Jeffrey LewisSome Think Size Matters

A few weeks ago — ok, maybe longer — I commented on an interesting proposal (that I nevertheless disagree with) by Crispin Rovere and Kalman Robertson to prohibit low yield nuclear weapons. I agued (“Size Doesn’t Matter“) that I didn’t agree with the problem as framed, didn’t think the treaty could be verified and, in any event, thought entry-into-force of the CTBT would be a better use of our time and energy.

Rovere and Robertson asked for the opportunity to respond, which I think is quite reasonable:

Size Matters 

Crispin Rovere and Kalman A. Robertson

The following is our response to Jeffrey’s challenge to our study and its conclusions. We apologise for this taking some weeks, Crispin has been lost in the internet black hole of the DPRK for much of that intervening period, and has only just rejoined the technological landscape.

Jeffrey critiques our study in precisely the right way, asking what problem we are trying to solve, examining our proposed solution, and then comparing it to the possible alternatives.  He posits that low-yield nuclear weapons are not uniquely dangerous, that a treaty prohibiting nuclear weapons with design yields below five kilotons would not be an effective way of dealing with them anyway, and that universalisation of the Comprehensive Nuclear Test Ban Treaty would achieve the desired result.

We shall frame our response to these arguments in precisely the way that Jeffrey so excellently presented them…


1.         What problem are we trying to solve?

We are concerned about the perceived usability of nuclear weapons, and we argue that low-yield weapons are more ‘usable’, on aggregate, than other kinds of nuclear weapon. Unlike nuclear weapons of higher yields, low-yield weapons make nuclear conflict more likely by weakening the firebreak that separates conventional and nuclear war. As the yields of nuclear weapons approach those of conventional weapons, the probability of conventional conflict escalating to the nuclear level increases.

It is true, as Jeffrey argues, that in the case of the United States the retention of low-yield weapons within its arsenal may not greatly increase the probability of America conducting a nuclear first strike. This is due to America’s overwhelming conventional superiority, because the contingencies in which the United States would consider a nuclear strike are likely to require higher yield weapons (see pages 36-37 of the full study).  Nevertheless, U.S. possession of low-yield weapons causes non-nuclear armed states to apprehend the risk of their use in counter-proliferating strikes and they may respond accordingly by developing their own nuclear arsenals. Indeed it is for these very reasons that low-yield weapons are utterly pointless for the U.S to possess, and yet they are distinctly dangerous for other states such as Pakistan and Russia to retain. It is therefore in America’s interests to support the restriction of these weapons in a multilateral framework – precisely where the proposed treaty comes in.


2.         Does our proposal solve this problem better?

 Jeffrey is concerned that “if the Russians build these things, they will do so not for deterrence but because they plan to use them.” If this is true, then Jeffrey is right – it doesn’t matter what treaty is in place because we’re all going to be in a nuclear war.

But this is not typically why states pursue low-yield weapons. Rather, low-yield weapons are sought principally in order to deter conventionally powerful rivals. In other words, they are designed specifically as first strike weapons, to be used against conventional forces. Jeffrey is sceptical of Pakistan’s claims to be developing low-yield artillery, but even if they aren’t, India apprehends that Pakistan’s nuclear doctrine incorporates first strike scenarios involving low-yield weapons and is responding on that basis.

On verification, Jeffrey states that ‘in order to verify that no state possesses nuclear weapons of a certain yield, very intrusive measures would be necessary’. That would be true if the aim was to eliminate the technological possibility that a nuclear-armed state possesses nuclear weapons of low yield. But this is not the role of adequate verification.

In any nuclear arms control agreement, adequate verification means that, if a nuclear-armed state commits a militarily significant violation, it will be detected in sufficient time for other parties to take action that will effectively neutralise any resulting military advantage. If any nuclear-armed state violates a treaty banning low-yield weapons, it gets at most a marginal discernible advantage, immediately neutralised by the higher yield weapons held by potential rivals and/or the legal mechanism of the treaty. Consequently, verification requirements are almost vanishingly small.

