[mcv]

It's not feasible to have a bus stop right in front of every house. It's unavoidable that most people are going to have to walk a bit. How far is reasonable, is a matter of trade-offs. It also depends on how fine grained the network is. If there are buslines every block, it's annoying if they don't stop there. But you have to walk a block or two to get to a bus line anyway, walking that bit more to get to the stop itself, matters a lot less.

[JoshTriplett]

> It's not feasible to have a bus stop right in front of every house.

And this is why point-to-point transportation is almost always faster and more convenient, if you can afford to use it. (That load-bearing "if" is important, though.)

[xmprt]

> And this is why point-to-point transportation is almost always faster and more convenient

Point-to-point transportation is faster and more convenient because:

1. we don't have bus lanes so buses are forced to sit in the same traffic as cars and 2. buses are often underfunded so have slow/infrequent service.

Point to point transportation is often slower and less convenient if buses and public transit is done right. I can count on my fingers the number of times I used an Uber or drove a car in the 1 month that I stayed in Europe - this was going out every day, in multiple cities, rural and urban, and across different countries.

This is a good thing! If more people use public transit when it's possible, it opens up the roads for the handful of people who actually NEED to use a car.

[AnthonyMouse]

Bus lanes still seem like the thing people who hate cars propose to intentionally screw over the people who have them. "Hey, we have this road with two or three lanes in each direction but it's fairly congested. Each of the lanes is carrying something like 50 cars per minute during the day! Why don't we impound one of them so we can have a bus carrying 40 people drive on it once every 15 minutes?"

If you have enough density to justify a bus lane, you have enough density to justify a subway.

[Karrot_Kream]

> If you have enough density to support a bus lane, you have enough density to support a subway.

Not at all. Building a subway in most US cities right now is very expensive. Raising the tax revenue alone is probably a non-starter.

Moreover you're going to have to close the road down anyway to do any form of cut-and-cover or even deep bore construction, which means every business on the corridor and every person who lives on it is going to get angry for as long as the subway is being built.

There's no painless way to do infill public transport. The problem is that nobody in the US is willing to compromise.

[AnthonyMouse]

> Building a subway in most US cities right now is very expensive.

This is true but seems like a problem worth solving. It's also true of more than subways; we have the same problem with bridges, housing and many other things. Better to get on with fixing it than use it as an excuse for doing something worse.

> Moreover you're going to have to close the road down anyway

That's a one-time cost, and you're not required to close a 500 mile stretch of road for years on end. Dig one block, install the tunnel, cover it, dig the next block.

[Karrot_Kream]

I agree with you (and importantly you can't make a subway political football the way you can make a bus lane), but my experience doing transit advocacy points otherwise. Americans in dense areas are feeling the HCOL pinch and are not very willing to float extra taxes to fund transit expansion.

IMO it comes back to the fact that Americans are just not willing to accept change of any kind right now. The economy feels too shaky, the electorate too divided (even within states and municipalities), and there's too little faith in government to architect the kind of change you'd need to build subways, underground metros, or even BRT. We need a larger feeling of unity even at a state level to make the changes necessary, which is why municipalities continue to do bare minimum maintenance of roadways and pretty much nothing else. The last big set of constriction in dense urban areas was funded by the Obama stimulus from the GFC which was passed 17 years ago.

[jltsiren]

> If you have enough density to justify a bus lane, you have enough density to justify a subway.

That assumes a linear city, where everyone lives within a short walking distance of the same street.

In actual cities, bus lines from different neighborhoods converge on main streets. While individual lines may have 10–15 minute intervals, bus traffic on the main streets may be high enough to justify dedicated bus lanes.

Then, as the city grows, it can make sense to replace the bus lanes with light rail and direct bus lines with collector lines connecting to the rail line. Which should be cheap, as a dedicated lane is usually the most expensive part in building light rail.

But you generally want to avoid building subways until you have no other options left. Subway lines tend to be an order of magnitude more expensive than light rail lines. Travel times are also often higher, as the distances between stops are longer and there is more walking involved.

[AnthonyMouse]

> That assumes a linear city, where everyone lives within a short walking distance of the same street.

Isn't that the assumption you're making? That there is a single primary street that everything converges and then diverges from which is common to every bus route? Meanwhile in practice any given person standing on the You Are Here dot could want to go in any of the eight directions from where they currently are.

A route that goes east-west isn't going to have much in the way of shared route with one that goes northeast-southwest except for the one point where they intersect, and isn't it better to have multiple routes intersecting in multiple places in terms of minimizing trip latency and maximizing coverage?

