Four Shibboleths of Nuclear Strategy

I’ve been meaning to write this post for awhile, based on all the podcasts Pranay, Austin, and I have done and after a series of CNSP seminars and workshops this semester, not to mention what has been appearing in the popular media on nuclear weapons. It is not fully-fleshed out and is just the beginning of perhaps a more detailed academic article on what I call the “four shibboleths of nuclear strategy,” or phrases/concepts that are often lazily thrown out and taken as given, or true, by not just the general public but nuclear scholars and practitioners as well. But are they? At the very least I hope this post will provoke some thought, maybe some outrage and hate mail, about the things we have taken as received wisdom, but which perhaps should not be. And what does it mean for nuclear deterrence and nuclear strategy if they are not.

“Strategic Stability is Good”

It is taken as an article of faith in the nuclear community and in popular discourse that “strategic stability” is good for American and global security. And that things that might undermine or threaten “strategic stability” are, on their face, dangerous and bad. The concept of “strategic stability” is poorly defined, has different meanings to different communities, is taken to be unquestionably good and (gasp!) god forbid if it is undermined, and increasingly obscures more than it helps.

First, there is no shared definition of “strategic stability.” For some, particularly senior national leaders, strategic stability may, and often does, mean the absence of great power war or war in general. In this conception, it is hard to argue that strategic stability is bad. But for the nuclear community, which often invokes the phrase, it has a much more specific and narrow meaning, one that is somewhat tangential from the broader conception most have: “no reciprocal fear of first strike,” or the absence of first (strategic nuclear) strike incentives because both sides have secure second strike forces capable of inflicting assured destruction—that is, a condition of Mutually Assured Destruction (MAD). This is a far narrower, and different, definition of strategic stability than foreign policy practitioners assume, and often leads them to accept arguments from nuclear proponents of “strategic stability” that may actually result in great power war, the very outcome they primarily seek to avoid. Why is this the case?

As Austin, Pranay, Charlie and I discussed, the implication of nuclear strategic stability is the so-called stability-instability paradox. That is, a condition of “strategic stability” at the strategic nuclear level disincentivizes escalation to that level—and strictly to that level—for fear of mutual suicide. But, it opens up space for instability up to that level, including conventional war or even limited nuclear use in the original Glenn Snyder formulation of the concept. Who is instability at lower levels of violence bad for? Well, for one, status quo powers who are not seeking territorial revision, because instability at lower levels of violence opens up the space for revisionist powers to attempt conventional aggression and use strategic stability at the nuclear level to stalemate or deter attempts to unwind territorial aggression. Who else is it bad for? Allies to whom a state may extend deterrence, because they will be the playground for the lower levels of violence. So the very logic of strategic stability, as narrowly defined by the nuclear community may actually be less than ideal for status quo powers and for their allies, and good for revisionist states seeking territorial revision. Since the United States seeks no territorial revisions in Europe or the Indo-Pacific but has a network of allies to whom we formally extend our nuclear umbrella in those regions, it is easy to see why the blanket acceptance of “strategic stability is good” is problematic. To be blunt, at the moment, strategic stability as defined in the nuclear sense is good for Putin, but dangerous for the United States and our NATO allies. One can imagine a similar dynamic with China in the future—the reason why China seeks nuclear strategic stability may be precisely to open up space for territorial aggression, as Putin has done.

Indeed, as we have discussed ad nauseum, the United States has actually sought to induce what Kent and Thaler in a RAND Report referred to as “optimum instability” at the strategic level—inducing enough doubt in the Soviet Union and now Russia (and China?) that they could have sufficient survivable forces to inflict unacceptable damage on the United States and its allies with a force remaining after US counterforce and counter-SSBN operations. As Austin and Brendan Green have shown, the US invested a lot in hunting Soviet subs late in the Cold War that gave Moscow sufficient anxieties about the robustness of “strategic stability.”

Just enough instability—what is “optimal” is something we should further explore, but my conception is it is a competitive process and there will be periods of more or less instability and it is not a static concept—at the strategic nuclear level does what? In theory, a little instability at the top should produce stability at lower levels of violence, deterring the revisionist power from even starting the music for fear that, at the very top, the United States has escalation dominance through meaningful damage limitation. The mere fact that the adversary may worry about this is what creates a deterrent effect that generates stability at the conventional and regional nuclear level—indeed reducing the probability that damage limitation options have to ever be considered or executed. This downdraft effect of “strategic nuclear instability” may have the counterintuitive effect of generating more peacetime strategic stability in the broader sense (of course, if deterrence should fail at lower levels, there are crisis instability effects that kick in but the logic of damage limitation options inhibiting adversary escalation to the top remains).

