Posts Tagged ‘Budget’


The price of the Manhattan Project

Friday, May 17th, 2013

There's been a little radio silence over here last week; the truth is, I've been very absorbed in NUKEMAP-related work. It is going very well; I've found some things that I thought were going to be difficult to be not so difficult, after all, and I've found myself to be more mathematically capable than I usually would presume, once I really started drilling down in technical minutiae. The only down-side of the work is that it is mostly coding, mostly technical, not terribly conducive to having deep or original historical thoughts, and, of course, I've gotten completely obsessed with it. But I'm almost over the hump of the hard stuff.

Two weeks ago, I made a trip out to the West Coast to hang out with the various wonks that congregate at the Center for Nonproliferation Studies at the Monterey Institute for International Studies. This was at the behest of Stephen Schwartz, who teaches a class over there and had me come out to talk to them about nuclear secrecy, and to give a general colloquium talk.

Atomic Audit

Stephen became known to me early on in my interest in nuclear things for his work in editing the book Atomic Audit: The Costs and Consequences of U.S. Nuclear Weapons Since 1940 (Brookings Institute, 1998). This is one of these all-time useful reference books; it is the only book I've read, for example, that has anything like a good description of the development of US nuclear secrecy policies. And the list of contributors is a who's-who of late 1990s nuclear scholarship. The book includes really detailed discussions about how difficult it is to put a price tag on nuclear weapons spending in the United States, for reasons relating both to the obvious secrecy issue, but also the fact that these expenses have not really been disentangled from a lot of other spending.

I've had a copy of the book for over a decade now, and it has come in handy again and again. I'm not a numbers-guy (NUKEMAP work being the exception), but looking at these kind of aggregate figures helps me wrap my head around the "big picture" of something like, say, the Manhattan Project, in a way that is often lost by the standard historical approach of tight biographical narratives. Of the $2 billion spent on the Manhattan Project, where did it go, and what does it tell us about how we should talk about the history of the bomb?

Here is a breakdown of cost expenditures for the Manhattan Project sites, through the end of 1945:

Site/Project 1945 dollars 2012 dollars %
OAK RIDGE (Total) $1,188,352,000 $18,900,000,000 63%
K-25 Gaseous Diffusion Plant $512,166,000 $8,150,000,000 27%
Y-12 Electromagnetic Plant $477,631,000 $7,600,000,000 25%
Clinton Engineer Works, HQ and central utilities $155,951,000 $2,480,000,000 8%
Clinton Laboratories $26,932,000 $430,000,000 1%
S-50 Thermal Diffusion Plant $15,672,000 $250,000,000 1%
HANFORD ENGINEER WORKS $390,124,000 $6,200,000,000 21%
SPECIAL OPERATING MATERIALS $103,369,000 $1,640,000,000 5%
LOS ALAMOS PROJECT $74,055,000 $1,180,000,000 4%
RESEARCH AND DEVELOPMENT $69,681,000 $1,110,000,000 4%
GOVERNMENT OVERHEAD $37,255,000 $590,000,000 2%
HEAVY WATER PLANTS $26,768,000 $430,000,000 1%
Grand Total $1,889,604,000 $30,060,000,000

I've taken this chart from here. The "current dollars" are 2012 dollars, with a "production line" labor deflator used (out of all of the options here, it seemed like the most appropriate to the kind of work we're talking about, most of which was construction).

To break the numbers down a bit more, K-25, Y-12, and S-50 were all uranium enrichment plants. Hanford was for plutonium production. "Special operating materials" refers to the raw materials necessary for the entire project, most of which was uranium, but also highly-refined graphite and fluorine, among other things. Los Alamos was of course the design laboratory. The heavy water plants were constructed in Trail, British Columbia, Morgantown, West Virginia, Montgomery, Alabama, and Dana, Indiana. Their product was not used on a large scale during the war; it was produced as a back-up in case graphite proved to be a bad moderator for the Hanford reactors.

I'm a visual guy, so I of course immediately start looking at these numbers like this:

Manhattan Project costs chart

Which puts things a little more into proportion. The main take-away of these numbers for me is to be pretty impressed by the fact that some 80% of the money was spent on the plants necessary producing fissile materials. Only 4% went towards Los Alamos. And yet, in terms of how we talk about nuclear weapons and the Manhattan Project, we spend a huge amount of the time talking about the work at Los Alamos, often with only token gestures to the work at Hanford and Oak Ridge as the "next step" after the theory had been worked out.

