Two weeks ago, The Atlantic published an article on its website that made an interesting and provocative claim about the history of the atomic bomb. The thesis, in short, is that the Manhattan Project officials deliberately misconstrued their own history to avoid the general public thinking that the atomic bomb had effects similar to the reviled and banned gas warfare of World War I. If true, that would be rather remarkable: while it is clear that the Manhattan Project personnel did care very much about their own history and how it would affect how people thought about the atomic bomb, an association with chemical weapons has not traditionally been hypothesized as one of the several motivations for this.
“Atomic Bombs,” the original name for the Smyth Report, was meant to be applied with a red stamp. But in the hurry to release it, this was forgotten, and its terrible subtitle became its actual title, hence everyone calling it “the Smyth Report.” It was, in other words, a report so secret that it forgot its own title! The only version with the red stamp applied was the one deposited for copyright purposes at the Library of Congress.
The author, Jimena Canales, is a professional historian of science who I’ve known for a long time. I’ve been asked about the article several times by other scholars who wanted to know whether the thesis was plausible or not. What’s tricky is that most people don’t know enough about the history of Manhattan Project publicity to sort out what’s new from what’s old, and what’s plausible from what’s not. Ultimately there are many parts of this article which are correct, but are not new (as Canales acknowledges in her article); the parts that are novel are, in my view, not likely to be true.
Canales’ article is about the creation of the Smyth Report. The argument is, essentially, that the Smyth Report is overly focused on physics at the expense of chemistry (which is the correct but not new argument), and that the reason it is focused on physics is so that people wouldn’t associate the atomic bomb with chemical weapons (which is the new but I think not correct argument). My problem with the piece is really the last part of it: I just don’t think there’s any evidence that this was a real concern at the time of writing the Smyth Report, and I don’t think it’s necessary to posit this as a reason for the way the report turned out the way it did (there are other reasons).
Those who have read this blog for a while probably know that I find the Smyth Report fairly fascinating and have written about itseveral times. It’s a highly unusual document that sits at several intersections: it hovers between secrecy and openness, it hovers between the end of the Manhattan Project and the beginning of the postwar era. In the remainder of this overly-long blog post, I am going to lay out a thumbnail sketch of the history of the Smyth Report as I understand it, what the key historiographical issues are, and why I disagree with the ultimate conclusions of Canales’ piece.
This is the third blog post I’ve written on the question of presidential nuclear authority. If you have unresolved questions, or feel like I’m taking some things for granted, you might first check out Part I (in which I introduce the problem) and Part II (in which I deal with a few common objections), if you haven’t already seen them.
One of the several projects I’ve been working on for the past several months has, at last, come to fruition. Way back in late November 2016, I got in touch with my friends at NPR’s Radiolab, Latif Nasser and Robert Krulwich, right after I had my Washington Post piece on the question of presidential nuclear weapons authority. The final product is now out, as a podcast given simply the title of “Nukes”:
If you are having trouble using the Radiolab website to get it (and the podcast starts after a 5 minute promotion for another podcast), you can download the trimmed MP3 here.
Radiolab, as many of you probably know, is a show about science and many other things. The pitches they like tend to revolve around interesting people who, traditionally, need to still be alive to be very effective at radio. (And as such, their concerns are often very different from those of historians, who prefer to traffic in the dead.) Latif and I have been friends for a long time now (we were in graduate school together), and have bounced ideas around for a long time, and he has pushed me in the past to find “living specimens” of the nuclear age that illuminate interesting questions.
One of the cases I mentioned in my Post piece was Harold Hering, the Major who was kicked out of the Air Force for asking a “dangerous question” while training to be a Missile Launch Officer at Vandenberg Air Force Base. Hering had asked, in essence, how could he, in his Minuteman missile bunker, know that an order to launch he received from the President had been a legal, considered, and sane one? (And if you want to know exactly what Harold asked, listen to the podcast, where we worked to make sure we really could nail this down as best we could, four decades after the fact.) The fact that his persistence in asking this question, and his lack of satisfaction with the answers, got him drummed out of the service was, I thought, and interesting comment on the nature of what “reliability” means in the context of nuclear weapons personnel. I had gotten interested in Harold’s story because it was discussed in Congressional testimony from 1976, during the only serious hearings that Congress had on this matter, and there was an article from Parade magazine about him appended to the hearings.
It had occurred to me that while Harold was likely quite old, he was probably still alive. I thought it might be worth seeing if I could track him down, and to see if he would be potentially willing to talk about his experiences with me, and to be recorded for the radio. In tracking him down, I thought I might have to utilize all of my Internet-searching, archive-crawling, database-accessing skills. A glance at Ancestry.com’s records made it clear he was born in Indianapolis, and helped me pin down his exact age. A good start, I thought, but with the elderly in particular it can be very hard to get further than that, since they are often not very wired into the modern world.
On a whim, though, before really starting the heavy-duty work, I would put his name into Facebook. Sure enough, there he was: the right age, the right place (still living in Indianapolis), and a Facebook profile photo of him as an USAF officer in the 1970s. So much for my searching skills.
I got in touch with Harold, got in touch with Latif and Robert, and thus started our multi-month process of researching, interviewing, and digging. There were a few issues that we thought would work best for the Radiolab format: the nuclear chain of command, the tensions between automation and human judgment, the question of how one might “remedy” the current situation (assuming one thought it was worth remedying, which I do).
One of the more dramatic sections of Hering’s 1973 journal — where the question he asked got finally translated into a disqualification, delivered in front of his family. “A more false statement has yet to be made,” writes Herald.
I sat in on a number of the interviews, and provided a lot of additional research. I’ve worked with Radiolab in the past, but never quite this close. It was fun. In the process, I got to talk and correspond a bit with not only Harold — which was a complete joy, as was the fact that he had kept a journal of his troubles in the 1970s, and was willing to provide it to us — but also with scholar and former missileer Bruce Blair, US Representative Ted Lieu, and the estimable William J. Perry, the former Secretary of Defense.
I also tried to see how far I could dig into a few of the lingering questions that had kept coming up after my other pieces. One that I really wished I could nail down more, what exactly is the nuclear chain of command? How many people are in between the President and the actual use of nuclear weapons? Where exactly is the “jump” between the “political” wing of the US government (e.g., the Executive Branch) and the “military” wing that actually implements the order?
