Countering Competitive Risk Compensation: Principles for Reducing Nuclear Risk on Net

In the context of strategic competition, efforts and interventions to reduce the risk of destructive nuclear or conventional conflict can change the incentives of leaders and institutions in ways that undermine risk reduction. As risk reduction is not the only objective that leaders may pursue, direct reductions in risk can accordingly embolden more aggressive behavior...

FacebookTwitterLinkedInEmailCopy Link

In the context of strategic competition, efforts and interventions to reduce the risk of destructive nuclear or conventional conflict can change the incentives of leaders and institutions in ways that undermine risk reduction. As risk reduction is not the only objective that leaders may pursue, direct reductions in risk can accordingly embolden more aggressive behavior as the leaders of opposing states seek to gain relative advantage and negotiating power. In nuclear strategy, phenomena such as brinkmanship, the security dilemma, and the stability-instability paradox pose a variety of complexities and trade-offs that can accordingly make it incredibly difficult to determine the net effect of various risk reduction efforts.

How then can we reason about the best ways to reduce risk in the presence of competitive behavior? Some experts argue that many interventions are essentially hopeless, with game theoretic models implying “the risk of inadvertent escalation is independent of position, posture, command, control, and technology.” In this article, I seek to characterize the complexities of nuclear risk compensation and to lay out principles for achieving net reductions in risk despite these challenges. In short, by focusing on the incentives, capabilities, and knowledge of decision-makers it is possible to reduce risk while limiting, mitigating, or entirely avoiding risk compensation.

The Nature of the Problem

Probability x Consequence

Risk is a matter of both probability and consequence: nuclear risk reduction involves both avoiding nuclear war entirely as well as mitigating worst-case consequences in the event deterrence fails. Would escalation be controllable? How many weapons would be used and to what effect? How likely would states be to attack targets in cities vs. remote military installations and would it matter? How survivable and recoverable would the long-term effects of worst-case scenarios be on global civilization and the climate?

Interventions that reduce the probability of nuclear war may not lead to a reduction in the *risk* of nuclear war if the probability of conflict is reduced merely by making the stakes of war more severe. As the credibility of deterrence rests on “the threat that leaves something to chance”, an extremely destructive deterrent with a clear zero percent chance of use under any circumstance cannot actually deter any behavior.

Reasoning About Probability Over Time

With no use of nuclear weapons in war since 1945, it is hard to know exactly how risky various nuclear close calls have been, let alone how different interventions have altered the overall risk of nuclear war over time, or how risk will unfold in the future. A hypothetical 20-fold difference between a 2% and 0.1% chance of nuclear war per year could be historically invisible: a difference between expecting one or more nuclear wars in one’s life going forward, and most likely avoiding nuclear war for centuries. While long-term estimates that compound probabilities over decades are inherently suspect given changing conditions, it is clear that for high-consequence events, small absolute differences in probability can be extremely important over time.

One practical implication of this insight is the trade-off between avoiding intense short-term nuclear crises and stepping back from strategic situations that pose an elevated risk over time. In both the Cuban Missile Crisis and the years prior to the Intermediate-Range Nuclear Forces (INF) Treaty, the U.S. and Soviet Union found themselves facing perceived decapitation threats to their leadership from land-based missiles that could arrive on much shorter notice than other nuclear weapons. If there had been no Cuban missile crisis, and instead both Russian missiles remained in Cuba and U.S. missiles in Turkey without forceful negotiation toward removal, both countries would have remained in an equilibrium with higher than usual fear of pre-emptive attack and accordingly higher risk of false alarm.

Negotiated Reductions vs. De-Escalatory Reductions

Unlike Russia’s missile deployments to Cuba, NATO’s 1979 double-track response to Soviet SS-20s deployments proposed up front to negotiate mutual deep reductions in intermediate-range missiles, while threatening to deploy more capable missiles should negotiations fail. In theory, this offered the Soviets relief from spending years worrying about the Pershing II, which they believed could hit Moscow with accurate earth-penetrating warheads with minimal warning (despite its shorter range design and the cancellation of penetrating warheads). In practice, the INF Treaty was delayed until 4 years after NATO made its missile deployments to Europe.

