- Champagne & K rations
- Estimation with high uncertainty
- Managing up: briefings & earned trust
- Managing down: authority & responsibility
- Scientists vs professional managers
- Information & competition
- The moment
- Deliverables & inevitability
Earlier this year I read Leslie Groves’ insider history of the Manhattan Project. It’s hefty, but bundled in with the history & science is an unexpected textbook on management. The bomb was over budget, over deadline, and created a product that people hate. It still ranks in the most successful R&D projects ever because of the size of what was accomplished.
Like most modern managers, Groves wasn’t a domain expert; he was a smart engineer who delegated as much as possible to the scientific and industrial leaders of the labs and factories.
A lot of his problems and solutions are applicable to the projects I work on now, 75 years later. Below I’ve excerpted brief clips from the book that are relevant to my job.
Champagne & K rations
In August, two months after D-Day, a small team of scientists & officers called Alsos landed at Rennes in France. Armed with geiger counters, they searched the university grounds for research concerning rare earths, isotope separation, and cyclotrons – the telltales of atomic science.
Afterwards they moved to the vicinity of Paris, which was still in enemy hands. Dodging mines and intermittent sniper fire, they searched the abandoned houses of Frederic Joliot-Curie and his associates, finding them stripped of all research materials. Joliot was a leading expert on the nuclear chain reaction.
Finally they left the American 12th Army and headed to the front line:
On August 25, they reached Paris, at the Porte d’Orleans, ahead of the French troops, and waited there for about half an hour until General LeClerc arrived with his armored division. The General led the triumphal entry, at 8:55 that morning, but tucked into the column, directly behind the first tank was an American jeep containing the first representatives of the U.S. Army: Pash, Calvert and two other Alsos agents.
Later, towards evening, they broke off again, and succeeded in reaching Joliot’s laboratory. There, on the steps of the university, they found Joliot and some of his staff, all wearing FFI arm bands. That evening they celebrated the liberation with Joliot by drinking some champagne he had reserved for the occasion. The American soldier’s staff of life, the K ration, served as the hors d’oeuvres. In keeping with the scientific surroundings, the champagne was drunk from laboratory beakers.1
The Alsos group was the scientific espionage wing of the Manhattan Project, the secret & far-reaching American research project to develop a nuclear weapon. Alsos was in Europe seeking to capture and control Nazi science, especially anything that could be used to build a nuke.
They found nothing of significance. The skeleton of a research effort was scattered across Europe, odds and ends of the basic science the Americans had finished two years earlier.
Why did the Americans succeed and the Germans fail?
Creating an A-bomb with 1940s technology while your country is at war is hard work.
There were endless technical, financial, human and political obstacles that could have ended the program before they delivered their 1 prototype and 2 complete bombs. This article is about those problems and Groves’ solutions.
Estimation with high uncertainty
Before Groves took the Project’s reins in 1942 he led construction of the Pentagon, a building with five identical sides where the biggest source of uncertainty was the soil conditions.
The nuclear project, by contrast, brought in the biggest brains in the world and required innovations in physics, metallurgy, chemistry and manufacturing. The path from beginning to end couldn’t be timed:
[it became] accepted practice to proceed vigorously on major phases of the work despite large gaps in basic knowledge.2
no one knew then just how much plutonium or U-235 would be required for an effective bomb; the most reliable estimates varied by a hundredfold from the minimum to the maximum.3
And there wasn’t even one path known in advance. The approach they chose on was like the technology tree from Sid Meier’s Civ games, but with numerous dead ends.
The founding team identified five methods to refine fissile material:
Dr. Conant … expressed the prevailing opinion that there were five basic production methods, each of which held out equal chances of success. U-235 could be separated by means of the centrifuge, diffusion and electromagnetic processes; while plutonium could be obtained from either the uranium-graphite pile or the uranium heavy-water pile.4
They started serious production research on all five methods simultaneously. Pruning the technology tree was a major priority, not just over the five methods but even for smaller decisions within one branch. For example, cooling the plutonium pile:
Four methods – using helium, air, water and heavy water – were under active study. It was essential that we concentrate on the most promising and more or less abandon work on the others.
