Good morning, and thanks for having me here. I'm a little embarrassed to admit it, but despite spending several years working nonproliferation and arms control issues at the State Department and in overseas diplomacy during the previous administration, this is actually my first visit to Oak Ridge. As a result, I'm particularly happy to have the chance to talk to you here today, and offer my thanks to our gracious hosts at the Oak Ridge National Laboratory, and to the for organizing this event.
I've been asked to say a few words about a recent project I undertook at with the generous support of the , looking into what I called "countervailing reconstitution," or "CR." This is my own term, which Preparatory Committee meeting of the Nuclear Nonproliferation Treaty, but the basic concept has been discussed for years, appearing variously as "weaponless deterrence" or "virtual nuclear arsenals."
The specific terminology used is less important, however, than the basic insight: that it may not always be necessary to rely entirely upon actual, existing "weapons-in-being" in order to take advantage of whatever stability nuclear deterrence can be said to offer. My project aimed to evaluate the merits and demerits of deliberately maintaining a nuclear weapons production capacity even after the abolition of nuclear weapons, as a means of providing some kind of nuclear deterrent stability even at "zero." Might the prospect of retaining reconstitution capabilities as a strategic "hedge" against regime breakdown make today's possessors more comfortable with abolition? And might the prospect of their doing so help deter "breakout" from a "zero"-based regime, because the party contemplating such a violation would know that he would quickly face nuclear competition from others, as they activated their reconstitution capabilities?
I. Crisis Stability Challenges
The first thing I think it's worth pointing out about discussions of CR and strategic stability is the close relationship between these issues and more common disarmament-community debates over nuclear force "de-alerting." At some important level they are essentially the same issue: both approaches seek to maintain deterrence with forces that require a degree of regeneration in order to be usable; they differ primarily in how difficult and time-consuming such regeneration is supposed to be.
In any event, the critique I have most often heard – and made – against both de-alerting and a CR-based "zero" concerns crisis stability. Specifically, this critique revolves around the fear that a world in which deterrence depends upon the capability to regenerate nuclear forces (e.g., by producing new ones from an abolition baseline) would be a world which created incentives to take the destabilizing step of beginning such reconstitution early in a crisis, for fear of the potentially catastrophic consequences of "losing" a regeneration "race.
Indeed, a CR-based deterrent system might even create incentives to use nuclear weapons preemptively, since "winning" a reconstitution race would offer a chance to strike first before one's adversary builds or can deploy (or use) his first weapons. At any rate, because a CR-based system would create incentives to ensure maximally rapid reconstitution if the order were given, it might ignite a new arms race in reconstitution technology and procedures, not to mention creating incentives to cheat by maintaining a secret weapons cache. All in all, it is not clear that such a system would be more stable than today's world.
II. Other Challenges
A. Deterring "Breakout"
Another problem lies in doubt about the degree to which a country's temptation to attempt "breakout" from an abolition regime would in fact be deterred by other countries' possession of reconstitution capabilities. The logic of reconstitution as an effective answer to the challenges of breakout would seem to presuppose what the disarmament community often takes as axiomatic, but what is in fact a highly contested issue: namely, that the only use of nuclear weapons is in deterring the use of similar weapons by others. To the degree that this assumption is not true, the case for CR as a deterrent weakens.
Any analysis that supposes a possibility of breakout, however, has to acknowledge that someone might want nuclear weapons for reasons other than balancing other such weapons, since otherwise "zero" would be self-stabilizing, as no one would ever want nuclear weapons again once the last one had disappeared. To worry about possible breakout, in other words, is already to concede that countries may want nuclear weapons for reasons not related to any other country's possession of such devices. But if this is the case, other countries' CR capabilities won't necessarily deter breakout.
A country afraid of a neighbor's overwhelming conventional military power, for example, might be quite willing to provoke a return to globally reconstituted nuclear arsenals by developing its own weapons, because the alternative for it would be to face that powerful rival without nuclear weapons. And a country seeking to overawe and dominate a region generally filled with less technologically capable powers might find breakout attractive even if this provoked arsenal reconstitution by former nuclear weapons possessors thousands of miles away. (Indeed, by leveling the initial nuclear playing field with former weapons states, an abolition regime might benefit such an aggressive regional power.) These examples suggest that it is an open question how well a CR-based regime would in fact deter breakout.
