[npunt]

I appreciate that argument and in some/many situations it is likely true, but in others it is not.

What matters isn't people per square foot, it's throughput. And the larger the vehicle, the more likely:

a) people are to want to stop at each stop,

b) multiple people are to get on and off at each stop,

c) the less likely it is to be full, and

d) the more awkward and slow it is in maneuvering on streets.

A/B/D all delay every other passenger, and make 3 mile trips take half an hour through a city.

Cruise Origin isn't the only autonomous bus. The great thing about autonomy is that it allows us more degrees of freedom to optimize transportation needs including offering a variety of shapes and sizes and densities, while removing the labor cost and the physical space cost of a driver.

You can just as easily design an autonomous bus to have density to match a larger bus while retaining the footprint of smaller vehicles, which improves all of the above issues with larger busses.

[jonstewart]

Bus dwell time plays a nonzero role in bus performance, but many multilane urban streets are choked with cars during rush hour. It’s often the case that congestion accounts for the lions share of bus transit time, yet there are typically more people on the bus than in all the cars on a given block. Creating a dedicated bus lane can dramatically improve bus performance, and has a follow-on effect that since congestion has been mitigated, the same set of buses (a fixed capital cost) can, in the same timeframe (fixed operating cost), make more circuit trips. So not only does a single trip get much faster, a bus lane magically produces more capacity.

[kbenson]

The Origin looks van sized, which is perfect. It doesn't take up significantly more space than a car while moving or stationary but can carry more people (and more importantly for congestion, likely will much of the time). Averaging three people per car/van sized vehicle would likely solve the vast majority of current congestion needs, as it would likely halve the vehicles on the road.

[Animats]

The Origin looks van sized, which is perfect.

Remember Supershuttle? They had vans. "Never more than 3 stops". Remember the long, long indirect routes of Supershuttle? Remember what happened to Supershuttle?

[usrusr]

Lack of network effect, perhaps compounded by inadequate routing (I have no idea if they were as good as possible or much worse than necessary).

Network effect must be huge for a system like that. Imagine a large fraction of the cars on the road was shared in the way of supershuttle, you'd already have in that pool a near-perfect itinerary to tie into for almost any trip. And the remainder could easily be fulfilled by assigning a new trip. If you just have a few cars and price for shared occupancy tours will inevitably be much worse. But once you have a critical mass network, route inefficiency will be just a load factor price/performance tradeoff like in a hash-map.

[npunt]

On the other hand, Lyft Line and Uber Pool seem quite efficient.

Supershuttle was concentrated pickup but distributed dropoff, whereas with network effects leading to more vehicles & considerably more efficient routing, you can even that equation out a bit more.

[kbenson]

I was thinking along the lines of alternating between a distributed pickup and concentrated drop off and concentrated pickup and distributed drop off. That is, you use these for last mile travel at either end of a commute artery.

Given there are many different bus/rail stops along a route (and many routes for buses), there's a relatively small geographic area to pick up in and drop of to when the other side of the trip is bus/rail. So you might have an automated van drive through an area and make a quick 4-6 pickups in a few block area, drive to a bus stop or rail station, drop those people off, and pick up 4-8 people for drop-off in close geographic proximity. Rinse and repeat.

The point doesn't have to be that they solve the last mile by replacing current mass transit, but by supplementing it in a way that allows for people that were far enough away that it was hard to use previously now have an easy and cheap way to do so, because you've expanded the coverage area of mass transit stations.

[jonstewart]

Yes, but. It's only perfect if there aren't other cars.

[willyt]

You are right, a bus trip I used to take regularly in London would take 15mins if you got the first bus at 5am, which was quite busy with shift workers. At 9am it would take 45mins with a similar level of crowding. A good portion of the journey was on bus lanes, but there were certain pinch points that killed the journey time in rush hour.

[saberdancer]

Throughput is an incredibly important one, but something that seems to get little attention from the tech world at least as far as I can tell. Is there work being done to improve throughput of normal cars at intersections, or at least semi autonomous ones. I am talking about the accordion effect that happens at each intersection/red light that there is. If a whole column of cars could start moving at the same moment, and then spread out with increase in speed it would result in an improved throughput at critical points in the cities.

Is there work in that area? I feel it shouldn't be a massive technical challenge, cars are stationary, lights are visible and in the future could even "talk" to each other to pass along critical information (if the car infront needs to brake for example).

[npunt]

Agree, those are the longer term efficiencies when we have primarily / only autonomous vehicles on the road. At that point we can optimize from the system POV not just from the vehicle POV. But that's a ways off.

[saberdancer]

But there is no reason why it couldn't be done today, even with non-autonomous vehicles. There are cars on the road that can automatically brake if they discover that you are incapacitated, or that something jumped out infront of you. Having some kind of "assisted red light start" which reduces accordion effect could be an easy way of improving throughput. At first only some cars would have it, and effect would be smaller from a system POV, but with time it would propagate.