Wiggle room in configuration and delivery date was the norm and big bets were made that intractable problems would yield to R&D.
its design, involving many acres of barrier, was based on this small piece less than two square inches in area. Even this practical foundation soon disappeared when it became known that the material used in the first filter could never be employed in the main plant.5
Always we assumed success long before there was any real basis for the assumption; in no other way could we telescope the time required for the over-all project.6
This element of doubt meant that we had to have fuses for four different height settings. By then most of us in the project were thoroughly inured to such uncertainties.7
Quantifying and adapting to uncertainty is how managers come to terms with it during R&D. I’ve worked with rare managers who have the vision to make flexible plans and parallelize their teams. Big plans can’t come to fruition without this skill.
In some shops R&D is at odds with managers, who think it diverts expensive hands from useful, measurable work. Groves, who in his prior life was basically a large-scale general contractor, somehow broke out of this mindset.
nothing would be more fatal to success than to try to arrive at a perfect plan8
Managing up: briefings & earned trust
When Groves needed the cooperation of a superior, it was his practice to send an expert from his team. This is partly from the need for secrecy and the low-fi communications tech they had, but also because educating upper management is a good way to get buy-in.
The use of officer messengers on highly secret missions, carrying no papers or very limited ones, was common practice in the Manhattan Project. Before their departure, I always instructed them carefully, going into the background and reasons for their missions so that they could answer any pertinent questions they might be asked and thus be of maximum assistance to the people with whom they talked. Written messages would have been extremely complex and confusing to anyone not thoroughly acquainted with our scientific progress in the atomic field.9
I love this rule. It’s my number one takeaway from the book. How often do we resent senior managers for making a slow or bad decision on a topic where they’re not read in? This is the solution to that.
Groves always preferred organizational power to flow from ability and earned trust rather than arbitrary rank. When his intelligence officers needed to commandeer Allied resources in Europe, he recognized that it was rank plus human and situational factors that let them succeed.
I always appreciated the co-operation given my representatives throughout the European Theater, when the only justification that they had for their apparently outlandish requests were simple memoranda signed by either Secretary Stimson, General Marshall stating that their mission was of the utmost importance.
these letters were most unusual and they realized that the matters involved must be of paramount importance. … while my officers were far from high-ranking, they were obviously of such ability … that they would have accomplished their missions no matter what obstacles stood in their way.10
In the R&D phase of the Project managing up often involved representing uncertainty and this was an area where Groves was willing to cut corners to get his way.
When pitching the Project to Congress:
Bush quite optimistically stated that there were no scientists in either Britain or the United States, associated with our project, who did not believe that we would be successful.11
When he recruited industrial partners (in this case DuPont) he gave a straightforward report on the challenges and risks:
Conant and I explained to them the entire atomic situation as it then existed, giving our views on the major problems we faced and the urgency under which we were working. We made no effort to hide our uncertainty about the feasibility of the entire project.12
Managing down: authority & responsibility
Groves understood that to operate largest R&D operation in history, every level of leadership would need to be competent and mostly independent. He seldom asked scientists (i.e. domain experts) to answer to paper pushers or arbitrary military leaders.
Authority was invariably delegated with responsibility, and this delegation was absolute and without reservation. Only in this way could the many apparently autonomous organizations working on the many apparently independent tasks be pulled together13
He was philosophically opposed to micro-managing:
instructions will be followed much more intelligently if they are general
mission orders must be used whenever there are many unknowns.14
He understood that decision-making is very difficult and wanted the best people to be making the important decisions in his organization:
there was constant interference by various people in matters that lay outside their spheres of responsibility. Throughout the life of the project, vital decisions were reached only after the most careful consideration and discussion with the men I thought were able to offer the soundest advice.15
And he wasn’t above begging. This is a telegram transcript between Groves and Curtis LeMay, the Air Force general running B-29 sorties against Japan.
Groves: Important that we make press announcement here immediately. Has there been any confirmation of the information reported by the plane in Farrel’s message to me this morning? Did you get that?
LeMay: I am not familiar with General Farrell’s message.
Groves: Can we guess?16
At one point in medieval Scotland demand for experienced engineers was so high that a siege commander had to kidnap one.17 Back then, the word meant ‘operators of siege machinery’, and it was already an old word.
A major part of Groves’ job was recruiting. Every phase of the Project exposed new challenges and the first solution to every one of them was bringing in new experts. Even when the first bomb was ready to go and the organization had its hands full, Groves spent a day in New York with senior leaders interviewing a newspaper reporter who would witness the explosion and write a release.
