Joshua PollackPief Panofsky on Missile Defense

For reasons not worth going into, lately I’ve been reading an awful lot about ballistic missile warfare. Today’s cutting-edge theater missile made in Russia, China, or the United States is a precision battlefield weapon, suitable for attacking military bases or formations. Or crushing the occasional civilian vehicle, but hey, stuff happens. (See: SS-21 Debris In Georgia, Revisited, June 10, 2009.)

It wasn’t always this way.

Until pretty recently, in fact, missile warfare was the ballistic equivalent of area bombing. Thousands of inaccurate V-2s — like the one shown above at Peenemünde White Sands, apparently — battered London and Antwerp in 1944. Iraq and Iran targeted each other’s cities with ballistic missiles in the 1980s, culminating in a series of demoralizing attacks on Tehran in March and April 1988. After the USSR withdrew its regular forces from Afghanistan, the Red Army unleashed Scud barrages against mujahideen fighters for a time. And Iraq famously employed its “stretched Scuds” against Israel, Saudi Arabia, Bahrain, and Qatar in January and February 1991.

This last episode is remembered as the baptism by fire of the Patriot missile. Despite an initial perception of success, upon close examination, the first time out was a ringing failure. Upgraded systems appear to have worked better in the next war against Iraq in April 2003, despite a handful of friendly-fire incidents. (Update: see also Patriot Performance in Iraq, March 1, 2005, and the DSB report on this subject.) But in the ceaseless competition between offensive and defensive weapons, perfection is simply too much to ask.

The late, distinguished physicist Wolfgang (Pief) Panofsky sketched out the implications of this fact for defense against nuclear attack as clearly and crisply as I can recall seeing anywhere in a June 2001 article in Arms Control Today:

Nuclear weapons, however, profoundly changed the relationship between offense and defense because they increased the explosive power of a payload of a given weight and size by a factor of one million—a very profound change indeed. The demands on the performance and reliability of defenses against an attack by even a single missile carrying a nuclear weapon must therefore be extremely high for the defense to be considered effective. When the Germans attacked Britain during World War II with primitive ballistic missiles, none were intercepted, but the damage was limited because the missiles carried conventional explosives. Had they carried nuclear warheads, a single missile would have devastated London. Defense against ballistic missiles is therefore a totally different problem depending on whether such missiles carry conventional or nuclear payloads.

One might say that missile defenses can be asked to reduce levels of tolerable harm. They cannot be asked to prevent any and all intolerable harm. For that, of course, we have nuclear deterrence.

Because of space limits, I couldn’t include the lengthy quotation above in my first column at the Bulletin of the Atomic Scientists, where Panofsky’s work was published many times. For that, of course, we have blogs.

(See also: “Pief” Panofsky, renowned physicist and arms control advocate, dies at 88, September 25, 2007; and In memoriam: Pief Panofsky, October 2, 2007.)


  1. yousaf

    There are great technical challenges to national missile defense, certainly.

    But there are also policy concerns.

    Either NMD will influence your deterrence calculus with respect to, say, Iran, or it will not. (The previous administration seemed to think it will, echoed in the Bush statement that NMD will yield “an added and critical dimension of contemporary deterrence”.)

    (1) If it does not change our deterrence calculus then why bother with the enormous monetary and political costs? If all our behavior towards Iran will be the same whether or not we have missile defense, then what is the point?

    (2) On the other hand, if NMD does embolden us to stake out policies that we otherwise wouldn’t (in the mistaken belief that it would be 100% effective at countering any possible attack these may provoke), then this would be misguided and downright dangerous. (Not only because missile defense will never be 100% effective, but also because any target countries have other options of delivering WMDs, besides missiles). In such a scenario, NMD would not have brought us any more security — rather it would only have encouraged us to undertake riskier behavior with the false confidence of (an ineffective) defense.

    Pavel Podvig has an interesting article in the Bulletin along the lines of (1) above — i.e. that NMD is strategically useless , however, I think we cannot exclude the possibility that it will be downright strategically dangerous.

    NMD will be useless, or it will be dangerous. Take your pick. One thing that is not up for debate, however, is that it will most certainly be expensive.

  2. Philipp Bleek

    Agreed, Pief P.‘s statement is wonderfully concise and lucid. But I’d quibble with your follow-on that “One might say that missile defenses can be asked to reduce levels of tolerable harm. They cannot be asked to prevent any and all intolerable harm. For that, of course, we have nuclear deterrence.”

    I wish I had as much faith in nuclear deterrence as is expressed here! (I’m assuming the comment cited above wasn’t tongue-in-cheek, although tone is notoriously hard to gauge on the interwebs.) Even taking the critiques of deterrence by the likes of Ward Wilson with a big grain of salt (and I do), and acknowledging that nuclear deterrence probably played a vital role in preventing large-scale war over the years, “the prevention of any and all intolerable harm,” even if read to refer only to nuclear harm, is a tall order, and unfortunately not one about which history provides abundant reason for confidence.