This is a place where people still had pushed me after my Post piece. How much could one really say about such things, as someone without a clearance? And on what evidentiary grounds could one say it?
The “farthest end” of the chain of command: “1st Lt. Pamela Blanco-Coca, 319th Missile Squadron missile combat crew commander, and her deputy commander, 2nd Lt. John Anderson, simulate key turns of the Minuteman III Weapon System Feb. 9, 2016, in a launch control center in the F.E. Warren Air Force Base, Wyo., missile complex. When directed by the U.S. President a properly conducted key turn sends a ‘launch vote’ to any number of Minuteman III ICBMs in a missileer’s squadron, with two different launch votes enabling a launch. (U.S. Air Force photo by Senior Airman Jason Wiese)”
Blair has insisted (in e-mail to me, and in our interview) that the whole “could the Secretary of Defense refuse an order” question was a red herring. The Secretary of Defense, he insisted, was completely dispensable with regards to the deployment of nuclear weapons. As I noted in my Post piece, there are several descriptions of the nuclear chain of command that imply that the Secretary of Defense is necessary, as the “conduit” (my term) between the political and military worlds. But is it true? Blair emphatically said no — but I never felt completely comfortable just taking his word for it. It’s not that I doubted Blair’s sincerity, or his long history of research and experience with this topic (aside from being a missileer himself, he also spent years researching command and control questions), but I’m a historian, I want a document to point to! Collecting good citations is what historians do.
What’s tricky, here, is that there are clear instances where the Secretary of Defense’s job is defined as translating a presidential order into a military result. And there are places in the descriptions of various components of the US nuclear command and control organization where the uppermost political “unit” is the National Command Authorities, which is defined as the President and the Secretary of Defense. Which has led a lot of authors to insist that there is a big role there, of some sort. And even I entertain the possibility in the Post piece, and in the Radiolab piece (my specific interview was recorded some months ago). The reason is pretty clear — DOD Directive 5100.30 states:
The NCA [National Command Authorities] consists only of the President and the Secretary of Defense or their duly deputized alternates or successors. The chain of command runs from the President to the Secretary of Defense and through the Joint Chiefs of Staff to the Commanders of the Unified and Specified Commands. The channel of communication for execution of the Single Integrated Operational Plan (SIOP) and other time-sensitive operations shall be from the NCA through the Chairman of the Joint Chiefs of Staff, representing the Joint Chiefs of Staff, to the executing commanders.
Which seems to set up the Secretary of Defense as an essential part of the chain. The directive in question is not especially recent (the unclassified version of the directive dates from 1974), and it doesn’t clarify exactly how important the Secretary of Defense might be.1
But over the last few weeks, while working on this episode and my own further digging into the matter, I have become convinced that the weight of the open evidence points to the idea that Blair is correct — the Secretary of Defense is not just unnecessary, but not even in the nuclear chain of command. What convinced me?
First, I found perhaps the only piece of military doctrine that actually explained, in a clear and concise fashion, how a nuclear order would be carried out. And it’s not some ancient Cold War archival document… it’s from 2015! On the website of the USAF’s (appropriately named) Curtis E. LeMay Center for Doctrine Development and Education, one can find ANNEX 3-72 NUCLEAR OPERATIONS, last updated in May 2015. It states, in a clarity that (after reading a lot of DOD doctrine) makes me want to weep with joy, despite the message:
The President may direct the use of nuclear weapons through an execute order via the Chairman of the Joint Chiefs of Staff to the combatant commanders and, ultimately, to the forces in the field exercising direct control of the weapons.
Which seems pretty definitive. The order jumps immediately from the President to the military, in the form of the Chairman of the Joint Chiefs of Staff, and from there percolates through the system of command, control, and communication to the various people who actually turn the keys and put the “birds” into the air.
Could the doctrine be wrong? Presumably such things are carefully screened before being offered up as official doctrine, and it seems about as clear as can be, but it’s always possible that something got mangled. But one other useful piece of evidence is that we asked Perry, the former Secretary of Defense, at point blank whether the Secretary of Defense was in the chain of command. The answer was a clear “no.” Perry explained that while, presumably the Secretary of Defense would express opinions and given counsel, the President was under no legal obligation to take such counsel, and the objection of the Secretary of Defense had no bearing either legally or practically.
I don’t know what your standard of evidence about such a question might be, but personally I find the testimony of a former Secretary of Defense, combined with a reasonably up-to-date piece of Air Force doctrine, to settle the case for me (at least, pending more evidence). No other assertions about the nuclear chain of command that I’ve seen have quite that kind of weight behind them.
Does this change our initial question, about who might say no? It shifts the attention away from the civilian Secretary of Defense (which is a civilian job, whether or not the person in the role is a retired General, as is currently the case) to the military position of the Chairman of the Joint Chiefs of Staff. Could such a person disobey the order? Perry suggested they might in practice try to, but there would be legal consequences (e.g., a court martial).
I gave a talk on these issues last week at the Belfer Center for Science and International Affairs at the Harvard Kennedy School (where I was a postdoctoral fellow in the Managing the Atom Program some years ago, and where I maintain an active affiliation), and two members of the audience (one an Air Force officer, the other my grad school colleague Dan Volmar, who works on the details of nuclear command and control history) pointed out that when doctrine says “the Chairman of the Joint Chiefs of Staff,” it is usually referring to a staff and not an individual person. Which is to say, it doesn’t necessarily indicate an individual personage, but instead indicates a web of people that are connected to the authority of that personage. I am not sure what would apply in this kind of extraordinary situation, but I thought it was an interesting point to bring up.
A slide from my Belfer Center talk on nuclear chain of command (in the talk, I end up removing, the SecDef from the chain, per the issues discussed in this post) — the graphical whimsy is a purposefully a little bit of levity on a serious topic. Graphics created using Keynote’s shape templates (yes, the hair is an upside-down speech bubble). And yes, I know I have “black boxed” C3 (command, control, and communications) operations in a “and now a miracle occurs” fashion.