Groves had an attentive eye for people, and not just at the top. Every level of the organization required the best. Once production ramped up, the MED’s uranium refinery was both one of the largest factory spaces in the world and one of the most exacting.
We encountered an extremely difficult problem in the welding of the steel plates surrounding the piles. This work had to be almost perfect. An average superior job would not do. … We created a special super-classification of welders with premium pay.18
The scale of their recruiting was tremendous:
Stone and Webster interviewed some 400,000 people for construction jobs and brought together a large force of experienced construction men19
we were unable to find enough pipe fitters to maintain our schedule. Investigation showed that there simply were not enough in the United States to fill the demands.20
For some specific hires, the team created specific profiles of the kind of experience they wanted, like the Done and gets things smart post.
To avoid any unnecessary stretch-out of our study, I wanted a man who was experienced in the oil industry, feeling that he would be used to making quick, conclusive decisions, based, if necessary, on very limited information. I did not want anyone who would always insist on 100 per cent proof before making a move.21
For interest: Oppenheimer wasn’t the favorite for his role. But at decision time, he was the only candidate qualified on paper to lead the science team who wasn’t already assigned to one of their labs.
The compatibility of the military staff with the scientists was a particular concern:
It was important that the officers whom we selected command the respect of persons already in the project, not only of the officers they would be thrown with, but particularly of the scientists. Experience had shown that the scientists were most critical of anyone whose mental alertness did not equal or excel theirs. Slowness of comprehension or inability to keep all the pertinent facts in mind, once they were explained, was fatal22
Scientists vs professional managers
If you’ve worked at a large company for any length of time you’ll recognize this moment: when Groves wanted to bring on DuPont to operate the refineries at scale, the scientists exploded.
if I would provide them with from fifty to one hundred junior engineers and draftsmen, they would then themselves design and construct the plutonium plant, rapidly and without delay. … The absurdity of such a proposal is apparent when it is remember that this was the plant where our construction forces reached a peak of forty-five thousand23
These were the brightest minds of the century, and they developed the key processes in their labs with small teams. Could they have bridged from that to a working gadget? Even with the assistance of the large industrial firms, the bomb wasn’t available in time to end the war in Europe.
A lot of teams have done a lot of soul-searching over the question of ‘can we do it smaller / faster’. Sometimes you need a big head count to get things done. Tesla learned this the hard way in 2018.
This may be a natural split in all R&D – being good at R doesn’t make you good at D. Certainly the scientists shined on the research side. From summer of 44 through the winter of 45 Oppie had the reins of the weapons design team, splitting their efforts on the gun & lens-type bombs, and delivering both.
NMR pioneer Isidor Rabi called the appointment of Oppenheimer “a real stroke of genius on the part of General Groves, who was not generally considered to be a genius”.
Take also the case of Deak Parsons, the ordnance expert who originally managed the bomb team at Los Alamos. Parsons favored the inefficient but well-understood gun-type bomb. When Oppenheimer took over, he reinvested most of the team’s resources in the implosion-type weapon even though nobody knew how to build it yet.
As things played out both devices were both ready circa July. The gun-type was so simple that they shipped and dropped one of these bombs without testing it. But it ran on rare U-235 – Little Boy used all 50kg of U-235 they had refined. The implosion-type weapon, which used plutonium and was far more powerful, was a riskier investment that worked out. Had they needed more than two bombs, the shortage of U-235 would have hobbled the gun-type assembly line.
Which path was right? I don’t see a general rule here. But when experts want to invest in something that seems out of reach, where possible we should let them.
Even though Parsons was demoted he and Oppenheimer stayed friends. When Oppenheimer’s security clearance was revoked after the war it made Parsons so mad that he had a heart attack and died.
Luck played as large a role in the success of the Project as scientist or consultants. A forward-thinking Belgian executive brought a huge shipment of high-content uranium ore to the US when war broke out and was pretty much waiting by the phone in Staten Island when the Project needed him.
KD Nichols, the project’s second-in-command, said:
Without Sengier’s foresight in stockpiling ore in the United States and aboveground in Africa, we simply would not have had the amounts of uranium needed to justify building the large separation plants and the plutonium reactors. 24
By eerie coincidence Sengier’s uranium comprised just half of a larger shipment. The other half spent the war in a rail yard in Belgium and was discovered by the Alsos group in the first days of the invasion of Germany. It had been in German hands for most of the war.