B. Defenses
How ballistic missile defenses and other countermeasures against nuclear weapons delivery would play into strategic stability in a world of CR-based deterrence is also very much open to question. On the one hand, one might suppose that strong defensive capabilities in the hands of the potential victims of a violator state would be stabilizing. Such defenses might, for example, buy time for the world to react by making it harder for a violator to threaten others with its new arsenal. (The malefactor wouldn't be able to hit anyone until he'd taken been able to build up his numbers beyond what others' defenses could handle.) By blunting the potential impact of an "entry-level" breakout arsenal, and by allowing other states to protect their own reconstitution capabilities from preemptive nuclear attack, defenses thus might contribute to strategic stability in a CR-based abolition regime.
On the other hand, defenses in the hands of the violator rather than the victim of breakout could exacerbate stability problems by making it seem easier for the malefactor to "win" any reconstitution race he would provoke. In such a context, defenses could even encourage nuclear use by fostering the belief that the aggressor could handle whatever small nuclear second-strike its victim might launch after its reconstitution infrastructure had been attacked by the "winner" of the reconstitution race. In any event, the effect of defenses upon stability seems quite variable and contextual; there is no a priori answer to their stability impact.
C. De facto CR for Everyone?
Another problematic issue for reconstitution theory is how stable CR-based deterrence would actually be if it were generalized – that is, if a great many countries acquired de facto reconstitution capabilities such that each needed to be factored into every other country's CR-deterrence calculations. This is, of course, the dilemma one faces when contemplating the spread of dual-use nuclear technologies that create a nuclear weapons "option" for their possessor. In such circumstances, as Thomas Schelling has suggested, every crisis would become, in a sense, a nuclear crisis, and every war a potential nuclear war.
A world of generalized CR would be one in which a great number of states faced each other in potentially very elaborate proto-deterrent relationships of bewildering complexity, each seeking to hedge against the reconstitutive capabilities (and reconstitutive speed) of its myriad potential adversaries, and with each of these overlapping and possibly mutually-interfering dyadic relationships subject to the crisis-stability challenges I mentioned earlier. This does not sound like a particularly stable world.
Of course, if the spread of dual-use technologies – and thus de facto reconstitution capabilities – is to a some extent inevitable anyway, this sort of world may be arriving whether or not anyone openly pursues CR. One might argue, therefore, that some deliberate maintenance of a reconstitution capability might end up being the "least worst" option available to former nuclear weapons state under an abolition regime. Nevertheless, a world in which many states maintain an inactive but fully-formed weapons-production infrastructure might be significantly different, in stability terms, from a world in which the same countries merely possess the "option" of starting to divert dual-use technologies and materials to military purposes. There's hovering on the brink, one might say, but then there's hovering on the brink. The details matter!
D. Survivability
Most analyses of reconstitution-based disarmament place great stock upon ensuring the survivability of CR assets against preemptive attack, either with conventional forces or with such nuclear weapons as may become available to the winner of a reconstitution race. How precisely one might ensure CR survivability, however, is somewhat less clear.
Some analysts have suggested that the availability of deep tunneling and hardening technology makes this a comparatively easy task. But I am not clear how far this argument can go, because one would need to ensure the survival and functional integrity not merely of disaggregated component elements – which might end up individually safe against and sound but yet entombed and isolated from each other in their deep caverns – but of an entire reconstitution process, as well as a means for potentially employing, and commanding and controlling, whatever weaponry that one reconstitutes.
One would, for instance, need to protect: weapons production and assembly facilities; warhead component and fissile material storage depots; delivery systems and the institutions and processes by which they are loaded with warheads, managed, targeted, launched, and controlled; and the logistics and communications linkages that tie together the system of arsenal reconstitution and weapons delivery and enable it to function. All of these things would have to be protected, lest the CR system be subject to paralysis by adversary preemption.
Through this prism, a pessimist might describe a regime of virtual nuclear deterrence as an opponent's dream, populated by quite a few of what one might call "showstopper" targets, any of which an adversary could hit in order to paralyze the reconstitutive system. An alerted and nuclear-armed ballistic missile deployed on a patrolling submarine, for example, can today be taken out of an adversary's threat equation only by direct destruction of the weapon or its submarine – neither of which is easy. By contrast, a disassembled or not-yet-extant warhead stored separately from a demobilized missile that is itself merely capable of being sent to sea on a submarine presently mothballed at pier-side can be effectively neutralized by attacking a long chain of necessary reconstitutive events at essentially any point: warhead reassembly, missile remobilization, re-mating, submarine loading, the port, transit to patrol areas at sea, or any part of the process that connects (and controls) these elements. Protecting all of this from conventional and nuclear attack would surely be very difficult.
