The Wall Street Journal recently reported that, in striving to curb inflation, the Federal Reserve has encountered a cost-of-living problem of its own. The Fed faces a $2.5 billion bill for remodeling its grand but decaying headquarters in Washington, up by a third from the pre-pandemic estimate.

In fairness to the central bank, the projected overrun is modest compared to the cost of, say, the Long Island Railroad East Side Access project in New York City, which allows suburban rail commuters from Long Island to ride directly to midtown Manhattan. In 2001 the LIRR proclaimed that the project would cost $4.3 billion and be ready by 2015. The final cost when the extension opened in early 2023 was at least $11.6 billion.

Cost overruns have become so common, even when fudge factors are built in to anticipate the unexpected, that many observers have become resigned to them. Runaway costs are not one end of a “normal distribution” (aka the bell curve) but represent what statisticians call a “fat tail.” This need not be, argue Bent Flyvbjerg, a Danish-born geographer at Oxford University and Dan Gardner, a journalist specializing in forecasting. Drawing on the psychology of decision making, systems theory, engineering and computer science, they outline a set of principles in their new book for keeping the behemoths on schedule and within budget.

Whose Fault?

If there is a number-one suspect in overruns, it is the mindset of politicians and corporate leaders who develop visions for great structures (or, for that matter, less tangible edifices like disruptive software) and embark on these projects without careful study. This tendency toward fiasco is almost inevitably compounded by the delays resulting from failure to plan for obstacles.

The visionaries often blame unpredictable events — so-called black swans — for the additional time and cost. As Flyvbjerg and Gardner observe, though, the more slowly an effort proceeds because of inadequate planning, the more likely it will encounter a dreaded bird. It does not help (especially in the corporate world) that there is a bias for action that can brand deliberate preparation as an enemy of getting things done.

A question follows: How to plan for problems that are truly difficult to foresee? The authors find a prototype in the practice of Frank Gehry, the architect who gave the world the stunning Guggenheim Museum in Bilbao, Spain, and the Walt Disney Concert Hall in Los Angeles, and in the methods of the animation studio Pixar. They call it “Pixar planning” — time-consuming iterative modeling of stages of a project that makes it much more likely to identify problems early and thereby save months or years of delay over the project’s life.

It also helps to use tested methods and technology, to entrust work to deeply experienced professionals, and to avoid experimental systems that are difficult to test in advance. And that means studying the history of similar projects.

Rush to Judgment

How Big Things Get Done is such a rare combination of impressive data-mining, social psychology insights, and plain old common sense that you’ll be inclined to accept the book as the last word. It is, indeed, indispensable reading for businesses and government agencies contemplating giant investments. But it is not the full story.

First, the authors neglect the role of government institutions in some of their examples. Consider the difference between the state of California and the governments of Japan and France, homes of the two most celebrated high-speed passenger rail systems, the Shinkansen and the TGV, respectively.

Japan and France, for their part, were able to literally bulldoze through local interests to get from point A to point B in a way that no state government in the U.S. can. For better and worse, California is a state of diverse communities and contentious elections. And to negotiate the political maze to build a new rail line connecting San Francisco to Los Angeles required a major detour through the lightly populated Central Valley. In fact, construction began in the middle with no hope of serious revenue generation until the entire route was completed — not because planners were ignorant of Flyvbjerg and Gardner’s modular approach, but because key state legislators demanded that approach.

A more economically rational route along the coast seemed to those politicians a giveaway to Silicon Valley moguls and their Hollywood peers. But set aside the issue of jump-starting rail use. Note that the choice of a longer route with more stops has probably doomed the high-speed train to being uncompetitive with flying between big cities. Indeed, the French national railroad company SNCF, which had been managing construction early on, gave up and turned to a high-speed rail project in Morocco where elitism was apparently not a four-letter-word. The train route along the dense corridor from Tangier to Rabat to Casablanca was ready for customers in 2018.

Second, some megaprojects are so advanced that conservative Pixar methods are of limited use, and yet may still be rational gambles. Flyvbjerg and Gardner observe scornfully that the James Webb Space Telescope took 19 years to build rather than the projected 12, and that its final cost was $8.8 billion, “an astronomical figure 45 percent over budget.”

To the authors, this excess illustrates the pitfall of the rush to commitment before a careful assessment of the hurdles that must be surmounted at each phase of the project. There was, indeed, a significant chance that the billions poured into a huge telescope slated to orbit the sun rather than the earth would prove to have been spent in vain. After all, the Superconducting Super Collider (SSC) particle accelerator in Texas, a pet project of President Ronald Reagan (who urged physicists to be “bold and greedy”!) was abandoned by Congress in 1993 after an expenditure exceeding $2 billion.

But does the failure mean that SSC should never have been authorized? From the project developers’ perspective, it would, of course, have been better not to start than to leave the particle accelerator unfinished and useless. However, there could never be true certainty about the outcome before the fact. And the potential prize was enormous: the SSC could have greatly accelerated progress in particle physics, as well as assuring U.S preeminence in the field for decades to come.

Actually, the failure to complete the project may have had further unfortunate consequences. It triggered a diaspora — half the technical staff left physics. And, if you’re willing to play this sort of game, one can also glimpse the origins of the Great Recession in the earlier collapse of SSC. Many of those exiled physicists, after all, became part of the nucleus of “quants,” the nerds who led Wall Street down the yellow brick road of higher mathematics to the ruin of the mortgage industry.

But back to the Webb Space Telescope. In identifying the project as a poster child of Big Project disease, Flyvbjerg and Gardner implicitly take on the burden of explaining how their template for efficient development would have brought the telescope online more quickly or for less money. And I doubt that would have been possible. There is no modular way to build a space telescope — or for that matter, a particle accelerator ring with a 54-mile circumference.

In any event, the fact that extra billions were spent to get Webb ready for prime time hardly means the project is a white elephant. According to Scientific American, “[t]he most powerful observatory ever made promises to produce some of the most incredible discoveries of our lifetime and beyond.”

The criteria laid down by Flyvbjerg and Gardner might have ruled out some of the greatest technological achievements of the past two centuries, including the first transatlantic telegraph cable and the Panama Canal. The canal was only built after the bankruptcy of the original French developer. Hard-charging, intuitive President Theodore Roosevelt subsequently overrode a blue-ribbon engineering committee’s recommendation of a sea-level canal in favor of the present lock canal, one of the most socially valuable (and profitable) public works ever. Sometimes an emotional vision of the future pays off big.

Third, even with the use of all the rational planning techniques the authors lay out, it may be impossible to avoid the aforementioned fat tail. Consider the status of the major project on which Flyvbjerg is serving as an advisor: a high-speed train linking London with the north of England. Now 40 percent complete, cost estimates have increased from £33 billion to £71 billion (approximately $86 billion). According to The Guardian, the government is now planning to reduce the frequency and speed of services and may delay completion of the project — as though inflated costs of materials and labor may drop again soon.

This is the point at which a cliché about glass houses and throwing stones might be warranted. But I’m inclined to be more charitable: the overrun might have been far worse without his team’s recommendations. Welcome to the club, Professor Flyvbjerg!