Information & competition
Groves’ account of hunting the German nuclear project reads like a spy novel. But everything the Alsos agents found was eerie and abandoned.
Following a tip, they covered the last mile at night, dodging the remaining German resistance and commandeering Allied resources when needed. They finally captured Europe’s top physics talent in a farmhouse.
The fear of a German nuclear project was a constant motivator. Competitive fear is the likely reason Roosevelt took the Szilard letter seriously and formed the Project. Anybody building a business has experienced these fears, where investment has to be balanced against defensibility & moat questions.
Our chief danger was that they might come up with relatively simple solutions to the problems we were finding so difficult.25
our enemies were continually harping on their proposed use of secret weapons. Although this was sometimes hard on our nerves, it did keep us from ever becoming overconfident of the superiority of American-British efforts26
had to assume that the most competent German scientists and engineers were working on an atomic program with the full support of their government and with the full capacity of German industry at their disposal.27
Once Alsos agents landed in Europe, they were able to act on these fears with an espionage process indistinguishable, in some places, from market research:
Lists were compiled of all of the precious metal refineries, the physics laboratories, the handlers of uranium and thorium, manufacturers of centrifugal and reciprocating pumps, power plants and other such installations as were known to exist in the Axis countries.28
Equal to the fear of the Germans winning the research race was leaking nuclear secrets via espionage. Security was a top priority for the Project, not just against the Axis but also for keeping back critical research from the Russians, French and even British & Canadians.
Even naming mattered. Groves hated that the spy group was called Alsos (Greek for ‘Groves’). He always assumed Oppenheimer had chosen ‘Trinity’ to blend in with the local place names, but when he asked, Oppenheimer sent him a line from a sermon that John Donne wrote while he was dying. Oppenheimer’s girlfriend sent it to him before her suicide.
The Trinity test was the moment when the impossible would either turn into the possible or not, and the Project didn’t have enough plutonium to try again without pushing back all their deadlines. After two years of R&D work pruning the technology tree, one huge question was still uncertain – what was going to happen when the bomb was triggered?
At the moment of truth even the weather rebelled:
for a considerable period, they had been making accurate long-range weather predictions for the test site. The only time they were not right was on the one day that counted.29
Fermi analyzed the likelihood the test would ignite the nitrogen in the atmosphere. Or the bomb could fizzle, scattering precious, toxic plutonium halfway across the state. The senior scientists took bets on the explosive yield of the device (someone won with 18 kt vs measurements of 18.6).
A lot was riding on a successful test, more than just reputations, careers, and dollars spent. The projected delivery date of the bomb had been factored into the Potsdam Declaration demanding the surrender of Japan. Valuable B-29s and critical personnel were diverted from the front to support the Project’s work.
Given the pressure, Groves’ job during the Trinity test was less technical oversight and more emotional leadership. As the weather turned, and the team had to decide whether to postpone the test, emotions ran high.
There was an air of excitement at the camp that I did not like, for this was a time when calm deliberation was most essential.30
I felt that no sound decision could ever be reached amidst such confusion, so I took Oppenheimer into an office that had been set up for him in the base camp, where we could discuss matters quietly and calmly. The only other persons taking part in this conversation were some weather forecasters whom we called in. Since it was obvious that they were completely upset by the failure of the long-range predictions, I soon excused them.31
Every five or ten minutes, Oppenheimer and I would leave the dugout and go outside and discuss the weather. I was devoting myself during this period to shielding Oppenheimer from the excitement swirling about us,32
Trinity was a success, and like railroad switch, many connected events were influenced by its outcome. Beyond the military & operational work of building and delivering another bomb, and the human consequences of using it, even negotiations between states were affected when the news arrived:
Truman was much fortified by something that had happened, that he had stood up to the Russians in a most emphatic and decisive manner33
A modern take on this:
They’re right, stakes matter, but they’re not all that matters. … Capabilities affect resolve.34
My point is that achieving the impossible changes the situation. This is the reason organizations engage in R&D. Forgetting the pervasive impact of success is a reason we flag or fail.
Every organization faces doubts about the power of change and the odds of success. Those doubts need to be addressed analytically with data & planning, but also emotionally through courage and leadership.
Deliverables & inevitability
Many of the Project scientists were refugees. Groves thought that they best understood what was at stake and had the most reason to succeed in the race for a bomb. Szilard, as much the creator of the Project as anybody, fled Hungary with his family.
