After New Start, What?


New Strategic Arms Reduction Treaty

One good reason to extend the Obama/Medvedev New Strategic Arms Reduction Treaty for five years is that it might take that long to negotiate and put in place what might come after its expiration. Since New START might not be extended, it makes sense to start brainstorming now about our nuclear future.

As I see it, there are four options.

One is a continuation of a numbers-based arms control and reduction regime. We’ve had one since 1972, beginning with the first Strategic Arms Limitation Accord, and it would seem strange and uncomfortable to be without one. This path isn’t easy: it’s likely that extending any numbers-based formula beyond New START would have to be far more complex and inclusive, both in terms of delivery vehicles and participating states. More on this below.

A second option is a combined numbers and norms based regime. This, too, would be more complex, inclusive, and difficult to do.

A third option would be a norms-without-numbers based regime. Not easy, but way less complex. This regime, like those above, would need to be more inclusive to be more effective.

The fourth option is to have no replacement for New START.

Why is it likely that any post-New START regime might have to be more inclusive?

To begin with, the hip bone is connected to the thigh bone. Even “mid-sized” nuclear powers now possess three-digit-sized arsenals — and some of these arsenals are growing. What one nuclear-armed state does has ripple effects on others, resulting in cascading nuclear requirements. Both the Kremlin and the Senate have said they want a more inclusive regime.

Another reason is that the more widely shared obligations are, the more effective a post New START regime will be. It’s high time to extend restraints to the nuclear arms competition in Asia. China, India and Pakistan have so far refused to do so, and their national security has been diminished as a result. Reducing nuclear dangers in Asia as well as Europe demands a formula that covers all seven of the largest nuclear arsenals — the United States, Russia, France, Great Britain, China, India and Pakistan.

Finding the right counting rules for a seven power agreement won’t be easy. Let’s presume to count, as we have become accustomed, each deployed intercontinental ballistic missile, submarine-launched ballistic missile and long-range bomber as adding “one” to each nation’s total. The baseline range here, dating back to SALT, has been 5,500 kilometers.

What about intermediate-range ballistic missiles?

Do we count them, as well? And does each missile beyond a certain, agreed range within this category count as one? I would be inclined to say yes.  One reason is because intermediate-range missiles are considered strategic delivery vehicles in Asia and Europe. Moreover, if we don’t count missiles of intermediate range and weigh them equally to longer-range missiles, we just invite an accelerated  competition at lower ranges. Negotiators would have to arrive at an agreed range threshold for counting purposes, and this could be a serious bargaining issue.

Another complication: If we count intermediate-range ballistic missiles beyond a certain range, do we count intermediate-range cruise missiles? Again, for the reasons noted above, I would argue in the affirmative. Do we include conventionally-armed cruise missiles as well as nuclear-armed cruise missiles meeting our range threshold? My tentative answer is no — as long as states are willing to accept intrusive monitoring that confirms that certain classes of cruise missiles do not carry nuclear weapons.

Do we count new types of weapon systems in national aggregates, such as hypervelocity “glide” delivery vehicles (regardless of their payload), other types of prompt global strike systems beyond an agreed range, and long-range, nuclear-armed torpedoes? Again, I’d answer in the affirmative. These weapon systems have strategic consequences, so they deserve to count. Moreover, to keep their numbers down and to reaffirm their “niche” status, it might make sense to count them equal to other strategic nuclear delivery vehicles. Otherwise, their deployment would not come at the expense of other weapon systems in national aggregates.

What about missile defense interceptors?

Do we count them, as well, if they meet certain performance parameters? In my view, the answer is, once again, yes, using the very same unit of account. That way, for every defensive interceptor a country deploys exceeding certain parameters, it would need to subtract one unit of offensive capability.

Under my notional and highly tentative thought process, states would be free to choose between offense and defense, and choose again within offensive categories according to their perceived national security interests. Cooperative, intrusive monitoring would presumably be required to affirm exclusions and agreed parameters. One possible way to deal with states that wish to avoid intrusive monitoring is to make them ineligible for exclusions.

What might the national aggregates look like?

They would depend in part on the ranges we might assign for accountable missiles below intercontinental range and what parameters would be set for counting missile defense interceptors. Those who are so inclined can help us all by doing the math using various parameters.

A multilateral, numerically based extension of New START would thus harken back to the naval arms limitation treaties between the first and second world wars. These treaties were bedeviled by workarounds, violations, exclusions, and — most of all — by the clear intent of two of the signatories — Japan and Italy — to disregard the national sovereignty of other states. Germany wasn’t a party to the Washington and London Naval treaties. It was supposed to forego having a navy by the punitive terms of the Versailles Treaty. That didn’t work out so well. Naval armament controls couldn’t possibly be sustainable when two major powers, Germany and Japan, were intent on dramatically changing the status quo in Europe and the Pacific.

