To me this feels like this could be done immediately, and at least we would have self driving for the most monotonous part of the driving experience, long hours of highway driving.
(2) all the things that "FSD" vehicles need are things that degrade quickly to the point where they don't work for said "FSD", so the highway maintenance costs would skyrocket
(3) Even if you were to say the costs of 1&2 were reasonable, that money has to come from somewhere, and what that will essentially mean is that the already underfunded infrastructure will have even less funding for things that are the also already underfunded freeway things, so your normal streets end up in even worse condition, and based on existing spending the brunt of _that_ cost will be born by low income people. The cost for low income being exacerbated by their presumably cheaper and poorer condition cars taking more damage from poorly maintained roads, and so increasing their direct maintenance costs.
(4) Fundamentally single car travel is an inefficient way to move people around, it simply does not scale as well as good public transport, and always will be: the amount of people a road system can transport is directly proportional to the area of the road that each person takes up, and cars are just not efficient. The average car is 8m^2 and in commute traffic the vast majority of them are single occupant, the average bus is ~30m^2. If we were extremely optimistic and assume an average of 2 people per car you have 4m^2/person vs 1m^2/person in a bus. This is excluding following distances, which if we were to include (say) 2m of following distance you get a conservative 3.5-4m^2 of additional wasted space so lets just say your total for a car is maybe 5m^2 per person. I think :D
We could then get them to automatically travel close together to reduce air resistance.
And you could charge them while they drive.
Maybe have a special low friction surface for the wheels to reduce energy use and air and waterway pollution from tyre, road and brake dust. Without tyre wear and range being a weak point in the design they could travel at much faster speeds.
Also if they are travelling together anyway you could have one big one with a more efficient motor and join them to it and use the charging network directly rather than needing them all to have individual batteries. We could call it a tesla chain or tchain. With then coupled together you could just have one person in the big-engine car supervising the 'full self driving' for all of the cars. Doesn't really matter then that it's in beta because it's just one or two people for 700 passengers.
Then there wouldn't be so many constraints on the size and shape so you could basically make the cars a box to fit more people.
Then with our self driving car network coming close to any given point we could just let them stay on the highly efficient part and get off it near our destination. We'd no longer need to dedicate 50-70% of the land area to collectors, arterials, extra wide streets and parking. What little time you spend walking would be more than made up for by not having to find a park, higher density and higher speeds. What a world! You wouldn't even have to dedicate a quarter of your home's lot to setbacks, driveways and storage for your car!
If only we'd had elon's amazing genius in 1879 we could have had a whole network of these by the beginning of the 1950s. We wouldn't have had to destroy half of our cities to make room for highways and parking then choke on exhaust fumes and toxic dust for 70 years.
Efficient transportation in cities is communal, and built using tax dollars. That's not Silicon Valley sexiness.
IIRC, they had some very successful prototype road systems back in the 90s (or maybe earlier) that were based on implanting magnetic markers into the road at regular intervals. The big problem was cost. There's miles and miles of roads and the cost to upgrade and maintain enough miles was prohibitive.
With that said, perhaps now that the cars themselves are 90% of the way there, the upgrades to the roadways could be more limited. I wonder if adding machine-readable markers to particularly important or problematic areas might be reasonable at this point.