You're not really making great counterarguments when you are pointing to Wikipedia articles and saying "hey so Japan is kinda lewd so you're wrong" - it borders on insulting. What makes you think I don't know that already? I've been studying transportation for the past few years, and that includes the transport economics that you're only scratching the surface of. I spent 59.5 hours in the best PT system outside of Japan this month alone. By all means disagree or challenge my assumptions - I don't know everything - but don't misinterpret my argument and then point me to Wikipedia when you're on my turf.

I wasn't saying mass transit was loathed by everyone (although there is plenty of evidence that people prefer driving most of the time). I was talking about transfers. Grouped transportation (PT, airlines) offer little to no door-to-door routes because there are incredible economies of scale when you group people that sorta go in the same direction. Look at the airline industry: what routes don't they fly? Well, there's no route from Amsterdam to, say, Huntsville Alamaba. But there are plenty of planes going across the Atlantic and already planes going from Huntsville to Atlanta, so with a stop in Atlanta you can just join the crowd and get there for a reasonable price. Hub-and-spoke is incredibly efficient, because it can reduce the number of routes you service by more than an order of magnitude while still serving everyone. And this hub-and-spoke model increases the number of passengers on the links you do serve, which means you have enough people to not lose all your money flying planes.

BUT it entails transfers, which are systematically disliked by everyone (yes, even in Japan). And the routes are almost always indirect and meandering. My internship is south from where I live. But instead of going south, I bike 10 minutes west to the train station, take a train that goes south but also to the east, and then take a subway that goes southwest to end up at my destination. If I would drive directly I'd always be faster, simply because it would be a more direct route. Which is what I was trying to argue: a taxi-service is small scale and direct and will beat out a fixed-line PT system from a service point of view, AV or not.

We're not all in taxis and Ubers though, because as you point out, the cost also matters. Here's a fun fact: on average, owning a car in the US costs $0.40 per kilometer, including taxes and maintenance but excluding parking. AV's, when they aren't owned by individuals but can be summoned like an Uber, are expected to cost one-fifth of that. Yes, in no small part because of large scale vehicle production and maintenance. But even if that is only achieved partially, the business case for AV-based taxi service is enormous because the largest expense (the humans) gets thrown out of the window. Imagine if you can get anywhere in your city for a dollar or two, faster and with more comfort than public transportation or driving. That is what efficiency is really about: reductions in cost, labor, energy and time that add up to such a significant change from what we have now that it's worth looking out for.

Maybe it will pick up a few people along the way, maybe you pay extra to be left alone, who knows. A lot of the talking points in here are clouded by uncertainties, including what I just said. For example, it is very likely that automated driving in complex urban environments (as opposed to the much easier problem of highway driving) is decades away. If that's the case, fixed long-distance highway drives might be the largest benefit in the short term: automated freight trucking and automated Greyhound / Flixbus instead of automated Uber. As I said to wasoxygen earlier this week...it depends. It all depends on a lot of factors.