> Which should be cheap, as a dedicated lane is usually the most expensive part in building light rail.

But that's the thing that makes the bus lane so expensive!

By the time you have an area with enough congestion to be considering a bus lane, the problem is generally that you can't add a lane because the land adjacent to the existing road is already developed and not available, otherwise you would just add an ordinary lane that buses could use too. But converting one of the existing lanes in an area which is already congested makes the traffic exponentially worse than putting the new thing underground.

Essentially, if you can add a lane then you add an ordinary lane and if you can't add a lane but need one then it's time to dig.

[jltsiren]

Public transit depends on the assumption that some trips are more common than others. If any given person is equally likely to go to any direction, public transit becomes too expensive to build. And it becomes impossible to make the city dense without turning the traffic into a nightmare.

A typical direct bus line starts from somewhere, goes through a number of neighborhoods, reaches a major street, and follows it to a central location. The number of directions that need a bus line is typically much higher than the number of streets reaching the central location. (For example, you need ~10-degree intervals at 10 km from the center to guarantee a reasonable walking distance to the nearest bus stop.) Hence the bus lines eventually converge.

Once you have enough bus traffic that a dedicated lane makes sense, transforming an ordinary lane into a bus lane will make the traffic faster for the average person. It's not a Pareto improvement, as the traffic will become worse for those who drive on that route. But it's not a huge loss for them either. If you already have 20+ buses/hour making frequent stops during the rush hour, the throughput for that lane will already be much lower than for the other lanes.

[hardolaf]

Heavy rail and light rail costs are very comparable unless you want to bury them. But it doesn't matter which you bury, they still cost about the same.

[pocksuppet]

We did that with computer networks. We had this high-quality voice call service, and then someone thought it should be switched to transmit data instead, of which voice calls were just one type. Now you have a minimum voice latency of a few hundred ms because voice traffic is competing with data traffic, and you didn't actually get much more data throughput because it was only one wire pair.

[JoshTriplett]

> Point to point transportation is often slower and less convenient if buses and public transit is done right.

Only if you're also intentionally making point-to-point worse.

Note that I'm not comparing to "get in your own car and drive", which has the disadvantage of having to park. I'm comparing the ideal taxi-shaped thing to the ideal bus-and-tram-and-train-shaped thing.

[xmprt]

> Only if you're also intentionally making point-to-point worse

I feel like you missed my last paragraph. If public transit is better then more people would use it and there would be fewer cars on the road. Can you imagine how terrible point-to-point traffic in SF would be if everyone was driving to work instead of relying on Caltrain or BART?

[JoshTriplett]

No, I didn't miss it. I'm saying that public transit can be better than it currently is, but it would take much more to make it better than point-to-point transit.

[njarboe]

Self driving cars for hire (Waymo, Tesla, others) can be that point-to-point system that is affordable. We will just have to build tunnels to deal with the increase in traffic. Hopefully the Boring Company or someone else can get tunneling costs way down.

[llbbdd]

I generally agree that self-driving cars are going to take this niche, but not with tunnels. Tunnels add the same dedicated infrastructure problems as mass public transit.

I'd suspect most car trips today are 1 or 2 passengers with the back seat and trunk empty; we'll eventually see new form factors of on-demand vehicle that trim off unneeded space. If you need to get from A to B alone, no cargo to speak of, you order a ride that covers that class and it's small. If you're taking a shuttle from the airport with your whole family and luggage, you order a ride with those specs.

[mulmen]

As long as one of those points is a transit stop then yeah, robotaxis make sense. In that model you don’t need the tunnels.

They make even more sense if they are a bit larger and can accommodate multiple people at once. Something like a large van or small bus.

[janalsncm]

If you are not being facetious, what you are describing is closer to a subway system, which has the disadvantage of being very expensive.

[JoshTriplett]

Hopefully someone else, so it actually happens and isn't overpromised and underdelivered.

(Also, tunnels are useful not just for the increase in traffic, but for moving car traffic away from non-car traffic, which makes both kinds of traffic safer, faster, and more efficient.)

[cyberax]

No, it's not unavoidable. Just ditch the buses and switch to cars, soon to be self-driving.

Even the rush hour traffic is trivially solved by mild carpooling (small vans for 4-6 people).

[wussboy]

Not Just Bikes makes a compelling argument that self driving cars are not the answer, and will almost certainly make things worse

[jacobgkau]

Would you care to summarize their argument?

