American Power Was Built on Cost Overruns
The construction of the Pentagon during WWII has lessons for today.
The construction of the Pentagon was more than 2x over budget, and - in a joke that writes itself - it represented the Pentagon’s first cost overrun. What followed would be a trend: the Manhattan Project, Apollo, etc. In addition to being massively over budget, these programs all have something in common…
They worked. And they delivered in time to meet the moment. Apollo inspired America at a time when kids were hiding under their desks. The Manhattan Project ended WWII when teenagers were preparing to invade Japan. The Pentagon, the world’s largest office building until earlier this year, was constructed during WWII in time for 20,000+ War Department employees who needed a place to plan and prosecute the war.
Nobody cares that these programs were over budget. They are priceless gems in American history that delivered eye watering capability on a timeline that mattered. Today, the Department of Defense would do well to reorient towards spending money to save time and deliver outcomes. This isn’t as easy as it sounds. It requires identifying leaders who are seemingly preternatural and funding them over those who are less capable.
Speaking of preternatural leaders - when Matt Damon Lt General Leslie Groves wasn’t on screen in Oppenheimer leading the Manhattan Project, he was busy leading the construction of a little known building called the Pentagon. His boss, General Brehon Somervell, the man who had conceived of the idea for the Pentagon, was also a good multitasker; he was running logistics for the entire Army the moment the U.S. entered the war.1
Groves and Somervell were men who understood the tradeoff between time and money. I’ll use their leadership constructing the Pentagon as a model example from which the entirety of the military-industrial-complex should draw inspiration as we seek to increase our technological baseline in time to deter conflict with our adversaries.
In 1941, the Pentagon was urgently needed. War was closing in as the War Department was spreading out. Some 24,000 employees were dispersed across dozens of buildings, and that number was set to increase to 30,000 by year’s end. This fragmentation was a national security threat. The inefficiencies of a scattered War Department were endangering American readiness and limiting the Army’s ability to coordinate mobilization. To mitigate the dire situation, Congress decided to construct temporary buildings and called on Somervell, then head of the Construction Division of the Quartermaster Corps, to figure out “an overall solution (pg 33).” Over the weekend, Somervell had his team draw up plans for a $35 million, 5.1 million square feet, 5-sided monstrosity to house the entire Department.
This was an incredibly bold move on Somervell’s part. Just one month earlier, a new, $18 million War Department headquarters, literally called The New War Department Building, had opened. Fortunately for Somervell, the new headquarters was considered completely inadequate, able to accommodate only 4,000 employees. It was also considered pitifully ugly, and Secretary of War Henry Stimson refused to move in. Even still, Stimson was not inclined to embark on yet another expensive headquarters.
But it wasn’t up to Stimson. FDR gave the green light for the Pentagon on July 24, 1941, just one week after Somervell drew up the initial plans. Somervell promised that the world’s largest office building would be ready for occupancy well within one year. And it was. The Quartermaster Corps broke ground on Sept 11, 1941, and the first employees moved in on April 30, 1942, less than eight months later.2 The entire building was completed by mid-February 1943, by which time the Pentagon had expanded to 6.1 million square feet (Somervell took advantage of the momentum of Pearl Harbor to add a fifth floor).
Everywhere, Somervell and Groves made creative decisions, bordering on the questionable, to accommodate the dual constraints of time and war shortages. Infamously, the Pentagon is a low building because Somervell needed to limit the amount of steel used (as well as aluminum, tin, copper, and most copper alloys used for galvanizing iron and steel…). Instead of steel, it’s constructed using reinforced concrete made from 680,000 tons of sand and gravel dredged from the Potomac river and supported by 41,492 concrete piles. Groves personally called engineering firms around the country desperately looking for concrete specialists.
Construction took place concurrently with architecture and structural design work. As the architects and engineers renovating the Pentagon learned more than fifty years later, many sections of the Pentagon were built without drawings or with drawings that bore little resemblance to reality. Somervell’s mandate to add a fifth floor was made with such haste that workers didn’t have time to fully remove the original roof.
Somervell and Groves had so spectacularly succeeded with the Pentagon that neither of them was particularly stressed about informing Congress in 1942 that they were $14.2 million over budget (they were actually $40 million over budget, as would be revealed in later investigations not made public).3 And Congress, for that matter, didn’t give them a hard time. The House Appropriations subcommittee chairman did remark to Groves that “you have overshot the mark by a pretty big margin” but that was the extent of the fuss (pg 221). “No protest was raised at the increased size or cost. Instead, much of the hearing dealt with complaints that the floors of the Pentagon were dusty (pg 221).”
Today, we have cost overruns for all the wrong reasons. It’s not because we’re building wonders in record time, but because poor-performing programs keep moving through the production pipeline irrespective of the value they’re generating (among other reasons that require a separate post). The former category of cost overruns is acceptable, as I’ve hopefully convinced you through the example of the Pentagon. The latter category of cost overruns is an abomination and an affront to American excellence.
Just deliver the value promised and nobody will remember the price tag.
Colonel and Brigadier General were the ranks Groves and Somervell held, respectively, when the Pentagon project began. Groves was promoted to Lieutenant General just before his retirement in recognition of his work on the bomb. Somervell earned his third star when he became Commander of the Army Service Forces, and he was given a fourth star upon his retirement.
Just two months later, on Dec 1, 1941, “Roosevelt signed legislation that stripped the Quartermaster Corps of its historical role directing Army construction, giving that responsibility to the Corps of Engineers. The Construction Division was being transferred to the Corps of Engineers (The Pentagon: A History, pg 158).” Despite the shakeup, Somervell remained boss of Pentagon construction for all practical purposes.
There are disputes as to the exact cost of the Pentagon. $63.6 million was the formally acknowledged figure by the War Department (and the DoD today). $86 million was what Congressman Albert Engel claimed was the true cost. Vogel asserts that the $75.2 million the Pentagon actually spent on construction is the best estimate, and it is the figure I’ve used (The Pentagon: A History, pg 331).
Interesting article. I have not read the book, or am very familiar with the construction of the Pentagon, but it is my understanding that military contractors in World War 2 were generally compensated using a Cost Plus model.
In other words, the company stated that a piece of equipment costs X dollars to produce, and the government compensated them with X + 0.10% (or whatever the actual number was). In this model, there was no such thing as a cost-overrun.
This is very different from our current system of competitive bidding before the project even starts. Our current system gives a strong incentive to make an unrealistic bid and then slowly roll out the real costs after it is too late to withdraw the bid without significant embarrassment.
Obviously, the WW2 model relies heavily on trust, but it seemed to work extremely well. My guess is that such a model would be very controversial, and perhaps unworkable, if used today.