  3. Josh (History)


    Your concern seems sound to me. In the interests of getting the main idea across here, I wasn’t able to convey everything that might be said about nuclear deterrence, to put it mildly. It’s no panacea. To the contrary, it’s uncertain.

    To clarify, let me put it this way: if even a single nuclear weapon goes off in a major city, that’s intolerable. So the appropriate role of a defense strategy isn’t to reduce 10x intolerable or 5x intolerable to the “merely” intolerable. It’s to reduce a 50% chance of the intolerable (let’s say) to a 0.00001% chance of the intolerable. In the ideal case.

    I don’t believe we’ve actually achieved anything close to that number, unfortunately.

  4. Pavel (History)


    The problem is that you don’t know if it is 50% and 0.00001% (setting aside the question of whether you could get that low or whether you can get confidence in this estimate). It may be 50% and 2% or 2% and 0.01% or some other set of numbers. We know that it is not going to be zero, so it will be intolerable in any case.

  5. Josh (History)


    In my experience, at least, the further one digs into this line of thinking, the more complicated it gets. There is a literature dedicated to the idea of acceptable and unacceptable risk. It defies easy summary, at least by me!

    For now, let’s set aside what the right number might be. We need only assume that the chance of nuclear use in any given period is higher than zero.

    One response is to try to reduce the probability of nuclear use to zero by eliminating nuclear weapons. But this won’t happen overnight even in the best case. The intervening period could last a very long time. Perhaps it will continue indefinitely, despite our best efforts.

    And even once nuclear weapons are successfully banned, do we really reach zero probability of nuclear use? I don’t think so, as we can’t be completely certain of compliance.

    In short, whether we consider a given >0 probability of nuclear use acceptable or not, it is difficult to imagine any situation in which it is reduced all the way to zero. But it would be a fine accomplishment to reduce that risk to negligible levels.

  6. Andrew Tubbiolo (History)

    The question I have with respect to NMD and small bit players is this. Does a NMD give one the option to not strike first? During a crisis with a small nuclear power alert levels will be high as a larger power will be tempted to srike first, and the smaller will be pressured to use it or lose it. Not to mention the limited conventional options of the smaller power. But then again we may be entering an age of limited conventional options with the global economy and all. But I think the real question is does a NMD give its owner the real option of waiting for the smaller power to launch first? Countering that, a first strike option is amplified for an NMD owner, as such a power could use the NMD to ‘clean up’ what ever emerges from a first strike. This analysis, no doubt, will encourage the smaller nuclear power to up their alert level even higher. This is assuming again that the NMD works, because if it does not there’s nothing to be gained but more encumbered debt and false options during a real crisis.

  7. Josh (History)


    Surely, the option not to strike first always exists. It’s served us well so far, at least.

    I see what you mean, though. One could argue that a defense gives a leader added confidence not to strike first in a crisis. Or, as you point out, one could argue the exact opposite: a defense gives a leader added confidence to strike first in a crisis.

    And yes, it does seem that the latter interpretation asks less of a defensive system than the former, since it would only have to contend with “ragged retaliation.” Still, I’m not too sure I’d like to take my chances, if it were up to me.

    The logic of a “reassuring” defense, or so it seems to me, requires either that the defense be seen as (virtually) perfect, or that a leader be inclined to distinguish between what Herman Kahn called “tragic but distinguishable postwar states.” On either score, I’ll defer to the Panofsky quote above.

  8. Jochen Schischka (History)

    In my opinion, the primary sense of a missile defense system is to make the other side reconsider their attack plans.
    It’s never a sure thing that the own attack missiles will work 100% (especially in a world without actual full end-to-end live-fire testing involving full-yield nuclear detonation over a specified target area); And, as ‘first-striker’, i’ll always have to consider the possibility that my opponent’s ABM-system plus his counter-strike-missiles will work 100% (‘worst case scenario’ or, if you like that better, ‘the safe bet’).

    Now, if i’m a first-striker with an ABM-system, i’ll always have to assume that my own attack missiles and my missile defenses fail at the same time while the other one’s attack missiles work perfectly…

    As far as i see, an added layer of safety against rash actions. And we should also not forget about accidentally launched nuclear weapons, which will have to be allowed to detonate where they’ll land (e.g. New York, Paris, Moskow etc.) without any intercept-option.

    Of course, such a (not neccessarily 100%-functional – but i’d say if i can save even only a single one of my multi-million population centers from thermonuclear devastation, it’s been fully worth the cost) ABM-system will only work against a) rational-thinking opponents and b) an enemy with less attack missiles than i can afford to buy interceptors (So it’s quite irrational of the Russians to feel threatened by NMD – unless they know for sure that most of their ICBMS will not work…).

