The MTA's Second Avenue line in New York comes out to more than AU$2Bn per km, on my reckoning (US$2.6Bn/mile). It's the current world champion for expensive holes.

The fact is that megaprojects basically never come in anywhere near their schedule or budget. That's because of a selection bias. When you crunch the numbers honestly, megaprojects come out as a terrible on a cost-benefit or ROI analysis. The ones that get built are built by people who choose to ignore the honest figures in favour of the shiny, optimistic ones. Et voila, another farce.

There are lots of other factors, but that's the core one that precedes the others.

Can you elucidate on that line of reasoning some more? I am unconvinced.

It has been my experience -- which some more "keepin it real" journalism concurs with -- that New York's outlandish infrastructure costs, as well as the majority of the USA's public projects, are primarily due to two main agents:

1. White collar bureaucracy

2. Blue collar "bureaucracy"

Both parties -- and the multitude of agents that operate within them -- are all too able to align their "piece" of the project towards their own interests.

These interests very much involve time and money.

For money, it's simply rerouting costs. And for time, it's "making one's mark," by misallocating resources to personally-enriching matters, that do not benefit the project as a whole, and usually hinder it.

It's late, and I'm not at 100% to go into detail, but surely this is well-known already?

I did structure my argument to say that the optimistic estimates precede the other problems. Before ordinary problems of inefficiency and corruption arise, the project has to be sold as a viable investment.

I take the argument about selection bias from Brent Flyvbjerg, who has been studying megaprojects for a fair while now. He calls it "survival of the unfittest"[0].

A similar resource is Merrow's Industrial Megaprojects[1], which is based on a large dataset of such efforts. He identifies many factors in project overruns, but essentially notes that most projects start off with fanciful estimates that will never be met. For a book about petrochemical plants, bridges and power stations, it's a fun read.

[0] https://arxiv.org/ftp/arxiv/papers/1409/1409.0003.pdf

[1] https://www.amazon.com/Industrial-Megaprojects-Concepts-Stra...

Megaprojects do come in near budget in many places. Not in the US for a number of reasons elucidated in this NYMag piece[0] by Josh Barro, whose takes I usually find surprisingly persuasive.

[0] http://nymag.com/intelligencer/2019/05/new-york-infrastructu...

Like the Berlin Airport

But it's just a subway tunnel, which have been built in cities since the 19th century. Is it really so much harder to do such "megaprojects" in the 21st century, with the vastly better equipment now available?

There are a raft of issues. The construction industry has not significantly benefited from large technological advancements that other industries have over the last century and is still human labor dependent. Wages have increased, we've decided it's not ok to kill people anymore, unions have removed exploitation in the immediate supply chains.

We also have more complicated, and are dependent on, soft and hard power structures to access and use capital. A great place to start looking into these issues is Bent Flyvbjerg at Said, Oxford

We’ve been building houses for millennia and construction is still notorious for not going according to plan. One-off projects are just hard to get to go as planned.

Lesson: don't do one-off projects. Find a project that meets you needs, and copy it. Make minor tweaks to their design to fit your needs, but copy as much as possible.

Houses typically go according to plan because so many of the processes and parts are standard. Even large mansions can go according to plan if you have good management (note that cheap management is rewarded for change orders and thus will ensure things don't go according to plan while better management will see that if they don't make a minor change now things won't go according to plan latter.

Imo big infrastructure needs like roads and subways shouldn't be done on a project basis at all. For instance, NYC shouldn't bid and build out massive contracts for a few km of subway. They should own a boring machine and add 10-100m of tunnels a week, every week[0]. (numbers arbitrary) You can shake so much overhead around project start-up and shut-down and it will be a long time before NYC (nevermind someplace like SF) has saturated the demand for public transit.

[0] Best example of this sort of logic that I have seen implemented is painting the golden gate bridge, just one properly sized and experienced crew painting from one end to the other and back again without the boondoggle.

I've thought the same for a while. A lot of the expense in upgrading rail cars, replacing frigates, building hospitals etc comes from the fact that every such project has massively expensive organisation constructed around it and then ... it's thrown away.

The Project Production Institute take a similar view: that instead of modeling a project as a fixed thing, you focus on it as a production process.

https://projectproduction.org/

But why there is a need to paint it all the time?

To make up some numbers: Assume that a paint job lasts 20 years. So if it takes a team of five 20 years to paint the bridge by the time they've reached the end the beginning will need a new coat.