Indeed, there are many treaties in international law that achieve their objectives without intrusive verification mechanisms. This is for one obvious reason – there is nothing substantial to be gained through a violation. The Anti-Personnel Landmine Convention, the Convention on Cluster Munitions, and the Convention on Certain Conventional Weapons have all created effective bans and yet they do not have any intrusive verification measures. Conversely, the NPT has stringent verification requirements (by way of IAEA safeguards) because there is a great incentive for some states to pursue nuclear weapons and their clandestine acquisition has profound consequences for the security of others. A treaty prohibiting low-yield weapons sits firmly in the former category because any state that has them, or gives them up, also has larger strategic alternatives.

To explain why, we can take Jeffrey’s Russia example. In a conventional conflict with the United States, Russia may calculate that the use of a low-yield nuclear weapon would not result in escalation to the strategic level, and that the United States will instead respond in-kind with low-yield weapons of its own. This could arguably benefit Russia, as its nuclear forces are far more capable than its conventional forces vis-à-vis the United States. But if a treaty prohibiting low-yield weapons were in place, then not only would such an attack be condemned by the international community, Russia would also be aware that America’s only option for retaliation would be to escalate. This understanding would no doubt factor seriously into the Russian calculus when deciding whether to cross the nuclear threshold in the first place. In other words, a treaty prohibiting low-yield weapons strengthens America’s nuclear deterrent, irrespective of whether Russia cheats by developing low-yield weapons.

It is through this lense that the verification issue must be viewed. Jeffrey makes the clever quip about “Frank Miller’s head exploding when he learns that we’re going to let the Russians examine all the electronics in our nuclear weapons.” Of course that’s not what we recommend, we simply say that any verification measures specifically for existing variable-yield weapons will depend on the electronics, and this may be the subject of separate negotiations between the states that possess them. For the reasons articulated above, the verification requirements are going to be modest, and will really depend on whether a state may have confidence that parties to the treaty aren’t deliberately fielding weapons that are specifically designed for nuclear first strike against conventional forces.

It is reasonable to assume that many thermonuclear weapons in advanced arsenals have primaries that, if they could be separately detonated without boosting, would result in a low-yield explosion (because such primaries tend to contain relatively small quantities of fissile material). This is an integral part of the physics of thermonuclear weapons and in this sense “swapping hydrodynamic codes”, to use Jeffrey’s expression, would indeed be worse than useless as a verification measure.

The fact that a state may have the capacity to detonate the unboosted primary of an otherwise high-yield weapon, such as an SLBM, to produce a low-yield nuclear explosion is unlikely to be a significant issue since no state could use it to deter a conventional attack without also openly violating the treaty.

3.         …than other, more likely outcomes? (such as entry-into-force of the Comprehensive Nuclear Test Ban Treaty)

Leaving aside the question of whether it is more likely or not, Jeffrey’s assertion is that universalisation of the CTBT would more effectively deal with the problems we seek to solve than the treaty we propose.  Although Jeffrey is right to point out the importance of the CTBT, and makes a good recommendation regarding confidence building measures at old Russian and US test sites, we must respectfully disagree with his assessment that the CTBT can substitute for a treaty prohibiting low-yield weapons.

To revisit the beginning, the problem we are trying to solve is as follows: to ensure the firebreak that exists between conventional and nuclear war is not undermined by the proliferation of low-yield weapons. The CTBT does not meet this objective.

Nevertheless, both a treaty prohibiting low-yield weapons and the CTBT multilateralise the arms control process in a manner that allows continued progress to be made on other arms control initiatives. Our proposed treaty actually makes the universalisation of the CTBT easier to achieve because it reduces incentives to test new low-yield designs. There is no reason why these two initiatives should not be pursued together, and we have previously written on the imperative of securing entry into force of the CTBT here, and here.

Jeffrey begins his critique of our proposed treaty by saying “it’s probably a bad idea”, but nothing in his argument actually suggests that it is. At most, he argues that it isn’t worth the political effort required, and on that point we differ, since taking non-strategic nuclear weapons into account in a multilateral framework is almost certainly a requirement for continued progress on nuclear arms control elsewhere.

We welcome the discussion that this proposal has generated, for it draws attention to the specifics of low-yield weapons in the broader arms control discourse. We believe that the introduction of low-yield weapons into unbalanced and unstable strategic rivalries carries greater risks of nuclear conflict than strategic weapons, which hitherto dominated nuclear arms control.