We can also break those numbers down a little finer, by turning to another source, Appendix 2 of Richard Hewlett and Roland Anderson's The New World. There, they have costs divided into "plant" and "operations" costs:

Site/Project Plant Operations Plant %
OAK RIDGE (Total) $882,678,000 $305,674,000 74%
K-25 Gaseous Diffusion Plant $458,316,000 $53,850,000 89%
Y-12 Electromagnetic Plant $300,625,000 $177,006,000 63%
Clinton Engineer Works, HQ and central utilities $101,193,000 $54,758,000 65%
Clinton Laboratories $11,939,000 $14,993,000 44%
S-50 Thermal Diffusion Plant $10,605,000 $5,067,000 68%
HANFORD ENGINEER WORKS $339,678,000 $50,446,000 87%
SPECIAL OPERATING MATERIALS $20,810,000 $82,559,000 20%
LOS ALAMOS PROJECT $37,176,000 $36,879,000 50%
RESEARCH AND DEVELOPMENT $63,323,000 $6,358,000 91%
GOVERNMENT OVERHEAD $22,567,000 $14,688,000 61%
HEAVY WATER PLANTS $15,801,000 $10,967,000 59%
Grand Total $1,382,033,000 $507,571,000 73%

They do not define how they differentiated between "plant" and "operations" expenses, but the most plausible guess is that the former are various start-up costs (e.g. construction) and one-off costs (e.g. big purchases of materials) and the latter are day-to-day costs (general labor force, electricity, etc.).

Looking at that percentage can tell you a bit about how much of the Manhattan Project was the building of a weapons production system as opposed to building three individual weapons. Nearly three-fourths of the expense was for building a system so large that Niels Bohr famously called it country-sized factory.1

The K-25 gaseous diffusion plant: the single largest and most expensive Manhattan Project site.

The K-25 gaseous diffusion plant: the single largest and most expensive Manhattan Project site.

Another way to look at this is to say that we usually talk about the atomic bomb as project focused on scientific research. But one could arguably say that it was more a project of industrial production instead. This is actually quite in line with how General Groves, and even J. Robert Oppenheimer, saw the problem of nuclear weapons. Oppenheimer, in testimony before Congress in 1945, went so far as to phrase it this way:

I think it is important to emphasize [the role of industry in the Manhattan Project], because I deplore the tendency of myself and my colleagues to pretend that with our own hands we actually did this job. We had something to do with it. If it had not been for scientists, there would have been no atomic bomb; but if there had been only scientists, there also would be no atomic bomb.

This is actually a very important point, and one which shines light onto a lot of other questions regarding nuclear weapons. For example, one of the questions that people ask me again and again is how close the Germans were to getting an atomic bomb. The answer is, more or less, not very close at all. Why not? Because even if their scientific understanding was not too far away — which it was not, even though they were wrong about several things and behind on several others — they never came close to the stage that would be necessary to turn it into an industrial production program, as opposed to just a laboratory understanding. That sheer fact is much more important than whether Heisenberg fully understood the nature of chain reactions or anything like that.

Why do we think of the bomb as a scientific problem as opposed to an industrial one? There are perhaps a few answers to this. One is that from the beginning, the bomb came to symbolize the ultimate fruits of scientific modernity: it was seen as the worst culmination of all of those centuries of rational thought. What grim irony, and what a standard story, that knowledge could lead to such ruin? Another reason is that scientific adventure stories are more interesting than industrial adventure stories. It is much more fun to talk about characters like Szilard, Oppenheimer, and Feynman running around trying to solve difficult logic problems in a desperate race against time, than it is to talk about the difficulties inherent to the construction of very large buildings.

Finally, though, there is the issue of secrecy. The scientific facts of the atomic bomb, especially the physics, were the most easily declassifiable. As discussed in a previous post (with many nods towards the work of Rebecca Press Schwartz), one of the main reasons the Smyth Report was so physics-heavy is because the physics was not terribly secret. Nuclear chain reactions, the idea of critical mass, the basic ideas behind uranium enrichment and reactors: all of these things were knowable and even known by physicists all over the world well prior to the bombing of Hiroshima and Nagasaki. The really hard stuff — the chemistry, the metallurgy, the engineering "know-how," the specific constructions of the massive fissile-material production plants — was silently omitted from official accounts.

Looking at the costs of the bomb help rectify this perception a bit. It still doesn't get us outside of the heroic narratives, for they are very appealing, but it can help us appreciate the magnitude of what is left out of the standard story.

  1. Bohr reportedly told Teller upon seeing Los Alamos and hearing about the entire project: "You see, I told you it couldn't be done without turning the whole country into a factory. You have done just that." []

Doomsday on the Cheap

Friday, January 4th, 2013

One of the really salient issues about nuclear weapons is that they are expensive. There's just no way to really do them on the cheap: even in an extremely optimized nuclear weapons program, one that uses lots of dual-use technology bought off-the-shelf, to make a nuclear weapon you need some serious infrastructure.

