Report from first person to give up their car for a robotaxi

A family riding in a Waymo One Robotaxi

I'm back from another electric car road trip -- more later on that -- but here's a story where I provide a report from a Waymo One user on how he sold one of his family's two cars and replaced it with robotaxi service. He's an early adopter, but he helps us examine just what some of the issues are around getting people to do that.

Read my new Forbes article at:

Giving up car ownership for Waymo Robotaxi


For a successful robotaxi service, you need to become more than a taxi to your customers, if you build your own self-driving car software. For a company that buys its self-driving cars from someone else, who makes cars (and other vehicles) for everyone, there's plenty of room be a successful service.

What I mean when I say you must be more than a taxi is that being just a taxi -- a supplemental service -- you will never be a huge service. Once you own a car, you have already spent a lot of money, and you have a strong incentive to use it for everything it can be used for, even times when it is not the best choice.

If you don't own a car, then you are going to use other transport for all your needs, and so you will make a lot more use of your chosen taxi service. That's how it makes a lot of money. Whether it builds its vehicles or not.

I'm not sure what you count as a huge service. Uber's gross bookings in 2018 were $50 billion. That's pretty huge, in my opinion.

I don't think the economics are going to work out for large robotaxi companies. Not unless governments ban people from hiring out their personally owned robocar (think Uber like it is today, with cars owned by individuals, except without the need for the car owner to actually drive or even be in the vehicle). So in that sense I guess you're right. But not because the market isn't there, but because it doesn't make economic sense to add a middleman where none is required.

Once it becomes possible to hire out your personally owned robocar, there will likely be lots of successful small robotaxi companies. Taxi service will remain a supplemental service. But it will grow in some areas (replacing some mass transit uses, replacing car ownership for people who drive infrequently, possibly transporting unaccompanied minors although a personally-owned robocar can handle that too) and it will shrink in some other areas (no need to hire a taxi to go out for a night of drinking, or to drive to/from a location with hefty parking fees or inconvenient parking such as the airport or stadium or casino or Manhattan). Looking at the most popular destinations for Uber, it'll probably shrink more than it grows.

Personally, I'll probably go from one car to two or three robocars. One regular-sized robocar, and one or two tiny robocars. I'll commute to work daily in a tiny robocar. I'll use either the second tiny robocar (if I have one) or the regular-sized robocar (if I don't) to take the kids to school daily. (Whether or not to have two tiny robocars or one one will depend on a cost-benefit analysis of the lower operating cost vs. the lower capital cost, factoring in the ability to hire out the robocars when not in use.) The regular-sized robocar I'll use whenever I need the space, either for people or for stuff. My tiny commute vehicle can pick up my lunch when I'm at work. Presumably most places that offer take-out will also have curbside delivery to my driverless robocar (I can remotely roll down the windows for them or give them a code to let them do it themselves). If vehicle-to-grid becomes a reality I'll also be able to subsidize the cost of the extra capital by renting out the battery to someone who needs it. If the cost of the car is mostly by-the-mile (it'll be interesting to see how long it is before car software is paid for by the mile), that four-seater's battery might get a lot of use as a Powerwall-on-wheels.

There are two strong reasons for a middleman. The first is that they are not a middleman -- they are a company like Waymo or Cruise or Uber who actually makes the self-driving software and designs the car. They have all the advantages of vertical integration.

The second is the consistency of service and single point of control that comes from a brand. Even if you do end up getting somebody's private BMW when you call Uber, Uber will take steps to assure the quality is maintained, the vehicle is clean, and that the BMW owner is booted out of their network if they don't live up to it. And if something goes wrong, you depend on Uber to fix it and get you there, not the random guy who owns the BMW. You want that middleman.

It is not clear to me why you would not use a taxi service to get your kids to school though, rather than buying another small vehicle for it. And can a car designed for people really compete for efficiency on delivery with either a Starship Robot, or a UDelv with 20 different people's orders in the van?

Yes, Uber is big (but a tiny piece of $5T in car transportation) but the point is that Uber went far beyond what the taxi industry was. They found a way to bring far more people into using taxi service than ever did before.

The tiny car (something like the Electra Meccanica SOLO, although the price would need to come down quite a bit) would serve multiple purposes. Yes, it would free up my regular sedan for taking the kids to/from school. But it also would cut the per-mile costs (in dollars and in pollution) of any time I needed to drive alone. Commutes, running errands, etc. It'd probably be a win, if the price were low enough, and if they charged for the self-driving software by the mile the price would probably be pretty low. If the cars could be rented out at least some of the time when I'm not using them, it'd be an even bigger win.

Taxi service to get the kids to school isn't going to be cheap. That's peak demand.

As far as using my car vs a delivery van, it'll probably be about the same cost, and the cost will be so small anyway. Again you're talking about a peak time. If they set the price lower than the cost of using a personal robocar, they'd have so much demand that they'd have to buy vehicles that just sit around all day waiting for peak demand times. It'd be less efficient than people just reusing what they already have.

Yes, a company like Uber will probably oversee the operation, for quality control. But I don't think it makes sense for them to own the cars. It's basically the same economic argument we've been having for months. Commutes tend to be in one direction, and it's too inefficient to double back with an empty car in the opposite direction. So to handle commutes with a robotaxi you almost have to have one for every commuter anyway. At which point they might as well own that vehicle.

One plus is that more and more cities have bidirectional commutes these days, but they also have a pattern where cars can be out in the burbs, come into the city during the day, act as city taxis all day, take commuters home at 5pm and be suburban taxis at night. Other cars stay in the city all day, of course, and others say in the burbs all day. But as people move, so does the need for off-peak rides in the places they go.

One part of his story stood out to me.
"Our mass transit here in Phoenix has the best of intentions, but it is ineffective if you are on a tight schedule"
My experiences with travelling, showed that in unfamiliar environments you need to leave extra time for orientation.
Even with smart phone navigation, going from one mode of transportation to another can be a pain for reasons other than infrequent/mismatched scheduling. A good example being large train and bus stations. Finding your way to the pickup point in a short period of time can be quite stressful. Even if smartphone navigation within a transport complex is available it's not a great interface for users who are unfamiliar with it.
Augmented reality directions via something like google glass would be much superior for a couple of reasons, no fumbling with orienting the device to match reality, no issues with multiple floors.
Even in familiar environments augmented reality could be useful for seamless transitioning by highlighting the actual vehicle within a carpark.

I have found that with a good app, foreign transit gets very good. Yes, not as good as for a native who goes through the station every day, but the good apps tell you exactly what door to use to exit the station, they tell you which end of the platform to stand on to make your transfer easy etc. The only thing I have yet to see is good underground localization.

However, none of that will fix the usability of transit in a town like Phoenix that does not have a critical mass of transit users.

Agreed, I really wasn't referring to Phoenix specifically, just making a comment on how seamless transitions could be aided by AR.
Any recommendations about a good general travel app, or does it differ country to country?

It does differ. I think it was citymapper I used in Paris which knew the Metro really well.

Sorry, comment was supposed to be reply to original article not to FKA's comment.

Add new comment