    As a side note:

    Josh, could you please modify the picture’s description? It’s definitely the TF-1-White-Sands-shot from August 22., 1951…

  9. Josh (History)


    Thanks for the correction on the picture.

    I’m afraid I don’t quite follow your reasoning. If the primary purpose of a missile defense is to make the other side reconsider their (nuclear) attack plans, well, that also happens to be the purpose of a nuclear deterrent. And from the perspective of a potential attacker, what’s a few detonations less on the other side’s cities (the effects of the defender’s BMDS, if it works somewhat) compared to whether or not one’s own cities are devastated (the effects of the defender’s nuclear retaliation) or left untouched? I think the answer is clear enough.

    Now look at it from the defender’s view. When it comes to nuclear weapons, the ability to confine the extent of horrific destruction is less meaningful than the ability to confine the probability of such an event. I would go so far as to say that the former factor, when compared to the latter factor, is trivial verging upon negligible. It’s the difference between more vs. less destruction, on one hand, and some vs. no destruction, on the other hand — where “more,” “less,” and “some” all describe the most devastating event in national history.

    We simply cannot discuss city-destroying weapons as if they were ordinary bombs.

  10. Jochen Schischka (History)


    Like i wrote before, this will not work with an opponent overabundant in warheads (and thus, i absolutely do not understand the Russians at all – PAC-3s and SM-3s are a much better bet against Iskander-Ms or a hypothetically revived Skorost, considering the european battlefield, while there will most likely never be enough GBIs to intercept anywhere near a sufficient number of russian warheads directed at the U.S. – unless the Russians have absolutely no confidence in their own missiles).

    ABM systems (irrelevant if working 100% or only 30% – so i’ll launch six missiles at that incoming warhead, just to be sure…so what? What is more irreplacable to me, those missiles or a whole city?) will only be of use in case of a limited number of attacking missiles, like in an accidental launch or in case of an attacker with only a few nuclear-tipped missiles in total.

    Since we are living in a world where obviously every psychologically-less-than-stable autocrat with enough money can buy all the weapons he likes (but characteristically can neither acquire nor sustain nuclear ones with medium to intercontinental range in large numbers!), i’d feel a lot safer if i knew that there is a second option to mutually assured destruction allowing e.g. a conventional answer to a (small-scale) nuclear attack (which can hopefully be intercepted). If such dangerous and devastating weapons get proliferated as broadly as is apparently the case nowadays, a nuclear exchange (with many million dead bodies) will be unavoidable in the long run (imagine not only every trigger-happy psychotic thug but everybody out there carrying an AK-47 on full-auto with an 75-shot-drum magazine – it’s only a matter of time until somebody pulls the trigger…and then it will get really nasty)!

    Let’s also not forget the political implications (if we’re talking about rational-thinking governments with nuclear weapons at their hands which will not use them ‘just for fun’): As we should have learned from the cold war, ICBMs are a ‘fleet in being’ – political decisions will get influenced by that threat, even if not a single missile is ever launched (and even if that flashy nice thing from that parade doesn’t work as advertised…). The possibility of intercepting (a small number of) incoming warheads offers a lot more freedom for decisions not pleasing anybody (not possible in the long run anyway in my opinion), especially considering all those proliferators with only a rather limited arsenal. And don’t forget about all those nations without a nuclear arsenal at all (like Japan, South Korea, Germany, Poland, Canada etc.) – they’ll be likewise influenced in their decision-making process, most likely not only to America’s, but also their own disadvantage.

    Last but not least, let’s consider the ‘accidental launch’ scenario (yes, i know, has not yet happened – but that is only a question of time, if you ask me): Would you say that it’s acceptable to just take a (thermonuclear) hit with several million dead? How should a nation taking such an accidental hit react? “We acknowledge that it was not your intention, but now we’ll have to eradicate one of your cities in return anyway, just to maintain our credibility”???

    Indeed, we cannot discuss city-destroying weapons as if they were ordinary bombs – i think we fully agree on that issue!

  11. Andrew Tubbiolo (History)

    Josh, your point of always taking the option not to fire first is unquestionable. Except in times of crisis. What happens at the times when policy, and diplomacy fail and military hostilities are imminent? I wonder if the dynamics of a first strike against a bit player like Iran would be far different than the dynamics of deciding to strike first against a more established power. Obviously, I think they would be. It’s an open ended question that my pea brain cannot close either way. The crater at the desert launch pad sure looks better than the crater at an urban setting. Then there’s the hornets nest to deal with … I can’t imagine the pressures of dealing with a real crisis.

    Running to extremes may be comforting in peace time, but in time of crisis will a future NCA be thanking his lucky stars he does not have the false option of an NMD, or cursing the lack of it? Then again, I’ve never read a weapons system described from the point of view of how it might enhance a NCA’s option to back out of launching during a crisis.

    One has to ask how many elected NCA’s ever really give it that much thought?


Pin It on Pinterest