The peculiar hazards of megaprojects

Originally published at:


Let’s follow this one from the start

See also:


Don’t forget the “American Dream”, Chris Christie’s non-trump-fluffing legacy:

Lots, lots more

Some do seem to have worked. The Interstate Highway project comes to mind. Possibly an outlier tho.


K-Mart was destroyed by a failed IT megaproject!

I thought it was private equity. We should blame HAL?

Your reflexive lack of faith in large government projects is disturbingly ungrounded in historical knowledge.

Major US government projects that are widely regarded as landmarks of success (partial list, off the top of my head):

The Manhattan Project ($2 billion in 1945)
The Apollo program ($25.4 billion in 1968)
The Marshall Plan ($12 billion in 1948)

And a few of the more famous ones from around the world, again off the top of my head:

The Suez Canal (100 million in 1869)
The Panama Canal (360 million in 1914)
The Channel Tunnel (£9 billion in 1994)


I suspect that something like the interstate highway project had the distinct advantage of compartmentalizing relatively neatly.

The “This one is going through, period” incentives probably led to some localized bouts of graft or incompetence that would have killed a local road project; but the project as a whole had the advantage of being made up of a lot of pieces that could be executed in parallel, mostly according to traditional methods(some standardization at the top; but pretty much just road construction happening in each location), and which were of nonzero use even before the whole project was finished.

The best project management death marches happen when you can’t parallelize(leaving the project spending a horrific amount of time waiting behind whoever is blocking the critical path); and where you don’t realize any gains until the thing is fully finished. “90% done, 90% to go” they say; so it really, really, helps if 90% done is, if not 90% of total utility, at least quite useful; because grinding out the last bit tends to be slow and painful. Projects that require major changes across a broad swath of people who don’t necessarily buy in also lose significant points. Exactly the sort of thing some rip and replace ERP nightmare does; and a 'how about you build a road that correctly intersects with another road when it hits the state line?" doesn’t.


I was going to say that this should also be counted as a US govt. project, but apparently wikipedia says:

France began work on the canal in 1881, but stopped due to engineering problems and a high worker mortality rate. The United States took over the project in 1904 and opened the canal on August 15, 1914.

So I don’t really know how to count that one. For France, it was a failed megaproject:

In France, de Lesseps kept the investment and supply of workers flowing long after it was obvious that the targets were not being met, but eventually the money ran out. The French effort went bankrupt in 1889 after reportedly spending US$287,000,000 and losing an estimated 22,000 lives to disease and accidents, wiping out the savings of 800,000 investors.[18][22] Work was suspended on May 15, and in the ensuing scandal, known as the Panama affair, some of those deemed responsible were prosecuted, including Gustave Eiffel.[23] De Lesseps and his son Charles were found guilty of misappropriation of funds and sentenced to five years’ imprisonment, though this was later overturned, and the father, at 88, was never imprisoned.[18]


I would add the LHC as one that worked well. Also any modern semiconductor fab costs many billions of dollars, and people routinely make those successfully.

I think the problems are more problems of corruption, and often the inherently corrupt incentives in the cost-plus business model, than in terms of large projects in general.


The interstate project actually fits many of the problems very well. The original proposal wasn’t actually declared complete until the early 90s and even then we have discontinuities in the system that make it technically incomplete. Large sections were cancelled mid project after a large outlay of funds, look at I-40, 80, 95, and 70 for some better examples. The costs also ended up running something like a half trillion dollars. The graft associated with the projects is too large to meaningfully summarize.

Sure it had lots of positive effects but it also ballooned way beyond the initial proposal and helped gut our cities. Consider what alternative projects could have been completed with a half trillion dollars and it is way less clear that the Interstate Highway System was the right choice.


Obviously they’re being rewarded for employing noted patriot Edward Snowden :slight_smile:

1 Like

It’s probably worth inquiring about what qualifies you as a ‘failure’.

All the projects @Glaurung listed got executed in the end; but the ones that had some sort of serious attempt at a cost estimate typically blew through it(channel tunnel by only 80%, Suez Canal roughly double original estimates; Apollo Program, James Webb deserves a dozen telescopes for tossing the $7 billion number and estimating $20 billion before presenting it; or it would have gone a lot more than 25% over); I’m not sure how much the Marshall Plan and Manhattan Project were intended to cost ahead of time; but in the latter case the time to delivery made it really lucky that the Axis weren’t having all that much luck with their own program; because it didn’t exactly ship early.