  1. G.Balachandran (History)

    Really should there even be such a detailed dissection of the issue?. Let us agree with the proposition of Crispin Rovere and Kalman A. Robertson that (i) “low-yield weapons are sought principally in order to deter conventionally powerful rivals”; (ii) US has overwhelming conventional superiority; and (iii) realistically speaking we are then talking about Russia and Pakistan confronted by conventionally powerful rivals.If I am not wrong,during the cold war, Soviet Union had conventional superiority and US to think of tactical low-yield weapons. Now they don’t need it and therefore it may even be true that “low-yield weapons are utterly pointless for the U.S to possess.”
    But from these self-evident propositions, how does one jump to the statement “If any nuclear-armed state violates a treaty banning low-yield weapons”? If is it accepted that it is not pointless for a “nuclear-armed state” to possess low-yield nuclear weapons, why should they sign any treating banning them? Or do the authors of the report suggest that it would be illegal for a state to do something which others have undertaken not to do, but which they have not accepted? i.e. a treaty can be binding even if someone is not a signatory? That would be an entirely a novel interpretation of international law and treaties. In that case India’s nuclear status would be illegal since almost all countries have signed the NPT and India has not. I don’t think this novel interpretation of the authors would find any takers in any sensible group of either international legal experts or IR specialists.
    It seems, therefore, that such a treaty is a non-starter from the very beginning. That being the case intricate discussion on the subject is not likely to be fruitful except as proposals for research grants.

    • I disagree (History)

      “a treaty can be binding even if someone is not a signatory? That would be an entirely a novel interpretation of international law and treaties.”

      No, untrue.

      Take this example:
      Here is a treaty, and 3/4 of the nations sign it.
      Axiomatically, 1/4 do not sign it.

      But all nations – signatories and non-signatories alike – incorporate the regulations contained with that treaty into their military manuals, etc. etc.

      In short: the gist of that treaty becomes the “state practices” of all nations, even by those who haven’t signed that treaty.

      Result: the regulations themselves become part of the “Laws and CUSTOMS of war”, and therefore become binding on all states.

      That isn’t a hypothetical case: it is how the “Hague Regulations 1907” took on a life of their own from the “Hague Convention 1907”.

    • Nathan (History)

      The real issue here is “does the world need nuclear weapons”? Who cares who might have the most or be most powerful? The US has always and still maintains that they are and yet they fell victim to rebels with no more than camels and guns, just determination. The US like Russia and all others who have nuclear weapons believe they are to be reckoned with rather than feared for what they are. Nuclear weapons are a terrorist threat to this planet and all those who have or threaten to use them are terrorists. There is no such thing as a treaty when nuclear weapons are involved, mentioned, threatened or used. Eventually the rest of the world will come to realize this then countries like the US and Russia will be asked or if needed forced to disarm. The rest of the world cannot afford for terrorists like the US or Russia to dictate the future and well being of the majority of humanity because they want money or power!

  2. John Maurer (History)

    On the deterrent logic of low-yield nuclear weapons:

    “Russia would also be aware that America’s only option for retaliation would be to escalate. This understanding would no doubt factor seriously into the Russian calculus when deciding whether to cross the nuclear threshold in the first place. In other words, a treaty prohibiting low-yield weapons strengthens America’s nuclear deterrent, irrespective of whether Russia cheats by developing low-yield weapons.”

    Two points:

    1) The posited asymmetry in US-Russian nuclear forces might further deter the Russians, *if* they think that the US threat to escalate to high-yield nuclear war is credible. If not, then the asymmetry of low-yield nuclear weapons might actually weaken the nuclear threshold, by convincing the Russians that they could use low-yield nuclear weapons with no real threat of US escalation beyond conventional means. After all, presumably the Russians will still retain a large arsenal of high-yield nuclear weapons of their own, which would make a US escalation to high-yield nuclear weapons massively irrational under virtually any circumstance, including Russian first-use of low-yield nuclear devices, especially if those devices were detonated outside of the continental United States. Under these circumstances, the United States’ “all or nothing” nuclear deterrent posture could actually be quite weak: what US leader would escalate to general nuclear war over a few low-yield detonations in Eurasia?