Otherwise every 20 years they would need to hire, train, then fire 200 painters for a 6 month-long paint job project. 200 painters, managers, and etc. who will invariably be less experienced with the finer points of painting said bridge since they don't do it every day.

So it doesn’t rust

Most houses are unavoidably one-off projects. The point of comparison would be an assembly line at a factory. Compared to that, most houses are one-offs; that particular design has never been built at that particular site by that particular group of people and contractors.

Guess we should have stopped at the wheel.

We have improved it a lot, but the wheel has been round for a very long time. We laugh at people who reinvent the wheel as opposed to doing incremental improvements on it.

We do have ships, rockets, and a few other things that don't need wheels (tough they tend to have wheels of some form inside), but the wheel itself is still just incremental improvements on that first one way back...

NIH and 'reinventing the wheel' are phrases for questioning the need to spend time and money to develop a thing that has the same purpose as an existing thing and doesn't improve on it, which I agree with. But to use ships as an example, at some point people decided it would be useful to cross bodies of water without getting wet so they invented the boat, and as a new invention it improved quality of life immeasurably. It feels like you're saying people should exclusively stop making things which have new and novel purposes and just focus on improving existing things, which is the part I disagree with. Also, I think as an example, gears and wheels which have similar bases of design are two completely different inventions because they have different purposes, this might be where we disagree.

This is a very regulated era, and the art of extracting money from megaprojects have been perfected by many parties.

It seems that China does not suffer from this "cost disease", but I don't know any real facts.

Wikipedia has a good summary of the construction techniques. Only 10% of it is just below street level using cut-and-cover technique.

https://en.wikipedia.org/wiki/Second_Avenue_Subway#Construct...

A large fraction of NYC's subway system was built using cut-and-cover, which is much cheaper than the deep tunneling being used for projects like the Second Avenue line or the East Side Access.

That's before adding other factors such as increase regulation, inflation and good ol' corruption.

That and there was no EPA, no environmental design reviews, not much permitting, lax or zero safety standards (today it is a huge deal when somebody dies on a construction project... no so much back then). There was no ADA accessibility standards, no elevators or required wheelchair access, there was very little underground.

If all the businesses along the block you are trenching through back then went under... so fucking what?

If your 24/7 jackhammers kept all the block up the whole night you’d tell them to take a hike.... etc, etc, etc.

If turns out if you care about any or all of those things, your costs go up. Back then it was all just negative externalities that got foisted on the neighborhoods around them. Think of the negative externalities causes by our interstate highway system. Good luck building that now! It literally divided entire neighborhoods. The entire thing would be require to go into a tunnel under the city. No way would people tolerate that shit now.

While this is all true, on a dollar-for-dollar basis, cut-and-cover is still cheaper than deep tunneling. There's no way around the fact that there is far more expertise, equipment, experience and competition for digging holes compared to digging tunnels.

I really don't understand the aversion to cut and cover these days. Whole blocks of street are routinely closed and resurfaced for long periods of time, especially if sewer work is involved. Businesses in urban environments aren't really reliant on the business from the street parking directly in front of their storefront anyway. Closing a couple blocks at a time wouldn't be all that bad, and it probably wouldn't take too long to throw a temporary grate for car traffic over the open cut anyway.

Recently there was squabble about the projected revenues of the LA olympics. The more recent figures were pessimistic, and naturally the media had their time writing pieces in different directions about this, blaming the city for incompetence and budget overruns, and the public formed their opinions and became ever more polarized. The reason why the figures had changed was simply that inflation had occurred over the years since the initial projection.

Here's talk about $3.5Bn/mile for Long Island Rail Road, including reasons for exploding costs

https://www.nytimes.com/2017/12/28/nyregion/new-york-subway-...

Have you seen any breakdown of why it costs that much? It seems like an unfathomable amount.

All sorts of reasons, but corruption (along-side organized crime) is a big issue in the construction industry. Imagine Oracle-style of service and business approach coupled with an npm-deep level of dependencies/middlemen whose palms need greasing and to have their cut of the project. Then on the other side you are boxed in by physics, thousands of regulations, zoning laws, complaints, city council, government bureaucracy.

"npm-deep" I love that phrase. anyone who suffers a dig around the node_modules folder knows the pain! Ugh! I also used Oracle in the early 2000's, I need hypnotherapy to get over that.

Because the tunnel is going underneath the Sydney harbour:

https://www.smh.com.au/national/nsw/under-wraps-sydney-s-fir...