Hacker News Reader

Comment by cyberax | Hacker News Reader

Sigh. Sprawl is _CHEAPER_ than dense cities for most infrastructure.

To give you a perspective, one mile of NYC subway now costs about as much as 1000 miles of 6-lane freeway. One mile of new light rail in Seattle will be enough to maintain all roads in Washington state for half a year.

The same story for water, sewer, and electricity. Dense cities are _incredibly_ expensive to maintain because projects become extremely complicated, with lots of project management overhead (more people-hours). And people are EXPENSIVE, compared to raw materials.

View on HN

Light rail is ~$27 million/mile and subway is ~$290 million/mile, if you look at the numbers for the latest finished projects in Finland. American infrastructure is unreasonably expensive because, for a number of reasons, Americans lack the ability and interest to build it efficiently.

Dense cities are common around the world. In most of them, the infrastructure is not particularly expensive.

> Light rail is ~$27 million/mile and subway is ~$290 million/mile, if you look at the numbers for the latest finished projects in Finland.

Finland is tiny. Even Helsinki is a bit less dense than Seattle, and almost 10 times less dense than Manhattan.

So yep, sparse cities are much cheaper and efficient than dense cities. That's exactly my point. The sparser, the better.

Your comparison is based on administrative borders rather than on the cities themselves. If you compare population densities in urban areas (as listed in Wikipedia), Seattle has 3600 people / square mile, Helsinki has 4800 people / square mile, and NYC has 6000 people / square mile.

Typical Helsinki consists of medium-density residential areas separated by large parks and forested areas. The urban core is dense but not particularly tall. The kind of suburban sprawl you often see in American cities, with massive areas with a low population density but little undeveloped land, is largely missing.

> Your comparison is based on administrative borders rather than on the cities themselves.

"Urban areas" are poorly defined in many cases. In some cases (e.g. "Greater Houston Area") they match what people think about "Houston", Seattle metro area does not.

> The kind of suburban sprawl you often see in American cities, with massive areas with a low population density but little undeveloped land, is largely missing.

Yes, and? Helsinki is still sparsely populated, so new infrastructure is cheap enough. The costs go up, once the density increases.

Administrative borders are arbitrary and rarely have anything to do with the actual borders of the city.

Population density figures should be reported as averages over people rather than averages over land. They are usually not, because it's difficult to calculate them that way. If you have a residential area with 20k people in a square mile next to an undeveloped square mile, the population density is usually reported as 10k people / square mile. But from the perspective of the people living there, as well as for the purposes of infrastructure construction, the actual population density is closer to 20k people / square mile.

Re: expensive. Hard disagree. To put things in technology terms roads don't scale to meet commuters needs. And as others have pointed out it is a unit cost problem with your reasoning (total cost per commute).

At some scale we aggregate processing and trade latency for bandwidth (busy poll versus hardware interrupts). We trade the ability for everyone to start their journey _right now_ for the ability for lots more of us to get to our destination via a common carriage (be this packets or people the pattern is the same).

Where land is cheap and commuter traffic isn't an issue maybe roads make sense but that doesn't mean sprawl is cheap in time-spent.

This isn’t a true cost analysis: NYC subway lines can simultaneously be more expensive in absolute terms and cheaper per commuter-mile-year.

The amortized cost of a commuter journey on the subway is a tiny fraction of the cost of and on roads, for all parties.

(This is true even when the subway is obscenely expensive to construct, as the 2nd Avenue line is. But treating deep-tunnel construction under some of the densest and most expensive real estate in the US as the norm is essentially dishonest.)

> This isn’t a true cost analysis: NYC subway lines can simultaneously be more expensive in absolute terms and cheaper per commuter-mile-year.

It's not.

> The amortized cost of a commuter journey on the subway is a tiny fraction of the cost of and on roads, for all parties.

Nope. Roads in sparse cities are cheaper on per-commute and the building cost.

> But treating deep-tunnel construction under some of the densest and most expensive real estate in the US as the norm is essentially dishonest.

Why? Ever-rising costs is the inevitable result of densification.

The NYC subway serves 1.8 billion rides in a down year, comprising tens of billions of passenger miles. Amortized over decades of 24/7 operation, there’s no road in the US that comes close to that, even with obscenely expensive deep tunneling.

Roads are cheap on paper, and expensive per commuter-mile. Mass transit is expensive on paper, and cheap per commuter-mile, especially in dense cities.

Do you think scary big numbers are significant in itself?

Well, Houston carries about the same number of commuters. At a fraction of the cost.

If the argument is about sustainable transportation, then yes: big numbers matter. Scale matters, and mass transit is the definition of scale.

I genuinely don’t believe that Houston carries “about the same number of commuters” as the NYC subway. It has a quarter as many people; you should cite a source. I can’t find one for that.

No. Scary big numbers don't matter by themselves without a context.

1.8 billion rides per year for a 8-million city works out to less than 0.7 rides per day per capita. I can guarantee that Houston has more road trips.

> Scale matters, and mass transit is the definition of scale.

Yes, and multi-dozen billion dollar corruption on public projects is also the definition of scale.

> I genuinely don’t believe that Houston carries “about the same number of commuters” as the NYC subway. It has a quarter as many people; you should cite a source. I can’t find one for that.

I'm using the data for the Greater Houston Area (pop: 7.2 million), which matches what people think when they say "Houston".

> 1.8 billion rides per year for a 8-million city works out to less than 0.7 rides per day per capita. I can guarantee that Houston has more road trips.

That's just the subway. Add another 100 million for MNRR and LIRR, plus another 100 million for PATH and NJT. And those are all post-COVID numbers; you can roughly double each to get an actual picture for the volume of mass transit needed to run NYC.

> I'm using the data for the Greater Houston Area (pop: 7.2 million), which matches what people think when they say "Houston".

The NYC Combined Statistical area has 23 million people in it[1]. Nearly as many people commute into just Manhattan as live in the actual city of Houston[2]. You simply do not get those numbers with road-based transit.

It doesn't matter how you cut this: Houston is operating at a fractional scale, and will continue to do so as long as roads dominate its development patterns. Houston's government appears to be aware of this[3].

[1]: https://en.wikipedia.org/wiki/New_York_metropolitan_area

[2]: https://wagner.nyu.edu/files/rudincenter/dynamic_pop_manhatt...

[3]: https://en.wikipedia.org/wiki/METRORail

> one mile of NYC subway now costs about as much as 1000 miles of 6-lane freeway

Is that freeway going through the middle of the city? I think not. Any construction in a dense city becomes expensive. But you need to divide that by the number of people using it as well. Everyone in the city can use a subway. Only that one house on top of the hill is using this pretty expensive road that goes to the top of the hill in the middle of nowhere.

> Any construction in a dense city becomes expensive.

Dude, that's literally the point I'm trying to make. Thank you.

Dude. Did you read the rest of my comment? It is obviously expensive to spend 100U (some hypothetical unit since I don't wanna waste time looking for the exact numbers) instead of 10U. But it is more expensive when each one of a 1000 people have to spend that 10U rather than all of those 1000 people contributing 0.1U to get 100U total.

Reply on news.ycombinator.com