    If we think the US-Russia example is incredible, then there may be other dyads of nuclear competition in which the same logic could apply: India-Pakistan, China-Russia, Israel-Iran. As long as neither side has the ability to credibly wipe out the other’s high-yield deterrent force, then low-yield use remains a rational option, if not a wise one. It might be that an asymmetry in low-yield nuclear weapons actually provides no significant advantage to either side. But this is far from proven. Absent a really credible argument for the strategic uselessness of low-yield weapons *for everyone,* not just for the United States, a comprehensive ban on such weapons would be meaningless, and perhaps contribute as much to the risk of nuclear war as alleviate it.

    2) If, in the case of the United States, we really believe that not having low-yield nuclear weapons will strengthen our nuclear deterrent (by tossing the steering wheel out the window in one of Schelling’s games of chicken), then why not disarm this component of our nuclear force unilaterally? If low-yield nuclear weapons are ultimately useless, then it seems we stand to gain by eliminating them from our own arsenal whether potential adversaries do or not.

    • Seb (History)

      Surely what matters most in the situations you are describing is the impact the weapon has, not the technology used to make the explosion.

      If one state starts wiping out conventional forces with abandon using low yield nuclear weapons, that has strategic implications which make some kind of strategic response rational and perhaps even necessary.

      The response will probably be driven by the precise context of how the weapons are used.

  3. anon (History)

    Is it just me, or is this something confusing about this argument. First, you contend that other nations will seek low-yield nuclear weapons because they are of value in deterring an adversary with conventional superiority. But they won’t violate a treaty on low yield nuclear weapons, even if it can’t be monitored with any level of certainty, because there would be no value to them of acquiring clandestine low yield nuclear weapons. Sorry. it doesn’t fly. If they value the weapons, and don’t fear the consequences of getting caught cheating, they’ll seek to acquire the weapons.

    Also, although you’re definition of adequate verification is absolutely correct, in a quaint, 1970s kind of way, it completely misses the debates and discussions about adequate/effective verification during the 1980s. Even if we could monitor to a level of certainty that would allow us to detect militarily siginificant cheating in time to respond, we would also need to be able to detect politically significant cheating so that we could challenge our adversary (i.e. the SU) about their intentions and reassure ourselves about our commitment to maintain our security.

    This was of primary importance during the Reagan Administration. You could never admit that you were willing to let some violations slip through. There is no conceivable way that the U.S. would sign a ban on some type of weapon if we had very low confidence in our ablility to detect a significant amount of cheating, even if that cheating wouldn’t undermine our national security (actually, Rumsfeld broke this rule in supporting the 2002 Moscow Treaty, saying we didn’t care what Russia did under the Treaty so we didn’t need a verification regime, but that treaty didn’t require us to do anything either…).

    If its important to stop nations from acquiring low yield nuclear weapons then its equally important to have at least some level of confidence that they have not acquired those weapons.

    I also find your U.S.-Russia case unconvincing because, well, if Russia uses low-yield nuclear weapons along its periphery, the size of U.S. nucler weapons isn’t really relevant to the U.S. response. Russia’s calculation on whether the U.S. (or more appropriately, NATO) will respond has far more to do with the question of whether Russia believes it has stomped on U.S. or NATO vital interests. It may believe it can keep its attack below that level so that NATO won’t respond, but has to recognize that, if it exceeds that level, NATO is likely to respond regardless of the yield of U.S./French/UK/NATO nuclear weapons.

    Its not about the size of the weapon, but about the size of the challenge.

    • Jeffrey (History)

      I agree with Anon.

      If Russia is cheating on the CTBT to develop low-yield weapons, they’ll also cheat on what amounts to an unverifiable promise to develop low yield weapons.

      And, if Russia is *not* developing low-yield weapons, an unverifiable promise doesn’t demonstrate that, particularly to suspicious minds.

      Our problem remains that we don’t know what the Russians are doing.

      The simplest solution to the problem of Russia’s possible development of low yield nuclear weapons is to bring the CTBT into force to prevent explosive testing, while creating provisions for additional US-Russia test site transparency and confidence building measures that follow the successful 1988 JVE.