Indeed, if “strategic stability is good” is the parent shibboleth, its child is “MAD is a fact.” For decades, the United States consciously sought to escape MAD and undermine strategic nuclear stability precisely so that the Soviet Union could not find space for aggression against NATO allies. MAD is a variable, not a fact, and one the United States sought to manipulate to achieve broader stability, by undermining the confidence the Soviet Union had that it possessed an assured destruction capability.

Thus, the notion that this or that posture adjustment may “undermine strategic stability,” which some deploy to shut a debate down because it is just so widely assumed that “strategic nuclear stability” is on its face and indisputably good…. may not be true. In fact, it may be good…but for Russia. Indeed, putting pressure on the strategic forces—either through offense (counterforce or counter-SSBN operations) or defense (e.g. missile defenses)—of a revisionist adversary may be both theoretically, and has been at least empirically, a more robust deterrent and extended deterrent/assurant, than an acceptance of (nuclear) “strategic stability” as an unadulterated good. Next time you hear someone try to end a conversation or debate by claiming “but that will undermine strategic stability,” it is worth asking them “and…why is that bad?” especially if we fancy ourselves as a territorially status quo power and purport to care about allies and extended deterrence (and by extension, nonproliferation by allies).

“Arms Races are Bad”

A related supposed truism that is often thrown out as an attempted conversation stopper is “but, that will cause an arms race” on the assumption that arms races and vertical proliferation are indisputably and monotonically bad. Arms races may be costly (though the cost to support one may actually improve the defense industrial base and the overall US economy, but I digress), but there is quite literally zero empirical evidence that “more nuclear weapons” increases the risk of nuclear employment. If we all agree that the goal of nuclear policy is reduce the risk that nuclear weapons are employed, then it is actually more empirically supportable that arms competitions have a much more subtle relationship to employment risk. Indeed, I have previously argued—in the decades after India and Pakistan first tested nuclear weapons in 1998—that the greatest risk of nuclear employment or use was when both sides had very small arsenals, because both sides acutely worried about survivability and use-them-or-lose them pressures. It is no coincidence that détente in the Cold War appeared at a time when the US and Soviet Union possessed their largest arsenals—and it was the Soviet’s relative inability to hold the US homeland at risk that triggered the Cold War’s most acute crisis in Cuba in 1962, not a possession of too many ICBMs.

Indeed, would anyone really credibly argue that a US-Russia nuclear relationship would be more stable if both sides only had 50 strategic nuclear weapons? Russia would have incredible incentives to use early and use everything at those numbers. No one is arguing that both sides should build up to 50,000 weapons again—at some point management risks as Scott Sagan famously argued begin to dominate the curve—but likewise, maybe there is an optimal number somewhere in the middle where both sides are more comfortable in some survivability but don’t possess tens of thousands of warheads, some Goldilocks zone in the middle, as it were.

What is that number? Who knows. 1550 is arbitrary. So is 2000 or 2500. But I have found that one response to Pranay’s and my Foreign Affairs article is handwringing that we are promoting an arms race. This was the essence of Bob Gallucci’s response to us. Without any evidentiary or empirical basis, many attempt to refute reasonable deterrence discussions with “but that will cause an arms race” assuming that that is the final word and on its face is an argument. It is not. It is also bizarre, and lazy, to refer to modest adjustments—moving, say 500 warheads, from the existing stockpile to deployed status on already active missiles—as an “arms race” when Russia and China are halfway down the straightway already while we are stuck in the starting block, but again I digress.

However, one sees these quotes repeatedly in the pages of the New York Times or Wall Street Journal as if any adjustments in posture—whether more deployed forces, or different forces in small numbers (like 50 SLCM-N)—are arms races and that they are indisputably risky, rapid, and uncontrollable. For one, there is no evidence again that more or different is necessarily more risky, or that fewer introduces less risk, when in fact there is more empirical evidence to support the opposite. But, second, that is not how arms “races” work—arms competitions are more like arms “walks” or even “crawls” given how long force posture adjustments anywhere actually take. The way arms competitions unfold allow states to make somewhat rational decisions and responses to each other, and while arms competitions are not static and there is undoubtedly a slow action-reaction cycle (it’s not like we can go from 1550 deployed to 3000 overnight, or to 10,000 even this century), there is nothing inherently risky about vertical proliferation and nuclear use. And they may even provide leverage for arms control. Again, arms competitions may be costly, but the relationship between vertical proliferation and nuclear use—the shared outcome we are all seeking to avoid—is far more complicated and uncomfortable than many arms controllers or disarmers would care to believe or accept.

