsurprising given he was erratically changing multiple lanes while people honked and I feared for my life.

Calvin P's avatar

Last time I was in an Uber, the driver ran a red light and almost t boned someone making a left turn.

Reese's avatar
Mar 6Edited

The institutional lag in responding to emergent technologies increasingly grates on me as the pace of change accelerates. That said, this post is a breath of fresh air, because any time I read anything related to “AI,” I feel like I'm taking crazy pills. LLMs, self-driving paradigms, and other applied statistical methods are only becoming more effective at pattern recognition and automating tasks. Labor displacement, surveillance, and transit anxieties are well-justified as a result, but here’s what isn’t: the regurgitated dismissal that it’s all just fake.

Thanks for writing this, because the usual vacuum of nuance on these subjects is, I think, extremely harmful... the polarization diminishes the likelihood and magnitude of any reasonable institutional response.

As you identified, there’s a tendency in which someone either thinks they can bend reality to their will with rhetoric, à la Umberto Eco’s concept of painting the “enemy” as simultaneously too strong and too weak, or is a full-on techbro accelerationist who wants to replace the entire consumer economy with agents trading through their own crypto wallets.

Anyone with a measured perspective who doesn’t assume they already know everything is left to watch in horror, lol.

Reece's avatar

I am writing a very long post on my other blog about more or less this exactly, AI, and nuance

Justin Gurr's avatar

Do you think the same tech could be deployed on buses to make it more feasible to increase frequency and coverage?

Dollyflopper's avatar

Not only can it but it is being deployed. The problem is Waymo today is rolling out $150K cars while a manufacturer of those mini robo buses is hoping to finally build an actual factory and roll them out for 3 to 4 times that. That talk is that the next generation of Waymo's will bring the per vehicle cost down to $100K.

Reece's avatar

Well, buses will obviously cost more, but thats ok because they move more people and last longer. If anything this will reduce the cost of buses relative to cars!

Myrtonos's avatar

I wonder how useful robotaxis could be in less populated areas where transit services that can be supported are less frequent.

Jon Boyd's avatar

Here is a suggestion for a big idea: we need to shape technology so that technology does not shape us. Morphine and heroin are wonder drugs for pain, but we need to limit their use according to strict rules. There is no question that the ICEV is a great technology, but the way we implemented the automobile in North America is highly destructive. Robotaxis could just result in a new form of car abuse.

Myrtonos's avatar

However, transit is fundamentally a way for a single operator to move more people at a time than each operator could with a more individual vehicle, like a car, but this means (even in Europe and Japan) that not all services are super frequent, it means people (especially those who cannot turn a steering wheel) moving to be near transportation rather than transportation moving to be near people and even in great transit cities this means the service is not normally door-to-door.

If self-driving makes running smaller vehicles more frequently more widely viable, it will allow more frequent service in less populated areas, and makes it easier to get around those areas (and at night also in cities) without a driver's license, to some extent this will be so even in Europe and Japan.

Brad Templeton (founder of the first ever internet based business) has written about this on templetons.com & 4brad.com .

Jon Boyd's avatar

AVs are not a solution at-scale.

Myrtonos's avatar

What do you mean "at-scale"?

Jon Boyd's avatar

Five percent of the population traveling in cars is fine; 95 percent of the population driving is urban suicide. Robotaxis do not scale any better than people driving their own cars.

Reece's avatar

I disagree. Shared assets are dramatically more efficient and scale much better.

Jon Boyd's avatar

Walking, cycling, and public transportation. If you have actual expertise on managing fleets, please share.

Myrtonos's avatar

Still confusing. I said that robotaxis would make it easier to get around less populated areas without a driver's license and might, even in many cities, make it easier to get around at night without a driver's license.

Jon Boyd's avatar

They don't scale in rural areas, either, and they are still the most expensive form of transportation.