    • Jonah Speaks (History)

      According to this article, Pakistan already has low-yield nuclear weapons: “The salvo launch demonstrates that Pakistan is steadily improving its counterforce capabilities against Indian armored thrusts as part of the Indian ‘Cold Start’ doctrine with the option of using low-yield, boosted fission, plutonium warheads in the possible range of 0.5 to 5 kilotons in case of a breakdown of conventional defenses,” he [Lt. Gen Khalid Ahmad Kidwai] said.

      Pakistan conducted only 6 tests over a 3-day period, while the Soviet Union/Russia conducted hundreds of tests over a period of decades. How is that Russia still needs to conduct more tests before fielding low-yield nukes? Or do the wonks disagree that Pakistan has low-yield nukes?

    • Jeffrey (History)

      I have long expressed skepticism towards Pakistan’s claim to have developed mininukes.

  4. shaheen (History)

    Replace “low-yield” with a specific value (1, 5, 10 kt, whatever) and the vacuity of the argument becomes more apparent.

    To be sure, for most political leaders, all things equal, lower-yield options would make it less difficult to be “comfortable” with the use of nuclear weapons than with higher-yield options.

    But that is valid only once you have decided to cross the threshold. It does not weaken the firebreak. For the vast majority of them, a nuke is a nuke, period. (And yes, I believe that it is true even in Islamabad.)

    Folks, stop channeling the 1960s.

    • MK (History)

      Verifying compliance with a specific low yield value can be more of a challenge than monitoring zero yield.

    • Jeffrey (History)

      I agree. All things be equal, zero is the easiest number to verify.

      One other point: The CTBT also prohibits explosions — a phenomenon that can be detected at distance — instead of possession of specific warhead designs, which would require unprecedented intrusiveness.

    • Mark Gubrud (History)

      I don’t see anyone arguing we should abandon the CTBT and only seek to ban explosions below a certain threshold. That would indeed make no sense.

      A low-yield nuke ban can be beneficial even if it is not strongly verifiable, which is to say, even if it is violated. But also, if a nation makes nuclear artillery a major component of its war plan, this will very likely be visible by independent means. In any case, complying states would not be accepting a great security risk.

      Very few arms control agreements are 100% verifiable. Numerical limits on nuclear weapons are not. The ban on chemical weapons is not.

      The appropriate questions to ask are not about absolute verifiability, effective verifiabiltiy, parity, etc. Rather, What are the benefits? What are the risks? Then you weigh one against the other.

      The proposed low-yield ban seems to me to carry very little risk other than that it may be a distraction from other objectives and it may be a target for crazies.

      Its benefit would be to establish a norm against the idea of using low-yield nuclear weapons as a counter to conventional weapons. It would further fence in the nuclear threat and be another step on the road to nuclear zero.

      It’s rather like the conceptually odd INF Treaty which I don’t think you would propose to do away with just because long-range missiles can almost always be brought down on closer targets.

      The strongest argument against is that this would be a distraction from other objectives. But maybe it’s not a bad idea to let a few more flowers bloom.

    • Jeffrey (History)

      It’s not simply a distraction, it would be a legal nightmare for the United States to take Spratt-Furse and turn it into a treaty obligation given the nature of primaries for modern thermonuclear weapons. There is simply no way to verify this agreement.

      What one can verify is the absence of nuclear weapons explosions, down to yields of a few tons, under the CTBT and set of supplementary test transparency measures. It would be much better to put in place such monitoring mechanisms to prevent the testing of low yield weapons.

      That might leave us with Pavel’s problem — adventurous Russian designers willing to deploy untested weapons — but I don’t see a solution to that problem given that the Russian’s certainly aren’t going to let us look at each weapon to determine its design type and yield. A ban on nuclear explosions of any yields gets us as near to a ban on development of new low-yield weapons as anything.

    • Pavel (History)

      The words I used were “competent designers,” not “adventurous”. I still think the problem of “untested weapons” is very much artificial. However, as I understand, the warheads that are being deployed now are based on a tested design, so even the Russians are not really adventurous at this point.

    • Jeffrey (History)

      Well, some “competent” designers are more risk averse than others, hence my use of the term “adventurous.”

  5. Cthippo (History)

    It seems to me that the whole argument turns on a value judgment.

    If you believe, as the original authors do, that smaller weapons are more likely to be used then it makes sense to try to limit them so as to increase the threshold for use.