I have even less faith than before in the idea that an order of such would be disobeyed. Not that I think the military is eager to deploy nuclear weapons — I’m sure they are not, and in fact I tend to feel that they have in the post-Cold War come to realize at some deeper level the risks associated with such weapons and the difficulties they impose on their services. But I do think that the nuclear command and control system is set up, both practically and doctrinally, to avoid asking the questions that are seen as being in the purview of the “political” side of the equation. From the same “Annex 3-72” (my emphasis):
The employment of nuclear weapons at any level requires explicit orders from the President. The nature of nuclear weapons — overwhelmingly more significant than conventional weapons — is such that their use can produce political and psychological effects well beyond their actual physical effects. The employment of nuclear weapons may lead to such unintended consequences as escalation of the current conflict or long-term deterioration of relations with other countries. For this reason above all others, the decision whether or not to use nuclear weapons will always be a political decision and not a military one.
Now, obviously conditions would dictate varying responses. I have faith that an “obviously bonkers” order would be somehow avoided (e.g., a frothing, “nuke them all, ha ha ha,” sort of thing). I’m not worried about that situation (it’s not outside the realm of human possibility — all humans are fallible, many develop various forms of mental illness, etc.), but I am worried about what I consider to be “ill-advised” orders, or “bad idea” orders, or “spur of the moment” orders that are considerably less apocalyptic (at least on their surface) than, say, a full nuclear exchange.
What would the military do in such a situation, if a correctly authenticated, correctly-formatted “execute order” came to them on their secure channels? I don’t have faith they’d abort it. Maybe you do — that’s fine, and I appreciate the company of optimists. But I just want to point out, the notion that thesystemwon’t work as intended is not a real “check.” It’s just hoping things will break in a way that would be convenient. I think we can do better, and I think that the consequences associated with the possibility of the rash use of nuclear weapons by an American President — any President — large enough to warrant trying to make a better (if not perfect) system, even if one thinks the probability of such a thing happening is low.
Blair’s interpretation of 5100.30 is that it was about removing the service chiefs and JCS from decision-making positions, and established the JCS as (only) an intermediary. He says that “on the inside” it is understood that the Secretary of Defense is just an advisor, like the Secretary of State. (E-mail to me.) On the history of 5100.30, see William Burr, ed., “Top Air Force Official Told JCS in 1971: ‘We Could Lose Two Hundred Million People [in a Nuclear War] and Still Have More Than We Had at the Time of the Civil War’,” National Security Archive Electronic Briefing Book No. 580 (February 15, 2017). Later (post-1970s) updates of DODD 5100.30 appear to be classified, and its last iteration was in 2006, and it was cancelled in 2014. See Reference (a) in DODD 3700.01, its successor document, which has nothing illuminating on the chain of command question in it.
I asked Blair if he had any further thoughts that I ought to share on this, and he sent me the following in an e-mail:
5100.30 mainly tried to change the priorities of command-control investment. Dep Sec David Packard (justifiably) feared that the U.S. nuclear command structure would collapse under attack and render U.S. retaliation impossible. He correctly understood that this acute vulnerability stemmed in large part from the chronic neglect of the national command centers and communications links by the chiefs of staff and civilian secretaries of the military departments [services] as well as the combat CINCs [unified and specified commanders], all of whom invested in their own service-specific command and communications networks at the expense of the national system and of interoperability. Packard promulgated 5100.30 in a vain attempt to reverse these priorities and compel the service chiefs and CINCs to invest seriously in fixing vulnerabilities at the top and make systems compatible up and down and across the chain of command. The directive makes the CJCS responsible for implementing this directive.
In operational respects, 5100.30 sought to reinforce this change of priorities by requiring the creation of robust capabilities at the national level to ensure that the NCA could survive a nuclear attack and effectively communicate nuclear war orders not only to the CINCs but also directly to the executing forces at the level of the individual commanders of strategic submarines, bombers, and land-based missiles. Again, the JCS was assigned responsibility for implementation — i.e., ensuring the transmission of the “go-code” (as well as termination) messages. The Deputy Director for Operations’s war room in the Pentagon was the assigned conduit. The service chiefs of staff and civilian secretaries were bypassed by this flow chart, as was the Secretary of Defense, who served merely as an advisor to the President, like the Secretary of State, and who might or might not be asked his opinion at the moment of truth. Even the CINCs became less critical in execution since the launch order was supposed to flow directly from the Pentagon war room (the National Military Command Center) to the firing crews at the bottom of the chain of command.
Five years ago today I introduced the NUKEMAP. It feels practically like yesterday — how fast that has flown! I occasionally get college students, not even brand new ones, who tell me that they used it in high school to do reports. That makes me feel… well, like I’ve contributed something, along with feeling old. So that’s not bad. I’ve been behind on posting for awhile now, and am behind on several things at the moment (lots of irons in the fire, plus the debilitating power of a news cycle that seems to change by the minute), but I wanted to put up something about the NUKEMAP.
NUKEMAP and NUKEMAP3D page views, exported from Google Analytics and cleaned up a bit, with a few of the “known” moments of virality indicated. Note how the “baseline” had steadily increased over time.
Some statistics: NUKEMAP has been the host of over 99 million virtual detonations, according to its internal logs. Every detonation, except for ones where people have opted-out of logging, is logged. As I’ve said before, I don’t record enough information for it to be non-anonymizing, but it’s interesting to see things like where people nuke, and what they do with the tool. According to Google Analytics, there have been (as of this checking) over 25 million pageviews, over 20 million of those unique pageviews (e.g., not people coming back and using it multiple times in one session). The usage of the site predictably flares up in certain moments of “virality” (for the 70th anniversary of Hiroshima, over 500,000 people used it over two days), and still have sharp moments of heavy traffic every few months. More interesting and important to me is that the site’s “slow days” are now not so slow. When it started, a “slow day” was a few thousand people using it. Today, it’s more like 15,000-20,000 people using it. And, for the most part, people are really using it: the average time on page is 5 minutes, which I think is pretty healthy for a web visualization used by tens of thousands of people a day. That means people are doing more than just clicking and glancing — they’re actually trying things out.