Most physical scientists realized that nuclear energy … might be used either to generate power for peaceful purposes or to create super-weapons. … it was the scientists who were personally acquantied with Hitler’s New Order who first became most interested in the possible military uses
many scientists came to doubt the wisdom of communicating the results of their work to scientist’s in the enemy’s camp. Again, this was particularly true of those who had come to America to escape Nazi persecution.35
The moral certainty about beating Germany to a bomb faded when Japan was the only enemy left. Oppenheimer hated that a second bomb was dropped, and got into such a fight with Truman that he was banned from the White House. He quoted the Gita line ‘brightness of a thousand suns’ at Trinity but is better known for quoting a later line, ‘now I am become death’, about the effect of war on moral behavior.
Even for those who had no doubts about using the weapon, its destructive power changed their outlook on war and peace. Groves ended his book recommending international control of nuclear arms and naming the United Nations as the best hope for peace.
After Trinity, the political consequences of having a bomb were instantly clear to everyone involved with the project, and to the French, German & Russian scientists who heard about it.
Agents secretly recorded the captured German scientists in 1945 and their conversations are a creepy inverse to the early certainty of the refugees:
[Otto Hahn] had contemplated suicide when he first saw the full potentialities of his discovery, and now that these had been realized, he felt that he personally was to blame.36
HEISENBERG: it’s got nothing to do with atoms … I am willing to believe that it is a high pressure bomb and I don’t believe that it has anything to do with uranium, but that it is a chemical thing where they have enormously increased the whole explosion.37
WEIZSAXCKER: I believe the reason we didn’t do it was because all the physicists didn’t want to do it, on principles.
HAHN: I don’t believe that, but I am thankful we didn’t succeed.38
There’s an argument that the bomb had no impact on the course of the war. It wasn’t available soon enough to use in Europe and the B-29, able to fly above enemy fighters, was the deadlier weapon.
I wonder if once you buy something for 2 billion dollars you have to use it. Early in the Project, in a meeting with an accountant, the Project leaders wondered what Congress would do when they found out about the Project and its expense:
“It won’t take me five minutes—I can give you my report in thirty seconds.” Intrigued by this, Mr. Patterson told him to go ahead, and Madigan reported, “If the project succeeds, there won’t be any investigation. If it doesn’t, they won’t investigate anything else.”39
- Some people think we don’t know how to manage knowledge workers. Did Groves know how? I don’t know, but given the increasing ratio of innovation as a fraction of all work, we should abandon organizational habits that no longer fit.
- Groves focused on top talent everywhere, not just in key roles. Getting great welders or sufficient pipe-fitters was as important to him as having Nobel Prize winners running the labs. Putting anything but the smartest officers next to the scientists ruins the pie.
- Coping with uncertainty matters, success here requires both analytical planning skills & emotional leadership.
- National research projects are passe in the US now but other countries will continue to invest in them. AI in China, cancer in Iceland, the enormously successful MITI in Japan. Research as much as roads is the basic infrastructure that enables an economy; assuming government can spend effectively, research isn’t a bad use of tax dollars.
- Can a politically repressive country engage in innovation research? The Project was subject to constant censorship for various reasons, and they managed to deliver.
- Success in R&D requires pain-tolerant managers.
Groves p 211 ↩
Groves, p 11 ↩
Groves, p 72 ↩
Groves, p 10 ↩
Groves p 113 ↩
Groves p 253 ↩
Groves p 269 ↩
Groves p 42 ↩
Groves p 202 ↩
Groves p 236 ↩
Groves p 363 ↩
Groves p 46 ↩
Groves p 26 ↩
Groves p 343 ↩
Groves p 328 ↩
John France / Western Warfare in the Age of the Crusades ↩
Groves p 84 ↩
Groves p 98 ↩
Groves p 101 ↩
Groves p 180 ↩
Groves p 373 ↩
Groves p 44 ↩
Groves p 187 ↩
Groves p 187 ↩
Groves p 187 ↩
Groves p 198 ↩
Groves p 291 ↩
Groves p 291 ↩
Groves p 292 ↩
Groves p 294 ↩
Groves p 304 ↩
Groves p 4 ↩
Groves p 333 ↩
Groves p 333 ↩
Groves p 335 ↩
Groves p 360 ↩