    If, on the other hand, you believe that smaller weapons are not more likely to be used then a treaty banning them is, if not pointless, then certainly a lower priority.

    The key to the whole thing is this value judgment as to whether yield affects usability.

    This gets into sort of a human nature effect where people disagree on the core value judgment, but instead of discussing that they argue endlessly about the peripheral issues and technicalities.

    • anon (History)

      Totally agree. If you start from the assumption that a nation would only use nuclear weapons in the most dire of circumstances (another value judgement where debate is possible, but I buy into this judgement), then the issue of nuclear use rests with the existence of that circumstance, regardless of the yield of the weapons. If Pakistan believed that its defeat was imminent, and it planned to use nukes as a last resort, then the use would occur regardless of the yield. Lower yield may make the threat of use more credible, and therefore possibly enhance Pakistan’s ablility to deter India from creating the “last resort” circumstance, but it does not make use more likely if deterrence fails and Pakistan ends up in this circumstance.

      As I said before, its the size of the challenge, not the size of the weapon, that determines whether a nuke will be used.

    • Jonah Speaks (History)

      One is a quibble about vocabulary: I see the term “value judgement” about what is really a “fact judgement.” More precisely, it is beliefs about the conditional probabilities that if certain events occur other events will follow. E.g., if nuke yield is low, nuke use is more likely (or not). The reason people disagree on these fact judgements is because there is currently no data upon which to build consensus estimates.

      Value judgements would be statements such as: Pakistan should not be the first to use nuclear weapons, even if Indian tanks invade, not even as a last resort. Pakistan needs to develop or purchase conventional tank-killer weapons, even if they cost more or are less effective against tanks than tactical nukes.

      Estimates of conditional probabilities can influence choices. Here is one way to analyze the choices:
      Pp = probability Pakistan provokes India (e.g. 20% per year).
      Pi = conditional probability India invades Pakistan after provocation. (This is India’s choice, but analyst’s probability estimate.)
      Pt = conditional probability Pakistan uses tactical nukes on invading tanks. (Pakistan’s choice, but analyst’s probability estimate.)
      Pf = conditional probability of full-scale nuclear exchange, if Pakistan uses tactical nukes.

      If yields are lower, Pf is likely to be lower, influencing Pakistan to set Pt higher. The effect on Pi is ambiguous, if the fear of nuclear war is what deters India. Conditional on India invades, the likelihood of full-scale nuclear war is PtPf, which could go either up or down, depending on by how much Pf goes down and by how much Pt goes up. Because the effect on Pi is ambiguous, so is the effect (if any) on Pp, and the ultimate probability of full-scale nuclear war between Pakistan and India.

      Without further information on how low yields impact PtPf, we cannot say whether this proposal to forbid low yields, even if fully enforceable, would either reduce or increase the odds of full-scale nuclear war.

    • Jonah Speaks (History)

      I would amend my definitions of Pt and Pf in the analysis above to include any nuclear use on any Indian target, not just tactical nukes on tanks. Hence,
      Pt = conditional probability Pakistan uses any nuke on any Indian target, and
      Pf = conditional probability of full-scale nuclear exchange, if Pakistan uses any nukes.

      Just to balance out the value judgements: India should not invade Pakistan in response to terrorist attacks. India should not order massive nuclear retaliation in response to limited nuclear attacks.

    • Jonah Speaks (History)

      Although not posted above, the authors view tactical nuke use as a very slippery slope to full-scale nuclear war. Here is a flavor of what they say: “The inability of political leaders to control the outbreak of a low-yield nuclear attack, direct ongoing operations, avoid escalation or bring hostilities to a conclusion vexed NATO planners for decades. There was always an unavoidable paradox that NSNWs [tactical nukes] had to be placed at risk of conventional attack in order to deter it.” “Strategists also concluded that, if nuclear weapons were used on a European battlefield, things would descend into chaos and anarchy in ‘very short order’, at which point escalation to all-out nuclear war might be ‘unavoidable’.”

      Amending my analysis again:
      Pt = conditional probability Pakistan uses any tactical nuke,
      Ptf = conditional probability of full-scale tactical nuke exchange,
      Ps = conditional probability Pakistan (or India) uses any strategic nuke,
      Psf = conditional probability of full-scale strategic nuclear exchange.