A piece of gold the weight of Little Boy would have cost between $5 and $6 million in 1945. The fissile material for Little Boy cost well over $1 billion. So it would actually have been a pretty good bargain at the time if Little Boy had cost its weight in gold. Also, I knew that making a highly-realistic model of Little Boy in Blender would come in handy someday.

A piece of gold the weight of Little Boy would have cost between $5 and $6 million in 1945. The fissile material for Little Boy cost well over $1 billion. So it would actually have been a pretty good bargain at the time if Little Boy had cost its weight in gold. Also, I knew that making a highly-realistic model of Little Boy in Blender would come in handy someday.

That's not to say that you need to redundantly overspend as much as the Manhattan Project did, or the US did during the Cold War, but even "cheap" nuclear weapons programs are pretty costly. There are a few multinational corporations that could probably pull it off if they were given carte blanche with the technology, but basically you're talking about a weapon that is made for, and by, states. (I'm not, of course, ignoring the possibility of hijacking someone else's infrastructural investments, which is another way to think about theft of fissile material.)

Solid gold B61s aside, this is a good thingIt actually makes nuclear weapons somewhat easy to regulate. I know, I know — the history of trying to control the bomb isn't usually cited as one of the great successes of our time, but think about how much harder it would be if you couldn't spot bomb factories? If every university physics department could build one? If they were really something you could do, from scratch, in an old airstream trailer?

Herman Kahn, 1968, by John Loengard, via Google's LIFE image archive

Herman Kahn, the great thinker of unthinkable thoughts, has a bit about the relationship between cost and doomsday in his 1965 book On Escalationwhich a someone in the audience of a talk I gave last month helpfully sent along to me:1

Assume that it were possible to manufacture a "doomsday machine" from approximately $10 worth of available materials. While it might be "unthinkable" that the world would be destroyed by such a "doomsday machine," it would also be almost inevitable. The only question would be: Is it a matter of minutes, hours, days, months, or years? About the only conceivable way of preventing such an outcome would be the imposition of a complete monopoly upon the relevant knowledge by some sort of disciplined absolutist power elite; and even then one doubts that the system would last.2

If the price of the "doomsday machine" went up to a few thousand, or hundreds of thousands, of dollars, this estimate would not really be changed. There are still enough determined men in the world willing to play games of power blackmail, and enough psychopaths with access to substantial resources, to make the situation hopeless.

If, however, the cost of "doomsday machines" were several millions or tens of millions of dollars, the situation would change greatly. The number of people or organizations having access to such sums of money is presently relatively limited. But the world's prospects, while no longer measured by the hour hand of a clock, would still be very dark. The situation would improve by an order of magnitude if the cost went up by another factor of 10 to 100.

It has been estimated that "doomsday" devices could be built today for something between $10 billion and $100 billion. [Multiply that by 10 for roughly current price in USD]3 At this price, there is a rather strong belief among many, and perhaps a reasonably well-founded one, that the technological possibility of "doomsday machines" is not likely to affect international relations directly. The lack of access to such resources by any but the largest nations, and the spectacular character of the project, make it unlikely that a "doomsday machine" would be built in advance of a crisis; and fortunately, even with a practical tension-mobilization base, such a device could not be improvised during a crisis.

In other words, since Doomsday Machines are phenomenally expensive, and thus only open as options to states with serious cash to spend (and probably serious existing infrastructures), the odds of them being built, much less used, are pretty much nil. Hooray for us! (Nobody tell Edward.) But as you slide down the scale of cheapness, you slide into the area of likelihood — if not inevitability — given how many genuinely bad or disturbed people there are in the world.

Cost and control go hand-in-hand. Things that are cheap (both in terms of hard cash as well as opportunity cost, potential risk of getting caught, and so on) are more likely to happen, things that are expensive are not. The analogy to nuclear weapons in general is pretty obvious and no-doubt deliberate. Thank goodness H-bombs are expensive in every way. Too bad that guns are not, at least in my country.

But area where I start really thinking about this is biology. Check out this graph:

Cost of sequencing a human-sized genome, 2001-2012. From the National Human Genome Research Institute.

Cost of sequencing a human-sized genome, 2001-2012. From the National Human Genome Research Institute.

This graph is a log chart of the cost of sequencing an entire human genome, plotted over the last decade or so. Moore's Law is plotted in white — and from 2001 through the end of 2007, the lines roughly match. But at the beginning of 2008, sequencing genomes got cheap. Really cheap. Over the course of four years, the cost dropped from around $10 million to about $10,000. That's three orders of magnitude. That's bananas. 