This thumbnail excursion raises central questions. First, what are U.S., Russian and Chinese intentions toward the status quo in Europe and in Asia? Does any major power seek changes in the status quo by means of aggressive or unconventional war? How instrumental are nuclear weapons to U.S., Russian and Chinese risk taking? And would the United States, Russia and China view multilateral arms control and reduction compacts to be in their national security interests?

If answers to these questions are not reassuring, then multilateral arms control and reduction compacts will not be reassuring, either. If, however, our assessments suggest that, at least in a preliminary way, these compacts are worth exploring, then let’s put our thinking caps on.

If agreements are reachable, they would likely take the form of political compacts rather than treaties, with all the attendant advantages and disadvantages.

Even if agreements are not reachable or inclusive, preliminary discussions with free riders could still be useful:  If any of the major nuclear-armed states do not wish to be a part of a multilateral regime, what other steps are they willing to take to reduce nuclear dangers and weapons?

If preliminary multilateral talks begin, several very hard problems would need to be addressed. Would a single unit of account be possible, or would states insist on different units of account for various offensive and defensive weapon systems? Which weapon systems would be counted and which ones would be excluded? And what seven-power ratios do the chosen units of account suggest?

Ratios could serve as a baseline to establish the principle of no aggregate increases of capability while modernization programs continue. Ratios could also serve as ceilings for proportionate reductions over time. Constraints on placing interceptors in space would seem necessary to make terrestrial limits and reductions possible.

My view is that comprehensive is better than partial constraints and that complexity invites workarounds, allegations of cheating and erosion of trust in any compacts reached. In other words, the simpler and broader the better.

This thought exercise leads me and perhaps you to the following conclusion: Negotiating multilateral, numerically-based arms control and reduction compacts will be a difficult and time consuming task. Nonetheless, it wouldn’t hurt to begin thinking through this approach and assessing different parameters and numbers.

This thought exercise might also lead us to think about a simpler but still inclusive approach. This might take the form of a norms-based regime.


How do we transform a dangerous deterrence-based system into a far less dangerous one?

A nuclear deterrence-based system will be with us for a long time — and that’s if we’re successful and lucky. The goal for me is how to create conditions so that a deterrence-based system doesn’t kill us.

How do we create these conditions?

One way is through numbers. Another way is through norms. The best way is by numbers in conjunction with norms.

I’ve written earlier about numbers. There’s certainly value in extending a bilateral numbers-based regime for the two states with the most nuclear arms. Since the United States and Russia exceed China’s nuclear holdings by a very significant margin, predicating an extension of New START to China’s inclusion makes no sense; it’s a thinly disguised way of killing New START’s extension. One of the many reasons to extend New START is the time it will likely take to agree to something better, with broader participation.

Let’s assume, optimistically, that New START is extended. What then?

Laying out a new set of very low numbers has been proposed. This is an important heuristic device — it helps point us in the direction of a safer future. But it’s hard to envision how the United States can get from here to minimal deterrence alone, safely, and without freaking out allies absent fundamental change in both domestic political and international conditions. Absent conditions conducive to actual transformation, those seeking minimal deterrence-based force structure will reinforce but not broaden mindsets.

Abolition and the Ban Treaty offer escape from a dangerous deterrence-based system but, alas, the United States, Russia, China and other nuclear-armed states are under the influence of Mars, not Venus. While I support the end state of a Ban Treaty and abolition, this is not the focus of my work. End states will remain distant goals unless we succeed in many battles between then and now.

In my view, the grassroots game only changes the inside game in the United States when the President is willing to override dangerous nuclear orthodoxy. This has happened very rarely, either when external circumstances reinforced domestic insistence (e.g., the ban on atmospheric testing and limits on national ABM deployments) or when external conditions are non-threatening (e.g., reductions in force structure and stockpile under Bush 41, and further reductions in the stockpile under Bush 43). When external conditions appear threatening, top-down decisions to buck nuclear orthodoxy can backfire (Carter) or are easily nullified (Obama).

Alternative theories of change in nuclear orthodoxy are welcome.

Another way to change status quo-oriented conditions so as to make deterrence less dangerous is to seek specific modifications in orthodoxy. Remove tactical nukes from Turkey, by all means. Reload the sealed Trident tubes and use this device to begin reducing deployed ICBMs. And please kill the low yield Trident warhead. This idea is the product of deliberation in rooms lacking proper filtration systems. It reminds me of when very smart people like Harold Brown and Bill Perry, operating from a de-oxygenated place of deterrence orthodoxy, tied themselves into knots trying to find a survivable basing mode for the MX missile.

I put No First Use in the category of a specific modification to make deterrence less dangerous. It’s very important. It’s also wonky. Anytime we challenge nuclear orthodoxy, it’s hard to avoid the weeds. My instinct is to try to avoid the weeds as much as we can.

Norms aren’t in the weeds.

I believe we can transform a dangerous deterrence-based system by championing norms that, over time, make nuclear weapons increasingly peripheral and less valuable. Norms that facilitate lower numbers and less exacting readiness standards. Ronald Reagan might phrase this as norms that make nuclear weapons impotent and obsolete.