During the late 1970s and early 1980s, the Soviet KGB and East German Stasi spent hundreds of millions aiming to subvert peace groups against NATO’s negotiating strategy. While years of activist pressure likely helped reach the final agreement, the Soviet prospect of unilaterally preventing or reversing NATO deployments and gaining concessions on other NATO forces likely increased the value of delay to Soviet negotiators in the first place. In the presence of Soviet misperception and opportunity for competitive behavior, NATO’s efforts toward lower-risk negotiation when combined with asymmetric activist pressure plausibly backfired into delaying mutual reductions following a period of elevated risk.

In general, competitive behavior presents the threat that risk reduction and arms control efforts can backfire due to mistrust and ambiguity of intentions. Signaling that one side is more desperate for risk reduction than the other can entice the other to extract concessions. Efforts that are perceived to have a secret competitive motive can inspire counterproductive reactions, such as the USSR accelerating its bioweapon efforts in response to the U.S. push for the bioweapons convention. In a state of mistrust, trust-building and transparency measures can be threatening: smaller nuclear arsenals meant to deter nuclear states with a larger number of weapons are relatively more dependent on secrecy to remain survivable.

The Security Dilemma

Military responses and investments designed to deter or build negotiating power toward reductions may instead foster insecurity and lead to arms races or escalatory spirals. This general situation, the security dilemma, highlights the inherent tension between conflicting risk reduction strategies. In an equilibrium of deterrence, even seemingly purely defensive systems can be escalatory as they reduce mutual vulnerability that deters either side from initiating direct conflict. While missile defense can complicate an adversary’s offensive planning, help reassure allies, and potentially allow for de-escalatory off-ramps in response to smaller-scale nuclear launches or accidents, increasing the reliability and size of such defenses could also instill escalatory fear of first strike by enabling less perfect offensive nuclear planning. The U.S. withdrawal from the anti-ballistic missile treaty, and the following race by Russia, China, and North Korea to deploy a variety of new nuclear systems to defeat missile defenses highlight this dynamic.

The Stability-Instability Paradox and Redirecting Competition

The nuclear stability-instability paradox is the well-evidenced observation that nuclear deterrence reduces the probability of war between nuclear states but, emboldens such states to engage in lower levels of conflict and proxy wars with greater confidence. While it can be observed in the dynamics of the Ukraine war or skirmishes between India and Pakistan, it can also be seen in Chinese opposition to certain risk reduction measures: perceiving them as guardrails for provocative behavior.

The general issue at stake is that providing stable incentives in one area of strategic competition tends to redirect each state’s competitive efforts in other directions that may or may not be lower risk. While effective deterrence by denial at lower levels of conflict likely dissuades some escalatory spirals from beginning, when underlying conflicts are not resolved, such restraints may build pressure for more discontinuous escalatory jumps. At higher levels of conflict, nuclear weapons are not the only mass consequence threat to deter; avoiding arms races and escalation with other weapons of mass destruction may become increasingly important as new kinds of mass precision threats and bioweapons become possible with Artificial Intelligence in addition to already existing WMDs.

Off-Ramps vs. Accountability

Another issue when thinking about risk over time is the conflict between de-escalatory off-ramps and accountability. Amid highly destructive conflict, face-saving interventions to allow leaders to gracefully de-escalate may inadvertently embolden future leaders to commit similar missteps, hoping to dodge repercussions. Coordinating norm enforcement relies on common knowledge of norm violation by those capable of enforcing norms, but the types of discreet and indirect dialogues that can enable leaders to reach agreement fundamentally rely on plausible deniability. In general, there is often a trade-off between creating and sustaining good incentives for leaders in the future, vs. their immediate actions. This is made more complex when there’s uncertainty about which norms are truly beneficial to enforce: efforts to scale moral coalitions and increase transparency can inadvertently foster groupthink and lock in undesirable norms.

