Bayesian Investor Blog

Book review: Now It Can Be Told: The Story Of The Manhattan Project, by Leslie R. Groves.

This is the story of a desperate arms race, against what turned out to be a mostly imaginary opponent. I read it for a perspective on how future arms races and large projects might work.

What Surprised Me

It seemed strange that a large fraction of the book described how to produce purified U-235 and plutonium, and that the process of turning those fuels into bombs seemed anticlimactic.

I was surprised that the two bombs used against an enemy were two very different bomb designs: different fuel, and different methods of reaching a critical mass. They considered their use sufficiently urgent that they couldn’t afford to test the uranium reaction (used on Hiroshima) before using it in combat.

The magnitude of the undertaking stretched the limits of US capacity enough that it must have impaired some of the other war efforts. Measured by the number of employees, it was bigger than Canada’s current military, bigger than Tesla or Facebook, but smaller than Apple.

Some of the difficulty came from the need to pursue multiple engineering paths in parallel, starting many of them with only vague guesses about what the path needed to accomplish. E.g. in October 1942, their estimate of how much fissionable material they needed per bomb was considered to be only accurate to within a factor of ten. A couple of months later, Groves got estimates of whether the leading strategy for producing plutonium would work: one expert was 99% confident, another saw only a 1% chance of success.

Security

The security considerations involved lots of problems that we rarely consider today. A hundred thousand ordinary workers can’t be hired without at least a moderate amount of ordinary comment in local newspapers. The press was mostly quite cooperative about censoring anything especially sensitive. But there was a tricky trade-off involved in telling the press what to censor, as too much discussion of what to censor would draw unwanted attention, even though the press was fairly trustworthy. (Today it sounds strange to use “trustworthy” and “press” in the same sentence.)

It seems that the security was sufficient to avoid alarming Germany, but not the Soviet Union.

Lessons

Nothing in recent memory has come close to being as ambitious as the Manhattan Project.

The Manhattan Project seems by far the most impressive example of a major innovation that worked the first time it was tested. When I try to come up with comparable examples, they all seem like smaller advances on existing technology: bitcoin, the space shuttle, the moon landing. The Manhattan Project roughly consisted of inventing alchemy in three years, after centuries of failed attempts.

The project required an unusual degree of trust and cooperation. The book documents that, but doesn’t attempt to understand the sources of that trust and cooperation.

I’m less clear on what I learned about arms races, beyond the fact that it can be both hard and important to figure out whether one is in an arms race.

The project started with little evidence as to whether there was an arms race. It became increasingly clear that Germany wasn’t mounting a serious effort, whereas the Soviets were at least serious about espionage related to the project. (No other countries appeared to have an ability to compete in any such race).

Obeying Orders

The book occasionally mentions problems with subordinates figuring out how to follow orders, given limited communication.

That reminds me of Infernal Corrigibility, Fiendishly Difficult:

Someone else in Aspexia’s position might wonder whether Asmodeus would be pleased, if she disobeyed Asmodeus’s orders [in order to achieve something Asmodeus might want] … Aspexia does not even consider it. One of the foremost ways in which a Grand High Priestess of Asmodeus is shaped, is to predictably not behave in ways that make it more expensive for Asmodeus to keep His compacts. Improvising circles around your orders can rather tend do that. If Aspexia was the kind of priestess to circumvent her orders, Asmodeus would have needed to take that nature into account in choosing her orders.

One example involved soldiers outside of the project being told to use two large cargo planes to transport two ordinary sized boxes across the Pacific. Given that planes were then in somewhat short supply, why couldn’t they fill the planes with other cargo? Security concerns made it unwise to tell them anything related to atomic bombs or the U-235 in one box.

Why the second plane? In case one plane went down, they wanted good info about where. Note that planes were much less reliable then than they are now.

The orders were mostly followed, in spite of complaints that they were idiotic.

But that doesn’t mean subordinates should always follow orders. The plane that dropped the bomb on Nagasaki had orders to drop the bomb only if the weather permitted visual sighting. The flight had a few snafus, which left it in a situation where it appeared that following orders would have left too little fuel to reach an airport (keeping the bomb would have meant using more fuel, and landing with it in the ocean, with vague hopes of a submarine rescue). The crew decided to drop the bomb using radar (Groves seems to approve). Their luck changed before they implemented that choice – they were able to drop it visually, and landed with too little fuel to taxi off the runway.

A key difference is that in the first situation, the subordinates had reasons to suspect that key information was being kept from them. In the second situation, the subordinates had information that would have surprised their superiors, and expected that their superiors had tried to provide fairly complete information. However, I wouldn’t want to generalize too much from these examples.

Groves hints that the German government was less effective at getting scientists to cooperate with its goals. I don’t see enough evidence there to draw any interesting conclusions.

Closing Thoughts

How much of this book is biased toward the author’s perspective? I see slightly fewer admissions of mistakes than I’d expect in an optimal book, but also few attempts to blame others. With a few exceptions, such as blaming the British for a key security failure, Groves praises nearly everyone that he names.

Here’s some odd praise for Groves from Edward Teller’s introduction:

He started with, and partially retained, thorough doubts about the feasibility of the project. Yet in convincing the leaders at DuPont that they should participate, he appeared totally confident in order to overcome the incredulity of those overly sane engineers.

There are some important policy questions related to this story that I’m mostly ignoring because the book added little to my understanding. But I will note that Groves hints that the cost of the project made it hard to not use the bomb. If Truman had decided against using it, other politicians would have asked uncomfortable questions about the resources spent on the project. Whereas actual events led to universal praise from US politicians. That likely wasn’t the deciding factor in Truman’s choice. It’s disturbing to think that under modestly different conditions, it could have been.