So are more or different deployed nuclear weapons bad? Not necessarily. They may even be stabilizing if they better deter two peer-ish adversaries and assure nervous allies from getting their own nuclear weapons. Is there a point at which more becomes “too much?” Perhaps. But that’s also what may ripen the conditions for arms control. But, at the very least, there is a goldilocks zone for vertical proliferation—and it is perhaps a wide one, and relative, and perhaps even arbitrary—but I would argue that the greatest risk of nuclear use in a nuclear relationship is when numbers actually get too low. Just something to think about the next time someone believes they have won a debate simply by saying “but, we must avoid arms races.” The goal is not to win or avoid an arms race, it is to avoid conflict and nuclear employment.

“Escalation is Uncontrollable”

This one drives me nuts. This phrase is thrown out often. In fact, my young self threw this out there a lot, so I am writing this against 2014 Vipin too. There is a widespread trope out there that any nuclear use will lead to “uncontrollable escalation.” There are two versions of this. The first is that any nuclear use will be catastrophic for humanity and end the world. This was certainly a concern in the event of a reciprocal strategic exchange between the US and the USSR during the Cold War where tens of thousands of nuclear weapons would have been lobbed at each other if escalation had reached that level.

But not all nuclear weapons are created equally. We have traded a very small risk of massive strategic nuclear exchange for what is, in my view, a higher risk of limited nuclear use in a regional conflict—Russia in Ukraine, Pakistan against Indian military targets, DPRK etc. In the October 2022 scare, the concern was not Russian strategic nuclear use of large warheads against US or allied targets, but the possibility that Putin may employ one or several much lower yield nuclear weapons, that may even be hard to attribute as nuclear weapons use initially, to prevent the collapse of Russian lines in the theater. Had Putin made the mistake of using nuclear weapons in this mode—had Russian forces collapsed, which they did not—it would not have been world-ending or even city-ending. Many of these nuclear weapons are perhaps an order of magnitude or more smaller than the weapons used on Hiroshima and Nagasaki and are roughly of the same order of magnitude in explosive power as some of our largest conventional weapons. Indeed, that is their point. As their conventional magazine depth depletes, in desperation, a Putin or a Pakistan or a North Korea could reach for one or several lower yield nuclear weapons to turn the tide of a conventional war they are losing, and keep it all on the battlefield.

It is here that one often hears—after first employment—the related argument that “escalation is uncontrollable” and that even a single nuclear weapon—even if it does not itself cause catastrophic damage—will lead to escalation that will. However, escalation is always a choice. It is never automatic. Had Putin employed nuclear weapons in 2022, the US and NATO response would have been severe but almost certainly non-nuclear—seeking to deter further employment of Russian nuclear weapons without provoking them. This was a choice the US could make. Russia could then similarly exercise a choice about how to respond. Given US regional and strategic nuclear capabilities and Russian fears about US damage limitation capabilities, there certainly would have been no automaticity to escalation, and hopefully robust deterrence across a range of capabilities and geographies to influence Putin’s choices.

Similarly, though I have previously expressed concerns about India and Pakistan throwing roundhouses at each other, most recently this past May, due to the risk of missiles going awry (because that’s never happened) or leaders being painted into a corner that makes escalation more likely, it is again never automatic. In fact, the related notion of “inadvertent” or “accidental” escalation if, for example, a nuclear system is accidently (or not accidentally) hit in a conflict by a conventional munition, similarly elides over the fact that leaders always have a choice about how to respond given the context. In the India-Pakistan case, India’s accidental Brahmos launch into Pakistani territory in March 2022 was embarrassing for India, but was accepted as an accident by a Pakistan which had no desire to escalate—so there was no escalation.

Leaders who want to escalate can escalate, and leaders who want to de-escalate can always find off-ramps to do so, even in the tightest of corners. My only point is that the notion of “accidental” or “uncontrollable” escalation is often ascribed as an existential risk in a nuclear world, and my own intellectual evolution on this point is that escalation is always a choice and is never automatic.

The popular myth that the next use of nuclear weapons will “end the world” is simply not true and this myth has allowed scholars and practitioners to avoid a very uncomfortable, but absolutely crucial and understudied issue about which we need much more careful thinking: escalation management. After all, if escalation is uncontrollable, there is little point in thinking hard in escalation management. But that view is wildly misguided and some of the most important knots today in nuclear strategy, and strategy more broadly, are precisely over managing escalation if nuclear weapons are employed in the midst of an intense conventional war. We have to think much more carefully and hard as a system about escalation management to ensure off-ramps are always available to leaders should they want to avail themselves of them and, likewise, that flexible conventional and nuclear response options are always available to calibrate responses with an eye on managing escalation, which is always a choice and never automatic or uncontrollable.