NUKEMAP3D is, for the moment, moribund. Google unceremoniously discontinued support for the Google Earth Web Plugin (the code on their end is just kaput), and no adequate substitute has yet emerged. There are some ways of crudely rendering a 3D planet on the web, but none that support buildings and skylines the way Google Earth did, and that is the whole point of NUKEMAP3D. However, I am developing a temporary substitute which is almost ready to roll out: it will allow you to export any NUKEMAP settings to a KMZ file which you can open in the Google Earths standalone program, and it will support mushroom clouds among other interesting features.
“Alas, poor NUKEMAP3D! I knew him, Horatio…” Don’t worry, NUKEMAP3D isn’t really dead, just waiting for better circumstances…
Some reflections: I still remain surprised that NUKEMAP has been as popular as it was. The idea of drawing concentric circles over a map is not a new one, and mine was not even the first web one. Heck, it wasn’t even the first web one for me — in 2005 or so I made a terrible crude version using MapQuest (remember them?) and PHP, and it wouldn’t have been sustainable to use (it literally used PHP to draw circles over static images from MapQuest, so it was very server-intensive by the standards of the day). But I did try to make a version that was easier to use than any of the other ones that were out there, and gave more intuitive, useful information. And when I upgraded NUKEMAP in the summer of 2013, I really did think it was contributing new possibilities: much more flexible detonation options, casualty estimates, a fallout model.
I still give talks about NUKEMAP all the time, whether to large groups (I was on a panel with Noam Chomsky a few years ago, talking about NUKEMAP), or to individual reporters (I did another interview on it just yesterday), or to small groups of students (I Skyped into a high school class a few weeks ago to talk about it, and how it was made, and how these students should not think of it as something beyond their capabilities to put together, something I don’t mind doing if I can make the time for it). I teach a course regularly (“Visualizing Society,” a sort of anarchistic data visualization/science and technology studies course) where I show students how to build NUKEMAP-like applications for other sorts of social phenomena. I still make updates and plans for updates to it: there are several projects in the works, including “refreshing” the interface a bit (don’t worry, it won’t end up looking painfully “trendy”; the blog could probably use a refresh, too), translating it into other languages (which requires more back-end coding than you might expect), and adding new substantive features (I have almost put the final touches on a nuclear burning model and better support for multiple detonations).
For me, the “holy grail” would be something that would let you see something like the Defense Civil Preparedness Agency made in 1973: a “personalized” view of what different damage looked like, from the street level. The technology for this isn’t quite here yet, but it’s not that far away, either.
I have a very long “wish list” of things that would be interesting to add: EMP features, a dynamic (time-sensitive) fallout model, support for the effect of terrain or dampening by buildings, and so forth. I do have some students who occasionally work for me, especially in the summer, on aspects of these issues, and some of this work may eventually make it into future versions of the NUKEMAP. I’m also interested in translating the NUKEMAP concept — this “personalizing” of nuclear weapons effects — into non-web domains as well. The main difficulty here is time: NUKEMAP is still a mostly one-man operation (imagine me in the salt mines, toiling out Javascript), and this one-man is (to his delight) admirably busy with a lot of things. I am very positively supported in this work by my university, I should say, and the College of Arts and Letters at the Stevens Institute of Technology has been paying the ever-increasing fees associated with running a popular website since I got here, and encouraging me to do even more with it.
I suppose one thing that I’m grateful for is that I’m not yet even slightly bored with any of it — I still find talking about it interesting, I still find it a model of how we might consider science communication to look in our present age. I strongly believe, and will evangelize about to anyone who asks me to (as many have found, probably without realizing what they were getting into), that there is something different about providing a sort of “simulation” to a user and saying, well, you figure out how this works, as opposed to a more didactic mode of education like lecturing. This has strong shades of “active learning,” but I’m not just talking about an approach to the classroom. One nice thing about tools like NUKEMAP is that I can see (through referring links) how people are using them. My favorite example, and this comes up all the time, is when people use it to argue with other people on the Internet. Someone will say, wouldn’t a nuclear bomb do X? And someone says, well, the NUKEMAP says it will be more like Y. And there’s this kind of “calibration” of understanding, as I think of it, that starts to narrow down what these weapons do and don’t do. (And it goes both ways: most people think they are more powerful than they are, but some think they are less powerful.) The NUKEMAP model, as I discuss in its FAQ, isn’t perfect by any means: in some circumstances it probably overestimates the effects (by not taking into account a lot of local variables), in others it probably underestimates them, and the “real world” is much more chaotic than a simple model that can run in your browser can account for, no doubt. But it helps to concretize the experience, the order of magnitude. I think there’s a lot of value in that, when we’re talking about something so removed from everyday human experience (thank goodness) as a nuclear weapon detonation.
And I think this is a model we need to really do more to export to other domains: nukes are one thing in our society that people have trouble really understanding on an intuitive level, but there are plenty more. This is what my “Visualizing Society” class is all about, at its core: finding ways to make interactive data visualizations or simulations that shed light on complex real-world issues. The technical bar for doing these things is lower than most people realize; if I can teach undergraduates (very good and often technically-inclined undergraduates, to be sure, but often ones with no coding experience) the basics of this over the course of a semester, then it can’t be that hard.
The original “NUKEMAP” — Hiroshima, before and after, from the view of a nuclear bombardier.
My main frustration with NUKEMAP as a communication tool is that the top-down, concentric-circles approach is the view of the military planner. It’s the view of the nuclear targeteer, or as a friend and collaborator put it earlier this week, it’s the view of real estate. It’s not the view of the person on the ground, it’s not the view of the survivor, it’s not the view of the victim. NUKEMAP3D did provide some aspects of that, but the Google Earth plugin, for its communicative benefits, was clunky to use (the 3D interface was not straightforward), required a special installation, and it was never as popular as the regular NUKEMAP. (I was, however, still impressed that some 3 million people used it over its lifetime.) I’m hoping that some future projects I have in mind (no spoilers, sorry) will address these issues more directly and more intensely.
