That's my point actually. You can't predict that, like you could predict the sudden demise of Diesel only 3 years ago.

3 years ago you could probably have done exactly the same reasoning with electric cars (eg: London 0-emission goal are for 2050). And I mean, if my country decided to go full electric in 10 years, I have no idea how I could own a car, unless charging speed, or street infrastructure, or millage vastly improved. Nothing of that has a clear 10 years roadmap. Yet here we are.

I think it will happen as soon as it's good enough. The incredible death rate of non-automated cars will make things happen faster than normal, imo.

Beyond gutnor's reply, cars are just fundamentally ripe for disruption. The benefits of personalized arbitrary-point-to-point mechanized transport are so compelling that society has been willing to put up with an enormous price for it, but that doesn't mean the current situation isn't still terrible and ready for a real tipping point effect the instant it becomes feasible. The only reason we've had manual driving at all was because humanity's development of mechanical tech greatly outstripped its development of information capture and processing tech, but that was always a temporary situation. Think about the incredible pressures on the current system:

- Well over a million deaths a year worldwide, if we're talking America then 30k-40k+ here alone. Tens of millions disabled/injured, at ruinous expense. Even more accidents that "merely" cause significant damage, again at major expense.

- The huge insurance and societal costs necessary to sustain the above.

- People who cannot drive in a society where that's immensely limiting. This sandwiches both sides of demographics. Particularly in the First World, demographic trends are resulting in an ever increasing number of elderly who cannot drive any longer. And it's always been a hassle and disruption for parents to need to shuttle around children, or arrange for the same, or just do without.

- A fear, justified or not, that terrorists will make increasing use of the fact that cars are an excellent way to deliver a tremendous amount of energy into a set of targets even without any sort of purpose-made weapons onboard and are not something that can be done without.

- The immense waste in incredibly valuable land resources needed for overbuilding parking, wide lanes, and other support infrastructure needed for human driven vehicles.

- The immense waste in human time and stress due to traffic and other conditions that could be ameliorated or even eliminated by automatic control.

Etc etc etc. The principle problem with both your arguments 1 & 2 is that you make them in a vacuum outside of the massive negatives that we live with right now. As with many disruptive new technologies, self-driving cars don't have to live up to any particular ideal, they merely need to be better then what we've got. And frankly that's just not a very high bar. It's not just economics it's politics, if you look through the above list you should be able to easily identify some of the most powerful interest groups and societal motivators in existence, touching almost all of society in general. Once the avalanche gets going feedbacks are going to reinforce pretty rapidly.

I don't dismiss the benefits of automated driving on society, literally everything you said is true and I agree with it 100% - I'm just saying that I feel like the tech isn't anywhere near ready for mass deployment, it feels like a gimmick or a really fancy adaptive cruise control at best, or that people who talk about it come from a Star Wars universe where being surrounded by super high-tech stuff is just normal and indistinguishable from magic.

Seeing the current state of the best-of-the-best from Tesla and Deimler, autonomous driving is decent, but it seems full of the same issues as image recognition - and that was predicted to take at most few years to perfect 50 years ago. Today in 2017 you can fool the best image recognition algorithm with a sofa in a zebra print - it comes with 90% confidence match for a zebra because the pattern is right. I firmly stand by my feeling that getting to anything approaching L5 autonomy will take half a century, and that's not near future in my mind.

Imagine in the middle of every American interstate a light-weight, elevated track holding up a lane in each direction designed for automated cars. Tech today could easily drive vehicles in such a controlled space at high speed and density. Switch to manual upon exit. At 100mph and 40ft per car one gets a rate of 13200 cars per hour. This is on par with the capacity of the current 5 lane Bay Bridge at 9,000 per hour [1]. This could be the future, if society could embrace it. (Electric?) automatic cars would be adopted very quickly if you got to use such a system for commuting and cross country travel. As driving tech got better cars could travel right next to each other on long trips to get the aerodynamic benefits of trains. Google, Apple - you have billions sitting around. How about building a demo track and start building the system on the 280 and 101 for a nice loop.

[1] Wish I had a better reference (http://www.sfgate.com/bayarea/matier-ross/article/Peak-hour-...)