“Deterrence is ‘The Threat that Leaves Something to Chance’”

This is my least well-formed and perhaps, at least for the academics out there, perhaps going to be my most controversial shibboleth, and not coincidentally comes after the above. Schelling famously argued that deterrence can be achieved by the “threat that leaves something to chance.” This is often invoked by those who seek to imply that “deterrence is easy” and that the mere risk of a nuclear threat being carried out is sufficient to deter an adversary—and is often used by those who support minimum deterrence postures or strategies, who argue a state doesn’t need certain survivability or assured destruction capability, when even an uncertain force will suffice. I think Schelling in fact meant the very opposite with this phrase than is popularly ascribed to it, that states had to bake in or hardwire third-party or external mechanisms to enhance the credibility of a nuclear deterrent threat, precisely because he believed nuclear deterrence was hard. In my view, Schelling was much more careful about the use of this phrase than those that followed. Many use the phrase incorrectly in my view, so my beef isn’t with Schelling so much as how scholars and practitioners have interpreted or misinterpreted his original meaning. To be fair, he could have been clearer too.

For many, the phrase implies the issuance of a threat where the mere chance that it may be carried out has a deterrent effect on the adversary—rolling the dice so to speak. But this was NOT how Schelling envisioned it. For Schelling it was not “I may or may not carry out the threat, you may not be sure,” it was importantly also that “I may not be sure!” He writes:

The key to these threats is that, though one may or may not carry them out if the threatened party fails to comply, the final decision is not altogether under the threatener’s control. The threat is not quite of the form “I may or may not, according as I choose,” but, has an element of, “I may or may not, and even I can’t be altogether sure.” Where does the uncertain element in the decision come from? It must come from somewhere outside of the threatener’s control. Whether we call it “chance,” accident, third-party influence, imperfection in the machinery of decision, or just processes that we do not entirely understand, it is an ingredient in the situation that neither we nor the party we threaten can entirely control.

In the extreme, an example of what Schelling was referring to here was actually some form of “deadhand” even in the event an adversary, for example, decapitates leadership a retaliatory response is guaranteed (think Perimtr). So what he is really talking about is a “threat that leaves nothing to chance” (Austin and I argue over which of us stole the phrase from the other, but I still maintain I used it first ). Contrary to how this concept is often used and abused in the field, Schelling was not, in my view, referring to the mere risk of escalation as a deterrent, but its certainty, or the intentional wiring of escalation as a deterrent.

This will be more familiar to at least some US nuclear policy officials than Schelling is often portrayed. This makes tripwire and forward deployed nuclear weapons Schelling (he uses these as examples of “inadvertent” war but it is really “advertent”) more consistent with “threat that leaves something to chance” Schelling than is often believed. Mechanisms to wire escalation into US plans was, in my view, what Schelling was referring to—at every level, whether limited nuclear options, or ultimately damage limitation (even though Schelling later recoiled at damage limitation options with the growth in Soviet forces, US planners never really walked away from this version of his concept). Stay tuned for a book my friends and colleagues Caitlin Talmadge and Brendan Green are working on, on the advertency of “inadvertent” escalation. Part of the confusion rests with Schelling’s initial word choice which has led to decades of scholars and practitioners arguing that “uncertainty” is what deters when in fact, I believe Schelling leaned closer to the view that “certainty” is what deters. More simply, perhaps, I think Schelling meant to imply here that it is not uncertainty which enhances the credibility of a threat, but certainty. The point is, I think many have been using the phrase “threat that leaves something to chance” incorrectly for decades. So next time you hear someone say “it’s the threat that leaves something to chance” which deters ask yourself, is it really, or is it “the threat that leaves nothing to chance” which deters.

Conclusion

If you’ve made this this far, how much did I piss you off? Just kidding—sort of. I anticipate some of these will be challenged, provoke outrage, maybe make you think a little. That’s good! I’m not even sure I’m fully right about any of these, and the arguments are unpolished and early, particularly my interpretation of Schelling. Much more work to be done on all of these issues, and this was just designed to get the wheels turning and a debate started on these long-held so-called “truisms” of nuclear strategy which may not in fact be true. Also let me know what shibboleths I missed. I am debating what, if anything, to do with this little op-ed here (there is a lot of work out there that I did not engage with or cite and that is simply a product of this medium and me finding myself in between kid stuff to muse finally for awhile, so no offense meant to all those that have tread on this ground previously), so let me know what you think. With that, happy holidays all!