Disentanglement, Red lines, and Seas of Ambiguity

Hoping to deter specific behaviors, leaders may communicate discrete “red lines” or instead keep their escalation thresholds more ambiguous to encourage their adversaries toward more continuous caution. Discrete thresholds are clear, but may lack credibility, or encourage an adversary to race right up to the boundary. Ambiguous escalation thresholds on the other hand may enable more opportunities for gradual escalation management, but at the risk of providing a slippery slope to higher intensity conflict. Disentanglement: separating nuclear and conventional warfighting systems (i.e. missiles, satellites, etc.), may be a way for states to signal more credibly that attacks on their nuclear systems would cross an escalatory threshold. At the same time, disentanglement comes at the cost of efficiency and reduced deterrence of attacks against conventional systems due to the lower risk of inadvertent nuclear escalation.

Is Risk Reduction Hopeless?

Reducing risk is a complex enterprise when competitors can take offsetting behaviors at different levels of competition, in different domains, at different times, and with very different types of threats. The probability of a threat can be traded for the intensity of a threat. Mistrust, misperception, and mistakes can make it hard to intervene to reduce risk decisively or to send unambiguous de-escalatory signals. Seemingly irrational risky behaviors sometimes have a longer-term logic in terms of reputation for resolve. These kinds of considerations paint an incomplete picture of the problem of adaptive competitive behavior for decisive risk reduction.

Decisive Risk Reduction: Principles and Interventions

Given the potential for states to respond to attempts at risk-reducing interventions with more competitive behavior or risk-manipulating strategies, what are the prospects for nuclear risk reduction on net? In this paper I reason through the particulars and limits of risk-compensating behavior, drawing analogies from the literature on risk compensation in the presence of safety regulations. In short, opportunities to decisively reduce risk can be found where risk compensation is absent or imperfect and where interventions are robust despite compensation.

When Risk Compensation is Weak

Invisibility, Illegibility, Secrecy, and Blindness

Strategic actors cannot precisely respond to that which they do not perceive or do not know about in the first place. When risk reduction efforts leverage secret, illegible, or invisible actions they may have a chance of reducing risk without any countervailing reaction. Such actions can be both cooperative or competitive in nature: in principle, a state could hold risk-reducing capabilities that would otherwise be fragile or generate a security dilemma if known, and it could have secret de-escalatory response plans that could only be exploited by an adversary for an advantage if known precisely.

Knowledge that the other side keeps secrets can itself however generate fear, and efforts to counter speculated unknowns. The process by which secrets are made, and secret makers are selected may be a potentially important signal: for both reassurance and deterrence alike.

Even without any secrets, attention is limited and achieving visibility and understanding can be a very costly process for large bureaucracies. Changes in one area of military policy or posture may be easily detectable in principle but could be missed when a country’s defense and intelligence bureaucracies fixate on other areas of defense intelligence.

Beyond bureaucratic inertia, motivated reasoning by leaders can still result in blindness from a failure to process easy-to-access, unambiguous analysis. Leaders who cultivate cluelessness of consequences for those they do not care about can more easily focus on their objectives while evading accountability for the unintended side effects of their actions.

To avoid competitive risk compensation, it is essential to assess simply how adaptive and responsive a competitor is: to understand their blind spots and biases, their culture, organizational behavior, and incentives. How and through what processes do they perceive risk? How are each side’s leaders psychologically similar or different, based on their individual histories, and the forces that attract and select people for leadership in their respective political systems? How internally or externally driven are their decisions? Sometimes, seemingly irrational decisions based on sunk costs or disproportionate vengeance have reasonable explanations in terms of political costs, status, reputation, and signaling commitment. Understanding these organizational irrationalities can help inform risk reduction strategies: even if a unilateral risk reduction measure is detected and could be exploited for an advantage, the time it takes for an unresponsive bureaucracy or political system to react can buy a window of decreased risk.