      In the event Pakistani policy is to use tactical nukes against an Indian invasion, the probability of full-scale nuclear exchange is PtPtfPsPsf. If we put reasonable numbers on this, e.g., (2/3)(3/4)(2/3)(1/2)=(1/6).

      In the event Pakistani policy is to use only strategic nukes against an Indian invasion, the probability of full-scale nuclear exchange is PsPsf. Reasonable numbers might be (1/3)(1/2)=(1/6). Same probability. The only difference is that using tactical nukes first brings an additional 50% chance of limited nuclear war. This is a very bad outcome, since “the mutual devastation would still be unprecedented.”

      The authors view strategic nuclear weapons as an adequate substitute for tactical nukes, even for purposes of deterring invasion. Perhaps it would be useful for the authors to explain this better. In particular, strategic nuclear use does not have to be unlimited nuclear exchange, as many people assume. It could instead be use or threat of use, perhaps with advance warning, of one or a few bombs. Since the defending side normally desires to keep its own territory more than an invader wishes to conquer it, the defender will plausibly “win” any limited strategic war that follows an invasion. Hence, a rational would-be invader will normally be deterred.

  6. Mark Gubrud (History)

    The point of a public agreement not to have low-yield nuclear weapons is to publicly renounce any notion that such nuclear weapons can be used without the likelihood of escalation to higher yield weapons.

    The possibility of cheating is almost irrelevant. The fact that you had publicly declared you did not have and would not use a low-yield nuke means that if you do pull one out and use it, there is even more reason to expect that the response from a nuclear-armed adversary would be the use of a higher-yield weapon. Or, if you use it against a non-nuclear state, that is an even more dastardly act that would earn even greater outrage from the rest of the world (which may or may not choose to intervene).

    On the other hand, not banning such weapons leaves open the possibility that nuclear states will deploy and come to rely on them as a counter to superior conventional forces. That will tend to be rather visible, integrated into operational planning, force structures and exercises.

    We might also be concerned about the possible use of low-yield warheads at sea, in space, or for missile or air defense. None of which would be unprecedented.

    I think one can reasonably judge that, if arms control efforts must be prioritized, this one might not be the highest priority. But in itself, it would be a good thing.

    • anon (History)

      So, a nation that used a low yield nuclear weapon (instead of a higher yield weapon) after pledging not to acquire that weapon, would be punished more greatly than a nation that used a nuclear weapon without having made such promises? This makes no sense to me. The international horror and reaction at the first use of nuclear weapons would be so great that the added “horror” at the breaking of a treaty vow would be diminishing small. Its the use of nuclear weapons, not the breaking of the treaty, that leads to the response.

    • Jeffrey (History)


      My concern is that the treaty would be so unverifiable that it would make the problem worse, if we define the problem (as I do) as uncertainty about Russian development of low-yield nuclear weapons.

      If Moscow makes a public agreement, a substantial fraction of the community will immediately accuse them of cheating on the agreement as they accuse them of cheating on the CTBT.

      Resolving this concern requires better verification. In that case, the CTBT plus test site transparency will build more confidence than an unverifiable pledge.

      If, however, the US and Russia want to make a joint statement that emphasizes the low yield nuclear weapons do not offer an escape from the truth that a nuclear war cannot be won and must never be fought, that would be fine.

    • Pavel (History)

      Jeffrey: Is it really “a substantial fraction of the community” that believes that Russia is cheating on CTBT? I thought it’s a fringe, even if a vocal one.

      Also, it’s not clear why the alleged cheating is linked to development of *low-yield* weapons. If one assumes that Russia is cheating than it is likely to work on high-yield weapons as well.

      But overall, this whole idea that one needs nuclear tests to develop new weapons is an artifact of the U.S. discussion. As I understand, Russians are pretty confident that competent designers don’t need new tests.

    • Jeffrey (History)

      How about “a vocal, slightly insane minority”?

    • Mark Gubrud (History)

      Perhaps we can agree that the “punishment” for using a low-yield nuke would not be less if a nation had pledged not to have them than it would be otherwise.

      It does seem to me that this would have a substantial impact on any estimation that you could use a low-yield nuke against an enemy that possessed only higher-yield nukes without expecting the latter to be used in response.