E. An Arms Race in Reconstitutive Speed
As noted earlier, one of the most vexing challenges of stability in a CR-based abolition regime is the danger of reconstitution "races." This puts an enormous premium upon how quickly reconstitution can be achieved, leading to the possibility of what one might term a "racing race": an arms race taking the form of a competition in rearmament methodologies and technologies, in which each participant aims to secure for itself the ability to constitute or reconstitute nuclear weapons faster than any potential adversary.
Harmonizing mobilization rates would seem essentially impossible. The time required to implement reconstitution would vary from one country to the next, and probably considerably. Some legacy warhead designs or delivery systems, for instance, might be capable of being returned to service more quickly than others. Some modes of nuclear infrastructure management also seem likely to be more reconstitution-friendly than others. A CR-based abolition regime would give governments incentives to fine-tune reconstitution capabilities with speed in mind, such as by switching to weapons designs optimized for rapid assembly, or pre-building weapons components designed for this purpose.
And with powerful incentives for haste could come recklessness. Countries unable to compete in a "racing race" by developing sophisticated solutions might be tempted to cut corners, perhaps through half-baked improvisational "MacGyver" solutions that could prove worryingly unsafe in operational practice.
III. Conclusion
What conclusions might one draw from all this? I began my project with a considerable degree of interest in reconstitution as at least a potential answer to some of the daunting theoretical and practical problems presented by the "nuclear zero" movement. Planning for a sort of reconstitution-based abolition seemed to offer a way to speak about "zero" in ways that didn't just seem stupid in the discourse of nuclear strategy and national security planning spoken by real-world decision-makers in possessor states. Accordingly, if things panned out, it seemed that CR might offer a program of action potentially "saleable" to security planners.
As you can probably infer by now, however, I am today a bit less optimistic. The game-theoretical obstacles seem to me very daunting, and the programmatic ones – that is, of actually maintaining a credible reconstitution capability over the indefinite future, a challenge that may resonate for weapons experts here at the Oak Ridge National Laboratory – hardly less so. But lest you take me for an unremitting pessimist, let me make two final points.
Given the problems that "weaponless deterrence" would still seem to raise, it may be difficult to imagine today's weapons possessors actually agreeing to it without there first having occurred some fundamental transformation in international politics. But many disarmament advocates do postulate just such a transformation, and indeed call for it: a wholesale transcending of the perceived need to rely upon strategic deterrence at all.
To this extent, it might be that the stability problem of pure-CR deterrence is either (a) so intractable as to be abolition-preclusive or in fact (b) a "self-solving problem" by virtue of reconstitutive "hedging" becoming unnecessary under the only politico-strategic conditions that would make abolition possible in the first place. In neither case, the argument might run, is there much need to worry about the puzzles and challenges that I think are presented by reconstitution theory.
I cannot resolve this question here, of course, nor say much about how one might hope to engineer such a miraculously transformed world. For the moment, it will have to suffice to suggest that CR doesn't appear to be as promising a map for the "road to zero" as one might have hoped.
But this leads me to my second and final point, and that is that nothing in all this analysis precludes still relying, and perhaps heavily, upon reconstitution capabilities as a means of persuading nuclear possessors to accept further reductions, even if this process cannot, at the end of the day, get us to the asymptote of actual abolition. The anticipation of CR, after all – in the form of U.S. planning for a more "responsive" production infrastructure as a strategic "hedge" to help handle unforeseen threats, or technical problems in our shrinking arsenal of remnant "weapons in being" – is already helping American planners be more comfortable with reductions.
Reconstitution is a principle that might be able to be pushed at least somewhat further in this vein, and which may well have salience for other possessor states as well – making some more comfortable with smaller arsenals, at least to a point, and others perhaps more comfortable with not building larger ones. This "facilitating reductions" angle, I think, is probably the most promising application of CR, and it is a subject that I hope to be able to explore further if I can secure funding for follow-on work.
Anyway, that should be enough to get discussions going. Thanks again for inviting me, and I look forward to your questions.