20 Comments
User's avatar
blaine wishart's avatar

Great report. I’m glad your brain and gut are aligned as I want them to takeoff.

Expand full comment
Hugh Knowles's avatar

Looks great fun and as you say magical. What sometimes seems missing from conversations about the technology is any reservations about rapid scaling of automated point to point transit. Just because we can, technologically, does not necessarily mean it is all good for cities or people. Arent robotaxis inherently less efficient at moving large numbers of people in dense urban areas compared to mass transit systems and can potentially exacerbate transportation inequalities? Not that robotaxis arent part of it but as so often a case of attention and investment...we need a more holistic approach to urban transportation that prioritises efficient mass transit, active transportation modes, and thoughtful urban design over individualised car trips, whether human-driven or autonomous.

Expand full comment
Hugh Knowles's avatar

As Jarrett Walker wrote in 2016 when critiquing Elon Musk's vision "Musk assumes that public transit is an engineering problem, about vehicle design and technology. In fact, providing cost-effective and liberating transportation in cities requires solving a geometry problem, and he’s not even seeing it. What’s more, he’s repeating a very common delusion, one I hear all the time in urbanist and technology circles.

Musk’s vision is fine for low-density outer suburbia and rural areas. But when we get to dense cities, where big transit vehicles are carrying huge ridership, Musk’s vision is a disaster. That’s because it takes lots of people out of big transit vehicles and puts them into small ones, which increases the total number of vehicles on the road at any time. The technical measure of this is Vehicle Miles (or KM) Travelled (VMT)."

https://www.citymonitor.ai/analysis/problem-space-why-elon-musk-doesnt-understand-urban-geometry-2281/?cf-view

Expand full comment
Azeem Azhar's avatar

Explain how you plan to build mass transit and fund it in the context of public finances and political reality.

If you can’t, it is, honestly, just complaining.

Expand full comment
Azeem Azhar's avatar

And i would add, i was in a Waymo. when you step in it, it becomes so obvious that cars should be autonomous independent of any thinking about public transport policy.

Heck / i want my personal car to my autonomous as much as i love having an auto-choke. (perhaps more.)

Expand full comment
Hugh Knowles's avatar

Not sure that is fair. I think the technology is impressive and interesting but it is also reasonable to have reservations and think there are systemic issues without having to solve them at the same time.

Expand full comment
Azeem Azhar's avatar

But without a real solution, it is just a complaint.

How do you propose stopping firms investing in particular innovations, specially the stack of technologies for this one?

Expand full comment
Hugh Knowles's avatar

I disagree. It is reasonable to talk about the technology in context. Doesn't taking about it without some of that contribute to not enough attention being paid to other solutions? Perhaps a small contribution to public transit investment is to not perpetuate idea that AVs are without issue.

Expand full comment
Hugh Knowles's avatar

It would be interesting to know how much has been invested in AVs in last decade (Perplexity suggests $200bn but no idea if that is anywhere close to accurate).

Expand full comment
Azeem Azhar's avatar

that seems about right. i used to think this money could be better spent lobbying to redesign cities … but of course there is no mechanism to do that so it’s just a pipe dream

whereas as magical cars that drive themselves and get cheaper can actually reshape cities. technology creates policy space.

Expand full comment
Saurabh's avatar

Hi Azeem, it’s great that you tried the two experiences (tesla and Waymo) in quick succession. It helps in comparing.

I tried full self-driving FSD (supervised) for my commute to work, for a month. I noticed something about the technology and my attitude towards it. Here is the headline “full self driving needn’t be one standard generic driving style. For it to be successful, it needs to learn and mimic every driver’s individual driving style (of course net of any inaccuracies)”. Why? Drivers (at least those who are like me) prefer to drive in a certain way. When FSD doesn’t behave the same way as we do behind the wheel (e.g. too much lane change, breaking etc), we tend to not trust it and take over the wheel.

This might become a big barrier for many people and prevent them from “owning” FSD. However, this might be less of a problem when one is renting a ride, like you did.

However, this point might be less important if, in the future, we fully transition from owning cars to renting rides.

Expand full comment
Tom's avatar

I was at a talk by a local university professor about five years ago. He covered all the leading-edge tech, absolutely objectively - he’d seen it all. But when he got to Waymo, he was genuinely taken aback by how advanced it was. He had been to google and had two drives a few years apart, and his mind was blown.

Expand full comment
Georgi Katanov's avatar

Waymo is much better for people who suffer from motion sickness even in mild forms because it is a much smoother ride. There are small nice touches such as remembering that you don't want the music to be on after switching it off several times. I also experienced a mini-ludite attack at a traffic light by a guy who turned around both side mirrors (not that the car needed them). The operator called immediately and asked if I wanted to continue the ride so they are keeping an eye in addition to all the sensors. Still a Waymo managed to surprise me after it made a move that would have been considered aggressive by most people so there may be a nascent personality somewhere deep in it.

Expand full comment
James's avatar

Traffic appeared light. Do you think it would pass a driver Turing test?

Expand full comment
Azeem Azhar's avatar

It drives far better than a human driver. So to that extent it would fail the Turing test. it obviously isn’t a human.

Expand full comment
Christian Graham's avatar

Fantastic. As a non-driver, I want these in the UK asap.

Expand full comment
Jeff Balash's avatar

How do you communicate with the car if you need to, such as if you'll be a few minutes late or if it hasn't arrived where it needs to be? Thank you.

Expand full comment
Azeem Azhar's avatar

Via the app. it waits five minutes then goes - a bit like an uber. I haven’t had a problem with it getting lost.

Expand full comment
Erich Joachimsthaler's avatar

GM also was out with the Bolt before Tesla and then Musk paid attention to EV, could this be the case here as well. Google got plenty of other problems, ie ChatGPT Search

Expand full comment
Azeem Azhar's avatar

waymo is an independent subsidiary with substantial external backing.

Expand full comment