      We should remember that the effects of high-yield nuclear weapons can be adjusted by choice of target and altitude. Therefore the use of a higher-yield weapon in response to the use of a low-yield one need not be escalatory in terms of the actual damage caused, if nuclear bursts can be used for signaling at all. OTOH, the use of many low-yield nukes would certainly seem to invite the use of some high-yield ones in response.

      Yes, it is crossing the nuclear threshold that most likely matters. An agreement like this would reinforce that assessment against any doubts.

    • Jonah Speaks (History)

      I see this proposal mainly as an attempt to establish a new international norm. Their proposal is to ban both the use of <5kt "tactical" nukes and the possession of same. Just as with chemical weapons, we can parse this into two main components: First, ban the use, second (if feasible), ban the possession.

      I would suggest the first treaty ban the use of tactical nukes, and that this be combined with certain transparency measures. In particular, I would suggest all nuclear-armed states declare both the number of warheads in their possession and the minimum and maximum design yields of these warheads. Later, one can work on additional treaties to verify these declarations, and to reduce or eliminate tactical weapons still in stock.

      I believe this proposed norm has merit and should be regarded as on a par with CTBT or fissile material cut-off. For nations that possess nuclear weapons, strategic nukes should be regarded as adequate for invasion deterrence, not necessarily "maximal" or "minimal", but adequate. If a nation uses tactical nukes against a nation that has strategic nukes, they are acting foolishly. Tactical nukes can also be used offensively and fools are everywhere, so elimination of stockpiles is also a worthy goal.

  7. Jonah Speaks (History)

    For the moment I won’t comment on regulating the low end of nuclear yields. I will simply ask the opposite question, why not regulate the high end of nuclear yields? Should there be a maximum yield on nuclear weapons?

    Risk is measured not simply by the probability of nuclear war, but also by the extent of damage if there is one. If we could regulate both the number of nuclear weapons and their maximum yields, then we limit (subject to targeting practices) the maximum damage from nuclear war, if such war occurs.

    • Tim (History)

      Practical concerns already limit the maximum yield, because 10 100kT weapons produce more destruction than 1 MT weapon.

      Beyond 10MT or so, the additional energy mostly just goes into heating up already destroyed objects to even hotter temperatures, and increasing the energy that escapes out the top of the atmosphere.

      There’s a reason why although arbitrarily large nuclear weapons are possible, they are generally not deployed.

    • John Schilling (History)

      Delivery systems also pose a practical constraint on warhead size. Aside from a few obsolescent Russian and Chinese ICBMs, there really isn’t anything suitable for delivering warheads of much more than about a megaton, and if we need to codify that in a treaty, missile size is more unambiguously verifiable than warhead yield.

  8. George William Herbert (History)

    I would connect some dots here, with the escalation analogies associated with CBW. The US’ variously implicit/explicit position was more or less “these are all WMD, and will be responded to in kind” more or less put everyone on record that we WOULD – or asserted we would – escalate to some yield of nuclear weapon should we be WMD’ed.

    Russia thinking “Oh, mininukes are ‘safe'” would be contrary to that. Perhaps they (or someone else) might believe it, but that would be a risky policy assertion.

    Two, terminology is hindering the discussion. “First strike” meant something specific in the US / USSR strategic situation, most commonly associated with counterforce or decapitation strikes. Using it to describe a first use by Pakistan on Pakistani territory or border regions after a massive conventional invasion is challenging abuse of the terms.

    It seems likely that a lot of the fear of small nukes is that Pakistan has quietly asserted that they would consider such use in the case of a successful “Cold Start” attack, and that that has worked to deter India from considering that a good strategy. It has worked because use of WMD defensively on ones own territory has not pushed many of the international fear buttons, and India can’t generate a good doctrine under which they would begin offensive nuclear war to respond to such a defensive strike.

    The US/USSR discussion is ultimately a red herring. The apparently successful deterrence of Cold Start is the issue here. The question is, whether that was bad or not…

  9. Jonah Speaks (History)

    Jeffrey, It looks like the http links for the authors’ articles have moved. The shorter Insight article can now be found here:

    I don’t know what happened to the authors’ longer study referenced therein and above on this blog-page and your previous blog-page on this topic. That link no longer works, and I can’t find it using Google Scholar.