[cozzyd]

Such a system would either be very low throughput or just as bogged down as the roads. Individual itineraries imply low-occupancy vehicles (because how many people from your starting point will go to the same out of thousands of destinations?).

If you assume an optimistic 4 people per vehicle with an optimistic 5 second interval between vehicles (which is probably not safe at the high speeds due to the frequent merges), that gives you ~50 people / minute / lane, or 3000 per hour. A single CTA 8-car train at crush capacity (admittedly uncomfortable) can carry nearly 1000 people.

Now a frequent system with high occupancy vehicles and transfers could work, but that's the same thing as a subway.

[tokipin]

The key is that this system can operate in 3D. You can have bundles of tunnels where you need higher throughput, and greater demand can just be met with more tunnels. The focus on small tunnels/cars as the core technology is part of this, since it makes it easier to reach "mass production" scales.

[cozzyd]

So to first order, you'd have to build as many tunnels as there are road lanes to get the same order of magnitude throughput? I get that there is more space underground, but this sounds completely cost ineffective given that many municipalities can't afford to maintain their roads.

[Firadeoclus]

5 second interval seems excessively large if you compare it to car traffic, and that's with human drivers.

[cozzyd]

Yeah but the cars aren't moving at $LUDICROUS_SPEED and aren't dealing with constant merges from feeder tubes.