Anyway, more is on the horizon, as ever, and it is just a matter of figuring out how to get it all done. More NUKEMAP, more NUKEMAP-like creations, more work. I’m grateful for NUKEMAP: what started out a literally two-day coding job (one resting, of course, on a decade of coding experience, and even some actual code that I had written a long time ago, to be sure) has turned into something of a guiding idea for a career. It definitely increased the popularity of my blog (whose traffic is admirably high for an academic, despite the fact that I am greatly remiss in updating it lately), and became a selling-point for the kinds of hybrid technical-historical-analytical projects that I never knew I had wanted to spend my life working on (though I did have some inklings). Anyway, much more is coming. When I go silent, don’t think, “what’s happened to him?” Instead, think, “what’s he getting ready for us, next?” There’s a lot in the pipeline.
by Alex Wellerstein, published December 23rd, 2016
This ended up being part one of a three part arc: Part I (in which I introduce the question), Part III (in which I talk about some “new” discoveries)
It’s been a busy month; aside from “regular work” sorts of duties (teaching, grading, writing, e-mailing, programming, book reviews, grant proposals, oh my!), I’ve been sucked into various discussions relating to presidential command and control after my last post, which got me a solicitation to write an op-ed for the Washington Post on presidential authority to launch nuclear weapons. I haven’t really gotten around to screening all the comments to my blog post, and Post piece has 1,200 comments that I’m just not going to bother trying to wade into. I thought though that I would post a few quick responses here to common comments I’ve gotten on both pieces.
The print version of my Post article (December 4, 2016). Thanks to my DC friends for sending me print copies — apparently one cannot buy the Post anywhere in Hoboken. You can read the online version here.
Both pieces are about the history and policy of presidential authorization to use nuclear weapons. In a nutshell, in the United States the President and only the President is the ultimate source of authority on the use of nukes. This is entirely uncontroversial, and the articles describe the history behind the situation.
The trickier questions come up when you ask, can anyone stop nuclear weapons from being used if the president wants to use them? Everything I’ve been able to find suggests that the answer is no, but there are ambiguities that various people interpret differently. For example, there are two separate questions hidden in that first one: can anyone legally stop the president, and can anyone practically stop the president? I will get into these below.
The most curious response that I’ve heard, both in person and second-hand, are people who have heard what I’ve said about this, and say, “that can’t be true, that would be a dumb/crazy way to set things up.” This is often a purely emotional response, not one based on any research or specialized knowledge — a pure belief that the US would have a “smarter” system in place. I find it interesting because it is a curious way to just reject the whole topic, some sort of mental defense mechanism. Again, everything I’ve found suggests that this is how the system is set up, and in both the blog post and the Post article I’ve tried to outline the history of how it got to be this way, which I think makes it more understandable, even if it’s still (arguably) not a great idea.
And, of course, because it featured the name “Trump,” a somewhat hyperbolic headline (which I didn’t write, but don’t really hate — it elides some caveats and ambiguities, but it’s a headline, not the article), and is in the Washington Post, there were a lot of people who wondered whether this was just a partisan attack. And amusingly had a number of people accuse me of being a Post employee, which I am decidedly not. My writing has also been sometimes referred to as “hysterical,” which is an interesting form of projection; to my eye, anyway, it is intentionally pretty sober, but I suppose we see what we expect to see, to some degree.
Interpreting a Trump tweet is no easy matter, and serves as sort of a political Rorschach test. The above is either completely in line with Obama’s nuclear modernization plan, or a call for something entirely different. I guess we’ll see…
It’s true that I think this issue is particularly acute with regards to Mr. Trump; if nuclear war powers are vested in the person of the executive, then the personality of the executive is thus extremely important. And I think even his supporters would agree that he has a reactive, volatile, unpredictable personality. He broadcasts his thin-skinnedness to the world on a daily basis. And a president who complains on a weekly basis on his portrayal on Saturday Night Live is, let’s be straight, thin-skinned. To be offended by comedic parody is beneath the station of the job, something that any American president had better get beyond.
But frankly I would be just as happy to have been talking about this in a Clinton administration, and would have been happy to talk about it during the Obama administration. The issue, and my position, still stands: I think that vesting such power in one human being, any human being, is asking a lot. And this has nothing to do with how reliable they seem at election-time: we have historical instances of presidents who had health problems (Wilson), were heavy users of painkillers (Kennedy) or alcohol (Nixon), or who were later discovered to be in the early stages of mental decline (Reagan). There is no reason to suspect that the president you elect one year will continue to be that person two or three years later, or that they will be totally reasonable at all times. And I don’t think anyone of any political party in the USA would go out on a limb to suggest that the US electorate will always elect someone whose can bear all that responsibility. So this doesn’t have to be, and shouldn’t be, a partisan issue, even if Trump in particular is getting a lot of people to start talking about it again.
One of the responses I’ve heard is that no further “checks” on presidential nuclear command authority are needed because any president who wanted to use nuclear weapons unilaterally, against the judgment of their advisors, would be agreed-upon as “insane” and thus could be removed from office under Section 4 of the Twenty-Fifth Amendment of the US Constitution. This is, I think, not adequate.
For one thing, the procedures are understandably complex and require a lot of people to participate — it is appropriately difficult for a president to be declared “unable to discharge the powers and duties of his office,” because if it were easy, that would be an easy way to dispose of an unpopular president. So it is not the sort of thing that can be made to go into immediate effect with quick turn-around, which really does not help us much in the nuclear situation, I don’t think.
Secondly, while the “insane president” idea often dominates the discussion here, that is an extreme and not entirely likely case. I am much more worried about the “president with bad ideas” approach, possibly a “president with bad ideas supported by a few advisors” approach. There are many nuclear-use scenarios that do not involve an attempted preemptive attack against Russia, for example. Not all will be “obviously insane.” And even a president who advocated first-use would not necessarily be “unable to discharge the powers and duties of his office” according to the people who need to agree to such a statement. This is a cumbersome, high-stakes approach; if the only way to stop a president from doing dumb things with nuclear weapons is to kick them out of office on the basis of them being medically unfit, that is a very difficult bar to climb to. (And we can add impeachment under the same objection — that is not a fast nor straightforward process, nor is it any kind of obvious deterrent to the kind of president who might consider using nuclear weapons in the first place.)