[mcv]

Self-driving cars still take up space on the road. Even more than human-driven cars, because now there will also be cars transporting 0 people. It's going to make congestion worse. Public transit is the solution to congestion. Well, one of the solutions, because bikes are probably a better solution for most people: they do start in front of your home, can park anywhere, and don't cause congestion the way cars do.

We're talking about cities, of course; in rural areas, nothing beats cars.

[cyberax]

> Self-driving cars still take up space on the road

This is a false argument. Think about this: a bus every 10 minutes is effectively 500-900 meters long! It easily "takes" as much space as 100+ cars. In other words, nothing would change from the traffic perspective if instead of 1 bus every 5 minutes, you had 100 individual cars.

The "people in the shape of a bus" argument makes sense only when you're talking about the performance in a very narrow case of transporting people in a steady, uninterrupted stream of buses. Or if you need to size your traffic bottlenecks.

Moreover, a bus route necessarily is unoptimal for at least some people on a bus. They are effectively "thicker" than other people because they take up more "effective space". But wait, there's more! Buses also necessarily move slower due to stops, so the "effective length" of a bus becomes even longer because cars will clear the road faster.

But wait, there's even more! A single bus needs about 3 drivers to be effective. So with the average daily busload of around 15 people, you have almost 20% of the bus taken by the drivers on average. This makes bus trips pretty expensive. Not quite to the level of Uber/Lyft, but surprisingly close.

And these problems are fundamental. That's why urbanists like NJB don't like to talk about that.

[mcv]

Nothing about this addresses NJB's argument that self-driving cars take up more space than regular cars, because there will now be cars with 0 people in them.

Ultimately the thing you want to transport is not cars, it's people. Walking fits the most people in a limited amount of space, then bikes, buses and other forms of public transport, then cars with 4 people in them, then cars with 3 people in them, then cars with 2 people in them, then cars with only 1 person in them, and finally empty cars. More cars will never reduce congestion.

But to address your point: A bus in a dedicated lane takes up more space than a bus that's stuck in car traffic, you are right about that. On the other hand, when congestion is so bad that cars simply don't move, no matter how many lanes they have, getting people out of cars into more efficient forms of transport, will also help cars. And a bus that actually goes, can do that. If you look in cities with good public transport, more people go by public transport than by car. In cities with good bicycle infrastructure, more people go by bike than by car. That means even cars are less likely to get stuck in traffic in those cities. Even if you take away a car lane.

I don't know where you got the idea that a bus needs 3 drivers.

[yunwal]

> a bus every 10 minutes is effectively 500-900 meters long!

uhhhhhh what. What does every 10 minutes have to do with this at all

> It easily "takes" as much space as 100+ cars.

are you ok??? have you seen a bus before??

> A single bus needs about 3 drivers to be effective

I have never ever seen a bus with 3 drivers in it. If you're talking about 3 drivers over the course of 24h, those drivers are not in the bus at the same time, and therefore don't make up 20% of the passengers on the bus. If you're saying the average bus route serves 15 people per day, you are certainly mistaken.

[ErroneousBosh]

> We're talking about cities, of course; in rural areas, nothing beats cars.

Where I grew up in NW Scotland, it's a five hour round trip to go to the supermarket. You pretty much need a car for that.

Where I live right now it's a five minute walk to the supermarket, but I still need a car because the things I work on are a long way from where I live, often up steep muddy mountain tracks.

When I lived in the middle of Glasgow people used to come up and have a go at me about driving a massive V8 4x4 in the middle of a city. What am I supposed to do with it? Bike to the suburbs and then go and drive up a mountain?

"But why not get a job where you don't need to drive hundreds of miles in a massive 4x4?"

Because then the things on the tops of mountains don't get fixed when they break, and the radios don't work properly, and then people like you die in a fire.

Sometimes it's hard for people to grasp that just because their not-really-a-job tapping numbers into an Excel spreadsheet all day can be done from home or from an easily walkable city centre location, it doesn't mean that everyone's job looks like that.

I do wish I could usefully use a cargo bike. Those things are awesome.

[sensanaty]

What people really misunderstand in these discussions is that no one is talking about completely killing off driving as an option, and no one says that public transportation works in literally 100% of circumstances.

We just want there to be viable public transportation options for situations where it makes sense. This even makes it easier for the people who do have to drive, like you, as there will be less congestion because a single bus can replace literally dozens of cars, combine that with a single tram and a single metro car and you're replacing literally hundreds of cars that would otherwise be on the roads instead.