Misaligned Incentives: Reducing Risk for Whom?

Interventions that do not impact decision-makers or their incentives are unlikely to shape their behavior. If the only path for a state to undermine risk reduction efforts for an advantage is for its leaders to take actions that are highly inconvenient, or at odds with their ideology, then compensatory behavior is likely to be restrained. Similarly, risk reduction efforts that target the negative externalities of competition and conflict can hold promise for avoiding risk compensation. Not every decision-maker cares about everyone, and interventions that reduce risks for those that don’t factor into decision-maker calculus can provide the opportunity for risk reduction without triggering a competitive response.

One potential example would be targeting and tailoring deterrence efforts toward decision-makers and what they care about; mitigating the risks to others to the degree possible without adversely impacting the incentives faced by decision-makers. As states pursue larger numbers of precision weapons, conventional and lower-yield nuclear weapons will be able to hold far more targets at risk without warheads in the hundreds of kilotons. Such developments may open future paths to sustain deterrence that are hundreds to thousands of times less destructive than alternative war plans. The backfire threat of such interventions is the degree to which they create instability: if countries see each other as more likely to engage in pre-emptive attacks, and perceive fewer unintended harms from their own attacks (i.e. by avoiding the use of high-yield weapons near cities), the likelihood of war may increase. Even if leaders don’t care about foreign civilians at all, unnecessary economic damage and pressure from the international community can still deter actions with high collateral damage. On the other hand, lower collateral damage deterrent threats are inherently more credible, and if all sides retain survivable deterrent forces, similar mutual deterrence regimes to today might be sustained with substantially lower societal and environmental risks overall. For the reasons above, the transition to such a globally safer deterrence equilibrium could be fraught with peril and would have to be pursued with great caution to prevent instability.

A more reliable way to exploit misaligned incentives for risk reduction is by searching for “low-hanging fruit” where few institutions have the incentive and funding to look. Compared to other institutions, states have the most money and military information with which to engage in comprehensive rigorous analysis, but also the tendency to have their attention contorted by the news, budget, and election cycles. Defense industry-funded analyses likewise can be rigorous, but also contain a bias toward profitability and the diffusion of jobs across more congressional districts. Independent private philanthropy provides some opportunity to search for future opportunities protecting people regardless of where they live.

Where is Competitive Risk Adaptation Impossible?

Even when people have the incentive and awareness to engage in risk compensation, there are not always opportunities to do so. An actor may have limited capabilities, few avenues for action, or be subject to tight controls and regulation. This situation however rarely applies to state leaders: in the absence of strong international institutions or other comprehensive domestic restraints, there are many ways for leaders to engage in risk compensation if they desire to do so. Accordingly, competitive risk adaptation is easiest to mitigate by denial on the domestic front: securing nuclear command and control to prevent unauthorized rogue individuals from manipulating the risk of nuclear use.

Toward Robust Risk Reduction

Even when competitive risk-compensating behavior is present, there are a variety of approaches that may still reduce risk. Interventions that are robust across many scenarios can help address uncertainty that would otherwise make attempts at optimization perilous.

Such robust approaches would employ holistic reasoning across multiple kinds of risks: both directly and in terms of decision-makers’ incentives toward escalatory behavior or arms racing in the most dangerous directions. Such approaches require integrative analysis and experts with broad situational awareness: likely quite analogous to at least in some ways to net assessment.

Differentially Defensive Strategies?

One path to net reduction in risk is to search for approaches to defense that are more inherently defensive, deterring, or more enabling of arms control than offense. However, as discussed earlier, such initiatives may fail if not credibly defensive or if too cost-inefficient to be practical. Economies of scale in short-range conventional systems may be one way to square this circle and effectively create defensive geography, but this may fail when the threat of invasion is the primary deterrent of a competitor. Defensive alliance structures likewise can add selective constraints on escalatory ally behavior. In the nuclear domain, boost-phase missile defense strategies that would only work against countries with minimal defensive depth, such as North Korea, may be relatively assuring to Russia and China compared to those with greater potential for deeper global coverage.