For ‘success’ do we just count delivery? Delivery at cost justified by ultimate utility, regardless of original budget? Insist on on time/on budget?


IIRC, James Webb is a victim of funding instability - every time Congress allocates not enough money to a project, that lack of funding creates delays (because there isn’t enough money to finish the next bit on time) and those delays create further cost overruns. Basically trying to pinch pennies for a complex project can actually backfire and cause costs to balloon for the project instead. I remember reading this WRT NASA projects like the James Webb telescope, but my Google fu is weak and I cannot locate the article today.

ETA: as to how we determine success, I really think looking at cost is ass backward. Huge projects are always going to cost huge amounts of money, and are always going to come in late and over budget. While they are underway, they’re always going to seem to be costing too fucking much. The question is, once they’re completed, in the long run, are we happy with the result of all that hard work, or not?

By that standard, the Marshall Plan and the Apollo Program were very successful. The Manhattan project delivered a decidedly mixed blessing, but the US government never seemed to begrudge doing it, especially during the Cold War.


The thrust of the paper seems to be that megaprojects always overrun massively and are therefore mis-sold, but you have to wonder, if it’s that much of a given (to the point of being called the “iron law” of big projects), then isn’t that consideration already baked into the system?

In other words, if everyone knows that a project slated to cost $10bn will probably cost at least $20bn, is the $10bn figure dangerously misleading people, or is it just understood as the table stakes for a process at whose ultimate cost it would be foolish to guess?

I realise it would be possible to simply apply a “large project multiplier” to the first estimate, controlling overruns (on a worldwide scale) by purely statistical means, but that multiplier itself is a guess, so people won’t be that much more confident of the adjusted $21.3bn estimate than the original $10bn estimate; they’ll just be put off.

The paper claims to refute that “but things would never get built” argument, but I’m not sure how. It’s not like a $10bn railroad gets built directly instead of 500 new $20m schools. If you were offered that literal choice, you would assume the railroad would end up costing more, but you would also assume that you wouldn’t get all the schools, because after the first 15 or so, someone would siphon the money elsewhere, and possibly even into their own pocket. With the railroad, the whole $10bn will ultimately go to what you are trying to use it for, even if it’s likely to suck in a lot more billions along the way.

(None of this, btw, is me endorsing giant weapons projects. A big bad project is still bad, and worse than a small bad project)


I find it hard to believe K-Mart would still be anything other than destroyed even if they hadn’t done some IT megaproject, and private equity kept their hands off completely.

1 Like

I think that we have to decide what our definition of “failure” is. Because many mega projects bankrupt the people making them, but do get finished. And once the companies have gone bankrupt, and the bond holders have been stiffed, there is a market clearing price at which they can operate if they don’t have to ever pay back the money used to build them. Those are examples, of that kind of “success.”
To which we can add. the Concorde, the Iridium satellite phone system, and many others.

The 747 was an actual successful mega project.


Also, the moon landings. We never made that money back via mining and permanent residences as promised.


The Big Dig in Boston is an epic saga of graft and corruption. It’s also pretty great. Really tied the room together


“…as promised”?

By whom?

Seriously, I was deeply into Project Apollo as it happened, and I never heard anyone “promise” – or even seriously suggest – that the cost of Apollo would be repaid by “mining and permanent residences.”

Neither mining nor permanent residences were ever part of the Apollo program, either as plans or as justification.

Idle speculations about future possibilities by futurists, science fiction writers, and the occasional aerospace engineer do not constitute “promises.”

(It is easy, BTW, to make the case that the economic benefits produced by the generously-funded R&D of the Apollo program have more than repaid its cost - just not from moon mining or colonization.)


Sorry, I was unclear about my meaning with regard to Webb. I didn’t have the cost of the telescope named after him in mind; I was thinking of the fact that he is (at least credited as) the person who took the $7 billion estimate for the Apollo program and had the good sense and honesty to advance the $20 billion estimate before the project was kicked further up the chain and sold to congress, the president, and the public.

They don’t normally name space telescopes after you for project management; but if anyone is eligible he deserves at least a few for that. Going 25% over on a cutting edge R&D project of massive scope is pretty good, a great many of us should be as lucky when remodeling our kitchens. Costing 3.5 times as much as estimated is somewhat less good; so he gets points for avoiding that.

Might have had to do with is background; he came to NASA via the Bureau of the Budget and some work for the treasury. Not much of a hard science CV; but probably knew something about cost estimates.