EDIT: If you think that AVs would help at the margins, they could, but one of the problems of cars is their spatial requirements, and AVs don't change that. In the simplest terms: WE HAVE TOO MANY CARS MAKING TOO ANY TRIPS.

alapan's avatar

Why can’t robotaxis just be considered part of the transit ecosystem? Private transport providers (buses, mini buses, etc) are available across the world, including countries with other forms of public transit. There are also private operators of smaller, shared transportation - eg auto rickshaws in India, collectivos in South America - that operate almost like taxis; and some even operate as cheaper forms of taxis. Robotaxis can just be yet another service; hopefully integrated to rail etc

Myrtonos's avatar

> There are also private operators of smaller, shared transportation

- eg auto rickshaws in India, collectivos in South America

- that operate almost like taxis; and some even operate as cheaper forms of taxis.

You don't get this (so much) in the developed world.

Vehicles like cybertaxis and robovans will make it easier to get around less populated areas without a driver's license.

alapan's avatar

You do get private bus operators, both intercity and municipal. And in many European countries, taxis are well integrated into the suburban rail network - so you can get off the train and get a bus or a taxi. Shared taxis are less common - but taxi drivers often accommodate; and Uber has shared cab rides in many countries also.

Reece's avatar

I don't think I'd equate the existence of taxis with integration of them into the system

Myrtonos's avatar

Yes but jitneys are not so common in the developed world.

If cheaper taxis (such as robotaxis) are a threat to transit, they could also be a threat to transit oriented development.

Dollyflopper's avatar

Can Waymo's compete on price? Seems unlikely unless they can cut the cost of their cars by another 70, 80%.

Kyle Bogosian's avatar

Cars are objectively more economically and environmentally efficient than mass transit when there is a group of 3+ travelers per car. Transit fares are already subsidized to cover more than half of internal costs; if you allow 3 people to ride for the price of 1 then you are basically giving away free tickets. There’s no reason to bother competing in these cases, you are offering a product that is worse in every way. Just focus on doing what mass transit does best, which is moving lone travelers.

Reece's avatar

Thats not true at all. The cost of roads is immense. Not dissimilar from mass transit and you need much more of them.

Kyle Bogosian's avatar

America's annual road spending is $250 billion for 3 trillion miles of driving, that's only $0.08 per vehicle mile! I know there are other reasons people don't like having cars (congestion in cities), but the infrastructure seems very economical.

Reece's avatar

I'm not confident in your numbers and I think it's pretty straightforward that public transport is a lot more cost-effective. 250 billion is an absolutely enormous number, and I'm not sure I would judge the outcome based on vehicle miles.

Reece's avatar

And I'm not saying this because I like transit or whatever, I quite earnestly believe it's much more cost effective when you look at the total system cost.

Dollyflopper's avatar

The driver for an Uber is not the most expensive part of the cost of the ride, that's the cost of the car. In the case of Uber, they offload that cost onto the driver. People see that drivers bring in about 60% of the fare and make the mistake of thinking the labor is 60% of the cost, it's not. That 60% includes money that is for the car.

Reece's avatar

That labor isn't more of the fare is just Uber being predatory. Its a huge portion of the cost and fare. Cars can be had relatively cheaply when renting for very short periods of time.

tian zhao's avatar

Honestly, the bus sucks as an experience. In North America we treat it mostly as charity. It's loud, every pothole shakes your brain and the windows, youre jammed up against someone else. It only works because it's cheap and convenient in Toronto, and funnels you to a service actually worth taking.

Even compared to a tram/streetcar the rider experience is majorly improved simply by being on rails.

We should just build cheap elevated automated Skytrains along each major arterial road and get bikes / escooters/ self driving cars to achieve that last mile.

The bus is basically outdated, and I'm kinda glad if the robotaxi kills it.

Reece's avatar

This is hyperbolic. Some of the buses are good.

More importantly though, you don't seem to be thinking at all about capital costs - buses are cheap, trams are not.

Myrtonos's avatar

Brad Templeton (founder of the first ever internet based business) has written about this on templetons.com & 4brad.com .