Results of a poll taken after the first presidential debate, on whether the candidates could be trusted with nuclear weapons. There are many ways to read this, but I think at a minimum we can say that when substantial percentages (much less a slim majority for one of them) of Americans believe that neither major-party nominee can be trusted with nuclear weapons responsibility alone, it’s time to rethink whether we should have a system that invests that decision completely in the president.
Another objection I’ve gotten is the “maybe there are checks that you just don’t know about, because they’re classified.” Fair enough, and I acknowledged this in my writings. Much about the procedures involved are classified, and there are good reasons for that. If an enemy knew exactly how the system worked, they could potentially plot to exploit loopholes in it. But I have two main responses on this.
First, if complete knowledge in the nuclear realm was necessary to talk about nuclear policy, then literally nobody could or would talk about it. Which means democratic deliberation would be impossible. So those of us without clearances can, and should, talk about what we do and don’t know, openly on both fronts. I try to make very explicit where my knowledge begins and ends. I would be completely thrilled if these discussions led to official clarifications — I think these issues are worth it, and if my understanding is wrong, that would be great.
Second, it strikes me as an act of tremendous optimism to assume that things in the US government are more rationally run than all signs indicate they appear to be. The study of nuclear history is not a study of unerring rationality, of clear procedures, or of systems set up to guarantee wise decision-making. It is easy to document that our command and control systems have been optimized towards three ends: 1. preventing anyone but the president from using nuclear weapons in an unauthorized fashion, 2. reliability of response to threats and attack, and 3. immense speed in translating orders to action. None of that suggests we should assume there are elaborate checks and balances in the system. My view: unless positive evidence exists that a government system is sufficiently rational, to assume it is sufficiently rational requires tremendous faith in government.
Lastly, to get at the strongest of the responses: the president is the only person who can order nuclear weapons to be used, but doesn’t the execution of that order require assent from other people to actually get translated into action? In other words, if the president has to transmit the order to the Secretary of Defense (as some, but not all, descriptions of the process say has to occur), and the Secretary of Defense then has to transmit it to the military, and the military has to transmit it into operational orders for soldiers… aren’t there many places in that chain where someone can say, “hey, this is a terrible idea!” and not transmit the order further?
In thinking about this, I think we have to make a distinction between a legal and a practical hinderance. A legal hindrance would be the possibility of someone being able to say, legally and constitutionally, “I refuse to follow this order,” and that would stop the chain of command. This is mentioned in the 1970s literature on presidential authority as a form of “veto” power. It is not at all clear that this is legally allowable in the area of nuclear weapons — it is, to be sure, an ambiguous issue of constitutional, military, and international law. I have seen people assert that the use of nuclear weapons would be unquestionably a war crime, and so any officer who was given such an order would recognize it as an illegal order, and thus refuse to obey it. I don’t think the US government, or the US military, sees (American) use of nuclear weapons as a war crime (a topic for another post, perhaps), and whether you and I do or not matters not at all.
And from a practical standpoint, we know the system is set up so that the people at the very bottom, the people “turning the keys” and actually launching the missiles, are trained to not question (or even deeply contemplate) the orders that reach them. They are trained, rather explicitly, that if the order comes in, their job is to execute it — not quite like robots, but close-enough to that. The speed and reliability of the system requires these people to do so, and they are not in a position to inquire about the “big picture” behind the order (and would not presume to be qualified to evaluate that). So if we take that for granted, we might ask ourselves, at what level in the hierarchy would people be asking about that? One can imagine a lot of different possibilities, ranging from a continuum of second-guessing that was fairly evenly gradated towards the “top,” or one that was really “band-limited” to the absolute top (e.g., once the order gets made by the president, it is followed through on without questioning).
I suspect that even the military is not 100% sure of the answer to that question, but I suspect that the situation is much more like the latter than the former. Primarily because, again, the US military culture, especially regarding nuclear weapons, is about deference to the authority of the Commander in Chief. Once you get beyond a certain “circle” of people who are close to the president, like the Secretary of Defense, I would be very surprised if the people in the nuclear system in particular would buck the order. The system and its culture was built during the cold war, focused on rapid translation between order and execution. Until I see evidence that suggests it has radically transformed itself since then, I am going to assume it acts in that way still. And again, everything I have seen suggests that this is still the case. As former CIA and NSA head Michael Hayden put it before the election: “It’s scenario dependent, but the system is designed for speed and decisiveness. It’s not designed to debate the decision.”
OK, but in practice, couldn’t the Secretary of Defense just refuse to act? Here it really becomes necessary to know how the system is set up, and I just don’t think enough information is out there to be definitive. From what I understand, the main role of the Secretary of Defense is to authenticate the order — to say, “yes, the president made this order.” Are there ways to get around that requirement, or around a stubborn Secretary of Defense? A practical one would be to just fire him on the spot, in which case the requirement for authentication moves down a notch in the Department of Defense succession rank. It could go onward and onward down the line, I suppose. But more practically, I have heard it suggested (from people who study such things) that there are protocols by which the president could bypass the Secretary of Defense altogether and communicate directly with the National Military Command Center to communicate such an order. I am still looking into what we can say about such things with conviction, but it would not surprise me if there were contingencies in place that allowed a president a more direct means of sending such orders, as part of the goal of making the system especially resilient in times of crisis (when appropriate representatives from the Department of Defense may not be available).
There is a famous anecdote about Secretary of Defense James Schlesinger requesting that all nuclear commands from Nixon in the final days of the administration be routed through him. It is not clear that it is true, and may be exaggerated greatly, but it is often cited as a means that a Secretary of Defense could override presidential authority. Even if it were true, it is ad hoc, probably illegal, and a pretty thin “check” to rest one’s hopes upon. And I would suggest that about any assertion that “practical” constraints exist in the form of individuals refusing to follow orders — they’re highly optimistic. Especially since there is a lot of options other than a raving president shouting “nuke them all!,” which would be pretty easy to disobey. There are many more scenarios that do not involve obvious insanity, but could still be terrible ideas.
The Lieu-Markey bill for limiting presidential power regarding nuclear weapon use.
