For everyone who has not read it: Aaron is not proposing the future of ground-based transportation. This is an analysis why he scraps his plans for a magnet-levitation, personal transportation system and a call for new ideas.
To clarify, there is an urban density limit on this conclusion:
"Time to get to a station (walking, or drive plus park) kills the effectiveness of most personal rapid transit and light-rail systems until you have a population density of 5000+ ppl/km2."
So for population densities of 5,000+ ppl/km2 this might work (emphasis on might).
Interestingly, urban areas of the developing world might qualify:
The problem is that many of these cities might not maintain their high densities as their economies grow. My guess is that a lot of the high density is due to extended families living together. But maybe not, there are quite a few developed world cities on the list (e.g. London, Madrid, Seoul, Tokyo).
Unfortunately, I cannot vouch for the reliability of the statistics presented here, there just a result of a very quick google search. I've got to stop spending so much time on HN.
5000 people per square kilometer is only 20 people per acre, and any standard rowhouse neighborhood has 14 lots per acre, so this is only 1.4 people per house, a much lower density than you would get with large families or with multistory apartment buildings. A lot of cities that fail to meet this density are only because their legal limits also include a lot of essentially undeveloped land as well.
The overall trend worldwide is toward higher urban density, not less. This is more so in the developing countries - it's unlikely to swing away from this. As long as population is growing, the cities tend to grow (and become more dense) with them.
That is correct in the aggregate: people are moving from sparse rural areas to denser metropolitan areas. However, the urban density of metropolitan areas themselves is decreasing in the developed world. Most of the growth occurs in relatively sparse suburbs.
But, also, in the detailed part of the proposal, he points out that with high densities (e.g. NY city) you need so many maglev cars to meet peak demand, that parking them off-peak is a significant problem.
Yes, but they could probably be automatically parked in tightly packed parking lots far out of town. As peak approaches, they could drive themselves back into town.
Because the customer doesn't require a specific car, these parking lots can be FIFO queues or FILO stacks and won't require maneuvering space.
Also, the cars could transition to a much cheaper, non-maglev parking system. This might even be done using a robot to lift and store the car into a slot in an otherwise "dead" grid. One expensive robot that can move in 2 dimensions and a lot of simple steel rails is a lot cheaper than heaps of maglev track.
Finally, the 333k car estimate was based on a subway where there is only one place that people are leaving from. There was an assumption that EVERY car must return EMPTY to the source of the trip afterwards to pick up the next passenger. This is not true, the percentage of returning cars is probably much lower (30%?) And even those cars could be useful for part of this trip.