Credible Communication

When up against skeptical, rationally paranoid competitors, being able to credibly signal intentions and set forth expectations is extremely important. Credible messages likewise may differ based on the kind of competitor they must convince: breaking through polarization or the information control strategies of authoritarian systems that may otherwise cause accidental self-deception. Sending clear credible messages in spite of such challenges requires better-calibrated strategic empathy: responding to sincere concerns, without being fooled by those invented for the purpose of risk manipulation.

Credibly anchoring intentions is vital when one’s side’s sheer military, technological, economic, or influence capabilities can erode the other side’s faith in one’s self-restraint. A similar theme can be found in the maxim to be “strategically predictable, but operationally unpredictable” to provide reassurance while remaining ready to deter aggression. When leaders face losses and deep mistrust toward the other side’s strategic intentions, they become far more likely to escalate, initiate wars, and gamble for redemption within conflicts. While often characterized as irrational, such behavior patterns of “risk seeking in losses and risk aversion in gains” are characterized by prospect theory and can be an optimal survival strategy in the face of uncertain risk environments. Accordingly, the degree to which credible communications provide a competitor with certainty that the status quo will not result in substantial losses is the degree to which they can be reassured and avoid escalation.

Credible Conditional Strategy

Credibly providing a path for a competitor to avoid loss is not reasonable if they position themselves to face loss in the event they fail to extract concessions or territory. Conditional strategies can provide paths toward positive-sum cooperation by default, while leaving open paths toward competitive strategy should certain conditions not be met. These sorts of dual-track negotiating approaches, when credible, help achieve the conflicting needs of concentrating negotiating power while still trying to build trust and de-escalate. The trade-off such strategies face is that sending credible signals while hedging against more than one possible future may be disproportionately costly, and benefit from having a larger economy.

For democracies, the higher audience costs that their leaders face for backing down can help make such strategy commitments more credible but may be a vulnerability if influence operations and disinformation campaigns sway public opinion. A public that is highly resistant to both foreign and domestically sourced disinformation accordingly helps to make both reassurance and deterrence more credible.

Independent Institutions and Third-Party Efforts

Even when the leaders of states and their institutions aim to speak as honestly and accurately as possible, their potential motive to persuade and access to private information will tend to result in perceptions of selective revelation and spin bias: engendering mistrust from their competitors. Independent institutions and third parties can help mitigate the problems this poses by providing a path for more credible information sharing, verification, and generating trusted common knowledge. To the degree such institutions or third parties do not carry obvious biases from their funding sources or affiliations, they may be able to offer mutually beneficial arms control and risk reduction proposals that are less likely to be viewed as a competitive trick.

The Obvious: Resolving or Realigning Political Conflict

While often intractable, another of the more obvious solutions to competitive risk compensation is simply to resolve the underlying political conflicts between competitors and to arrive at more mutually beneficial relations. Given the prevailing difficulty of halting state competitions entirely, another productive avenue can be to redirect competition into less threatening directions, such as relatively positive-sum economic competition.


Nuclear risk reduction is a complex task, especially when faced with the potential for state leaders to reactively manipulate risk for an advantage. Efforts that do not consider risk probabilities and consequences over time or that don’t analyze the incentives that result from risk reduction interventions are unlikely to fully achieve their intended effects. Despite the power of state leaders to engage in many forms of competitive risk compensation, such behavior can be avoided with risk reduction efforts that avoid leaders’ attention and incentive to react. Likewise, strategies that improve understanding to reduce unnecessary sources of conflict or that combine credible positive-sum intentions with deterrence can help to reduce risk on net by undermining the logic of competitive risk compensation in the first place. While many types of risk reduction intervention can be undermined or even zeroed out by competitive behavior, risk reduction is far from hopeless.

FacebookTwitterLinkedInEmailCopy Link