In my Post piece, I discussed possible resolutions. None could be completely satisfying; the nuclear age is defined by that lack of total certainty about outcomes. But there have been proposals about requiring positive assent from more people than just the president for any kind of first-use of nuclear weapons. I discuss the Federation of American Scientists’ proposal in my Post piece, and the fact that you could use that as a template for thinking about other kinds of proposals. I am not wed to the idea of getting Congressional approval, for example. Frankly I’d be happier if there were a legal requirement that would codify the Secretary of Defense’s veto power, for example. One can productively debate various options (and I’m still thinking about these questions), and their legality (it is a tricky question), but I still think it would be a valuable thing to give people at the highest levels something non-ad hoc to fall back upon if they wanted to actively refuse to obey such an order. The common objection I’ve heard to such an idea is, “maybe we don’t need it, there might be hidden checks in place,” which is not much of an objection (relying on optimism in the system).
Congressmen Ted Lieu (D-CA) and Senator Ed Markey (D-MA) introduced a bill last September that would require a Congressional declaration of war before first-use of nuclear weapons would be allowed. I don’t think that’s necessarily the right approach (Congress has not issued a declaration of war since World War II, so this is effectively just a prohibition on first-strike capability, which will lead the military, defense establishment, and most security scholars I know to definitively oppose the idea), but I hope that this might serve as a place to revisit and discuss these issues formally in a legislative setting. I am not convinced I have the perfect policy solution, but for me an idealized law would have provisions that allowed for first-use in emergency circumstances, with at least one other human being (preferably more, though not an impractical number) having to actively agree with the order (and having veto power over it). I think this is a rather modest suggestion. It would not completely rule out first-strike possibilities — nothing would, save lack of nuclear capability altogether, and that’s a separate can of worms — but it would allow the American people, military, and political establishment to know that no single human being would be shouldering that responsibility alone.
by Alex Wellerstein, published November 18th, 2016
This ended up being part one of a three part arc: Part II (in which I address several specific criticisms), Part III (in which I talk about some “new” discoveries).
I’m in the process of writing up something more substantial about nuclear weapons and the 2016 Presidential election, but I keep getting asked one thing repeatedly both in person, over e-mail, and online: “Are there any checks in place to keep the US President from starting a nuclear war?”
What’s amazing about this question, really, is how seriously it misunderstands the logic of the US command and control system. It gets it exactly backwards.
A recent Tweet by the USAF expresses US nuclear doctrine in a nutshell: “Always on the ready is an understatement when you are providing #POTUS with the ability to launch ICBMs.” (November 17, 2016) Hat tip to Alexandra Levy of the Atomic Heritage Foundation for bringing this one to my attention.
The entire point of the US command and control system is to guarantee that the President and only the President is capable of authorizing nuclear war whenever he needs to. It is about enabling the President’s power, not checking or restricting him. As former Vice President Dick Cheney put it in 2008:
The president of the United States now for 50 years is followed at all times, 24 hours a day, by a military aide carrying a football that contains the nuclear codes that he would use and be authorized to use in the event of a nuclear attack on the United States.
He could launch the kind of devastating attack the world has never seen. He doesn’t have to check with anybody, he doesn’t have to call Congress, he doesn’t have to check with the courts.1
This isn’t new; it shouldn’t come as a surprise to anyone. This has been discussed since the 1940s. And yet, people today seem rather shocked to hear it, even very educated people.
To be sure, the official doctrine that I have seen on the Nuclear Command Authority implies that the President should be given as much advice as possible from the military, the Department of Defense, and so on. But nothing I have seen suggests that this is any more than advisory — and the entire system is set up so that once the President’s order is verified and authenticated, there are meant to be only minutes until launch.2
Diagram of the various US Nuclear Command, Control, and Communication (NC3) Systems, as of 2016. From Nuclear Matters Handbook (2016).
It isn’t entirely intuitive — why the President, and not someone else, or some combination of people? Why not have some kind of “two-man rule,” whereby two top political figures were required to sign off on the use before it happened? The two-man rule is required for commanders to authorize nuclear launches, so why not the Commander in Chief?
To understand why this is, you have to go back and look at the history of how this doctrine came about. Today we tend to discuss this in terms of the speed in which a retaliation would be necessary in the event of a crisis, but the debate wasn’t originally about expediency at all, but about an understanding of Constitutional power and the inherently political nature of the bomb. I see the debate about the (un-)targeting of Kyoto, in mid-1945, as the first place where some of these questions started to get worked out. Presidents generally do not pick targets in war. That’s a general’s job. (Like all things in history, there have, of course, been exceptions.) But when it came to the atomic bomb, the civilian branch of the executive government (personified here by the Secretary of War, Henry Stimson), demanded veto power over the targets. The military (here, General Leslie Groves) pushed back, asserting that this was a military matter. Stimson insisted, and eventually got the President’s personal ear on the matter, and that was that. Truman, for his part, while he did not authorize the actual bombing in any explicit way (he was shown the bombing order, but he did not issue it nor was his approval required, though he could have vetoed it), did, on August 10th, re-assert nuclear authority by prohibiting future bombing activity without his explicit permission.
One can tell that the relationship between General Groves (left) and David Lilienthal (right) was not exactly the smoothest. Photo by Ed Westcott.
From that point forward, the President made very explicit that his office was in charge of the atomic bomb and its uses, not the military. It was not a “military weapon,” which is to say, it was an inherently political weapon, one that needed to be handled by that most inherently political office, the Presidency. This became the framework for talking about domestic control over nuclear weapons in the 1940s, the civilian vs. military split. It was believed that only an elected civilian could make the call for this of all weapons. Truman himself put it to David Lilienthal in 1948:
I don’t think we ought to use this thing unless we absolutely have to. It is a terrible thing to order the use of something that, that is so terribly destructive, destructive beyond anything we have ever had. You have got to understand that this isn’t a military weapon. It is used to wipe out women and children and unarmed people, and not for military uses. So we have got to treat this thing differently from rifles and cannons and ordinary things like that.3
In the early days, this civilian-military split was actually enforced at a physical level, with the non-nuclear parts of the weapons kept by the military, and the nuclear parts (the pits) kept by the civilian Atomic Energy Commission. By the end of the Eisenhower administration, changes in doctrine, technology (sealed-pit weapons), and fears (e.g., a Soviet “sneak attack”) had led to 90% of the nuclear weapons transferred into the hands of the military, making the civilian-military distinction a somewhat theoretical one. Eisenhower also “pre-delegated” the authority to start nuclear war to several military commanders on the front lines, on the idea that they would not have time to call back to Washington should Soviet tanks start pouring into Western Europe. (So while the President is the only person who can authorize a nuclear attack, he can also extend that authority to others if he deems it necessary.)