What norms make nuclear weapons less usable and subject to further reductions?

Above all, the norm of not using nuclear weapons in warfare.

The norm of not threatening to use nuclear weapons in warfare.

The norm of not testing nuclear weapons.

The norm of nonproliferation.

The norm of safety and security for nuclear weapons and bomb-making materials.

The norm of reducing nuclear excess.

These norms can’t stand alone. They would need reinforcement with norms of responsible interstate relations, harkening back to McCloy-Zorin, Nixon-Brezhnev, and the Helsinki Final Act. (More on these in subsequent posts.) My thinking is still fuzzy and needs refinement, but my basic point here is that the nuclear norms we seek won’t be reinforced over time if nuclear-armed states are intent on changing lines on maps by the use of force or subversion.

This is obviously problematic, given Vladimir Putin’s behavior in Ukraine, Georgia and elsewhere, and given China’s muscle flexing toward its eastern neighbors. More such actions could be ruinous to nuclear norm building. Past actions need not be: The United States negotiated with the Soviet Union over the control and reductions of nuclear capabilities while refusing to recognize the Kremlin’s engorgement of the Baltic States and its control over unwilling eastern European populations. Likewise, the West would not recognize Russia’s annexation of Crimea and can push back against its misbehavior elsewhere while seeking to reduce nuclear dangers.


Russia and China are not pre-World War II Germany and Japan.

They want to increase their spheres of influence at the expense of the United States, not engage in aggressive wars to control their neighbors. Russia has inherent weaknesses. China has great potential. The United States has diplomatic, economic and military instruments that, if wisely employed, can counter these ambitions. The battlefield use of nuclear weapons is not among them.

Is “No First Use” the right way to characterize the most important norm of responsible nuclear stewardship? A broader formulation — something like ‘no use’ or ‘no nuclear use,’ or ‘no battlefield use’ might be better. Joan Rohlfing, who has given this much thought, prefers ‘prevent nuclear use.’ The bumper sticker version of this might be Don’t Start a Nuclear War. More brainstorming would be helpful.

However this norm is formulated, it would cause some additional angst within states that rely psychologically on the U.S. nuclear umbrella, especially in the Pacific where U.S. conventional, naval, and air power capabilities — not to mention U.S. diplomacy —  are not entirely reassuring. But neither do allies find the threat of nuclear first use very reassuring.

The weakest link of extended deterrence isn’t the absence of new low-yield options with assured penetrability that can reach targets from long distance within thirty minutes. The weakest link of extended deterrence is the notion of first use. First use is the failure of deterrence; retaliatory use is the essence of deterrence. Under the norms proposed here, there will be no shortage of U.S. retaliatory capabilities. This message needs to be conveyed to jittery allies alongside the core message of seeking to prevent any crossing of the nuclear threshold by diplomatic and conventional military means.

There is no greater or more important challenge before us than to extend the norm of not using nuclear weapons on battlefields. This norm is now almost three-quarters of a century old. Borrowing from Lew Dunn, our task is to extend this norm to the century mark: No Nuclear Use after Hiroshima and Nagasaki for the following 100 years. And to extend the norm of not testing nuclear weapons to 2045, as well. Tests are confirmations of utility and affirmations of power. Instead, we seek to convey the message that nuclear weapons are a breed apart. So far apart from other kinds of weapons that responsible states don’t even test them.

Sounds hard?

Sure. But not as hard as stopping tests, which took four decades of hard labor, or getting to the three-quarters-of-a century mark without battlefield use despite intense crises, impulses toward preemption, and despite nuclear-armed states losing wars against non-nuclear-armed foes and despite limited warfare between nuclear-armed states.

How do we succeed?

One day, one month, and one year at a time. Just like those who worked the problem before us succeeded. By making it through every crisis without the use of a nuclear weapon. By repeatedly challenging every state, including those led by authoritarian rulers in possession of nuclear weapons, to be responsible stewards. By hammering away until we are blue in the face about norms that define responsible nuclear stewardship.

How useful will nuclear weapons be if they haven’t been used in warfare for 100 years? 

How many nuclear weapons might national leaders think they absolutely need if they haven’t been used on battlefields for 100 years?

In the First and Second Nuclear Ages, we’ve defined progress or backsliding mostly in numerical terms. We’ve also been legalists. We seek numbers embedded in treaties. There is still great value in numerically based compacts and treaties. It’s well worth trying to extend and expand numerical limitations whether they are embedded in treaties or executive agreements. But we may well be entering a period where norms matter more than numbers.

We can succeed by championing norms even if we do not succeed with numbers, and it will be absolutely essential to champion a norms-based system if the numbers fall away. Along the way, it’s essential to push back against nuclear excess generated by over zealous proponents of dangerous nuclear orthodoxy.

Yes, this is daunting. But it’s also achievable.