I was already reeling at this graph when I saw that Kathleen Vogel has a very similar chart for DNA synthesis in her just-published book, Phantom Menace or Looming Danger?: A New Framework for Assessing Bioweapons Threats (John Hopkins University Press, 2012). (I haven't had a chance to read Kathleen's book yet, but flipping through it, it is pretty fascinating — if you are interested in WMD-related issues, it is worth picking up.)

Everything regarding the reading and writing of DNA is getting phenomenally cheap, really quickly. There's been a blink-and-you-missed-it biological revolution over the last five years. It's been caused by a relatively small number of commercial players who have made DNA sequencing into an automated, computer-driven, cheap process.4 It will probably hit some kind of floor — real-world exponential processes eventually do — but still.

I don't have anything much against DNA sequencing getting cheap. (There are, of course, implications for this, but none that threaten to destroy the world.) DNA synthesis makes me pause — it is not a huge step from DNA synthesis to virus synthesis, and from there to other bad ideas. But as Kathleen emphasizes in her book (and in talks I've seen her give), it's not quite as easy as the newspapers make out. For now. We're still probably a few decades away from your average med school student being able to cook up biological weapons, much less biological Doomsday Machines, in a standard university research laboratory. But we're heading down that road with what seems to me to be alarming speed.

Don't get me wrong — I think the promises of a cheap revolution in biology are pretty awesome, too. I'd like to see cancer kicked as much, and maybe even more, than the next guy. I'm not anti-biology, or anti-science, and I'm not in fan of letting a purely security-oriented mindset dominate how we make choices, as a society. I don't necessarily think secrecy is the answer when it comes to trying to control biology — it didn't really work with the bomb very well, in any case. But I do think the evangelists of the new biology should treat these sorts of concerns with more than a knee-jerk, "you can't stop progress" response. I'm all in favor of big breakthroughs in medicine and biology, but I just hope we don't get ourselves into a world of trouble by being dumb about prudent regulation.

What disturbs me the most about this stuff is that compared to the best promises and worst fears of the new biology, nuclear weapons look easy to control. The bomb was the easy case. Let's hope that the next few decades don't give us such a revolution in biology that we inadvertently allow for the creation of Doomsday Machines on the cheap.

  1. Herman Kahn, On Escalation: Metaphors and Scenarios (Transaction Publishers, 2010 [1965]), 227-228. []
  2. Note the implicit connection here between knowledge and the importance of cheapness — when materials are cheap, knowledge becomes everything. Or, to put it another way, this is why computer viruses are everywhere and atomic bombs are not. []
  3. Here he cites his own On Thermonuclear War, page 175, but in the copy I have, it is page 145, footnote 2: "While I would not care to guess the exact form that a reasonably efficient Doomsday Machine would take, I would be willing to conjecture that if the project were started today [1960] and sufficiently well supported one could have such a machine by 1970. I would also guess that the cost would be between 10 and 100 billion dollars." $10 billion USD in 1960, depending on the conversion metric you use, is something in the neighborhood of 100 billion dollars today, with inflation. []
  4. I thank my friend Hallam Stevens for cluing me in on this. His work is really must-read if you want to know about the computerized automation of sequencing work. []

Origins of the Nuclear Black Budget

Monday, December 5th, 2011

How much does the nuclear complex cost the American taxpayer? In a post last Thursday, Mia Steinle and Danielle Brian of the Project on Government Oversight wrote about the fact that the exact cost of the U.S. nuclear complex remains an unknown figure to taxpayers. ("Taxpayers Left in the Dark When it Comes to Nuclear Weapons Spending," POGO, December 1, 2011.) They describe a debate between the Obama administration and groups trying to come up with realistic estimates for the numbers:

So, how much does the U.S. spend on nuclear weapons? The only way to know for sure—and the only way for Congress to make informed decisions about funding—is for the administration to be more transparent about its nuclear spending, and to make a complete, detailed budget available to the public that includes operations, tactical nukes and other costs borne by the taxpayer. We also need a GAO audit of that budget, because right now, the one thing we do know is that we do not know enough.

There's a way in which billions and billions just seem like lots of numbers with zeros attached to them, though the difference between "$700 billion over ten years" (one estimate) and "$200 billion" (official estimate) is pretty large. Half a trillion is quite a lot of zeros.

What accounts for the disagreement? Interestingly, classification and secrecy appear to play a very small role. While all sides acknowledge that there are some classified programs that aren't being listed, it seems to me that everybody is more or less in agreement that most of the expenditures are more or less in the open. The question is not so much a matter of what's being paid for, but how much everything will eventually cost, and whose figures you trust on that point.

So that's interesting to me. The old debate on the price of the nuclear complex was based on classification and inconsistent record keeping. The new debate is a question of how much these large technical systems are actually going to cost and how many factors over budget they will be — POGO claims that a DoD project costing six times its initial estimate is not unusual.

Read the full post »