The Kennedy administration, looking to assert more positive control over the beginning of a nuclear conflict (especially after the Cuban Missile Crisis, which raised the real possibility of a low-level misunderstanding “escalating” in times of uncertainty), requiring the weapons themselves to have sophisticated electronic controls (Permissive Action Links) that would prevent anyone without a coded authorization to use them. There is more to these stories,4 but I just want to illustrate a bit of what the “control” debate was really about: making sure the President, and only the President, was ultimately the one making decisions about the bomb.
A retired “nuclear football” suitcase, from which the President can authorize a nuclear attack. Photo credit: Smithsonian Institute/Jamie Chung, via Wikimedia Commons.
I have been asked: would the officer carrying the “football” actually go forward with a nuclear attack, especially if it seems heedless or uncalled for? (The “nuclear football” is the special computer that, once the nuclear “codes” are inputted into it, somehow electronically starts the sequence of events that leads to the weapons being used.) Which I find lovably optimistic. The entire job of the person carrying the football is to enable the President to launch a nuclear attack. They would not presume to know the “big picture” of why the President was doing it — they are not a high-level military or policymaker. They are going to do their job; it is what they were chosen to do.
Would the military second-guess the President, and override the order? I mean, anything is possible — this has just never happened before, so who knows. But I am dubious. In 1973, Major Harold Hering was fired for asking, “How can I know that an order I receive to launch my missiles came from a sane president?” Not because it is a fireable offensive to imply that the President might not, at all times, be entirely capable of making such an order, but because to start to question that order would mean to put the entire credibility of the nuclear deterrent at risk. The entire logic of the system is that the President’s will on this point must be authoritative. If people start second-guessing orders, the entire strategic artifice breaks down.5
So is there any check on the President’s power to use nuclear weapons? Well, technically the US election process is meant to be that check — don’t elect people you don’t trust with the unilateral authority to use nuclear weapons. And this, indeed, has been a theme in numerous US elections, including the most recent one. It is one issue among many, of course.
There is, of course, no big red button. There are lots of other, smaller buttons, though. Source: Ren and Stimpy, Space Madness.
Do I personally worry about an unhinged, unthoughtful President using nuclear weapons heedlessly? Sure, to some degree. But not as much as I worry about other damage that such a President will do to the country and the world (the environment, economy, social fabric, international order, and human rights are higher on my list of concerns at the moment). Which is to say, it’s on the list of things one might worry about (for any President, but certainly the next one), but it’s not my top worry. Ultimately I do have some faith, perhaps unearned, that even someone who is woefully under-educated about world affairs, strategic logic, and so on, will come to understand rather rapidly that it is in the United States’ best interests not to break the nuclear taboo.
The United States benefits from the taboo disproportionately: should the threshold for nuclear use be lowered, we would be the ones who would suffer the most for it, because we tend to put our cities and military forces and everything else in centralized, easy-to-take-out-with-a-nuke sorts of arrangements, and because we enjoy a powerful conventional military power as well. We have the luxury of a nuclear taboo, in other words: we don’t have to use nukes to get what we want, and indeed in many situations nukes are just not as useful as they might at first appear.
So only a true idiot would think that using nukes foolishly would actually be a useful thing, aside from the collateral damage, moral issues, and so on. Take from that what you will.
I am not interested in having political arguments (one way or the other) in the comments of this blog post — I am burned out on online political debates for the moment. If you want to have a political debate, have it elsewhere. I will only approve constructive, interesting, non-obvious comments. Trolls will be banned and blocked. We will be coming back to this topic again, don’t worry. (Or do.)
Actual doctrine is understandably hard to get one’s hands on. The Nuclear Matters Handbook 2016, created by the Office of the Deputy Assistant Secretary of Defense for Nuclear Matters, is an extremely useful resource in this respect. Chapter 6 is about the Nuclear Command and Control System, and describes the many procedures, organizations, and technologies used to provide “the President with the means to authorize the use of nuclear weapons in a crisis and to prevent unauthorized or accidental use.” [↩]
The context of this snippet, recorded in Lilienthal’s journals, is a meeting between the President, Lilienthal (Chairman of the Atomic Energy Commission), and several military men and his cabinet. Lilienthal noted parenthetically that in the gap between the two “thats” in the second sentence, Truman looked “down at his desk, rather reflectively,” and that he (Lilienthal) “shall never forget this particular expression” when Truman said was not a “military weapon.” The military men, in Lilienthal’s account, then immediately began to talk about how important it was to get used to handling the bombs. General Kenneth Royall asked, “We have been spending 98% of all money for atomic energy for weapons. Now if we aren’t going to use them, that doesn’t make any sense.” Lilienthal’s commentary: “If what worried the President, in part, was whether he could trust these terrible forces in the hands of the military establishment, the performance these men gave certainly could not have been reassuring on that score.” Account of a conversation with Harry Truman and others on July 21, 1948, in David E. Lilienthal, The Journals of David E. Lilienthal, Vol. 2: The Atomic Energy Years, 1945-1950 (New York: Harper and Row, 1964), 391. [↩]
I gave a talk at the History of Science Society’s Annual Meeting in Atlanta a few weeks ago, on the topic of Command and Control systems and the ways they encode different visions of Constitutional authority and responsibility, and I am working to turn that into some kind of publishable paper. [↩]
There is an anecdote that is often repeated that states that Secretary of Defense James Schlesinger supposedly altered, on some level, the nuclear command authority during Watergate — telling his staff that he had veto power over any nuclear commands by Nixon. It is not something that has been historically substantiated, and, even if true, would technically have been un-Constitutional. It doesn’t mean it isn’t possible — but I find the whole thing, again, fairly dubious, and it certainly was not in line with the official regulations. There are instances (Stanislav Petrov, for example) of officers in nuclear situations not following regulations in an effort to avoid escalation. But they are by definition ad hoc, not to be relied upon. [↩]